24888
April 2002
E THE WORLD BANK






Government at Risk






Government
at Risk
CONTINGENT LIABILITIES
AND FISCAL RISK
Hana Polackova Brixi
Allen Schick
editors
A COPUBLICATION OF
THE WORLD BANK AND
OXFORD UNIVERSITY PRESS



C) 2002 The International Bank for Reconstruction and Development /
The World Bank
1818 H Street, NW
Washington, DC 20433
All rights reserved.
1 2 3 4 04 03 02 01
A copublication of the World Bank and Oxford University Press.
Oxford University Press
198 Madison Avenue
New York, NY 10016
The findings, interpretations, and conclusions expressed here are those of the
author(s) and do not necessarily reflect the views of the Board of Executive Direc-
tors of the World Bank or the governments they represent.
The World Bank cannot guarantee the accuracy of the data included in this work.
The boundaries, colors, denominations, and other information shown on any map in
this work do not imply on the part of the World Bank any judgment of the legal
status of any territory or the endorsement or acceptance of such boundaries.
Rights and Permissions
The material in this work is copyrighted. No part of this work may be reproduced
or transmitted in any form or by any means, electronic or mechanical, including
photocopying, recording, or inclusion in any information storage and retrieval
system, without the prior written permission of the World Bank. The World Bank
encourages dissemination of its work and will normally grant permission promptly.
For permission to photocopy or reprint, please send a request with complete
information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers,
MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, www.copyright.com.
All other queries on rights and licenses, including subsidiary rights, should be
addressed to the Office of the Publisher, World Bank, 1818 H Street NW, Washington,
DC 20433, fax 202-522-2422, e-mail pubrights@worldbank.org.
Library of Congress Cataloging-in-Publication Data has been applied for.



Contents
Foreword                                        ix
Nicholas H. Stern, Senior Vice President and Chief
Economist, World Bank, and
Gobind T Nankani, Vice President, World Bank
Acknowledgments                                  xi
Introduction                                     1
Hana Polackova Brixi, World Bank,
and Allen Schick, University of Maryland
PART I LEARNING TO DEAL WITH
FISCAL RISKS IN GOVERNMENT PORTFOLIOS
POSSIBLE ANALYTICAL AND INSTITUTIONAL FRAMEWORKS
1. Dealing with Government Fiscal Risk: An Overview  21
Hana Polackova Brixi, World Bank,
and Ashoka Mody, IMF
2. Accounting and Financial Accountability to Capture Risk  59
Murray Petrie, The Economics and Strategy Group
3. Budgeting for Fiscal Risk                        79
Allen Schick, University of Maryland
4. Institutional and Analytical Framework for Measuring
and Managing Government's Contingent Liabilities  99
Suresh M. Sundaresan, Columbia University
V



vi                                             CONTENTS
5. Analytical Techniques Applicable to Government
Management of Fiscal Risk                        123
Krishna Ramaswamy, the Wharton School
6. Fiscal Sustainability and a Contingency Trust Fund  143
Daniel Cohen, Universite de Paris,
Ecole Normale Sup&rieure
7. A Framework for Assessing Fiscal Vulnerability  159
Richard Hemming, IMF, and Murray Petrie,
The Economics and Strategy Group
COUNTRY EXAMPLES
8. Evaluating Government Net Worth in
Colombia and R6publica Bolivariana de Venezuela  181
William Easterly and David Yuravlivker, World Bank
9. The Challenges of Fiscal Risks in Transition:
Czech Republic, Hungary, and Bulgaria            203
Hana Polackova Brixi, World Bank, Allen Schick,
University of Maryland, and Leila Zlaoui, World Bank
10. Analyzing Government Fiscal Risk Exposure in China  235
Kathie L. Krumm, World Bank, and Christine P. Wong,
University of Washington
11. Dealing with Contingent Liabilities in
Indonesia and Thailand                           251
Hana Polackova Brixi, World Bank,
and Sudarshan Gooptu, World Bank
12. Dealing with Contingent Liabilities in Colombia  269
Juan Carlos Echeverry, Ver6nica Navas, Juan Camilo
Gutierrez, and Jorge Enrique Cardona,
Government of Colombia
PART II DEALING WITH SPECIFIC
SOURCES OF GOVERNMENT FISCAL RISK
ANALYTICAL AND MANAGERIAL TOOLS
13. Pension Guarantees: A Methodology for Assessing
Fiscal Risk                                      283
George G. Pennacchi, University of Illinois



CONTENTS                                              vii
14. Measuring and Managing Government Contingent
Liabilities in the Banking Sector                311
Stijn Claessens, University of Amsterdam,
and Daniela Klingebiel, World Bank
15. Government Insurance Programs:
Risks and Risk Management                        335
Ron Feldman, Federal Reserve Bank of Minneapolis
1 6. Contingent Liabilities of the Central Bank:
Analyzing a Possible Fiscal Risk for Government  355
Mario Blejer, Central Bank of Argentina,
and Liliana Schumacher, IMF
PRACTICE
17. Contingent Liabilities in Infrastructure:
Lessons from the East Asian Financial Crisis     373
Ashoka Mody, IMF
18. Monitoring Fiscal Risks of Subnational Governments:
Selected Country Experiences                     393
Jun Ma, Deutsche Bank
19. Guarantees as Options: An Evaluation of Foreign
Debt Restructuring Agreements                    419
Sweder van Wiinbergen, University of Amsterdam
and the Centre for Economic Policy Research,
and Nina Budina, World Bank
20. The Fiscal Risk of Floods: Lessons of Argentina  451
Alcira Kreimer, World Bank
Conclusion: Toward a Code of Good Practice
on Managing Fiscal Risk                          461
Allen Schick, University of Maryland






Foreword
IN PUBLIC FINANCE, it no longer suffices for analysts and institutions
to focus solely on budget revenues and expenditures. Recent history
demonstrates that fiscal performance and, in turn, economic develop-
ment can be seriously disrupted by the sudden, unexpected costs of
hidden contingent liabilities and other unanticipated fiscal risks.
During the second half of the 1990s, unreported contingent liabili-
ties and related fiscal risks contributed to economic crises and disrupted
growth in a number of developing countries, motivating stepped-up
efforts at the World Bank to devise new concepts and tools for analyz-
ing and managing public finance. With the aim of improving the analy-
sis of fiscal risks and supporting policy advice in this area, the Economic
l'olicy Unit of the Bank's Poverty Reduction and Economic Manage-
ment Network established the Quality of Fiscal Adjustment Thematic
Group. This book was produced as part of the effort by this Thematic
Group to promote new thinking about public finance.
We now know that conventional frameworks for fiscal analysis that
concentrate on direct, explicit liabilities fail to address contingent fiscal
risks. For example, fiscal sustainability analysis that focuses, as is typi-
cally the case, on the officially reported budget deficits fails to detect
possible future increases in government debt and payments that may
emerge from both explicit and implicit government guarantees on en-
terprise credit, state insurance schemes, exchange rate guarantees, and
commitments to assist failed banks. Similarly, the government budget
process and documentation generally fail to scrutinize the substantial
claims on public resources that are associated with contingent liabili-
ties, realized and potential.
This book is a notable step forward in filling gaps in our under-
standing of fiscal risks and in developing suitable frameworks for man-
aging them. Through country cases and advances in conceptual design,
the book provides a menu of practical ideas for policymakers and
scholars for bringing fiscal risk within the ambit of public finance. It
demonstrates that government fiscal analysis needs to cover the entire
ix



x                                                   FOREWORD
portfolio of direct and contingent liabilities, as well as assets and the
revenue base. This in turn requires the government to identify, classify,
and assess its fiscal risks so that it can provide reliable estimates of future
payments that may ensue from past and pending liabilities. Only by iden-
tifying and measuring its exposure can a government bring its risks un-
der effective control. This task has been facilitated by the availability of
new methodologies, such as value-at-risk analysis and options pricing.
But despite these advances, governments still face technical challenges in
dealing with risk. However, the greater challenges are political and infor-
mational. Governments must care enough about fiscal and economic per-
formance beyond the short term to impose limits on future risk-taking
and to invest resources in identifying and controlling fiscal risks.
The essential challenge for governments is to launch a full and forth-
right effort to avoid excessive risk-taking and to prudently manage the
risks that they do take. Doing so typically requires governments to com-
mit themselves to greater transparency and broader fiscal discipline
than they have had in the past.
The benefits for governments that make the effort are enormous,
not only with respect to their future fiscal stability, but also with re-
spect to their capacity to achieve broader policy objectives. The ideas
and cases presented in this book should prompt governments to under-
take this effort.
At the World Bank, the analysis and control of fiscal risks has now
become an integral part of its assistance to member countries. Working
closely with policymakers, the Bank tailors its analytic support and
policy advice to country-specific circumstances, taking into account the
technical challenges (few countries have a reasonably complete inven-
tory of government contingent liabilities and other fiscal risks), the in-
stitutional set-up (fiscal and quasi-fiscal institutions and relationships),
and the political sensitivity of the issues (a full accounting of fiscal risks
often shows a government to be in a less favorable financial condition
than is reported in official statements).
There is no single approach to dealing with contingent liabilities and
fiscal risk. As the studies in this book indicate, governments that have
sought to control their risk exposure have taken several different ap-
proaches. This book points the way ahead by setting out general prin-
ciples of sound fiscal management and by providing specific examples
of innovative country practices.
Nicholas H. Stern             Gobind T Nankani
Senior Vice President         Vice President and Head of Network
Development Economics         Poverty Reduction and Economic
and Chief Economist           Management Network
The World Bank                The World Bank



Acknowledgments
WVE WOULD LIKE TO EXPRESS our sincere gratitude to all the distin-
guished authors for their valuable contributions to this book. We also
thank our many colleagues and friends who provided helpful advice
and suggestions. We acknowledge in particular the support and feed-
back on the papers that we received from the Quality of Fiscal Adjust-
ment Thematic Group (QFA TG) of the Economic Policy division of
World Bank's Poverty Reduction and Economic Management (PREM)
Network. Assistance from other World Bank staff, staff of other inter-
national institutions, government officials, academics, and researchers,
who provided valuable feedback on earlier drafts of the chapters, is
also gratefully acknowledged. Finally, we would like to thank the anony-
mous reviewers of this book whose challenging comments and sugges-
tions have greatly improved this text.
We acknowledge with thanks the financial support we received for
this book and related research and dissemination from the Economic
Policy division of World Bank's PREM Network through the QFA TG.
Support received from the Public Expenditure Thematic Group of the
Public Sector Division is also gratefully acknowledged.
The views expressed in this book are those of the authors and should
not be attributed to any particular institution, including institutions
with which individual authors may be associated, including the World
Bank, International Monetary Fund, the Federal Reserve System, and
others.
xi






INTRODUCTION
Government at Risk:
Contingent Liabilities
and Fiscal Risks
Hana Polackova Brixi
World Bank
Allen Schick
University of Maryland
ONE DOES NOT HAVE TO SEARCH far to see evidence of governments at
risk. In some countries, a long spell of seemingly sound fiscal manage-
ment was followed by a fast-moving crash that forced the government
to spend unbudgeted resources to pay obligations that few knew ex-
isted. When economic conditions stabilized, the government was left
with elevated levels of debt and other obligations. Similarly, in other
countries government debt soared much faster than the official deficit
would have suggested, because the government was compelled to make
good on obligations that had been assumed to be outside the budget-
ary framework. Today, in still other countries, the public finances are
facing a difficult future as the off-budget obligations accumulated in
the past come due in the decades immediately ahead. In all these sce-
narios, governments at risk have faced or are facing major fiscal chal-
lenges as a result of their contingent liabilities, which tend to remain
outside the framework of conventional public finance analysis and
institutions.
When risks come due in developed countries, they impose costs on
government budgets and sometimes temporarily reduce economic out-
put. When governments in less-affluent countries take imprudent risks,
the effects of such risks may spread quickly across the economy, cause
1



2                                          BRIXI AND SCHICK
capital to flow out to safer havens, possibly compel the government to
change economic course, and so retard or even reverse the country's
development. Moreover, if the mechanisms for information disclosure
are weak and market institutions not well developed, the risks assumed
by government generate a bias in the behavior of economic agents and
moral hazard in the markets, and thus work against development even
before they are realized.
In many countries, the reality or prospect of unbudgeted fiscal risks
coming due has been a wakeup call to extend fiscal management be-
yond the budget to all actions and transactions that put the govern-
ment in financial jeopardy. Doing so is difficult, however, because it
requires government to enter uncharted fiscal territory, where analyti-
cal frameworks are sometimes difficult to apply, accounting standards
are underdeveloped or poorly enforced, and the data are inadequate or
hidden from public scrutiny. And, because contingent liabilities often
grow from fiscal opportunism, when policymakers seek to hide the
real fiscal cost of their decisions and to reduce the reported budget
deficit, bringing them under control may become first of all a question
of political will.
This book aims to provide motivation and practical guidance to
governments seeking to improve their management of fiscal risks.
Among other things, it addresses some of the difficult analytical and
institutional challenges that face reformers tooling up to manage gov-
ernment fiscal risks, and it describes the inadequacies of conventional
practices as well as recent advances in dealing with fiscal risk. It also
presents several untested ideas for developing new instruments for regu-
lating and valuing fiscal risks. In so doing, the authors recognize that
some novel schemes are not yet sufficiently developed to warrant im-
mediate application by governments. But pushing forward the frontier
of public finance today, the book aims to enhance the practice of fiscal
analysis and management in the future.
This volume has grown out of World Bank initiatives to assist coun-
tries that are working to understand and manage the fiscal risks facing
their governments. These initiatives, led by the World Bank Quality of
Fiscal Adjustment Thematic Group during 1998-2001, covered over
40 interested countries and involved partnerships with many govern-
ment agencies, leading universities, research institutions, international
agencies, and associations of practitioners around the world.
The Magnitude of the Problem
We define fiscal risk as a source of financial stress that could face a
government in the future. The book focuses particularly on the fiscal
risks that are realized when uncertain events occur-such fiscal risks



INTRODUCTION                                               3
are often associated with government contingent liabilities. Recent his-
tory has brought with it many examples of contingent liabilities that
challenge government finances. The explicit and implicit government
insurance schemes in the domestic banking sector that emerged from
the 1997 financial crisis in East Asia added some 50 percent of gross
domestic product (GDP) to the stock of government debt in Indonesia,
30 percent in Thailand, and over 20 percent in Japan and Korea. In the
1980s, similar schemes generated a fiscal cost of over 40 percent of
GDP in Chile and around 25 percent of GDP in C6te d'Ivoire, Uru-
guay, and Republica Bolivariana de Venezuela.' In the 1990s, Brazil
and Argentina saw their government debt escalate when the central
government had to bail out commitments made by subnational gov-
ernments. Government debt in Malaysia, Mexico, and Pakistan soared
from unexpected defaults on government guarantees that had been
issued to promote private participation in infrastructure. Several chap-
ters of this book illustrate that contingent liabilities may become the
most critical factor in a country's fiscal performance.
Empirical analysis of past increases in the stock of government debt
confirms that realized government contingent liabilities account for a
large share of those increases. Kharas and Mishra (2001) illustrate
across nearly 50 countries that large increases in the stock of govern-
ment debt cannot be explained by the governments' reported budget
deficits (see Figure 1). Calling the annual increase in government debt
that is in excess of budget deficit a "hidden deficit," Kharas and Mishra
show that hidden deficits have stemmed mainly from the cost of real-
ized contingent liabilities and realized risks in the government debt
portfolio (particularly the currency risk of government foreign debt
instruments). In some developing and transition countries, contingent
liabilities have contributed on average to hidden deficits of more than
2 percent of GDP annually over a period of more than 10 years. The
analysis by Kharas and Mishra also indicates that contingent liabilities
tend to be associated with speculative attacks and currency crises.
In the recent past, several factors have worked to increase govern-
ments' exposure to fiscal risk and their tendency to incur hidden defi-
cits. The rapidly increasing volumes and volatility of international
private capital flows have accelerated the growth of domestic finan-
cial systems but also have made these systems, and thus implicitly the
countries' fiscal authorities, more vulnerable. This condition was clearly
illustrated during the three years following the 1997outflow of foreign
capital from East Asia. Privatization and reduction of the explicit fi-
nancial role of the state allowed many governments to cut their bud-
geted expenditures, but required either explicit or implicit promises
that the government would come to the rescue should the private sec-
tor fail to deliver expected outcomes. Such guarantees and promises,
in turn, have increased the uncertainty of future public financing



4                                             BRIXI AND SCHICK
Figure 1. Average Annual "Hidden" Deficits
(percent of GDP over different 5- to 20-year periods from 1970s to 1990s)
Developed
Developing and Transition Countries       Countries
8
7
6
5
4
3
2
1
-3
-4
U                                 O       
Note: Graphs represent average annual increases in government debt unexplained
by reported deficits.
Source: Kharas and Mishra (2001).
requirements. Furthermore, these guarantees have boosted moral haz-
ard in the markets. Loans and investments with a full guarantee suffer
from insufficient analysis and supervision by creditors. Moreover, the
beneficiaries of poorly designed state insurance schemes tend to ex-
pose themselves to excessive risks. For example, in the United States
the generous benefits of the federal flood insurance program have re-
sulted in the excessive construction of homes in flood-prone areas (U.S.
GAO 1998). This market behavior makes it more likely that later the
government will be asked to provide financial support.
The Political Economy of Contingent Liabilities
and Fiscal Risk
Often fiscal risks, particularly those in the form of contingent liabili-
ties, arise from politics and fiscal opportunism rather than economic
policy. Policymakers tend to build up government contingent liabili-
ties to avoid difficult adjustment and painful structural reforms. In



INTRODUCTION                                                s
this process, credit guarantees replace budgetary subsidies; take-or-
pay contracts come in lieu of liberalizing prices and restructuring the
energy, water, and other vital sectors; "letters of comfort" allow insol-
vent enterprises and banks to avoid bankruptcies; and so on.
In some instances, government support in the form of contingent
liabilities may be justified. In Europe and the United States, few would
argue against the immediate provision of a government guarantee to
cover the legal liability of airlines after the September 2001 terrorist
attacks in the United States. Government support was deemed justi-
fied for the political risk negatively affecting the airlines. Government
guarantees, as the form of support, also were appropriate because the
airlines would, in their own interest, try to avoid further terrorism
onboard.2
In many cases, however, governments have assumed contingent li-
abilities either in pursuing low-priority objectives-that is, on pro-
grams that would not have withstood public scrutiny-or in using
off-budget support when other forms would have been more appropri-
ate. Some governments have provided letters of comfort to cover the
commercial risk of foreign investors that have taken a position in do-
mestic financial institutions or enterprises. But with the ensuing moral
hazard, bailouts have often followed, frequently financed directly from
special government funds rather than through the budget, and have
tended not to be a good use of public money.
A common example of off-budget forms of support that are not
appropriate is the provision of credit guarantees to enterprises that
continually incur losses. While the government may have a good rea-
son to support some of such enterprises (for example, the national
railways, if their losses are the result of government fare pricing policy),
budgetary subsidies or direct government loans would sometimes be
more effective and almost always less expensive.
Whether for railways or airlines, often the best response to calls for
government support is to encourage restructuring, privatize enterprises
and financial institutions (and recapitalize them in the process if needed),
break down monopolies, and liberalize prices. But the ease of issuing
government guarantees and other promises of future possible govern-
ment support allows the government to postpone these sometimes dif-
ficult and, in the short term, costly actions.
In most countries, government is able to offer a promise of future
contingent support without seriously considering the future cost to the
taxpayer. Governments doing their accounting and budgeting on a cash
basis have particularly wide scope to behave opportunistically.3 In con-
trast to commercial accounting practices, which in most developed
countries require firms to recognize the future cost of pensions and
other risks on their balance sheets, few governments disclose the pro-
spective costs of their off-budget commitments. At the time of signing



6                                          BRIXI AND SCHICK
a letter of comfort or a guarantee contract, the government is able to
claim no cost to the budget. In some countries, a variety of govern-
ment entities are able to take on such commitments-sometimes with-
out even informing the ministry of finance or any other authority. Other
countries have centralized the guarantee-issuing authority at the min-
istry of finance or similar entity, but still do not require such an entity
to report the government's off-budget commitments until they fall due.
And even after they fall due, government may just issue more debt or
find alternative ways to cover them, without ever recording the event
in any reports. As a result, guarantees and similar forms of off-budget
support that create contingent liabilities turn out to be a relatively
easy way for government to avoid scrutiny of the risks inherent in
channeling its support.
As these examples of the inappropriate use of off-budget forms of
government support indicate, fiscal opportunism that gives rise to gov-
ernment contingent liabilities tends to grow out of a narrow scope of
conventional fiscal analysis and fiscal management. Scrutiny that fo-
cuses solely on the government's cash-basis budget and deficit invites
policymakers to generate contingent liabilities and other fiscal risks
outside the budgetary framework (see Box 1). Many countries have
learned firsthand that a narrow focus on cash-basis budget, deficit,
and debt compels governments to delay investments and structural
reforms, run down public assets, raise temporary revenues (sometimes
by assuming long-term liabilities in exchange for cash), and distort
spending priorities and the timing as well as the form of government
support (see, for example, Forte 1998, Polackova 1998, and Easterly
1999). As reforms (such as pension reform, the downsizing of public
employment, and enterprise and bank restructuring) that may require
higher deficits in the short term are put on hold, fiscal opportunism
puts economic growth at risk. Whether contingent liabilities are as-
sumed in the effort to maintain the status quo and avoid reforms, or
just to provide government support outside the budget and thus con-
ceal its cost and financing, their existence generates uncertainties about
future public financing requirements and so threatens future fiscal sta-
bility and the country's development. In this context, Selowsky (1998)
has emphasized that reported deficit reduction does not necessarily
imply "quality" of fiscal adjustment, which has the dimension of
sustainability as well as efficiency.
Overall, development tends to be correlated with a shift in risk from
individuals and individual economic agents and sectors to the state. As
governments promote development and economic conditions improve,
policymakers are pressured to take on commitments they may have
avoided earlier. Social security programs, various state insurance
schemes (targeting various beneficiaries, including enterprises, devel-
opers, farmers, and depositors), umbrella guarantees covering agen-



]NTRODUCTION                                                       7
Box 1. Fiscal Risks as a By-product of Deficit Targeting
When loose rules for fiscal management are accompanied by pressure
for fiscal adjustment, vote-seeking politicians and budget-seeking bu-
reaucrats have additional cause for opportunism. Paradoxically, the
incentive to mask the true cost of risks often rises when government
comes under pressure to tighten its budget constraints. When pressure
for adjustment is slack, the government may have little incentive to
hide the costs of its financial commitments. But when stringent fiscal
targets are imposed, wily spenders have incentive to substitute illusory
adjustments for actual ones. And when proper accounting rules and
strong enforcement mechanisms do not accompany the targets, the
spenders have ample opportunity to evade the controls.
Various studies have shown that weak enforcement produces bud-
getary opportunism. In 1985 the United States enacted the Gramm-
Rudman-Hollings law, which promised to progressively reduce the size
of the deficit and to produce a balanced budget by 1991. But, as Schick
(1995) points out, instead of genuine austerity, the law spawned bud-
getary legerdemain that increased the government's exposure to fiscal
risk. The volume of loan guarantees escalated, the government was
slow in responding to a costly crisis in the banking sector, and it adopted
policies (such as asset sales) that weakened its long-term fiscal posture.
In effect, as Rubin (1997) shows, faced with the Gramm-Rudman con-
straint on fiscal deficits, the U.S. Congress has reduced direct lending
by $50 billion and increased loan guarantees by $178 billion, replacing
budgetary outlays by explicit contingent liabilities.
In the process of fulfilling all the criteria for EU membership, the
Maastricht criteria on government deficit and debt were applied by
some countries in ways that escalated fiscal risk. The ploys, well de-
scribed by Forte (1999), included defining government narrowly, so
that the finances of various state-owned or controlled institutions were
not included in the calculation; hiding a portion of the governments
debt in various nongovernmental accounts; substituting guarantees for
loans and grants; recording subsidies as purchases of assets; devising
off-budget expenditures in lieu of direct financing; and deferring ex-
penditures on infrastructure and maintenance. Some European Union
governments received one-time payments from enterprises in exchange
for assuming future pension liabilities; others reduced their reported
public debt by reclassifying certain state enterprises as private entities.
in Italy, the railways have raised funds through the financial markets
to cover their deficits for many years with government agreement and
an explicit guarantee from the treasury. Yet, those operations had no
impact on the measured fiscal deficit or on the measured stock of gov-
ernment liabilities (Glatzel, 1998).
(Box continues on the following page.)



8                                              BRIXI AND SCHICK
Box 1 (continued)
Creative accounting and budgeting practices have been used also in
developing countries to portray their fiscal condition in a much more
favorable light than is warranted. When pressured by adjustment pro-
grams administered by the World Bank or IMF, some developing coun-
tries have privatized assets, disinvested in infrastructure and other public
goods, and replaced subsidies by directed credit and credit guarantees.
Widespread recourse to illusory adjustments has led Easterly (1999) to
conclude that when outside institutions demand a reduction in the deficit
or debt, the affected government often responds by creating a fiscal
illusion: achieving more favorable deficit and debt figures while di-
vesting assets, accumulating contingent liabilities, and in other ways
eroding the government's net worth.
cies in particular lines of business (for example, agricultural credit and
guarantee funds, housing funds, and export credit funds), and specific
guarantees that cover anything from borrowing by state-owned enter-
prises to commercial risks facing private investors, all tend to grow
significantly as countries progress in their development.
Transition and emerging market economies face particularly large
fiscal risks. Their dependence on foreign private financing, weak regu-
latory and legal enforcement systems, opaque ownership and distorted
incentive structures, inadequate information disclosure, and the weak
disciplinary effects of the international financial markets tend to in-
crease the incidence of failures in the financial and corporate sectors.
Such failures in turn often generate political pressure on governments
to intervene through bailouts. A history of bailouts, particularly if
coupled with a lengthy tradition as a paternalistic state, only contrib-
utes to the spread of moral hazard in the markets.
In addition, transition and the emergence of new markets involve
enormous risks: by entrepreneurs in starting new businesses or ac-
quiring old ones; by investors in providing venture capital; by im-
porters and exporters in building new trade opportunities; by farmers
in facing volatile prices and competition; by state-owned enterprises
in taking on excessive risk pursuing profit or being barred from charg-
ing market prices; by workers in seeking employment free of govern-
ment intervention. Understandably, some economic agents seek to
transfer the risk to government explicitly by obtaining guarantees or
other forms of assurance that government support will be forthcom-
ing. Without extensive guarantees, there is some likelihood that pri-
vate enterprise will be stillborn or stunted, the inflow of capital will



INTRODUCTION                                               9
be inadequate, investors will be unwilling to acquire state enterprises,
and depositors will be reluctant to place their money in domestic
banking institutions.
Markets that have a short history and offer limited information
restrict the understanding that investors as well as politicians have of
the risks they are taking. Imperfect knowledge induces both investors
and politicians to underestimate the future potential cost of their deci-
sions. Underestimation tends to be greatest when costs are contingent
on future occurrences, such as repayment of loans or the performance
of enterprises, and when the government bears implicit obligations
that depend on future decisions, such as on whether to make good on
uninsured bank deposits. This factor explains in part why in many
transition and emerging market economies creditors have tolerated
excessive risk exposure by domestic financial institutions and enter-
prises before fleeing. During the early years of change in an economic
system, it is tempting for politicians to take the position that risks are
justified because they enable the economy to grow more robustly. Later,
politicians often feel that they have no choice but to assist troubled
enterprises and financial institutions. As economies integrate with the
international markets, more reliable data become available and more
scrutiny is demanded. This tendency enhances the ability of both gov-
ernments and investors to estimate risks with standard methodologies.
Investors are then more likely to become more cautious in buying gov-
ernment debt instruments and thus to subject governments to greater
cliscipline.
Scope for Fiscal Opportunism
Across countries, the main sources of fiscal risk and their underlying
political economy tend to be similar. We now review the most com-
rnon examples and highlight the scope for fiscal opportunism that ex-
ists in the various cases. We do not claim, however, that fiscal
opportunism actually arises in all countries and in all such cases.
The largest scope for fiscal opportunism is traditionally offered by
the financial sector. Governments are accustomed to using financial
institutions, private or state-owned, to finance various projects and
support programs. Development banks, policy banks, and credit and
guarantee funds are authorized by the government to borrow in the
markets to finance its programs. They raise resources to build roads,
power plants, or schools; provide credit to farmers, enterprises, or
health insurance funds-sometimes for new investment projects, other
times to cover operating losses; and offer guarantees to exporters or
developers. Because many such programs, although sometimes justifi-
able on policy grounds, are not profitable financially, financial institu-
tions accumulate liabilities without securing the revenues to pay them



10                                         BRIXI AND SCHICK
off. For the government, financial institutions appear to be a conve-
nient channel for promoting various agendas without directly burden-
ing the budget. But later, when the financial institution is unable to
roll over its debt, it ultimately needs government resources. If these
resources are provided directly from the proceeds of government bor-
rowing, the budget deficit remains unaffected. In some countries, gov-
ernment exercises substantial influence over the banking sector, and
financial institutions do not pursue the interest of their creditors and
depositors (without relying on government bailout). As noted earlier,
the result is widespread losses and possibly a banking crisis. In resolv-
ing a banking crisis, many governments again use the financial sector
to create a fiscal illusion. Often asset and management companies are
created for the sole purpose of raising revenues to recapitalize banks
outside the budgetary framework of the government (Klingebiel 2000).
In some countries, state-owned enterprises are the vehicle used to
implement programs of a fiscal nature. By giving state-owned enter-
prises the responsibility for providing unemployment benefits, pen-
sions, schooling, housing, and such, the government may again ensure
service delivery without directly burdening its budget. Later on, should
enterprises be short of resources to cover the cost of all these services,
the government would provide support-possibly through a financial
institution. When privatizing an enterprise, the government may have
to take over its obligations, which sometimes can be done in the form
of a guarantee issued by an autonomous privatization fund or credit
and guarantee fund (for example, by an environment fund to cover the
future environmental liabilities of the enterprise). On the other hand,
the government may be able to obtain a cash payment for assuming
some of the enterprise obligations, as was recently done in France for
enterprise pension obligations (Forte 1998). The government also may
require enterprises to charge artificially low prices for "necessities," in
effect avoiding the need to pay family transfers from the budget. To
cover the ensuing losses, the government may then issue a guarantee
on a credit to be taken by the enterprise from a commercial bank.
Ultimately, when the government has to provide financing it may do
so via a guarantee fallen due-again outside the budgetary framework.
Subnational governments are able to devise similar routes for op-
portunistic fiscal behavior. Many create their own financial institu-
tions to raise revenues for off-budget programs, issue guarantees, or
borrow directly in the financial markets. Because their obligations
appear to be backed implicitly by the central government, they find it
possible to raise funds even if the financial sustainability of their ac-
tivities raises doubts. Fiscal risks taken by subnational governments
are complicated by the fact that subnational policymakers may them-
selves rely on an implicit promise of the central government's help.
Depending on the political clout of each individual subnational gov-



INTRODUCTION                                               11
ernment, it may be untenable for a central government to let a
subnational government go bankrupt. Because most countries do not
have clear regulations or a monitoring system for subnational govern-
ment risk-taking, as financial markets develop, subnational govern-
ments tend to accumulate excessive obligations that eventually may
compel the central government to provide a bailout.4
In recent years, government promotion of private participation in
infrastructure, although often justified on policy grounds, has become a
major source of fiscal risks. The justifiable objective of promoting pri-
vate initiative may be diluted by the lack of political will to establish an
adequate pricing mechanism, unbundle monopolies, and introduce a
risk-sharing mechanism with the private developers and creditors. Ex-
plicitly through build-operate guarantee contracts, or implicitly through
a perceived responsibility for the provision of core services, the govern-
ment may be called on to step in with financing in case of failure.5
Public pension and health schemes are another common, albeit pre-
dictable, source of fiscal risk for governments. Arguably, society is
better off when such risks are pooled and when the pool is expanded
to cover all citizens or residents. But whatever its advantages, pooling
transfers the risk to future rather than current government budgets. In
an aging society, a promise of high pension and health benefits affects
future government finances enormously. Most governments, however,
still do not consider this intergenerational impact.
Reforms
Reforming any of these areas or, more broadly, how government deals
with fiscal risk is likely to be costly politically, because constituencies
from pensioners to bank owners have reasons to oppose such reforms.
The average taxpayer would benefit-but he or she usually lacks lob-
bying power. Furthermore, such reforms imply that policymakers might
have considerably less scope for fiscal maneuvering in the future. Chap-
ters of this book address the various aspects of fiscal opportunism as
they arise in different circumstances and make recommendations for
improvements in the light of their political feasibility.
This book has been prepared in recognition of the probability that
national governments will continue to shoulder various fiscal risks.
The contributors to this volume recognize that risk-taking by the gov-
ernment is justified in some instances. But they take the position that if
risks are here to stay, they should be properly regulated and managed,
with appropriate information and oversight and full accounting for
the costs that may be imposed on government.
It is assumed here that a necessary first step toward fiscally prudent
policies is for policymakers to identify, classify, and understand the
fiscal risks facing the government. Comprehension of the fiscal risks



12                                          BRIXI AND SCHICK
and their consequences may encourage governments to avoid the risks
that are bound to surface within a politically meaningful time horizon.
For risks that extend beyond that time frame, achievement of fiscally
sound behavior may depend on market discipline. In particular,
policymakers are more likely to gravitate toward fiscally sound deci-
sions if the media, the public, investors, credit-rating agencies, and
multilateral institutions understand the government's fiscal performance
in its entirety and if there are sanctions when politicians expose the
state to excessive risks and then conceal those risks.
The contributors to this volume seek to identify institutional mecha-
nisms that can be applied domestically and internationally to optimize
the amount of risk-taking by government. Domestically, an agency
that is insulated from direct political pressures-for example, a su-
preme audit institution or an autonomous government debt manage-
ment office-can assess and report on the direct and contingent fiscal
risks of each government agency and of government as a whole. Al-
though voters do not necessarily care about government fiscal risk,
public explanation of the fiscal risks by an independent audit office
may encourage the international forces of restraint. To be effective,
international restraint should be used to ensure that the government
applies the international rules for fiscal analysis not only to the budget
and debt, but also to contingent liabilities. Specifically, international
pressure may compel the government to meet certain quality stan-
dards: to define, measure, and monitor its full fiscal performance, us-
ing sound indicators and methods as defined by international authorities
such as the International Monetary Fund, World Bank, European Com-
mission, or sovereign credit rating agencies and investors.
Organization of This Book
This book explores the problem of fiscal risk along two dimensions.
One dimension is that of the entire government portfolio of fiscal risks;
the other is that of selected specific sources of government fiscal risk.
Accordingly, the book is divided into two parts, each offering a con-
ceptual treatment of the issues along with country examples.
Part I begins with an overview of different approaches to dealing
with government fiscal risks (Chapter 1 by Hana Polackova Brixi
and Ashoka Mody). The overview offers a classification of fiscal risks
(the Fiscal Risk Matrix) and, with extensive references to the exist-
ing literature and country practice, summarizes various analytical
and institutional approaches toward government management of fis-
cal risk. In particular, it outlines an approach to managing govern-
ment fiscal risk in the context of the portfolio of government assets,
sources of future revenues, and direct and contingent liabilities; sets



1rNTRODUCTION                                            13
policy formulation in the context of fiscal risk management; and of-
fers some guidance on structuring guarantees and government pro-
grams to minimize their risk.
Next, the book explores the institutional factors affecting the op-
portunistic behavior of policymakers and suggests corrective measures
writh examples of good practice across countries. In this context, Chap-
ters 2 and 3 by Murray Petrie and Allen Schick, respectively, illustrate
the inadequacies of conventional cash-basis reporting, accounting, and
budgeting, and call for more comprehensive and sounder approaches.
These chapters particularly highlight the usefulness of requiring gov-
ernment to publish a statement of contingent liabilities and fiscal risk.
T hey also outline the benefits of accrual-basis accounting and budget-
ing for government fiscal risk management.6 These chapters provide
many examples of country practice from developed, transition, and
developing countries.
Subsequent chapters explore the practice of risk management in the
private sector and its applicability to government. Chapter 4 by Suresh
Sundaresan provides an overview of the analytical tools and practices
used by financial institutions and corporations to manage their risk
exposures, and then applies the methodology to valuing government
guarantees. Similarly, Chapter 5 by Krishna Ramaswamy applies a
factor model to government risk analysis-particularly for risk-taking
by public sector entities, including state-owned enterprises.
Then, linking the discussion of government and private sector prac-
tice, the chapters that follow focus on approaches to expanding fiscal
analysis to incorporate fiscal risks and to bringing market incentives
into the government's thinking about fiscal sustainability. Chapter 6
by Daniel Cohen integrates contingent liabilities with the traditional
fiscal sustainability analysis and offers an institutional arrangement to
introduce market discipline into government risk-taking. Chapter 7 by
Richard Hemming and Murray Petrie further expands the fiscal
sustainability framework and introduces a framework for assessing
the exposure of a government's future fiscal performance to risks.
The country examples in Part I offer additional conceptual ap-
proaches and illustrate some of the discussion in the earlier chapters.
Reflecting on the ability of policymakers to generate fiscal illusion,
Chapter 8 by William Easterly and David Yuravlivker applies a com-
prehensive approach to fiscal analysis in the form of evaluating the net
worth of the governments of Colombia and Republica Bolivariana de
Venezuela. Chapter 9 by Hana Polackova Brixi, Allen Schick, and Leila
Zlaoui recognizes the special challenges in fiscal risk management fac-
ing transition countries and evaluates various aspects of the quality of
fiscal adjustment and fiscal management related to government con-
tingent liabilities in the Czech Republic, Bulgaria, and Hungary. In Chap-
ter 10, Kathie Krumm and Christine Wong incorporate contingent



14                                         BRIXI AND SCHICK
liabilities into the fiscal sustainability analysis for China. Looking at
the portfolio of contingent liabilities and risks from the perspective of
government debt, Chapter 11 by Hana Polackova Brixi and Sudarshan
Gooptu presents scenarios for government debt management in Indo-
nesia and Thailand. In a similar spirit, Chapter 12 by Juan Carlos
Echeverry and others discusses reforms in dealing with contingent li-
abilities as implemented under the leadership of the Colombia Treasury.
Part II presents analytical and institutional approaches that gov-
ernments might consider when facing risks in specific government
programs or sectors. Chapter 13 by George Pennacchi focuses on the
risk of guarantees that are often taken on by governments imple-
menting pension reforms and utilizes an option-pricing methodology
to value a government's risk exposure. In Chapter 14, Stijn Claessens
and Daniela Klingebiel analyze the ways in which to measure and
reduce the government's risk exposure in the banking sector. Chapter
15 by Ron Feldman discusses risks arising from government insur-
ance programs. Recognizing the implicit responsibility of the fiscal
authorities (and thus of the government budget) for the central bank's
positive net worth, Chapter 16 by Mario 1. Blejer and Liliana
Schumacher utilizes a value-at-risk approach to assessing the central
bank's own risk exposure.
Looking at country experience and practice, Chapter 17 by Ashoka
Mody analyzes the lessons of the 1997 East Asian financial crisis for
private participation in infrastructure and associated government con-
tingent liabilities. For fiscal risks taken on by subnational governments,
Chapter 18 by Jun Ma offers a framework that would allow the cen-
tral government to monitor and discipline the subnational governments'
risk exposure and thereby reduce the associated exposure by the cen-
tral government. Chapter 19 by Sweder van Wijnbergen and Nina
Budina applies option-pricing methodology to evaluating government
foreign debt restructuring agreements for Bulgaria. Finally, the fiscal
risk of floods, particularly in the case of Argentina, is explored in
Chapter 20 by Alcira Kreimer.
Reflecting on available country experience and on the new con-
cepts presented in the book, the concluding chapter by Allen Schick
draws together a list of policy recommendations for governments seek-
ing to bring their fiscal risks under control.
The descriptions and discussions in this book of the concepts and
practices of dealing with contingent liabilities and other fiscal risk
suggest that a broad range of approaches for governments to use in
analyzing and managing such risks are available. In this respect, the
book illustrates that contemporary practices have yet to be standard-
ized. Under these circumstances, the book seeks to motivate policy-
makers and policy analysts to pay attention to the fiscal risks govern-
ments face, and it provides a rich menu of practices that may be applied
in countries that are serious about confronting such risks.



INTRODUCTION                                                   15
Notes
1. For an overview and analysis of the cost of banking crises, see Honohan
and Klingebiel (2000).
2. In addition, countries argued that they could not afford to let several
big airlines go under simultaneously, because the effect on jobs and confi-
dence could be too great. A temporary, targeted government subsidy to over-
come the temporary financial shock may be appropriate to smooth job
reductions and allow the strongest to survive. It should be acknowledged,
however, that the credit guarantees for which the airlines lobbied may have
the additional effects of delaying the restructuring in the airline industry
that, given the increasing losses of many airlines even before the attacks, was
considered overdue. For analysis of the pros and cons of various scopes and
forms of government support to the airline industry after the terrorist at-
tacks, see, for example, the Economist ("More Pain Ahead," September 22,
2001, and "Uncharted Airspace," September 28, 2001).
3. In cash-basis accounting, expenses and liabilities are accounted not when
the obligation is incurred, but only when the government makes the actual
cash transfer. Thus a government collecting a fee for assuming a liability (for
example, when it issues a guarantee or accepts the pension liability of an
enterprise under privatization) reports the transaction as a net revenue gain.
4. Dillinger (1999) discusses the economics and political economy of cen-
tral government bailouts of subnational governments in South America.
5. Irwin and others (1998) provide examples and analysis of public risk in
private infrastructure.
6. An accrual-basis accounting system without accrual budgeting will not
ensure that governments adequately consider contingent fiscal risks in policy.
Although this system encourages governments to prepare a statement of con-
tingent liabilities and financial risks, it generally does not require that the
liabilities be included in the balance sheet and that the associated risks be
evaluated and quantified. International accrual accounting standards require
that liabilities be accounted only when the obligation is due with certainty.
For a discussion of the rules of probability and risk assessment, see Interna-
tional Accounting Standards Committee (1997).
References
Dillinger, William. 1999. "Fiscal Management in Federal Democracies: Ar-
gentina and Brazil." Policy Research Working Paper 2121. World Bank,
Washington, D.C.
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy (April).
Forte, Francesco. 1998. "Accounting and Financial Practices in Light of Con-
text of the Maastricht Treaty." In European Union Accession: The Chal-
lenges for Public Liability Management in Central Europe. Washington,
D.C.: World Bank.



16                                            BRIXI AND SCHICK
Honohan, Patrick, and Daniela Klingebiel. 2000. "Controlling Fiscal Cost of
Banking Crises." Policy Research Working Paper 2441. World Bank, Wash-
ington, D.C.
International Accounting Standards Committee. 1997. International Account-
ing Standards 1997. London.
Irwin, Timothy, Michael Klein, Guillermo Perry, and Mateen Thobani, eds.
1998. Dealing with Public Risk in Private Infrastructure. World Bank
Latin American and Caribbean Studies. Washington, D.C.: World Bank.
Kharas, Homi, and Deepak Mishra. 2001. "Fiscal Policy, Hidden Deficits,
and Currency Crises." In S. Devarajan, E H. Rogers, and L. Squire, eds.
World Bank Economists' Forum. Washington, D.C.: World Bank.
Klingebiel, Daniela. 2000. "The Use of Asset Management Companies in the
Resolution of Banking Crises: Cross-country Experience." Policy Research
Working Paper 2284. World Bank, Washington, D.C.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Rubin, Irene. 1997. The Politics of Public Budgeting: Getting and Spending,
Borrowing and Balancing. Chatham, N.J.: Chatham House.
Schick, Allen. 1995. The Federal Budget: Politics, Policy, and Process. Wash-
ington, D.C.: Brookings Institution.
Selowsky, Marcelo. 1998. "Fiscal Deficits and the Quality of Fiscal Adjust-
ment." In The Challenges for Public Liability Management in Central
Europe. Washington, D.C.: World Bank.
U.S. GAO (General Accounting Office). 1998. Budgeting for Federal Insur-
ance Programs. Washington, D.C.: Government Printing Office.



PART I
Learning to Deal
with Fiscal Risks in
Government Portfolio






Possible Analytical
and Institutional Frameworks






CHAPTER I
Dealing with
Government Fiscal Risk:
An Overview
Hana Polackova Brixi
World Bank
Ashoka Mody*
International Monetary Fund
IMANY GOVERNMENTS HAVE FACED serious macroeconomic instabil-
ity as a result of obligations that were not recorded in any fiscal docu-
inents. Governments may have taken advantage of guarantees and
financial companies to implement their policies outside the budgetary
system, or they may have just been blind to risk spreading in the mar-
lcets. Whether the result of fiscal opportunism to conceal the true fis-
cal cost of government programs, or of an effort to find more efficient
ways to achieve policy objectives (by, for example, offering guarantees
instead of direct loans and cash subsidies), or of lenience toward moral
hazard in the behavior of market agents, such obligations often turn
out to be very costly. At some point in time, many guarantees fall due,
state insurance programs require subsidies, and banks involved in policy
lending or exposed to excessive risk with the hope of government bail-
out eventually file for such a bailout. And, as illustrated in Mexico in
1994, in East Asia in 1997, and, to some extent, around the world
after the September 2001 terrorist attack on the United States, these
hidden obligations tend to surface and require public resources all at
once in times of crises and economic slowdown.'
The author was with the World Bank when he wrote this article; he is currently with
the IMF.
21



22                                            BRIXI AND MODY
But the conventional approaches to public finance analysis fail to
reveal hidden government obligations and the associated fiscal risks.
Similarly, public finance institutions, including systems for government
budget management, debt management, accounting, financial control,
and public scrutiny, often remain blind to government contingent li-
abilities. Thus a string of years in which government has reported a
balanced budget and low public debt suggests neither that the govern-
ment has been fiscally prudent nor that it will enjoy fiscal stability in
the near future.2
This chapter begins by discussing a simple typology of sources of
fiscal risk-the Fiscal Risk Matrix. It then introduces the Fiscal Hedge
Matrix and expands on the standard government asset and liability
management framework. Using this framework, it outlines principles
for dealing with fiscal risks in their policy and institutional contexts.
The chapter then explores the possibilities for governments to reduce
their risk exposure by enabling market agents to better manage their
own risks, developing risk-sharing mechanisms, hedging, and building
reserves. Finally, the chapter offers a set of questions to assist policy-
makers in learning about fiscal risks in their own country. Overall, this
chapter introduces a number of topics that will be further elaborated
in the rest of the chapters in this volume.
The Fiscal Risk Matrix
The Fiscal Risk Matrix presented in Table 1.1 divides sources of fiscal
risk-that is, sources of future possible financing pressure on the fiscal
authorities of a country-into four groups according to the following
characteristics: direct versus contingent, and explicit versus implicit.3
Direct liabilities are predictable obligations that will arise in any event.
Contingent liabilities are obligations triggered by a discrete but uncer-
tain event. For government policies, the probability of a contingency
occurring and the magnitude of the required public outlay are exog-
enous (for example, the occurrence of a natural disaster) or endog-
enous (for example, the implications of market institutions and of the
design of government programs on moral hazard in markets). Contin-
gent liabilities also rise with weaknesses in the macroeconomic frame-
work, financial sector, and regulatory and supervisory systems, and
with weak information disclosure in the markets. They also emerge
from so-called quasi-fiscal activities-that is, activities of a fiscal na-
ture that the government pursues outside its budgetary framework.4
Explicit liabilities are specific government obligations defined by law
or contract. The government is legally mandated to settle such an ob-
ligation when it becomes due. Implicit liabilities represent a moral
obligation or expected burden for the government not in the legal sense,
but based on public expectations and political pressures.



DEALING WITH GOVERNMENT FISCAL RISK                                 23
7able 1.1 Government Fiscal Risk Matrix
Direct liabilities           Contingent liabilities
Sources of          (obligation                   (obligation if a
obligations        in any event)              particular event occurs)
Explicit      * Sovereign debt (loans    * State guarantees for non-
Government      contracted and securities  sovereign borrowing by and
liability as    issued by central          other obligations of sub-
recognized      government)                national governments and
by a law      * Expenditure composi-       public and private sector
or contract     tion (nondiscretionary     entities (development banks)
spending)                * Umbrella state guarantees
* Expenditures legally      for various types of loans
binding in the long term   (mortgage loans, student
(civil service salaries    loans, agriculture loans,
and pensions)              small business loans)
* Trade and exchange rate
guarantees issued by the state
* State guarantees on private
investments
* State insurance schemes
(deposit insurance, income
from private pension funds,
crop insurance, flood in-
surance, war-risk insurance)
Implicit      * Future public pensions   * Default of a subnational
A moral         (as opposed to civil       government or public/private
obligation      service pensions),         entity on nonguaranteed
of govern-    * Social security schemes-   debt/obligations
ment that     * Future health care       * Banking failure (support
reflects        financing,                 beyond government in-
public and    * Future recurrent costs     surance, if any)
interest group  of public investment     * Cleanup of liabilities of
pressures       projects                   entities being privatized
* Failure of a nonguaranteed
pension fund, employment
fund, or social security fund
(protection of small investors)
* Possibly negative net worth
and/or default of central bank
on its obligations (foreign
exchange contracts, currency
defense, balance of payments)
* Other calls for bailouts (for
example, following a rever-
sal in private capital flows)
* Environmental recovery, dis-
aster relief, military financing
a. In this framework, these services fall in the category of government direct implicit
liabilities if their provision is not mandated by law. If mandated by law, then these services
fall in the category of government direct explicit liabilities.
Source: Polackova (1998).



24                                         BRIXI AND MODY
Government direct explicit liabilities are legal or contractual obli-
gations of the government that will arise in any event. These obliga-
tions constitute the main subject of conventional fiscal analysis. They
are: the repayment of sovereign debt, expenditures based on budget
law in the current fiscal year, and expenditures in the long term for
legally mandated items such as civil service salaries and pensions and,
in some countries, even the overall social security system. Among these
items, recent literature has focused on risks embedded in the size and
structure of the government debt portfolio (see Nars 1997, World Bank
and IMF 2000, and Dooley 2000 for an overview).
Government direct implicit liabilities will also arise in any event,
but the government will not be legally obliged to act on them. Such
obligations often arise as a presumed consequence of public expendi-
ture policies in the longer term. Given their implicit nature, these ob-
ligations are not captured in government balance sheets. Typically,
they are high for demographically driven expenditures. For example,
future pensions payable in a public pay-as-you-go scheme, unless guar-
anteed by law, constitute a direct implicit liability. Its size reflects the
expected generosity of and eligibility for pensions and future demo-
graphic and economic developments. Among direct implicit liabilities,
recent literature has particularly explored public pension liabilities
(World Bank 1994; IMF 1996; OECD 2000; Bodie and Davis 2000).
Contingent explicit liabilities are government legal obligations to
make a payment only if a particular event occurs. Because the fiscal
cost of contingent explicit liabilities is invisible until they are trig-
gered, they represent a hidden subsidy, blur fiscal analysis, and drain
government finances only later. For that reason, state guarantees and
financing through state-guaranteed institutions look politically more
attractive than budgetary support even if they are more expensive later.
In the markets, contingent government obligations may immediately
create moral hazard, particularly if the government guarantee covers
the whole rather than a part of the underlying assets and all rather
than selected political or commercial risks. State insurance schemes
often cover uninsurable risks of infrequent losses that are enormous in
total magnitude. Thus, rather than financing themselves from fees,
they redistribute wealth and rely on government net financing. To date,
research has focused on issues of measurement and management of
loan guarantees (Mody and Patro 1996; Lewis and Mody 1997), in-
vestment guarantees in infrastructure (Chase Manhattan Bank 1996;
Irwin and others 1998), state development finance institutions (Yaron
1992), pension guarantees (see Chapter 13 by Pennacchi in this vol-
ume), deposit insurance (Leaven 2000; World Bank 2001a), crop in-
surance (Hueth and Furtan 1994), and other state insurance schemes
(U.S. GAO 1998, and Chapter 15 by Feldman in this volume).



DEALING WITH GOVERNMENT FISCAL RISK                         25
Contingent implicit liabilities depend on the occurrence of a par-
ticular future event and on government willingness to act on them.
Such obligations are typically not officially recognized until after a
failure occurs. The triggering event, the cost at risk, and the required
size of government outlay are uncertain. In most countries, the finan-
cial system represents the most serious contingent implicit government
liability. Experiences have indicated that markets expect the govern-
ment to help financially far beyond its legal obligation if stability of
the financial system is at risk (for examples, see Chapter 14 by Claessens
and Klingebiel in this volume and World Bank 2001a). Fiscal authori-
ties also are often compelled to cover losses and obligations of the
central bank, subnational governments, state-owned and large private
enterprises, budgetary and extrabudgetary agencies, and any other
agencies of political significance.5
The effect of fiscal risks arising from government direct and contin-
gent liabilities can best be analyzed in the context of an extended gov-
ernment balance sheet that includes future revenues as well as contingent
liabilities, and assets as well as direct liabilities. This approach builds
on the assets and liabilities management literature (Cassard and
Folkerts-Landau 1997; OECD 1999; World Bank and IMF 2000) and
can be referred to as an extended assets and liabilities management
framework. In this approach, a Fiscal Hedge Matrix (Table 1.2) comple-
ments the Fiscal Risk Matrix to illustrate the different sources of po-
tential revenues that can serve to cover government obligations. Sources
of government financial safety also can be divided into direct and con-
tingent, explicit and implicit. Direct explicit sources reflect the
government's legal power to raise income from its existing, tangible
assets. Direct implicit sources also are based on the existing assets, but
these are not under the government's direct control and thus may off-
set fiscal risks to a limited degree only. Contingent explicit sources
relate to the government's legal power to raise finances in the future
from sources other than its own assets. Finally, contingent implicit
sources are not available to the government until a particular situation
occurs and, even then, require the government to make a special case
for their utilization. The two matrixes (Tables 1.1 and 1.2), once filled
with country-specific items, would help a government to identify the
exact scope for its fiscal analysis and management.
The value of government assets and future revenues and the cost of
government obligations are associated with different types of risk. Clearly,
the government's residual, unhedged exposure to fiscal risk is the result
of the correlation among, rather than a simple summation of, the effects
of the different types of risks on the individual items in its extended
balance sheet. The types of risk include: refinancing risk (constraints on
the government's ability to issue debt-exacerbated particularly by short



26                                                BRIXI AND MODY
Table 1.2 Government Fiscal Hedge Matrix
Contingent sources of safety
Sources of     Direct sources of safety   (dependent on future events,
financial        (based on the stock         such as value generated
safety            of existing assets)           in the future)
Explicit      * Assets recovery (work-  * Government revenues from
Based on        out and sales of non-     resource extraction and sales
government      performing loans and    * Government customs
legal powers    sales of equity)          revenues
(ownership    * Privatization of state-  * Tax revenues
and the right   owned enterprises and     - minus tax expenditures
to raise        other public resources      (exclusions, exemptions,
revenues)     * Recovery of govern-         and deductions, which
ment loan assets            reduce taxable income)
(resulting from earlier   - minus revenue commit-
direct government           ment (to subnational
lending)                    governments)
- minus revenues sold
forward (commodity
forward sales) and pledged
as collateral (partly at risk)
* Hedging instruments and
(re-)insurance policies
purchased by the government
from financial institutions
Implicit      * Stabilization and contin- * Profits of state-owned
Based on        gency funds'              enterprises
government    * Positive net worth of   * Contingent credit lines and
indirect        central bank              financing commitments
control                                   from official creditors
* Current account surpluses
across currencies
a. Stabilization and contingency funds may be designated for a general or very specific
purpose and can be under direct or indirect government control. Thus their classification
may be different in each case.
Source: The authors.
maturities of and maturity bunching in government obligations), liquid-
ity risk (risk of having to sell assets at loss-intensified by maturity
mismatch between assets and liabilities and by rigidities in the
government's capacity to raise revenues and cut expenditures), currency
risk (exchange rate risk and cross-currency risk, exposure to short-
term exchange rate volatility-arising from the currency structure of
government debt and exchange rate guarantees, which is partly offset
by the currency structure of government assets and revenues), interest
rate risk (particularly associated with floating interest rates), commodity



DEALING WITH GOVERNMENT FISCAL RISK                       27
price risk (swings in the price of oil, rice, and similar commodities),
derivative risk (risk of large losses from the use of derivative instru-
ments), medium- and long-term sustainability risk (risk arising from
adverse trends that underlie government finances), political risk (risk
of policy reversals and political instability), and operational risk (poor
valuation and risk assessment, system errors, poor organizational struc-
tures, corruption, and fraud). The literature has mainly explored the
impact of most of these types of risk on government direct liabilities
(World Bank and IMF 2000) and of selected types of risk on govern-
rnent revenues and expenditures (see, for example, Larson, Varangis,
and Yabuki 1998 on the impact of commodity price). To analyze the
overall government risk exposure and its sensitivity to different risk
types, taking into account possible correlations among risks across the
entire extended government balance sheet, public finance will need to
build on tools developed in finance, such as portfolio optimization
and factor analysis (see Chapter 4 by Sundaresan and Chapter 5 by
Ramaswamy in this volume).
Dealing with Risk in Fiscal Analysis
and Fiscal Management
Because it is impossible for governments in a market environment to
avoid fiscal risk, they need to control and manage their risk exposure.
Dealing with fiscal risk is important not only from the perspective of
future fiscal stability. With respect to allocative efficiency, for example,
only with a view toward its likely full fiscal cost in the future can a
proposed government guarantee be properly scrutinized against other
competing programs. As for operational efficiency, only with a full un-
derstanding of the various types of risk involved can such a guarantee be
structured in a way to provide the desired support without unnecessar-
ily generating moral hazard and exposing the government to risk.
But do governments have the incentives and capacity to reflect fis-
cal risks in their policy choices and to carry out a good fiscal risk
management strategy? It depends on how well they understand the
issues and on the kind of scrutiny and pressures policymakers face in
dealing with fiscal risks and their consequences. There are good ex-
amples to build on. Australia and South Africa use a medium-term
expenditure framework to enhance predictability of fiscal performance
and, particularly in the case of South Africa, to make their govern-
ments accountable for their risk analysis as well as macroeconomic
and demographic assumptions.6 Canada, the Netherlands, and the
United States have incorporated analysis of selected contingent liabili-
ties and tax expenditures into their budgetary frameworks, requiring
budget allocations and reserve funds to reflect the present value of



28                                          BRIXI AND MODY
future potential outlays and foregone revenues (Congressional Research
Service 1998). Sweden and Colombia have authorized their govern-
ment debt management agencies to track and manage the risk of con-
tingent liabilities, and they require the beneficiaries of government
guarantees to pay the full present value of their expected fiscal cost
up-front into a reserve fund, which also is managed by the debt man-
agement agency (see Calderon Zuleta 1999 and Chapter 12 by
Echeverry and others in this volume). In India, the Federal Reserve
Bank and Ministry of Finance have carefully assessed the government's
exposure to fiscal risk across the entire Fiscal Risk Matrix and, as a
consequence, established a Guarantee Redemption Fund to cushion
the future fiscal cost of central government's guarantees. They also
implemented rules on subnational government guarantees and on
subnational guarantee funds to provide for the future expected cost of
subnational guarantees (Commonwealth Secretariat 2001).
The analysis and management of government exposure to fiscal risks
have three dimensions: the macroeconomic context, specific fiscal risks,
and the institutional framework. The macroeconomic context of the
government's exposure to fiscal risks relates to its capacity to absorb
financial pressures that it may realize in the future. How much room
for maneuver does the government have to absorb fiscal risks? Limits
on the government's absorption capacity for fiscal risks are determined
by possible macroeconomic constraints, such as the existence of a cur-
rency board and fixed exchange rate arrangements, and by the trends,
rigidity, and sensitivities of the general government's expenditures and
revenues, the size and liquidity of its assets, the structure of its obliga-
tions, and its future possible borrowing constraints. For example, does
the government have access to reliable sources of financing, such as
deep domestic bond markets, or is it critically dependent on the confi-
dence of foreign investors?
In this context, fiscal performance relates to developments in the
government's entire extended balance sheet (across the two matrixes
presented above). Thus fiscal sustainability analysis would be replaced
by an analysis of government net worth and of the future financial
pressures and financing options. In this volume, Chapter 8 by William
Easterly and David Yuravlivker applies the net worth approach to
Colombia and Republica Bolivariana de Venezuela. Focusing on obli-
gations, in Chapter 10 Kathie Krumm and Christine Wong integrate
contingent liabilities with fiscal sustainability analysis for China. In
Chapter 9 Hana Polackova Brixi, Allen Schick, and Leila Zlaoui try to
overcome the shortcomings of conventional deficit measurement and
calculate the "true fiscal deficit" and "hidden debt" of the govern-
ment of the Czech Republic, reflecting the cost of contingent liabili-
ties. For Bulgaria, they discuss the impact of the currency board



DEALING WITH GOVERNMENT FISCAL RISK                         29
arrangement on the government's vulnerability to fiscal risks. Chapter
7 by Richard Hemming and Murray Petrie outlines a framework for
analyzing fiscal vulnerability, which is broadly defined as the ability of
government to achieve its fiscal policy objectives.
Dealing with specific fiscal risks requires addressing the following
three areas of questions: First, what are the sources of fiscal risk? Par-
ticularly, which budgetary and off-budget government programs may
generate unexpected financial pressure on the government in the fu-
ture? Second, what types of risk is the government exposed to? For
example, are the cost of the government's debt service, the value of its
revenues, or the likelihood that government contingent liabilities will
become reality affected by movements in the exchange rate, domestic
or foreign interest rate, and commodity prices, or by the risks of its
own fiscal mismanagement and operational failure (institutional risks)?
Third, how sensitive is the overall fiscal position to the various sources
and types of risk? What are the possible stress scenarios linking the
realization of different fiscal risks, and what are the possible conse-
quences for the government fiscal position?
The goal of government fiscal risk analysis (and management) is to
ensure that government has the cash available to meet its obligations
and deliver on its policies, under any more or less likely conditions.
Therefore, less likely stress scenarios require as much attention as the
most likely baseline scenario.7 Results of stress testing, reestimated
periodically for changes in the underlying assumptions (to mark the
expected fiscal outcomes to market), will be critical for government
risk management in deciding, for example, on an appropriate level of
reserves and hedging strategy. In Chapter 11 in this volume, Brixi and
Gooptu, taking into account the correlation in the sensitivity of the
different items in the Fiscal Risk and Fiscal Hedge Matrixes to the
different types of risk, build a stress scenario for the government of
Indonesia.
For major contingent liabilities, how does one assess their size, prob-
ability of realization, and possible future fiscal effects? The literature
on valuing credit and project guarantees and state insurance programs
has built on contingent claims analysis (see Mody and Patro 1996,
Lewis and Mody 1997, Arthur Andersen 2000, and Marrison 2001
for an overview). Contingent claim analysis uses two basic concepts:
the expected costs (that is, the most likely or average cost) and the
unexpected costs (that is, the maximum likely cost with a particular
small probability, also referred to as value at risk or cost at risk) of the
contingent liabilities.
To reveal both the average and the maximum likely fiscal cost of
government contingent liabilities, one would generate a large number
of random scenarios (for example, using Monte Carlo simulations, as



30                                            BRIXI AND MODY
described in Arthur Andersen 2000 and Marrison 2001). In the distri-
bution of outcomes of the different scenarios, scenarios that occurred
most often would indicate the average fiscal cost and scenarios (also
referred to as stress scenarios) that occurred at a predetermined
probability (usually 5 percent or 10 percent) would indicate the cost
at risk. (In this volume, valuation of contingent liabilities using value-
at-risk and option-pricing approaches are discussed particularly in
Chapter 4 by Sundaresan, Chapter 13 by Pennacchi, and Chapter 19
by van Wijnbergen and Budina. For an overview of different approaches
as they have already been applied, see U.S. GAO 1993, 1998.)
As such, the average cost of contingent liabilities is a measure of
government subsidy implied by the issuance of a government guaran-
tee or other program of contingent government support. Thus from a
policy point of view, the average cost estimate can be used to judge
whether the government would be willing to support the project through
an equivalent up-front cash subsidy (in which case, as further discus-
sion will point out, the government also should be willing to budget
for the expected cost of contingent liabilities at the time of assuming
them). The stress scenario implies a fiscal cost that the government has
to consider in its risk management approach and has to be prepared to
pay in the future.
The institutional framework for dealing with fiscal risks mainly
relates to the rules and practice of information disclosure, monitoring,
fiscal planning, and budgeting. The institutional framework affects
the government's incentives and ability to constrain, control, and man-
age its fiscal risks. The framework must be such that it promotes a
risk-awareness culture in government and minimizes the scope for fis-
cal opportunism. Whether or not policymakers extend the focus of
public finance institutions to cover fiscal risks depends on several fac-
tors, including the definition and measurement of internationally rec-
ognized fiscal indicators (that is, the pressure of institutions such as
EUROSTAT, the International Monetary Fund, the World Bank, the
Organisation for Economic Co-operation and Development, and the
United Nations), public pressure (by the media, independent audit in-
stitutions, watchdog agencies, and legislators), investors' demands and
preferences (for example, to what extent sovereign credit rating agen-
cies pay attention to fiscal risks, and investors punish the government
for concealing relevant data and exposing the country to excessive
fiscal risk), and whether or not fiscal risks attract the attention of
reform-minded policymakers.
Fiscal risks are likely to attract more attention if the government is
required to disclose them. Disclosure can be in the form of a simple
statement of contingent liabilities and tax expenditures, or full-fledged
financial statements based on an accrual accounting system. The gov-
ernments of Australia, Canada, the Netherlands, New Zealand, and



DEALING WITH GOVERNMENT FISCAL RISK                              31
the United States offer good practices to follow. The fiscal statements
of these countries list the various sources of fiscal risk; discuss their
nature and sensitivities, implications for future fiscal position, and
allocative efficiency (compared with budgetary spending); and, where
applicable, provide their face value or estimated future fiscal cost or
both (see Chapter 2 by Petrie for examples and references). Govern-
ments also should enforce requirements that market agents disclose
information about their risk exposures. Particularly agencies that may
appear to be implicitly guaranteed by the government should be sub-
ject to strict disclosure requirements. In this context, for example,
subnational governments should report on their guarantees and on the
activities of their own financial enterprises; state-owned enterprises
should report on their environmental commitments; and financial in-
stitutions should report on their off-balance sheet items. Such broad
disclosure would allow market agents as well as the government to
conduct proper monitoring and analysis and to react to possible moral
hazard.
Although the organizational setup for fiscal risk management will
be specific to every country, some general principles apply (see Box
1.1). These include: centralize the risk-taking authority (possibly the
Box 1.1 Division of Responsibilities for Fiscal Risk
Management
Large banks, including J. P. Morgan and Deutsche Bank, have divided
the functions of designing and authorizing new transactions, analysis,
and record-keeping among three different offices. The front office serves
as the central point for the design of financial instruments, and it has
the exclusive authority to enter into new derivative and debt transac-
tions. Its objective is to ensure the required levels of available cash and
optimize the overall return-risk ratio. The middle office provides analysis
of future obligations and payoffs and their sensitivities for the entire
portfolio. Finally, the back office is responsible for record-keeping and
maintaining comprehensive databases.
Maintaining these functions independent of each other improves
transparency and the control of portfolio risks and prevents the front
office from exceeding its predetermined risk exposure limits. The gov-
ernments of Ireland and Sweden, for example, have successfully ap-
plied such a division of responsibilities to their debt management.
Many governments have centralized the authority to issue debt,
guarantees, tax exemptions, and other off-budget programs. Now they
should expand the scope of risk management and adjust organizational
structures and responsibilities accordingly. See Nars (1997).



32                                          BRIXI AND MODY
ministry of finance to oversee any risk-taking in the public sector);
separate risk-monitoring (for example, internally by the debt man-
agement office and externally by the supreme audit institution) from
risk-taking; and connect risk-taking with budgeting and debt man-
agement practice. Obviously, accountability structures are crucial to
ensure the best efforts and reduce the scope for fraud and corruption.
Policymakers and civil servants need to be accountable for the ad-
equacy of their risk analysis and the assumptions underlying deci-
sions that involve fiscal risks and for managing the overall government
risk exposure. Therefore, the role, of the independent audit (and the
supreme audit institution), as it is in the case of the U.S. General
Accounting Office, would extend beyond its conventional limits to
cover all aspects of government risk analysis and risk management
(see Chapter 2 by Petrie).
Budgeting and accounting rules influence the allocation of resources,
affect the timing and recognition of transactions, and may provide
opportunities and incentives to shift costs and risks from one period to
another and from one part of a government to another. Cash flow
budgeting, for example, makes guarantees and other contingent forms
of government support look more attractive than direct loans or cash
subsidies. It treats direct loans and subsidies as outlays, but does not
recognize contingent liabilities until default occurs, at which point the
government has little choice but to make good on past commitments.8
Direct loans and subsidies thus appear expensive and the contingent
form of support cheap. To make matters worse, in cash flow budget-
ing income earned from origination fees on guarantees is booked as
current revenue, making it appear that the government is profiting by
taking these risks.
In contrast, an accrual-based budgeting and accounting system re-
quires that the net present fiscal cost associated with individual gov-
ernment programs, including programs that generate a contingent
liability, be included in budget documents and thus be made visible
from the moment government decides to launch them. Moreover, for
contingent liabilities it encourages the government to set aside resources
up-front at the time of their launching (see Box 1.2 for a private sector
analog and Chapter 3 by Schick for a detailed discussion).9 To strengthen
accountability, the budget would be set in the context of a publicly
announced medium-term fiscal framework, which would later make
any departures from the original risk analysis apparent.
The experience of a number of countries (most of which is dis-
cussed in this book), including Canada, Colombia, Hungary, the Neth-
erlands, South Africa, Sweden, and the United States, has indicated
that a comprehensive shift of the accounting and budgeting systems to
accrual basis is not necessary in order for government to take control



'DEALING WITH GOVERNMENT FISCAL RISK                           33
Box 1.2 Budgeting for Risk in the Private Sector
Programs of contingent support are often akin to put options, which
create the obligation, but not the right, to buy an asset at certain pre-
defined strike levels. Charging the full option price when writing (sell-
ing) an option, as is typical in the private sector, amounts to immediately
fully provisioning the expected cost of contingent support. The price
of an option reflects the present value of the future possible loss, which
may be incurred by the underwriting institution. As illustrated by the
Black-Scholes formula, the price increases with the time to expiry (for
example, maturity of the guaranteed loan) and with the volatility of
the underlying asset (for example, share price of the enterprise whose
debt is under the guarantee). Financial institutions charge the full op-
tion price immediately at the time of selling the option. The amount is
then used either to build reserves or to buy a hedge. See Hull (1997).
of its fiscal risks.'0 Tables 1.3 and 1.4 summarize measures to enhance
policymakers' understanding of fiscal risks and their incentives as well
as capacity for dealing with them. These measures can be built upon
any decent public finance management system.
Reducing Government Risk Exposure
A number of commodity-exporting nations have tried to manage their
risk exposure by building reserves in booms and creating stabilization
funds, but the results have been mixed (Davis and others 2001 provide
an overview). With more success, several governments have used de-
rivatives and "exotic" debt instruments (such as debt instruments linking
the amount of debt repayment to commodity prices) to hedge risks in
the government assets and liabilities portfolio." Other governments
also have successfully purchased reinsurance for risks (such as weather
risks) from a large international reinsurer.'2 In recent years, increasing
market integration has made it possible to pool risk across countries
and thus has enabled financial markets to provide insurance against
risks, such as crop risks and disaster risks, that had been considered
uninsurable before. Financial markets have typically welcomed gov-
erinent risk management initiatives (in the early 1990s, for example,
the government of New Zealand witnessed a rapid improvement in the
terms of its sovereign borrowing once it announced and implemented



34                                                BRIXI AND MODY
Table 1.3 Systemic Measures to Reduce Government
Exposure to Fiscal Risk
Fiscal policy                              Fiscal management
First-                            First-
* Identify and reconsider govern-  * Assign responsibility for identifying,
ment programs and promises        recording, and reporting government
that imply significant risks, no  obligations, assessing the future
longer serve a significant social  likely fiscal cost, and monitoring
or economic purpose, or can be    government exposure to fiscal risk.
replaced by market instruments  * Analyze the sensitivity of the
such as private insurance,        government's fiscal position to the
derivatives.                      different types of risks and create
* Identify, classify, and analyze all  a list of early warning indicators
major sources of fiscal risk and  that signal possible future fiscal
financial safety (build an extended  pressures (for example, exchange
government balance sheet).        rate movements, rate of credit
* Announce that only those contin-  growth in the banking system,
gent liabilities included in a public  and demand growth vis-a-vis the
statement of fiscal risk will be  existing capacity in infrastructure).
honored.                        * Establish nominal, nonbudgetary
* As a context to policy decision-  control mechanisms for fiscal risks
making, introduce the concept     such as information disclosure on
of fiscal vulnerability.          government obligations and tax
expenditures, exposure limits (for
example, maximum amount of
guarantees outstanding), and ear-
marking of future funds to cover the
likely costs of contingent liabilities.
Later-                            Later-
* For policy decisions, analyze full  * Recognize and disclose information
fiscal performance and its vulner-  about government full fiscal perfor-
ability to risks and expand the   mance, including exposure to
medium-term fiscal framework      fiscal risk.
to cover the effects of fiscal risks. * Scrutinize fiscal risks in the budget
* Identify the government's risk    process before they are taken.
absorption capacity and (accord-  * Consider government risk exposure
ing to its risk preference and its  in structuring programs.
capacity to absorb and manage   * Build capacity for analyzing and
risk) determine the government's  managing risk (including mitigation
optimal risk exposure and reserve  of the risk at source, transferring
and hedging policy.               the risk to parties better able to
* Develop a government risk manage- bear the risk, and monitoring and
ment strategy that would guide    managing any residual risk that
policymakers and staff in day-to-  cannot be mitigated or transferred)
day decisions involving govern-   and for auditing government
ment risk exposure.               dealing with fiscal risk.
* Consider private sector coverage  * Build a mechanism to enforce the
to replace existing and proposed  disclosure, monitoring, and regula-
government programs.              tion of risks in both the public and
private sectors.
Source: Polackova (1998) and the authors.



DEALING WITH GOVERNMENT FISCAL RISK                              35
Table 1.4 Measures to Control the Fiscal Risks of Individual
Government Programs
Fiscal policy                              Fiscal management
Before accepting-                 Before accepting-
* Assess how the program fits with  * As part of fiscal planning and
policy objectives.                budgeting, evaluate the risks and
* Consider the program's risks and  estimate the likely future fiscal
likely future fiscal cost and com-  cost.
pare alternative forms of govern-  * Design the program well, provid-
ment support, including actions to  ing for risk mitigation and risk
enable private sector coverage.   transfer, to minimize government
* Outline and announce the limits  risk exposure and moral hazard in
of government responsibility with  the markets.
respect to the program in order to * Decide on the management of
minimize moral hazard.           residual risk (for example, set a
hedge and an additional reserve
requirement).
When accepted-                    When accepted-
* Stick to the preset limits of the  * Report the risk (for example, issue
scope of the program and of the   a statement of fiscal risks).
associated government           * Budget for the net present value of
responsibility.                   the expected fiscal cost of the
* Reconsider the program's          program.
relevance in the context of the  * Continuously monitor the perfor-
evolving needs, changing structure  mance under the program, including
of the economy and role of the    the program's risk factors, reserve
government, and advances in       adequacy, and behavioral effects.
technology and financial markets. * Update the risk analysis and risk
management strategy.
* Audit the validity of the risk analysis
and the quality of risk management.
* Draw consequences if a bias in the
original risk analysis and govern-
ment decision is revealed.
* Prepare contingency plans for
dealing with the program (whether
explicit or implicit) if fiscal risks
are realized.
Upon execution-                   Upon execution-
* Execute strictly within the     * Evaluate performance under the
preset limits of government       program, compare and report the
responsibility.                   actual fiscal cost versus the original
* If implicit, assess the fit with  cost estimates, and punish for
policy priorities and possible   failures (including a bias in risk
moral hazard effects before       analysis and deficiency in risk
executing.                        management).
Source: Polackova (1998) and the authors.



36                                            BRIXI AND MODY
its risk management strategy). Since the 1997 crisis in Asia, invest-
ment banks and sovereign credit rating agencies have increasingly dis-
cussed government risk exposure in their country risk analyses (see,
for example, Standard and Poor's 1997).
For governments and for enterprises, the objective of risk manage-
ment is to align the demand for funds with revenues (Froot, Scharfstein,
and Stein 1994). Private companies and financial institutions have
benefited from using enterprise-wide risk management strategies and
financial markets to manage and hedge risk. In addition to learning
from their experience, which is discussed in Chapter 4 by Sundaresan,
governments often also have the option of supporting broader reforms
that reduce risks or make them insurable in the markets.
For governments, the management of risk entails three complemen-
tary tasks: involving the private sector (mitigating the risk at source
and developing financial risk markets), transferring the risk to parties
better able to bear the risk (creating risk-sharing arrangements), and
managing any residual risk that cannot be mitigated or transferred
(monitoring, building reserves, and hedging).
Ultimately, risk mitigation with private sector involvement is the
most desirable long-run strategy, because it not only reduces the
government's exposure to fiscal risk but also reduces the vulnerability
of the economy to shocks. Instead of assuming risk, governments would
enable markets to deal with it. For example, a power sector that is
organized to permit competitive generation and distribution will fos-
ter efficient use of resources while at the same time lowering the risk
arising from excessive installation of capacity (for other examples, see
Table 1.5).
Similarly, by supporting the development of the markets for risk
instruments, the government can effectively withdraw from its direct
role in dealing with many risks. In this regard, policymakers need to
ask: Are the risks for which government coverage is sought truly unin-
surable in the private sector? How can these risks be made insurable?
For example, regulatory changes can encourage large international
insurers to access the local market and pool risks uninsurable in a
small economy and can make derivatives accessible to local market
agents. Private risk markets reduce the need for the traditional govern-
ment programs such as disaster risk insurance, crop insurance, and
minimum price policies. Similarly, new financial instruments may help
domestic financial institutions to better manage risk, thereby reducing
their demand for government guarantees. However, risk mitigation
strategies and markets for risk instruments may require fundamental
sectoral reforms and thus cannot be developed overnight. (For a dis-
cussion linking fiscal risk with sectoral reform, see, for example, Irwin
and others 1998). Therefore, more effective risk-sharing may be the
practical short-term strategy.



DEALING WITH GOVERNMENT FISCAL RISK                                 37
Table 1.5 Risk, Possible Fiscal Cost, and Private Sector
Solution
Type of risk     Coverage     Possible fiscal cost  Private sector solution
Credit        Debt service and  Principal plus    Credit enhancement
guarantees      losses due to     interest plus   Risk-sharing with
default           possible penalties  creditors
by the creditor
for default
Guarantee     Minimum ab-       Guaranteed        Sound regulatory frame-
on minimum      solute amount     amount times       work for pension
return of       (monetary value)  the number of     funds and the overall
pension funds                     pensioners         financial markets
Minimum relative  Average wage
amount (share     share times the
of average wage)  number of
pensioners
Project       Design and        Very large if not  Correcting for any
guarantees      development       capped             market failures that
Construction risk                     reduce access of
Operating risk                        investors to adequate
(cost overrun,                      risk protection
delays)                              mechanisms in the
Demand/revenue                         markets
risk                              Risk-sharing with
Financial risk                        investors and
(exchange rate,                     creditors
interest rate)
Force majeure
Environmental risk
Political and policy
risk
Disaster      Losses due to     Very large if      Environment allow-
Insurance       disasters         notcapped          ing direct access to
international in-
surers and reinsurers,
catastrophe bonds
and derivatives
Deposit       Bank deposits     Face value of      Regulations allowing
Insurance                         all deposits if    bank access to risk
not capped        protection and risk
markets
Disclosure of informa-
tion on bank perfor-
mance and manage-
ment, international
competition, low
limits on the deposit
amounts guaranteed
(Table continues on the following page.)



38                                                 BRIXI AND MODY
Table 1.5 (continued)
Type of risk    Coverage      Possible fiscal cost  Private sector solution
Price        Minimum price      Guaranteed min-   Encourage direct
support        of a product/      imum price        access to interna-
commodity          minus actual      tional derivatives
price, multiplied  markets
by quantity
Implicit     Explicitlv: none   Almost unlimited  Sound regulatory frame-
guarantee    Implicitly: obliga-                    work for accounting,
on various     tions of subnation-                 information dis-
obligations    al governments,                      closure, and audit
state-owned finan-                 Credible announce-
cial institutions,                   ments and actions by
enterprises and                      the government to
funds, large mar-                    minimize expecta-
ket agents, etc.                    tions of a bailout
Source: The authors.
Risk transfer and a good risk-sharing mechanism require clear policy
objectives and understanding of all the underlying risks in a program.
In the private sector in the last 20 years, the possibilities for transfer-
ring risk have been growing rapidly. Derivative products have allowed
firms to subdivide, isolate, and swap various risks, but they also have
created new risk exposures that are not easy to quantify (Garber 1998).
So far, for governments the primary method of transferring risk has
been through risk-sharing provisions in their guarantees and insurance
contracts. Recent practice and literature have suggested that carving
out commercial risk from government coverage significantly reduces
the negative behavioral effects of such government programs in the
markets as well as limits government risk exposure (U.S. GAO 1998;
World Bank 1999c, 2000). For example, in the banking sector, should
the government provide deposit insurance, then a relatively low cap
on government protection would increase the incentives of bankers to
improve due diligence and project selection, lowering risk and the
wasteful use of resources. Similarly, price support programs require
their own risk-sharing arrangement with, for example, farmers (see
Box 1.3).
For implicit contingent liabilities, risk-sharing tends to be applied
ex post. As Honohan (1999) argues, however, fiscal cost is lower (and
government crisis management more efficient) if government has an
ex ante, confidential contingency plan (for example, deciding ex ante
which stakeholders-domestic depositors in local currency, domestic
depositors in foreign currency, foreign depositors, creditors, and share-
holders-to assist and how much before a crisis occurs). Similarly, the



r)EALING WITH GOVERNMENT FISCAL RISK                            39
Box 1.3 Government Price Support Programs for Farmers
International markets are not always able to offer adequate instruments,
such as futures, options, or insurance policies, to protect farmers against
the volatility of their particular product. In reality, it may be difficult for a
farmer to sell (short hedge) futures on his or her product in order to make
the final selling price certain and thus protect against losses from a pos-
sible future reduction in its price. Thus the government may still be the
only source of protection. (Even when markets do not offer instruments
of adequate protection to individual farmers, they may offer hedging or
reinsurance instruments on a larger scale and customized basis to govern-
ment-for example, customized derivatives contracts over-the-counter or
a reinsurance policy). In many countries, government protection comes in
the form of a price support program, which guarantees farmers a mini-
mum price for their output. Such programs generate for the government
an obligation to pay farmers the difference between the market price and
the guaranteed minimum price should the price of their product drop
below the minimum price.
If price support turns out to be the preferred choice of government
support, how should the program be designed to minimize both moral
hazard on the side of the farmers and the future fiscal cost on the side of
the government? To deter overproduction, the amount of product guar-
anteed must be limited-by imposing a nominal ceiling and quotas and
by charging the farmers a fee per unit of guaranteed product. (For a par-
ticular minimum price, the Black Scholes options-pricing formula allows
one to determine the fee. As Cox and Rubinstein 1985 explain, the only
required variables for the calculation will be the given minimum price,
the actual price, the volatility of actual prices over past years, and the
time to expiration-that is, the number of days ahead for the minimum
price to apply. Alternatively, the government may sell the limited amount
of per-unit price guarantees in an auction). If the objective is to provide
subsistence to temporarily impoverished farmers and encourage them to
sustain a limited amount of land or number of bushels or livestock (rather
than flood the market with excess production), the program may be effi-
ciently designed as a put spread, setting not only the minimum price but
also the maximum amount of support paid to farmers per unit of product.
For example, if the minimum price is set at 100 and the maximum amount
of government support at 20, government pays 15 if the actual price is 85,
but no more than 20 even if the actual price drops well below 80.
Changing tastes that may cut demand for a particular commodity (such
as lamb and beef in the early 21st century) and cause it to drop in price
erode the rationale of the commodity's strategic importance. Thus the
risk of a continual drop in demand should not belong to the government.
Also, risks that the quality of (and thus the price that can be charged for)
a domestic product drops compared with the quality and price of that
sold by international competitors, or that new technology and fertilizers
drive prices down permanently, should not belong to the government.
These are reasonably well under the farmers' control. Thus a price sup-
port program would become effective only if the reasons for a low price
are temporary and clearly and entirely out of the farmers' control.



40                                            BRIXI AND MODY
central government can impose limits, either confidentially or pub-
licly, for its responsibilities in the event a subnational government goes
bankrupt (for example, to ensure the water supply but not the mayor's
salary).
As for dealing with any residual risk that cannot be mitigated or
transferred, this also is best done within the expanded assets and li-
abilities framework (for literature on the portfolio approach to risk
management, see Cassard and Folkerts-Landau 1997; World Bank 2000;
World Bank and IMF 2000). Across the two matrixes presented in
Tables 1.1 and 1.2, analysis and stress testing of the impact of various
types of risks help to account for the correlation in the value of gov-
ernment spending pressures and proceeds under different scenarios,
and to identify the government's residual, unhedged risk exposure (the
types of risk that have the strongest overall fiscal effect).
The government then has two basic approaches to dealing with its
risk exposure and to protecting itself against rare events: building up
reserves and using financial hedges. Once a government determines
the level of loss it is capable of and willing to absorb, it can establish
reserves against unexpected losses. Several concerns, though, are asso-
ciated with provisioning. First, setting reserves on the basis of portfo-
lio risk analysis (as opposed to the additive loss exposure of each
government program) is advantageous but it has yet to be fully tested
by governments (Lewis and Mody 1997). Second, reserves have an
opportunity cost-and one that is particularly high in bad times and in
poor countries. The challenge is to acknowledge the opportunity cost
of not having reserves-when government is stuck, unprepared for a
sudden increase in its obligations or a drop in its revenues. For govern-
ments that find themselves unable to obtain favorable credit in such
situations, fiscal crises become a reality with all their negative conse-
quences. Furthermore, how should reserves be invested, and who will
be responsible for investment decisions? How does government ensure
reserve adequacy? And how does government prevent the misuse of
reserves? Arguably, politicians will always find ways to tap reserves
even for purposes other than those originally intended. The experience
in many countries, particularly with stabilization funds, indicates that
neither laws nor rules prevent misuse.
Analysts have been searching for possible approaches to improving
reserve adequacy and reducing possible misuse. In Chapter 6 of this
volume, Daniel Cohen proposes that countries with developed capital
markets create a transparent reserve fund and sell its shares to private
owners. Market mechanisms would serve to discover the share price
of the reserve fund, primarily reflecting on the adequacy of its capitali-
zation and of its use. For countries with underdeveloped capital mar-
kets, a reputable foreign institution may be entrusted with the task of



DEALING WITH GOVERNMENT FISCAL RISK                            41
Box 1.4 Margin Calls to Collateralize Risk
Learning from the practice of margin calls applied by investment banks,
government may be able to reduce moral hazard in the markets and its
own risk exposure by requiring beneficiaries of its programs to make
collateral payments when their performance deteriorates. The collat-
eral "penalty" would be calculated as the increase in the mark-to-market
fiscal cost. This practice would encourage the beneficiaries of govern-
ment programs to limit their own risk exposure and generate resources
for the government contingency reserve fund when the government's
risk exposure increases. But it would demand tight monitoring of per-
formance in the real and financial sectors.
Investment banks periodically monitor their credit risk exposure on
clients' portfolios against predetermined, uncollateralized limits and
require clients to make collateral "penalty" payments for the excess
mark-to-market value of a potential loss over the limit. The limit is
defined ex ante, as part of the contractual agreement between the bank
and the client, often in terms of both the most likely loss and value at
risk facing the bank with respect to a specific sector, region, or market
segment. For a specific portfolio, when assumptions underlying its risk
analysis deteriorate, the bank requires the client (a "margin" call) to
immediately make a collateral payment equal to the excess of the mark-
to-market potential loss over the limit.
managing the reserves under predetermined parameters of risk and
returns. A contract would specify permissible claims on reserves and
make other claims subject to a penalty and ex ante public disclosure.
Both approaches could be complemented with margin calls to collaterize
risk (see Box 1.4).
Hedging does not fully substitute for reserves, because all contin-
gencies cannot be foreseen and market hedges are not available for
contingencies that can be visualized.'3 However, where hedging is pos-
sible, it may be superior to building reserves for governments with
good capacities and control mechanisms. The fiscal costs of different
government programs may be negatively correlated,'4 but creating
additional government programs with the sole objective of risk-pool-
ing would be a questionable practice for government. Therefore, hedg-
ing and purchase of reinsurance may be needed to complement pooling.
Take, for example, a government that largely depends on tax rev-
enues from copper sales. When the price of copper goes down, the
government is short on revenues. If its access to borrowing is limited,
it has to cut public expenditures abruptly. Instead of building a revenue



42                                           BRIXI AND MODY
stabilization fund, the government can look for possibilities to stabi-
lize its fiscal performance by structuring its obligations to reflect cop-
per prices. For example, the government may attempt to link its
liabilities to the source of volatility-that is, to issue bonds that offer a
yield inversely linked to copper price. Structuring liabilities according
to the sources of volatility in the government portfolio of contingent
liabilities and revenues reduces the overall volatility in the future fiscal
outlook and thus offers an alternative to stabilization reserve funds.
When revenues are low, debt service will become less expensive and
vice versa. Private insurance companies have similar experience in is-
suing catastrophe bonds, which offer lower yields when a hurricane
occurs than when it does not.i" The government also may purchase
customized derivatives that will deliver a positive payoff, inversely
related to the copper price, when the price of copper drops below a
specific threshold.
Similarly, governments have found ways to hedge their risks through
derivatives applied to their obligations portfolio. Many governments-
most notably, Belgium, Colombia, Hungary, Ireland, New Zealand,
and Sweden-have experimented with asset and liability management
approaches in order to match currency, interest rate, and maturity risks
in the portfolios of their liabilities and assets. They have utilized inter-
est rate swaps, currency swaps, currency forwards, and other deriva-
tives to achieve the desired risk profile in their debt portfolio.
Possibilities to hedge risks of contingent liabilities have been ex-
plored mainly with respect to minimum price support policies and
crop insurance. To hedge their risk exposure, several governments (for
example, of Colombia and Mexico) have been purchasing futures and
customized forward contracts. (For an example of the use of options
to hedge price support policy, see Box 1.5). Ideally, the price of the
hedge would be passed on to the program beneficiaries (for example,
as a fee to be charged the farmers discussed in Box 1.3). Some finan-
cial instruments, such as catastrophe bonds, which link their yield in-
versely to the occurrence of a particular catastrophe, would help to
hedge risks of associated contingent liabilities and revenue decline.
For more examples, using a combination of the above approaches, see
Table 1.6.
The borderline between hedging and speculation is, however, some-
times difficult to draw. Recent experiences of companies and hedge
funds have reconfirmed that derivatives that provide less than a per-
fect hedge may generate risks on their own. Also, the use of derivatives
is dangerous if there is not a clear strategy on which specific risks to
hedge and to what extent to hedge them. For government as well as
for private companies, the ad hoc availability of new hedging instru-
ments, and the attractions of financial engineering, should never drive
risk management decisions. Potentially risky hedging techniques in-



DEALING WITH GOVERNMENT FISCAL RISK                            43
Box 1.5 Securitizing Government Risk
How does a government hedge the risk of price support policies? Sup-
pose the government offers a minimum price guarantee on a commod-
ity. Assuming the floor price is set at $10 per unit, the government pays
the difference between the floor and actual price if the actual price falls
below $10 per unit (see graph).
This payoff exactly illustrates that providing a price support policy
is equal to shorting a put option (selling the right to sell a commodity
at a specified minimum price). To hedge against possible losses, the
replicating strategy suggests buying puts from international financial
intermediaries. The cost to the government is the difference between
the total fees collected by the government from the commodity pro-
ducers (if possible in an auction) and the price paid by the government
for the put. The strategy allows the government to convert its fiscal
cost from the form of an unknown contingent liability to a fixed, up-
front payment.
clude a dynamic hedging strategy, which requires readjusting the hedg-
ing mechanism as its underlying assumptions change (Hull 1997).
Therefore, before a treasurer or debt managers launch sophisticated
risk management techniques, top policymakers need to decide to what
extent the government should be exposed to risk-that is, to what
extent should the government require that its expected fiscal pressures
be matched with its actual resources? Addressing this question, in turn,
will depend on the extent to which government can rely on ad hoc
borrowing and tax increases and on the extent to which the govern-
rnent can afford to restructure or default on its budget programs (are
arrears permissible?) and on its contingent and direct liabilities. And
what is the government's risk management capability?



44                                                BRIXI AND MODY
Table 1.6 Reducing Government Risk Exposure
Source of risk  Reduce risk in design   Reduce exposure for risks taken
Guarantees Cover only selected       Adjust risk exposures in direct
risks such as political/  liability and assets portfolio (for
policy risks.            example, reduce exposure to the
pertinent currency if exchange rate
risk is covered by the guarantee).
Establish a reserve fund for all
guarantees.
Limit total benefits paid to the
amount available in the reserve
fund.
Ensure reserve adequacy by trans-
forming the reserve fund into a
public company with shares
freely traded.,
Disaster    Cap maximum benefit.     Issue catastrophe bonds (possibly
insurance   Insure middle rather than  for a basket of likely disasters).
first portion of loss.  Purchase reinsurance for risks in
excess of a threshold that is
deemed fiscally bearable.
Deposit     Require information dis-  Establish separate reserve fund.
insurance     closure, sound regula-  Limit total benefits paid to the
tion, supervision, and   amount available in the reserve
enforcement before in-   fund.
troducing deposit      Ensure reserve adequacy by
insurance.               transforming the reserve fund
Cap maximum benefit.       into a public company with
Insure middle rather than  shares freely traded.'
first portion of loss.
Price       Auction policies.        Purchase payoff-replicating
support     Cover only selected risks  derivatives.
such as political/policy  Purchase reinsurance.
risks.
Implicit    Make announcements       Seek contingent credit line from
guarantees    and act to minimize      IMF.
to banks and  moral hazard.
enterprises
Privatiza-  Find a strategic investor  Use proceeds to reduce govern-
tion and      (future revenue).        ment liabilities or future
asset sale   Maximize privatization    obligations.
revenue.
(Table continues on the following page.)



DEALING WITH GOVERNMENT FISCAL RISK                                 45
Table 1.6 (continued)
Scurce of risk  Reduce risk in design    Reduce exposure for risks taken
Purchase     Pay only estimated market Sell bad assets quickly.
o0: bad        price (a part of recap-
assets         italization objectives).
Allow for market mech-
anisms to deal with bad
assets (strengthening the
position of creditor, bank-
ruptcy processes, and so
forth).
Make announcements
and act to minimize
expectations of future
possible repeated
purchases.
Commod-      Demand base payment       Issue commodity-linked
ity tax        independent of com-       bonds.
modity price.            Purchase commodity-linked
derivatives.
Purchase insurance.
Repay-       Require collateral.       Purchase default insurance.
inent of
direct
lending
a. The interests of the fund shareholders would contribute to ensuring reserve adequacy-
that is, to charging guarantee beneficiaries, such as banks, adequate premiums. This ar-
rangement loosely imitates the arrangement suggested in Chapter 6 by Daniel Cohen in
this volume.
Source: The authors.
The Fiscal Risk Questionnaire: An Example
For many governments, the objective to understand and manage their
exposure to fiscal risk is likely to require major efforts throughout an
extended period of time. The fiscal risk questionnaire that follows
provides a set of questions that analysts and policymakers may find
useful in dealing with government fiscal risk.
0 Fm 0m



46                                            BRIXI AND MODY
Fiscal Risk Questionnaire
A. Macroeconomic Context
1. What are the macroeconomic constraints on the government's
future fiscal performance?
Identify constraints such as a currency board or other inflexible
exchange rate arrangement, high public debt levels, low sover-
eign credit rating, or shallow domestic debt markets that risk the
government's future ability to issue debt.
2. What trends are affecting the government's general fiscal position?
Analyze medium-term and long-term trends to provide a con-
solidated picture inclusive of all levels of government, off-budget
funds, public assets, and liabilities.
3. How sensitive are these fiscal trends to the underlying macroeco-
nomic, demographic, and policy assumptions?
Build stress scenarios for the medium-term and long-term fiscal
outlooks with respect to the underlying assumptions.
B. Sources of Fiscal Risk
1. What are the major sources of fiscal risk? What are the largest
and riskiest contingent government liabilities in the country? Are
they explicit or implicit? Fill in the Fiscal Risk Matrix (Table
1.1) with specific items.
2. What are the main types of risks determining the size and ur-
gency of the items in the Fiscal Risk Matrix?
Consider currency risk, interest rate risk, commodity price risk,
refinancing risk, operational risk, political risk, policy risk, and
similar risks.
3. Overall, are the items identified in the Fiscal Risk Matrix likely
to raise significantly the future financing requirement of the gov-
ernment? If so, under which circumstances mainly?
Consider the overall size of contingent liabilities (face values)
and a stress scenario if these obligations are realized. Compare
the overall face value of contingent liabilities with the reported
government debt and with the government's future borrowing
capacity.
C. Analysis of Selected Risks
1. How clearly defined are the public sector and the spheres of gov-
ernment responsibility? Is there a precise legal delineation of the



DE_ALING WITH GOVERNMENT FISCAL RISK                        47
public sector (for example, in the form of a full list of public
sector agencies) and of government responsibilities? If not le-
gally defined, is it clearly understood which services are guaran-
teed by the government?
2. State-guaranteed institutions and directed credit
Identify institutions that fulfill orders of the government to ex-
tend financing to enterprises, banks, agencies of any kind, or
households. Review their balance sheets and statements of con-
tingent liabilities.
What type of government support do these institutions receive
(for example, privatization revenues, cheap financing via the cen-
tral bank, state guarantee for borrowings)?
3. Guarantees
Review government guarantees, their issuer (the ministry of fi-
nance or other government agency), beneficiaries, creditors, face
values, the type of risks and their shares covered, currency of
denomination, and risk estimates if any.
4. State-owned enterprises and banks
Review the largest state-owned enterprises and their balance sheets
and risk statements and the largest state-owned banks and their
balance sheets and risk statements.
D. Sources of Government Financial Safety
1. What are the major sources of government financial safety? What
are the largest tradable government assets and other sources of
future revenues? Fill in the Fiscal Hedge Matrix (Table 1.2) with
specific items.
2. How sensitive are these financial sources to the respective types
of risks? What are the likely scenarios for the future government
revenue stream?
3. Taking into account the correlation in the sensitivity of govern-
ment finances to the different types of risks, compare the sce-
narios of the likely future government revenue stream with the
likely financing pressures to emerge from the items identified in
the Fiscal Risk Matrix. What is the worst-case scenario?
E. Recording and Reporting: Transparency
1. For each item in the matrixes, which agencies are responsible for
final approval, recording, monitoring, and data consolidation?



48                                            BRIXI AND MODY
2. Which agencies and authorities are informed ex ante about con-
tingent liabilities and, overall, about the fiscal risks associated
with programs under government consideration?
* The issuing agency only?
* The related sector ministry?
. The finance ministry?
* The cabinet?
. The parliament?
* Others?
3. Which agencies can instantaneously retrieve up-to-date informa-
tion about the items listed in the matrixes? Which documents
report such information? What is the time lag in reporting?
4. Which sources of fiscal risks are not reported to the:
. Ministry of finance
* Cabinet
* Central bank
. Parliament
* Foreign investors
* Public?
F. Institutional Arrangements: Accountability
1. Do any legal requirements apply to the government with respect
to estimating, accounting, and reporting the future fiscal costs
associated with its direct and contingent liabilities?
No
Yes, in the budget process
* when the government is called on to pay
* when cash is transferred
* other.
2. Which of the obligations identified in the Fiscal Risk Matrix are
not regulated by any law and depend fully on ad hoc government
decisions?
3. What is the legal and regulatory framework for:
* State guarantees: the requirements for their design (the type of
risks that can be covered, the extent of the required risk-
sharing), issuance (is only the ministry of finance authorized?),
government control mechanism (required reports from the
creditor and beneficiary, audit and valuation requirements),
and the realization mechanism if they fall due.



DEALING WITH GOVERNMENT FISCAL RISK                       49
* Subnational governments, public sector agencies and enter-
prises, and state-guaranteed institutions: the financial man-
agement and reporting requirements and government control
mechanism.
* Demands on the government to extend ad hoc previously un-
foreseen financial support: the legal requirements and prac-
tice for deliberation in government decisionmaking.
4. Is the government legally required to explain increases in public
liabilities (particularly increases above the levels explained by
budget deficit figures)?
No
Yes,
* to the parliament
* to the public
* other.
5. What is the authority and capacity of the supreme audit institu-
tion with respect to the government's risk exposure and risk
management?
Is the supreme audit institution authorized to review and ca-
pable of reviewing the quality and assumptions of the
government's fiscal risk analysis, fiscal risk management strat-
egy, and fiscal risk management practice?
G. Policy: Practice
1. When the government considers new promises of contingent gov-
ernment support, how much attention does it pay to risk analysis,
to the issues of moral hazard in the markets, and similar things? In
particular, do the ministry of finance, cabinet, central bank, or
parliament quantify the future fiscal cost of alternative options in
a single medium-term fiscal framework? Describe the risks of al-
ternative options (direct versus contingent support program, pos-
sibilities for risk-sharing in guarantee contracts, and others).
2. In which areas and under what circumstances do the public or
interest groups expect the government to provide financial sup-
port beyond the budget?
3. Identify examples of times at which the government withstood
political pressure and did not provide financial support beyond
the budgeted figures (for example, when the government refused
to solicit financial support for a failed enterprise or bank).



50                                          BRIXI AND MODY
4. Are public enterprises and banks, state-guaranteed institutions,
and creditors and beneficiaries under state guarantees "rewarded"
and "punished" for the quality of their management of risk?
Provide examples.
H. Fiscal Risk Management: Capacities
1. Describe the capacities of the ministry of finance, other govern-
ment agencies, public sector institutions and enterprises, and state-
guaranteed institutions to analyze, monitor, and control the risks
of government programs and contingent liabilities.
What methodologies have been used to analyze the size and risks
of specific contingent liabilities? What have been the challenges
and results in applying these nmethodologies? What has been the
process of monitoring and controlling risk?
2. Describe the process of designing a state guarantee, a state insur-
ance program, or any other program that involves fiscal risk.
Particularly focus on the risk analysis and treatment of moral haz-
ard in the design process. For example, to what extent are risks
shared between the government and the program beneficiaries?
3. What measures have been implemented by the parliament, cabi-
net, ministry of finance, sector ministries, and other public agen-
cies to prevent and reduce fiscal risks arising from the public and
private sectors?
For example, are there any limits on enterprise debt, subnational
government obligations, or central bank obligations? Are any
actions taken if they appear too high? Does the government have
an explicit risk management strategy with respect to its overall
fiscal risk exposure?
4. Does the government build contingency reserves, purchase rein-
surance, or hedge to mitigate its fiscal risk exposure?
Who determines the required size of the contingency reserves
and how? What is the hedging practice?
Conclusion
Dealing with contingent liabilities and other fiscal risks has recently
surfaced as an increasingly important issue in public finance analysis
and public finance management. Learning from country experience
and from new research (both discussed in the chapters of this book),
this chapter has outlined a framework to guide analysts and
policymakers in their attempts to understand and improve the man-



DEALING WITH GOVERNMENT FISCAL RISK                            51
agement of fiscal risk. Further research and long-term efforts by gov-
ernments will be needed, however, for public finance analysis and public
finance institutions to truly come to terms with government fiscal risk,
and for policymakers and civil servants to acquire the incentives and
capacity to optimize government exposure to risks.
Notes
1. The literature discussing the ultimate fiscal and economic costs of hid-
(fen government obligations is large. Kharas and Mishra (2001) illustrate the
fiscal cost of contingent liabilities across over 30 countries. Works by Brock
(1992), Kryzanowski and Roberts (1993), Caprio and Klingebiel (1997), and
World Bank (2001a) discuss government bailouts in the banking sector. Free-
man and Mendelowitz (1982) illustrate a bailout in the private corporate
sector. U.S. GAO (1998) draws lessons from the past fiscal cost of credit
guarantees and government insurance programs. Townsend (1977), Salant
(1983), and Bardsley (1994) analyze adverse behavioral effects and the ex
post fiscal cost of price-fixing schemes. Dillinger (1999) and Fornasari, Webb,
and Zou (2000) analyze bailouts of subnational governments by the central
government.
2. The conventional approach to the analysis of fiscal sustainability is
limited in two ways: first, it looks only at the liability side of the public
sector balance sheet, and, second, it considers only direct liabilities, ignoring
contingent liabilities, both explicit and implicit. Under the conventional ap-
proach, the actual deficit is compared with the estimated sustainable deficit
level that will keep the debt-to-GDP (gross domestic product ) ratio constant
for feasible rates of growth, real interest, and inflation. This approach as-
sumes that keeping a constant ratio of public debt to GDP will ensure public
sector solvency and avoid debt crises in the future. Another, less-stringent
requirement is to test for the no-Ponzi-scheme condition for public debt,
followed up by the neoclassical solvency approach. This methodology checks
for public solvency by comparing the annual rate of growth of the govern-
ment debt-to-GDP ratio with the real interest rate. If the debt ratio system-
atically grows faster than the real interest rate, the public sector is considered
insolvent. Conventional fiscal analysis also tends to neglect the sensitivity of
the fiscal position to risks, such as macroeconomic volatility, called contin-
gent liabilities, and unclear expenditure commitments, and the ability of the
government to cope with shocks.
For a critique of standard approaches to fiscal sustainability analysis and
deficit measurement, see Eisner and Pieper (1984), Buiter (1985), Bean and
Buiter (1987), Fischer and Easterly (1990), Blejer and Cheasty (1991), East-
erly, Rodriguez, and Schmidt-Hebbel (1994), Selowsky (1998), Easterly (1999),
and Kharas and Mishra (2001). These studies reflect the fact that, depending
on the definition and measurement methodology, deficit measures may mean



52                                               BRIXI AND MODY
very different things. They explore the impact of the coverage of a govern-
ment budget, the definition of government, and the accounting and budget-
ing frameworks, as well as the impact of inflation, seasonality, structural
changes in the economy, and the business cycle on the results of fiscal analy-
sis. Doubts have been expressed about the feasibility of incorporating hidden
government obligations in fiscal analysis. For example, Eisner (1984) notes
that the valuation of contingent liabilities of government in the calculation
of deficits is subject to the criticism that the government can legislate actions
that may seriously change their future value.
For a critique of how public finance management frameworks treat con-
tingent liabilities and fiscal risk, see, for example, World Bank (1997), Schick
(1998, 2001), and Schiavo-Campo and Tommasi (1999).
3. This section draws on Polackova (1998).
4. In Chapter 7 of this book, Richard Hemming and Murray Petrie try to
differentiate quasi-fiscal activities from contingent liabilities, using a nar-
rower definition for contingent liabilities.
5. Dillinger (1999) provides examples of the subnational government risk
of Argentina and Brazil. Hutchinson (2000) and World Bank (2001b) illus-
trate the fiscal risks of state-owned enterprises. In the United States, an inter-
esting body of literature has been developed around the so-called government-
sponsored enterprises (GSEs), which provide guarantees in housing and
agriculture markets. The best-known GSEs are Fannie Mae and Freddie Mac.
Creditors have perceived GSEs as beneficiaries of an implicit guarantee of
the U.S. government and thus have been willing to offer cheap financing. For
an overview, see Stanton and Moe (2001) and Van Order (2000).
6. The Medium-Term Expenditure Framework (MTEF), as first applied in
Australia, tackles fiscal opportunism with respect to government expendi-
tures and revenues by requiring policymakers to analyze and disclose the
assumptions and medium-term (three to five years) implications of their budget
proposals and to be accountable for any departures from their targeted levels
in year-by-year decisionmaking. In South Africa, MTEF extends further to
cover consequences of off-budget items such as government guarantees. For
a discussion, see World Bank (1997), Schiavo-Campo and Tommasi (1999),
Kruger (1999), and Schick (2001).
7. Stress scenarios illustrate the sensitivity of expected fiscal outcomes to
normal and abnormal changes in the underlying assumptions. For a specific
program, stress testing will show how normal and extreme changes in the
underlying factors (such as commodity price) over the next 3, 6, and 12
months will affect its fiscal cost. More broadly, applied to the government's
fiscal position stress testing helps detect key fiscal vulnerabilities of govern-
ment. A convenient way to generate less likely scenarios and stress scenarios
is the value-at-risk (cost-at-risk) approach. For an introduction to the value-
at-risk literature, see, for example, Best (1998) and Butler (1999); for an
application to contingent liabilities, see Blejer and Schumacher (1998).



DEALING WITH GOVERNMENT FISCAL RISK                             53
A baseline scenario would reflect the expected, most likely outcomes and
utilize actuarial, econometric, or contingent claim analysis methods (see Lewis
and Mody 1997 and Mody and Patro 1996 for a summary and Hull 1997 for
a textbook introduction).
8. Many governments pay the cost of defaulted contingent liabilities di-
rectly from borrowing proceeds. In such cases, contingent liabilities never
affect the budget and budget deficit. After they fall due, they are only re-
flected in an increase in government debt.
9. The use of present value budgeting may or may not affect cash-based
estimates of the government's fiscal deficit. It depends on whether the effect
on the deficit is recorded when money is transferred from the budget to a
contingency fund (then no effect is recorded when a guarantee is called and
paid for from the contingency fund) or only when actual cash payments are
disbursed from the program account. For more discussion on accounting and
budgeting for risk, see also Lewis and Mody (1997), Brixi, Ghanem, and
Islam (1999), Brixi, Papp, and Schick (1999), and World Bank (1999b).
10. A full accrual-based accounting system, though desirable, is neither
necessary nor sufficient to ensure that governments adequately report and
consider contingent fiscal risks in policy. Although an accrual accounting
system encourages governments to prepare a statement of contingent liabili-
ties and financial risks, it generally does not require that contingent liabili-
ties be included in the balance sheet and that the associated risks be evaluated
and quantified. International accounting standards, for example, require only
probable contingent liabilities (contingencies with a relatively high probabil-
ity of realization) to be included in the balance sheet, leaving the others in a
separate statement of contingent liabilities without requiring analysis of the
underlying risks. In the case of budgeting, budgeting for the present value of
the future fiscal cost of contingent liabilities may be combined with cash
budgeting for budgetary spending (cash-based) programs. Countries success-
fully combining reporting on contingent liabilities with cash accounting in-
clude the Czech Republic, Hungary, and South Africa, and those budgeting
for contingent liabilities within a cash-based budgeting system include Canada,
the Netherlands, and the United States.
11. Derivatives can be designed in many different ways to fit the risk
management objectives. Basic derivatives include futures and forwards, swaps,
and options. A future contract is an arrangement between two parties to buy
or sell an asset at a predetermined future time and price. These contracts are
normally traded on an exchange. A similar arrangement, a forward contract,
can be made with a financial intermediary over-the-counter, without involv-
ing an exchange. A swap is an arrangement in which two parties exchange
the stream of payment of two different assets. For background on futures,
forwards, and swaps, see Hull (1997). An option contract gives the holder the
right but not the obligation to buy (call option) or to sell (put option) an asset
at a predetermined future time and price. This predetermined price is known



54                                              BRIXI AND MODY
as the strike price and the predetermined date is known as the expiration or
maturity date. At the time of purchase, the buyer of an option contract pays an
option price to the option writer. Option-pricing analysis, most often employ-
ing the Black-Scholes formula, serves to set the option price. For background
on options, see Cox and Rubinstein (1985). On the actual and possible use of
derivatives as hedging instruments by governments, see, for example, Claessens
(1993), Nars (1997), World Bank (1999a, 2000), and Patterson (2001).
12. Mody (2000), for example, describes the experience with rainfall in-
surance purchased by the government of Nicaragua from a private reinsurer.
13. Hedging refers to the strategy of protecting oneself against losses aris-
ing from changes in market conditions. It involves relatively safe perfectly
offsetting transactions (that would "perfectly" offset gaines/losses arising
from changes in market conditions) and relatively risky dynamic hedging
strategies (that require continual rebalancing as market conditions evolve).
For an introduction, see Hull (1997).
14. Pooling less than perfectly correlated risks (for example, hurricane,
drought, and fire) would allow the government to reduce the volatility in the
total long-term cost of its insurance programs.
15. Catastrophe bonds are obligations whose interest and principal pay-
ments are linked to a catastrophic event. For example, they could call for a
reduction in the interest or principal, or for extension of maturity if losses
related to the underlying event exceed the trigger level. This arrangement is
called reinsurance protection (see Insurance Services Office 1999).
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
Arthur Andersen. 2000. General and Specific Methodologies for Valuing
Contingent Liabilities. Washington, D.C.
Bardsley, Peter. 1994. "The Collapse of the Australian Wool Reserve Price
Schemes." Economic Journal 104 (426): 1087-1105.
Bean, Charles, and William Buiter. 1987. "The Plain Man's Guide to Fiscal
and Financial Policy." Employment Institute, London, October.
Best, Philip. 1998. Implementing Value at Risk. Chichester; New York: John
Wiley.
Blejer, Mario, and Adrienne Cheasty. 1991. "The Measurement of Fiscal
Deficits: Analytical and Methodological Issues." Journal of Economic
Literature 29: 1644-78.
Blejer, Mario, and Liliana Schumacher. 1998. "Central Bank Vulnerability
and the Credibility of Commitments: A Value-at-Risk Approach to Cur-
rency Crises." IMF Working Paper WP/98/65. International Monetary
Fund, Washington, D.C.



EEALING WITH GOVERNMENT FISCAL RISK                           55
Bodie, Zvi, and Philip Davis. 2000. The Foundations of Pension Finance. 2
vols. Cheltenham; Northampton: Elgar Reference Collection.
Brixi, Hana Polackova, Hafez Ghanem, and Roumeen Islam. 1999. "Fiscal
Adjustment and Contingent Government Liabilities." Policy Research
Working Paper 2177. World Bank, Washington, D.C.
Brixi, Hana Polackova, Anita Papp, and Allen Schick. 1999. "Fiscal Risks
and the Quality of Fiscal Adjustment in Hungary." Policy Research Working
Paper 2176. World Bank, Washington, D.C.
Brock, Philip. 1992. "External Shocks and Financial Collapse: Foreign Loan
Guarantees and Intertemporal Substitution of Investment in Texas and
Chile." American Economic Review 82 (2): 168-73.
Buiter, Willem. 1985. "A Guide to Public Sector Debt and Deficits." Eco-
nomic Policy (November): 14-79.
Butler, Cormac. 1999. Mastering Value at Risk: A Step-By-Step Guide to
Understanding and Applying Var. London; New York: Financial Times/
Prentice Hall.
Calderon Zuleta, Alberto. 1999. "Valuing and Managing Risk Associated with
Government Contingent Liabilities." In Proceedings of the Second Sovereign
Debt Management Forum. Washington, D.C., November 1-3. World Bank.
Caprio Jr., Gerard, and Daniela Klingebiel. 1997. "Bank Insolvency: Bad
Luck, Bad Policy, or Bad Banking?" In Michael Bruno and Boris Pleskovic,
eds., Annual World Bank Conference on Development Economics 1996.
Washington, D.C.: World Bank.
Cassard, Marcel, and David Folkerts-Landau. 1997. "Risk Management of
Sovereign Assets and Liabilities." IMF Working Paper WP/97/166. Inter-
national Monetary Fund, Washington, D.C.
Chase Manhattan Bank. 1996. Risk Management Handbook, Private Sector
Participation in Urban Infrastructure. Washington, D.C.: U.S. Agency for
International Development.
Claessens, Stijn. 1993. "Risk Management in Developing Countries." Tech-
nical Paper 235. World Bank, Washington, D.C.
Commonwealth Secretariat. 2001. External Sector Contingent Liabilities-
A Study on India. London, January.
Congressional Research Service. 1998. Tax Expenditures. Compendium of
Background Material on Individual Provisions. Committee on the Bud-
get, U.S. Senate, Washington, D.C.
Cox, John, and Mark Rubinstein. 1985. Options Markets. Upper Saddle
River, N.J.: Prentice Hall.
Davis, Jeffrey, Rolando Ossowski, James Daniel, and Steven Barnett.
2001. "Stabilization and Savings Funds for Nonrenewable Resources.
Experience and Fiscal Policy Implications." Occasional Paper 205. Inter-
national Monetary Fund, Washington, D.C.
Dillinger, William. 1999. "Fiscal Management in Federal Democracies: Ar-
gentina and Brazil." Policy Research Working Paper 2121. World Bank,
Washington, D.C.



56                                              BRIXI AND MODY
Dooley, Michael. 2000. "Debt Management and Crisis in Developing Coun-
tries." Journal of Development Economics 63 (October): 45-58.
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy (April).
Easterly, William, Carlos Alfredo Rodriguez, and Klaus Schmidt-Hebbel. 1994.
Public Sector Deficits and Macroeconomic Performance. Washington,
D.C.: World Bank.
Eisner, Robert. 1984. "Which Budget Deficit? Some Issues of Measurement
and Their Implications." American Economic Review 74 (2):138-43.
Eisner, Robert, and Paul Pieper. 1984. "A New View of the Federal Debt and
Budget Deficits." American Economic Review 74 (2): 11-29.
Fischer, Stanley, and William Easterly. 1990. "The Economics of the Govern-
ment Budget Constraint." World Bank Research Observer 5 (2).
Fornasari, Francesca, Steven Webb, and Hcng-fu Zou. 2000. "The Macro-
economic lImpact of Decentralized Spending and Deficits: International
Evidence." Annals of Economics and Finance 1 (November): 403-33.
Freeman, Brian, and Allan Mendelowitz. 1982. "Program in Search of a
Policy: The Chrysler Loan Guarantee." Journal of Policy Analysis and
Management I (summer): 443-53.
Froot, K. A., D. S. Scharfstein, and J. C. Stein. 1994. "A Framework for Risk
Management." Harvard Business Review (November-December).
Garber, Peter. 1998. "Derivatives in International Capital Flows." Working Pa-
per 6623. National Bureau of Economic Research, Cambridge, Mass., June.
Honohan, Patrick. 1999. "Fiscal Contingency Planning for Banking Crises."
Policy Research Working Paper 2228. World Bank, Washington, D.C.
Hueth, Darrell L., and William H. Furtan. 1994. Economics of Agricultural
Crop Insurance: Theory and Evidence. Norwell, Mass.: Kluwer Academic
Publishers.
Hull, John. 1997. Options, Futures, and Other Derivatives. Upper Saddle
River, N.J.: Prentice Hall.
Hutchinson, Gladstone. 2000. "The Challenge of Fiscal Sustainability: Pub-
lic Enterprise Financial Relationship with Government in Jamaica."
Lafayette College. Processed.
IMF (International Monetary Fund). 1996. "Aging Populations and the Fis-
cal Consequences of Public Pension Schemes." Washington, D.C.
Insurance Services Office. 1999. "Insuring Catastrophic Risk." Processed.
Irwin, Timothy, Michael Klein, Guillermo Perry, and Mateen Thobani. 1998.
Puiblic Risk in Private Infrastructure. Washington, D.C.: World Bank.
Kharas, Homi, and Deepak Mishra. 2001. "Fiscal Policy, Hidden Deficits,
and Currency Crises." In S. Devarajan, E H. Rogers, and L. Squire, eds.
World Bank Economists' Forum. Washington, D.C.: World Bank.
Kruger, Coen. 1999. "Valuing and Managing Risk Associated with Govern-
ment Contingent Liabilities." Presentation at the Second Sovereign Debt
Management Forum, Washington, D.C., November 1-3. World Bank.



DEALING WITH GOVERNMENT FISCAL RISK                           57
Kryzanowski, L., and G. S. Roberts. 1993. "Canadian Banking Solvency."
Journal of Money, Credit and Banking 25 (3): 361-76.
Larson, Donald F., Panos Varangis, and Nanae Yabuki. 1998. "Commodity
Risk Management and Development." Policy Research Working Paper
1963. World Bank, Washington, D.C.
Leaven, Luc. 2000. "Banking Risks around the World: The Implicit Safety
Net Subsidy Approach." World Bank. Processed.
Lewis, Christopher, and Ashoka Mody. 1997. "The Management of Con-
tingent Liabilities: A Risk Management Framework for National Gov-
ernments." In Timothy Irwin, Michael Klein, Guillermo Perry, and
Mateen Thobani, eds., Dealing with Public Risk in Private Infrastruc-
ture. World Bank Latin American and Caribbean Studies. Washington,
D.C.: World Bank.
M\4arrison, Christopher. 2001. "Risk Measurement for Project Finance Guar-
antees." Capital Markets Company. Processed.
Mody, Ashoka. 2000. "Nicaragua: Rainfall Risk Management Project." World
Bank. Processed.
Mody, Ashoka, and Dilip Patro. 1996. "Valuing and Accounting for Loan
Guarantees." World Bank Research Observer 11 (February): 119-42.
Nars, Kari. 1997. Excellence in Debt Management. London: Euromoney.
OECD (Organisation for Economic Co-operation and Development). 1999.
How Should Governments Invest Financial Assets and Manage Debt.
PUMA/SBO/RD (99). Paris: OECD.
. 2000. Reforms for an Aging Society. Paris: OECD.
Patterson, Linda. 2001. "The State of Texas Hedging Program: A Long Posi-
tion and a Long Perspective." Presentation delivered at World Bank, June
25. Patterson and Associates. Processed.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Salant, S. W. 1983. "The Vulnerability of Price Stabilization Schemes to Specu-
lative Attack." Journal of Political Economy 91: 1-38.
Schiavo-Campo, Salvatore, and Daniel Tommasi. 1999. Managing Govern-
ment Expenditure. Manila: Asian Development Bank.
Schick, Allen. 1998. A Contemporary Approach to Public Expenditure Man-
agement. Washington, D.C.: World Bank.
. 2001. "Does Budgeting Have a Future?" Presentation at OECD Se-
nior Budget Officials' Meeting, Paris, May. Organisation for Economic
Co-operation and Development.
Selowsky, Marcelo. 1998. "Fiscal Deficits and the Quality of Fiscal Adjust-
ment." In The Challenges for Public Liability Management in Central
Europe. Washington, D.C.: World Bank.
Standard and Poor's. 1997. "Criteria: Financial System Stress and Sovereign
Credit Risk." Sovereign Ratings Service (December).



58                                             BRIXI AND MODY
Stanton, Thomas, and Ronald Moe. 2001. "Government Corporations and
Government Sponsored Enterprises." In Lester M. Salamon, ed., The Tools
of Government: A Public Management Handbook for the Era of Third-
Party Government. New York: Oxford University Press.
Townsend, R. M. 1977. "The Eventual Failure of Price Fixing Schemes."
Journal of Economic Theory 14: 190-99.
U.S. GAO (General Accounting Office). 1993. Bank Insurance Fund: Review
of Loss Estimation Methodologies. Washington, D.C.
. 1998. Budgeting for Federal Insurance Programs. Washington, D.C.
Van Order, Robert. 2000. "The Structure and Evolution of American Sec-
ondary Mortgage Markets, with Some Implications for Developing Mar-
kets." Freddie Mac, Washington, D.C. Processed.
World Bank. 1994. Averting the Old Age Crisis. Washington, D.C.
1997. Public Expenditure Management Handbook. Washington, D.C.
1999a. "Dealing with Commodity Price Volatility in Developing
Countries: A Proposal for a Market-based Approach?" Discussion Paper
for the Roundtable on Commodity Risk Management in Developing Coun-
tries, Washington, D.C., September 24.
1999b. Malaysia: Public Expenditure Review. Washington, D.C.
1999c. Toolkils for Private Participation in Water and Sanitation.
Washington, D.C.
-    . 2000. Proceedings of the Second Sovereign Debt Management Fo-
rum. Washington, D.C., November 1-3.
- 200 la. Finance for Growth: Policy Choices in a Volatile World. Policy
Research Report, May.
. 2001b. "Parastatal Performance and Strategy in Bangladesh."
Processed.
World Bank and IMF (International Monetary Fund). 2000. "Guidelines for
Public Debt Management" <www.worldbank.org/fps/>.
Yaron, Jacob. 1992. "Assessing Development Finance Institutions-A Public
Interest Analysis." World Bank Discussion Paper 174. World Bank, Wash-
ington, D.C.



CHAPTER Z
Accounting and Financial
Accountability to Capture Risk
Murray Petrie
The Economics and Strategy Group
THE TRADITIONAL CASH-BASED accounting and reporting systems used
by many governments provide inadequate information within the ex-
ecutive for the management of fiscal risks.' They also do not produce
enough information for the legislature and the public to hold govern-
ments accountable for the management of fiscal risks. The traditional
focus on cash has been consistently associated with fiscal management
practices that are short term and reactive. Poor information has inter-
acted with poor incentives to discourage decisionmakers from taking
a longer-term view of fiscal risks and their management.
In recognition of these shortcomings, a range of initiatives has been
introduced in different countries and by international agencies that,
while not motivated solely by a desire for better risk management, has
important implications for the management of fiscal risk. The initia-
tives taken include: greater overall transparency in fiscal management,
supplementary reporting of noncash information in budget documents
and the final accounts, changes to the basis of accounting, and moves
by governments to tighten control over fiscal risks.
This chapter discusses these initiatives. It does so within a frame-
work that focuses on risk to the aggregate fiscal position-that is, on
sources of variability in the overall level of government spending, rev-
enues, the fiscal balance, and value of assets and liabilities. While these
sources of variability manifest themselves through variability in indi-
vidual spending, revenue, and financing programs, it is the impact on
the aggregate fiscal position that is the chief object of concern.
Risk is defined here as any situation in which there is a range of
possible outcomes around the expected fiscal position. Inherent uncer-
tainty about which of a number of different possible states of nature
59



60                                           MURRAY PETRIE
will apply in the future creates a range of possible fiscal outcomes.
This approach captures forecasting risk, in addition to uncertain obli-
gations such as guarantees and other contingent liabilities.2 The focus
in this chapter is on the annual budget and the short- to medium-term
fiscal position.
In general, the objective of financial risk management for any entity
is to improve the entity's financial position and performance, while
protecting the entity from unacceptable variance in returns-and in
particular from the risk of unacceptable losses. It is clear that the speci-
fication of what is an unacceptable level of fiscal risk will vary from
country to country. It will depend on the initial fiscal position, the
degree of fiscal flexibility, the nature and extent of fiscal risks, the
quality of the information available, the capacity for risk manage-
ment, and on the perceived "return" to risk-taking. Assessing what a
government's appetite for risk should be is an extremely complex is-
sue, which at present is at the boundary of public finance theory, let
alone practice (for a discussion of these issues, see OECD 1999 and
Polackova 1998). For most governments, determining optimal risk
exposure will not be of practical relevance until major progress has
been made in identifying, analyzing, quantifying, reporting, and man-
aging existing fiscal risks. This is the subject of the rest of this chapter.
The chapter is structured as follows. The next section describes the
weaknesses in traditional cash-based budget management systems with
respect to the management of fiscal risks. That section is followed by
one that outlines a framework for fiscal risk management, comprising
both macro-level and micro-level elements, and provides country ex-
amples. The chapter ends with some concluding remarks.
Weaknesses in the Treatment of Risk under
Traditional Cash-based Accounting Systems
Governments have historically reported their financial performance
using cash-basis accounting.3 The reasons for this include simplicity,
compliance with legal requirements, and the government's borrowing
requirement and the macroeconomic impact of the budget.
Cash-basis accounting has, however, a number of well-recognized
weaknesses. Some relate specifically to inadequacies in the treatment
of fiscal risks. For example:
o The cash basis results in the incomplete or inaccurate measure-
ment of current transactions. In many instances, there is a discrepancy
between when a commitment is entered into, when resources are used
and the economic effects are felt, and when a cash payment is made.
For example, contingent liabilities are not reflected in pure cash-basis



ACCOUNTING AND FINANCIAL ACCOUNTABILITY                   61
financial statements until and if they have to be paid.4 The real effects
of government lending and insurance programs and of civil service
pension schemes are distorted by cash-basis reporting.5
* There is a lack of information on stocks of assets and liabilities
and the relationship between flow and stock variables. Although most
governments provide at least some information on public debt, depre-
ciation is not reported, and risks relating to an impending need for
capital replacement may remain hidden.
In addition, there has generally been no reporting of the sensitivity of
the fiscal position to changes in key forecasting assumptions.
More seriously, the deferred recognition of expenditures can hide
an accumulation of problems until they reach massive proportions.
The classic example of this is regulatory forbearance in banking super-
vision. Regulators and politicians may be tempted to defer dealing
with insolvent banks in the hope the banks will recover, or, if not, that
the cost will fall on a future administration.6 With the expectation of
future government support, the managers of distressed banks may take
on even higher levels of risk in the knowledge they are in effect gam-
bling with public money. The cost of financial sector reconstruction in
the last two decades has been massive in many countries and consti-
tutes a major source of fiscal risk.
In general, therefore, pure cash-basis accounting and reporting re-
sult in inadequate information for use by the executive in managing
fiscal risks and for use by the legislature and the public in holding the
government accountable for such management. Moreover, it often
leaves financial markets guessing about the true state of public finances
and the likely (as opposed to the budgeted) fiscal deficit. In recogni-
tion of these shortcomings, some countries have introduced in recent
years a range of initiatives in fiscal risk management. These include
overall transparency in the public accounts7; supplementary reporting
of information on fiscal risks in budget documents, in the final ac-
counts, or in both; a change to the basis of accounting; and a central-
ized approach to implementing fiscal risk management within the
government. These initiatives are discussed in detail in the next section.
Recent Initiatives to Build Accountability
in Fiscal Risk Management
The approach to fiscal risk management outlined below contains what
might be thought of as both macro- and micro-level elements. The
macro-level element covers broad transparency to the public in the
conduct of fiscal policy. The more micro-level elements comprise specific
reporting of information on fiscal risks, the choice of an appropriate



62                                             MURRAY PETRIE
basis of accounting, and a centralized approach to the implementation
of fiscal risk management.
Greater Overall Fiscal Transparency
Conducting fiscal policy in a transparent and open manner is a high-
level approach that has the potential to reduce significantly fiscal
risk and improve risk management. Making information available
publicly requires institutional capacity and systems within govern-
ment so that the information is available within government in the
first instance.8 Such "internal transparency" is capable on its own of
producing major gains in the management of fiscal risk through im-
proving the information base within the executive for fiscal
decisionmaking and strengthening accountability within government
for risk management.
In the absence of public transparency, however, there is limited ex-
ternal accountability and less assurance that there will be a sustain-
able improvement in the management of fiscal risk. And in the absence
of external accountability, there may be less incentive for the govern-
ment to implement sound internal risk management systems in the
first place.
Construed broadly, fiscal transparency requires much more than
just the timely publication of the budget and final accounts. It also
covers clarity of roles and responsibilities; commitment to the timely
publication of complete information on the past, present, and pro-
jected fiscal position; open budget preparation, execution, and report-
ing; and independent assurances of the integrity of fiscal information.9
Transparency in all these dimensions is a critically important ele-
ment of a medium-term fiscal risk management strategy. While there
are elements of transparency that relate specifically to the reporting of
fiscal risks (see below), there are also broader elements that can sig-
nificantly reduce fiscal risk. For example, the main source of fiscal risk
in many countries is debt servicing. Transparency in the conduct of
debt management is an important element in effectively managing risks
and ensuring the accountability of those responsible. Transparency in
this respect involves clarity of roles and responsibilities and clear ob-
jectives for debt management; open processes for conducting debt
management operations; and regular and timely reporting of a range
of details of government debt.'0
A further problem in many countries has been the lack of clarity of
roles within the public sector. Directives to state-owned enterprises
(SOEs) to conduct a proliferation of quasi-fiscal activities have often
resulted in these institutions accumulating losses and needing to be
bailed out."1 In formal terms, such potential spending can be seen as a
type of implicit contingent liability (as illustrated by the Fiscal Risk



ACCOUNTING AND FINANCIAL ACCOUNTABILITY                   63
Matrix in Chapter 1). A transparent, arms-length governance frame-
work for SOEs, with separate identification (and funding from the
government budget) of noncommercial obligations, can reduce what
is in many countries a major source of fiscal risk. A lack of clarity in
the respective roles of the central government and lower-level govern-
ments and ad hoc mechanisms for determining intergovernmental trans-
fers are also sources of fiscal risk in many countries.
A more general point is that greater transparency in the conduct of
fiscal policy reduces the risk of errors. In a situation in which func-
tions such as macroeconomic forecasting, fiscal forecasting, and
macrofiscal analysis are conducted entirely within the government,
there may be no independent check on the quality and accuracy of
fiscal information and analysis.
Supplementary Reporting of Information on Risks
]:n view of the deficiencies in cash-basis reporting, a number of coun-
tries have in recent years initiated supplementary reporting of infor-
mation on fiscal risks. Some international standard-setting agencies
also have been revising or developing standards for fiscal reporting
that bear on the management of fiscal risks.
For example, the Code of Good Practices on Fiscal Transparency
issued by the International Monetary Fund (IMF) contains a require-
ment that countries publish a fiscal risk statement with the annual
budget. The Fund's "Manual on Fiscal Transparency" indicates that
the statement should contain information on specific fiscal risks-such
as contingent liabilities-and on general forecasting risks.
The International Federation of Accountants (IFAC) is undertaking
a medium-term project to develop a core set of standards for financial
reporting by governments. The standards are for the cash and accrual
bases of accounting. As part of this project, IFAC published the study
Governmental Financial Reporting: Accounting Issues and Practices.
Rather than prescribing particular accounting treatments, it describes
the types of additional disclosures that some governments using cash-
basis reporting make on contingent liabilities. It also notes that report-
ing of information on contingencies is required under accrual accounting
(see IFAC 2000: 47-49, 171).
The IMF is also revising its Government Finance Statistics Manual.'2
Some major changes are proposed, including use of an accrual basis of
recording and compilation of information on contingent liabilities in a
supplement to the balance sheet.
There are different ways in which to categorize fiscal risks as a
useful way of organizing supplementary reporting. One of the possi-
bilities would be according to whether they are forecasting risks (these
are related to the types of risk discussed in Chapter 1) or specific fiscal



64                                            MURRAY PETRIE
risks (discussed as sources of risk in the Fiscal Risk Matrix of Chapter
1). Forecasting risks are the inherent risks involved in forecasting the
fiscal aggregates. Budget forecasts are normally highly sensitive to the
assumptions made about a small number of key parameters. Govern-
ments also are typically exposed to specific fiscal risks. These include
contingent liabilities such as guarantees, indemnities, uncalled capital,
and legal action against the government.
Forecasting Risks. Forecasting risks should be disclosed in the cen-
tral government's annual budget documents. The realism and reliabil-
ity of the budget are generally highly dependent on the quality of the
underlying macroeconomic forecasts on which the budget forecasts
are based. Typically, there also will be a small number of key forecast-
ing assumptions related to particular revenue sources or expenditure
programs (for example, the price of oil or the exchange rate). Variabil-
ity in the annual cost of debt servicing due to factors such as exchange
rate, interest rate, and maturity structure risk can be a major source of
exposure as well.
The budget documents should therefore provide information on the
key assumptions on which the budget forecasts are based, and they
should illustrate the sensitivity of the budget to variations in these key
assumptions. Periodic assessments also should be published of the reli-
ability of budget macroeconomic and fiscal forecasts compared with
outturn, with an analysis of deviations by major category. For example,
information should be provided on the effects on forecast revenues,
expenditures, and the overall balance of, say, a 1 percent decrease in
growth of the gross domestic product (GDP) from that assumed in the
budget. Half of the member countries of the Organisation for Eco-
nomic Co-operation and Development (OECD) publish a fiscal sensi-
tivity analysis."3 U.S. budget documents, for example, contain detailed
information about and discussion of the economic assumptions under-
lying the budget, including comparisons with the assumptions devel-
oped by the Congressional Budget Office and with the assumptions
contained in the administration's budget for the previous year (see
United States 1999: chap. 1, Economic Assumptions).
In addition to sensitivity analysis, it is desirable to publish some
alternative medium-term scenarios in which different economic growth
and aggregate fiscal developments are combined. These can illustrate
the robustness of the budget in the face of broad alternative develop-
ments. An example of scenario analysis reporting by the New Zealand
government is shown in Table 2.1.
In this context, a government might go the further step and discuss
its fiscal strategy in the event that the economic and fiscal outlook
turns out to be less favorable than that contained in the budget fore-
casts. Discussion of broad fiscal contingency plans could help to re-



ACCOUNTING AND FINANCIAL ACCOUNTABILITY                        65
Table 2.1 Fiscal Scenario Reporting: Summary of Alternative
Scenarios, New Zealand
1997/98 1998/99 1999/00  2000/01  2001/02
actual forecast projection projection projection
Production GDP (%)
Central forecast            2.0    (0.3)   2.9      3.5      3.0
Export-led recovery         2.0    (0.3)    3.8     4.4      2.7
Wreak recovery              2.0    (0.4)    2.0     1.8      3.7
Nominal expenditure GDP (%)
Central forecast            3.3     0.4    3.8      5.2     4.6
Export-led recovery         3.3     0.4    4.9      6.5     5.1
Wleak recovery              3.3     0.3    2.7      2.6     5.2
Operating balance (billion $)
Central forecast            2.5     2.2    (0.0)    0.8      1.5
Export-led recovery         2.5     2.2    0.3      1.7     2.8
Weak recovery                2.5    2.1    (0.6)   (0.7)    (0.3)
Source: Based on Table 3.1 in Government of New Zealand (1999: 59).
duce market uncertainty about the possible path of fiscal policy. Pro-
viding markets with a broad indication of what sort of fiscal adjust-
ments will be made in response to possible adverse developments-for
example, spending cuts or deferrals, tax increases, a bigger deficit, or
some combination of these elements-may reduce the risk of abrupt
market reactions to adverse market developments. Such an indication
also may make it more likely that a government will be ready to take
quick action when and if an adverse event does occur.
Specific Fiscal Risks. For contingent liabilities, supplementary re-
porting should take the form of a statement of the outstanding stock of
contingent liabilities of the central government. In addition to its pub-
lication with the final accounts, a statement of contingent liabilities
should be included with the annual budget documents in order to pro-
vide a complete picture of the fiscal position at the time the budget is
presented. To qualify as a contingent liability, the amount of expendi-
ture at risk should be material,'4 and the likelihood that the item will
result in future expenditure should be more than remote. Under ac-
crual accounting, to qualify as a contingent liability a possible future
expenditure must also be less than likely."
For each such contingent liability, or pooled program of contingent
liabilities, information should be presented on its nature and potential
fiscal significance. For example, for a loan guarantee (or a portfolio of
similar loans) reporting should cover the amount of the loan, to whom



66                                           MURRAY PETRIE
the loan has been advanced, and the duration of the loan. Whether
there have been any changes in the details of the item since the previ-
ous reporting date also should be noted. Where possible, an estimate
should be provided of the expected cost of each contingent liability-
desirably in the form of a range rather than just a point estimate. Some
contingent liabilities are nonquantifiable, however, while an estimate
of expected cost for others would not be sufficiently reliable. An ex-
ample of contingent liability reporting, taken from the first trial bal-
ance sheet of the Japanese government, is shown in Box 2.1.
Other specific, short-term fiscal risks that should be reported with
the budget include the following:
* Where there is an unusual degree of uncertainty about the likely
cost of a material expenditure item in the budget, this should be dis-
closed. For example, perhaps the government has made provision for
meeting the costs of reconstruction following a major disaster. At the
time of finalizing the budget the cost allowed in the budget may be
very uncertain.
* Where items have not been included in the budget at all because
of the extent of uncertainty about their timing, magnitude, or eventu-
ality, these items should be disclosed. For example, the government
may have announced its intention to increase food subsidies or public
pensions, but the details of the decision may not have been finalized
sufficiently for inclusion in the budget. The government of New Zealand
reports these sorts of specific fiscal risks in its Budget Economic and
Fiscal Update report accompanying the annual budget (Government
of New Zealand 1999).
A Change to the Basis of Accounting
The basis of accounting refers to the set of accounting principles for
recording transactions that determine when the effects of transactions
or events are recognized for financial reporting purposes. The accrual
basis of accounting entails recognition of transactions or events at the
time the transaction or event occurs rather than at the time a cash
payment is made. Accrual accounting also entails the production of a
full balance sheet. These two differences between cash and accrual
accounting have important implications for the management of fiscal
risk. In addition, the disclosure of supplementary information on con-
tingent liabilities and commitments is required under IFAC's Gener-
ally Accepted Accounting Practice (as reflected in international
accounting standards)."6
A balance sheet encapsulates a longer-term perspective on the
government's financial position. In principle, it represents the estimated
present value of future cash flows-provided they meet the accounting



ACCOUNTING AND FINANCIAL ACCOUNTABILITY                         67
Box 2.1 Reporting of Contingent Liabilities in Japan
The following table is an extract from the Japanese government balance
sheet (preliminary trial), as of March 31, 1999 (Japan, Study Group on
Explanatory Methods of Fiscal Position 2000).
1. Contingent liabilities.
(1) Liabilities for loan guarantees and loss compensation
contracts (million yen)
Amount in     Amount
Items                        foreign currency  in yen
Compensation for Nuclear
Energy Business                             798,000
Guarantee for principal and
interest payment for bonds
of Japan Finance Corporation
for Small Business                           400,350
Total                                       48,928,627
(2) Claims for damages in pending cases (miUlion yen)
Cases                                  Amount claimed
Claim for injunction in Amagasaki Air
Pollution Case (Kobe District Court, (wa)
No. 2217, 1988; (wa) No. 1766, 1995)      12,168
Appeal case for restitution of unjust enrich-
ment (Endless Money Chain Case)
(Fukuoka High Court, (administrative; ko)
No. 11, 1996)                             10,396
Total                                       88,510
Note 1: All the amounts claimed are mentioned in this table, whether
or not the Government is expected to win the case.
Note 2: In cases where the claimed amount is more than 1 billion yen,
case names are mentioned.
(3) Other major contingent liabilities
Title            Outline
Project for     When natural disasters occur and prefectural
providing aid    authorities provide aid money to the head
money to assist  of each household affected by disasters, the
victims' recovery  Government shall be liable to share part of
from disasters,  the aid money under the provisions of
Cabinet orders.



68                                            MURRAY PETRIE
definition and recognition criteria and can be reliably measured. A
balance sheet therefore provides significant additional information
about the future implications of current policies. For example, under
accrual accounting information is provided on the full cost of current
civil service pension policies, the accumulating burden on future bud-
gets, and the variability in the value of the liability from year to year.
Such information can help focus debate on the appropriate design of
pension schemes. The choice between defined-contribution and de-
fined-benefit pension schemes, for example, involves significant differ-
ences in the amount of fiscal risk borne by the government.'7
However, accrual accounting on its own does not provide all the
information that a fiscal economist would wish. This can be a source
of confusion because of the sometimes different use of the word ac-
crual by economists and accountants. To some economists, accrued
expenditures mean any real consumption or outflow of resources. To
an accountant, an accrued expenditure must meet the definition of a
liability and the recognition criteria of a liability-chiefly, that it can
be reliably measured. The effect of the definition and recognition cri-
teria in accrual accounting is that there is a significant difference be-
tween the information that would be contained in a "comprehensive
balance sheet"'8 and that contained in a balance sheet produced in
accordance with current internationally recognized accounting prac-
tices. For example, the expected cost of possible obligations such as
one-off guarantees is not generally recognized as a liability and inte-
grated into the budget by governments that have adopted accrual bud-
geting. A guarantee does not meet the recognition criteria for an
expense, unless it is judged more likely than not that the expense will
in fact occur and the expected cost of the guarantee can be estimated
with sufficient reliability. In general, this means that one-off guaran-
tees will not normally be recognized as expenses in the budget.
The desirable approach to accounting (and budgeting) for specific
fiscal risks requires balancing two important principles in public fi-
nance. First, decisionmaking is best informed, and incentives are best
aligned, if governments recognize the cost of commitments at the time
they are made. Second, budget appropriations should be based on re-
liable information compiled on the basis of widely accepted account-
ing policies and supported by credible institutional arrangements."9
The best approach seems likely to depend on the particular circum-
stances in individual countries, including the significance and nature
of specific fiscal risks, the existing accounting system, and the country's
financial management capacity. Introducing an annual contingency fund
and expanding the reporting of information on fiscal risks-both fis-
cal sensitivity and specific fiscal risks-can be readily done in the con-
text of a cash-basis accounting and budgeting system. Malaysia and
Japan are examples of countries using a modified cash basis of ac-



ACCOUNTING AND FINANCIAL ACCOUNTABILITY                    69
counting, but which report supplementary information on guarantees.20
Much of the information on fiscal risks is in any case provided through
such supplementary reporting even in those countries that have adopted
full accrual accounting. The adoption of accrual accounting is a large
undertaking, and it should be guided by broader considerations than
the management of fiscal risk alone.
A Centralized Approach to Implementing
Fiscal Risk Management
The nature of the fiscal risk management function lends itself to a
centralized policy setting and oversight. The links to the budget pro-
cess, and the need (at least in principle) to take a portfolio approach to
the analysis of risk, suggest that a completely decentralized approach
would not be effective. Depending on the overall public sector man-
agement framework, however, some combination of central oversight
and decentralized accountabilities will be appropriate.
A further important source of fiscal risk in most countries, as noted,
is the sensitivity of the fiscal position to changes in the economic envi-
ronment. An early step in a risk management strategy should be re-
porting the fiscal sensitivity analysis. This requires close coordination
between those responsible for fiscal and macroeconomic forecasting,
particularly in generating realistic alternative macroeconomic scenarios
to test the sensitivity of the fiscal baseline.
In generating alternative fiscal scenarios, allowance should be made
in one scenario for a combination of adverse events to stress test the
medium-term fiscal baseline. Such a scenario might include a fall in
economic growth, a drop in government revenues, an increase in rou-
tine government spending, a shortening of the maturity structure of
public debt, the calling of some guarantees, and expenditure demands
from implicit contingent liabilities. The likely correlations between
these adverse developments mean that while such a scenario may not
be likely, it will nevertheless be very much within the range of possi-
bilities (as is all too often demonstrated by actual country experiences).
At the same time, information should be aggregated across the cen-
tral government on the specific fiscal risks to which individual govern-
ment agencies are currently exposed. The information should include,
to the extent possible, an estimate of the likely range of expected cost
to the government for each risk or pooled program of risks. This task
might best be achieved as a component of the fiscal reporting indi-
vidual agencies must provide to the ministry of finance.21 This requires
a clear, common framework defining fiscal risks-for example, in the
government's accounting policies (for contingent liabilities) and in the
budget instructions sent to departments. It requires communication of
a clear expectation that agency heads will be accountable for complying



70                                          MURRAY PETRIE
and of a means to verify compliance. It also requires careful central
monitoring. For example, the maturity profile of the different risks
and guarantees must be analyzed to avoid any undesirable bunching
of exposures. Identification of priorities is required to facilitate choices
between competing proposals for government guarantees. And, once
sound governance arrangements are in place, ongoing central moni-
toring is required of the performance of individual agencies in mitigat-
ing the fiscal risks to which they are exposed. Compliance by individual
agencies will, in turn, mean they must have the necessary accounting
and information systems to capture and report the relevant informa-
tion, and an effective internal control environment.
Kazakhstan provides a good example of a centralized approach to
the ongoing management of existing guarantees (see Box 2.2). Every
entity whose foreign borrowing is guaranteed by government is re-
quired to provide quarterly financial statements to a special monitor-
ing unit in the Ministry of Finance. They also must lodge the funds
required for loan repayment installments into a restricted account one
month before each installment is due and maintain the equivalent of
one installment in the account at all times. This provides scope for the
government to take action in the event of a failure to lodge a payment,
before a call is made on the budget to honor the guarantee.
Another measure that would strengthen the accountability of
policymakers and civil servants for prudent management of risks would
be to incorporate risk-taking and risk management within the scope of
the external audit. This would mean that the supreme audit institution
would audit the information provided by each government agency on
the fiscal risks to which it is exposed. In New Zealand, each govern-
ment department must maintain a Register of Contingent Liabilities in
which the details of all contingent liabilities are recorded. Each minis-
ter responsible for a department is required to sign a certification twice
each year that the department's schedule of contingent liabilities is
complete and accurate to the best of his or her knowledge. These docu-
ments are subject to external audit. The New Zealand controller and
auditor general also has reported to Parliament on the performance of
individual entities in managing specific fiscal risks (see New Zealand
Controller and Auditor General 1999). Similarly, the U.S. General
Accounting Office reports to the U.S. Congress on contingent liabili-
ties and other risks facing the federal government and on the analysis
and management of the risks by the responsible departments.22
There may still be a problem with guarantees "in the bottom drawer"
that never see the light of day until they are presented as a fait accom-
pli to the ministry of finance. This is likely to be a problem in coun-
tries with a history of hidden guarantees and off-budget activities. One
way to try to break out of this low-level equilibrium might be to take
the approach sometimes adopted for expenditure arrears-that is, the



ACCOUNTING AND FINANCIAL ACCOUNTABILITY                        71
Box 2.2 Management of Fiscal Risks in Kazakhstan
Kazakhstan provides a good example of a concerted approach to strength-
ening the management of fiscal risks (Kazakhstan 2000). The impetus for
reform came after a proliferation of guarantees culminated in a large call
on the budget during a difficult period of fiscal consolidation. This situa-
tion focused attention on the need to bring guarantees under effective
centralized control. Kazakhstan also faces substantial variability of bud-
get revenues, in part because of its reliance on oil revenues.
Reforms introduced to improve the management of fiscal risks include
the following:
* All guarantees now require the prior approval of the minister of
finance.
* Every entity whose foreign borrowing is guaranteed by government
is required to provide an annual business plan and quarterly financial state-
ments to a special monitoring unit in the Ministry of Finance. They also
must lodge the funds required for loan repayment installments in a restricted
account one month before each installment is due and maintain the equiva-
lent of one installment in the account at all times. The Ministry of Finance
reports quarterly to the government on the financial condition of guaran-
tee recipients. This reporting provides scope for the government to ensure
timely action is taken to prevent a call on its budget for covering such debts
in the event of a failure by the guarantee recipient to lodge a payment.
* The government has introduced an annual ceiling on the stock of
new guarantees that can be issued. This ceiling has resulted in careful
scrutiny of all new guarantee proposals. A number of foreign-financed
new capital projects have not gone ahead because, on closer examina-
tion, it was decided that it was not appropriate for the government to
issue a guarantee.
* An allowance is made within net lending in the central government
budget for the estimated cost of guarantees in the next year. This allow-
ance comprises a mixture of one-off guarantees that are very difficult to
estimate and some large guarantee programs for which some historical
loss information is available. If the funds are not required, the spending
authority lapses at the end of the year.
* A revenue-dependent contingency fund has been established within the
budget. It contains a list of worked-up capital projects that can only pro-
ceed as and when interim budget revenue targets are met during the year.
government announces that any government guarantees that have not
been declared to the minister of finance by a particular date will not
be honored. If combined with a commitment by the government to
publish a statement containing all guarantees, this approach might be
sufficiently credible to flush out the legitimate government guaran-
tees. The recipients of legitimate guarantees would have both a strong



72                                            MURRAY PETRIE
incentive to ensure that the guarantee's existence was declared and the
means to monitor whether it was declared. The success of such an
approach would, however, be critically dependent on the presence of
sufficient political will to enforce the strategy.
The next step would be the imposition of a control framework for
the management of existing fiscal risks. This step might be taken on a
decentralized basis, in which each agency would be responsible and
accountable for managing its own fiscal risks in a prudent manner.
Central agencies would monitor performance, but might require that
a prior clear and comprehensive risk management strategy be in place
in each agency. Alternatively, or in addition, the ministry of finance
might impose specific risk management policies and practices for ge-
neric or significant risks (for example, a foreign exchange risk man-
agement policy to which each agency with significant foreign exchange
exposure must adhere). Specific responsibilities should be assigned
within government for monitoring fiscal risks across the SOE and fi-
nancial sectors.
Central controls also should be imposed on taking on new specific
fiscal risks. Because of the potential for budget discipline to be circum-
vented, and for a loss of fiscal control through the issuing of guaran-
tees, centralized processes should be put in place to enable the
government to control the issuing of new guarantees and indemnities.
Depending on the country concerned, this control might mean a re-
quirement for prior approval of the minister of finance, the cabinet, or
parliament.23 In Japan, the Parliament approves an annual ceiling on
the face value of new guarantees issued in each fiscal year (Japan,
Ministry of Finance 2001).
Approval of a new guarantee also might depend in part on the cir-
cumstances of the recipient of the guarantee-such as its financial sound-
ness or the quality of its governance. In South Africa, policy restricts the
issuing of guarantees to certain situations (see Box 2.3). For example,
no guarantees are provided to private institutions unless management
decisions can be influenced directly by the government. Processes also
might be put in place to constrain the issuing of new guarantees to the
annual budget round. Such a step would at least allow some comparison
of the merits of individual spending and guarantee proposals in terms of
cost-effectiveness. It also could facilitate broad judgments about the
consistency of the total "spending and guarantee package" with the
government's budget and medium-term fiscal strategy.
Some Concluding Remarks
There is a hierarchy of approaches to improving the management of
fiscal risk. A fundamental first step is developing an understanding of



ACCOUNTING AND FINANCIAL ACCOUNTABILITY                           73
Box 2.3 Management of Fiscal Risks in South Africa
South Africa's approach to fiscal risk management comprises both a
macro-level framework and provisions at the micro level (Kruger 1999).
The macro-level framework includes:
* full transparency of fiscal management to ensure accountability (fail-
ure to comply with reporting requirements is a criminal offence);
* a medium-term expenditure framework, which enhances transpar-
ency and predictability;
* intergovernmental fiscal arrangements involving constitutional re-
strictions on the ability of provincial governments to borrow and powers
for the national government to intervene in the event a province incurs an
unauthorized expenditure;
* a coordinating and supervisory role in borrowing by state-owned
enterprises (SOEs); and
* a regulatory environment for the banking and financial sector so
that systemic risks do not pose a threat to planned fiscal outcomes.
The framework for managing fiscal risks at the micro level includes:
* quantification of all financially related assets and liabilities;
* a clear distinction between contingent liabilities and actual liabilities;
* strict guidelines for issuing guarantees-no guarantees to assist pri-
vate institutions unless management decisions can be influenced directly;
guarantees may be provided where there is an obligation in terms of inter-
national treaties, or where foreign loans are considered to be in the na-
tional interest (guarantees for commercial entities are in the process of
being phased out);
* guarantee fees to act as a disincentive to use guarantees and to create
a level playing field where SOEs are competing with the private sector;
* a Public Finance Management Act that establishes full account-
ability, clear reporting responsibilities, and the use of accrual accounting
principles, including the production of a consolidated balance sheet; and
* management of implicit fiscal risks through classifying SOEs on the
basis of the tolerable risk appetite per institutional type-for example,
through restructuring commercial enterprises for privatization; through full
cost recovery in public utility pricing with cross subsidization to be trans-
parent; and through insurance providers charging risk-related premiums.
the main sources of fiscal risk in a particular country. Such an under-
standing requires aggregation and centralization of information across
the central government on the specific fiscal risks to which individual
government agencies and the central government itself are currently
exposed (using, for example, the Fiscal Risk Matrix presented in Chapter
1). This step in itself is a demanding exercise. From this step should



74                                            MURRAY PETRIE
follow attempts to assess at least the broad order of magnitude of the
most significant risks and how they affect different elements of the
government's revenues, expenditures, assets, and liabilities. This as-
sessment enables ongoing monitoring of risks, identification of prior-
ity areas for reducing exposure to existing risks, and some control over
taking on new risks.
Major gains might be made at this stage in many countries through
reexamining some basic policies from the perspective of fiscal risk
management. For example, the framework for intergovernmental fis-
cal relations, the need to retain government ownership of some SOEs
and banks, the management of public debt, and the quality of pruden-
tial supervision of the financial sector are all areas where good policy
design and transparency can make a major contribution to reducing
fiscal risk.
Reporting detailed information on specific fiscal risks, and on the
sensitivity of the fiscal position, should be an early part of a risk man-
agement strategy. Another objective is to publish information on broad
alternative macrofiscal scenarios. Such reporting requires a support-
ing public management infrastructure. In addition to ensuring better
information for the executive on which to base fiscal policy, it pro-
vides the crucial added discipline of external accountability. Efforts to
improve the management of fiscal risks also might benefit from incor-
porating risk management into the scope of the external audit. At the
same time, improvements should be initiated on key deficiencies in
broader fiscal transparency arrangements.
Central controls over who is authorized to take on new risks, such
as issuing guarantees, should be put in place. The possibility of avoid-
ing the risk altogether, or of shifting it partially to other agents, should
be examined at the outset. Consideration of many guarantees also might
be held over for deliberation alongside other fiscal priorities in the
annual budget round. Specific responsibilities should be assigned within
the government for monitoring fiscal risks across the SOE and finan-
cial sectors and for monitoring the financial position of agencies with
government-guaranteed debt. Charging a fee for the issuance and main-
tenance of a guarantee might be a useful additional means of ensuring
sound scrutiny of proposals for new guarantees. Moreover, some at-
tempt should be made to compare the merits of guarantees against
competing fiscal priorities. This might involve setting a limit on the
value of new guarantees entered into.
A change in the basis of accounting toward accrual accounting can
result in better information and accountability for fiscal risk manage-
ment. From an implementation risk perspective, however, it would in
general seem prudent to move first from pure cash-basis reporting to
supplementary reporting of fiscal risks, and the implementation of a
general budget contingency reserve, before considering the establish-



ACCOUNTING AND FINANCIAL ACCOUNTABILITY                         75
ment of dedicated reserve funds for contingent liabilities or introduc-
ing full accrual accounting.
There is great scope for improvements in fiscal performance in many
countries through better management of fiscal risks. The deficiencies
of traditional pure cash reporting have reinforced the tendency of gov-
ernments to take a short-term and reactive approach to fiscal manage-
rnent. New and expanded sources of fiscal risk increase the imperative
for better fiscal control, and new techniques offer the possibility of
improved management and better outcomes.
Notes
1. The author would like to acknowledge helpful comments on earlier
clrafts from David Webber, Ian Ball, Jon Blondal, and Istvan Szekely.
2. Contingencies are defined by the International Accounting Standards
Committee (IASC) as conditions or situations whose ultimate outcome, gain
or loss, will be confirmed only on the occurrence, or nonoccurrence, of one
or more uncertain events. See IFAC (1998: paras. 692-701).
3. See IFAC (1998) for a discussion of the definitions of the different bases
of accounting (cash, modified cash, modified accrual, and accrual accounting).
4. With the exception of origination fees, which, if charged, show as a
favorable cash impact at the time the guarantee is issued, further distorting
the picture.
5. See U.S. GAO (1998a: 5) for a discussion of the distortions of insur-
ance and lending programs under cash accounting. "Cash-based budgeting
for federal insurance programs may provide neither the information nor
incentives necessary to signal emerging problems, make adequate cost com-
parisons, control costs, or ensure the availability of resources to pay future
claims."
6. See U.S. GAO (1998a: 6): "Many analysts believe that the cash-based
budget treatment of deposit insurance exacerbated the savings and loan crisis
by creating a disincentive to close failed institutions. Since costs were not recog-
nized in the budget until cash payments were made, leaving insolvent institu-
tions open avoided recording outlays in the budget and raising the annual
deficit but ultimately increased the total cost to the government."
7. The move to greater overall fiscal transparency, supplementary report-
ing, and the choice of accounting basis are, of course, motivated by many
considerations in addition to risk management objectives.
8. Effective accountability also may require a change in attitudes to embed
a feeling of personal responsibility for fiscal risk management in the public
service culture. Such a change should perhaps be seen in the broader context of
civil service reform. This is beyond the scope of this chapter, but may be a key
element in bringing about an improvement in fiscal risk management.



76                                                MURRAY PETRIE
9. These are the four general principles of the International Monetary
Fund's Code of Good Practices on Fiscal Transparency. See the IMF website
<www. imf. org/external/n p/fad/trans/index. htm>.
10. See the IMF's "Fiscal Transparency Manual" for a discussion of report-
ing public debt. The manual can be found on the IMF website <www.imf.org/
external/n p/fad/trans/index. htn>.
11. Quasi-fiscal activities (QFAs) are activities undertaken under the di-
rection of government by a central bank or state-owned financial or com-
mercial enterprise that are fiscal in character-that is, the effects of the activity
could in principle be duplicated by budgetary measures in the form of a tax,
subsidy, or direct expenditure. Examples are guarantees, subsidized lending,
and financial sector bailouts. These activities have similar economic effects
whether they are undertaken by a central government agency or a central
bank or a public financial enterprise. They can be very large, and they need
to be taken into account in assessing fiscal performance and fiscal risk.
12. See IMF (1996: 85-87). Also available on the Fund's website <wwwimf.
org/external/pubs/ft/gfs/manual/index.htm>.
13. Some OECD countries take additional steps to ensure the integrity
and quality of the macroeconomic forecasts. These range from an expert
independent review panel that comments publicly on the forecasts; to a re-
view of the macroeconomic assumptions by the National Audit Office and a
legal requirement to publish the entire Treasury macroeconomic model (United
Kingdom); to basing the government's official forecasts on private sector
consensus forecasts (Canada); to fully contestable fiscal forecasts produced
by a separate entity, the Congressional Budget Office, which reports directly
to the legislature (United States).
14. It would seem desirable, however, to report contingent liabilities where
the likelihood of actual expenditure is very small but the amount potentially at
risk is very large.
15. Where accrual or modified accrual-basis accounting is used, only those
events that are judged less than likely to result in future expenditure are in-
cluded in supplementary reporting as contingent liabilities. Those events judged
likely to result in future expenditure are recognized immediately as a liability-
that is, they are defined as liabilities rather than contingent liabilities.
16. See IFAC (2000: 171). In accounting, commitments are defined as a
government's responsibility for a future liability based on an existing contrac-
tual agreement. Examples include long-term leases or multiyear contracts for
the purchase of capital equipment.
17. Under a defined-benefit scheme, the government bears the risk of a
mismatch between the return on any scheme assets and the defined pension
obligation. Under a defined-contribution scheme, the contributor bears the
risk of uncertain return on pension fund assets.
18. That is, a balance sheet containing all the prospective cash inflows
and outflows compiled using realistic projections based on current policies.



ACCOUNTING AND FINANCIAL ACCOUNTABILITY                         77
An example of a comprehensive balance sheet is the Fiscal Risk Matrix and
Fiscal Hedge Matrix shown in Chapter 1.
19. Valuing risk may be a demanding exercise; it requires a modeling
capacity that is not readily available in government offices. And even the
results of the best models need to be treated as rough estimates of future values
rather than predictions. The U.S. General Accounting Office, for example, has
commented critically on the ability of U.S. federal agencies to reasonably esti-
mate subsidy costs in their credit programs: "Until weaknesses are addressed
the credibility of loan program cost information they submit will continue to
be questionable" (U.S. GAO 1998b). Therefore, scenario analysis may be more
useful for policymakers than a single expected cost figure.
20. See IFAC (2000: 51) for a discussion of the additional reporting pre-
pared in Malaysia in the context of a modified cash basis of accounting. A
feature typical of the modified cash basis system is holding the books open for a
specified period after year-end to overcome some of the timing problems of pure
cash-basis accounting.
21. References to the ministry of finance are a generic reference to the
ministry or department with primary responsibility for fiscal policy coordi-
nation and budget management.
22. For examples of such reports, see U.S. GAO (1998a, 1998b, 1998c) and
the U.S. General Accounting Office website <uww.gao.gov>.
23. A recent survey of OECD countries found that in three the approval of
only the minister of finance was required for the granting of a guarantee, while
in the great majority the approval of parliament was required (Blondal 1999).
References
Blondal, Jon. 1999. "Management of Fiscal Risk in OECD Member Coun-
tries." Presentation at World Bank Course on Managing Fiscal Risks, Wash-
ington, D.C., June 8-11.
Government of New Zealand. 1999. Budget Economic and Fiscal Update 1999.
Wellington.
IFAC (International Federation of Accountants), Public Sector Committee. 1998.
Draft Guideline for Governmental Financial Reporting. New York.
.2000. Governmental Financial Reporting: Accounting Issues and Prac-
tices. New York.
IMF (International Monetary Fund), Fiscal Affairs Department. 2001. "Code
of Good Practices on Fiscal Transparency" and "Manual on Fiscal
Transparency"<www.imf. orglexternal/nplfad/translindex.htm>.
IMF (International Monetary Fund), Statistics Department. 1996. Government
Finance Statistics Manual: An Annotated Outline. Washington, D.C.
Japan, Ministry of Finance, Financial Bureau. 2001. FILP Report 2000.
February.



78                                               MURRAY PETRIE
Japan, Study Group on Explanatory Methods of Fiscal Position. 2000. "The
Japanese Government Balance Sheet (Preliminary trial)." Tokyo, October.
Kazakhstan. 2000. "Rules for Carrying Out the Monitoring of the Financial
Condition of Legal Entities that Have Received Non-governmental For-
eign Loans under Republic of Kazakhstan Government Guarantees." Astana.
Kruger, Coen. 1999. "Managing Fiscal Risks: The South African Approach."
Presentation at World Bank Course on Managing Fiscal Risks. Washing-
ton, D.C., June 8-11.
New Zealand Controller and Auditor General. 1999. How Are State-Owned
Enterprises Managing Foreign Exchange Risk. Chapter 8, First Report for
1999. Wellington.
OECD (Organisation for Economic Co-operation and Development). 1999. How
Should Governments Invest Financial Assets and Manage Debt? PUMA/SBO/
RD (99). Paris.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington D.C.
U.S. GAO (General Accounting Office). 1998a. Budgeting for Federal Insur-
ance Programs. GAO/T-AIMD-98-147. Washington, D.C.
. 1998b. Credit Reform: Greater Effort Needed to Overcome Persistent
Cost Estimation Problems. GAO/AIMD-98-14. Washington, D.C., March.
.1998c. Report on US Government's Consolidated Financial Statements
for FY97. GAO/AIMD-98-127. Washington, D.C.
United States. 1999. Budget of the United States Government, Analytical Per-
spectives. Fiscal Year 1999. Washington, D.C.: Government Printing Office.



CHAPTER 3
Budgeting for Fiscal Risk
Allen Schick
University of Maryland
As MEASURES OF FISCAL RISK, conventional budgets are deficient on
two counts. First, they have a short time frame-one year in countries
that have only annual budgets, three to five years in countries that
budget within medium-term fiscal frameworks. These time horizons
are too short to account for the downstream risks taken by govern-
rnents when they establish pension systems and other entitlements,
issue or guarantee loans, or promise to make good on shortfalls in
financial performance. Second, conventional budgets record only cash
flows; they do not account for the buildup of liabilities, contingent
obligations, or the future cost of past commitments.
Because budgets measure cash flows rather than liabilities and short-
term payments rather than long-term risks, politicians have an incen-
tive and opportunity to provide benefits to those who seek assistance
from government in ways that mask the true cost, leading to policies
and actions that worsen future fiscal conditions. Governments create
fiscal illusions, beneficiaries behave in morally hazardous ways, and
the upshot is escalation in fiscal jeopardy. This pattern is widespread;
it occurs in developed, transitional, and developing countries. In de-
veloped countries, the failure to properly control and account for risk
may burden future budgets and take a bite out of economic growth. In
transitional and newly developed countries, however, the failure may
retard development and reverse recent economic gains. In developing
countries, failure to deal adequately with risk may diminish already
d1im economic prospects.
The study of fiscal risks held by government is in its infancy. Build-
ing on work done by Hana Polackova Brixi and others, this chapter
discusses means by which national budgets might be transformed into
more effective instruments to control fiscal risk (Polackova 1998a).
79



80                                            ALLEN SCHICK
The first section briefly outlines alternative approaches for dealing
with fiscal risk, and the two sections that follow examine alternative
approaches for incorporating risk into financial statements and bud-
get decisions. The final sections consider options for controlling fiscal
risks through market-type arrangements.
Approaches to Managing Fiscal Risk
Governments determined to manage the risks they take can choose
from a variety of approaches. Some have been tried by a few govern-
ments, some by none. This section clarifies the various approaches,
which are elaborated later in this chapter.
The first approach is for government to be open about the types of
risks it faces, the volume and possible costs of these liabilities, and the
probability that various commitments will come due. This approach is
in line with the contemporary drive for transparency in fiscal matters,
but it requires distinguishing between explicit and implicit risks.
The second approach is to incorporate decisions on risks into the
ongoing budget process, thereby enabling the government to compare
direct and contingent expenditures without biasing the outcome in
favor of one or another type of transaction. Not all risks can be man-
aged through the budget, however. The more direct and explicit the
risk, the greater is the suitability of budgeting for estimating the cost
to government and setting aside resources for this purpose.
Third, government can manage risk by limiting risks before they
are taken. This approach would entail establishing criteria for deter-
mining whether the government should issue guarantees or enter into
other contingent commitments, assessing the degree of risk in the light
of these criteria, and refusing to take on risks that do not meet the
government's standards.
Finally, government may rely on market-type mechanisms to shift
all or a portion of the risk to private entities. Some of the innovations
in this area have had little or no application in government. Although
they are commonly used by commercial enterprises, the aim here is
not to recommend particular reforms but to stimulate innovative think-
ing on how governments might come to grips with practices and con-
ditions that may jeopardize their future fiscal health. Table 3.1
summarizes features of the four approaches.
The four approaches are differentiated not only in method but in
objective as well. This first is based on the notion that government
should be informed before it takes on new risk. The second empha-
sizes budget neutrality-the rules of budgetary accounting should not
bias policy in favor of any particular instrument. The third approach
rests on the notion that government should be risk-averse and should



BUDGETING FOR FISCAL RISK                                       81
Table 3.1 Comparison of Four Approaches to Managing
Fiscal Risk
Approach         Main objective             Limitation(s)
1. Reporting   Transparency: fuller  Many governments do not have
on financial  account of financial  accurate, comprehensive financial
statements   condition and risks.  statements. These statements can-
not cover implicit risks or risks with
low probability. Just publishing
statements does not itself change
risk-taking behavior by government.
2. Cost-based  Budget allocations   Costing methodology is not well
budgeting    reflect prospective  developed, and cost estimates may
cost to government;  be unreliable. Does not cover
risk-taking competes  implicit risks. May be treated as a
with other claims    technical exercise rather than
on budget.           as a real allocation of resources.
3. Rules for   Applies to guarantees  Political pressure may override the
taking fis-  and other contingent  criteria. Very few governments
cal risks   liabilities; criteria  have rules determining when to
applied before risk  enter contingent liabilities.
is taken.
4. Market-type  Relies on market to  Governments typically take on
arrangements reduce risk held by  contingent liabilities because of a
government and to    decision not to rely on the market.
more accurately
estimate the cost of
risks taken by it.
Source: The author.
accept new risk only when stringent criteria have been fulfilled. The
fourth approach takes the view that in the best of circumstances gov-
ernment is inherently a poor assessor and regulator of risk, and so it
should turn these tasks over to the market.
Transparency in Reporting on Fiscal Risk
]Liabilities and expectations, which are neither known nor recorded,
cannot be effectively controlled. An essential first step, then, in mak-
ing risks held by government transparent is to inventory the array of



82                                             ALLEN SCHICK
risks and liabilities borne by it. This is not an easy task, however, and
so the first effort may not yield a comprehensive or completely accu-
rate account. Several governments, including that of South Africa, have
used the Fiscal Risk Matrix presented in Chapter 1 to identify the risks
they are holding and the policy remedies that might be applied. In
South Africa, the official charged with this responsibility has indicated
that striving to fill in each of the boxes in the matrix made government
aware of significant risks that were previously unknown. Table 3.2
shows the results of this work. Note that for most of the entries the
government has been unable to estimate the downstream costs it might
face. Nevertheless, filling in the matrix has been a useful early step in
mapping out the government's exposure and response.
In compiling an inventory, it is necessary to canvass state entities
and programs in order to identify the agencies authorized to enter into
commitments, the transactions or conditions that have been insured,
the contingencies that would trigger government payments, and the
volume of outstanding liabilities. Although a comprehensive account
may be out of reach, government can identify most of its direct and
contingent obligations by concentrating on those sectors and programs
in which these risks typically occur: agriculture, housing finance, bank-
ing, small business, state enterprises, imports and exports, insurance
schemes, and infrastructure development.
There are two schools of thought on whether government should
acknowledge implicit liabilities. One urges transparency; the other
counsels fuzziness. Both aim to discourage moral hazard, but they dif-
fer on how this is best accomplished. The case for openness and disclo-
sure rests on the argument that moral hazard abates when the
government clearly and credibly signals how it will respond to pos-
sible failures and events-for example, whether it will indemnify un-
insured depositors or pay for essential services provided by bankrupt
municipalities. It might declare that depositors will be compensated
for losses only up to a certain amount or that it will finance certain
municipal services but not others. These signals, it is argued, will deter
affected parties from behaving in ways that add to the government's
potential liabilities. Of course, signaling its intentions is effective only
if the government holds a credible position and acts as promised. Wrong
signals provide added incentives for misbehavior.
It would be appropriate for the government to take further action
to discourage moral hazard when it makes implicit liabilities explicit.
For example, if the government steps in to pay for the ongoing services
of insolvent municipalities, it also should enact laws or regulations
that deter local governments from spending beyond their means, as
well as legislation extending bankruptcy rules and procedures to these
entities. If it accepts responsibility for expected (but not legally re-
quired) future pensions, the government should restructure the pension



BUDGETING FOR FISCAL RISK                                      83
Table 3.2 South African Policy Approach
Type of risks                        Policy approach
Explicit direct risks
Sovereign borrowing      Identify risks and formulate risk-averse strategy.
(ZAR380 billion)
Medical schemes          Adjust policy in budget.
Civil pensions           Adjust policy in budget.
Explicit contingent risks
Loan guarantees          Phase out guarantees.
(ZAR73 billion)        Revise authority to borrow and issue guarantees.
Cap borrowing authorities and approve and
coordinate borrowing strategies.
Guarantees on private    Share risk (contracts).
investment             Establish joint project limits.
Establish country limits.
Cap limits per institution.
State insurance schemes  Cap government risk exposure.
Share risks (also offshore).
Implicit direct risks
Socioeconomic expen-     Analyze policies.
diture                 Establish medium-term expenditure framework
also to reflect contingent liabilities.
Better reflect cost in annual budget.
Recurrent expenditure   Incorporate in fiscal planning and budgeting.
of public investment   Introduce "corporate governance" in projects.
(also state-owned
enterprises)
Implicit contingent risks
Default of subnationals  Monitor and introduce ex ante warning signals.
Systemic risks           Monitor.
Liabilities and risks of  Consider fiscal risks when restructuring.
policy failure from    Monitor.
privatization/commer-
cialization
Disaster relief/unavoidable Build contingency reserves.
expenditure            Establish contingent credit lines and purchase
reinsurance.
Monetary/exchange        Rethink interest rate and exchange rate policy
management              to contain government risk exposure.
Monitor central bank reserve management,
derivative use, and risk exposure.
Source: The author.



84                                             ALLEN SCHICK
system to put it on a sound financial basis. Without corrective mea-
sures, making implicit costs explicit would almost certainly worsen
the fiscal posture.
The counterargument is that the government should not divulge its
intentions on implicit risk because doing so would greatly increase moral
hazard. This is the position taken by the International Monetary Fund
(IMF) in statements elaborating on its new Code of Good Practices on
Fiscal Transparency (IMF 1999). The IMF code vigorously promotes
openness on fiscal matters, but it nevertheless recommends that "im-
plicit guarantees, such as the possibility that a government may in the
future bail out a public enterprise or private sector bank" be excluded
from statements on contingent liabilities (paragraph 67). The statement
adds that implicit guarantees should be excluded "because of the poten-
tial moral hazard to which being transparent about such provisions could
give rise" (emphasis in original).
When a government considers whether to make implicit liabilities
explicit, it must take account of the credibility of its position. More
than one government has announced that it will not pay depositor
claims on failed financial institutions only to be compelled by political
or economic circumstances to provide assistance.
Indeed, there can be no blanket rule for all implicit liabilities. In
some cases, maintaining a fuzzy position will shift a portion of the risk
to market actors if they bet wrong on what the government will do. In
other cases, fuzziness will leave the government with higher costs in
the end when it makes good on implicit commitments. Implicit pen-
sion liabilities should not be treated the same as implicit exchange rate
guarantees. For the former, the government might do well to recognize
the liabilities and to provide for them in the budget; for the latter, the
government might be better off keeping importers and exporters un-
certain about its intentions.
Accounting for Contingent Liabilities
Public accounting systems generally recognize direct liabilities, not
contingent ones. The Draft Guideline for Governmental Financial
Reporting issued by the International Federation of Accountants (IFAC)
in 1998 takes the position in paragraph 443 that "commitments and
contingencies are items which do not meet the definition and recogni-
tion criteria" for incorporation in financial statements (IFAC 1998).
These criteria define a liability as "a present obligation of the enter-
prise arising from past events, the settlement of which is expected to
result in an outflow . . . of resources." According to IFAC, a liability
should be recognized in financial statements when "it is probable that
an outflow of resources . . . will result from the settlement of a present
obligation and the amount at which the settlement will take place can



BUDGETING FOR FISCAL RISK                                  85
be measured reliably" (paragraph 443). Contingent liabilities do not
satisfy these criteria, because they depend on future rather than past
events and cannot be reliably measured. Nevertheless, the IMF Code
of Good Practices on Fiscal Transparency specifies in Statement 2.1.3.
that "[s]tatements should be published with the annual budget giving
a description of the nature and fiscal significance of contingent liabili-
ties, tax expenditures, and quasi-fiscal activities" (IMF 1999). (For a
more detailed discussion of accounting and reporting of contingent
liabilities and other fiscal risks, see Chapter 2 by Petrie in this volume.)
Measuring Risks
The IMF fiscal transparency code urges that each nation present in its
annual budget a statement of fiscal risks, including those deriving from
guarantees and insurance schemes, and that, where feasible, these risks
be quantified. Doing so is a challenging task, for risk assessment and
measurement are much more advanced in the business sector, where
various statistical tools and hedging strategies are widely used. Few
governments have had much experience in this area, but there is no
reason why they cannot adapt relevant commercial practices to their
needs. As governments gain experience in assessing risks, the quality
of their estimates is likely to improve.
Reporting the estimated costs of contingent liabilities and other fis-
cal risks in budgets or on financial statements may spur governments
to produce point estimates, which specify a definite cost but are al-
most always wrong and misleading. Few (if any) governments have
the capacity to measure accurately the probability that future contin-
gencies will come due and the cost they will incur if they have to make
payments pursuant to these obligations. Less than a decade after it
spent more than US$100 billion resolving widespread insolvency in
the banking sector, the U.S. government issued in fiscal 2000 a provi-
sional balance sheet estimating the present value of future deposit in-
surance liabilities at only $1 billion (United States, OMB 1999: Table
2-1). It may be that measures adopted to regulate the lending practices
of financial institutions and fees paid for deposit insurance will pro-
tect the U.S. government against future losses. But surely one cannot
rule out the possibility of widespread distress in the banking industry
or capital markets that would cause insured losses to escalate.
Ideally, risks should be estimated in terms of a range, with the key
assumptions and probabilities published alongside the estimates. In
addition, given the difficulty of estimating future costs, the following
suggestions may be useful for governments with limited capacity to
rnanage risks. First, concentrate on the riskiest endeavors, the ones
likely to account for most downstream liabilities. Second, make pre-
cise cost estimates only when warranted by experience and when the



86                                             ALLEN SCHICK
risks are pooled rather than concentrated. And, third, report fiscal risks,
even when it is not possible to quantify costs.
Formulating Financial Statements
Although IMF recommends that a statement on contingent liabilities
be included in a country's budget, there is a modest trend toward pub-
lishing them as notes to financial statements. Three contemporary de-
velopments have given new prominence to government financial
statements: (a) the shift underway from cash-based public accounting
to the accrual basis; (b) the growing reliance on financial statements to
report on a government's financial condition; and (c) the broadened
role of auditors in reviewing these financial statements and in assess-
ing the government's performance.
New Zealand pioneered in accrual accounting and budgeting a
decade ago when its Public Finance Act (1989) mandated that all gov-
ernment entities apply commercial accounting principles. The conver-
sion was swift and relatively painless and has led to the publication of
audited financial statements for all departments as well as a combined
financial statement for the government; a supporting schedule to this
statement lists both quantifiable and nonquantifiable contingent li-
abilities. Over the past decade, approximately a dozen national gov-
ernments have followed suit, and more are likely to join the movement
to the accrual basis under prodding from international organizations.
In commercial practices, contingent risks have an impact on the
balance sheet when provision is made for expected losses. To the ex-
tent, therefore, that losses on contingent liabilities can be measured, it
would be appropriate to make provision for them on the balance sheet.
But at this stage, cost estimates for these liabilities usually lack the
precision associated with recording direct liabilities. The most sen-
sible approach, therefore, may be to list contingent liabilities in the
notes, not in the financial statement.
Budgetary Neutrality
The rationale that has led some governments to introduce accrual ac-
counting also applies to the budget. Cost-based budgeting is designed
to make a government responsible for the resources it uses and strength-
ens the government's capacity to manage risks that are taken in cur-
rent budgets but payment for which emerges in subsequent fiscal years.
Polackova (1998b: 50) correctly argues that an "accrual-based account-
ing system without accrual budgeting is neither necessary nor suffi-
cient to ensure adequate policy consideration for contingent liabilities
and other fiscal risks." But there are several impediments to budgeting



BUDGETING FOR FISCAL RISK                                  87
for contingencies on a cost basis. One is that the development of ac-
counting standards for government budgets is still in its infancy. An-
other, already noted, is that cost estimates tend to be less reliable for
contingent rather than direct liabilities. As long as cost is based on
budget estimates rather than market prices, combining direct and con-
tingent expenditures in a single cost measure may be highly mislead-
ing as to the resources actually expended or risked by government. A
third issue is that the time horizon of budgeting may be too short for
allocating costs arising from contingent liabilities. Even in countries
that have a medium-term timeframe, budget projections typically ex-
tend out only three to five years. Fiscal risks often spill over well be-
yond this period.
There are four basic approaches to budgeting for contingent liabili-
ties. One is to present background information on contingent liabili-
ties and other financial risks in the budget, but to make budget decisions
only for direct expenditures and for payments pursuant to existing
commitments. Another is to devise a separate budget for contingent
liabilities and risks. The third is to integrate direct and contingent
liabilities on a cash basis. And the fourth option is to integrate the two
types of liabilities on a cost basis. The first three options are discussed
briefly, and the final one is considered in the light of novel U.S. bud-
getary procedures for direct and guaranteed loans.
Presenting Information in the Budget
The first option is to publish information on contingent liabilities and
risks in the supporting schedules, but not in the budget estimates. The
inclusion of supplementary information is a common practice in bud-
geting. Many national budgets provide background data on the
economy, grants to subnational governments, programs and activities,
and other matters. Although the supplementary information is not
voted, it assists parliament, attentive groups, and the public in assess-
ing the government's budgetary intentions. Similarly, background in-
formation could be provided on contingent liabilities, but these data
would not be combined with the estimates. For most governments,
providing such information would be a significant advance, even if the
budget listed only the various risks and liabilities but did not provide
cost estimates.
Parallel Budgeting for Contingent Liabilities
Alternatively, the government would compile a parallel budget for
contingent liabilities and related risks. Although separate from the regu-
lar budget, the "contingent liabilities" budget would be voted by par-
liament. This parallel budget would, like a regular budget, specify all



88                                             ALLEN SCHICK
the commitments authorized for the fiscal year; it might also limit the
amount of contingent liabilities outstanding and set aside cash resources
for expected calls on contingent liabilities during the year.
During the 1980s, a parallel budget system was introduced by the
U.S. government for direct and guaranteed loans. This parallel budget
set a total limit on the amount of new loan commitments and allo-
cated this total among particular programs and agencies. This separate
system was replaced in the early 1990s by the integrated arrangement
described below. This U.S. system is limited to risk associated with
direct and guaranteed loans; it is not applied to other contingent
liabilities.
Clearly, few governments have sufficient information to compile a
comprehensive budget for all contingent liabilities. It may be appro-
priate, therefore, to budget only for those contingencies for which rea-
sonably reliable data are available. For example, government may have
reasonably accurate and complete information on the loans it guaran-
tees, but know little about the debt incurred or guaranteed by state-
owned enterprises. In such a situation, it would be prudent for the
budget to cover only the former, even though the latter may pose greater
risk. Of course, government should endeavor to progressively improve
its coverage of fiscal risk by investigating areas where information is
meager but the risk of loss is great.
An alternative approach would be to budget for changes in the vol-
ume of known contingent liabilities and to concentrate on those pro-
spective losses that can be reasonably estimated rather than all such
liabilities. To do this, the government might construct a baseline (simi-
lar to the baselines used for expenditure projections in medium-term
frameworks) that would estimate future payouts for previously autho-
rized or outstanding contingent liabilities. Each year the baseline would
be adjusted for changes in projected payouts stemming from new gov-
ernment actions (such as the issuance of additional guarantees, changes
in economic conditions affecting the probability that the government
will have to make future payments, and reestimates of the losses ex-
pected from existing contingent liabilities). The baseline data would
be presented for government decision. Each year the government would
decide on new explicit contingent liabilities by adjusting the baseline.
But the government would not control total contingent liabilities; in-
stead, it would focus on year-to-year changes in estimated payouts for
those risks included in the baseline.
Integrating Contingent Liabilities into the Cash-based Budget
The third option would go further and combine payments on contin-
gent liabilities with the conventional cash-based budget. In this ar-
rangement, the government would set aside resources in the budget to



BUDGETING FOR FISCAL RISK                                 89
pay for losses expected during the year or over the medium term. It
also would use the budget to regulate the total volume of guarantees
or the amount of new guarantees to be issued during the fiscal year.
A comparison of the approach initially taken by the Netherlands
and that applied by Hungary shoiws various ways of integrating guar-
antees with direct expenditures. When it initially incorporated guar-
antees in the budget, the Netherlands recorded total new guarantees
as expenditures. Thus within a fixed budget constraint, the issuance
(or authorization) of new guarantees crowded out an equivalent amount
of direct expenditure. In effect, the budget made provision for the full
value of guarantees, not for expected payouts. Clearly, this treatment
was intended to discourage the issuance of guarantees in lieu of con-
ventional grants and subsidies. Once this practice was developed, the
Netherlands shifted to a "cost" basis that budgeted for estimated
payouts.
The government of Hungary has adopted seven interlocking budget
controls on guarantees (see Chapter 9 by Brixi, Schick, and Zlaoui in
this volume). First, the volume ol: guarantees authorized in the budget
is limited to a certain percentage of state revenues. Second, the volume
of outstanding guarantees issued by various state entities (such as the
Hungarian Development Bank) is limited by law. Third, the budget
sets aside funds for payments expected to be made during the fiscal
year pursuant to existing guarantees. Fourth, guarantee contracts are
reviewed by the Ministry of Finance, which closely monitors the issu-
ance of guarantees by departments and other entities. Fifth, material
information on new guarantees (such as the amount, conditions, justi-
fication, lender, and borrower) are published in a government resolu-
tion. Sixth, the annual budget reports the probability of default and
expected payments on each guarantee program. Finally, the issuance
of guarantees is reported to the State Audit Office.
Although the system is not perfect-for example, some major guar-
antees are exempted from the limit on the total volume issued each
year-it has greatly reduced the exposure of government to losses from
contingent liabilities. In fact, disbursements for calls on defaulted guar-
antees have been below the amounts set aside in the budget for this
purpose. The main problem for Hungary has been dealing with im-
plicit contingent liabilities. These are not covered by the new budget
control system, and payments to cover these losses have far exceeded
those made for explicit liabilities.
Budgeting for the Cost of Contingent Liabilities
Budgeting for contingent liabilities on a cash basis may result in either
too much or too little control of the government's risk of loss. The
early approach taken by the Netherlands overstated the cost, because



90                                             ALLEN SCHICK
the government was likely to be liable for only a fraction of the losses
it was guaranteeing. By contrast, the Hungarian method may under-
state cost, because the government will have to make good on its com-
mitments even if the amounts provisioned in the budget are inadequate.
Moreover, when contingent liabilities are budgeted on a volume
and cash basis, politicians may have an incentive to substitute them
for grants and other disbursements. While cash payments are bud-
geted as outlays in the year they are made, payments for guarantees do
not appear as outlays until later years, when default or other events
occur. On a cash basis, guarantees are inexpensive relative to grants,
even though they may cost more in the long run.
To deal with this problem, the U.S. government introduced new
budgetary rules for direct and guaranteed loans in 1992. These rules
are designed to neutralize budgetary incentives and to make politi-
cians indifferent to whether they choose grants, direct loans, or guar-
antees. These three types of transactions are budgeted on a cost basis,
rendering the timing of cash payments less relevant in allocating gov-
ernment resources. The Netherlands now uses a similar system.
The current U.S. system, which has not been altered since its intro-
duction in 1992, shifts the budgetary basis of loans and guarantees
from cash flow to subsidy cost. This cost is defined in law as "the
estimated long-term cost to the Government of a direct loan or a loan
guarantee, calculated on a net present value basis, excluding adminis-
trative costs." Net present value is calculated by discounting estimated
future cash outflows (loan disbursements and payments on defaults)
and inflows (origination fees, repayment of principal and interest on
direct loans, and recoveries), using a discount rate equal to the interest
rate paid by the U.S. government on borrowings of a comparable ma-
turity. A separate appropriation is made for the projected subsidy cost
of each loan program; this appropriation is included in the budget as
an outlay, even though money might not be disbursed until years later.
Subsidy costs are included in the computation of total budget expendi-
tures and in the surplus or deficit.
Budgeting for direct and guaranteed loans entails complex proce-
dures for estimating subsidy costs and new accounting procedures for
recording the cash flows associated with loan transactions. The pro-
cess is designed to differentiate between the subsidized portion of loans
that is budgeted as a cost and the unsubsidized portion that is bud-
geted as a below-the-line transaction. Budget resources are provided
only for the subsidy cost, which almost always is significantly less
than the face value of the loan or guarantee. The subsidy cost of direct
loans is the present value of the amounts not repaid (minus fees and
recoveries) and the difference between the interest rate charged bor-
rowers and the cost of money to government. The subsidy cost of guar-



BUJDGETING FOR FISCAL RISK                                91
anteed loans is the present value of the difference between cash pay-
ments for defaults and cash received from fees and recoveries.
The subsidy cost is estimated at the time a direct loan is obligated
or a loan guarantee commitment is made. Actual loan performance,
however, often varies from early estimates, sometimes significantly.
Accordingly, the subsidy cost is annually reestimated during the life-
tirne of loans and guarantees, and an automatic appropriation is pro-
vided to cover overruns. Because this appropriation is automatic,
government agencies are not penalized in their budgets if they under-
estimate subsidy cost. Since fiscal 1992 when the credit system was
first implemented, underestimates h,ave been relatively minor, but it is
important to bear in mind that this entire period has been largely one
of robust economic growth in the UJnited States.
As noted, a key objective of the subsidy approach is to ensure that
budget decisions are neutral-that is, they are not skewed in favor of
or against any particular type of transaction. Budgeting for subsidy
cost puts direct loans, guaranteed loans, and grants on an equal basis.
All are budgeted in terms of cost to government rather than in terms of
cash exchanged. The hypothetical example below contracts the cash
basis and subsidy cost treatment of these transactions (in millions of
U.S. dollars).
Transaction              Amount budgeted
Type                Amount     Cash basis  Subsidy cost
Loan                 $100        $100         $15
Guaranteed loan        100          -2         30
Grant                  20          20          20
This hypothetical case compares a direct loan of $100 million, a
guaranteed loan of $100 million, and a grant of $20 million. On a
cash basis, the direct loan would appear in the budget as the most
costly transaction, because the entire amount disbursed is recorded as
an outlay. By contrast, the guaranteed loan would be budgeted as a
revenue gain for the government, because it receives income from origi-
nation fees in the year the guarantee is issued. On a subsidy cost basis,
however, budgeted outlays for the clirect loan would be reduced from
$100 million to $15 million, because projected repayments of princi-
pal and interest would be included in the measurement of subsidy costs.
Budget outlays for the guaranteed loan would rise from -$2 million to
$30 million, because projected defaults in later years would be in-
cluded in the subsidy cost, making it the costliest type of transaction.
The amount recorded for a grant would not change, but the cost basis



92                                             ALLEN SCHICK
would make it more readily comparable to direct and guaranteed loans.
Note, however, that the hypothetical amount shown here for a grant is
based on disbursements, while the cost reported for a loan and a guar-
antee is based on projections of future discounted cash flows.
Conversion to the subsidy cost basis entails maintaining separate
budgetary accounts for the subsidized and unsubsidized portions of
loans and guarantees. Program accounts receive appropriations for
subsidy costs; financing accounts handle the cash flows associated with
the nonsubsidized portion. Program accounts are included in the bud-
get; financing accounts are recorded as "means of financing" and their
cash flows are not included in budget receipts or outlays.
The subsidy cost basis is currently used in the United States only for
direct and guaranteed loans, not for other contingent liabilities. How-
ever, legislation tabled in the U.S. Congress in 1999 (but not enacted)
would have shifted all U.S. government insurance programs to this
basis (H.R. 853, sec. 604, 106th Cong., 1st sess., 1999). The legisla-
tion provided that beginning with fiscal 2006, insurance commitments
would be made only to the extent that budget resources were appro-
priated to cover their "risk-assumed cost." In the legislation, this cost
is defined as "the net present value of the estimated cash flows to and
from the Government resulting from an insurance commitment or
modification thereof." Inasmuch as the volume of insurance commit-
ments is many times greater than that of loan guarantees, enactment
of this legislation might have an enormous impact on the budgetary
treatment of contingent liabilities. Although no action has been taken,
this type of proposal is likely to be revived in the future.
Criteria for Contingent Liabilities
The two broad approaches discussed thus far-accounting and budget-
ing for fiscal risks-would significantly enhance government's under-
standing of the contingent liabilities it faces. But they do not deal with
the critical issue in risk management: whether government should com-
mit itself to the contingent liability in the first instance. In the business
sector, financial institutions rigorously partition decisions on risk from
risk assessment. One decisionmaking entity is responsible for negotiat-
ing insurance contracts, loans, and other commitments; the other as-
sesses the credit-worthiness of borrowers, the risk inherent in the activity,
and other risk factors. Moreover, the institution would finalize its com-
mitment only if the risk factors were assessed to be within acceptable
parameters. In government, however, risk commitment and risk assess-
ment are often done by the same entity. In the United States, for ex-
ample, the government agency that guarantees loans also estimates the
subsidy cost by assessing the probability of default.



BUDGETING FOR FISCAL RISK                                 93
C(anadian Principles for Regulating Risk
The best time to manage risk is before the commitment is made; after-
ward, all government can do is to account for the risk it has under-
taken in financial statements or the budget and to manage its portfolio
of liabilities in ways that mitigate losses. The government of Canada
introduced a set of principles in the mid-1980s to regulate the risks it
takes on loans and loan guarantees. For this discussion, the most sa-
lient principles are the following:
* In the case of loans, any concessional terms, such as a below-
market interest rate, are treated as budgetary expenditures. In some
cases, the subsidy is so high that the entire loan is budgeted as an
expenditure.
* Before a loan or guarantee is tendered, the sponsoring depart-
ment must analyze the project and demonstrate that it cannot be fi-
nanced without government assistance, and that cash flow will be
adequate to cover repayment of the debt as well as interest and operat-
ing costs and yield a satisfactory rate of return.
* Risk must be shared with private equity sponsors who supply a
substantial portion of the required funds from their own resources.
Moreover, guarantees must provide that in the event of default, the
government shall recover its losses from private equity sponsors.
* Bankers should share risk by bearing a minimum of 15 percent of
the net loss associated with any clefault. This arrangement would give
them an incentive to undertake a rigorous assessment of their risk
exposure.
* Interest rates on loans should be set to cover the government's
cost of money and estimated future losses on loan guarantees. Fees
should be imposed to recover estimated future losses and to defray
administrative expenses.
* Provision should be made fo:r loans and guarantees at the time they
are issued. The amount provisioned should be based on an assessment
of risk, and sponsoring departments must pay for these provisions out of
fees earned in issuing guarantees or their annual appropriations.
* New loans and loan guarantee programs must be approved by
the minister of finance and authorized by Parliament.
* Departments and Crown corporations are required to report on
their contingent liabilities. These reports are published as notes to the
government's annual financial statement. Moreover, estimates of con-
tingent liabilities and losses are audited by the auditor general who
reports directly to Parliament.
Certain exceptions to these rules exist, but the overall effect has
been to compel government to consider its risk exposure before guar-
anteeing loans.



94                                            ALLEN SCHICK
Limiting the Government's Liability to the
Amount Provisioned for Losses
The U.S. and Canadian methods for budgeting for fiscal risks rely on
loss estimates made at the time loans are issued or guaranteed. Inasmuch
as future defaults cannot be known perfectly ex ante, it is possible they
will be underestimated. If this occurs, the amount appropriated for sub-
sidy cost in the United States or the amount provisioned for losses in
Canada will be immediate. In fact, politicians may have a strong incen-
tive to underestimate risks, especially in the American system where
an automatic appropriation is available to cover unbudgeted losses.
They do not pay a budget penalty when costs are underestimated.
In Chapter 6 of this volume, Daniel Cohen has proposed a novel
scheme to impel government to disclose the true risk deriving from
guarantees and other contingent liabilities. He would have the gov-
ernment provision for projected losses by setting aside money equal
to its estimated liability in a reserve fund. Once this money is re-
served, the government would have no further liability. All claims
against guarantees would be paid by the reserve fund, not by govern-
ment. Most important, claims would be limited to the resources avail-
able in the reserve fund; if the reserve fund were depleted, no further
claims would be paid.
This arrangement would provide lenders and others seeking gov-
ernment guarantees with a strong incentive to demand that risk be
accurately assessed. An underestimate would devalue guarantees and
shift risk from the government's reserve fund to private parties. Cohen
also has suggested that the reserve fund be privatized, with its shares
traded publicly. The share price would reflect the market's assessment
of whether adequate provision has been made for the risk insured by
the reserve fund.
The Cohen proposal has not been implemented by any government,
though doing so would not be a difficult technical feat. It is not cer-
tain, however, that his mechanism would effectively constrain the
government's liability to the amount provisioned. Ex ante valuation of
risk is inherently inaccurate, and barring government from adjusting
the amount provisioned as it gains additional information may be im-
practical. Firms often adjust the amount provisioned for bad debt and
other losses in response to new information, and there is little reason
to expect the government to behave otherwise. One can foresee enor-
mous political pressure on a government to bail out underfunded re-
serve funds; in fact, it is likely to be blamed for the underfunding. Of
course, if government has the option of replenishing the reserve fund,
it would lose much of the incentive to provision adequately for losses.



BUDGETING FOR FISCAL RISK                                   95
Even if it were to curtail explicit risk, the Cohen arrangement would
not curtail the government's exposure to implicit contingent liabilities.
In these cases, there may be strong jpressure on the government to com-
pensate for losses even if it were not: legally obligated to do so, and even
if the reserve fund lacked sufficient resources to cover these claims.
Using the Market: to Regulate Risk
An alternative to the Cohen plan would be to allow the market to
assess and allocate risk rather thati the government. The main advan-
tages to bringing the market into play would be less political oppor-
tunism, diversified risk, and lower government losses. Every risk insured
by government can be insured cormmercially at some cost, or not at all
if: the probability of loss were so high that no private insurer would be
willing to take it. When government replaces the market, it shifts the
cost to itself. Understandably, private risk takers actively search for
opportunities to shift the cost to government, and they often have little
difficulty recruiting politicians to go along. Any market-type remedy,
therefore, must return all or a portion of the cost to private risk-takers.
The easiest way to accomplish this would be for the government to
refrain from tendering guarantees: another would be to adopt the rig-
orous screening criteria applied by Canada. But assuming the govern-
rnent was bent on accepting risk, it might take several measures to
reduce its exposure.
Risk-sharing
One simple way for government to reduce risk is to share it with lend-
ers, borrowers, importers/exporters, enterprises, or others seeking guar-
antees. If the government were to enforce a rule that it would assume
no more than 50 percent of the risk, the private parties holding the
other half (or more) of the risk would have to think about their own
exposure before proceeding with the transaction. A bank would have
to consider the credit-worthiness of borrowers; it would no longer
suffice to care only about the quality of the guarantee. Some guaran-
teed transactions might still proceed, but others surely would be aborted
if the government limits its liabiliity.
In a variation on this approach, the government would insure the
[ast rather than the first loss. Risk-sharing would be promoted by the
use of high deductibles, which would require the insured party to pay
for the loss up to a certain monetary value. Government liability would
take effect only after the deductible is satisfied.



96                                             ALLEN SCHICK
Risk-based Premiums
One of the anomalies of risk-taking is that government often charges
less for its riskiest guarantees. A well-established enterprise might be
charged market interest rates or an initiation fee; startup ventures
might receive concessional interest rates and have the fee waived. As
perverse as this seems, there is a certain political-economic logic to
this behavior. Risky borrowers, the argument runs, need government
assistance because they have no recourse to private markets. They
cannot afford to pay up-front fees or at-market interest rates. There-
fore, the government should assist them by forgoing these charges or
offering concessionary terms. In effect, the guarantee serves as a sub-
sidy. Arguably, guarantees are an inefficient form of subsidy. Assis-
tance might be better provided through grants rather than contingent
liabilities that mask the true cost of government. On this basis, gov-
ernment would do well to charge risk-adjusted premiums for its guar-
antees. Some lenders and borrowers, importers and exporters, and
other risk-takers would be deterred by high premiums, thereby re-
ducing government's exposure.
Reinsurance of Government Risk
One of the most common means used in the private sector to limit
liability is the purchase of reinsurance. This practice is rarely applied
in government, however. There is no technical impediment to govern-
ment purchasing reinsurance when it guarantees loans, agricultural
prices, or any other event or outcome. As in markets, the amount paid
by government would reflect the risk it has taken. The cost of reinsur-
ance not only would give government a powerful signal about the risk
it is holding, but also might dissuade it from taking risks that would
require very high reinsurance premiums. In other words, the practice
of immediately reinsuring itself would deter government from under-
writing transactions adjudged by the market to have the highest prob-
ability of loss.
The reinsurance model can accommodate several variations. For
example, government could reinsure only a portion of risk, or it could
base premiums on the cost of reinsurance. In both cases, reinsurance
would be a means of promoting risk-sharing.
Conclusion
Risk management is still in its infancy in public finance. Governments
have little of the experience and few of the instruments used by firms
and markets to assess and control risk. Some techniques perfected in



BUDGETING FOR FISCAL RISK                                      97
the private sector may be adapted for government use in the years
ahead, and some of the cutting-edge practices introduced by Canada
and the United States might be tried in other countries. But dealing
with contingent liabilities is not easy in any country and is especially
difficult in developing and some transitional countries where the mar-
kets and insurance sectors are relatively undeveloped and where the
nmargin for error is narrow.
Some critics have urged that the best posture is for governments to
take few risks. Clearly, risk-sharing can be more broadly applied by
insuring only a portion of possible loss, levying risk-adjusting premi-
ums, and purchasing reinsurance. But in modern times, even well-
managed governments in sturdy economies are called on to accept
risks that in an earlier age might have been held by the household or
enterprise. Many positive things, including economic improvement,
have happened because governments have taken fiscal risks. It is pre-
cisely because governments will continue to expose themselves to vari-
ous contingencies that they should be more transparent about the risk
they face, more willing to make provision for these risks in their bud-
gets, and more insistent on sharing the risk with others.
References
IFAC (International Federation of Accountants), Public Sector Committee. 1998.
Draft Guideline for Governmental Financial Reporting. New York.
IMF (International Monetary Fund), Fiscal Affairs Department. 1999. "Manual
on Fiscal Transparency." Washington, D.C.
Polackova, Hana. 1998a. "Contingen-t Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
. 1998b. "Government Contingent Liabilities: A Hidden Risk to Fiscal
Stability-A Consideration for EU Accession." In European Commission,
European Union Accession: The Challenges for Public Liability Manage-
ment in Central Europe. Washington, D.C.: World Bank.
United States, OMB (Office of Management and Budget).1999. Analytical Per-
spectives, Budget of the United States Government, Fiscal Year 2000. Wash-
ington, D.C.: Government Printing Office.






CHAPI'ER 4
Institutional and Analytical
Framework for Measuring and
Managing Government
Contingent Liabilities
Suresh M. Sundaresan
Graduate School of Business, Columbia University
THE MANAGEMENT OF RISK in tthe process of earning an attractive
return on the capital employed is a challenge that confronts many
organizations in the private and public sectors. For private corpora-
tions, which are widely regarded to be profit-maximizing (or value-
maximizing) entities, this task has assumed paramount importance
with the increased level of competition, globalization, and securitization
of the markets where risks are priced and traded. The institutional and
analytical frameworks used to manage and measure the risks of corpo-
rations and set the right incentives for managers have been studied
extensively. As for the public sector, economists have widely investi-
gated the problems associated there with the measurement of deficits
and contingent liabilities, but no operational guidelines have emerged
on how contingent liabilities should be reported in any measure of
government deficit.
The measurement of risk and its accurate reporting in the calcula-
tion of federal, state, and city deficits are important to many organiza-
tions such as the International Monetary Fund (IMF), World Bank,
lenders in the private sector, and ultimately citizens of different gen-
erations. Fiscal liabilities tremendiously influence the deficit and thus
the actual risk and opportunities faced by a country. An example of a
liability that may result in intergenerational transfers is the social
99



100                                  SURESH M. SUNDARESAN
security system and the aging of the population. The percentage of
retired people in the population is increasing in many countries. This
increase will lead to a greater burden on the working population even
when other factors are held constant.
This chapter reviews the literature on measurement of fiscal defi-
cits, government liabilities, and the analytical tools for measuring gov-
ernment contingent liabilities. It also discusses the merits of measuring
a liability accurately in the context of the fact that governments typi-
cally have much discretion to legislate actions that can drastically af-
fect the future value of its liabilities (which cannot be predicted ahead).
Despite valid concerns about relying on a single measure of deficit, I
argue that an accurate measure of liability is very important to lending
institutions and legislative bodies. In this context, I argue that the valu-
ation of contingent liabilities in reporting the fiscal risk of govern-
ments is qualitatively different from a standard loan guarantee (put
option-pricing) problem. This is rooted in the observation that the
presence of a guarantee (an example of a contingent liability of the
government) alters the stochastic process of the variable on which the
guarantee is provided in the first place.
An example should amplify this point. Suppose the government
decides to guarantee loans of private sector companies in an infra-
structure area such as power. This guarantee covers the repayment of
interest and principal, which may be denominated in a foreign cur-
rency. The presence of the guarantee results in an excess supply of
power, leading to a supply response. In anticipation of this supply
response, the consumer demand for power-intensive goods also goes
up, leading to a demand response. In equilibrium, this response will
affect the price of power at which the demand is equal to supply.
This situation makes the valuation of loan guarantees a challenging
problem.
This chapter is organized as follows. The next section reviews the
economics literature on deficits as well as the operational concerns
of development institutions. It is followed by a description of the
institutional features that monitor the risks taken by corporations.
The next section contrasts the institutional arrangements that gov-
ernments face in addressing the risk management problem. It also
explores the differences between the objectives of a corporation and
of a government and how such differences may affect their propen-
sity toward risk-bearing. A description of the tools of risk manage-
ment used in the private sector follows, along with my argument that
many of these tools need to be recast in a fundamental way to be
relevant to managing the risks faced by governments. The final sec-
tion outlines a model for measuring the exposure of a government's
contingent liabilities. Most of the technical arguments are presented
in the Appendix.



MEASURING AND MANAGING CONTINGENT LIABILITIES             101
Economic Theory on Deficits
Economic theory has much to say about the measurement of deficits,
the relevance of any deficit measures, and the role of deficits in the
welfare of the economy. This section briefly reviews the literature.
Accounting, Reporting, and Policy Flexibilities
The construction of an ideal balance sheet for a country has engaged
the attention of economists for over 20 years. Papers by Buiter (1983)
and Bean and Buiter (1987) have attempted to lay down some concep-
tual groundwork to address this issue. The conceptual and method-
ological issues in the measurement of fiscal deficits, which are materially
affected by the way in which the liabilities are measured and reported,
have been discussed in an extensive review paper by Blejer and Cheasty
(1991). At the heart of measuring liabilities is the issue of measuring
revenues and expenditures of governments, on the one hand, and fi-
nancing, on the other. Blejer and Cheasty (1991) point out that "the
government debt criterion" assumes that if a transaction extinguishes
a liability or creates a liability, then it is considered as financing. Un-
der "the public policy criterion," t[ransactions that further the goals of
policymakers are classified as taxing and spending. Depending on the
doctrine used, then, the deficit measures may mean very different things.
Another item that introduces a variation in the deficit measure is the
choice between cash accounting and accrual accounting. Under strict
cash deficit accounting, only those government outlays for which cash
has been distributed within one year is part of the budget balance.
Likewise, only those actual cash revenues received within the year go
to the budget balance. In the accrual measure of deficits, the actual net
resource preemption is accounted for regardless of the cash flows dur-
ing the year. Depending on the concept used (which may be closer to
either the cash or the accrual concept), the measure of deficit will
represent different things.
The issues surrounding the measurement of deficit and their impli-
cations have been articulated in papers by Eisner (1984) and Eisner
and Pieper (1984). An insight that emerges from this strand of litera-
ture is that a single measure of deficit may never be adequate for pub-
lic policy debate or cross-country comparisons. Moreover, deficits can
display seasonal patterns, may vary with the growth rate of the economy,
and are easily changed through legislative actions and tax policies.
Eisner (1984) notes that the valuation of a government's contingent
liabilities in the calculation of deficits is subject to the criticism that
the government can legislate actions that may seriously change the
future value of its contingent liabilities. Inflation also can significantly
affect the reported budget deficit in a significant manner. Finally, when



102                                  SURESH M. SUNDARESAN
the economy is growing and undergoing structural changes, any mea-
sure of deficit is likely to be not very informative. Thus the measure of
deficit should control for factors such as seasonality, the stage of the
economy, and inflation levels.
Ricardian View of Deficit Measurement
The standard approach to budget deficits makes the following case: if
the government borrows money (and effectively reduces taxes), it may
promote aggregate demand by the consumers. An implication is that
the private savings will be lower than the implied tax cut. This will
call for an increase in the real rate of interest to restore savings. But
the higher real rate will crowd out investment, leading to a lower stock
of capital. In this sense, a budget deficit induced by borrowing is a
burden, especially for future generations who will face a diminished
stock of capital.
At the heart of any discussions of government liabilities and defi-
cits lies the Ricardian equivalence theorem, which argues for the per-
fect substitutability of tax and debt financing under some simplifying
assumptions. In a seminal contribution, Barro (1974) argues that
intergenerational altruism is a key factor in restoring the neutrality
of fiscal policies. The present value of government expenditures
(broadly defined to include hidden contingent liabilities) must equal
the present value of tax revenues, because the government budget
constraint must hold intertemporally. It then follows that the present
value of taxes cannot change unless the present value of all expendi-
tures also changes. In short, any deficit-induced tax cut must result
in a future tax increase with the same present value. An important
implication of the Ricardian equivalence theorem is that any current
budget deficit leads to an increase in private savings, which exactly
offsets the decrease in the government's savings. This has important
implications for the role of deficits and their measurement. Under
the Ricardian view, an accurate measurement of a deficit may help
forecast future taxation policies. (For discussion, also see Barro 1979,
1989; Bean and Buiter 1987; Bernheim 1987; Bernheim and Bagwell
1988; Abel and Bernheim 1991.)
Corporations have the choice of issuing debt or equity, and the
Modigliani-Miller theorem sets the benchmark for any debate on
whether a specific combination of debt and equity is value-maximizing
from the perspective of the corporation. The government can run a
deficit or cut taxes, and the Ricardian equivalence theorem defines the
circumstances under which fiscal policy choices matter. Because de-
viations from the Ricardian equivalence result may be expected, whether
there is a budget deficit or a tax cut may no longer be a matter of
indifference.



MEASURING AND MANAGING CONTINGENT LIABILITIES           103
Effects of Government Guarantees
The volume of government contingent liabilities is of concern to pri-
vate sector lenders, the IMF, the World Bank, and future generations
of citizens. Loan guarantees, which are an example of a state contin-
gent liability, may lead to moral hazard and excessive risk-taking and
an oversupply of loans. The impact: of government guarantees on for-
eign investments was examined by Wu (1950), who argued that the
threat of expropriation and nonconvertibility of currency earnings were
the biggest risks of concern to the guaranteeing institutions. The possi-
bility that government actions may affect the allocations of resources
and the response by private sector entities has been well recognized in
the literature. Brock (1992) points out that government guarantees on
foreign loans have led to investment booms that have affected the level
and the path of the outputs in the economy whose risks are being
measured. Brock (1992) also notes that "loan guarantees may also
distort an economy's macroeconomic adjustment by permitting a post-
ponement of the liquidation process." He presents case studies from
Chile and Texas to show that sorme delays in the closure of insolvent
financial institutions have had disastrous macroeconomic consequences.
A similar argument has been advanced by Kryzanowski and Roberts
(1993) to suggest that the absence of bank runs in the Canadian bank-
ing system may not be due to the bank branching system but to "the
forbearance of regulators coupled with an implicit guarantee of all
deposits." The fact that loan guarantees and subsidies lead to actions
by entities in the private sector h,as been well articulated by econo-
mists. Insightful papers by Townsend (1977) and Salant (1983) note
how price-fixing schemes are doorned to fail and are prone to specula-
tive attack by agents in the econormy. Bardsley (1994), who examined
the collapse of Australia's Wool Reserve Price Scheme, points out that
price-fixing schemes can fail even when they are not backed by a fixed,
exogenously set financial limit as assumed by Townsend (1977). It is
clear that lenders would like to have accurate information about what
the valuations of contingent liabilities are even if the government is
able to pursue tax and other fiscal policies in the future to dramati-
cally change the future valuation of these liabilities. The point is that
any current valuation of such liabilities must be viewed within the
context of dynamic actions that can be taken by the government to
significantly affect the valuation. With this caveat in mind, we turn to
some important papers in the realm of valuing contingent liabilities.
Valuation of Loan Guarantees
The presence of contingent liabilities in the private sector and their ef-
fects have been studied in a variety of different contexts. In a pioneering



104                                  SURESH M. SUNDARESAN
paper, Merton (1977) investigated the effect of deposit insurance and
its valuation. His insightful analysis showed that loan guarantees are
basically put options written on the underlying assets backing the loans.
Later Merton (1978) extended his analysis to value deposit insurance
when there are costs of surveillance. In a related contribution,
Borensztein and Pennacchi (1990) examine the valuation of interest
payment guarantees on debt issued by a developing country. They ex-
ploit the market prices of debt in the secondary market to infer an
unobservable state variable whose realizations depend on whether there
will be contractual debt service or default. Borensztein and Pennacchi
assume that the stochastic process followed by the state variable is
unaffected by the presence of a guarantee.
A direct application of the insights of such papers to the measure-
ment of the contingent liabilities of government is precluded for sev-
eral reasons. First, the announcement of a guarantee produces a supply
response. For example, government loan guarantees end up increasing
the stock of debt because of the supply response. This is not modeled
in option pricing. Second, and perhaps more important, because of
this supply response, the level and the path of outputs and prices change,
leading to an endogenous shift in the underlying path of the economy.
This implies that the standard option-pricing paradigm, which assumes
an exogenous stochastic process for the underlying assets that is in-
variant to the presence of the guarantees, cannot be applied in a "cook-
book" style to valuing the government's contingent liability. The moral
hazard problems associated with price subsidies and guarantees make
the valuation much more difficult.
Summary of Insights from the Economics Literature
In summarizing the insights of the economics literature, I argue for the
following priorities in research on fiscal risk measurement in general
and the valuation of contingent liabilities in particular:
* The neutrality of fiscal policy under some conditions implies that
a dollar of debt financing is no different from a dollar raised through
taxes. Although this may not strictly hold, it forms a useful theoretical
basis for examining how one should assess a federal deficit. The abil-
ity of the government to legislate future actions that may significantly
alter the current valuations of contingent liabilities makes any isolated
valuation measure less relevant for policy debates.
* Government actions such as subsidies and loan guarantees lead
to private actions and responses that may affect the level and path of
the outputs of the economy. They may induce moral hazard-for ex-
ample, a government guarantee on debt issued by entities in the pri-
vate sector may reduce the incentive of these entities to stay current in



MEASURING AND MANAGING CONTINGENT LIABILITIES               105
their debt obligations. (But this is mitigated by the fact that private
sector issuers may care about their reputations if they are repeated
participants in credit markets.) Another possibility is that the guaran-
teeing entity is able to extract significant rents from the debtors be-
cause of the dire need to get the scarce capital. These observations
underscore the need to evaluate the costs of government actions in a
general equilibrium or a more inclusive setting than just viewing the
measurement of contingent liability as an isolated problem.
* Some government actions such as loan guarantees and price-fixing
schemes may lead to a harmful response by agents in the economy,
such as speculative attacks on buffer stocks and subsidy programs and
inefficient liquidation of insolvent firms in the presence of guarantees.
This observation leads to the possibility that the conventional mea-
sures of valuing such contingent liabilities may underestimate their
true costs.
* A corollary to this observation is that the valuation of contingent
liabilities cannot be conducted in isolation of their effects on the asset
side, as well as the liability side, of the government's balance sheet.
'rhis is particularly challenging because the present value of benefits
and other costs generated by government actions is often difficult to
quantify but nonetheless very important.
Operational Guidelines for Measuring Risk
Economic theory thus offers important guidelines in the measurement
and management of risks assumed by government's implicit and ex-
plicit guarantees and subsidies. T he focus of this literature is to pro-
vide a conceptual and a methodological basis for understanding and
reporting deficits. However, for policymakers and institutions such as
the World Bank and the IMF, which extend loans and assistance to
developing and underdeveloped economies, the issue of correctly mea-
suring the risks assumed by such economies is of practical and opera-
tional concern because it affects their lending and assistance policies.
A recent contribution by Brixi and Zlaoui (1999) and Chapter 3 by
Allen Schick and Chapter 9 by 1-iana Polackova Brixi, Allen Schick,
and Leila Zlaoui in this volume have explored risk measurement is-
sues applied to specific countries as well as to the general issue of
contingent and hidden government liabilities. The key contributions
in this strand of work can be surnmarized as follows:
* Practical measures of liabilities can be put in the context of a
Fiscal Risk Matrix (see Polackova 1998 and Chapter 1 of this volume).
Four types of risks are recognizecd in this framework: (a) direct explicit
liabilities, (b) direct implicit liabilities, (c) explicit contingent liabili-
ties, and (d) implicit contingent liabilities. Direct explicit liabilities



106                                   SURESH M. SUNDARESAN
are government obligations such as government employee wages that
are clearly identifiable and fall under the rubric of the national legal
framework. Direct implicit liabilities are future obligations of gov-
ernment that do not necessarily constitute legal obligations but rather
moral obligations. Explicit contingent liabilities cover state guaran-
tees on loans issued by nongovernment institutions. Finally, implicit
contingent liabilities include hurricane damage, earthquake damage,
and so forth.
a Specific proposals for budgetary controls as a means of control-
ling and managing fiscal risk include: (a) improved reporting of the
actual performance of programs under the state guarantees; (b) pro-
viding the right incentives by requiring significant risk-sharing-this is
similar to reinsurance contracts in which the reinsurer requires the
insurance company to take the first layer of damages before insuring a
prespecified second layer; and (c) modifying and supplanting the ac-
crual-based accounting procedures to shed light on hidden liabilities.
* Risk management tools that are already prevalent in the private
sector can be applied to the public sector, such as prudent provision-
ing, stress testing, and organizational allocation of risk management
responsibilities.
The challenge that remains in risk management is to improve the
practical task of risk reporting by more completely tapping into the
insights of economic theory that were summarized earlier. This chap-
ter attempts to do precisely that.
Institutional Framework for
Risk Measurement in the Private Sector
One place to look for some guidance on risk measurement and man-
agement is the private sector. Any firm wishing to actively manage risk
has to pass the hurdle that firms need not manage the risks that are
easily diversified by stockholders. The basis for this hurdle arises from
capital market theory and asset-pricing models in the finance litera-
ture. Unless active risk management results in a higher value to the
stockholders, there is no need for a firm to manage risk. Typically, two
factors make risk management by a firm value-maximizing. The first
is the cost of financial distress: if the firm's risk of financial distress is
increased as a consequence of not hedging market risks and the result-
ing process of financial reorganization or liquidation is costly, then
risk management adds value to the firm. Second is the underinvestment
problem: if the benefits of a project largely accrue to the bondholders,
then the firm may underinvest. This situation can be mitigated by hedg-



MEASURING AND MANAGING CONTINGENT LIABILITIES           107
ing under some circumstances. There also may be tax and regulatory-
related incentives for firms to hedge risk.
A fundamental aspect of corporate organization is that there are
markets in which the claims issued by corporations are actively traded.
Such markets include equity markets such as the New York Stock Ex-
change as well as bond markets in both exchanges and the dealer com-
munity. These claims are exchanged by sets of investors many times in
any day. For this process to operate efficiently, considerable informa-
tion has to be produced daily about the risks assumed by corporations.
Thus corporations operate in an economic environment in which in-
formation is produced by many independent outside groups. In well-
developed economies, the following organizations produce information
about corporate risks.
Stock analysts follow the earnings potential of corporations. For
big corporations such as Microsoft, thousands of analysts assess the
economic future of the company and produce earning forecasts and
generate recommendations about whether the stock price of the com-
pany (which summarizes the risk-return tradeoff to the stockholders)
is "fair."
Credit rating agencies such as Moody's and Standard and Poor's
analyze the economic environment faced by the companies that issue
debt capital. Such agencies often place some faltering companies un-
der "credit watch" and downgrade companies whose economic stand-
ing has deteriorated or is in danger of deteriorating.
Commercial banks monitor the activities of the borrowing corpo-
rations actively. Bank loans are typically senior and secured, and
banks attempt to anticipate any economic difficulty that the bor-
rower may face.
External auditors certify the financial standing of the borrowing
corporation and provide credible reports on the risks faced by the com-
pany. By enforcing high accounting and reporting standards, auditing
firms are able to issue financial statements (including statements about
off-balance sheet liabilities) that provide useful guidelines to stake-
holders in the company. Although this is generally the case, from time
to time auditors appear to perforn their functions poorly.
Finally, external markets, which fall into three categories, apply
here. The first category of markets is those in which the claims issued
by corporations are traded. Perhaps the most important source of in-
formation about the risks taken by corporations is these markets them-
selves. Equity and bond prices respond quickly in an efficient market
to "news" about the riskiness of companies. The second category of
markets is those for corporate control. A fall in equity prices may
reflect in part a company's declining economic fortunes. In such circum-
stances, the company may becomne the target of a takeover attempt,



108                                    SURESH M. SUNDARESAN
hostile or otherwise, in markets for corporate control. The possibility
that such a takeover might result in a reorganization of the company
gives the right incentives to senior managers of the company to seek
the right balance between risk and returns. Finally, there are markets
in which corporations can participate to manage their exposure. Ex-
amples are those for securitization, derivatives securities, and credit
derivatives.
The private sector therefore has several institutions and markets
that produce, disseminate, and update information about the risks taken
by corporations. Within this rubric of institutional factors, the task of
articulating the risks taken by corporations boils down to a determi-
nation of which risks the corporation decides to bear (based on its core
competence) and which risks it can either hedge or parcel out to play-
ers in the capital markets. The actual reporting of risk takes the form
of quarterly earnings reports and annual audited financial statements,
which provide stakeholders with much information. Independent re-
ports by stock analysts and credit rating agencies also provide a wealth
of information. Finally and perhaps most important, markets provide
almost continual information on the risks of companies.
The risk management tools used in the private sector include the
following:
* Measuring market risk. This category includes the methods used
to assess the exposure of a company to interest rates, foreign curren-
cies, commodity prices, and macroeconomic factors. The concept of
value at risk (VAR) is becoming a popular way to measure and report
the risk of financial institutions and even nonfinancial entities.
* Measuring and provisioning for credit risk. The credit risk of a
borrowing company is summarized in its credit rating reports, the prices
of its loans and bonds, and its stock prices. The financial statements of
the company also report its credit exposure.
* Stress testing. Subjecting the borrowing company to stress testing
by simulating extreme economic environments will reveal how well it
might perform under adverse circumstances.
* Contractual tools: marking position to markets, requiring exten-
sive credit screenings, establishing margins, and arranging for collat-
eral requirements. By requiring that swaps be marked to market on a
daily basis, swap dealers reduce their credit exposure to daily losses.
In the absence of marking to market, such losses may accumulate to a
level that may threaten the survival of one of the counterparties. Of-
ten, contractual provisions stipulate that positions must be settled in
cash if one of the counterparties is downgraded by a rating agency.
Such provisions try to internalize in contractual provisions future in-
creases in credit risk.



MIEASURING AND MANAGING CONTINGENT LIABILITIES           109
Institutional Framework for Governments
In sharp contrast to the private sector, governments have far fewer
independent institutions and markets that monitor and produce infor-
mation about the risk and contingent liabilities assumed by the gov-
ernment. This section begins by identifying the institutions that
articulate the risks of government and contrasting them with those in
the private sector.
Because governments do not issue stock, there are no stock analysts
who produce information about the risks assumed by them. The ab-
sence of this source of information is a major difference between pri-
vate sector entities and governments.
Governments do borrow money in public bond markets. The debt
issues of government are evaluated by the credit rating agencies, such
as Moody's and Standard and Poor's, which analyze the economic and
political environment faced by the governments. This can be a poten-
tially useful source of information. Over the last decade sovereign bond
markets have grown at a much more rapid rate than sovereign loan
markets. This means that there are market rates that reflect informa-
tion about the risks of countries.
Commercial banks and bank syndicates extend loans to countries,
as do organizations such as the ]:MF and World Bank. These institu-
tions monitor the activities of the borrowing countries actively and
produce a wealth of information on the risks of governments.
As for external markets, the category of markets that contains in-
formation about government risk is limited to the sovereign bond and
loan markets. Governments do have access to derivatives markets, and
they use them to manage some of their market exposure.
Broadly, the risks facing a government arise from the political and
economic environments. Such risks are present whether the govern-
ment in question represents a developed economy such as the United
States or a developing economy such as India. The risks that arise
from the political climate facing the country include the possibility of
war or dealing with large-scale immigration arising from instability in
neighboring regions or natural dlisasters. Such risks inevitably create
fiscal problems of considerable significance. Risks that arise from acts
of God such as earthquakes or hurricanes also may contribute to ma-
jor fiscal strains. The rest of this chapter will explore risk management
tools that comprise or address the following:
a The institutional framework for risk management issues that
deal with moral hazard issues--that is, the accountability of govern-
ment actions-and the legal framework for and enforcement of con-
tracts. Any risk management system is dependent on the underlying



110                                  SURESH M. SUNDARESAN
institutional framework for its efficient implementation. This chapter
will specify the institutional framework needed to effectively measure
and manage government risk.
* The contingent liabilities of government. Examples of
government's many contingent liabilities include various guarantees,
pension liabilities, indexed wage contracts, and health insurance. Us-
ing some of the existing framework for classifying these liabilities, this
chapter will highlight the risk measurement problem in each category.
* The unique risks of government: lender of last resort, disaster
relief, public health, famines, financing the growth of impoverished
sections, and real options in the economy. An example of real options
is the tremendous growth and foreign currency earning possibilities
that arise from information technology for a developing country. Sub-
sidies such as tax exemption may accelerate use of the Internet in re-
mote villages, which can form the basis for promoting primary
education and eventually eradicating illiteracy.
This discussion is prefaced, however, with a brief review of the
methods by which corporations measure and manage risks. This is a
useful starting point, because many of the risk management tools and
techniques have been developed in the context of corporations. The
chapter then articulates the differences between the contingent obliga-
tions and fiscal risks of corporations and governments. This allows a
focus on the key differences between corporate and government risk
management practices. The institutional framework that underlies risk
management in the corporate sector will be contrasted with the mecha-
nisms in place for monitoring the risks of government.
Risk Management Tools in the Private Sector
There are two distinct approaches to managing and measuring risks in
the private sector. The first approach is to identify those risks that the
organization believes it is particularly good at managing and earning
an attractive return on in the process. Such risks are fully borne by the
firm. Risks of the firm that are more efficiently borne by the markets
are securitized and parceled out where such markets are well devel-
oped. In their absence, such risks are laid off in listed or dealer deriva-
tives markets such as bond futures contracts or swaps. An example
should serve to illustrate this point. Mortgage companies extend
fixed-rate mortgages (FRMs) and adjustable-rate mortgages (ARMs).
To fund such a loan portfolio, such firms issue liabilities that typically
are short term in nature. Some of the liabilities may be noncontingent;
others may be contingent. Let us first focus on a simple noncontingent
liability in the context of the simple balance sheet shown below.



MEASURING AND MANAGING CO:NTINGENT LIABILITIES             1ll
Assets                             Liabilities
Fixed-rate mortgages (FRMs),       Six-month certificates
6 percent, 30-year, $500 million   of deposit, $500 million
principal amount
Adjustable-rate mortgages (ARMs)   One-year certificates
indexed to one-year Treasury bill  of deposit, $500 million
yield, $500 million principal amount
Consideration of the risk assumed by the mortgage company re-
quires a look at the exposure of its liabilities in relation to that of its
assets. If the company decides to hold both FRMs and ARMs in its
portfolio, then its risk picture turns out to be quite different: FRMs
have a long maturity in relation to the short-term certificates of de-
posit (CDs) that were issued by the company to help fund the FRM
portfolio. In other words, there is a maturity gap (or a duration gap).
T he precise nature of the gap depends on the risk of prepayments, but
that is not the principal concern now. On the other hand, the ARMs
are reset every year, and their market risk is closer to the market risk of
the liability that was issued to create the ARM portfolio. Hypotheti-
cally, assume that the gap measure is zero for the ARM portfolio. In
this illustration, the mortgage company may decide to securitize the
FRM portfolio, sell it in the mortgage-backed securities (MBS) mar-
ket, and use the proceeds to retire the liabilities and keep the rest as
profits. It may decide to maintain the ARM portfolio in which the
market risk is relatively low; the remaining exposure is the credit risk
that the company may believe it is in a better position to manage and
earn an attractive return on. In such a situation, the risk management
and reporting boil down to two actions. First, securitize and dispose of
the risk that the market is able to bear more effectively. This reduces
the size of the balance sheet and puts a lesser burden on the regulatory
capital requirements. Second, manage the risk of the remaining liabili-
ties in relation to the remaining assets on the balance sheet to earn a
better return.
Although here the balance sheet is illustrated with direct (non-
contingent) liabilities, the same argument applies to contingent liabili-
ties. A government agency such as the Government National Mortgage
Association (GNMA or Ginnie Mae) in the United States securitizes
pools of mortgage loans and sells them to institutional investors. But
at any time it may have a "pipeline" of mortgage loans that are yet to
be securitized. On this pipeline, Ginnie Mae is exposed to serious risk
of prepayments. By issuing callable bonds (a contingent liability), Ginnie
Mae can attempt to align the prepayment risk of its pipeline assets-
that is, if interest rates fall and there is a prepayment, then Ginnie Mae



112                                    SURESH M. SUNDARESAN
can use the cash flows from prepayment to call the liability back. To
report the value of this contingent liability without simultaneously
recognizing the asset it is hedging would be a mistake. Of course, in
these examples the market values of contingent liabilities and the as-
sets they help generate are easier to measure than in the case of a
government. Herein lies the challenge of government's fiscal risk
measurement.
Regulatory agencies also have a major interest in making sure that
financial institutions have a prudent risk management practice. In the
United States, institutions such as banks are under the guarantee of the
Federal Deposit Insurance Corporation (FDIC), and the government
has a major interest in ensuring that these institutions remain solvent.
In other instances, the government is interested in ensuring that the
risks are managed in a way so that in the event of a "financial conta-
gion," markets and institutions do not collapse. With these in mind,
central banks have coordinated mechanisms for managing market risk,
credit risk, operational risk, and liquidity risk in the world's financial
markets.
To summarize, the risk management actions that can be taken by
firms in the private sector can be placed in distinct categories:
* Customize the assets and liabilities so that the overall market
exposure is kept at a minimum.
* Securitize risks through the use of special-purpose vehicles (SPVs)
and sell the cash flows from the assets in the SPV to institutional inves-
tors. This will be done by firms that believe that the cost of securitization
(which requires credit enhancements, liquidity enhancements, and other
legal expenses) is less than the benefits that arise from securitization.
Such benefits may include regulatory capital relief from a reduced bal-
ance sheet, an ability to manage the firm better because of the reduced
balance sheet, and an increased ability to focus on managing those
assets in which the firm believes it has a comparative advantage.
* Use derivative markets to manage risk when appropriate. There
are two types of markets in which firms hedge their risks. The first is
listed markets, where derivative securities such as options and futures
contracts are traded in open outcry markets. Examples of such mar-
kets in the United States include the Chicago Board of Trade and the
Chicago Options Exchange. Listed markets such as these tend to be
standardized and offer extensive liquidity. Some Central American and
Latin American countries hedge their external floating dollar debt by
trading in the eurodollar futures contracts at the Chicago Mercantile
Exchange, for example. Second, firms also use dealer markets to man-
age their risks. Dealer products tend to be more customized and illiq-
uid. Examples of dealer derivatives include interest rate swaps, foreign
currency swaps, and interest rate caps.



MEASURING AND MANAGING CONTINGENT LIABILITIES            113
* Engage in prudent provisioning for "credit events" that may hap-
pen to counterparties in the future. Besides provisioning capital for
future contingencies, firms may also enter into contractual safeguards.
Contractual provisions may include collateral requirements, marking
positions to market periodically and on a contingent basis, and
optionality to terminate or renegotiate if the credit rating of the
counterparties is in jeopardy.
* Explore the growing credit derivatives market. In this market
firms can obtain full or partial insurance for specified credit events.
Credit default swaps, credit-linked notes, collateralized loan obliga-
tions (CLOs), and collateralized bond obligations (CBOs) are examples
of such derivatives.
Formulating Risk Management
Practices for Governments
In formulating the risk management practices of government, it is im-
portant to first determine the incentives that will encourage the pri-
vate sector to assume the risks that the government is ill-equipped to
bear. In addition, the government should attempt to mitigate any sig-
naling costs that may inhibit optimal provision of liquidity. For ex-
ample, as is often the case, suppose that the central bank, as the lender
of last resort, provides credit under the discount window. But the bor-
rowings under the discount window may fall because of the percep-
tion that other institutions regarcl discount window borrowing to be a
signal of liquidity or credit problems of some significance. Govern-
ments can design mechanisms that may result in a better allocation of
risks. A few examples will illustrate these points. In all these actions,
there are costs and benefits that have to be articulated before such
actions are taken.
Often, governments have to face the risks associated with acts of
God such as earthquakes or hurricanes. Currently, such risks are in-
sured by insurance companies and reinsurance firms. By helping to
create a market for the insurance of such events and securitization of
such risks, government is able to transfer at least part of that risk to
the capital markets. The development of property and casualty in-
surance companies, reinsurance companies, catastrophe-based futures
contracts, bonds, and so forth in certain markets does in fact suggest
that such risks can be managed in the private sector to some degree.
The role of such markets has been discussed in Braun, Todd, and
Wallace (1998). The growth of catastrophe-linked markets also has
complemented the reinsurance markets and helped to change the in-
centives in the reinsurance markets, which are widely perceived to be
noncompetitive.



114                                   SURESH M. SUNDARESAN
By issuing guarantees and by supporting loan programs through
direct borrowings, the federal government has improved the flow of
credit into the housing sector. The development of mortgage-backed
securities markets where the risk of prepayments is parceled out to
institutional investors such as pension funds and insurance companies
has freed the commercial banks from their exposure to this risk. This
is an example of a government policy in which the risks are reallo-
cated in the economy so that institutions bear only those risks in which
they perceive a comparative advantage. On the other hand, such deci-
sions may prove to be costly to the government. In the 1980s, when
the interest rates shot up and the yield curve became inverted, some
U.S. federal agencies such as the Federal National Mortgage Associa-
tion (FNMA, or Fannie Mae) had to be rescued by the government
through "regulatory forbearance." In the last decade, agencies such as
Fannie Mae have grown so rapidly and have leveraged their equity
capital so much that the risk exposure implicitly facing the govern-
ment could be quite substantial in the event of a failure. Moreover,
given the fact that agencies such as Fannie Mae are privately owned
companies held by stockholders, it is reasonable for the government to
reevaluate whether explicit and implicit subsidies that are currently in
place should be continued. For example, Fannie Mae does not pay
taxes, it has a direct line of credit with the Treasury, and its securities
are exempt from certain regulatory restrictions that make them more
attractive to institutional investors.
In response to the potential demand for liquidity during the Y2K
period, the central bank of the United States came up with a proactive
plan rooted in options pricing. As the lender of last resort, the central
bank has a responsibility to extend credit to vulnerable sectors of the
economy, but by selling liquidity options the central bank signaled
ahead of time that it was ready to extend credit should there be a crisis
related to Y2K. Such an action has several effects. First, it signals that
the central bank is ready to extend credit. Second, by inviting sealed
bids, it protects the identity of the potential buyer of the liquidity op-
tions. This is important because there is increasing evidence that bor-
rowing in the "discount window" has reputational costs for the
borrower. By specifying the securities that are acceptable as collateral
in the options contract, the Federal Reserve Bank lets investors know
the terms under which it is offering liquidity.
In addition, the government can follow the practice of the private
sector in providing more transparent accounting and reporting of its
balance sheet where there is a concerted effort to identify and measure
all liabilities (both contingent and noncontingent) and relate them to
the asset side of its balance sheet. For example, the guarantees ex-
tended by the government to help develop a securitized market for



MEASURING AND MANAGING CONTINGENT LIABILITIES            115
mortgages should show up as a liability. But so must the present value
of the benefits that have accrued to the taxpayers as a result of the
growth of these markets. The cost of obtaining credit for housing might
be considerably higher in the absence of the development of
mortgage-backed securities markets that grew because of the govern-
ment guarantees. This should be stressed along with the cost of gov-
ernment guarantees.
A Framework for 'Valuing Guarantees
A model for valuing government guarantees is sketched out in the
Appendix to this chapter. This section stresses the intuition behind the
general approach. Figure 4.1 provides the demand and supply curves
for an underlying product or service in the absence of a guarantee. The
equilibrium price level is P = 27. Once the government introduces a
guarantee to the suppliers, more firms may enter the market, leading
to an increased supply and possibly an increased level of risk because
some weaker firms may attempt tc take advantage of the guarantee to
enter the market. Figure 4.2 shows this effect.
As noted, the Appendix to this chapter outlines a model in which
one can examine quantitatively the effect of a price guarantee on the
vvelfare of the economy and the allocation.
Figure 4.1. Equilibrium Price
Demand/supply
120
0oo -    Demand
80 -                                  ~~~~~~~~~~~~~supply
60 -
40 
20 -
20
O        I  I  I  I   I  I  I  I  I  I  I  I  -I  I  I  I  L   .   .I .I. .
0      10      20      30     40      50      60      70
Price
Source: The author.



116                                  SURESH M. SUNDARESAN
Figure 4.2. Equilibrium Price
Demand/supply
120
100o     Demand
100 -  Demand                     Supply (under guarantee)
80 -
40Supply (nooguarantees))
40-
20 -
O  1   r_I I        I      L I  I  I  L_   I  I   I  I  I  I  I  Ij
0      10      20     30      40      50     60      70
Price
Source: The author.
Conclusion
This chapter has presented an overview of some risk management
practices in the private sector and the extent to which they may be
useful for managing the fiscal risk of government. Problems that are
unique to the risk management in government were identified. They
included the potential supply responses to government guarantees
and the ability of government to influence significantly the value of
any guarantee by its future actions. In recent times, some innovative
approaches have been used by governments to manage risk. The use
of liquidity options by the central bank of the United States, the se-
lective use of guarantees to promote the development of private mar-
kets for securitization, and incentives for the development of risk
insurance markets are some examples of innovations in which gov-
ernment has helped develop markets where risks are parceled out to
investors who are willing to bear them.
Annex 4.1. A Model of Valuing Government Guarantees
A standard model of valuing contingent liabilities views it as a put
option. Consider a simple closed economy in which the price of the
output follows an exogenous process:



MEASURING AND MANAGING CONTINGENT LIABILITIES           117
dP/P = (xdt + aFdz
where cx and a are positive scalars and {z, t > 01 is a Brownian motion
process describing the uncertain evolution of the price of the good in
the absence of any government contingent liability. Assume that the
supply curve in the economy is elastic to the price of the good, so that
the supply S is described as the constant elasticity of the supply curve
in
S  aPb
where
dS
b=   S
dP
p
is the elasticity of the supply with respect to the output price. If b = 0,
then S = a and is inelastic. In this economy, with no price guarantees,
one can characterize the welfare of the consumer by deriving the value
function and the optimal consurnption (demand). What happens to
this stylized economy when the government introduces a subsidy or a
guarantee to the price level? There will be a shift in the optimal con-
sumption (demand) and in the value function of the consumer. Study-
ing these questions requires modeling two things. First, identify the
supply response to the guarantees. Second, identify how such supply
responses affect the equilibrium price dynamics through the optimal
demand of agents in the economy. To study these questions formally,
let us investigate the optimal allocation in this economy and measure
the value to the economy in the absence of any contingent liability.
This will serve as a benchmark ifor evaluating the effect of such a li-
ability on the price process through (a) a supply response and (b) opti-
mal demand allocation.
Bencbmark Allocations
Let the economy be described by a representative agent who maxi-
mizes the expected lifetime utility of consumption of the good as
Max,E 1[ eFes u(cs)ds]
The wealth of the consumer who also holds all the stock of the
goods supplied is simply
W =.PS = aPb, I



118                                 SURESH M. SUNDARESAN
The dynamics of the wealth process is
dW = W [(b + l)ax + I b(b + 1)a2]dt - Pcdt + W(b + 1)adz.
2
The optimal consumption problem of the consumer may be specified
in terms of the Hamilton-Jacobi-Bellman (HJB) equation, where J is the
value function associated with the optimal consumption allocation:
0  u(c) - PJ + Jw ((b + 1)aW + ½12b(b + 1)c22W - Pc) +
| /2jww(b + 1)2W2o2+Jp cP + 1/JppP2 +J  WP(b +
The first order condition for oprimality is
us = PJW
This problem cannot be solved in closed form in general unless ad-
ditional restrictions are placed on the problem. To get concrete results,
assume that the utility function is of the constant relative risk aversion
(CRRA) type, so that
u(c) =-
y
For this problem one can solve explicitly for the optimal consump-
tion policy and the value function of the economy. This is stated as a
proposition below.
The optimal consumption (demand) is linear in wealth and nonlin-
ear in the price level
c= [ho4]y- w
where 4D is a function of the price. The function 4 is found by solving
the partial differential equation
0=  r  - p[y<>] + [yFD]((b + 1)x + ½b2b(b +  p)2-yPLY]' )
+ 1/2[yD](b + 1)2a2 + [F)pcP] + I/2DPPo2P2 + D>PYP(b + J)G2
The value function of the economy is
J = W¢I(P).
The function (D(P) is given by
m/ P) = a,p-r



MEASURING AND MANAGING CONTINGENT LIABILITIES           119
and
1  [ p  yba - 2b(b + 1)ya2 d (y + 1)yyc2 + y2a2(b + 1)]
Note that combining the expressions for J and 'D(P) yields
J=a[    ] .
Allocation with a Guarantee
Consider a simple guarantee scheme that is specified as follows: the
government will pay the amount P - P whenever P < P. In such a
situation, there is the following supply and consumption response. The
supply function will be
S =aP for P> P
and
S = aP b for P < P.
This supply response is consistent with the expectations of pro-
ducers that the government will compensate them in states of the
world where the actual price of the good is less than the guaranteed
price. As the price goes below the guaranteed level, the supply stays
fixed and the only uncertainty that faces the consumer is the wealth
level. Corresponding to these two regions, the wealth dynamics evolves
as follows:
dW= W[(b + 1)a + ½12b(b + 1)a2]a!t- Pcdt + W(b + 1)adz, when P > P
and
dW = Wadt - Pcdt + Wadz, when P < P.
The optimization problem facing the consumer now can be stated as
[ u(c) - pJ + Jw ((b + 1)(xW + ½12b(b + 1)a2W - Pc) + Jp aPl
0 ax  + ½/2JWW(b + 1)2W2a2+ ½/2JPPo2P2 +JpwWP(b + 1)a2]'
when P > P and
0 = Maxc [u(c) - pJ + Jw (aW - Pc) + i/2½WWW2G2], when P < P.



120                                      SURESH M. SUNDARESAN
It turns out that the value function can be written as J = WYTI(P)
and ] = a2WY.
Enforcing the conditions of value matching and smooth pasting
conditions yields
J (P, W) = J(P) =* ¢(P) = a2
and
JW(P, W) = Jw(P) > 'P(P) = a2.
This framework, using the value function and the optimal consump-
tion rule, provides a guideline for a numerical solution.
References
Abel, Andrew, and Douglas Bernheim. 1991. "Fiscal Policy with Impure
Intergenerational Altruism." Econometrica 59: 1687-1711.
Bardsley, Peter. 1994. "The Collapse of the Australian Wool Reserve Price
Schemes." Economic Journal 104: 1087-1105.
Barro, Robert, J. 1974. "Are Government Bonds Net Wealth?" Journal of Po-
litical Economy 82: 1095-1117.
- 1979. "On the Determination of the Public Debt." Journal of Political
Economy 87: 940-71.
- 1989. "The Ricardian Approach to Budget Deficits." Journal of Eco-
nomic Perspectives 3 (spring): 37-54.
Bean, Charles, and William Buiter. 1987. "The Plain Man's Guide to Fiscal and
Financial Policy." Employment Institute, London, October.
Bernheim, Douglas. 1987. "Ricardian Equivalence: An Evaluation of Theory
and Evidence." In Stanley Fischer, ed., NBER Macroeconomics Annual (Na-
tional Bureau of Economic Research). Cambridge: MIT Press.
Bernheim, Douglas, and Kyle Bagwell. 1988. "Is Everything Neutral?" Journal
of Political Economy 96: 308-38.
Blejer, Mario, and Adrienne Cheasty. 1991. "The Measurement of Fiscal Defi-
cits: Analytical and Methodological Issues." Journal of Economic Literature
29: 1644-78.
Borensztein, E., and G. Pennacchi. 1990. "Valuation of Interest Payment Guar-
antees on Developing Country Debt." IMF Staff Papers 37 (December): 806-
24, International Monetary Fund, Washington, D.C.
Braun, Anton, Richard Todd, and Neil Wallace. 1998. "The Role of
Damage-Contingent Contracts in Allocating the Risks of Natural Catastro-
phes." Research Department, Federal Reserve Bank of Minneapolis, April.
Brixi, Hana Polackova, and Leila Zlaoui. 1999. "Managing Fiscal Risk in
Bulgaria," Draft paper. World Bank, Washington, D.C., November.



MEASURING AND MANAGING CONTINGENT LIABILITIES                 121
Brock, Philip. 1992. "External Shocks and Financial Collapse: Foreign Loan
Guarantees and Intertemporal Substitution of Investment in Texas and
Chile." American Economic Review 82 (2): 168-73.
Buiter, Willem. 1983. "Measurement of the Public Sector Deficit and Its Im-
plications for Policy Evaluation and Design." IMF Staff Papers (June):
306-49, International Monetary lFund, Washington, D.C.
Eisner, Robert. 1984. "Which Budget Deficit? Some Issues of Measurement and
Their Implications." American Economic Review 74 (2): 138-43.
Eisner, Robert, and Paul Pieper. 1984. "A New View of the Federal Debt and
Budget Deficits." American Economic Review 74 (2): 11-29.
Kryzanowski, L., and G. S. Roberts. 1993. "Canadian Banking Solvency."
Journal of Money Credit and Banking 25 (3): 361-76.
Merton, Robert C. 1977. "An Analytical Derivation of the Cost of Deposit
Insurance and Loan Guarantees." lournal of Banking and Finance 1: 3-11.
- 1978. "On the Cost of Deposit Insurance When There Are Surveillance
Costs." Journal of Business 51 (July): 439-52.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Salant, S. W 1983. "The Vulnerability of Price Stabilization Schemes to Specu-
lative Attack." Journal of Political Economy 91: 1-38.
Townsend, R. M. 1977. "The Eventual Failure of Price Fixing Schemes." Jour-
nal of Economic Theory 14: 190-9.
Wu, Yuan-Li. 1950. "Government Guarantees and Private Foreign Investment."
American Economic Review 40 (1): 61-73.






CHAPTER 5
Analytical Techniques Applicable
to Government Management
of Fiscal Risk
Krishna Ramaswamy
The Wharton School, University of Pennsylvania
INSTITUTIONS, CONTRACTS BETWEEN institutions, and entire market-
places have developed to facilitate the transfer of risks from those un-
willing or unable to bear them to those who are both willing and able
to do so.
A privately owned corporation typically has a portfolio of projects
or activities, each of which contributes a random amount per period
to its overall profits. It will, in general, finance these projects with
forms of debt or equity capital or with both. In deciding between alter-
native investment projects, the privately owned corporation looks to
add value, net of the cost of the investment. The computation of that
value involves discounting the risky stream of cash flows from the
project at a rate that is adjusted to account for the risk of the project.
In deciding between alternative ways of financing its portfolio of
projects (for example, with short-term or long-term debt, or convert-
ible debt, or preferred stock), the corporation seeks to maximize the
value of its shareholder ownership. The shareholders have limited li-
ability-a feature that permits shareholder-appointed managers to take
greater risks in search of profitable opportunities. The celebrated
]\Vodigliani-Miller theorem offers assurance that, under certain condi-
tions, the total value of the projects undertaken is unaffected by the
way in which the firm's activities are financed. This result ensures that
the manner in which the risk of the aggregate cash flows from all the
projects is partitioned does not affect the firm's current market value.
123



124                                     KRISHNA RAMASWAMY
Absent bankruptcy costs, the use of derivatives to "hedge" the cash
flows from the activities does not in itself add value.
A government, or state-owned enterprise, also has a portfolio of
activities, although the scale of some of these may be dictated and thus
better recognized as commitments. These activities will most often be
paid for with general tax revenues, and in some cases, the entity might
float a public debt issue with a guarantee from another source. It is in
comparison with its counterpart in the private sector that several im-
portant differences emerge. First, the choice between alternative ac-
tivities as expenditures or "investments" is generally not made
actively-rather, it is set exogenously by law or by a political process.
Second, where the private corporation can avail itself of market prices
in imputing (current) values to lines of activity or even entire divi-
sions, the state-owned entity cannot perform valuations of equal reli-
ability. Indeed, the notion of "valuing" departments or activities within
a government budget-by computing the present value of future net
outlays or receipts, thereby acknowledging the intertemporal nature
of these activities-runs at odds with the conventional cash flow-based
budgeting that forms the basis of decisions in the public sector. Third,
where the private corporation's shareholders can seek protection from
its creditors, the public entity cannot avoid its explicit and implicit
commitments in the event of a crisis. Here I have highlighted the three
major differences that an academic finance theorist would see; readers
with experience in these contexts will recognize other differences.
The purpose of this chapter is to describe the analytical techniques
that can reasonably apply to the measurement of risks within a fiscal
budgeting context and to discuss, in conceptual terms, the analytical
methods that enable a decisionmaker to analyze them.
What are the main analytical tools for measuring fiscal risk that
can be brought to bear from financial theory? A popular paradigm in
finance employs mean-variance analysis, in which undesirable fluc-
tuations in wealth or cash flow, as measured statistically by the vari-
ance (or equivalently by the standard deviation), are balanced against
the desirable characteristic of the mean or expected value. Researchers
have developed models of risk measurement in this context-known
as factor models-in which the randomness in the fluctuations in the
value of an investment is related to some basic economic variables
such as the gross national product (GNP) or industrial production.
This class of models enables computation of the risk (standard devia-
tion) of an asset or a cash flow as it derives from more basic and
fundamental influences; it also enables identification of those influ-
ences that are more important than others. This chapter explores an
application of factor models within a mean-variance framework and
applies it to a budgeting context in which random (cash flow) expen-



ANALYTICAL TECHNIQUES                                      125
clitures from several distinct units (departments or subnationals, say)
create the fiscal risk for a national body.
What are the main tools for controlling fiscal risk? This question is
riot easily answered for the following reasons. First, fiscal risk is an
outcome of commitments that are either explicit or implicit, and the
immediate control measure, to scale back these commitments or to
"deleverage" them, may not be a realistic alternative in many cases.
Second, the normal prescription-to purchase insurance and effect the
transfer of the risk to those who a:re willing and able to bear the risk-
may require the use of markets that are not yet developed and that in
many cases remain closed for fundamental reasons. That said, forms
of risk transfer in the public sector, such as loan guarantees and insur-
ance coverage, do constitute valicl methods.
For the application of the relevant analytical methods (such as op-
tion pricing), some conditions mlust be met.' In order to provide a
useful treatment, I will
* Stay (as much as an academic can) within the practical context in
which most government decisionmakers find themselves.
* Take the basic situation to be one in which a parent government
unit (a federal or national authority) oversees the budgets of several
state-owned enterprises, or subnational units, or even departments.
* Take as given a common period for which the parent and sub-
units are assessing the fiscal risks.
* Take the typical budget process to be one in which each subunit
prepares its budget for the future time period with a common set of
forecasts-of, for example, the expected rate of growth in per capita
income, an average exchange rate or interest rate, or expected fuel
prices-that are determined by the parent government unit or that are
agreed on in preparing the budget.
* Assume that the parent government takes up the net deficits that
occur at the level of the subunits. This means that surpluses and defi-
cits at the subunit levels are passed through to the parent unit.
It is possible to relax most of these assumptions and employ the
ideas in other contexts as well. The final assumption, however, is more
critical. I have taken as a working assumption (because it simplifies
the analysis considerably) that the parent government receives the sur-
pluses and is responsible for the deficits. Such a symmetrical pass-
through of surpluses and deficits rnight make the analysis more suitable
for a parent unit that oversees several subunits or departments that
have little or no autonomy. In the conclusion of this chapter, I have a
few remarks about the one-sided case in which only the (gross) aggre-
gate deficits (perhaps exceeding some level) are borne by the parent
unit and what methods can be applied in this case.



126                                    KRISHNA RAMASWAMY
Risk Measurement in a Fiscal Context
Contributors to this book (see Chapter 1 by Brixi and Mody and Chapter
18 by Ma) and others (see, for example, Polackova 1998 and Lewis
and Mody 1997) have discussed the classifications of risk in a fiscal
context and recommendations about its measurement and control. In
this section, I add to this practical discussion by describing the analyti-
cal tools that can be applied from the field of finance. To do so, I
introduce some symbols to simplify the nature of the analysis, but I
surround them with their definitions and explain in words the impor-
tant concept to be retained.
It is useful discipline for every government unit to recognize all its
activities and departments nominally as sources and users of funds.
Because some units may be both, it is recognized that in making a
quantitative tally one might equally rely on the net source of funds to
describe a department's contribution to the overall fiscal surplus or
deficit. Within a government unit i, I denote the net cash flow amounts
in a budget period t as (ft, which can be thought of as the period's net
expenditure. I treat positive numbers as a use of funds (as is typical of
most expenditure-oriented activities) and negative numbers as a source
of funds (as would be normal for a surplus unit or a revenue author-
ity). The squiggle above the letter C serves as a reminder that the
projected cash flow amount is a random variable. Like any random
quantity, it has an anticipated component-the expected value-which
is almost always the basis for the departmental budget allocation. In
what follows I use the terms anticipated budget or expected cash flow
interchangeably to apply to the expected value, denoted by E(C,).
The realized cash flow in period t can take values that are higher or
lower than expected, and the unanticipated component is denoted as
u,, and is the difference between the realized and expected values:
,= C,, - E(C1)
Note that ex ante-at the beginning of period t-the value of u;, is
random, but when it is revealed that the i-th activity has a larger (or
smaller) deficit than had been budgeted, ii,, is positive and is an un-
pleasant (or negative and a pleasant) surprise, ex post.
I characterize the fluctuations in the unanticipated shocks iit as the
fiscal risk contributed by the i-th government unit. By its definition,
the average value of these shocks is zero.
An individual unit might further recognize that there are some domi-
nant sources that drive the fluctuations in the values of u;t. For ex-
ample, the costs of operating a network of schools will involve energy
use and related transportation costs, where both the amount used and
its price are weather-dependent and outside the control of the oversee-
ing department. In other contexts, the demand for government relief



ANALYTICAL TECHNIQUES                                      127
services might be related inversely to the aggregate output and em-
ployment, or heavily dependent on the international price of a locally
produced and exported commodity. A classification of the underlying
influences that affect the unanticipated component of the i-th unit's
net expenditure is valuable not only in helping design systems to moni-
tor and control those risks, but also in permitting the supranational or
overseeing entity to assess its exposure to those influences.
A risk model of this type is conventionally called a factor model.
For example, in assessing the risk of a portfolio of common stocks, a
financial theorist would assess the stocks' individual risks (standard
deviations) as well as the correlations between every pair of stocks in
the portfolio. A more parsimonious description of the risk structure of
the stocks in this portfolio is a model that assumes there are a few
common factors that affect each stock's returns and that a given stock
has an elasticity or a sensitivity to each of these factors. The fluctua-
tions in a given stock's returns depend on the sum of the influences
contributed by each of these factors, plus a risk that is unique or idio-
syncratic to that stock and uncorrelated to the returns of the other
stocks. Anyone summarizing a factor model for a useful representa-
tion of the risks in stock market portfolios would need for a given
st:ock a list of the elasticities to each of the K common factors and a
measure (standard deviation) of the risk of its unique component. In
practice, such a model can be statistically estimated from historical
data using the technique of factor analysis in which the data reveal
what the common influences are, or it can be conducted with a
prespecified list of factors that are imposed a priori (for a clear and
useful exposition of such a procedure, see Sharpe 1981).
In the context of the application here, the lack of adequate histori-
cal data related to budget expendlitures (especially in a government
context in which departments and activities are being added over time)
will not permit the statistical estimation of such a factor model. How-
ever, the very procedure of constructing a periodic budget will point
up the factors (such as fuel prices or exchange rates) that are influen-
tial and lead to the estimation of the elasticities.
Tbe Measurement of Risk at the Level of the Unit
The probability distribution of the unanticipated shock ii, to the i-th
department's budget in period t associates with each (possible) level of
the random shock an assessment of the likelihood of its occurrence.
The standard measure of risk in this context is the variance of the
related quantity, the standard deviation. When the likely shocks are
symmetrical around their average value of zero, the standard devia-
tion measure reveals that approximately two-thirds of the realized
outcomes will lie in an interval of one standard deviation on either



128                                    KRISHNA RAMASWAMY
side of zero. The graph that follows shows such a distribution; it en-
compasses extreme values that are more than two standard deviations
away and thus occur with lower probability (likelihood), but the bulk
of the outcomes occur in the middle.
I      I      I
-3     -2    -1      0      1      2     3
Subunit i's deficit ;,,
Arriving at a forecast of the standard deviation of ii, is a difficult
but necessary step. It can draw on historical experience, or it can be
derived from a subjective forecast of, say, three or more scenarios for
the most significant influences on the department's activity.
One could expect that the larger the department, the larger would
be its expected cash flow E(C;,) and the larger would be the standard
deviation of its unanticipated component. There is no reason to ex-
pect, however, that there is a special relationship-proportional or
linear-between the expected values and the risks of a departmental
cash flow when viewing these quantities across departments. One could
(reasonably) expect that the risk or standard deviation of a particular
department is related positively to the growth rate in the department's
budget-the year-on-year growth in the anticipated budget item E(Ci,).
Note, however, that the level of the anticipated budget E(C) may be
a policy-driven variable that can be taken as exogenously given or as a
forecast, and its growth rate might also reflect an active decision com-
ing from a political process.
The Aggregation of Cash Flows of Constituent Units
At the supradepartmental or governmental level, the aggregation of
the cash flows requires adding up the contribution from the N con-
stituent units:
Aggregate Cash Flow =   + C, + C3, +   + CN,
Within the period, and working with a cash flow-based budget pro-
cess, one can anticipate the fiscal deficit at the governmental level to
be the expected value of



ANALYTICAL TECHNIQUES                                    129
E(Aggregate Cash Flow) = E(C,,) + E(C2,) + E(C3,) + *  + E(CN,)
which works out to the sum of the anticipated budgets of the N con-
stituent units. If this aggregate nlmber is positive, then the govern-
ment anticipates and plans for a deficit, given that I chose to use positive
numbers to represent net expenditures or uses of funds.
What can be said about the risk at the level of the parental govern-
mnent unit? Typically, it is committed to meeting the needs of unantici-
pated net expenditures at each department level, and I shall assume
that it does. It might have the ability to draw on the unanticipated
surplus from those units that has come in with lower-than-expected
net expenditures, or higher-than-expected net revenues, perhaps with
a lag. In other, perhaps more realistic situations, the ability to use the
cash flow from surplus units might be limited. I proceed here with the
assumption that the government can draw on these resources fully.
This assumption enables me to write that the unanticipated compo-
nent of the government deficit labeled in period t is U,, which is the
sum of the unanticipated components at all the departmental levels
Ut    ,+ "21 q I3t +  + UN,
If each departmental unit stayed just at its anticipated budget level,
then the realized value of each unanticipated component would be
zero, and, as a result, the government deficit also would be right on
target. However, this is an unlikely outcome; in practice, some units
would show net expenditures higher than targeted and others might
come in at below-target levels.
In this scheme is a diversification effect. The risk (or standard de-
viation) of the aggregate unanticipated deficit is not equal to the sum
of the risks (or standard deviations) of the individual unanticipated
components. The standard textbook analysis would point up that if
the unanticipated components were less than perfectly correlated-
assuming for the sake of argument that some pairs were negatively
correlated-the aggregate component would show a lower level of
fluctuation than the sum of component unit risks would lead one to
believe.
But a more useful analysis can be brought to bear conceptually.
Suppose, as was argued earlier, that for the i-th unit the unantici-
pated net expenditure iu,-the deviation of actual expenditures from
those that were anticipated and budgeted-is weakly and positively
influenced by the realized growth rate in per capita income, but
strongly and positively related to the deviation of fuel prices from
the values that were used in planning the anticipated budget. These
driving "factors" are the sources of risk, and the sensitivities of the
net excess expenditures over the anticipated budget to these factors



130                                     KRISHNA RAMASWAMY
will be large in relation to fuel prices but small in relation to the
income growth rate.
Other budget units might face exposures to these two "factors"-
the growth in incomes and fuel prices, factor sensitivities of different
magnitudes-and they also might be exposed to an additional and
different group of factors altogether. The exposure of the suprana-
tional government to these factors is the sum of the exposures of the
individual units to each of the separately identified factors. The pres-
ence of common and pervasive factors-including aggregate demand
levels, exchange rates, interest rates, international commodity price
levels, and so on-in each departmental unit means that the variances
and covariances of the unanticipated net expenditures across depart-
ments will display a "factor" structure. The Appendix to this chapter
shows the computation of the aggregate risk in the case in which one
can identify a few common and pervasive factors that affect the net
expenditures of the constituent units; a stylized numerical example
with two common factors is provided.
The computation of the aggregate risk reveals the following proper-
ties, in each of the following conditions:
* Case 1. The individual units are exposed to diverse and unrelated
risk factors, so that there is little commonality among them. In this,
the risk reduction achieved depends on the levels of the relative planned
expenditures (the anticipated budgets) within the units.
* Case 2. The separate units are exposed to the same common fac-
tors in the same way; the sensitivities to these factors across the units
are all of the same sign but of different magnitudes. In this case, the
exposure at the government level to the factors is the sum of the expo-
sures across the units. Here the magnitudes of the budgets remain rel-
evant, but the important point is that the exposure to the factor-oil
prices or interest rates-is averaged across the units.
* Case 3. There is commonality among the factors to which the
individual units are exposed, but the sensitivities to these factors are of
opposing signs, so that, for example, one unit has a positive exposure
to fuel price rises while others have negative exposures. It is in this
case that the maximum reduction in exposures to the factors can be
anticipated, but it would be an atypical occurrence.
What can be said of the risks that are unique to the subunits, the
realizations of Z? If the sums have been done right and all the common
factors have been considered, these risks should be unrelated. How-
ever, it can be anticipated that some departments' expenditures would
be hit by "acts of God"-earthquakes or unforeseen emergencies, and
so on-and that the incidence of such "catastrophic" risks might be
modeled further, perhaps in a Monte Carlo simulation context.



ANALYTICAL TECHNIQUES                                      131
Is it possible to compute a figure to serve as a barometer of future
fiscal shocks? Given a distribution for the aggregate fiscal risk, it is
natural to ask what is the likelihood of an unanticipated shock of a
given size to the deficit? This calculation would be similar in spirit to
the daily evaluations of the prospective risk exposure of the portfolio
of a large financial institution. In these organizations, there is a daily
calculation of the value at risk (VALR), or of a related number, which
indicates the level of loss one can expect over the next day or week
with a prespecified probability level. For example, a 10-day, 95 percent
VAR of $45 million would indicate that a loss of more than $45 mil-
lion could be expected over the next 10 days with 5 percent probabil-
ity. In the context here such a calculation is possible, not on a short-term
basis-which would not really be relevant-but over the future inter-
val for which the budgets are prepared.2
To summarize, the nature of fiscal risks that derive from an aggre-
gate exposure to several subunit budgets can be assessed quantitatively
by employing a factor structure in which the factors as economic in-
fluences are imposed from prior knowledge of the subunits' activities.
A.n estimation of such risks is a first step to computing the exposures
and the overall liability of the parent unit.
The Control of Fiscal Risks
A shareholder-managed corporation in the private sector can exploit
the diversification of the cash flows from the aggregation of the com-
ponent cash flows from all the projects and divisions on its balance
sheet in precisely the same way as described here for the government.
However, the corporation in the private sector can alter the alloca-
tion of its investments across its divisions, thereby controlling the
risk (and the expected profitability) of the overall enterprise to its
current shareholders as owners. But such changes in the allocation of
budgets to the underlying departmnents are not generally feasible for
the government unit.
This brings us squarely to the issue of how the risk at the aggregate
governmental level can be controlled. It is by now well recognized
(and it has been stressed repeatedly elsewhere in the chapters in this
volume) that budgeting for government fiscal risks is made more diffi-
cult when the subunits under the government's direction exploit the
fact that excess unanticipated expenditures will be financed by the
parent unit. An explicit (or implicit) coverage of this type provides the
wrong incentives for the subunits: they may no longer take pains to
prepare accurate budgets and forecasts, and they may not monitor
expenditures and control risks to the extent that they would if they
were residual claimants as are the shareholders of a private sector



132                                      KRISHNA RAMASWAMY
corporation. When a subunit has the ability to affect the level of the
net expenditures, any coverage of the excess net expenditures over the
anticipated budget will raise issues related to moral hazard. In this
sense, the assessment of the probability distribution of each unit's i,. is
made doubly difficult.
In an individual's risk asset portfolio, the most immediate and di-
rect way to reduce risk is to allocate part of the portfolio wealth to the
risk-free asset.3 That way, the overall risk is reduced. When an insurer
(the example here is of the government unit) assesses the actuarial or
expected value of any future liabilities arising from unanticipated ex-
penditures, and it incurs and provisions for it, it effectively invests in a
risk-free asset to reduce its risk. In practical terms, this is equivalent to
limiting the size of its activities.
As we shall see in the rest of this chapter, in cases in which the
liability incurred is related to an economic quantity that is effectively
a traded asset, it is possible to actively control the liability.
Market-Based Vehicles for Controlling Risk
What are the other ways in which a privately held corporation can con-
trol its risks? Exchange-traded and over-the-counter methods for risk
control undertaken by corporations take myriad forms, but their diver-
sity (and complexity) can be reduced to one or more of the following:
* Forward and futures contracts. These are methods that enable
the user to lock in a fixed price or fixed rate at a future date.
* Options contracts. These are essentially the same in this context
as acquiring insurance. They enable the user to purchase protection
against unfavorable outcomes. The purchase of put options would in-
sure against unfavorably low prices for an asset or commodity that
one wishes to sell. The purchase of call options would provide insur-
ance against unfavorably high prices for an asset or commodity that
one is preparing to buy.
* Swaps. These are methods that provide protection against risk
over several periods, being in essence a portfolio of forward positions.
Note that these contracts can apply to risks related to economic
quantities such as exchange rates, interest rates, equity values, com-
modity prices, including energy prices, and default premiums. (Other
mechanisms for risk management, such as securitization and sale of
related securities, are discussed elsewhere in this volume.) Often the
risk that a particular company faces might not be identical to a traded
underlying asset-but this "basis risk" poses only occasional difficul-
ties to most risk managers' operations.



ANALYTICAL TECHNIQUES                                      133
A conventional question in this context posed to the risk manager is
related to the cost of risk reduction or hedging. The fact that no outlay
is needed to engage in forward contracts, or for that matter in closely
related swap contracts, cannot be taken to imply that there is no cost
to hedging. By locking in, via a forward contract, a fixed price for the
acquisition of heating oil in the future, the consumer reduces the fluc-
tuations in expenditures, but knows that the fixed price that was quoted
is adjusted for the risk of the fluctuations that someone else is now
bearing. In the case of options contracts, the up-front fee for the option
is the actuarially fair price for insuring against unfavorable future prices.4
In looking at the risk inherent in its fiscal deficit, a government
typically sees the risk of low growth as the most important factor. It
would be impractical for the government to hedge against this "fac-
tor."5 But there are other factors that drive at least part of the fiscal
(deficit against which the government can hedge, such as fuel prices,
commodity prices, and interest rates.
One contingent liability that is discussed repeatedly in this volume
relates to a loan guarantee, or an insurance guarantee such as that
given for deposits at financial institutions. It is well known that these
guarantees are isomorphic to (typically) put options. The celebrated
Black-Scholes model provides the analytical basis for the valuation of
these guarantees; it is based on the assumption of the tradability of the
underlying risk and the feasibility of a dynamic hedging strategy. When
these assumptions do not hold (as they may not in the context of many
developing countries' guarantee provisions), it does not mean that the
valuation cannot proceed; in these cases, one can reasonably use the
technique of computing the discounted expected value.6
Here I must stress that the recognition of the liability and the com-
putation of its "fair value" take on the usefulness of an amulet, be-
cause these actions in themselves do not limit the liability or even help
to control it. A government thai: provides guarantees is in the same
position as an individual investor who has written a put option-and
that investor typically will not remain sanguine in the knowledge that
he has carefully recorded the opcion's Black-Scholes value as a liabii-
ity. It is only in the constructive act of taking offsetting positions that
the risk is reduced.
The Factor Model and Contingent Liabilities
The factor model described earlier highlighted the fact that a parent
government unit faced the sum of the exposures to the common eco-
nomic influences that affected the subunit budgets. Some of these com-
mon influences may be traded economic quantities, a fact that the
parent unit can exploit in one of two ways.



134                                    KRISHNA RAMASWAMY
The parent unit can sell explicit guarantees to the subunits against
unfavorable outcomes in these quantities. Such guarantees could be
hedged by the explicit purchase of marketed insurance contracts at the
level of the parent. Under this scheme, the parent serves as a ware-
house that manages a book of these liabilities. Incentives at the sub-
unit level are maintained-they would still need monitoring-because
the economic quantity that is the basis of the guarantee is observable
to both parties, so long as the parent is not also the guarantor of the
deficit from all sources of risk.
An alternative scheme is for the parent to take up the exposures to
the factors from the subunit levels, but purchase insurance against the
resulting aggregate factor exposures to traded underlying risks directly
from the market. Consider the following example: a state government
oversees the expenditures of several units, including the Department
of Highways, the Department of Prisons, the Department of Unem-
ployment Relief, and the Department of Public Education. Each sub-
mits a budget and prepares estimates of the quantity of fuel and heating
oil it would need, using projections for the demands for its services
and using forecasts of forward prices. A severe winter would lead to
higher consumption of heating oil at both the schools and the prisons,
perhaps at higher-than-expected prices. Extremely severe winters would
require increased use of fuel to clear the highways, but savings in heat-
ing oil at the schools. A prolonged bout of bad weather can hurt agri-
cultural output and affect future unemployment levels. Both heating
oil and gasoline are traded commodities, and it is possible to structure
risk management of the unanticipated expenditures in relation to a
fuel-related factor. Weather derivatives are available, as are insurance
contracts against severe weather. An assessment of the exposures of
the separate departments to fuel price fluctuations and to extreme
weather scenarios would permit the parent state government to either
hedge the unanticipated demands on its resources or provide explicit
insurance where it is feasible to do so.
Conclusion
Analytical techniques for risk measurement and risk management that
have been developed in finance are applicable to the management of
government fiscal risks; the most well-known example among these is
the provision of guarantees or insurance in the context of pension ben-
efits, deposit insurance, and protection against default by a public or
private corporation.
In this chapter I have highlighted the role of risk measurement in
the fiscal context by employing a factor model, which forms the basis
of risk measurement in portfolio analysis. The model can be adapted



ANALYTICAL TECHNIQUES                                     135
to the fiscal context where a parent government oversees the expendi-
tures of several subunits with separate budgets; the risk factors in this
context may be recognized by examining the economic influences within
these budgets. Whereas the typical application of the factor model in a
portfolio context leads to the allocation of investments, in the context
of fiscal risks the application is confined to measuring and monitoring
risks. However, this is an importan't first step in the use of techniques
to control for these risks. The factor model is particularly suited to the
aggregation of risks and the recognition of the net exposure of the
parent unit to latent risks in the departmental or subunit budgets.
Annex 5.1
This Appendix spells out the technical details of the analytical model.
The model corresponds to a K-factor model of the net expenditures of
N separate entities, which were described as departments or subunits
earlier in this chapter. The parent government unit aggregates the cash
flows across these subunits and is assumed to be responsible for the
aggregate net expenditures that are unanticipated, implying that at the
end of the period the parent government will meet the net demands for
aid from each of the subunits.
Let C,, represent the cash flow of the i-th subunit, which is first
written as
C,, = E(C,,) +
This equation says that the ranclom cash flow from the i-th unit in
period t is (tautologically) decomposed into its expected value plus a
mean zero error, iu,. The expected value corresponds to the antici-
pated budget or forecast budget for department i; the error term uit
corresponds to the surprise or what earlier in the chapter I called the
unanticipated net expenditure.
Factor Structure
Now I shall impose a linear factor structure on the unanticipated net
expenditure; K pervasive common factors will affect the i-th depart-
ment's budget expenditures:
K
it= E (1ikfk} + Eji
k = i
This relation says that the K separate factors are underlying sources of
risk that drive the risk at the department's budget level. The sensitivity
of the budget to the particular factcr k-say, the fuel price, representing



136                                     KRISHNA RAMASWAMY
the deviation of the realized fuel price from the value that was used in
designing the department's budgeted expenditure for fuel-is captured
by bik. Large positive values for b,k would imply that the department's
net expenditures amplify the fluctuations in fuel prices around their
forecast values; small and positive values indicate a lower sensitivity,
while a negative value indicates that the effect on the department's
budget is to promote a surplus (deficit) relative to budget estimates of
total net expenditures when fuel prices are higher (lower) than ex-
pected. Note that some factor sensitivities may be zero-the subunit
may have no exposure to exchange rates or interest rates. The final
term, e,, represents the risk that is unique to the i-th subunit and
unrelated to any of the chosen factors; it is the idiosyncratic risk whose
variation will be the only source of variation in the subunit's realized
budget figures if each of the factors takes on the value zero (which
would mean, for the example of the k-th fuel price-related factor, that
the realized fuel price was equal to its expected value, used in prepar-
ing the anticipated budget).
The following remarks apply to the choice of the factors and the
factor sensitivities:
* The factors should be chosen to represent the major common
influences to which the subunits are exposed. The examples I give: fuel
prices, foreign exchange rates, interest rates, aggregate economic ac-
tivity-related variables such as employment or per capita income, the
international price of a locally produced or imported item or raw
material, and so on.
* The factor sensitivities can actually be traced through the budget
process as a budgeting variance. By tracing the effect of a 1 percent
change in the forecast fuel price, one can ask what the effect on the
overall budget would be, keeping other things constant.
* One assumption I place on the model says that the expectation of
the unique or idiosyncratic shock at the department level, conditional
on the factor realizations, is zero-that is,
E (ii,tif, f2, f3, .. I f) = °.
This relation says that the unique or idiosyncratic term is just that:
unpredictable noise that is unrelated to the factors.
* Note that I do not make the assumption that the factors are
uncorrelated to each other. In fact, I expect that several assumptions
about the factors have been used in devising the budget items, including
assumptions about average fuel prices, unemployment levels, and ex-
change rates. And I expect that these economic quantities are related.
* Now consider the factor structure for the net unanticipated ex-
penditures for two separate subunits, i and j. If I have extracted all the
common factors between the net random expenditures of the subunits,



ANALYTICAL TECHNIQUES                                     137
then I can conclude that the correlation (equivalently, covariance) be-
tween the residual or idiosyncratic risks, £it and i,,, is zero.
In typical finance applications, the factors are extracted from the
data, and the usefulness of the factor model comes from the properties
of the correlation between these "residual" risk terms. There, it is
usually argued that the correlations between the idiosyncratic risks are
weak, or zero.
In the application that is the fccus here, the factor model is being
employed to capture commonalities in net unanticipated budget ex-
penditures. Here I do not have the luxury of a sample of past data
from which to extract common factors: rather, I will impose such fac-
tors and attempt to capture as many important economic influences
that can make the subunit's actual expenditures depart from budgeted
expenditures. Of course, one outcome of this is that if my specification
cf these factors is partial or incomplete, then the residual risks can be
correlated across the subunits.
The variance of the unanticipated net expenditures at the level of
the subunit is then
Var(Cit) = Var(iu,,) = Var |, bjk4k } + Var(ij,)
Notice that a degree of "diversification" is taking place within the
subunit itself: the net budget shortfall or surplus will be driven by the
aggregate influence of K factors. The ratio of the variance due to the
factors to the total variance captures how much of the variation in
the budget expenditures is captured by the K chosen factors in the
decomposition.
Aggregation to the Government Level
The parent unit serves to absorb the sum of the unanticipated net ex-
penditures at the level of the subunits, so let us look to find the prop-
erties of
N
ut= E1Ujt
Using the factor decomposition that was developed for ii,, yields
XN        l    XN       l           N             N 



138                                      KRISHNA RAMASWAMY
which shows that the unanticipated demand on the parent's deficit
coverage (a) depends on the same K factors, and (b) is also exposed to
the sum of the idiosyncratic shocks to each department's budget that is
unrelated to the K factors.
Defining the K separate sensitivities at the parent level as
N
Bk=(     bikJ  k = 1, 2, . . ., K
reveals that the sensitivity to the k-th factor at the aggregate level is
the sum of the sensitivities at all the subunit levels. The aggregate
unanticipated deficit can then be written as
K          N
ut XBk fk + (     E£).
k   =  I   i= I
The following observations regarding U,, the unanticipated budget
expenditures at the aggregate level, are now in order:
* Summing the exposures across each of the K factors would reveal
that for some cases the risk exposures are reduced in the aggregate.
* Assuming the residual risks E, at the subunit levels are not corre-
lated across subunits, one finds that the risk unrelated to the factors at
the aggregate level has a smaller effect on the aggregate risk.
* What are the properties of the distribution of the aggregate bud-
get risk? Earlier in the chapter, I define in broad terms a number (by
analogy to value at risk, VAR) that applies to a percentile of the distri-
bution of U,. Unlike the corresponding VAR measures, which are ap-
plied to the potential cash flow gains and losses over the next day or
week, the fiscal risk measures described here apply to the risk levels
faced by the government unit over the next budget period.
Aggregate Budget Deficit (Positive) or Surplus (Negative)
Prob p
-3     -2     -1      0      1      2   C  3
Aggregate deficit ($ billions)



ANALYTICAL TECHNIQUES                                    139
In the graph, an aggregate deficit figure larger than 4 = $2.5B is
seen with probability p. I have used the distribution of the aggregate
deficit as a bell-shaped curve only as an example.
This measure is best suited to timne periods and situations in which
the risks do not change-that is, the VAR number is likely to be accu-
rate only for short periods and for portfolios whose risk characteristics
do not change during the periods. If the portfolio has options posi-
tions, or nonlinear characteristics, then these assumptions can be vio-
lated. In the context of government budgets and fiscal risks, the budget
periods are typically annual (at best quarterly), and there may be con-
tingent liabilities that have option-like characteristics. For these rea-
sons, the at-risk figure computed as described above will not be accurate
and should only be taken as an approximation.
* Some of the K factors (say the first m factors, k = 1, 2, . . ., m)
may be related to traded assets, and the remaining factors (k = m + 1,
m + 2, . . . , K) may be important economic influences that are not
traded. In that case, what would be the risk levels if the decision were
made to hedge some of the traded risks-perhaps with the use of de-
rivatives for which there is a market?
7'he Factor Model: A Numerical Example
A, simple numerical example will illustrate the workings of a factor
model. Assume that there are four government units or state-owned
enterprises (SOEs), and that two common economic factors affect the
cash flows of these units-GNP growth ( f) and the average level of
energy prices (f2). Each government unit prepares a budget for the
coming year, and each unit's budget process builds in certain baseline
assumptions about GNP growth and the expected energy price level.
The realized GNP growth rate and, the actual energy prices will consti-
tute departures from these baseline assumptions and will therefore lead
to surpluses and deficits relative to the budget amounts in each unit's
planned cash flow. These baseline assumptions are provided by a cen-
tral forecasting service, and, for the sake of argument, suppose that
these assumptions include a forecast of the standard deviation of the
forecast percentage of GNP growth (a, = 0.5%) and the standard de-
viation of forecast error in baseline energy prices (aY2 0.7%), and that
these factors are uncorrelated.
Suppose now that the expected budget amount for Unit 1, say, is
$50 million, in accordance with the baseline assumptions. An analysis
of the budget process of this unit reveals that the effect of a 1 percent
departure of GNP growth rate from the baseline expected growth rate
corresponds to an unanticipated $5 million in expenditures, and a 1
percent departure from assumed baseline energy prices gives rise to an
unanticipated $1.2 million in expenditures. These numbers are derived



140                                     KRISHNA RAMASWAMY
by tracing their effect through the budget planning process. In the no-
tation of the previous section, b, = 5 and b,2 = 1.2 for Unit i = 1. In
addition, other idiosyncratic factors affect the cash flows of Unit 1
that are unrelated to the two common factors. Assume that the stan-
dard deviation of the idiosyncratic risk in the budget for Unit 1 is $2
million. In the notation established above, the idiosyncratic risk in
Unit l's budget is Var(1, ) =(1,,) = $2M.
These three quantities-the sensitivities b.,, b, to the two factors, plus
the idiosyncratic standard deviation Y(i,, )-constitute the numbers that
must be assessed for each unit, from the information embedded in the
budget process. Listed in the table below are the numbers for the nu-
merical example.
GNP            Energy      Idiosyncratic    Budget
sensitivity     sensitivity      risk          level
Item        b.1              b,           o(e)          E(C,)
Unit 1        5              1.2           2.0          -25
Unit 2        1              0.5           0.5           10
Unit 3        0              1.5           2.5           15
Unit 4      -12              0.1           4.0           30
Total        -6              3.3       Not applicable    30
An explanation of the entries in the table follows:
* The last column of the table contains the expected budget level of
the unit, in millions of dollars. Only the first unit is revenue-produc-
ing; the others expect to incur deficits. However, the factor model
applies to the unanticipated component-the risk in the aggregate defi-
cit-which is the departure from the expected levels. The total (aggre-
gate) expected budget deficit is $30 million, as shown in the last row.
* Unit 4 has a negative elasticity to the GNP growth rate; if the
economy grows faster than expected, by 1 percent, then the budget
deficit actually declines in that unit by $12 million. But the forecast
also acknowledges that the idiosyncratic risk in that unit's cash flows
is also highest, while it is least sensitive to energy prices. This unit
might have responsibilities that devolve, for example, on health or
education, or welfare, so that its expenditures might be reduced if the
economy does well.
* Unit 4 has no exposure to GNP; its risk draws entirely from energy-
related and idiosyncratic risks.
* The last row provides the total aggregate exposure to the two
common factors. Note that the aggregate exposure to the energy price
factor simply adds up and that no unit provides a diversifying expo-
sure to energy prices. In contrast, Unit 4 has a large negative exposure



ANALYTICAL TECHNIQUES                                     141
to GNP, so that the aggregate exposure is actually negative. In the
notation of the section above, B, = -6 and B2 = 3.3.
* Finally, note that the standard deviations of the idiosyncratic risks
do not add up, so it makes no sense to sum them.
To compute the budget risks for each of the units and for the parent
unit, take the numbers corresponding to the standard deviations of the
two factors provided by the central forecasting service, aY = 0.5% and
G2 = 0.7%. Then the variance of Unit l's budget can be computed as
Var(Cit) = b2 I 121 + b12(  +  (i,
which computes to
Var(C,,) = (52 X 0.52) + (1.22 x 0.72) + 22 = 10.96,
implying that Unit l's budget on a stand-alone basis (its own total
risk) has a standard deviation of ^F10.96 = $3.31M. It can be verified
using the numbers in the table that the stand-alone risks of Units 2, 3,
and 4 are given by the standard deviation levels of $0.79 million, $2.71
million, and $7.21 million, respecrively. As expected, Unit 4 shows the
highest overall risk level.
Thus the expected budget expenditure for Unit 1 is revenue of $25
million, which translates into the statement that the realized revenue is
forecast to fluctuate in the region $25M ± 3.31M = $21.69M to $28.31M
two-thirds of the time, assuming a normal distribution of factor and
idiosyncratic influences. Similarly. Unit 4's budget will show a deficit
ranging from $22.79 million to $37.21 million in two out of three years.
But the common factor exposures at the individual unit level ag-
gregate to provide total exposures as shown in the last row of the
table. The aggregate deficit can now be computed as
i=4
Var(Aggr Def.) = B 2ol + B2C2 +  a (E
which when we use the numbers yields
40.84 = ((-6)2 x 0.5 2) + ((3.3)2 x 0.72) + {22 + 0.52 + 2.52 + 42)
The standard deviation at the aggregate level works out to 4084
= $6.39M, implying that the parent will feel a deficit in the range of
$30M ± 6.39M = $23.31M to $36.39M in two out of three years. As
this stylized example shows, an averaging of factor exposures reduces
the exposure to GNP fluctuations, and the diversification of idiosyn-
cratic influences occurs as well. Indeed, the risk at the parent level is
less than the total stand-alone risk felt by the managers at Unit 4.



142                                       KRISHNA RAMASWAMY
Notes
1. See, for example, Chapter 4 by Suresh M. Sundaresan and Chapter 13
by George Pennacchi in this volume. I will restrict the discussion in this
chapter to analytical methods that follow up on the factor model just intro-
duced.
2. There is a close correspondence between the practice of stress testing
the budgets and this effective VAR number.
3. The most immediate way to increase risk is to leverage the portfolio by
borrowing.
4. In some arrangements, the contract for insuring the risk buries the cost
of hedging in a net payment that occurs later-that is, the cost of insurance is
paid at the end, in favorable outcome states, so no up-front fee is needed. But
in those cases, the insurance is not free and must still be budgeted for in a
fiscal context.
5. It would require the government to take a "short" position in the mar-
ket-to enable it to receive cash when aggregate output is below expecta-
tions-and in any case this would require a reliable positive correlation
between output and market values, which is not typically observed.
6. In the event that the no-arbitrage assumptions are met, one can com-
pute the expectation of the payoff under the guarantee in a risk-neutral
economy and discount it at the risk-free rate.
References
Arrow, K. J. 1971. Essays in the Theory of Risk Bearing. Chicago: Markham.
Arrow, K. J., and R. C. Lind. 1970."Uncertainty and the Evaluation of Pub-
lic Investments." American Economic Review 60: 364-78.
Lewis, Christopher, and Ashoka Mody. 1997. "The Management of Contin-
gent Liabilities: A Risk Management Framework for National Govern-
ments." In Timothy Irwin, Michael Klein, Guillermo Perry, and Mateen
Thobani, eds., Dealing with Public Risk in Private Infrastructure. World
Bank Latin American and Caribbean Studies. Washington, D.C.: World
Bank.
Merton, R. C. 1990. Continuous Time Finance. Cambridge, Mass.: Blackwell.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
for Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.
Sharpe, W. F. 1981. "Some Factors in NYSE Security Returns: 1931-79."
Journal of Portfolio Management 8 (4): 5-19.



CHAPTER 6
Fiscal Sustainability and
a Contingency Trust Fund
Daniel Cohen
Universite de Paris, Ecole normale superieure
Tlins CHAPTER SETS UP A simple framework to analyze the conse-
quences of contingent liabilities for the management of public finances.
From this framework, it suggests the design of an institutional solu-
tion to the problem of fiscal opportunism in government when dealing
with contingent liabilities.'
The idea developed here is that government should handle contin-
gent liabilities by setting up an independent trust fund that itself will
have limited liabilities. The government is expected to be the key, but
not necessarily the only, shareholder of the fund. When the fund is
established, it is endowed with capital that corresponds to the overall
size of the government commitmenit. A government would finance this
endowment by raising correspondingly its debt. This first stage has the
merit of imposing on government decisionmaking a one-to-one rela-
tionship between the government's commitments to the private sector
and its own liabilities to the financial market. As a shareholder of the
trust fund, on the other hand, the government can expect to receive
dividends. The dividends should be set so that they keep constant (say,
in real terms) the resources that are managed by the trust fund-in
effect, maintaining a constant level of guarantees that the fund is en-
titled to offer. Under these circumstances, the government needs only
to raise taxes by an amount that corresponds to the difference between
the debt it has issued and the "market value" of the trust fund. The
market value of the trust fund corresponds in theory to the difference
between the resources with which it has been endowed and the cost it
is expected to cover. Raising taxes to the difference between the initial
143



144                                           DANIEL COHEN
endowment of the trust fund and what it should be expected, on aver-
age, to pay is exactly what the theory of optimal taxation suggests.
There are two merits to and two critical questions about this scheme.
The first merit is that it forces the government to internalize immedi-
ately the cost of the scheme it wants to implement. The second merit is
that the limited liability nature of the trust fund puts an explicit ceiling
on government involvement. If risks are initially underreported, this
would not represent an additional burden to the government, unless it
willingly accepts the idea of sharing the extra burden with the private
sector. This leads to the first question. How can the government ap-
propriately assess the market value of its commitments? This question
can be partially answered by dragging in the private sector. If shares of
the trust fund were to be marketed, the information revealed by the
market could then be channeled to the government, allowing it to fi-
nance adequately the net burden that this commitment adds to those
of its other commitments. The second question is associated with the
nature of the risks involved. Should the government set up one trust
fund for all contingent liabilities, or one for each category of risk? In
principle, it is optimal to set up one trust fund for each category of
risk. As I will argue, however, pooling risks tends to lower the cost to
the government and is therefore bad from an allocative perspective,
but good from a financial perspective.
The chapter proceeds as follows. I first briefly summarize the frame-
work of the theory of government solvency. I then examine how it
should be extended to encompass contingent liabilities. Finally, I ana-
lyze the merits (and the limits) of the institutional framework I have
sketched.
The Basic Analytical Framework
An Intertemporal Approach to Government Solvency
The simplest framework for analyzing fiscal sustainability is to follow
the law of motion of government debt over time (all technical details
are given in the Appendix to this chapter). The debt buildup of the
government stems from the primary deficit (which is simply the differ-
ence between government expenditures other than interest payments
and government revenues) and the interest bill itself. While the former
can, in principle, be changed at will, the latter is the legacy of the past
and can only be changed with time. This explains how a profligate
government can exhaust the solvency of its successors by leaving out a
stock of debt that might turn out to be "unsustainable" (on these is-
sues, see, for example, Buiter 1985 and Cohen 1991).



FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND        145
A direct way to assess whether a government is solvent is to write,
in present value terms, all items that will contribute to the buildup or
to the reduction of the debt in the future. (The present value terms
simply weight any item forthcoming at a time in the future by the
inverse of the compounded interest that applies up to that time.) The
increase in the present value of the debt between any initial time and
the future can then be written straightforwardly as the sum of the
present values of forthcoming primary surpluses. Solvency thus can be
defined as follows: a government is solvent if and only if it is able to
generate in the future a stream of primary surpluses that are sufficient
to repay (in present value terms) t:he stock of outstanding debt that it
has inherited from its predecessors. It is a straightforward analytical
exercise to demonstrate that such a condition will hold if and only if
the present discounted value of the debt will itself converge toward
zero in the long run. Indeed, if this is the case this condition implies
that the debt is "repaid" asymptotically and that there will always be
a chain of investors willing to rel'inance the government (so long as
they are certain that the government will continue to implement the
set of primary surpluses that generate such an outcome).
Incorporating Contingent Liabilities
The above principles address the case in which a government can com-
rnit itself to a stream of primary surpluses that can handle the debt it
has inherited. Let us now see how these principles should be amended
to encompass the case of contingent liabilities. In order to do so, let us
assume that the government grants the private sector a guarantee against
a one-off event such as an earthquake or a banking crisis. To simplify
t:he presentation, let us assume here that the guarantee will cost, if
exercised, a fixed value that is known in advance. Clearly, the cost of
the guarantee must take account of the probability that the shock will
occur, but also, and most important, the likely timing of the shock
itself. Assume that the shock is rnost likely to occur in 10 years, al-
though it might also occur, with a lower probability, in 2 or 20 years.
The government therefore should budget the contingent liabilities based
on the present value, mostly for a 10-year horizon, of the expenses
that will be involved on average.
So long as interest rates are positive, this present value will be less
than the expected value of the cost itself. Clearly, there are only two
situations in which the expected cost (guarantee times probability of
exercising it) may not be the right measure of the burden involved:
when the time horizon over which the contingent liability is expected
to be exercised is very short, or when the interest rate is very low. In
each of these two extreme cases, the government should simply budget



146                                            DANIEL COHEN
the expected value of the cost (cost times probability of occurrence).
In all other intermediate cases, it should budget a lower number.
This simple result is at odds with the conventional idea that any
contingent liability should be incorporated in the budget either at face
value or at a lower number that is measured simply by the likelihood
that the guarantee will be exercised. The basic principles just presented
reveal that the expected timing of the event is crucial. Events that are
certain to occur sometime in the future may cost less than events that
might occur with a probability lower than one, but fairly soon if they
do occur. Take the example of a financial crisis affecting the banking
sector. It is a sure thing in most developing countries that such crises
will occur. No one expects it to happen tomorrow morning, however,
which allows the government to take actions progressively in order to
meet the expected cost (see below). Imagine instead that the govern-
ment extends a guarantee against the risk that the Y2K bug will dis-
rupt the payment system. Here the probability is certainly smaller than
one, but the timing of the event is very sharp: it might cost the govern-
ment more to provision the latter risk than the former.
Practical Implications
How should the government handle the financing of contingent li-
abilities in practical terms? Two different methods come to mind. One
option is to wait until the guarantee is exercised. At that point, the
debt is increased by an amount corresponding to the guarantee to be
exercised. Another option is to raise taxes preventively in order to
match the market value of the expenses that are generated by the con-
tingent liability. If government pays any attention to smoothing taxa-
tion, as is usually the case when taxation is distortionary, this latter
option is bound to dominate the former one. At this point, however,
the complexity of the political process plays a critical role. From the
perspective of a current government that does not internalize the wel-
fare of its successors, the first scheme is likely to appear preferable, at
the expense of social efficiency. (See Polackova 1998 and Kharas and
Mishra 2001 for a more detailed discussion of contingent liabilities
and their fiscal implications.)
Government Risk and Institutional Designs
The Risk of Government Repudiation
In most developing countries, a state of fiscal crisis is almost a fact of
life, making it difficult for the government to guarantee contracts in
the future, when its own debt is quite often at risk. In order to give



FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND       147
some empirical insights into the extent of such risk, I calculated the
primary surpluses of three European countries-Belgium, Italy, and
Ireland-that faced deep financial problems in the 1980s but were
successfully dragged out of their crises. To summarize, in the 1980s
Belgium's government surplus as a fraction of the tax receipts of the
high debtors of the Organisation for Economic Co-operation and De-
vetopment (OECD) was 10 percent. Italy's corresponding surplus was
12 percent and Ireland's 15 percent.
These numbers, which are not far apart, point to the idea that the
"financial crisis" of a state originates from a relatively low ratio of
government surpluses to tax receipts. Contingent liabilities such as
protecting the banking sector or guaranteeing some major environ-
mental damage often involve costs that are typically larger than the
immediate financing capacity of thie government. In such cases, the
government would not be willing to honor its previous commitment
and the entire scheme would collapse.
Institutional Designs
If the government's signature is at risk, the guarantee that it granted
will not necessarily be honored. Take the extreme case in which the
risk that is protected corresponds to the risk of government default
(such as a "systemic" risk that threatens the banking sector and simul-
taneously endangers public finances). The private sector will then never
be protected ex post and government guarantees will be pointless. In
such a case, there is room for institutional improvements that will
protect the private sector against the risk of government default. One
simple way to address this issue is to set up an independent trust fund
that is endowed by the government by an amount that is explicitly
geared toward protecting the privare risk. Let us now review how such
a scheme should proceed.
In terms of credibility, the easiest way to proceed is to endow the
trust fund with an amount that corresponds to the risk that is pro-
tected and to make the government the shareholder of the fund. This
one-off transfer is financed through an increase in government debt.
As long as the risk does not materialize, the fund makes profits that
can be captured by the government (essentially in the same way that a
central bank's profits are redeemed by the government). Once the risk
materializes, the trust fund extends its guarantee to the private sector,
regardless of the government's financial situation. The fact that the
trust fund capitalizes on its assets as long as the risk has not material-
ized means that the government receives a transfer every period, as
dividends, as long as the risk does not show up. Through such a policy
(assumed to be designed in real terms), the fund's resources are kept to
the initial level. In present value terms, this strategy corresponds to a



148                                          DANIEL COHEN
net receipt, which can be subtracted from the direct cost originally
involved in the transfer. This leaves the government with a net burden
that obviously corresponds with the number involved in the first evalu-
ation of the cost, and that must be matched by raising taxes by the
corresponding expected amount. Clearly, then, the optimal taxation
scheme that I alluded to earlier in this chapter can be readily delivered
by this scheme.
Bailing in the Private Sector
There is no reason why the government should be the only shareholder
of the trust fund. In particular, when there is uncertainty about the
nature of the risk involved it may be critical that the government build
on the private sector's knowledge.
Assume, for example, that the risk covered by the government ma-
terializes with a probability that is common knowledge among private
agents but not shared with government officials. If shares of the trust
fund are marketable, the government can immediately observe the full
value of the trust fund and needs only to finance the residual. In prac-
tice, obviously, the government needs only to trade a few shares in
order to capture the information on the value of the risk. This infor-
mation can be immediately acknowledged in the government's account.
In that case, the government simply has to raise taxes to take account
of the net burden. Clearly, however, nothing prevents the government
from going further and perhaps selling the entire project to the private
sector. In such a situation, the net cost is readily measured by the one-
off difference between the level of the guarantee and the market value
of the trust fund, and the optimal taxation scheme is easy to imple-
ment. This situation corresponds to many practical examples such as
that of an export credit agency. The government could well help the
private sector to create such an agency that insures exporters against
the risk of foreign default, and then privatize the agency itself. The
government's involvement is limited to a one-off endowment that can
be properly financed. The same might be true of a guarantee on the
banking sector. The government could set up a version of the U.S.
Federal Deposit Insurance Corporation (FDIC) aimed at guaranteeing
deposits and then let the banking community manage and sustain the
agency later on.
Pay as You Go
The trust fund scheme also can be applied to the cases in which the
government is involved in repeated transactions (for example, taking
care of pensions) rather than one-off mechanisms only. If the govern-



FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND        149
ment rolls over every period new guarantees as the old ones are exer-
cised, the nature of the game becomes different. But one can still rely
on a trust fund to protect goverm-nent finance from the risk that the
scheme is misconceived. Assume that the government is willing to com-
mit itself up to a given amount and is willing to renew its commit-
ments in order to keep that amount constant over the years. Rather
than self-managing this set of guarantees, the government could as
well create a fund, endowed appropriately each period, that would
monitor the risks. This arrangement would have the virtue of protect-
ing government finances from unwvanted shifts in the underlying prob-
abilities. If the projects involved are riskier than initially thought, the
private sector will bear the consequences, not the public finances. This
may not be an optimal risk-sharing method, but when the government's
solvency is threatened by an unwanted increase in public expenditures,
it might be the best way to go. Similarly, one could drag in the private
sector. Think of a trade credit agency that the government wants to
help by granting guarantees against adverse events. The government
could endow the trade agency wit:h a given flow of resources and let
the private sector bring some additional collateral. If the risk is ap-
propriately measured, then the scheme is neutral to the private sector.
If it is underreported, then the collateral brought by the private sector
will be lost (at least in part). This could form the basis of a truthful
exchange of information. At a minimum, the private sector would not
want to underestimate the fundamental risk. In fact, it would want to
overestimate that risk, but this would then immediately raise govern-
mnent involvement and would have a deterrence effect.
Merging Risks
Until now, it was implicitly assumed that the government had to deal
only with one category of risks. Assume instead that there are two
classes of risk (the general case would proceed immediately): good
risks and bad risks. Good risks that fall due later (on average) than
bad risks are less costly to guarantee. In theory, the optimal strategy
for the government would be to pledge these two risks, each with a
different trust fund. There is, however, an interesting paradox here:
not being able to monitor the two classes of risk reduces the cost of
operating the trust fund, in a situation in which each class of risk is
managed separately. The intuition behind this apparently paradoxical
result is that the bad risks disappear more rapidly than the good ones,
so that the process by which the risks are reallocated regularly benefits
the good ones. It is then a safe strategy, from the government perspec-
tive, to pool risk, provided, of course, that the usual adverse selection
biases are taken care of.



150                                           DANIEL COHEN
Conclusion
Whatever the precise institutional setting that one has in mind, it should
be clear that the proper management of contingent liabilities imposes
two imperative constraints. One is a full reporting of the cost involved.
Another is the proper internalization of these costs by the government.
The specific mechanism proposed here is one in which the full cost is
determined by setting up a limited liability framework that imposes,
by fiat, an upper limit to the scope of government intervention. The
proper internalization by government proceeds from the fact that the
financing should be up-front. Although other, somehow less extreme
methods may be possible, it is nonetheless critical that the government
acknowledge ex ante the fiscal implications of the guarantees involved.
Intervening ex post is not only bad from the point of view of the tax
system (which would suffer from the strains of increased expenses),
but also bad for the sectors that might feel protected by government
guarantees to discover only too late that governments do default re-
peatedly on earlier commitments.
Annex 6.1
Fiscal Sustainability and Contingent Liabilities
The simplest framework for analyzing fiscal sustainability is one in
which all government debt is short term, the interest rate is a constant
r, and the economy is deterministic.2 Before relaxing these assump-
tions, one by one, let us first investigate how fiscal sustainability can
be analyzed in this highly simplified environment.
Call D, the stock of government debt at time t, G, government ex-
penditures, and T, the amount of tax collections. The law of motion of
government debt can then be written as
(6.1)              D, = (1 + r)D,, + G,-.
(G,- T ) is simply the primary deficit of the government, and rD,1 +
G,- T, represents the overall deficit. At this stage, equation 6.1 is noth-
ing but an accounting identity that states that the increase in govern-
ment debt amounts to the overall deficit. To see how this equation can
be turned into a constraint imposed on government finances, let us
write (6.1) in present value terms, from the perspective of an initial
(and arbitrary) initial time t = 0. Dividing both sides of the equation by
the discount factor 1/(1 + r)' produces
(6.2)             D.T       D' X_ I  GI - T,
(6.2)       ~~(1 +r)t  (+ r)'-   (1 +r)t



FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND      151
which has a straightforward interpretation: in present value terms the
increase in debt is worth the present value of the primary deficit. Add-
ing up all such equations 6.2 from ttie initial time t = 0 to an arbitrary
time T in the future yields
DT          ~T GI- T
(6-3)          (   )T = D_1(1 + r) = E I t)
(1 +r)    -         =0 (1 +r)'
The increase of the debt, in present value terms, is the sum of the
present values of the primary deficit. When T is allowed to go to infin-
ity, this yields
lim   DT          + r),
(6.4)            (   T = D 1( + r) = E
T-- (1  r)T       t = 0 (1 +r)
The asymptotic value of the present value of the debt is worth the
sum of all past primary deficits and the initial debt.
The straightforward definition of solvency that is then imposed on
the government is called the "transversality condition"-namely
( 6.5s )              lim   D T    O
(6.5)          ~~T-- (1 +r)T
This condition states that the government should let the present
value of its debt go to zero in the long run. Once this condition is
granted, one can see from equation 6.4 that the government is solvent
in the following sense that
T                        GT
(6.6)        E i        = D_1(i + r) + E  -
(6.6)    ~t =0 (1 +r)t          t= 0 (1 + r)t
The sum of all tax collections 1:o come, when written in present
value terms, must match the initial value of the debt accumulated by
the government and the sum of all present values of future government
expenditures. The beauty of the transversality condition is that a simple
requirement on long-run debt is equivalent to a complex analysis of
the determinant of future tax and expenses.
The Case of a Growing Econonly
Consider the simple case of a growing economy in which the tax capa-
bility of the government is driven by an exogenous law
(6.7)                  T, = T0(1 + n)'



152                                          DANIEL COHEN
in which n is the underlying growth rate of the economy. Assume fur-
thermore that the interest rate is larger than the growth rate-that is,
(6.8)                      r > n.
In order to simplify the analysis, also assume that government ex-
penditures grow at the same exogenous rate n-that is,
(6.9)                  G, = GO(1 + n)'
One can then write the present value of the primary deficits as
(6.10)           Il        = (Go - To) 1 +-r
0 (I+ r)t            -
so that the government is solvent if and only if
D  To - Go
(6.11)                 D  <
Assume that G is lifted up to the limit of what the government can
afford. Then the primary surplus of the government must be exactly
equal to
(6.12)               To - Go = (r- n)D .
One can straightforwardly see that debt is exactly growing at the
rate n, so that at any later time one has
(6.13)               T,- G, = (r-n)D,
and
(6.14)                D, = (1 + n)D,,.
Clearly, this implies that
(6.15)    ~lim _  __= Do lim ~      I0
(6.15)        to -  (1 + )t =   Do t -(I + r ) = °
The solvency condition does not then imply that debt should be
stabilized. Indeed, in this critical example, debt is growing to infinity.
It does impose, however, that debt should not be growing faster than
tax receipts. In other words, it imposes that the debt-to-tax ratio should
be bounded.



FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND      153
Contingent Liabilities
Assume that the guarantee has pledged to insure a risk that is expected
to be exercised according to a Poisson process of parameter k. This
means simply that the probability that the guarantee will be called
upon at time t is ?I(1 + X)'; t 2 1. Call K the value of the government
guarantee. From a market perspect:ive, the value of the grant can then
be simply written as
(6.16)               V=ET( + r)T 
in which Er represents the expected value with respect to the time T at
wvhich the government will have to extend its guarantee-that is,
(6.17)       V = E           _-  K =         K
I (1 +X   (+ r)'     (+ r)
The market value V of the guarantee coincides with the amount
involved, K, in two limiting cases: when 1 goes to infinity or when r
goes to zero. In the first case, 1 infinitely large, the guarantee is ex-
pected to be delivered very rapidly; in the second case, the fact that it
will be granted sooner or later is irrelevant if the market puts no weight
on present events compared with later ones. In either of these two
extreme cases, the government should fund the guarantee immediately
and include it at face value as part of government expenses.
In less extreme cases, however., the government should set aside a
lower number that takes account of the fact that the guarantee will be
exercised at some time in the future, thus corresponding to a lower
cost than the face value itself.
In that case, taxes must be raised to the point where they solve for
the new intertemporal budget constraint
(6.18)     i     t   + Do +--x     KT
1 (1 + r)t       (r + X)     1 (+ r)t
where
(6.19)              T =T F+          K
(r +)
Repeated Shocks
Let us now investigate the dynamics that are involved when the gov-
ernment extends continually the set of guarantees that it grants to the



154                                          DANIEL COHEN
private sector. For simplicity, assume here that all the guarantees per-
tain to the same class of risks, parametrized by X (this hypothesis is
relaxed below). Call K, the value, at any given time t, of the outstand-
ing stock of guarantees that are granted by the government. In market
terms, the value V, of these commitments can be written simply as
(6.20)        V, =     I         K,   + V,
(1 +r) [(+?.)
the sum of the present value of the payments that the government is
expected to make in the next period and of the remaining commit-
ments one period later. Furthermore, if X, is the flows of new contin-
gent liabilities that the government extends every period, the outstanding
stock of government guarantees can be written as
(6.21)           Kt+, - Kt = -      Kt + Xt
(I +X)
since a fraction 2I(1 + X) is redeemed each period and a new flow X, is
added to the total.
Referring to (6.20) and (6.21), specifically expressing K,, in the
(6.21) and substituting it for K,, in (6.20), one can write a general
formula that measures the cost (in present value terms) to the budget
of such a government policy:
(6.22)  Vo =  K         1~+~ Y         '
(r + X)   t =1(1 + r)t(1 + k)' [s= I (1 + X)]
The intuition behind such a formula can be readily conveyed by
examining a few relatively simple cases. For example, when X, = 0, the
government lets the stock of guarantees be depleted as time passes.
This yields the same result obtained in the previous subsection:
(6.23)                 V=         K
(r + X)
Now consider the case in which the government keeps replenishing
the stock of contingent liabilities as they are exercised in order to main-
tain the stock of long-term liabilities at a given level K. The overall
cost of such a policy can be simply written as
(6.24)                 V=         K
r(l + X)



FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND      155
which then has a straightforward policy implication: the government
has to raise taxes by the amount
(6.25)               AT=rV=-         K
(1 + X)
to be solvent. This is a version of the pay-as-you-go system, because
now the government pays the contingent liabilities as they get going.
Two polar cases are thus emerging. When the government is com-
mitted to a one-off guarantee, funding its scheme is essential to avoid
a one-off increase in taxes when the time comes to exercise the guar-
anree. But when the government is continually rolling over the amount
of its contingent liabilities, then a pay-as-you-go system is fine in theory,
provided that an explicit limit is set on the amount of the government's
guarantees.
Merging Risks
Assume that there are two classes of risk: good risks that fall due with
a Poisson process 1,, and bad risks that fall due with a Poisson process
?2 with X2 > , (bad risks come earlier on). If K, is the amount commit-
ted to the good risk and K2 the amount committed to the bad risk, then
the government must endow each fund with
(6.26)             Fl =   1  ; F2 =  r22
rr
or equivalently pledge to each of the two trust funds annual transfers
z1=rF1=XK1 and Z2 =rF2 =XK2.
Assume, however, that the government cannot differentiate between
the two groups, and furthermore assume that the government is main-
taining a constant level of commitments, as they were exercised, to the
private sector, no matter which one of the two risks is covered. The
value of such commitments can be written as
(627a)=1 + r) IVI 1 +        ;K (t + 12) K2 (t + 1)
(.7   t+)=   1-K t :             I(+X1)
(6.27b)      K, (t + 1) = - __ - K, (t) +XI (t + 1)
(1 + X1)



156                                         DANIEL COHEN
(6.27c)      K, (t + 1)= k-     K2(t) +X2(t+1
(6.2 7c)         ~~~(1 + K2)I(()X(t+1
(6.27d)     X(t) _ Xl(t) + X2(t) = X,K1(t) + X2K2(t).
In this model, the outstanding level of government guarantees K(t)
+ K2(t) = K(t) is maintained at a constant through government deci-
sions, but the internal composition of these guarantees is left to its
own dynamics. Call 0 the proportion of good risk in the economy and
assume that the reallocation of new guarantees is extended accord-
ingly-that is,
(6.28a)                X,(t) = OX(t)
(6.28b)              X,(t) = (1 - 0)X(t)
in which
(6.29a)           X,(t) = 01X,K,(t) + X2K2(t)]
(6.29b)        X2(t) = (1 - 0)[?,K,(t) + k2K2(t)].
Asymptotically, the composition of the fund will evolve toward
(6.30a)             ()       - 0) + 2K
(6.30b)           K1(=    0)K
The total asymptotic cost of the fund will then be
(6.31)     ~r [=           l (1 - 3 ) + k2]
while monitoring the two risks separately would involve the cost
(6.32)            F' = 1 10?,l + (1 -O)X2]K.
The cost of managing the merged trust fund is strictly below the
cost of monitoring each of the two funds separately. The intuition is
given in the text.



FISCAL SUSTAINABILITY AND A CONTINGENCY TRUST FUND            157
Notes
1. The problem of fiscal opportunism encompasses several aspects. Gov-
ernments whose time horizon is rarely far beyond the next election usually
enjoy extending contingent liabilities because they are not counted as a cur-
rent cost. The short time horizon also implies that governments tend to un-
clerrate the risk of their contingent liabilities and, particularly, make inadequate
provisions for future contingent pavments. Similarly, contingent liabilities
expose the government to a risk of default that is not necessarily a source of
concern for financial markets. A government's repudiation of its contingent
liabilities does not necessarily imply that it will default on other debts. There-
fore, under standard conditions no strong incentives are working against
government risk-taking and toward proper provisioning for risk. In this vol-
uime, issues of fiscal opportunism are discussed in more detail in the Intro-
duction and in Chapter 1 by Hana Polackova Brixi and Ashoka Mody and
Chapter 3 by Allen Schick.
2. I thank Hana Polackova Brixi for very helpful comments on an earlier
draft. Errors are mine.
References
Buiter, Willem. 1985. "A Guide to Private Sector Debt and Deficits." Eco-
nomic Policy (November): 14-79.
Cohen, Daniel. 1991. Private Lending to Sovereign States. Cambridge, Mass.:
MIT Press.
Kharas, Homi, and Deepak Mishra. 2001. "Fiscal Policy, Hidden Deficits,
and Currency Crises." In S. Devarajan, F. H. Rogers, and L. Squire, eds.
World Bank Economists' Forum. Washington, D.C.: World Bank.
Polackova, Hana. 1998. "Contingent Government Liabilities: A Hidden Risk
to Fiscal Stability." Policy Research Working Paper 1989. World Bank,
Washington, D.C.






CHAPTER 7
A Framework for Assessing
Fiscal Vulnerability
Richard Hemming
International Monetary Fund
Murray Petrie
The Economics and Strategy Group
FISCAL VULNERABILITY GOES BEYOND a situation in which a govern-
ment currently pursues inappropriate fiscal policies, or it lacks the
ability to implement better policies.' While obviously bad policies and
a lack of implementation capacity will almost always signal vulner-
ability, governments that seem well positioned in relation to both may
nonetheless find that they are vulnerable in two respects. First, under-
lying weaknesses may be present that do not affect fiscal outcomes
today but could at some time in the future prevent a government from
achieving its fiscal policy objectives. And, second, such weaknesses
may limit a government's ability to respond to future fiscal policy chal-
lenges, such as the need for fiscal consolidation as part of a coordi-
nated policy response to an external shock.
The focus in this chapter is on fiscal vulnerability from a macro-
economic perspective, and the suggested framework for assessing vul-
nerability highlights four macrofiscal aspects of vulnerability: (a)
incorrect specification of the initial fiscal position; (b) sensitivity of
short-term fiscal outcomes to risk; (c) threats to longer-term fiscal
sustainability; and (d) structural or institutional weaknesses affect-
ing the design and implementation of fiscal policy. The framework is
intended primarily to be a basis for identifying situations in which an
anticipatory response to potentially poor macrofiscal outcomes is
159



160                   RICHARD HEMMING AND MURRAY PETRIE
required. But it also goes beyond this point, because a clear link
exists between fiscal vulnerability and economic vulnerability more
generally.2 This being the case, the importance of assessing fiscal
vulnerability also derives from its contribution to effective macro-
economic surveillance. The framework suggested in this chapter can
be viewed as providing guidance that will assist countries in assess-
ing their fiscal vulnerability.
Work on fiscal vulnerability has paralleled work on financial sector
vulnerability (see Downes, Marston, and Otker 1999). Strong links
exist between fiscal and financial sector vulnerability, because fiscal
vulnerability can manifest itself as a financial sector problem (such as
government preemption of bank lending on concessional terms), while
addressing a financial sector problem can be a source of fiscal vulner-
ability (such as government support for bank restructuring). A new
financial architecture should pay attention to both fiscal and financial
sector vulnerability, although the former has so far attracted much less
attention.
Reflecting the fact that a lack of transparency can be a major source
of vulnerability, there is a considerable overlap between assessments
of fiscal vulnerability and assessments of fiscal transparency. For this
reason, many of the issues discussed in this chapter are also covered by
the IMF's Code of Good Practices on Fiscal Transparency (the trans-
parency code) issued by the International Montetary Fund (IMF) and
taken up in more detail in IMF's Manual on Fiscal Transparency (the
transparency manual)-see IMF (2001). Indeed, it is inevitable that
vulnerability assessments will be informed by fiscal transparency as-
sessments included in Reports on the Observance of Standards and
Codes (ROSCs), for those countries where they are available, or un-
dertaken independently of the ROSC process. Transparency is also
important, because a prerequisite for a vulnerability assessment is a
reasonable degree of transparency. It would be difficult to assess the
vulnerability of a totally nontransparent fiscal system. This chapter
covers certain parts of the transparency material that are absolutely
central to vulnerability assessments, both in the sense of identifying
sources of vulnerability and in facilitating assessments.
The chapter is organized as follows. The next section specifies the
fiscal policy objectives that provide a benchmark against which vul-
nerability assessments are made. The third section discusses different
macrofiscal aspects of vulnerability and the methodology for vulner-
ability assessments. That section is followed by one that suggests vul-
nerability indicators and provides guidance on their interpretation,
and then by one that addresses some aspects of vulnerability that are
microstructural, as distinct from macrofiscal, in nature. The last sec-
tion offers concluding comments.



A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY            161
Fiscal Polic:y Objectives
Fiscal policy can be viewed as operating at three levels. The first is the
aggregate level, where the concern is with total expenditure and taxa-
tion (or the revenue effort more generally), the overall fiscal balance
and the associated deficit financi[g or use of fiscal surpluses, and the
fiscal consequences of accumulated liabilities and assets. The second is
the sectoral level, where there is a strategic focus on the broad struc-
ture of spending (across major programs) and revenue (mainly across
major tax bases). And the third is the program level, where the empha-
sis is on the microeconomic efficiency of individual spending and tax
programs.3
Fiscal vulnerability can manifest itself and can therefore be assessed
at any of these three levels. This chapter focuses on fiscal vulnerability
in a situation in which a government is exposed to the possibility of
failing to achieve its aggregate fiscal policy (or macrofiscal) objec-
tives. These objectives are:4
1. Foremost, a government should seek to avoid excessive fiscal
deficits and debt; they could directly threaten short-term macroeco-
nomic stability and longer-term fiscal sustainability.
2. A government should ensure that fiscal policy contributes to ef-
fective demand management by retaining sufficient flexibility to re-
spond in an appropriate and timely way to domestic and external
macroeconomic imbalances.
3. A government should raise revenue in a manner consistent with
maintaining reasonable and stable tax rates, which contribute to an
environment that encourages economic activity.5
Fiscal vulnerability could reflect a possible inability to meet any or
all of these macrofiscal objectives. Certainly each is important in its
own right. Thus a fiscal vulnerability assessment might suggest that,
given a government's expenditure plans: the money creation necessary
to finance the fiscal deficit may lead to inflation or a debt sustainability
problem may emerge; fiscal policy will have to be undesirably tight,
and maybe pro-cyclical, during an economic downturn; or tax rates
will have to increase over time to levels that are likely to have signifi-
cant disincentive effects. Any onie of these possible macrofiscal out-
comes would be a source of concern. However, in most circumstances
such outcomes will not be independent. So if deficits and debt are a
concern, providing a fiscal stimulus during a recession will usually be
costly given the high interest rate premia that are imposed when the
fiscal position is weak. And if the need instead is to contain a boom,
the scope to do so may be limit-ed if the room for tax increases has
been exhausted before deficits and debt become a concern.



162                   RICHARD HEMMING AND MURRAY PETRIE
There are also clear interactions between macrofiscal objectives and
what might be called (to distinguish them from macrofiscal objectives)
the microstructural objectives of fiscal policy-that is, objectives set
at the sectoral and program levels. The causality can run both ways.
Weaknesses in the design and operation of spending and tax programs
can and do contribute directly to poor macrofiscal outcomes. At the
same time, if poor macrofiscal outcomes are possible, then it is likely
that some of a government's microstructural objectives will go unmet
as a consequence. There is also the possibility that the aggregate fiscal
policy approach could indicate that a country is not especially vulner-
able, yet at the same time the country is falling short of meeting, for
example, a high-priority poverty alleviation target. Clearly, such an
outcome can create vulnerability in the obvious sense that a govern-
ment that fails to meet such a critical equity goal may have to respond
in a manner that either compromises its macrofiscal objectives (if it
has to spend more on poverty alleviation programs and raise already
high taxes or borrow excessively to pay for the additional spending),
or, if there is little unproductive spending, it will have to cut some
other high-priority programs or shift some spending off-budget (which
simply shifts the source of vulnerability). If a government does neither
of these things, its political survival may be threatened.
Macrofiscal Aspects of Vulnerability
The starting point for meaningful analysis and discussion of fiscal vul-
nerability from an aggregate fiscal policy perspective is a clear view of
the initial fiscal position, both in terms of whether macrofiscal objec-
tives are initially being met and in terms of the quality of the informa-
tion that is available about the initial fiscal position. In particular, it is
important that the initial fiscal position describes the full range of
fiscal activity in the economy. The next step is to develop an under-
standing of the range of possible short-term macrofiscal outcomes by
assessing their sensitivity to underlying risk. The focus should then
shift to government exposure to medium- and longer-term adverse trends
or influences that may affect fiscal sustainability. Finally, attention
should be paid to vulnerabilities that arise from weaknesses in the
structure of public finances, the institutional capacity for fiscal man-
agement, and the broader effectiveness of government. These four as-
pects of fiscal vulnerability provide the organizational structure of the
framework for assessing vulnerability that is suggested in this chapter.
Because assessing vulnerability is a forward-looking endeavor, it
necessarily requires that a view is formed about future economic de-
velopments in general and future fiscal developments in particular. In
this connection, the position taken in this chapter is that fiscal vulner-



A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY            163
ability assessments need to be prudent, in the sense that they should be
prepared using a framework that has a bias toward downside risks. A
prudent approach can be justified by asymmetries in the economic
implications of unfavorable versus favorable outcomes, especially given
the weak initial fiscal positions or deficit bias in many countries., It is
also necessary to lean against a systematic tendency toward optimism
arising from elements of the political environment, such as the short
time horizon of elected officials.
*The Initial Fiscal Position
The initial fiscal position is clearly a source of vulnerability to the
extent that the macrofiscal objectives described above are initially not
met. Thus if the starting position is characterized by high deficits and
debt, an inability to respond to macroeconomic imbalances (for ex-
ample, because automatic stabilizers are small and discretionary fiscal
policies take time to formulate and implement), or very high tax rates,
then there would usually be a presumption of vulnerability. However,
the concern here is more with the possibility that the available fiscal
information may not relate to the full range of fiscal activity that is
undertaken by or on behalf of the government.
It would typically be the case that a description of the initial fiscal
position takes the latest central government budget as its starting point,
and this is reasonable if the budget is realistic. If the budget is not
realistic, the estimated outturn for the preceding year should be the
starting point. If a vulnerability assessment is being made some time
after the budget has been formulated, it would be appropriate to begin
with an estimate of the budget outturn rather than the budget or the
preceding year's outturn. This would allow several factors to be taken
into account, including: new data for the outturn for the preceding
year that render a budget based on an earlier estimate unrealistic; re-
vised macroeconomic forecasts affecting the assumptions underlying
the budget; and other fiscal policy developments (new programs, offi-
cial statements about budgetary policy, and so forth).
From such a starting point, it is crucial to go beyond the budget,
because it is most unusual that the budget captures all fiscal activity:
* Where lower levels of government are large, focus on the general
government.7
* Include extrabudgetary activities of the central government and
lower levels of government.
* Cover quasi-fiscal activities undertaken outside government. The
central bank, public financial institutions, and nonfinancial public
enterprises to varying degrees are all involved in such activities, which
tend to generate implicit or explicit obligations for the government. In



164                   RICHARD HEMMING AND MURRAY PETRIE
many cases, precise quantification may be difficult. A qualitative state-
ment about quasi-fiscal activities would in the first instance suffice,
but rough orders of magnitude should be provided for the main quasi-
fiscal activities and the need for better information should be empha-
sized. Where quasi-fiscal activities are significant, the focus should
shift to the public sector.
Another problem is that fiscal activity is often measured in an unre-
liable or incomplete manner because of weak accounting and control
systems. With poor fiscal data, there will often be large discrepancies
in the fiscal accounts, for example between above-the-line and below-
the-line fiscal balances. The cash accounting traditionally used by gov-
ernments, while having a number of advantages, also has inherent
weaknesses as a measure of fiscal activity. Most notably, it fails to
reflect activities that give rise to arrears and noncash-based provisions
to clear arrears (for example, netting of expenditure arrears and tax
offsets).
It is important to take into account changes in the stock of govern-
ment liabilities and assets. Knowing a government's gross debt is a
minimum requirement for longer-term sustainability analysis, while
information on the structure of debt (that is, maturity, fixed versus
variable interest rates, and currency composition) is needed to assess
short-term fiscal risk. If a government has sizable financial assets, its
net financial debt is more relevant than its gross debt to longer-term
sustainability. And where available, even partial information on other
asset transactions (most usefully on privatization and government in-
vestment in productive assets) would provide some basis for determin-
ing whether the effects of fiscal policy actions that show up as changes
in aggregate revenue and expenditure, and the resulting changes in the
fiscal deficit, are being accompanied by changes on the government's
balance sheet that might undo their impact.8
Explicit and implicit contingent liabilities also should be covered.
The Fiscal Risk Matrix in Chapter 1 of this volume outlines the classi-
fication and examples of such government obligations. Where there is
provisioning against contingent liabilities-for example, a deposit in-
surance scheme-the level of provisioning should be noted and the
focus should be on the uncovered contingent liability. Again, the quan-
tification of contingent liabilities may be difficult. If so, they should be
handled in the same way as quasi-fiscal activities.
Finally, the difficulty in interpreting fiscal balances can be a source
of vulnerability. This difficulty derives in part from the extent of fiscal
activity to which the fiscal balance refers. But it also relates to the way
in which the fiscal balance is measured. Because assessing vulnerabil-
ity will necessarily become a comparative exercise-assessments for
some countries will inform assessments for other countries-there are



A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY             165
advantages to using an internationally comparable measure of the fis-
cal balance (and other fiscal aggregates) in all countries. The overall
balance, appropriately adjusted in response to its most obvious short-
comings (for example, to reflect expenditure arrears, the use of privati-
zation proceeds), is best suited to this purpose. However, the overall
balance should be supplemented by other fiscal balance measures where
they provide a better indicator cf the stance of fiscal policy from a
macroeconomic perspective.9
,Short-term Risks
Fiscal outcomes are exposed to variations in key underlying assump-
tions and other factors. (Chapter 5 by Ramaswamy in this volume
addresses this fact by applying a factor model.) To assess the sensitiv-
ity of fiscal outcomes, it is necessary to move beyond the initial fiscal
position, because the one-year time horizon of a budget, and by impli-
cation of the initial fiscal position, does not do justice to the full extent
of the short-term risks to which fiscal outcomes are exposed. For this
reason, the initial fiscal position should be accompanied by a short-
term forecast that looks at least two years ahead. The short-term fore-
cast should be based on unchanged policies, in the sense that policy
intentions that have been announced but not implemented should be
excluded, and it should be purged of temporary measures affecting the
initial fiscal position. This forecast should be fairly detailed, but possi-
bly less so than the description of the initial fiscal position. Both the
initial fiscal position, which typically itself has a forward-looking com-
ponent, and the short-term forecast should then be subjected to sensi-
tivity analysis.
The principal short-term risks that need to be addressed are the
following:10
* The initial fiscal position and the short-term forecast are sensi-
tive to changes in macroeconomic variables and other sources of eco-
nomic risk. Unanticipated changes in growth of the gross domestic
product (GDP), unemployment, inflation, interest rates, external trade,
capital flows, the exchange rate, and other macroeconomic variables
affecting macrofiscal outcomes give rise to forecasting risk affecting
revenue, expenditure, and financing. The structure of debt is impor-
tant in this regard, because it affects the fiscal risk associated with
short-term movements in interest rates and the exchange rate. Rev-
enue and expenditure are also subject to other risks affecting the tax
base (such as corporate profitability), nontax revenue (such as mineral
prices), and spending programs (such as government wage increases).
* It is possible that contingent liabilities will be called when no
budget provision has been made to meet them. The appropriate way



166                   RICHARD HEMMING AND MURRAY PETRIE
to provide for contingent liabilities, and the practicalities of doing so,
raise issues that have resulted in limited provisioning (see Chapter 2
by Petrie in this volume for further discussion). Implicit liabilities will
always entail more substantial risk, since provisioning is usually judged
inappropriate because of moral hazard problems.
* There may be a lack of clarity about the size of specific expendi-
ture commitments in that provision may be made in the budget for
spending on an activity (such as bank restructuring), but there is less
than the usual precision about the cost implications of that activity.
* Finally, some fiscal policies may be defined imprecisely. This would
be the case where a government announces a policy intention (such as
providing incentives for saving and investment) but either the details
of the way in which the policy is to be implemented are not well devel-
oped or the implications of an announced method of implementation
are unclear.
Fiscal analysis should therefore involve an examination of the range
of possible short-term macrofiscal outcomes, with a focus on varia-
tions in underlying assumptions and other parameters to which a prob-
ability or likelihood of different events can be attached, albeit in some
cases only approximately. However, while the realization of a typical
risk might signal the need for policy adjustment, it would not neces-
sarily imply vulnerability because the consequences may be easily
accommodated. This being the case, there is merit in subjecting a short-
term forecast to more aggressive stress testing, especially when there are
reasonable grounds to consider that a substantial shock to the economy,
be it global, regional, or country-specific, is more than a remote short-
term possibility. This situation is addressed in the next section in the
context of stress testing a baseline medium-term projection."
Longer-term Fiscal Sustainability
Even if fiscal outcomes are not exposed to significant short-term risks,
running persistent fiscal deficits may result in debt levels that become
a source of fiscal and broader macroeconomic difficulties over the
medium term. The standard debt dynamics analysis is the usual basis
for identifying the impact of deficits on indebtedness, and more gener-
ally for assessing the implications of past and current fiscal policies for
longer-term sustainability. Where available, market-based indicators
(for example, a government debt rating or interest rate premia) can
supplement debt dynamics analysis.'2
The starting point for an assessment of longer-term sustainability is
a baseline, medium-term fiscal projection, which typically would look
at least five years ahead. It should assume a continuation of current
policies, which will often require difficult judgments as to what is and



A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY           167
what is not a current policy.'3 Alt'hough it encompasses the initial fis-
cal position and the short-term forecast, such a baseline does not need
to be as detailed. To assess vulnerability, the baseline medium-term
projection should be supplemented by alternative scenarios. These sce-
narlos should illustrate the responsiveness of the medium-term fiscal
outlook to different specifications of the initial fiscal position and short-
term forecasts. In particular, it should be recognized that short-term
risks may persist, and some could be greater, in the medium term. For
example, certain contingent liabilities, such as pension guarantees, are
probably a bigger source of medium-term risk than of short-term risk.
Stress testing would then be used to assess the impact of short-
term and medium-term shocks (for example, a global interest rate or
business cycle shock, a regional or country-specific reversal of mar-
ket sentiment, or a sharp deterioration in the terms of trade). Stress
testing should identify how the fiscal outlook would change under
circumstances ranging up to the fairly extreme and shed light on why
the outlook changes in the way it does.'4 It is therefore important to
identify the key transmission mechanisms through which the main
fiscal aggregates are affected. MAoreover, the likely correlation be-
tween different transmission mechanisms should be recognized. For
example, a major macroeconomic shock such as an output collapse
may cause revenue to fall, expertditure to increase (for example, on
the social safety net), and contingent liabilities to come home to roost,
all at the same time. A downside scenario should envisage some of
the worst things that can happen (for example, full-blown crises)."5
In this way, stress testing, in combination with identification of the
policy responses that reduce vulnerability, provides the basis for sys-
tematic fiscal contingency planning, which has to be a key outcome
of effective surveillance.
Long-term projections and scenarios are a natural extension of
medium-term analysis. However, they can be even less detailed given
their more speculative nature. It is particularly important to take into
account long-term expenditure pressures. The impact of demographic
developments on pensions and health spending is an obvious source of
such pressure in many countries, The possible exhaustion of a natural
resource that generates substantial revenue, and any associated envi-
ronmental degradation, will also be relevant to long-term fiscal
sustainability in some countries.
Structural Weaknesses
The composition of expenditure and revenue is important in assessing
vulnerability. A principal source of vulnerability is a high proportion
of nondiscretionary spending to total spending, which limits a
government's flexibility to adjust spending levels downward when it



168                   RICHARD HEMMING AND MURRAY PETRIE
needs to do so. Nondiscretionary spending is that for which a govern-
ment is under a legal or other strong obligation to meet. The most
notable examples are interest payments, formula-based transfers to
lower levels of government, and public pensions.16 However, the com-
ponents will vary from country to country, and classifying all spending
as either nondiscretionary or discretionary may require difficult judg-
ments. In many countries, the distinction may boil down to that be-
tween spending on transfers (broadly defined) and spending on goods
and services.
Some large expenditure items are a source of vulnerability, because
they are resistant to adjustment given the powerful interest groups
they serve. Military spending is a case in point, although a large gov-
ernment wage bill may be every bit as entrenched where public sector
unions are strong or the government is an employer of last resort.
There also may be latent expenditure needs that do not manifest them-
selves until triggered by a shock or discontinuity of some kind. Any
significant gap in expenditure, compared with established norms, could
be exposed in this way. For example, the need for a social safety net
may become apparent only after an economic crisis, but once in place
it will almost certainly become permanent.
A good tax structure is one in which revenue derives from a range
of taxes with broad bases, ideally large macroeconomic aggregates
(that is, wages, profits, and consumption, including imports of
consumables). Not only will such a structure tend to result in reason-
able tax rates, but it also will ensure a moderately elastic tax system,
which is desirable from the point of view of facilitating countercyclical
fiscal policy through the operation of automatic stabilizers. A revenue
composition dominated by just one or two taxes, especially if they
have narrow bases, is a source of vulnerability, both in terms of in-
creasing a government's exposure to unexpected fiscal developments
because revenue from just a few taxes is likely to be volatile (for ex-
ample, trade tax revenue is highly sensitive to exchange rate changes),
and in terms of limiting its capacity to respond when the need arises
because tax rates probably have to be very high. Frequent tax law
changes, especially when they result in more exemptions, tax holi-
days, and other relief, as is often the case, can add to vulnerability by
progressively undermining the tax base. A government's capacity to
respond is also constrained by extensive earmarking, which limits the
scope for discretionary tax changes. Finally, a heavy reliance on nontax
revenue, the main sources of which (grants, royalties, privatization
proceeds, central bank profits) may not be stable or particularly re-
sponsive to policy intervention, also can contribute to vulnerability.
The institutional capacity for fiscal management is a major deter-
minant of fiscal vulnerability. Numerous aspects of the institutional
capacity for fiscal management could be relevant to fiscal vulnerabil-



A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY              169
ity-most notably, the administrative capability for expenditure man-
agement and revenue collection, the roles and responsibilities of gov-
ernment and within government, the public availability of information,
the budget process, and the integrity of fiscal information collection.
The emphasis should be on particular weaknesses that can signal that
poor macrofiscal outcomes are possible (for example, expenditure and
tax arrears, ineffective audit procedures).
Finally, a government that has a general reputation for being inef-
fective will usually be vulnerabi'e. Thus a government that gets in-
volved in too many activities that should be left to the private sector,
whose agents (that is, public servants, public enterprise managers) have
a relationship with politicians that is inconsistent with the arm's-length
principle, or that is characterized by extensive nontransparency and
corruption, cannot be expected to meet its macrofiscal objectives on a
consistent basis.
Clearly, the four aspects of vulnerability just highlighted are closely
related. Obviously, an incorrect specification of the initial fiscal posi-
tion makes it difficult to assess both short-term risks and longer-term
sustainability. Moreover, a chronic misspecification of the initial fiscal
position is likely to be a manifestation of weak institutional capacity
for fiscal management. But some of the interactions are subtler. For
example, weak fiscal institutions can act to amplify rather than dampen
macroeconomic volatility, as would be the case when a country pur-
sues pro-cyclical fiscal policy because an inability to save fiscal re-
sources generated by a buoyant economy-reflecting political pressures
for such savings to be spent-leads to a pattern of tax cuts during
expansions and tax increases during recessions.17
Vulnerability Indicators
There are clear gains in convenience if the results of the analysis of the
initial fiscal position, short-term risks, longer-term sustainability, and
structural weaknesses could be summarized in a few key vulnerability
indicators, and the preceding discussion suggests some obvious candi-
dates. However, selecting a set of vulnerability indicators involves trade-
offs. While the inclusion of a large number of indicators increases the
probability that fiscal vulnerability will be identified, they may make
it difficult to identify the main sources of vulnerability. A large num-
ber of indicators also will tend to increase the information require-
ments of vulnerability assessments, which would create difficulty in
ensuring comparability across countries and generally make assess-
ments less manageable. In the end, then, a set of vulnerability indica-
tors that is fairly parsimonious is needed.
The set of indicators described in general terms in Box 7.1 and in
more detail in the Appendix focuses on the aspects of vulnerability



170                     RICHARD HEMMING AND MURRAY PETRIE
Box 7.1 A Possible Set of Fiscal Vulnerability Indicators
* Fiscal position indicators-weak initial fiscal position; incomplete
coverage of government fiscal activity; poor accounting and control; in-
sufficient balance sheet information; sizable uncovered contingent liabili-
ties; and significant quasi-fiscal activities.
* Short-term risk indicators-high sensitivity of short-term fiscal out-
comes to changes in key macroeconomic variables; inappropriate debt
structure; variable revenue sources and expenditure programs; calling of
uncovered contingent liabilities; and other expenditure risks.
* Longer-term sustainability indicators-unfavorable debt dynamics;
low government debt rating or high interest rate premia; adverse demo-
graphic trends; and rapid natural resource depletion or serious environ-
mental degradation.
* Expenditure indicators-large share of nondiscretionary spending
or transfers; excessive military spending; and significant gaps in expendi-
tures (for example, social security, safety net, health and education, infra-
structure).
* Revenue indicators-inelastic revenue system; highly concentrated
tax revenue; frequent tax law changes; extensive earmarking; and reli-
ance on grants and other major nontax revenue sources.
* Fiscal management indicators-large expenditure arrears and use
of netting arrangements; marked deviation between the original budget
and the budget outturn; nonexistent or weak medium-term budget plan-
ning; long delays in preparing and auditing final accounts; large tax ar-
rears and use of tax offsets; a large stock of tax refunds, especially for the
value added tax (VAT); an out-of-date taxpayer register; and an ineffec-
tive tax audit program.
* Government effectiveness indicators-poor results from surveys of
public sector performance, corruption, and so forth.
discussed in this chapter and avoids reference to features of the fiscal
system that are unrelated to vulnerability at the aggregate level. Even
so, for a number of reasons these indicators should still be regarded as
provisional. First, the acid test of their suitability will come only when
they are implemented in the context of actual vulnerability assessments.
Country experience may suggest that there are more useful alterna-
tives to the suggested indicators. Second, country experience may also
suggest a need for additional indicators. Political factors, for example,
are not included, despite the fact that a specific political event, such as
an approaching election, may generate an unusual amount of uncer-
tainty about short-term fiscal policy in many countries. There also
may be a case for looking at characteristics of the fiscal management
system that give rise to vulnerabilities beyond those directly related to



A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY            171
aggregate fiscal policy, because indirect effects (for example, where
macrofiscal outcomes are affected by the need to meet microstructural
objectives) will often be important. Finally, the scope for greater use of
survey-based indicators that capture important supplementary infor-
mation bearing on fiscal vulnerability should be explored."8
The fiscal vulnerability indicators can be used in a range of ways.
At the least demanding extreme, they can be viewed as no more than a
checklist that someone assessing fiscal vulnerability could use to keep
track of whether he or she has covered all the relevant aspects of vul-
nerability. At the other much more demanding extreme, it is easy to
envisage them as a stepping-stone on the way to the type of systematic
assessment that could ultimately produce an index of vulnerability. In
the first instance, the indicators should probably be used more as a
checklist, albeit an evolving one as the indicators are modified in re-
sponse to experience with their use. But once a final set of indicators
begins to emerge, they can be used in a more ambitious way, perhaps
with the objective of developing a small number of broad categories of
vulnerability. In this connection, however, it will be necessary to as-
sign the indicators priority, to take account of differences in fiscal in-
stitutions across countries, and perhaps to develop indicators relevant
to particular groups of countries (or weight indicators differently across
groups of countries).'9
In whatever form they are used, fiscal vulnerability indicators have
to be interpreted with caution. It has already been noted that indica-
tors of short-term fiscal risk may point to the need for a relatively
small fiscal correction within the context of an otherwise strong un-
derlying fiscal position. It is important to distinguish this from a situ-
ation in which the underlying position is also weak.20 Although it will
often be the case that a high level of exposure to short-term fiscal risk
will be a leading indicator of longer-term vulnerability, it is important
that fiscal vulnerability assessments avoid judgments based only on a
few indicators.
The policy significance of a similar level of fiscal vulnerability also
will vary across countries, depending on the broader economic con-
text in which fiscal policy is placed. For example, evidence of short-
term risk may be a serious concern in a country with a currency board
or a fixed exchange rate, where the scope for discretionary monetary
policy is limited. Government deficits and debt also may be more of a
concern in a country with relatively low national savings and a high
external debt, or with high inflation and an underdeveloped financial
sector. And fiscal vulnerability in a country where an external crisis
could have contagion effects should warrant particular attention. Fi-
nally, a given level of fiscal vulnerability also may be a cause for greater
concern if combined with a low level of adherence to a broader set of
standards related to government performance.



172                  RICHARD HEMMING AND MURRAY PETRIE
Microstructural Aspects of Vulnerability
This chapter has focused so far on1 a government's aggregate fiscal
policy objectives. As noted, however, a government has to pursue
other microstructural objectives, and failure to meet such objectives
can create vulnerability just as surely as failure to meet macrofiscal
objectives. A government can have a wide range of microstructural
objectives. Abedian (2000) focuses on those microstructural objec-
tives that are most closely linked to a government's socioeconomic
legitimacy. Equity goals related to poverty alleviation, income redis-
tribution, and improvements in other indicators of human develop-
ment fit into this category. So too do many of the efficiency goals
that traditionally (that is, in a welfare economics sense) justify gov-
ernment intervention in the economy, such as providing public goods
and merit goods, taking account of the external costs and benefits
associated with private decisions, and addressing other aspects of
market failure.
Two microstructural aspects of vulnerability are discussed in Abedian
(2000). The first aspect is those undesirable features of the environ-
ment supporting government activity that can lead to government (or
bureaucratic) failure. This failure in turn manifests itself in the pursuit
of inappropriate equity and efficiency goals. Weaknesses in this area
extend beyond those noted above as relevant to macrofiscal vulner-
ability and cover such things as the political process, the public sector
culture or ethos, and the general approach to governance. The second
aspect is the fiscal management system, where weaknesses that again
extend beyond those noted above result in delivery failure, that is a
large gap between a government's equity and efficiency goals and what
it actually achieves.21
Concluding Comments
The aim of fiscal vulnerability assessments is to identify those features
of a country's fiscal system that compromise the ability of the govern-
ment to meet its aggregate fiscal policy objectives. They also provide a
basis for managing vulnerability in order to limit the government's
exposure to possible adverse outcomes, in particular by enhancing its
ability to respond to fiscal and broader economic developments. Man-
aging vulnerability requires a preemptive effort to address potential
problems revealed by vulnerability assessments.
It is primarily the organizational framework that vulnerability as-
sessments provide that is new. However, the true value added will not
become clear until the framework is implemented. In particular, the
vulnerability indicators will almost definitely need fine-tuning and
possibly more extensive reconsideration.



A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY                  173
Finally, although this chapter has discussed government liabilities
and assets in terms of the importance of having information on their
levels and structure as indicators of vulnerability, it has not discussed a
government strategy for their management. Yet, as Chapter 1 by Hana
Polackova Brixi and Ashoka Mody indicated, government liabilities
and assets can be managed in a way to contain vulnerability by reduc-
ing government risk exposure. The sophistication of a government's
risk management strategy may not be done justice by the indicators
suggested above. More work is rneeded to offer practical guidance on
optimal risk-reduction strategies.22
Annex 7.1:
Measurement of Fiscal Vulnerability Indicators
Indicators                              Measures
Fiscal position indicators
Weak initial fiscal     Overall fiscal balance as a share of GDP
position              Other fiscal balance measures as a share of GDP
(where relevant)
Net financial debt as a share of GDP
Size of aatomatic stabilizers (small/average/large)
Average and maximum rates of tax (for each
main tax)
Incomplete coverage     Revenue covered in fiscal data as a share of
of government fiscal    general government revenue
activity              Expenditure covered by fiscal data as a share
of general government expenditure
Poor accounting and     Fiscal balance measured from above-the-line
control                 relative to the fiscal balance measured from
below-the-line
Insufficient balance sheet Gross debt
information           Net financial debt
Other balance sheet data
Sizable uncovered       Gross contingent liabilities as a share of total
contingent liabilities  revenue
Net contingent liabilities as a share of total revenue
Or
Descriptrion of main contingent liabilities and
quantification of largest net contingent liabilities
Significant quasi-fiscal  Quasi-fiscal activities as a share of total revenue
activities            Or
Description of main quasi-fiscal activities and
quantification of largest quasi-fiscal activities
(Table continues on the following pages.)



174                     RICHARD HEMMING AND MURRAY PETRIE
Indicators                              Measures
Short-term fiscal risk indicators
High sensitivity of short-  Impact of variations in forecasted GDP growth,
term fiscal outcomes to  inflation, balance of payments, exchange rate,
changes in key macro-   and interest rates on the fiscal balance
economic variables
Inappropriate debt      Maturity (short, medium, and long term), interest
structure               rate structure (fixed versus variable rates), and
currency composition of debt
Variable revenue sources  Impact of variations in other economic and
and expenditure         noneconomic determinants of revenue and
programs                expenditure on the fiscal balance
Calling of uncovered    Net contingent liabilities as a share of GDP;
contingent liabilities  expected payments in connection with
guarantees and other contingent liabilities
Other expenditure risks  Description of programs and policies that give
rise to risks
Longer-term sustainability indicators
Unfavorable debt        5-10 year projection of gross or net debt as a share
dynamics                of GDP, and change in the primary balance as a
share of GDP required to stabilize the debt ratio
at the current level or at a specific target level
Low government debt     The Bloomberg website <www.bloomberg.com>
rating and/or high      provides information on how to calculate
interest rate premia    interest rate premia.
Adverse demographic     Long-term projection of retirement age and
trends                  school-age population relative to total and
working population; impact on expenditure as
a share of GDP and on tax rates
Rapid resource depletion  Years of usable reserves at current exploitation
rate; resource-related revenue as a share of total
revenue; resource-related financial assets as a
share of GDP; serious environmental
degradation
Expenditure indicators
Large share of nondiscre-  Nondiscretionary spending and transfers as a
tionary spending and/   share of GDP
or transfers
Excessive military spending Military spending as a share of GDP
Significant gaps in     Programs for which spending as a share of GDP
expenditure             is significantly below the average for compa-
rable countries



A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY                  175
Indicators                              Measures
Revenue indicators
Inelastic revenue system  Tax elasticity or buoyancy
Highly concentrated     Revenue composition, particularly trade tax
tax revenue             revenue as a share of total tax revenue
Frequent tax law changes  Major tax changes, especially new exemptions and
other reliefs, every year or every two years
Extensive earmarking    Revenue from earmarked taxes as a share of total
revenue
Reliance on grants and  Nontax revenue as a share of total revenue;
other unstable nontax   compcsition of nontax revenue
revenue sources
Fiscal management indicators
Large expenditure arrears  Expenditure arrears as a share of total revenue;
and use of netting      significant netting of arrears; inability
arrangements            to repDrt on sizeable arrears
Marked deviation between Expendit:ure outturn relative to original expendi-
the original budget and  ture; resort to large supplementary budgets
the budget outturn
Nonexistent or weak     Effective medium-term budget planning
medium-term budget
planning
Long delays in preparing  Length of time between end of fiscal year and (a)
and auditing final      preparation of final accounts and (b) release
accounts                of audited accounts
Large tax arrears and   Tax arrears as a share of total revenue; sharp
use of tax offsets      increase in tax arrears; significant tax
offsets
A large stock of tax    Stock ol: tax/VAT refunds as a share of tax/VAT
refunds, especially for  revenue
value added tax (VAT)
Out-of-date taxpayer    Currentiess of taxpayer register by main tax
register
An ineffective tax audit  Coverage of tax audit; targeting of tax audit
program
Government effectiveness indicators
Poor results from surveys  Information available from the Institute for
of public sector perfor-  Management Development's World Com-
mance, corruption, and  petitiveness Yearbook; the Transparency
so forth                International's Corruption Perceptions Index



176                     RICHARD HEMMING AND MURRAY PETRIE
Notes
1. This chaper was originally issued as IMF Working Paper WP/00/52 by the
International Monetary Fund (IMF), Washington, D.C. It draws on a companion
paper by Iraj Abedian and has been prepared with significant input from William
Allan, John Crotty, and Steve Symansky. It also has benefited from comments
made by many colleagues in the Fiscal Affairs Department at the IMF and by
seminar participants at the European Central Bank and the World Bank.
2. This is illustrated by recent work that points to variables related to fiscal
imbalance-for example, high domestic credit growth and large current account
deficits-as being among the strongest predictors of an external crisis (IMF 1999).
3. The World Bank (1998) refers to these as the level-1, level-2, and level-3
operations of fiscal policy.
4. In addition to these general objectives, a country may have specific fiscal
policy objectives that the government has set for itself (for example, as reflected
in a fiscal rule) or that may have been agreed on with others (for example, as
part of IMF conditionality).
5. Stability of tax rates is also justified by the fact that the distortionary cost
of taxation is reduced by smoothing tax rates over time. Tax smoothing is also
consistent with countercyclical fiscal policy.
6. Prudence is consistent with the usual approach in accounting, where fi-
nancial statements consistently err on the side of caution in recording events or
transactions that are likely to have a favorable impact, while being less cautious
when the results are likely to be unfavorable.
7. Although lower levels of government that set their own objectives and
are subject to market discipline can be viewed as independent of central govern-
ment from an economic perspective, from a vulnerability perspective consolida-
tion is desirable (although vulnerability could be assessed independently for lower
levels of government).
8. Easterly (1999) finds evidence from an analysis of countries borrowing
from the IMF and the World Bank and of European countries covered by the
Maastricht Treaty that fiscal adjustment often takes the form of privatization
and cuts in government investment, so that changes in reported fiscal deficits
represent illusory rather than real adjustment because there is an offsetting bal-
ance sheet transaction.
9. Depending on a country's circumstances, other such measures might in-
clude the structural balance, the operational balance, the primary balance, or
the augmented balance.
10. These risks are the same as those discussed in the fiscal transparency
manual, because a requirement of the fiscal transparency code is that all govern-
ments should publish a fiscal risk statement with the annual budget.
11. However, if the focus of a vulnerability assessment is on short-term fiscal
outcomes alone, stress testing should be part of such an assessment.
12. They cannot, however, substitute for such analysis, or for vulnerability
assessments more generally, because there is little evidence that either debt rat-
ings or interest rate premia adequately reflect fiscal sustainability. They are in-



A FRAMEWORK FOR ASSESSING FISCAL VULNERABILITY                 177
fluenced more by external sustainabihty, level of development, and the depth of
the market for a country's debt.
13. The difficulty is illustrated by the discussion of fiscal policy in Asia in
IMF's World Economic Outlook, October 1998 (IMF 1998), where, in distin-
guishing discretionary from nondiscretionary fiscal measures, it had to be de-
cided whether holding nominal spending constant when spending had in the
past increased represents changed or unchanged policy.
14. Stress testing clearly goes beyond the usual scenario analysis, which in
the fiscal area tends to involve producing higher-growth and lower-growth sce-
narios to illustrate the benefits of stronger fiscal policies and the costs of weaker
fiscal policies than in the baseline.
15. There is a parallel here with risk assessment in the private sector. Stan-
dard value-at-risk methodologies used in financial analysis show how much a
bank or firm could potentially lose over a specified time period for likely market
movements. Stress testing is used to assess and manage extreme risks.
16. Not all nondiscretionary spending is necessarily a problem. For example,
spending on unemployment compensation is cyclically sensitive. It therefore acts
as an automatic stabilizer during a cyclical downturn, reducing the need for
discretionary fiscal policy.
17. Talvi and Vegh (2000) find that fiscal policy in developing countries is
for such a reason highly pro-cyclical. They suggest that attention be paid to
designing fiscal arrangements (such as stabilization funds) aimed at ensuring
that fiscal savings generated during good times is saved for bad times.
18. The weakness of survey-based indicators is that surveys fail to reflect the
strength with which views are held and thus the weight of opinion. However,
they can incorporate information frcm a wider range of sources.
19. In this connection, Bird and Banta (1999) suggest indicators for transi-
tion economies that take account of their special circumstances.
20. This is analogous to the differentiation often applied to companies to
distinguish those with short-term cash flow problems but positive net worth
from those with negative net worth.
21. These sources of microstructural vulnerability also affect what Tanzi
(2000) refers to as the quality of the public sector.
22. For some discussion of the issues in this area, see Skilling (1999).
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
Abedian, Iraj. 2000. "Micro-Structural Aspects of Fiscal Vulnerability." Inter-
national Monetary Fund, Washington, D.C. Processed.
Bird, Richard M., and Susan M. Ba:nta. 1999. "Fiscal Sustainability and Fiscal
Indicators in Transitional Countries." Paper presented at the USAID Confer-
ence on Fiscal Reform and Sustainability, Istanbul, June.



178                    RICHARD HEMMING AND MURRAY PETRIE
Downes, Patrick T., Dewitt D. Marston, and Inci Otker. 1999. "Mapping Fi-
nancial Sector Vulnerability in a Non-Crisis Country." IMF Policy Discus-
sion Paper PDP/99/4. International Monetary Fund, Washington, D.C.
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy 28 (April): 57-86.
IMF (International Monetary Fund). 1998. World Economic Outlook, Octo-
ber 1998. World Economic and Financial Surveys. Washington, D.C.
. 1999. World Economic Outlook, October 1999. World Economic
and Financial Surveys. Washington, D.C.
IMF (International Monetary Fund), Fiscal Affairs Department. 2001. "Code
of Good Practices on Fiscal Transparency" and "Manual on Fiscal Trans-
parency" <www. imf. org/external/np/fad/trans/index. htm>.
Skilling, David. 1999. "How Should Governments Invest Financial Assets and
Manage Debt?" Paper presented at the 20th Annual Meeting of Senior Bud-
get Officers, Organisation for Economic Co-operation and Development,
Paris.
Talvi, Ernesto, and Carlos A. Vegh. 2000. "Tax Base Variability and Procyclical
Fiscal Policy." NBER Working Paper 7499. National Bureau of Economic
Research, Cambridge, Mass.
Tanzi, V. 2000. "The Quality of the Public Sector." International Monetary
Fund, Washington, D.C. Processed.
World Bank. 1998. Public Expenditure Management Handbook. Washington,
D.C.



Country Examples






CHAPTER 8
Evaluating Government Net
Worth in Colombia and Republica
Bolivariana de Venezuela
William Easterly
and Davicd Yuravlivker
World Bank
ONE OF THE MOST IMPORTANT questions to ask about public finance
in any given country is: Does the government's balance sheet contain
assets (treasures), such as oil or valuable public enterprises, that can
help the government meet its long-run budget constraints? Or does the
government's balance sheet contain large implicit liabilities (time
bombs), such as a pension system temporarily in surplus but headed
for big future deficits? Even if the government's balance sheet contains
treasures, are those treasures being depleted to finance current con-
sumption, at the expense of futuLre generations?
The conventional approach to assessing the sustainability of a given
fiscal deficit is to compare it with the deficit necessary to keep the
public debt-to-GDP (gross domestic product) ratio constant. This is
justified by the reasoning that if a country is not already in a debt
crisis, then keeping the public debt ratio constant will be sufficient to
avoid a debt crisis in the future.'
The main objectives of this chapter are: (a) to present the assets/
liabilities/net worth approach to evaluating the true financial position
of the public sector in order to achieve a better assessment of longer-
term fiscal sustainability; and (b) to present applications of this ap-
proach to the cases of Colombia and R6publica Bolivariana de Venezuela
in order to illustrate its use and to help their authorities in the process
of fiscal adjustment. In particular, this study could help ensure that the
181



182                              EASTERLY AND YURAVLIVKER
fiscal measures that governments are implementing to address present
fiscal gaps are coherent and consistent with achieving longer-term fis-
cal sustainability. In addition, the conceptual framework developed
here and its application to these two country cases could be useful in
analyzing similar problems in other countries.
The next section of this chapter presents the assets/liabilities/net
worth conceptual framework and methodology. The final section sum-
marizes the main findings of the chapter and offers some concluding
remarks.
Assets/Liabilities/Net Worth Approach
Conceptual Framework
The limitations of conventional sustainability analysis call for broad-
ening the sustainability analysis to consider the evolution of the
government's complete balance sheet (Table 8.1).2
Which components of the government's fiscal policy package improve
the government's net financial position and which ones worsen it?
Anyone considering the longer-term sustainability of a government's
finances would find it helpful to think first about the balance sheet
that the government must manage.3 Although one cannot always mea-
sure all of the items in this balance sheet, the balance sheet approach is
a useful way to think of which government actions imply progress or
regress toward sustainability.4 Investing in a project with an economic
rate of return higher than the discount rate, for example, improves the
government's net financial position. (The government has to devise a
way to capture the revenues that accrue to the society from this project,
however.) Forward-looking individuals and firms (and credit rating
agencies) will react more favorably to a fiscal adjustment that pre-
serves high rate of return activities while cutting consumption than to
a myopic package that cuts projects with high returns just to improve
this year's fiscal balance.
Note that in Table 8.1 the present values of government consump-
tion or government tax revenue do not appear in the balance sheet.
The government's intertemporal budget constraint is that the present
value of government consumption minus government revenues be less
than or equal to the government's net worth (more on the daunting
task of implementing this in a moment). If the government's inter-
temporal budget constraint is violated under current revenue and ex-
penditure plans, then fiscal sustainability does not hold. The future
fiscal adjustment required to satisfy the intertemporal budget constraint
is a measure of the distance from sustainability in government finances.
A more sophisticated analysis would refer to a desired net worth level,



EVALUATING GOVERNMENT NET WORTH                               183
Table 8.1 Government Balance Sheet for Fiscal Sustainability
Assessment
Assets                                         Liabilities
Government-owned public goods (such as  Public external debt
infrastructure, schools, and health clinics
that generate an adequate econoniic rate
of return and an indirect financial rate of
return through tax collection)
Government-owned capital that is finan-  Public domestic debt
cially profitable (anything for which gov-
ernment can charge user fees to generate
an adequate financial rate of return)
Value of government-owned naturall re-  Contingent liabilities (for
source stocks (such as oil and minerals)  example, bank deposit
guarantees, net present
value of pension scheme,
guarantees of private debt)
Expected net present value of loans to pri-  Government's net
vate sector and other financial assets  worth
Source: The authors.
but that would depend on the social preferences of each particular
society, and it goes beyond the scope of this chapter.
Methodology: Applying the Concept
Sustainability can be evaluated using the balance sheet approach in
two ways. The first is to estimate the stocks in the government's bal-
ance sheet and assess whether the public sector net worth is positive or
negative. If it is negative, then sustainability will require that the present
value of tax revenues minus government consumption be sufficient to
cover the negative net worth.
The second approach would look at sustainability in flow terms.
The criterion for sustainability would then be to maintain a constant
ratio of net worth to GDP. The idea is that if there is no payments
crisis today, then keeping the net worth-to-GDP ratio constant will
avoid a payments crisis in the future. So fiscal sustainability would
require keeping government net worth to GDP constant. If there is a
payments crisis today, then the r ule would imply increasing the ratio
of net worth to GDP.
If one cannot measure government net worth directly, a pragmatic
approach would be to first measure all balance sheet items that can be



184                              EASTERLY AND YURAVLIVKER
measured to calculate the net debt of the government and then evalu-
ate individual government actions for their effect on fiscal and exter-
nal sustainability.
Note that analyzing this broader notion of fiscal sustainability re-
quires many assumptions to calculate concepts such as the implicit
pension debt and the value of oil reserves. This kind of exercise should
be thought of as illustrative of how the government's long-run finances
will evolve if certain assumptions hold.
What about balance sheet items that cannot be measured? All we
can say is that sustainability is worsened by any action that reduces
government assets or increases contingent liabilities. For example, a
cut in operations and maintenance (O&M) spending that lowers the
value of government-owned highways will worsen sustainability, even
though this spending cut improves the conventional measure of the
fiscal deficit. (If the proceeds from cutting O&M are used to repay
foreign debt, and if the rate of return on O&M is the same as the
interest rate on foreign debt, the spending cut would have a neutral
effect on fiscal sustainability. However, estimates of returns to O&M
are usually well above the interest rate on foreign debt.) A cut in prof-
itable public investments will worsen fiscal sustainability, even though
this cut improves the conventional measure of the fiscal deficit. Ex-
pansion of deposit insurance increases the government's contingent
liabilities and so worsens fiscal sustainability, even though it does not
show up in conventional deficit measures. Deposit insurance may bring
benefits such as increased deposits in the banking system, but the con-
tingent liability of the government should still be accounted correctly.
Consuming the revenues from extraction of a nonrenewable resource
or from privatization worsens sustainability.
In summary, balance sheet accounting can produce a better long-
run perspective on fiscal sustainability than can be obtained from con-
ventional deficit measures. Although the task of measuring government
assets and liabilities is always difficult and sometimes impossible, the
balance sheet approach still encourages clear thinking about what ac-
tions will improve or worsen the government's long-run finances. That
is particularly important when governments need to implement urgent
stabilization measures, to ensure that the measures are consistent with
longer-term fiscal sustainability.
Results from Colombia and
Republica Bolivariana de Venezuela
This section reviews the application of the assets/liabilities/net worth
approach to Colombia and Republica Bolivariana de Venezuela. These
are summaries of two very thorough studies conducted in both coun-
tries by local teams that had a unique access to the data required to



EVALUATING GOVERNMENT NE'T WORTH                          185
carry out such a task. The Colombian study expands the traditional
government balance sheet by calculating other assets as well as contin-
gent and implicit liabilities of the public sector such as pension liabili-
ties and the potential costs of a peace agreement with the guerrillas.
The Venezuelan study, rather than looking at stocks, calculates the
changes in the main public sector assets and liabilities since 1970, and
then looks forward and calculates the net present value of the key
assets and liabilities of the public sector. Both studies conclude that
fiscal adjustment measures are needed to ensure longer-term fiscal
sustainability.
Colombia
Echeverry and others (1999) perform a comprehensive estimate of
public sector stocks, including contingent assets and liabilities (Table
8.2). Because their estimates are fraught with uncertainty and re-
quire many special assumptions, one should think of this exercise as
mainly illustrative.
Table 8.2 estimates that the Colombian public sector has 162 per-
cent of GDP in assets. The most important assets are investments,
plant and equipment, and natural resources. (The estimate does not
include the value of national infrastructure because of the difficulties
of valuation.) However, the Colombian public sector has liabilities
amounting to 232 percent of GD)P, implying a negative net worth of
70 percent of GDP. The most important liability by far is the implicit
pension debt, which amounts to 156 percent of GDP. In contrast, the
total public debt, which is usually the focus of sustainability analyses,
amounts to only 20 percent of GDP. The advantage of the balance
sheet approach is that it identifies a "hole" in the intertemporal public
finances that would not have been apparent using conventional deficit
or debt measures.
How could Colombia's finances be made sustainable if the public
sector's net worth is negative? To cover the negative net worth, the pub-
lic sector could run a perpetual primary surplus (government revenues
minus government noninterest spending). Here is a rough idea of the
amount of surplus necessary: the present value of a perpetual surplus of
x percent of GDP at a discount rate of r and a GDP growth rate of g
would simply be x/(r - g). If r = 0.10 and g = 0.04, for example, this
would imply a required perpetual primary current surplus of 4.2 percent
of GDP to cover negative net worth of 70 percent of GDP.
Given the importance of the implicit pension liability in the overall
picture, it is worthwhile to look at it in more detail. Table 8.3 breaks
down the pension liability into its components.
The most striking finding is that the central government's pension
obligations are more important than the obligations of the general
social security system (although the latter are far from trivial). Another



Table 8.2 Comprehensive Public Sector Balance Sheet, Colombia, 1997
(percent of GDP)
National level           Territorial level      Total
Decentralized  Central   Decentralized  Central   Public sector
Assets                                                     48.4         42.5        25.9         23.6       162.3
Current assets                                             15.8          5.6         4.5          3.6        29.S
Cash                                                         1.7         1.3         0.5          1.0         4.4
Investments                                                  3.5         2.2         1.6          0.7         8.1
Rents                                                        0.0         0.8         0.0          0.6         1.4
Accounts payable                                             9.0         1.2         1.9          0.8        12.9
Inventories                                                  0.9         0.0         0.4          0.0         1.3
Other                                                        0.6         0.0         0.3          0.4         1.3
Fixed assets                                               32.6         36.9        21.4         20.1       132.8
Investments                                                  1.4        17.5         2.0          6.9        27.8
Rents                                                       0.0          0.0         0.0          0.2         0.2
Loans                                                        6.3         1.9         2.6          0.2        11.0
Plant and equipment                                         13.8         2.4        10.8          5.6        32.5
Public goods                                                 3.3         0.1         0.4          3.3         7.1
Natural resources                                            1.0        14.4         0.0          1.1        38.4
Other (mostly electromagnetic spectrum)                     6.9          0.5         5.6          2.8        15.8



Liabilities                                                  34.9        20.8         15.4          7.1        232.3
Current liabilities                                          14.4         5.3          3.4          3.1         26.1
Required deposits                                             5.2         0.0          0.1          0.0          5.3
Public debt                                                   0.1         2.8          0.3          0.4          3.6
Financial obligations                                         2.4         0.0          0.2          0.3          3.0
Suppliers' credits                                            3.7          1.1         1.2          1.5          7.6
Labor obligations                                             1.1         0.2          0.8          0.7          2.8
Bonds                                                         1.2         0.3          0.0          0.0          1.5
Other                                                         0.5         0.9          0.7          0.2          2.4
Long-term liabilities                                        20.4        15.5         11.9          4.0        205.9
Public debt                                                   1.2        14.2          1.8          2.6         19.8
Financial obligations                                         3.1         0.0          2.5          0.2          5.7
Suppliers' credits                                            0.8          1.0         1.2          0.1          3.1
Labor obligations                                             0.0         0.0          0.1          0.1          0.2
Bonds                                                         2.4         0.1          0.2          0.1          2.7
Other                                                        12.9         0.2          6.1          1.0         20.2
Contingent liabilities                                                                                         154.1
Pension liability                                                                                            156.5
Net other contingent liabilities                                                                              -2.4
Other                                                         0.1         0.0          0.1          0.0          0.2
Public sector net worth                                      13.5        21.7         10.6         16.5        -69.9
Public Treasury                                               0.0        23.9          0.2         16.2        -92.0
Institutional                                                13.0          0.0         9.8          0.0         22.8
Other                                                         0.5        -2.2          0.6          0.3         -0.7
Total liabilities, including net worth                     48.4        42.5         25.9         23.6        162.3
Sources: Echeverry and others (1999) and the authors of this chapter.



188                                   EASTERLY AND YURAVLIVKER
Table 8.3 Estimates of the Implicit Pension Liability,
Colombia, 1997
(percent of GDP)
Entity                                             Pension liability
National central government,                            69.76
Teachersb                                             30.10
Armed forces,                                         15.92
Rest of central government'                           23.74
National decentralized governmentd                       5.76
Caprecom'                                              2.99
Ecopetrol'                                             2.28
Ecocarb6n'                                             0.03
Caja Agraria'                                          0.46
Territorial,                                            32.54
Social security,                                        48.48
Total                                                156.54
a. Liabilities assumed directly by the national government.
b. Based on actuarial estimates done for Echeverry and others (1999).
c. Based on actuarial estimates done in 1997.
d. Liabilities assumed by state enterprises and decentralized agencies.
e. Based on actuarial estimates done in 1998.
Sources: Ministerio de Hacienda y Credito Piblico, Colombia, and calculations of
Echeverry and others (1999).
way to reach sustainability would be to reform the government's pen-
sion system in order to raise contributions, postpone retirement, or
take other measures to lower the net present value of pension obliga-
tions. The same reforms to the general social security system could
also help reach sustainability.
Other contingent liabilities do not place a very heavy strain on pub-
lic sector finances. Table 8.4 shows that nonpension contingent liabili-
ties only sum to 14.5 percent of GDP. Among the most important of
the other contingent liabilities are the fiscal guarantees to the metros
(urban subway systems), the cost of achieving peace with the domestic
guerilla movements, and the expected present value of the bailout of
the financial system.
It is important to note that the public sector also has assets of condi-
tional value. The Colombian government has partial claim to oil, gas,
and coal reserves, whose value is contingent on world prices. (The price
assumptions are US$10 per barrel for oil, $30 per ton for coal, and
$1.30 per cubic foot for gas.) Note that these asset calculations subtract
off the unit costs of exploration and extraction. The public sector also
owns the electromagnetic spectrum, which it sells to cell phone compa-



EVALUATING GOVERNMENT NET WORTH                            189
T'able 8.4 Other Contingent Liabilities of Public Sector
(Excluding Pensions), Colombia, 1997
Contingent liabilities                       Percentage of GDP
Natural disasters                                  1.10
Earthquakes                                      1.09
Floods                                           0.01
Bailing out financial system                       2.18
Infrastructure guarantees                          5.89
Roads                                            0.10
Airports                                         0.70
Electricity                                      0.50
Telecommunications                               0.23
Metros                                           4.36
External debt guarantees                           0.69
Judicial findings against public sector            0.16
Guarantees of municipal and provincial debt        0.48
Cost of reaching peace agreement                   4.04
Total                                            14.54
Sources: Echeverry and others (I1999) andl the authors of this chapter.
nies. As of 1997, these contingent sources of future government rev-
enues were: petroleum reserves, 14.6 percent of GDP; natural gas re-
serves, 9.0 percent; coal reserves, 11.9 percent; and electromagnetic
spectrum, 18.3 percent. They totaled 53.9 percent of GDP.
Note that the value of the stock of natural resources is not that high
relative to GDP, only amounting to 36 percent of GDP. Later it is
revealed that the value of natural resources in Republica Bolivariana
de Venezuela is much larger.
In summary, the Colombia experience illustrates the value of the
balance sheet approach to public sector finances. This approach finds
negative net worth of the public sector, requiring a fiscal adjustment
whose need would not have been apparent from conventional deficit
measures.
Republica Bolivariana de Venezuela
For Republica Bolivariana de Venezuela, Garcia, Balza, and Villasmil
(1999) use a more dynamic approach to look at longer-term fiscal
sustainability. They calculate the annual change in the net worth of
the public sector during the past 30 years. That change was used to
finance public consumption and to subsidize domestic consumption of



190                               EASTERLY AND YURAVLIVKER
oil products. They also calculate the implicit and contingent liabilities
of the public sector and project the evolution of public assets, liabili-
ties, and net worth into the future in order to estimate the maximum
nonoil fiscal deficit that would be consistent with longer-term fiscal
sustainability. Furthermore, these projections are done under a range
of assumptions to test for their sensitivity to various domestic and
external conditions. Finally, Garcia, Balza, and Villasmil discuss key
policy measures that would help the fiscal accounts move in the right
direction to achieve a solid financial position on a sustainable basis.
The Evolution of Public Assets, Liabilities, and Net Worth, 1970-
98. Garcia, Balza, and Villasmil (1999) describe five major develop-
ments that affected the financial position of the Venezuelan public
sector in this century:
o 1930-70-the discovery of oil and use of its revenues to finance
investment in infrastructure and the provision of public services;
o 1973-83-the decision to invest heavily in public enterprises by
means of growing external indebtedness;
o 1983-88-the decision to reduce investment in social sectors and
public infrastructure in order to maintain a high level of investment in
public enterprises and to service external debt;
o 1989-95-the use of privatization revenues to finance recurrent
public deficits; and
o 1996-98-the decision to increase investment and production
in the oil sector, financed by cuts in expenditure and investment in
other areas.
The sharp increases in oil prices in 1973-74 and in 1978-79, and
the expectation that prices would continue their upward trend, led to
a growing reliance on external debt to finance public expenditure and
advance ambitious investment projects. At the same time, the avail-
ability of huge oil revenues and foreign credit delayed implementation
of the structural reforms that were necessary to improve the efficiency
and competitiveness of the economy. Eventually, the volatility of oil
prices led to large ups and downs in the Venezuelan economy in gen-
eral and in the fiscal accounts in particular.
Estimating the evolution of public assets and liabilities is a very
difficult exercise. Garcia, Balza, and Villasmil (1999) focus on four
main sources that add liabilities or reduce assets: changes in the exter-
nal debt, changes in the domestic debt, privatization revenues, and the
value of oil production, net of production and investment costs. They
also identify three main uses that add assets or reduce liabilities: in-
vestment in nonfinancial public enterprises, other public investment,
and changes in international reserves. The difference between varia-
tions in uses and sources would then reveal the changes in the net



EVALUATING GOVERNMENT NET WORTH                           191
worth of the public sector. Tables 8.5 and 8.6 show the evolution of
these variables from 1970 to 1998.
According to Table 8.5, the net worth of the public sector (in net
present value terms using a discount rate of 5 percent) declined by
nearly US$300 billion (287 percent of 1999 GDP) over the period
1970-98. It should be noted, however, that this estimation does not
include the discovery of 50 billion barrels of new oil reserves during
those years, which brought the total level of proved reserves to 76
billion barrels. Depending on the price of oil, these new discoveries
would represent $60-$180 billion in net present value terms. The de-
cline in the net worth of the public sector shows the drop in assets, or
increase in liabilities, used to: (a) finance current public consumption
(net of nonoil revenues) amounting to $177 billion and (b) subsidize
the domestic consumption of oil products (which were sold not only at
below-border prices but even below their production costs for most of
the period) by $86 billion in net present value terms. The remainder,
nearly $32 billion, is classified as errors and omissions, which reflect
rneasurement errors in public expenditure accounts and transfers to
the private sector that were not properly accounted for.
The loss of public net worth could be overstated, because current
public expenditures include the provision of education and health ser-
vices, which are in fact an investment in human capital. From that
point of view, there was an increase in private assets that compensates
for the decline in public assets. However, the magnitude of the drop in
net public worth is well beyond the fraction of current public expendi-
ture that could be accounted for as investment in human capital. Fur-
thermore, the drop in the quality of public services during the period
as well as the decline in real per capita GDP of over 10 percent suggest
that the public sector "jewelry" was not used in the most efficient
way. That calls for an evaluation of the use of public resources to
ensure that they do not undermine the financial position of the public
sector and that they contribute efficiently to the development of the
physical and human capital base of the country.
The loss of public net worth also could be understated because many
of the public investments turned out to be unprofitable. For example,
the heavy investment in steel production in the 1970s and 1980s, at a
time of overcapacity in the world steel market, almost surely did not
increase the value of public capital very much.
Contingent and Implicit Liabilities. The calculation described above
does not take into account the existence and evolution of contingent
or implicit public sector liabilities. Therefore, after discussing short-
term fiscal stabilization measures for 1999-2000, Garcia, Balza, and
Villasmil (1999) identify and estimate the net present value of the pub-
lic sector liabilities embodied in: (a) the social security system; (b) the



Table 8.5 Net Worth of the Public Sector, R6publica Bolivariana de Venezuela, 1970-98
(millions of U.S. dollars)
Value of
oil prod- Investment
uction (net by non-
Change   Change          invest"ment financial        Change             Net    Subsidies
in net  in net   Privati-  and oper-  public  Other  in inter-  Change in  current to domestic  Errors
external  domestic  zation  ational  enter-   public  national  public  expenses  oil con-  and
debt     debt    income  expenses)  prises  investment reserves  net worth  (nonoil)  sumption  omissions
(8) = (5) + (6) +
(7) - (1) - (2) -           (l 1) = -(8)
(1)     (2)      (3)       (4)     (5)      (6)      (7)    (3) - (4)  (9)      (1 0)  -(9) - (10)
1970                   152       81              940      225      573        85    -289      -16       75      230
1971                  270       31              1,569     442      610      444     -375      180       87      108
1972                 1,084      43              1,701     867      605      218   -1,138      339      100      699
1973                   135       99             2,665      859     782      724     -534      389      114       31
1974                 -112      502              9,204      816    1,932    3,842  -3,004     -707      132    3,579
1975                 -301      331              7,843    1,534    3,028    2,613    -698     -196      151      742
1976                 2,667     342              6,127    1,566    3,702    -286   -4,154    1,436      174    2,544
1977                 5,322    1,446             8,256    3,985    3,494    -425   -7,970    2,328      635    5,008
1978                 4,779     803              6,668    3,927    3,172   -1,707  -6,858     1,836     253    4,768
1979                 6,721     308             11,485    3,457    2,977    1,302  -10,778   3,184      917    6,677
1980                 4,656        1            16,083    3,762    3,529    -715  -14,164    5,040     1,976   7,148



1981               2,331   1,535          16,099   4,681   3,460    1,594 -10,231   6,526   2,814     890
1982              -1,281   1,090          11,175   5,564   4,830    1,420    830    5,846   1,984  -8,660
1983               2,167    975           12,245   4,964   4,781    1,110  -4,531   4,418   2,749  -2,635
1984              -1,127  -2,032          16,130   1,971   3,756    1,320  -5,924   3,050   3,363    -488
1985              -1,470   1,704          15,303   2,152   3,933    1,281  -8,171   2,462   3,838   1,871
1986                293    1,565           5,780   3,241   3,880  -3,892   -4,409   2,211     898   1,300
1987               2,313  -3,840           9,864   2,407   2,311    -483   -4,102   3,484   2,586  -1,968
1988                645      128           7,142   2,843   3,933  -2,705   -3,844   4,111   1,748  -2,015
1989              -1,321  -1,960          11,176   2,121   1,499     740  -3,535    4,282   2,883  -3,630
1990               1,746   -323       10  15,018   1,993   2,162   4,348   -7,948   6,455   3,592  -2,100
1991                874    1,497   2,278  11,340   1,515   4,118   2,346   -8,010   5,842   2,447    -280
1992               3,306    -272      30   8,566   1,679   2,261   -1,104  -8,793   5,749   2,315     729
1993              -7,571   1,309      32   8,238   1,068   1,788    -345     503    4,101   2,070  -6,674
1994                174    2,058      18   8,603     644   1,391  -1,149   -9,967  11,770   2,413  -4,216
1995              -2,365   1,299      20   9,400     962   2,390   -1,784  -6,785   5,797   3,002  -2,015
1996               -988   -3,996   1,159  15,120     854   1,731    5,506  -3,205   3,397   3,504  -3,696
1997                212     -843   2,425  12,538   1,154   3,332   2,589  -7,256    4,547   2,896    -187
1998               -666   -1,338     110    271    1,135   3,313   -2,969   3,102   1,672  -1,573  -3,202
Total           22,645   2,543    6,082 266,549  62,388  79,273  13,919 -142,238  99,533  48,143  -5,443
NPVa              71,586  12,948   7,643 546,088 132,279 168,260   35,527 -295,199 177,181  86,365  31,653
a. In 1999 net present value, at a 5 percent discount rate.
Sources: Garcia, Balza, and Villasmil (1999) and the authors of this chapter.



Table 8.6 Net Worth of the Public Sector, R6publica Bolivariana de Venezuela, 1970-98
(percent of GDP)
Value of
oilprod- Investment
uction (net by non-
Change    Change           investment financial           Change              Net     Subsidies
in net   in net    Privati-  and oper-  public  Other   in inter-  Change in  current to domestic  Errors
external  domestic  zation   ational   enter-    public   national  public  expenses   oil con-  and
debt      debt    income   expenses)  prises  investment reserves  net worth  (nonoil)  sumption  omissions
(8) = (5) + (6) +
(7) - (1) - (2) -              (I 1) = -(8)
(1)       (2)      (3)       (4)       (5)      (6)       (7)   (3) - (4)    (9)     (10)   -(9) - (10)
1970                    1.1      0.6                 7.0      1.7       4.3      0.6      -2.2     -0.1       0.6       1.7
1971                    1.8      0.2               10.3       2.9       4.0      2.9      -2.5      1.2       0.6       0.7
1972                    6.5      0.3               10.2       5.2       3.6      1.3      -6.8      2.0       0.6       4.2
1973                    0.7      0.5               13.2       4.3       3.9      3.6      -2.6      1.9       0.6       0.1
1974                  -0.4       1.7               30.6       2.7       6.4     12.8     -10.0     -2.4       0.4      11.9
1975                  -0.9       1.0               24.2       4.7       9.3      8.0      -2.2     -0.6       0.5       2.3
1976                    7.1      0.9               16.4       4.2       9.9     -0.8     -11.1      3.8       0.5       6.8
1977                   12.2      3.3               18.9       9.1       8.0     -1.0     -18.2      5.3       1.5      11.4
1978                    9.9      1.7               13.9       8.2       6.6     -3.6     -14.3      3.8       0.5       9.9
1979                   11.7      0.5               19.9       6.0       5.2      2.3     -18.7      5.5       1.6      11.6
1980                    6.7      0.0               23.2       5.4       5.1     -1.0     -20.4      7.3       2.8      10.3



1981                  3.0      2.0              20.6      6.0     4.4      2.0     -13.1      8.4     3.6       1.1
1982                 -1.6      1.4              14.1      7.0      6.1      1.8      1.0      7.4      2.5    -10.9
1983                  2.7      1.2              15.1      6.1      5.9      1.4     -5.6      5.4      3.4     -3.2
1984                 -1.9     -3.4              26.9      3.3      6.3      2.2     -9.9      5.1      5.6     -0.8
1985                 -2.4      2.8              24.7      3.5      6.3      2.1    -13.2      4.0      6.2      3.0
1986                  0.5      2.6               9.6      5.4      6.4     -6.4     -7.3      3.7      1.5      2.1
1987                  4.8     -8.0              20.5      5.0      4.8     -1.0     -8.5      7.3      5.4     -4.1
1988                  1.1      0.2              11.9      4.7      6.5    -4.5      -6.4      6.8     2.9      -3.3
1989                 -3.0     -4.5              25.7      4.9      3.4      1.7     -8.1      9.8      6.6     -8.3
1990                  3.6     -0.7              30.9      4.1      4.4      8.9    -16.4     13.3      7.4     -4.3
1991                  1.6      2.8      4.3     21.2      2.8      7.7      4.4    -15.0     10.9     4.6      -0.5
1992                  5.4     -0.4      0.0     14.1      2.8     3.7     -1.8     -14.5      9.5      3.8      1.2
1993                -12.6      2.2      0.1     13.7      1.8      3.0    -0.6       0.8      6.8      3.4    -11.1
1994                  0.3      3.5      0.0     14.7      1.1      2.4     -2.0    -17 1     20.2      4.1     -7.2
1995                 -3.1      1.7      0.0     12.2      1.2      3.1    -2.3      -8.8      7.5      3.9     -2.6
1996                 -1.4     -5.7      1.6     21.5      1.2     2.5       7.8     -4.6      4.8      5.0     -5.3
1997                  0.2     -1.0      2.7     14.2      1.3      3.8      2.9     -8.2      5.1      3.3     -0.2
1998                 -0.7     -1.4      0.1      0.3      1.2      3.5    -3.1       3.3      1.8    -1.7      -3.4
NPV.                 69.7     12.6      7.4    531.3    128.7    163.7     34.6    287.2    172.4     84.0     30.8
a. In 1999 net present value, at a 5 percent discount rate.
Sources: Garcia, Balza, and Villasmil (1999) and the authors of this chapter.



196                               EASTERLY AND YURAVLIVKER
guaranteed minimum pension for public sector employees; (c) the labor
liabilities of the public sector; and (d) the public guarantee of bank de-
posits. Those estimations are later used to assess the long-term fiscal
sustainability of the public sector in Republica Bolivariana de Venezuela.
The pay-as-you-go social security system that was in place until
1998 collapsed because of various factors, including a growing infor-
mal sector that currently accounts for 53 percent of employment, eva-
sion of payment of contributions to the system, and the use of social
security funds to cover the deficits of the health care system. Had it
not been reformed, that system would have gone bankrupt, affecting
over a half-million people and costing the public sector the equivalent
of 77 percent of GDP in present value terms. The 1998 reform created
a mixed system of personal accumulation funds and a solidarity pen-
sion fund administered by the state. The retirement age of women was
raised to 60 years (as it is for men), contribution rates were doubled to
12-13 percent, and the minimum contribution period was raised from
15 to 20 years.
The fiscal cost of the reform has three components: (a) the existing
retirees and those retiring in 1999 under the old system; (b) the "rec-
ognition bond" to people who contributed to the old system but who
have not yet reached retirement age; and (c) the solidarity pension
fund. Under certain assumptions, the total fiscal impact of the new
system was estimated, in present value terms, at nearly US$60 billion,
equivalent to 65 percent of GDP at the end of 1998.
There are special pension regimes in the public sector that are not
contributive, including those of the central government, the local gov-
ernments, most public enterprises, autonomous agencies, the judicial
system, the national universities, and the armed forces. These regimes
are financed directly from the budgets of these agencies, whose statis-
tical and financial information base is in many cases quite deficient.
At the end of 1998, these regimes had nearly 250,000 retirees and nearly
1 million current public sector employees under their umbrella. A con-
servative estimate of this public liability is US$12.4 billion in present
value terms, equivalent to 13.5 percent of GDP at the end of 1998.
The labor liabilities of the public sector include the accumulated
benefits of one month's salary per year of work per employee. The
1997 reform of the Organic Labor Law abolished the indexation of
those benefits to the last salary and the doubling of the benefits when
the employee was laid off for "unjustified reasons," but extended the
benefit to two months per year. The new law also called for payment
of the arrears of this benefit, which were accumulated under the old
law, over a period of five years, including an additional bond for the
transfer from the old to the new system. By the end of 1998, the total
public debt on this account amounted to US$7 billion, equivalent to
7.6 percent of GDP.



EVALUATING GOVERNMENT NET WORTH                            197
R6publica Bolivariana de Venezuela had two systemic banking crisis
in 1961-63 and 1994-95 and three large bank bankruptcies in between.
In all cases, the central bank (and the deposit insurance guarantee agency,
or FOGADE, in 1994-95) stepped in to cover all deposits, and in the
last crisis it extended that coverage up to a limit of VEB10 million, two
and a half times higher than the VEB4 million established in the original
deposit insurance scheme. At the end of 1998, the total deposits in the
banking system amounted to US$16.3 billion, or 17.7 percent of GDP.
Of those, $7.5 billion consisted of deposits under VEB4 million, which
are guaranteed by FOGADE, and $9.8 billion of deposits under VEB10
million, which was the actual coverage limit after the last banking cri-
sis. FOGADE, on the other hand, had assets of only $760 million. There-
fore, the implicit public guarantee amounts to about $9 billion, equivalent
to nearly 10 percent of GDP. The estimate of the contingent liability
would be lower, depending on the probability of banking losses.
Table 8.7 shows the assets, liabilities, and net worth of the Venezu-
elan public sector in terms of net present value (US$ millions). Assets
include reserves of oil, gas, and coal and international reserves, leaving
out other mineral reserves as well as the value of public utilities and
public enterprises because of estimating difficulties. Liabilities include
external and domestic debt, social security, the labor and pension li-
abilities of the public sector, and bank deposit guarantees. The calcula-
tions assume a growth rate of 4 percent and a discount rate of 9 percent.
As Table 8.7 shows, the difference between assets and liabilities
results in a public sector net worth of US$43.5 billion, equivalent to
4.5 percent of GDP. This net worth could be consumed by allowing the
public sector to have primary noncoil deficits on the order of 2 percent
of GDP on a permanent basis.
lable 8.7 Net Worth of the Public Sector in
R.epublica Bolivariana de Venezuela
(net present value, millions of U.S. dollars)
Assets                                     Liabilities
Oil,                 120,000     External debt           23,613
Gas                   16,000     Domestic debt            3,980
Coal                   2,300     Deposit insurance       10,961
International reserves  14,849   Labor debt and public
pensions              20,032
Seigniorage           10,834     Social security debt    61,921
Total              163,983       Total                120,507
Net worth              43,476
a. Assuming a price of oil of US$16 per barrel.
Sources: Garcia, Balza, and Villasniil ( 19 99) and the authors of this chapter.



198                               EASTERLY AND YURAVLIVKER
This analysis has certain limitations, because the implicit assump-
tion is that the public sector has perfect access to international credit
markets and can borrow on the basis of the net present value of its net
worth. However, as recent history has shown, access to international
credit is anything but perfect, and it depends on changing global con-
ditions. Furthermore, the volatility in oil prices introduces a lot of
uncertainty into any calculation of the value of that asset. At the same
time, these calculations show that continuing the trend of the past 29
years, when the nonoil primary deficit averaged 12.8 percent of GDP,
is clearly unsustainable. Even nonoil primary deficits on the order of 7
percent of GDP, the average of the 1990s, would require an average oil
price of nearly US$20 per barrel. Thus Garcia, Balza, and Villasmil
(1999) concluded that, even under relatively optimistic assumptions,
the Venezuelan public sector could not maintain, on a permanent ba-
sis, nonoil primary deficits of over 6 percent of GDP.
Overdependence on future oil revenues is risky, because within 30
years or so there could be a technological change that would bring
about a sharp decline in the demand for oil. Furthermore, exploiting
and consuming an exhaustible resource raise the issue of intergen-
erational appropriation of the national wealth. In that respect, there
are two approaches to oil revenues: one is to treat them as any other
source of revenue; the second is to treat them as an asset being de-
pleted, making sure that the proceeds of oil are invested and generate
other income in the future, resulting in a constant rent over time.
Apart from the principle of equal rights to the national wealth by
all generations of Venezuelans, the second approach also would pre-
vent a major adjustment in the future, whenever the production of oil
goes down. To implement this approach, the government could invest
part of the oil revenues abroad, to cushion the county from permanent
oil shocks (temporary shocks are addressed by the oil stabilization
fund). Keeping these resources abroad also would reduce overvalua-
tion pressures, which have negative effects both on the fiscal accounts
and on the real sector of the economy. However, the country would
have to balance these considerations with the need to expand infra-
structure investment and human capital accumulation within the coun-
try itself, which means that the economic rate of return on oil revenues
could well be, at least in the next few years, higher when invested
internally than when kept abroad.
Finally, it should be noted that the results are sensitive to the as-
sumptions about GDP growth and the discount rate. Robust, sustain-
able growth is essential for long-term fiscal sustainability, and reducing
the volatility of economic activity (by limiting the impact of oil price
shocks, for example) would have a positive effect in terms of a lower
country risk and the risk premium that Republica Bolivariana de Ven-
ezuela pays on its external borrowing. Also, the quality of fiscal ad-
justment is very important in order not to undermine the infrastructure



EVALUATING GOVERNMENT NET WORTH                           199
development of the country. Thus, apart from improving the efficiency
and effectiveness of public investment and expenditures, there is a need
to increase nonoil revenues and to reduce the contingent and implicit
liabilities of the public sector. In terrns of levels, the nonoil fiscal deficit
should on average be below 4 percent of GDP in order to ensure longer-
term fiscal sustainability.
Summary and Conclusions
This chapter describes two ways to evaluate sustainability using the
balance sheet approach. The first is to estimate the stocks in the
government's balance sheet and assess whether public sector net worth
is positive or negative. If it is negative, then sustainability will require
that the present value of tax revenues minus government consumption
be sufficient to cover the negative net worth.
The second approach is to look at sustainability in flow terms. The
criterion for sustainability would t:hen be to maintain a constant ratio
of net worth to GDP or, if a desired level of net worth can be deter-
mined, to maintain that ratio above the desired level. The basic idea is
that if there is no payments crisis today, then keeping the net worth-to-
GDP ratio constant will avoid a payments crisis in the future. Thus
fiscal sustainability would require keeping the government net worth-
to-GDP ratio constant, or above a minimum desired level. If there is a
payments crisis today, then the rule would imply increasing the ratio
of net worth to GDP.
This chapter presents the conclusions of two country studies, for
Colombia and Republica Bolivariana de Venezuela. The first study
uses the first approach, calculating stocks of assets and liabilities, and
the second study illustrates the second approach, focusing mostly on
changes of assets and liabilities over time. Both, however, have clear-
cut conclusions about the actions needed to achieve longer-term fiscal
sustainability.
For Colombia, Echeverry and others (1999) estimate that the pub-
lic sector has 162 percent of GDPI in assets. However, the Colombian
public sector has liabilities amounting to 232 percent of GDP (the
most important by far is the implicit pension debt, which amounts to
156 percent of GDP), implying a negative net worth of 70 percent of
GDP. In contrast, the total public debt, which is usually the focus of
sustainability analyses, amounts to only 20 percent of GDP. Given this
estimate of net worth, achieving longer-term fiscal sustainability would
require a perpetual primary current surplus of 4.2 percent of GDP.
For Republica Bolivariana de Venezuela, Garcia, Balza, and Villasmil
(1999) reveal that in 1970-98 thie net worth of the public sector de-
clined by nearly US$300 billion, the equivalent of 287 percent of 1999
GDP. This decline shows the drop in assets, or rise in liabilities, used



200                                  EASTERLY AND YURAVLIVKER
to: (a) finance current public consumption (net of nonoil revenues)
amounting to $177 billion, and (b) subsidize the domestic consump-
tion of oil products (which were sold not only at below-border prices
but even below their production costs for most of the period) by $86
billion in net present value terms.
The difference between Venezuelan assets and liabilities results in a
public sector net worth of US$43.5 billion, equivalent to 45 percent of
GDP. That would allow the public sector to have nonoil deficits on the
order of 2-6 percent of GDP on a permanent basis, depending on the
assumed future oil price.
These calculations show that continuing the trend of the past 29
years, when the nonoil primary deficit averaged 12.8 percent of GDP,
is clearly unsustainable. Even nonoil primary deficits on the order of 7
percent of GDP, the average of the 1990s, would require an average oil
price of nearly US$20 per barrel. Thus Garcia, Balza, and Villasmil
concluded that, even under relatively optimistic assumptions, the Ven-
ezuelan public sector cannot maintain, on a permanent basis, nonoil
primary deficits of over 6 percent of GDP.
In conclusion, our picture of public finance in two test cases, Co-
lombia and Republica Bolivariana de Venezuela, was dramatically al-
tered by using the balance sheet approach to public finances. We found
both treasures and time bombs in the governments' balance sheets.
Notes
1. This approach is used as well to assess the consistency between a bud-
get deficit and inflation targets, where not only the debt ratio is presumed
constant but also the money-to-GDP ratio. Revenues from money creation
are calculated as the sum of the inflation rate times the money-to-GDP ratio
(the inflation tax) and the growth rate times the money-to-GDP ratio (seignior-
age). Economists have used these calculations on many occasions to assess
the consistency of the fiscal stance with avoiding a debt crisis and excessive
inflation. The classic references are Buiter (1983, 1985) and Anand and van
Wijnbergen (1989). Other examples of references are Marshall and Schmidt-
Hebbel (1994) for Chile, Haque and Montiel (1994) for Pakistan, and
Morande and Schmidt-Hebbel (1994) for Zimbabwe.
2. For an illustration of the shortcomings of conventional sustainability
analysis, see Blejer and Cheasty (1991), Polackova (1998), and Easterly (1999).
3. Buiter (1983, 1985) pioneered the use of the balance sheet approach to
fiscal accounts. The government of New Zealand has pioneered the use of
the balance sheet approach in its fiscal accounting (Scott 1996).
4. Some assets are omitted because they are impossible to measure, such
as the value of the human capital embodied in the government's work force.



EVALUATING GOVERNMENT NET 'VORTH                              201
References
The word processed describes informally reproduced works that may not be
commonly available through libraries.
Anand, Ritu, and Sweder van Wijnbergen. 1989. "Inflation and the Financ-
ing of Government Expenditure: An Introductory Analysis with an Appli-
cation to Turkey." World Bank Economic Review 3 (January): 17-38.
Blejer, Mario I., and Andrienne Cheasty. 1991. "The Measurement of Fiscal
Deficits: Analytical and Methodological Issues." Journal of Economic
Literature 29: 1644-78.
Buiter, W 1983. "Measurement of the Public Sector and Its Implications for
Policy Evaluation and Design." IMIP Staff Papers. (Updated and reprinted
in Mario 1. Blejer and Andrienne Cheasty, eds. 1993. How to Measure the
Public Deficit: Analytical and Methodological Issues. Washington, D.C.:
International Monetary Fund.)
-    . 1985. "A Guide to Public Sector Debt and Deficits." Economic Policy
(November): 14-79.
Easterly, William. 1999. "When Is Fiscal Adjustment an Illusion?" Economic
Policy (April): 57-86.
Echeverry, Juan Carlos, Maria Victoria Angulo, Gustavo Hernandez, Israel
Fainboim, Cielo Numpaque, Gabriel Piraquive, Carlos Rodriguez, and
Natalia Salazar. 1999. "El Balance del Sector Publico y la Sostenibilidad
Fiscal de Colombia." Archivos de Macroeconomia. No. 115. National
Planning Department, Bogota.
Garcia Osio, Gustavo, Rafael Rodriguez Balza, and Ricardo Villasmil. 1999.
"Ajuste y Sostenibilidad Fiscal de Largo Plazo-El Caso de Venezuela."
Oficina de Asesoria Economica y Financiera del Congreso de la Repub-
lica, Caracas. Processed.
Haque, Nadeem Ul, and Peter Montie:i. 1994. "Pakistan: Fiscal Sustainability
and Macroeconomic Policy." In William Easterly, Carlos Alfredo
Rodriguez, and Klaus Schmidt-Hebbel, eds., Public Sector Deficits and
Macroeconomic Performance. Washington, D.C.: World Bank.
Marshall, Jorge, and Klaus Schmidt.Hebbel. 1994. "Chile: Fiscal Adjust-
ment and Successful Performance." In William Easterly, Carlos Alfredo
Rodriguez, and Klaus Schmidt-Hebbel, eds., Public Sector Deficits and
Macroeconomic Performance. Washington, D.C.: World Bank.
Morande, Felipe, and Klaus Schmidt-Hebbel. 1994. "Zimbabwe: Fiscal Dis-
equilibria and Low Growth." In William Easterly, Carlos Alfredo
Rodriguez, and Klaus Schmidt-Hebbel, eds., Public Sector Deficits and
Macroeconomic Performance. Washington, D.C.: World Bank.
Polackova, H. 1998. "Contingent Liabilities: A Threat to Fiscal Stability."
World Bank PREM Notes. No. 9. Washington, D.C.
Scott, G. 1996. "Government Reform in New Zealand." IMF Occasional
Paper 140. International Monetary Fund, Washington, D.C.






CHAPTER 9
The Challenges of
Fiscal Risks irn Transition:
Czech Republic, Hungary,
and Bulgaria
Hana Polackova Brixi
World Bank
Allen Schick
University of Maryland
Leila Zlaoui
World Bank
AN ESSENTIAL STEP IN controlling the expansion of government con-
tingent liabilities and reducing fiscal risk is being able to identify and
measure them. In this chapter we discuss how this may be done, and
we demonstrate how an assessment of fiscal adjustment may change
substantially when a broader picture of government obligations is in-
cluded. The chapter is based on a 1999 analysis of fiscal adjustment in
the Czech Republic, Hungary, and Bulgaria.
The Czech case provides an example of officially balanced govern-
ment budgets and of the extensive use of guarantees and other forms
of off-budget support. The case study shows how to deal with some
(lifficult conceptual and measurenient issues when estimating govern-
inent contingent liabilities and the unreported portion of fiscal deficit.
In contrast to the Czech Republic, Hungary internalized most fiscal
r isks in the government debt and constrained off-budget fiscal activi-
ties. While the Czech Republic enjoyed "budget balance," Hungary
203



204                               BRIXI, SCHICK, AND ZLAOUI
faced high budget deficit and debt levels. Less-visible aspects of fiscal
adjustment, however, pull the comparison of the fiscal performance of
thc two countries in the opposite direction. Somewhat like Hungary,
Bulgaria maintained low, transparent government exposure to fiscal
risk. And somewhat like the Czech Republic, Bulgaria has been com-
mitted to low budget deficits and macroeconomic stability-since 1997
at least, after having introduced a currency board arrangement. In
contrast to both the Czech Republic and Hungary, however, Bulgaria
has been slow in the transition process and has yet to meet the fiscal
challenges of the needed enterprise restructuring and investment in
infrastructure and environment.
Although each of the three countries has taken a different path to
fiscal adjustment, economic realities and opportunities may lead to
more similar behavior in the future. As the Czech Republic's hidden
liabilities came to light in 1998, the government faced pressure to im-
pose discipline in resolving old and taking on new fiscal risks. As for
Hungary, the favorable fiscal performance in the second half of the
1990s emboldened the government to take on greater liabilities than it
had been able to handle during the adjustment period. With the Czech
government taking a tougher stand on contingent liabilities and the
Hungarian government showing signs of loosening its established con-
trols, a comparison of the two countries in the future may show a
different pattern than that evident at the end of the 1990s. Bulgaria's
recent favorable fiscal performance has yet to be tested by the tradeoff
of fiscal prudence versus a more aggressive strategy toward the country's
development and accession to the European Union (EU). All three coun-
tries consider their accession to the European Union a key policy pri-
ority, and their motivation to meet the EU accession requirements
(including requirements on the quality of infrastructure and environ-
ment) is high.
The three sections that follow apply the Fiscal Risk Matrix (pre-
sented in Chapter 1 of this volume) and analyze the "true" fiscal posi-
tion of the Czech Republic, Hungary, and Bulgaria. The main
conclusions and suggestions for future work are summarized in the
final section.
Measuring the True Fiscal Deficit
of the Czech Republic
The Czech Republic has been known for its balanced government bud-
gets. In contrast to most countries, however, fiscal performance in the
Czech Republic encompasses a significant number of government ac-
tivities financed outside the budgetary system. These activities gener-
ate fiscal risks. Recently, these off-budget fiscal risks have become more



THE CHALLENGES OF FISCAL RISKS IN TRANSITION                           20s
visible, because state guarantees and agencies that are either explicitly
or implicitly guaranteed by the government have generated significant
claimis on the budget (see Table 9.1).
Given the magnitude of off-budget activities, fiscal analysis in the
Czech Republic should identify all the main activities of a fiscal nature
Table 9.1 Fiscal Risk Matrix, Czech Republic
Contingent liabilities
Souirces of        Direct liabilities              (obligation if a
obligations    (obligation in any event)      particular event occurs)
Explicit
Government     * Foreign and domestic   * State guarantees
liability as      sovereign debt        * Liabilities and other obligations
recognized     * Budget expenditures      of Konsolidacni Banka
by a law or    * Future legally bind-   * Obligations of Ceska Exportni,
contract         ing expenditures         EGAP (an export guarantee fund),
and Deposit Insurance Fund
Implicit
A "moral"      * Future investment ex-  * Obligations of National Property
obligation of    penditures to meet EU    Fund (own debt, guarantees,
government       accession requirements   and obligations to Ceska Inkasni,
that reflects  * Future recurrent ex-     Land Fund, and similar entities)
public and       penditures related to  * Liabilities of Ceska Financni and
interest group   public investment        Czech National Bank (result of
pressures        projects                 the central bank's nonstandard
* Military expenditures    operations)
as required by North   * Liabilities of banks (Komercni
Atlantic Treaty Organ-   Banka, Ceska Sporitelna)
ization (NATO)          * Further losses and defaults of
large enterprises (Czech Railways)
* Obligations of PGRLF (an agri-
cultural credit and guarantee fund)
* Liabilities and other obligations
of subnational governments
* Liabilities of credit unions
(Kampelicka) and private
pension funds
Note: The liabilities listed refer to the fiscal authorities of the central government.
Because the government is legally obliged tc pay future public pensions (a public pay-as-
you-go pension scheme), future pensions constitute a direct (expected with certainty)
explicit (legal) liability. The expected investrnent expenditures that are needed to meet EU
accession requirements are the major direct implicit liability. State guarantees and financ-
irig through state-guaranteed institutions ale key examples of explicit contingent liabili-
ties. And, like that of many other countries, the financial system represents the most
serious source of implicit contingent liabilities for the Czech government.
Source: The authors.



206                                    BRIXI, SCHICK, AND ZLAOUI
in order to determine the "true fiscal deficit." Excluding the quasi-
fiscal activities of the central bank, the Czech National Bank, the "hid-
den" part of the fiscal deficit comprises two main components: (a) net
spending on programs of a fiscal nature by special off-budget institu-
tions-Konsolidacni Banka (KOB), Ceska Inkasni (CI), Ceska Financni
(CF),' and the National Property Fund (NPF); and (b) implied subsi-
dies extended through state guarantees. For the financial relationships
of the special institutions, see Figure 9.1 (developed by the Ministry of
Finance of the Czech Republic).
For any given year, net public spending by these institutions in-
cludes cash outlays on new programs in the form of directed credits
and asset purchases,2 and interest expenditures. This spending is ad-
justed for debt collection, interest revenue, and other revenue from
programs. Table 9.2 shows the components of the "hidden" deficit. In
the remainder of this section, we explain each row of this table in
deta