_V/Ps r"(q
POLICY RESEARCH WORKING PAPER    1769
Information, Incentives,                                        Writteri perornrna-
c ontra3cts,^'i   tao.t;- t. -is C'
and Commitment                                                  way ?heipsta:fo C
enterprise irrprpove
An Empirical Analysis of Contracts                              perfoir   sem f
W\Xhy?   ei-,-. '  itrot rTnltia,;n
between  Government and State                                   asytc ryie;,v,  of
Enterprises                                                     qovo-n ri -t rC,'
and lc-k  ;f  -
Mary M. Shirley                                                   lfrlt     "d ,d tC w'A
L. Colin Xli                                                   incerntrvJ   nir q16
The World Bank
Policy Research Department
Finance and Private Sector Development Division
May 1997



PoI,ICY RFSEAI'(1H WORKING  PAPER 1769
Summary findings
Shirley and Xu anal ze experience with written                  Performance contracting assumes that government's
performance contra.its between developing country             objectives can be maximized, and performance
governments and the maniagers of their state-owned            improved, by setting targets that take into account the
enterprises. Such contracts have been a vogue since the       constraints placed on managers. For this to occur, the
mid-1980s, and substantial resources have been sunk into    principals must be willing to explicitly state their
their design and eniforcement, yet the few assessments to     objectives, assign to them priorities and weights, translate
date show mixed results.                                      them into performance improvement targets, provide
Using a simple agency model, they identify how              incentives to meet those targets (or monitor the agents
problems of weak incentives stemmning from information        without incurring significant costs), and credibly signal
asymmetry. lack of government commitment, and lack of    their commitment to the contract. These conditions
managerial commitment can lead to shirking.                   failed to materialize
TIlhey appiv the model to a sample of 12 contracts with       Why would governments adopt contracts to which
monopoiy eitterpriscs in six developing countries             they were not committed or that were politically
(Ghana, India, the Republic of Korea, Mexico, the             unrealistic? Sometimes because it enabled them to get
Philippines, ancd Senegal). All suffer from serious           foreign assistance.
contracting problems. They find no pattern of improved          How explain the managers' lack of commitment? Not
performance that can be attributed to the contracts.          surprisingly, managers with information advantages and
Only three of the 12 case-study companies showed a          bargaining power, and with no strong incentives or
turnarounlld in total factor productivity after contracts     commitment from the government, used their advantages
were introduced, six c,ontinued past trends, and three        to manipulate the targets so as to ensure that their
perfornmed substantially worse under contracts than they      performance would be judged satisfactory.
had before. Labor productivity improved at a faster pace        Shirley and Xu outline the conditions under which
m four cases, and deteriorated in nonie, but the              performance contracts might succeed in improving
mrproveoenrt predated the co; tFact.                          performance.
This paper -  a prodLuct of the Finance and Private Sector Development Division, Policy Research Department -  is part
of a larger effort in the department to analyze the role of institutions and incentives in public sector management, as well
as in the private sector. The study was funded by the Bank's Research Support Budget under the research project "The
Changing Role of the State" (RPO 678-69). Copies of this paper are available free from the World Bank, 1818 H Street
NW, Washington, DC 20433. Please contact Paulina Sintim-Aboagye, room N9-030, telephone 202-473-8526, fax 202-
522-1155, Internet address psintimaboagye(ofworldbank.org. May 1997. (48 pages)
Ike P'ohcy Research Working Paper Series disseminates the findings of uwork in progress to encourage the exchange of ideas about
deveiopment issues. An ohbective of the series is to get the findinrgs out quickly, even if the presentations are less than fully polished. The
papers carry the namnes of the authors anrd should be cited accordingly. I*e findings, interpretations, and conclusions expressed in this
pap.er are entirely those of the authors. They do not necessarily represent the vieuw of the World Bank, its Executive Directors, or the
,ou,r'tres they represent.
Pro(duiced bv the Policy Research Disseminiation Ccnter



Please do not quote or cite
without
permission of the authors.
Information, Incentives and Commitment: An Empirical Analysis of
Contracts between Government and State Enterprises
by
Mary M. Shirley and L. Colin Xu
The authors are grateful for useful comments from Gerard Caprio, and for the
contribution of Luke Haggarty, as well as the assistance of Zeny Kranzer, Bill Moore and
Polly Means. The paper draws on work done by Mary Shirley for World Bank,
Bureaucrats in Business: The Economics and Politics of Government Ownership,
(Washington, D.C.: Oxford University Press, 1995) and the acknowledgements in that
work apply here as well.



2
I. Introduction
Written contracts between governments and state-owned enterprises (SOE) have
enjoyed a vogue since the mid- 1980s. Our survey of developing countries found 565 such
contracts in 32 countries plus another 103,000 in China as of June, 1994 (table 1). We
defined these contracts, which we termed performance contracts, as negotiated, written
agreements between governments and the managers of their SOEs which specify explicit
targets that SOE management pledges to achieve in a given time frame, and where
performance is measured at the end of a specified period. I Substantial resources have
been sunk into the design and enforcement of these contracts, yet the few assessments to
date show mixed results (see, for example, Shirley 1989, 1991; Nellis 1989; Trivedi
1990).
SOEs face more than the usual principal-agent problems. Typically, the agent
(SOE management in this case) has more information than the principal (government)
and can choose how much effort to expend. (For a review of the theoretical arguments
see, Ross 1991; Stiglitz 1974; Sappington 1991.) The principal can only observe
outcomes and cannot measure accurately the effort expended by the agent or distinguish
the effects of effort from other factors affecting performance (Laffont and Tirole 1986).
The assumption in much of the literature is that the principal is a residual claimant whose
objective is to maximize his or her returns, and will monitor management effort
accordingly. If there are multiple principals, they will share the same objective (although
smaller stakeholders may try to free ride on the monitoring efforts of larger ones). If the



3
shares of the enterprise are traded in a highly liquid stock market, the stockowners may
choose to exit, in which case a large shareholder may take over the firm, oust the old
manager and appoint a new, more capable one. The threat of such a takeover may act as a
check on the behavior of incumbent managers.
SOEs, however, have no clear residual claimants; they are subject to multiple
principals who impose multiple, often conflicting objectives and constraints on the
enterprise (Jones, "The Linkage Between Objectives and Control Mechanisms in the
Public Manufacturing Systems" in Trevedi, 1990). These multiple principals may derive
benefits from objectives which run contrary to improving economic performance. For
example, politicians may benefit politically from instructing SOEs to maximize
employment (Shleifer and Vishny 1994). Bureaucrats might benefit from instructing
SOEs to engage in activities that increase their power, prestige or perks (Buchannan et. al
1980). Besides the lack of a clear residual claimant, SOEs are often monopolies not
subject to bankruptcy, and are also not vulnerable to takeover (Vickers and Yarrow
1988). As a result, it is especially difficult to judge managerial performance or to
motivate managers to exert effort.
A written performance contract can be seen as a forcing mechanism to clarify
objectives and make it easier to judge performance and motivate managers. The contract
is negotiated between representatives of the multiple principals and the SOE, and
translates the multiple goals into targets measured by specified criteria (Jones,
"Performance Evaluation for State-Owned Enterprises" in Ramamurti and Vernon 1991).
Targets can be weighted to reflect government priorities and provide further clarity.



4
Management can be motivated to improve performance through bonuses based on
achievement of the targets, while poor performance can be penalized through firing or
demotion.
An implicit assumption underlying such contracts is that even as government's
social or political objectives are being maximized, performance on economic measures
can still be improved, by setting targets that allow for improvements within the
constraints imposed by these noneconomic objectives. For example, even though a
government requires a SOE to keep its labor force at current levels despite overstaffing,
the company could still achieve contractual targets aimed at, for instance, improving
quality or minimizing materials costs. Constraints are taken into account by selecting
targets and/or criteria that measure only those factors which management can control.
For instance, performance can be judged against past trends, rather than against some
objective standard, to take account of factors which management can't change much from
year to year. Thus, even if management can't reduce overstaffing they may still be able
to improve the trend in labor productivity by expanding production through better use of
plant and equipment. Or, to give another example, trends in profitability may be
measured in constant prices to take account of management's inability to raise prices in
real terms.
Drawing on the work of Sappington (1991), Lafont and Tirole (1987) and
Willamson (1976 and 1985),2 we hypothesized that how well a contract does in
improving SOE economic performance depends on how well it addresses three
interrelated problems: information asymmetry, incentives, and commitment. Information



5
problems arise because the contracting agents have different sets of information, and it
may be costly to get information about whether the terms have been honored; thus each
side can use the information it holds exclusively to improve its position at the expense of
the other. At the same time, because not all contingencies can be fully anticipated and
accounted for, it is impossible to design a contract that will cover all eventualities. To
induce the contracting parties to comply with the contract's provisions, contracts include
promises of incentives (rewards and penalties). But where information is imperfect and
it is costly for government to pay the incentive, the reward may be set too low to motivate
managers to improve performance. Since managers face a disutility of effort, under such
low incentives they will shirk.
Each party must be convinced of the commitment of the other to honor the
contract now and as circumstances change, but when government is one of the
contracting parties it may be difficult or impossible to force government to comply.
There may be no neutral third parties with the power to compel the government or its
successors to meet its promises. Government in developing countries may face especially
serious problems making credible commitments because the institutions (checks and
balances, reputation, etc.) which curb arbitrary actions and bind successor administrations
to the promises of their predecessors are weak (North and Weingast 1989; Levy and
Spiller 1996). If managers believe that there is a high probability that government will
renege on its commitments (such as the commitment to pay an incentive), they will not
invest in learning or exert effort to improve results. Managerial commitment is also



6
important. If managers are not committed to the contract, they may try to bargain that
they are not punished regardless of the firm's performance.
The next section of this article presents a simple, agency theoretic model to
illustrate how these contracting problems may affect performance. Section III examines
how a sample of contracts dealt with each of the contracting problems and Section IV
measures the success or failure of the contracts against objective, quantitative measures of
firm performance. Section V concludes with an assessment of the findings and the policy
implications.
II.   The Model
To illustrate how the problems of performance contracts affect SOE performance,
we first develop a static model to show that, when enterprises have an information
advantage the government will choose lower-powered incentives, which lead to an
efficiency loss compared to when the government has complete information. Next we
give two examples under a dynamic setup to show how the lack of commitment by either
of the contracting parties can hurt SOE preformance. The theoretical arguments offered
here emphasize the role of the manager in enterprise performance, and synthesize
standard arguments of information economics in the context of SOE performance
contracts.
A.   Information Asymmetry, Incentives and Performance: Static Analysis



7
We make use of Freixas et al (1985) to compare the performance of an enterprise
with and without an information advantage.3 Consider an enterprise which produces
output y of a product whose social value is p. The technology is characterized by
(1)    y = eO
where e is managerial effort, and 0 is the firm-specific efficiency level, beyond the
control of the manager. Other inputs such as capital and labor are omitted based on the
assumption that such inputs are observable to the government. By working hard the
manager incurs disutility y (e), with y '(e) > 0, and tj "(e) > 0
The government tries to motivate the manager using the following performance
contract:
(2)    R(y) = a + by = a + beO
where R is the reward to the manager: a is the manager's base salary independent of the
level of output and assumed to be positive, and b represents the power of incentives.
Linearity of the performance contract is assumed to simplify the analysis.4 In transferring
$1 to an enterprise, the government incurs costs of $(1 + ?), where k is the additional
costs associated with distortionary taxes and with administrating enterprises by
bureaucrats (such as planning, monitoring, and verifying inputs and output). The
manager maximizes his utility function,
(3)   7t F =a+ beO -(e),
while the social welfare of the rest of the world is
(4)   7r ° = peO - (1 + X)(a + beO)
So the objective function of a benevolent government is



8
(5)    W=peO-y,(e)-X(a+beO)
The Optimal Static Incentive Scheme under Complete Information
There are two types of firms: firms with better technology (9 ), and the others (0). With
complete information the government knows the type of an enterprise, and since it is
costly to transfer funds to the enterprise, the government will maximize W while making
the manager's utility just equal to his or her reservation utility, which without loss of
generality is assumed to be 0. So the government hopes to solve the optimal incentive
intensity b from the following problem:
Max      W = peO -y (e) - k(a + beO)
=6_(1+)[ P  eO-(a+beO)]+(a+beO -y(e))
(6)                         
s.t.  a + beO = v (e)
v '(e) = bO
The last equation is the rationality constraint -- the first order conditon for the manager to
maximize i F; the equation before the last one reflects the government's intention to
leave no rent to the firm. Substituting the first constraint into the objective function,
taking a derivative with respect to b, then substituting y '(e) = bO , we get the optimal
incentive intensity with perfect information:
(7)   b =  P    5
The right hand, denoted as s, is the shadow price of output to the government: the
government incurs a social cost of $ (1 + X) to induce the enterprise to produce $1 worth
of output.
Static Incentive Scheme Under Incomplete Information



9
Now suppose the enterprise manager has an information advantage. While the manager
knows his or her enterprise's type (6 orO ), the government presumes that a firm's
technology parameter is:
0 O with probability (1- v)
=    with probability v
Denote e(b), e(b) as the utility-maximizing efforts when the bonus intensity is b for firms
with type 0 and 0 . Since firms with better technology (0 type) have a higher marginal
product, the first-order condition of the enterprise implies that e(b) > e(b).
The maximization problem of the government now becomes
max W = v{p0e(b) -y (e(b)) - X[a + bOe(b)]}
+((  - v){p0e(b) -W (e(b)) - X[a + bOe(b)]}
(9)  s.t  v (e(b)) = a + bOe(b)
(10)           bO W=i'(e)
bO =  '(e)
Constraint (9) says that the government will design a payment scheme so that the
manager of an inefficient type will receive his or her reservation utility level and nothing
more: the government hates to transfer any more than necessary to its SOEs since it costs
the society an extra $ X . Constraint (10) implies that managers of both types of enterprise
will maximize their utilities.
Substitute (9) and (10) into (8), we get
W = v{jp0e(b) - (1 +  )Wi (e(b))} + (1 - v){pOe(b) - v (e(b)) - X' (e(b)) - X b[Oe(b) - 0e(b)]}
Taking the derivative with respect to b and setting it to zero, we get:5



10
X__                 Oe
(11)  s-b*=       (l-v)        +-e Oe
(11) implies that b* < s . So when the firm 's manager has an information advantage, the
government will set a lower incentive than in the case ofperfect information (in which
b *=s). Since e'(b) > 0 and e'(b) > 0, it follows that the performance of both types of
firms will be lower under imperfect information than under perfect information.
B. Commitment and Effectiveness of Performance Contracts
The effectiveness of contracts depends not only on information and incentives but
also on commitment. Although there are various ways in which lack of commitment is
important to contracts, here we illustrate two types of lack of commitment: the
government's commitment not to appropriate firm-specific investments of the manager;
and the manager's commitment to stick to the terms of the performance contract.
Expropriation Risks due to the Lack of Commitment by the Government
In this example the lack of commitment by the government reduces the returns of
firm-specific investment and the manager's efforts. Consider a two-period setup: t=O, 1.
In period 0, the manager sinks a firm-specific, non-contractible investment of I, which
will improve output of period 1 by K(I). The manager's investment can be viewed as his
or her time spent on learning about the firm (its technology, personnel, and market),
which is lost when he or she leaves the firm, as well as his or her efforts in managing the
firm. We assume K'(I) > 0, and K"(I) < 0; So returns to investment has diminishing
returns. The output in period 1 is e,O, + K(I).



11
The government announces its plans to reward the enterprise with the static
optimal compensation scheme: R, = a + by,. This is justified by a well-known result in
regulation theory that in a stable dynamic setup with asymmetric information, it is
optimal to repeat the optimal static incentive scheme in each period.6 The government,
however, deviates from the optimal in that it has a commitment problem: based on past
observations, the manager expects the government to renege on its performance contract
with probability X (m) . In other words,
with probability  the government will compensate the manager regardless of
X (m):          outcomes, R, = R,, a constant exogenously set.
with probability the government will abide by the performance contract so that
I -Tr (m) :     R, =a+ by,.
A larger m implies a stronger commitment to performance contract,  (m) < 0. The
dm
manager decides to work at the intensity e, at the beginning of period t, and decides to
invest I at the beginning of period 0; the government rewards the firm at the end of each
period.
Assume the intertemporal discount rate is , , then the utility function is
WF = (1-t(m))(a +beo00) +t(m)R0 - .i(eo)-I
+3 {(I -T m.)[a + b(eiO, + K(I))] + T (m)R, - x(e,)}
The manager's optimal investment and effort are solved from the first order conditions:
1 = [1 -t(m)] P bk'(I)
(13)  N4 '(eo) = [1 -t(m)]b0O0
T   '(e  ) = [1-st (m)] bO 
The comparative-static analysis yields that



12
a   > 0, and A-    > 0, where X = I, eO, e,
It then follows that ' > 0, t = 0,1 . So a stronger commitment of the government will
am
improve the performance of enterprises by encouraging investment and inducing larger
effort per period.7
The Manipulative Firm
In this subsection we consider an example in which the manager can use his or her
bargaining power to ex post reduce the threshold for punishment in his or her own favor.
We might not observe the manager reducing the punishment threshold in firms with
substantial bargaining power because they can manipulate the performance criteria ex
ante (though this manipulation does not fall under the category of commitment). In the
extreme, a manipulative firm can set itself the targets within easy reach. In general, the
more bargaining power a firm has (in part due to its information advantage), the easier the
targets the firm can obtain; as a result, the lower the effort and learning the manager and
the employees will yield, thus the worse the performance.
In firms without such strong ex ante bargaining power, or firms with such powers
yet can still bargain down performance criteria ex post, we will see managerial non-
commitment when the manager expects to be able to bargain down the performance
criterion ex post. The government hopes to use the following performance contract,
which punishes the manager with no rewards when R, < R, a constant:
Event:                Reward to the manager R,:
Period 0:  a+ beoO < R           O



13
a+be0O 2 R            a + beO
Period I   a+b(e, + k(1))e < R   0
a+b(e, +k(I))0 > R    a+b(e, +k(I))0
The manager ex ante agrees to the terms, knowing that in the second period he or
she can bargain for a lower performance criterion R < R, and thus avoid punishment as
long as R, > R. Then, the manager is motivated by the expectation of the ex post rewards
and punishment, summarized as follows:
Event:                 Reward to the manager R,:
Period O:  a+be0O <R             0
a+ beOo 2 R           a + beO3
PeriodI   a+b(e +k(I))O <R       0
a+ b(e, + k(I))O > R  a + b(e, + k(I))O
A crucial but plausible assumption for the following proof is that 09(0) is
monotonously increasing over the relevant range when 0 belongs to the range defined by
K < R, < R R. In our context, since the government is likely to set R well below the mean
or the median of 0, and R is even lower, 0p (0) is most likely increasing over the range
bounded by the two thresholds.
Now compare the objective functions with and without managerial commitment.
The manager is committed when he or she exerts effort to exceed the punishment
threshold R in both periods, and is not committed when he or she expects to bargain
down the punishment threshold to R (R < R) in the second period. The managerial
objective functions with and without commitment differ only when R < RI < R.
Therefore,



14
R-a
h(e,+K(I))
(14)  UNC -UC        f  (a +Kb( (e, +K(I))(pO)dO
R-a
b(e1+K(I))
Obviously, optimal e( is the same with or without managerial commitment. Yet,
optimal e, and I will decrease, as the following can be proved:9
MRNC - MRC1 <0
MR I  - MRI <0
NC      C
Thus in this example, relative to managerial commitment, managerial non-
commitment leads to declining efforts of managers over time, and a suboptimal
investment. Empirically, managerial non-commitment to contract should be shown in
decling stringency of targets over time.
III. Empirical Application
Our model suggests that information asymmetries lead governments to provide
low powered incentives that don't induce managers to exert as much effort as under
perfect information. It further predicts that lack of commitment will also lead to lower
performance. To test these hypotheses we first examined how performance contracts
addressed problems of information, incentives and commitment in a sample of twelve
companies in six countries, as shown in table 2. (For ease of reading we have used
simplified enterprise names; see table 2.) We then measured performance trends after the
year of the contract and compared them with pre-contract trends.



15
Our analysis is based on the contract documents and reports from the monitoring
agency, a questionnaire administered to each sample company and government
monitoring agency, World Bank file data, interviews in the field and with knowledgeable
World Bank staff, and the enterprises' own audited accounts and reports.'I The sample is
small and not random, so care must be used in generalizing from the results." I Also there
are weaknesses in the data: accounting is weak in Ghana and Senegal, and in some cases
our precontract period is short and the length of the postcontract period varies, which
makes it hard to measure trends.12 Even so, the sample does include countries at very
different levels of income that employed varying approaches to performance contracting.
An improvement in postcontract performance across such differing country and contract
experiences would suggest that the contracts can reduce the three contracting problems
(unless other exogenous factors were at work, something we also investigated).'3
A. Information Asymmetries and Incentives
Our first hypothesis is that in the presence of information asymmetries governments
are likely to use lower powered incentives. Competition between firms for the market or
bidding between potential managers for the privilege of operating the firm increases
government's information by allowing comparisons between the firms' performance or
the managers' bids. However, all of the sample enterprises were monopolies and none of
the contracts were competitively bid, so this form of information revelation was not
employed.
The information disadvantage of governments was also manifested by that all of the
government agencies responsible for negotiating, monitoring and evaluating (henceforth



16
supervisors) used SOEs' annual, audited accounts for the information, but the quality of
these data varied, and in five of the cases (three in Ghana and two in Senegal) were very
poor. For example, in 1991, twelve of the seventeen SOEs supervised by the State
Enterprise Commission of Ghana (or SEC) were seven weeks late in delivering their 1990
fourth quarter results. Only five of the firms had current audited accounts; the most
recent audited accounts for the others ranged from one to three years old (State Enterprise
Commission 1991). Similar problems were cited in interviews in Senegal, where
enterprises were also reported to evaluate themselves.'4 Moreover, as we show below,
the enterprise managers had considerable bargaining power in all the cases except the two
SOEs in Korea, which made it difficult for the government supervisors to demand better
information. The low status and pay of the government bureaucrats responsible for
supervising the contracts meant that the skills available to assess and process the
information were also low. Korea, and to a lesser extent, India, used outside experts to
supplement the skills of the bureaucrats which helped reduce its information problems
(see section III.C).
As expected, the information problems led to low powered incentives.15 Managers
and staff could receive a pecuniary bonus for good achievement of their targets in only
two of the 12 sample contracts (Korea Electricity and Korea Telecoms). In the two
Korean cases, the staff of the company could get a bonus of up to three months salary for
"A" graded performance. Bonuses are also part of the contracts in the three sample
contracts in Ghana, but according to survey responses promised bonuses have not been
paid. In two other contracts (India Electricity and Oil) managers could receive an award



17
for good performance in a public ceremony, but interviews suggest that this was not
viewed as a very strong motivation.
Thus, with the exception of Korea, governments opted for low powered incentives.
The bonuses in Korea were high powered incentives in part because they were paid to all
staff; managers reported feeling pressured to improve performance so as not to disappoint
their employees (Shirley 1990). Moreover, bonus ratings were reported in the media,
which managers stated was a strong motivating force (Ibid.). Since Korea had less
information problems than the other countries, its experience lends support to our
conjecture that information asymmetry leads to lower power incentives.
B. Lack of Government Commitment
Our model next suggests that the expectation that government will renege will lead
managers to underinvest in improving the performance of their firms. While the model
suggests one way governments might renege is by not paying the promised bonus, the
government could also renege by not taking the actions necessary for management to
achieve the targets. Governments made explicit promises in all but four of the contracts
(the exceptions are the two Korean and two Philippines contracts) to take actions that
were crucial to the successful achievement of the targets. Although the contracts in the
Philippines did not include such obligations, many of the targets (such as financial and
profitability measures in current prices or labor productivity) partly depended on
government actions (such as allowing tariff adjustments or lay offs) to be achieved, which
could be taken as an implicit promise. Only in the cases of Korea Telecoms and
Electricity were the targets - for example, transmission losses, thermal efficiency,



18
minutes of outages per household, and profitability in constant prices-not directly
dependent on government decisions. The governments in our sample not only controlled
prices of factors and products, decisions about wages and layoffs, procurement above
minimal levels, and all senior appointments, they also strongly influenced whether the
SOE was paid by its customers (and had to pay its suppliers or creditors). Eighty eight
percent of respondents to our questionnaire cited one of the factors under government
control (hiring, firing, payment of government's and other SOEs' receivables to the firm,
pricing policy, etc.) as the number one obstacle interfering with the achievement of the
objectives of the contracts.16
To assess whether government was credible in its commitment to provide the
promised incentives and to honor its promises under the contract, we looked for the
presence of enforcement mechanisms. We also checked whether governments took
costly actions to signal commitment to managers (see Spence 1973; Lupia and
McCubbins 1996) for discussions of costly action to signal credibility). Since the sample
contracts were all repeated games (most were annual, see table 2), we also looked at
government's compliance in the early contracts as a signal of its credibility in later
contracts.
Enforcement. None of the contracts specify an enforcement mechanism in the sense
of a neutral third party, insulated from politics, with the power and information to compel
both parties to comply. In several cases, however, government obligations were
potentially enforceable by an outside party. The World Bank included government
contractual obligations as covenants in its loans for SOEs in Ghana, Mexico and Senegal.



19
(Although these covenants were drawn up on the basis of project or sector needs rather
than because they were part of the contracts, governments' agreement to the World Bank
loan could be interpreted as a signal of commitment to the contract as well). Interviews
suggested that the influence of this outside involvement was mixed. It may have helped
the contracts' credibility in Ghana and Mexico, but it appears to have reduced the
contracts' credibility in Senegal. Survey respondents and other observers agreed that the
contracts in Senegal were viewed by the signatories as donor driven and not as credible
obligations. The suspension of disbursement of a World Bank loan to Senegal Electricity
after government reneged on its contract obligations to settle arrears to the company did
not change government behavior and penalized the company as much as, if not more
than, the government. The contract expired in 1989; by mid- 1995 the government and
Senegal Electricity management had yet to agree on a new contract. The role of outside
pressure from agencies such as the World Bank is necessarily a limited one, since such
agencies are not party to the contract and have, at best, only an indirect stake in the
outcome.
The participation of private individuals (non-government accountants, lawyers, and
academics) and a prestigious government think tank, the autonomous Korea Development
Institute, may have increased the credibility of government's commitment to the contracts
with Korea Electricity and Telecoms. Knowledgeable outsiders may have informally
monitored the government supervisors and acted as a check on government actions that
would have violated the contract. Moreover, they had a more direct stake in the outcome,
than, for example, the World Bank; such outsiders stated in interviews that they felt their



20
reputation was on the line in the highly publicized evaluation process in Korea (Shirley
1991). India also used outside evaluators, although their role was more limited then in
Korea where the outside team had primary responsibility for evaluation.
Costly Action. Many of the government actions promised in the contracts were
politically costly. For example, since SOE jobs are used to reward political supporters in
many countries (Shleifer and Vishny 1994), allowing managers to lay off redundant
workers would be a strong signal of commitment to the contract. It would also be
politically costly to divert revenues from other uses to pay government bills to the firm
and to fund government mandated investments, allow the firm to pursue its creditors and
apply legal penalties if they did not pay, and regulate monopoly prices according to clear
and reasonable rules that would allow the firm enough revenue to service its debt,
maintain and improve service, upgrade technology, and expand to meet demand (which
would mean price increases for powerful consumers). Costly signals were sent only in
the contracts with the two SOEs in Korea. When performance contracts were introduced
in Korea in 1983-84, managerial autonomy of the two SOEs in our sample (and of all
other so-called Government Invested Enterprises) was increased.'7 This seems to have
boosted the contracts' credibility with managers and employees. However, some of
those interviewed cited waning government commitment to performance contracting in
Korea in recent years, raising doubts about whether this credibility had been sustained.'9
Reneging. The model suggests that if governments renege by failing to pay the
promised incentives to managers who achieve their performance targets, the SOEs will
yield sub-optimal performance. In three of the five contracts where bonuses were part of



21
the agreement (the contracts with the three SOEs in Ghana) government reneged on its
promised incentive payments. Governments also reneged on the promises they made
from the beginning of the contract process in Ghana, India and Senegal (table 3). This
early reneging hurt the credibility of government commitment to the subsequent
contracts. For example, every year the contract for India Electricity promised
government help in enforcing prompt payment from its customers, the State Electricity
Boards, yet accounts receivable went from 149 days in 1986, the year before the contract
was introduced, to 207 in 1990; in 1991 they fell to 150 days when a power plant was
turned over to the SOE in partial payment of arrears. Similar reneging occurred in the
first contracts in Senegal. We already described Senegal Electricity's refusal to sign a
second contract. The government's failure to honor its obligations to Senegal Telecom
evoked this comment in a 1989 company report: "It must be remembered that the
contract plan was never considered a binding contract by the public powers."
Government credibility may have been greater in the contracts in Ghana, since the
government met some of its early promises. For example, even though the government
of Ghana did not always approve Ghana Water's tariff increases promptly and required
the company to serve some nonpaying customers, it did increase tariffs in real terms and
support management in reducing the work force. Furthermore, all three contracts in
Ghana provided for recalculation of targets if government reneged on its contractual
obligations. Although this betrays a disconcerting expectation that government will
renege, it also means that SOEs were not penalized for government misbehavior. As we
mentioned, the Mexican government met its obligations in the contract with the



22
Electricity Company, albeit with some delay.20 Finally, the Government of the
Philippines reneged on its implicit promises by, for example, not requiring the public
sector to promptly pay its bills to the SOEs.
In sum, due to weak enforcement, lack of costly signals and/or early reneging,
government commitment to the performance contracts in Ghana, India, the Philippines
and Senegal was not credible, while the Mexican contract may have been marginally
credible. Only Korea managed to formulate targets that were not directly dependent on
government action, institute an enforcement mechanism that used outside evaluators to
enhance credibility, and take costly action to signal commitment.
C. Lack of Management Commitment
The model suggests that if managers lack commitment and have bargaining power
they will try to avoid punishment when their performance falls below target. We looked
for evidence of managerial bargaining power by comparing the status of the SOE
negotiators with the status of the government supervisors responsible for negotiating,
monitoring and evaluating the contracts. We also analyzed the targets for evidence that
managers were manipulating the performance criteria.
Managerial Bargaining Power. In all but Korea and India (where outsiders played a
role, described above) contracts were negotiated, monitored and evaluated by middle- or
low-level civil servants, while the SOE's representatives were better paid and had much
higher status. For example, in Ghana the SOEs were often represented at the bargaining
table by a managing director with status equivalent to a cabinet minister (and better pay),
while the government supervisory agency was usually represented by a low-level



23
functionary.21 In contrast, Korea's contracts are evaluated by an ad hoc task force
composed of private individuals (nongovernment accountants, lawyers, academics, etc.)
and a prestigious government think tank, the autonomous Korean Development Institute.
However, the negotiations in both Korea and India were done by the government
bureaucrats. The lower status of the government supervisors made it hard for them to
negotiate tough targets and demand the information necessary for evaluating performance
or to give the SOEs an unsatisfactory score. Outsiders were more likely to rival the status
of the SOE representatives and be able to overcome some of the managers' bargaining
power. However, since the outsiders in both Korea and India were brought into the
process on an ad hoc basis their lack of continuity may have reduced their opportunity to
learn about the company's performance over time.
Asymmetry in the status of the government supervisory agencies vis-a-vis the
managers was aggravated in some of our sample contracts because supervisors were
weakened by frequent changes in responsibilities and authority. Specifically, the
supervisory agency in Senegal was moved twice after contracting began (from the
Presidency to the Prime Minister's office to the Ministry of Finance). According to
interviews, these moves diluted its authority vis-a-vis the enterprises. Similar changes
undermined the authority of the responsible agencies in the Philippines (where the
supervisory agency also has a high rate of staff turnover) and Ghana.22 In contrast, the
contracts in India, Mexico and Korea were negotiated, monitored and evaluated by more
stable agencies.



24
We expected that if managers are not committed they would manipulate the
targets to avoid punishment. Punishment was avoided by most of our sample SOEs. Only
in the case of the two Korean contracts was there any evidence that management
achievements had some effect on managerial careers (Shirley 1991); interviews in the
other countries revealed that managers' job assignments were largely politically
determined and not driven by performance.23 None of the sample received a less than
satisfactory grade against the contract criteria, so there was no basis for government to
discipline the managers or not pay the promised incentives. This suggests that an
important manifestation of low managerial commitment was bargaining for targets that
could be attained without increased investment or effort.
Targets. There is anecdotal evidence that some targets are easily attained:
*  Senegal Telecom's target for call completion rates was lower in the 1986 performance
contract than in the company's business plan of ten years before. In 1992 the call
completion rate was a mere 55 percent in Dakar and less than 40 percent for inter-
urban calls, about half the international industry standard.
*  In India, negotiations on targets sometimes dragged on so long that the targets were
set to be the same as actual performance.24
*  In Ghana contract targets were set by the companies themselves; the supervisory
agency considered many targets too low and established a penalty for under-targeting
(State Enterprise Commission 1991).
* In the Philippines, one of Philippines Electricity's most important and difficult
challenges was to improve its reliability, but its reliability indicator (the percentage of



25
electricity which its customers have contracted for but the company never supplied)
became less important in its performance score, falling from 30 percent of total
targets in 1990 to 10 percent in 1991 and 15 percent in 1992. This is not because the
firm became more reliable; on the contrary, by 1992-93 outages of seven hours per
day were common in many parts of the country. 25
Korea used a comprehensive target for assessing performance of the sample
enterprises, public profitability in constant prices. (This indicator follows a trend
very close to total factor productivity and can be calculated using shadow prices to
account for price distortions; as a target it would be had to manipulate.) 26 The share
of public profitability in the contracts for Korea Telecoms and Electricity fell from 20
percent to 12 percent from 1985 to 1992. Field interviews suggest that this occurred
because SOE managers successfully negotiated the addition of other targets at the
expense of public profitability.
Besides this anecdotal evidence there is comprehensive evidence that all of the
sample contracts had targets which put the government supervisors at a serious
disadvantage vis-a-vis the enterprise representatives. All but two of the sample contracts
had either many targets (seven had more than 20 targets on average) or targets which
changed frequently (four changed more than 20 percent of their targets a year on
average), or both (figure 1). For example, Korea Telecoms' contracts had about 40
targets, most with weights of 1 percent or less. On average, over one third of the targets
in the contracts with the three SOEs in Ghana changed every year. Each time a new
criterion is added, evaluators had to decide on the criterion value-what constitutes good,



26
fair, or poor performance. Recall that the targets in performance contracts for SOEs were
based on improvements over past performance, rather than some objective standard,
which required the government supervisors to learn about what constituted good
performance for each individual firm. Learning which values are hard and which soft is
more difficult when criteria are numerous or constantly changing.27
One way to make it easier to judge performance is to assign a weight to each target so
they add up to 100 percent and the supervisor can calculate an aggregate score. While
most of our sample used weighted targets (the exceptions are in Senegal and Mexico),
fluctuations in the weights frequently offset this advantage, adding another layer of
complexity and uncertainty. For example, the weight assigned to India Oil's financial
targets went from 20 percent in 1989 to 12 percent in 1990, then climbed to 40 percent by
1992.28
Such numerous, unstable, and increasingly less stringent targets indicate that
uncommitted managers were exploiting their bargaining power to negotiate targets that
maximized their information advantage. And indeed, field interviews suggested that
managers often negotiated for changes or additions to targets. However, another factor
may also be partly responsible. As mentioned, SOEs faced numerous objectives from
multiple principals; the government supervisors may have had trouble mediating among
the different vies and opted for numerous, relatively evenly weighted targets. Since the
political strength of the principals changed from time to time, this could account for some
of the fluctuations in the targets.29



27
In sum, the managers in our sample had bargaining power and used it to
manipulate the performance criteria to avoid penalty or to assure reward, as the model
suggested.
D. Contracting Problems: Summary
Our model suggests three ways in which performance contracts might fail; as figure 1
shows, all but the two contracts with the SOEs in Korea evidence all three kinds of
failure: low power incentives stemming from information asymmetries; lack of
government commitment; and lack of managerial commitment. Korea does better than
the rest of the sample on the first two, but fails on managerial commitment. The next
sections examines how the contracts did in improving SOE performance.
IV. The Impact of the Contracts on Performance
Our next task was to determine whether enterprise performance changed in ways
that could be attributed to the contract (or, more weakly, in ways that the contracts could
not be ruled out as explanatory factors). To assess the companies' economic performance
we compared trends in profitability (return on assets, or ROA), labor productivity (LP)
and total factor productivity (TFP) before and after the introduction of the contract.30
ROA was calculated as sales minus cost of goods sold and depreciation over revalued
fixed assets.3' TFP was calculated as the constant value of production over the constant
costs of all production factors (labor, intermediate inputs and capital), using company
volume data to construct company-specific price indices for each factor and output where
available; otherwise we used the relevant country-specific price indices. LP was
calculated as the real value of production divided by the number of workers.



28
Our counterfactual was based on the trend in the three measures before the
contract was introduced. We assumed that if the contracts made a difference we would
see a noticeable kink in trend in the period after the year in which the contract was
introduced. A noticeable kink was defined as a move from a negative trend to a positive
trend (or vice versa), or a more than doubling of the rate of change. We had to use
relatively stringent criteria because of the small sample size and the lack of a control
(such as an SOE without a contract).
A frequent question is why did we not assess how well the firms did on their non-
economic objectives, especially since some observers consider these more important for a
state enterprise than its economic objectives, and success in these goals may be
responsible for (and outweigh) failures in economic performance measures? Since 75
percent of the respondents to our questionnaire saw improving profitability or efficiency
as the number one objective of the contracts and 88 percent, as objective number one or
two, and since two thirds of the score of the enterprises were based on economic
achievements, we deemed improvement on ROA, LP and TFP as better measures of the
success of the contracts than non-economic achievements.32 Moreover, as we noted in
the introduction, an important rationale for performance contracts is to maximize
economic performance withjn the constraints of non-economic goals.33
Profitability. None of the twelve sample companies improved their trend in ROA
after the contract. Nine continued to improve or deteriorate without a noticeable change
in their trends, while three showed worse ROA trends after contracts than before
(Philippines Electricity, Senegal Electricity, and Senegal Telecoms; see figure 2). ROA



29
trends in state-owned enterprises may reflect government behavior more than managerial
behavior, since the sample's product prices were controlled by the government. (Some
important factor prices were also controlled; for example, government set prices for
Senegal Electricity's petroleum inputs at about twice world market prices in the 1980s
without always allowing it to raise its price to the consumer.)34 The profitability trends
in the sample suggest that the contracts had little impact on government's pricing
behavior. Only one company did better because of government mandated price increases
(Philippine Water), and these gains were not enough to change its almost flat trend in
ROA. As for the deteriorating firms, one (Senegal Telecom) shows worse ROA because
of government pricing decisions, mainly because the government reduced its very high
base tariffs.
Labor Productivity. Labor productivity growth was positive in all twelve cases, but
the data suggest that the contracts had little impact on this trend. Only two companies
show a kink in their labor productivity trends (declining labor productivity began to
increase in Senegal Telecoms, and the rate of increase in labor productivity more than
quadrupled after the contract in Ghana Water; see figure 3.); in the rest, labor productivity
continued its precontract trend upward. In only four of the twelve cases was labor
productivity growth primarily due to the shedding of excess workers (plus the removal
from the rolls of "ghost workers"). Consistent with this finding, 53 percent of
respondents to the questionnaire (N= 17) ranked inability to dismiss unnecessary workers
as the number one obstacle to achieving the contract's goals. In the other eight,
production gains outstripped employment increases. Since labor productivity could be



30
raised by increasing the use of other factors, rather than by improving overall efficiency,
we considered TFP a more reliable indicator of efficiency improvement.
Total Factor Productivity. Performance on TFP is mixed. In three of the twelve
companies (Ghana Water, Mexico Electricity, and Senegal Telecoms) declining TFP
began to increase after the contracts were introduced, while three others (India Electricity,
Philippines Water and Senegal Electricity; see figure 4) suffered a downturn in what had
been an improving TFP trend. The six remaining firms continued their precontract trends
upward. This ambiguous finding shows no strong relation between the contracts and TFP.
Exogenous Factors. Our performance measures suggest that the contracts had little
positive impact, as figure 5 shows. To assess whether this could be attributed to other
factors which might have blunted or negated the influence of the contracts, we first
checked for the presence of any exogenous factors that could have affected the
performance trends. Specifically we looked for important changes in input or output
markets or prices, natural or man-made disasters, major work stoppages or any other
major events reported in the companies' annual reports or cited by country and company
experts. We also compared the companies' trends with trends in other variables (notably
with GDP growth) to see if an acceleration or deceleration in growth might have affected
performance.35
In two cases we found exogenous factors which might have influenced trends --
Philippines Electricity and Senegal Telecoms. The postcontract return on assets of
Philippines Electricity was adversely affected by increases in the cost of imported oil, but
this cannot explain its poor performance in constant priced TFP. TFP might have been



31
adversely affected by disruptions caused by the Luzon earthquake, plus devastating
typhoons in 1989 and 1990; however, observers knowledgeable about the company
regard its performance as consistently weak. Moreover, other disasters (the Mt. Pinatubo
eruption and Ormoc flood in 1991) had not prevented the firm from recovering its
productivity levels.
Senegal Telecom may have benefitted from the fact that it was split off from the
postal service in 1986, the first year of the contract. Although we have tried to isolate the
telecommunications side prior to this split, that may not always have been possible, and
could cause the period before the contract to look worse than it otherwise would. (The
split itself is not strictly exogenous; even though it was not explicitly part of the contract,
it was part of the agreement with the World Bank to implement the two together and a
case could be made that they were interdependent.) In any event, if the Senegal Telecom
improvement is indeed overstated, it would strengthen our finding that the contracts had
no discernable positive impact on performance.
V. Conclusions
Only three of the twelve case study companies showed a turnaround in TFP after
contracts were introduced, six continued their past trends, and three performed
substantially worse under contracts than before. Labor productivity improved at a faster
pace in four cases, and deteriorated in none, but for most SOEs the improvement long
predates the contracts. ROA deteriorated in three firms; the rest showed little change.



32
Thus, the evidence from this sample gives little support to the premise that explicit
contracts helped improve SOE performance.
If we compare the contracting problems in figure 1 with contract performance we also
see no clear patterns. The contracts with the least problems (Korea Electricity and
Telecom) were with companies which showed no change in performance; the two SOEs
continued their pre contract (improving) trends. Since these contracts had problems of
managerial commitment, this may be a necessary condition for performance contracts to
have a positive impact on performance.
As we mentioned, performance contracting assumes that government's objectives can
be maximized, and performance improved, by setting targets which take into account the
constraints placed on managers. For this to occur, the principals will need to be willing
explicitly to state their objectives, assign priorities to multiple objectives (by assigning
weights), translate these objectives into targets aimed at performance improvement,
provide incentives to meet these targets (or monitor the agents without incurring
substantial costs), and credibly signal their commitment to the contract.
All these assumptions failed to materialize in our sample, First, with the exception of
Korea, the governments negotiating our sample contracts chose to hide their true
objective functions and pledge actions that they were not motivated to implement.
Second, with the exception of India, Korea and Mexico, the supervisory agencies were
given neither the capacity nor the incentives to negotiate, monitor and evaluate well.
Their information disadvantage was reinforced by other government actions, such as
giving-them low pay and status, frequently moving them within the government and



33
failing to force the SOEs to comply with their information requirements. Most of the
contracts fell victim to information asymmetries which led governments to set incentive
mechanisms too low to persuade managers to comply.
Why would governments introduce the contract process and then not try to make it
work? Some governments may have been motivated to make promises that were
politically unrealistic because the contracts allowed them to earn foreign assistance. Such
assistance might not have been forthcoming if they could only promise the marginal
improvements possible without changes in government behavior. This may have been
the case in the contracts in Senegal and Ghana where the negotiating agents pledged
actions (regular increases in tariffs, end to arrears) that were not politically possible, and
in the Philippines, where target achievement depended on government actions that were
politically unlikely. That they did this knowingly is suggested by the fact that the
problems they pledged to correct were long standing and had proved immutable.
Less hard to interpret is the lack of commitment of the managers. In contrast to a
private firm, the SOE managers in our sample had little choice but to sign the contract
(the case of Senegal Electricity notwithstanding). Not surprisingly, managers with
information advantages and bargaining power, and without high powered incentives or
credible commitment from government, chose to use their advantages to manipulate the
targets so as to assure that their performance would be judged as satisfactory. In essence,
the sample performance contracts failed to reduce the problems caused by the absence of
a clear residual claimant with a direct financial stake in the SOE's surplus.



34
This analysis has policy implications for developing country governments considering
performance contracts. Such contracts will only succeed if the contracting problems
described here can be overcome. This suggests that contracting governments will need
to:
* Reduce information asymmetries by improving accounting practices and the skills of
negotiators and evaluators. Where market failures don't rule out competition,
information could be improved by breaking up monopolies; where products are
homogeneous or easily compared, and technology is relatively simple and stable,
information could be enhanced by allowing managers to bid for the contract.36
*  Give the bureaucrats and other actors involved the contract discussions independence,
adequate pay, stability and the power to punish intentional misinformation or
withholding of information.
*  Clearly specify and enforce high powered rewards and penalties when the government
has adequate information.
*  Take politically costly actions at the outset and throughout the process to signal to
managers that they will have enough freedom to achieve their targets, such as
allowing layoffs of redundant workers or by regulating prices to permit an efficient
firm to earn a reasonable return.
*  Set up and abide by a neutral, third party enforcement arrangement (such as an
arbitration board).
*  Design a contract with few (two or three) comprehensive, weighted targets, which are
stable over a period of time.



35
Design self-enforcing contracts, by, for example, making managers post a bond that
they will forfeit if they fail to meet the agreed targets (as was done in China, for
example).
Whether a performance contract will succeed if all of these measures are implemented is
an open question. Our sample is small, and it is not representative of the contracts used in
transitional economies for competitive SOEs; we lacked controls (SOEs without
contracts), and we used relatively stringent criteria for assessing performance change. 37
Since all the contracts failed to improve performance, we cannot say with certainty
whether or not a well designed contract would have done better. Also unknown is
whether alternatives which create a residual claimant, such as contracts with private
managers or sale of the enterprise to a private owners, might produce better results
(although findings in World Bank 1995 are very suggestive). What is clear is that, if the
contracts in our sample are representative of the performance contracts in use with natural
monopolies worldwide, then considerable time and effort is being expended on an
exercise with neither theoretical nor empirical justification.



36
Appendix A: The Derivation of Equation 11
- =v{p0e'-(1+X)bQe }+(1-v){p0e'-b0e'-XbOe -X(0 e-Oe)-Xb(6e'-Qe')
= v(1 +X)(s-b)Oe +(I -v){pOe' -(I +X)bOe'+ Xb6e' -XbOe -kb(Oe' -f3e )-x(/  e-0e)}
=v(l+X)Oe +(l-v)(l+X){(s-b)Oe'-X(1-v)(0 e-0e)}
=(1+X){(s-b)[vfe +(1 -v)e']-    A(1-v)( e-0e)}
Setting this to zero yields equation (11). Q.E.D.
Appendix B: The Proof of the Two Inequalities
The Proofs for the two inequalities are mirror images, so we shall only prove the last one.
R-a      -      R-a               -
Denote 0 =    -       , 0 =            ; note that 0 > 0 . Then take derivative with
-b(e1 + K(I)) b(el + K(I))
respect to I (for equation (24)), and we get
MR'   MR(p   R - a   -K'(I)                                        R -a   -K'(I)
fb   '(I) p(0)dO +Rp (0)  bB(e1K(I))2
A!J)Ps y)UIcJ      K'(I)p(0)dO -       ~b (e + K(I)) -         b  (e_  + K(
= fb3               -tb0K{ p(0)d0 -bK'(I)p(0)d--Rp(0)0 e K()  + 0K(  l K(I)
= bK'(I) I [gf)9(0)dO _0 (0)]                           b(e,v(0)  -  b(  + K(I)) {fQ(p) 09 (
We now show that the terms in both {. } are negative, thus the above equation is negative. The
term in the first {.} is negative because fOEp(0)dO - x2T (x) is monotonically decreasing in x:
a [  p(HO)do_-x2p(x)] = -X(p(x) +xip'(x)) <O
the right hand side is non - positive because by assumption, x > 0, and Ow(O) increases in 0, thus,
p(0) +OTp'(0) > 0.
The second {.} term is negative because of the same assumption: Osp (0) increases in 0, and
0 >0.  Q. E. D.



37
References
Barron, David. 1988. "Design of Regulatory Mechanisms and Institutions," in R.
Schmalensee and R. Willing (eds). Handbook of Industrial Organization.
Amsterdam: North-Holland.
Besanko, David and David Sappington. 1987. Designing Regulatory Policy with Limited
Information. London: Harwood Academic Publishers.
Dyer Ciss6, Nichola. 1994. "The Impact of Performance Contracts on Public Enterprise
Performance." Background paper. Policy Research Department, World Bank,
Washington, D.C.
Freixas, Xavier, Roger Guesnerie, and Jean Tirole. 1985. "Planning under Incomplete
Information and the Ratchet Effect." Review of Economic Studies. LII: 173-191.
Jones, Leroy. 1975. Public Enterprise and Economic Development: The Korean Case.
Seoul: Korea Development Institute.
. 1981. "Towards a Performance Evaluation Methodology for Public
Enterprises: With Special Reference to Pakistan." Paper presented at the
international symposium on Economic Performance of Public Enterprises,
Islamabad.
. 1985. "Public Enterprise for Whom? Perverse Distributional
Consequences of Public Operational Decisions." Economic Development and
Cultural Changes 33 (2): 333-47.
Laffont, Jean-Jacques, and Jean Tirole. 1986. "Using Cost Observation to Regulate
Firms." Journal of Political Economy. 94, Part 1:614-41.
. 1993. A Theory of Incentives in Procurement and Regulation. MIT
Press: Cambridge, MA
Levy, Brian, and Pablo Spiller, eds. 1996. Regulations, Institutions, and Commitment:
Comparative Studies of Telecommunications. New York: Cambridge University
Press.
Lupia, Arthur and Mathew D. McCubbins. 1996. The Triumph of Reason: Knowledge
and the Foundation of Democracy. New York: Cambridge University Press.
Nalebuff, Barry and Joseph Stiglitz. 1983. "Information, Competition, and Markets."
American Economic Review. 73(2):278-283.



38
Nellis, John 1989. "Contract Plans and Public Enterprise Performnance." Policy, Planning,
and Research Working Paper 118. World Bank, Washington, D.C.
North, Douglass C. 1990. Institutions, Institutional Change, and Economic Performance.
New York: Cambridge University Press.
, and Barry R. Weingast. December 1989. "Constitutions and
Commitment: The Evolution of Institutions Governing Public Choice in
Seventeenth-Century England." The Journal of Economic History. Vol. XLIX.
No. 4.
Sappington, David E. M. 1991. "Incentives in Principal-Agent Relationships." Journal of
Economic Perspectives 5 (2): 45-66.
, and Joseph E. Stiglitz. 1987. "Privatization, Inforrnation, and Incentives."
National Bureau of Economic Research (NBER) Working Paper 2196. Harvard
University, Cambridge Mass.
Shirley, Mary. 1989. "Evaluating the Performance of Public Enterprises in Pakistan."
Policy, Planning, and Research Working Paper 160. World Bank, Washington,
D.C.
. 1991. "Improving Public Enterprise Performance: Lessons from South
Korea." Annales de l'economie publique sociale et coop*ative 62 (1). De Boech
Universite, Brussels.
Shleifer, Andrei and Robert W. Vishny. 1994. "Politicians and Firmns." Quarterly Journal
of Economics. 109:995-1024.
Spence, A. Michael. 1973. "Job Market Signalling." Quarterly Journal of Economics.
353-374.
State Enterprises Commission (SEC). 1991. SOE Performance Evaluation Report. Accra,
Ghana.
. 1993. The SOE Reform Program 1984/1992: Review and
Recommendations. Accra, Ghana.
Stiglitz, Joseph E. 1989. The Economic Role of the State. Basil Blackwell, Inc.
Carnbridge, MA.
Trivedi, Prajapati, ed. 1990. Labor Redundancy in the Transport Sector: The Case of
Chile. Washington, D.C.: World Bank.



39
Vickers, John and George Yarrow. 1988. Privatization: An Economic Analysis.
Cambridge, MA.:The MIT Press.
Weitzman, Martin L. "The Ratchet Principle and Performance Incentives." The Bell
Journal of Economics. 11 (1):302-08.
Williamson, Oliver. 1975. Markets and Hierarchies. New York: The Free Press.
. 1976. "Franchise Bidding for Natural Monopolies - in General and with
Respect to CATV." Bell Journal of Economics 7 (1): 73-104.
. 1985. The Economic Institutions of Capitalism: Firms, Markets,
Relational Contracting. New York: Free Press.
World Bank. 1995. Bureaurcrats in Business. The Economics and Politics of Government
Ownership. Washington, D.C.: Oxford University Press.



40
State owned enterprises were defined as government owned or government controlled economic entities
that generate the bulk of their revenues from selling goods and services. (See Jones 1975). This definition
includes all government owned enterprises where government controls the appointment of management
either through its ownership stake or, in cases where government owns a minority of the shares, the
distribution of the remaining shares gives government control. It excludes state owned activities such as
education or health services, that are financed in other ways, such as from government's general revenues.
(World Bank, 1995, Box 1. 1)
2 See also Milgrom and Roberts 1992.
3
. Also see Laffont and Tirole (1993)
. Since y could be any other performance criterion, such as log(output), the linear assumption is
not as restrict as it looks.
. If interested, see appendix A for the derivation.
6
, See Laffont and Tirole (1993).
. In another subtle way, the government can show its non-commitment by taking advantage of
new performance data to ratchet up its performance criteria (Weitzman, 1980). Consider a performance
criterion set at period t; if a firm performs better this period, the government raises the target for the next
period. The manager anticipates the negative externality of current efforts on the future stream of rewards,
and reduces current efforts relative to the level of effort that prevailed when the govermment was commited
to repeating the optimal static performance contract. In this example when the government refuses to abide
by the optimal static contract for each period, and instead takes advantage of new information, the firm
refrains from working at the socially optimal level.
8
. A sufficient condition for this assumption to hold is that 0 belongs to the range where
(p '(0 ) 2 0, satisfied for the uniform distribution, and for the lognormal distribution when ln(O ) is below
its mean. Lognormal distribution is (and normal distribution is not) a relevant distribution in our example
because we constraint 0 to be positive, as required by a lognormal distribution. More generally, since
most commonly-used density functions usually do not begin to decrease until the random variable hits its
median or mean, the assumption is most likely to be satisfied when 0  is below the mean or the median of
the random variable.
9. See Appendix B for details of the proof.
10 There were 17 respondents to the questionnaires, one from each of the sample companies and monitoring
agencies, with the exception of the Mexican Government which did not respond. Field interviews were
done in Ghana and Senegal by Dyer Cisse (whose interview reports were used for this article), and in India
and Korea (during several earlier visits from 1990 to 1994) by one of the authors. Staff from the
Philippines government responsible for the contracts were interviewed in Washington, D.C. One of the
authors was also involved in the initial discussions of the Mexico contracts.
I We included one electricity company from each country to test for industry effects (none were
found).
12 Korea Telecoms was created as a separate company only two years before its contract started.
Senegal Telecoms was created in the same year that it signed its first contract; we extrapolated
backwards two years, using data from prereorganization telecom units. We included Senegal
Electricity's second contract (starting in 1990), even though it was never signed. The conclusions
would not change if that period were omitted. We also took 1987 as the start date for India Oil's
contracts even thought it signed a pilot form of contract in 1986 (based on the multi-year French
contract plan). The government later switched to a very different sort of contract (closer to the
Korean model, using yearly weighted targets). The conclusions would not change if the earlier
contract period were included.
13 Performance improvement should not be confused with good performance in some absolute sense.
For example, Ghana Water and Sewerage increased TFP by an average of 12.5 percent a year during



41
the three years of contracts for which we have data (1990, 1991 and 1992). Yet this is still a company
with almost 50 percent of its water unaccounted for and only about two-thirds of the population
served with water.
14 Independent auditors for Senegal Electricity concluded that "the accounting and financial management
show such an amount of anomalies and deficiencies that the auditor will not be able to release an opinion
on the matter." World Bank, "Senegal: Note on The Institutional Study for the Power Sector", Memo,
1992.
'5 A study of Pakistan (Shirley 1989) found that poor inforrnation also led to low powered incentives there.
16 Other obstacles were unrealistic objectives, poor management, factors outside the management's and
government control (such as world markets), failure of international agencies to release funds and public
attitudes towards the company. Number of respondents was 17.
17Standing boards of directors were replaced by executive boards; government representation on the
boards was reduced to two members; only board approval of the budget was required when previously
the supervising ministry, the Economic Planning Board and the cabinet also had to approve;
responsibility for most personnel decisions was shifted from the supervising ministry to the enterprise;
procurement through a centralized office was made voluntary instead of compulsory; all oversight
was centered in the contract with one yearly inspection, compared to the extensive system of controls
and inspections used before (in one year before the contracts Korea Electricity had eight different
inspections lasting 108 days); preference was shifted to internal candidates for senior positions
(previously over half of all such appointments were from outside the firm) and an explicit merit
assessment was introduced (Shirley 1991, 11-12).
18 An early opinion survey of 750 employees in all ranks of the SOEs under contract found that 93
percent thought that management had improved, thanks to the performance evaluation system; 55
percent saw substantial improvement; 94 percent of the executive directors said that there had been
substantial or significant improvement (Song 1988).
9 A sign of this waning commitment may be the move away from public profitability, as mentioned
earlier.
20 Mexico Electricity's experience under the contract differs from that of other SOEs in Mexico
in that the government followed through on its commitments under the contract. The Mexican
government distinguishes between the Financial Restructuring Pact applied to Mexico's Electricity
and performance contracts. A World Bank assessment of Mexico's standard performance contracts
with five other SOEs concluded that the results were disappointing. The study attributed the
poor outcomes to a lack of adequate consultation between government and the SOEs;
government attempts to treat the contracts as one more instrument of control among many; and
targets which lack coherence, simultaneously overlapping and leaving gaps (World Bank 1992).
21 Interview with Hafeez Shaikh (1994).
22 In Ghana, government negotiators and monitors in the State Enterprises Commission (SEC) lost
power when responsibility for energy contracts was shifted to the sector agencies. Although the shift
only affected two SOEs directly (including one of the three in our sample), staff in the SEC believe
that the change adversely affected the credibility of all monitoring and negotiating. According to an
SEC report "This situation tends to create considerable uncertainty about who can require
performance information from the SOEs. This has affected the ability of the Commission to perform
its reporting and evaluation function." (SEC 1991).
23 In Korea, evaluations of how different departments, divisions and offices contributed to contract
achievements are taken into account in internal promotions (Shirley 1991).
24 Field interviews.



42
25 In addition Philippines Electricity's targets for new lines, main lines, income, rate of regurn on
revalued assets and power generation per employee were reduced even though the company had
exceeded the target the previous year. Delays in getting information on achievements may be a reason
why government agents agreed to set targets well below the previous year's achievements.
26 Public profits are private profits adjusted to measure returns to society as a whole and to exclude
factors beyond the control of managers (Jones 1991).
27 See Jones (1993) for more on the concepts of criterion and criterion values and an excellent review
of the issues of performance contracting.
28 Interviews with government evaluators and knowledgeable observers also suggest that the managers
in our sample negotiated for more and softer targets and were often responsible for changes in the
criteria.
29 Some observers have suggested in discussions of this paper that the weaknesses in the targets are due to a
lack of capacity on the part of the government agencies or are part of the early phase of a learning process
about how to contract successfully. These explanation do not seem credible to us because numerous or
fluctuating targets are found in the entire sample, in countries where govemnment capacity is strong (India,
Korea or Mexico) and in countries where the contract process has been in place for more than five years
(Korea and Senegal).
30 The data were taken from enterprise audited accounts and put into comparable formats where
necessary. Although every effort was made to verify the accuracy of the data, in some cases the
underlying information systems are weak. However, it seems plausible that errors are not correlated
over time and do not greatly affect the trend analysis.
31 Where assets were not revalued by the company, we revalued them, using company figures for
depreciation and the GDP deflator for inflation.
32 Out of a total of 17 respondents to this question.
33 Some commentators have asked us why we did not judge each firm's performance by its attainment of
the economic targets specified in its contract. As mentioned, we did investigate performance against
targets and found that all of the sample SOEs achieved at least satisfactory ratings where a score was
assigned (Senegal had no such score). However, since many of the targets (as we analyzed above) were
subject to manipulation, firms might attain their economic targets under the contract but not necessarily
operate more profitably or productively. (For example, 30 percent of India Electricity's score in 1991-92
depended on the volume of electricity they generated; output went up but the use of material inputs rose
even faster. India Electricity achieved its target and received a score of excellent that year, but its TFP fell
below precontract levels.
34 Its prices are nevertheless about one-third less than those charged to the private sector, but private
producers usually have more freedom to change producer prices.
35 We also compared performance trends with trends in inflation and interest rates.
36 This is not likely to work in every potentially competitive sector. Financial enterprises, for example,
may still have strong information advantages despite competition because of implicit or explicit safety nets
(such as deposit insurance) and the adverse effects of competition on managers' incentives to act prudently.
37 The authors are currently studying the experience with performance contracts for competitive and
monopoly SOEs in China. This study will correct some of these problems, since it will be a much larger
sample with controls, and we have both productivity measure and better descriptions of provisions of
contracts. Econometric analysis will be used.



Table 1. Nunber of Performance Contracts in Developing Countries, by Sector
Latin     Mid-East and    Central
Sector                   Africa    Asia     America    North Africa    Ewuopea       Total
Transport                  26        8          4             6             2         46
Telecom and Post           15        2          1             1             0          19
Extractive industries      6        26          2             2             3         39
Agriculture                13        3          2             0             0          18
Water                      4         4          0             1             0          9
Electricity                11        8          6             1             1         27
Other                      61       160         1             1             4        227
Total                    136      211        16             12           10        385 a
a. Data for Romania only. Contracts are also being used in Bulgaria, but no details are available.
b. Total figures cover thirty-one countries. In addition, Indonesia reports 180 firms, and China 103,000;
no breakdown by industry was available. Data reflects situation as of June 1994. Based on a world-wide
search using World Bank and other sources. In some countries additional contracts may have been awarded.
Source: Survey of World Bank reports and staff.
Table 2. Case Study Enterprises
First contract
Country (contract type)              Enterprise name: (NAME USED IN TEXT)      Contract duration     vear
Ghana (performance contract)    Electricity Corporation of Ghana (ECG):                              1989
GHANA ELECTRICITY                                   Yearly
Ghana Water and Sewerage Corporation (GWSC):                         1989
GHANA WATER
Ghana Posts and Telecommunications (GP&T):                          1990
GHANA TELECOM
India (memorandum of            National Thermal Power Corporation (NTPC):                           1987
understanding)                  INDIA ELECTRICITY                                   Yearly
Oil and Natural Gas Commission (ONGC):           (published)
INDIA OIL                                                             _
Korea (performance evaluation   Korea Electric Power Corporation (KEPCO):                            1984
and measurement system)         KOREA ELECTRICITY                                List of yearly
Korea Telecommunications Authority (KTA):          targets
KOREA TELECOMS
Mexico (convenio de             Comision Federal de Electricidad (CFE):            3 years           1986
rehabilitacion financiera)      MEXICO ELECTRICITY
Philippines (performance moni-  Metropolitan Water and Sewerage System                               1989
toring and evaluation system)    (MWSS): PHILIPPINES WATER                       List of yearly
National Power Corporation (NPC):                   targets
PHILIPPINES ELECTRICITY
Senegal (contrat plan)          Societe Nationale d' Electricite (SENELEC):                          1987
SENEGAL ELECTRICITY                                3 years
Societe Nationale des Telecommunications du                         1986
LSenegal (SONATEL): SENEGAL TELECOMS   I
Source: World Bank data.



Figure 1 Summary of Sample Contracting Problems
Information and Incentives                     Govemment commitment                                   Manageral commitment
No                                  Fluctuating
Unskilled,    Unskilled,    Low        Lack cf                    No         Managerial   performance   Examples   Numerous          targets
Poor       unstable      unstable    power of      costly                  3rd party     bargaining    impact on      of soft      targets     (>18% per
accounftng    negotiators   evaluators   incentives    signals    Reneging   enforcement     power         career       targets       (>20)         year)
Ghana  Elericity                                   *            *                                    c                                                      0             0
Water   *c                                                                 C             c *                         0                          0 *
Telecoms                    0             0    _                   C           c            c             0              0            0            0            0
India   Elericity     0                          0            0                                    C)                           0                          0
Oil  0                             0 C                       0    0                   0  C                        0             0    0                    0
Korea Elecdidty   0                                0            0           0           0             C             C                           0            0            0
Telecoms                                 0             0           0           0            0             C              C            0                         0
Mexico Elericity    0                 0                         C 0         0                         C             C                                       00
Philippines Elericity    0                *0                        0            0           0            0 0           0             0                          0            0
Water                          0 *     *                                   0             0*0                         0                         0             0
Senegal   Eledicity       0                                       * 0                     * 0
Telecoms   0                0             *            *           *           0            *             0              0            0            0            0
0 = Major problem
C = Somewhat of a problem
0  = Utte or no problem



Figure 2. Pre- and Postcontract Performance: Net Rate of Return on Revalued Assets
ercent                        SOEs showing deteriorating trends
40
30  _  . _ ~  "- , - -                 Senegal Telecom
30 _,,-                                                                                  _
20
10
.      .     .........................................................  Senegal Electricity
/Philppines Electricity
0
I  I      I      I       I                I      1       i             i          I     I
-5    -4    -3    -2    -1 Base year +1          +2    +3      +4      +5     +6     +7
Source: Company data and World Bank estimates.
Figure 3. Pre- and Postcontract Performance: Labor
(indexed to base year)
3               S OEs showin  turnaround or acceleratiNg_provement__
2 -                                          Gh    W                      Seea
0                    I                I         I       III
-3      -2       -1   Base year  +1        +2       +3       +4       +5       +6      +7



Figure 4. Pre- and Postcontract Performance: Total Factor
(indexed to base year)
SOEs showing iturnaround or accelerating improvement
4-
3
/Mexico
Electricity
2-
l   L  <  ,VGhana Water  Senegal Telecom
0
-2      -I    Base year  +1         +2       +3       +4       +5       +6       +7
SOEs showing deterioration
1.5,
1.0                        /
Senegal
\         /  ~~~~Electnc,                                                 .
0.5-
Philippines Water
India Electricity
0  l       l       l
-5     -4       -3      -2      -I   Base year  +1      +2       +3      +4      +5
Source: Company data and World Bank estimates.



Figure 5. Performnance Changes after the Introduction of Performance Contract
Number of state-owned enterprises
12-
Improved
Improved
8       No
No~~~N
chm                     . ff 
4 -
Return on assets          Labor           Total factor
productivity       productivity
Source: Comapany data and World Bank estimates.



Figure 6 Summary of Contract Outcomes
Changes in Trend in:
ROA          LP          TFP
Ghana   Electrcity       0            0            0
Water       0            +            +
Telecoms       0            0            0
India   Electricity    0            0
Oil      0            0            0
Korea   Electricity      0            0            0
Telecoms       0            0            0
Mexico   Electricity      0            0           +
Philippines   Electrcity                   0           0
Water       0            0
Senegal   Electuicity      -            0
Telecoms                    +            +
0 = No change in trend
+ = Trend improved
- = Trend deteriorated



Policy Research Working Paper Series
Contact
Title                            Author                  Date              for paper
WPS1746 The Role of Long-Term Finance:    Gerard Caprio, Jr.      April 1997         P. Sintim-Aboagye
Theory and Evidence              Asli Demirgu,c-Kunt                        38526
WPS1747 Protection and Trade in Services:    Bernard Hoekman      April 1997         J. Ngaine
A Survey                         Carlos A. Primo Braga                      37947
WPS1748 Has Agricultural Trade Liberalization  Merlinda D. Ingco  April 1997         J. Ngaine
Improved Welfare in the Least-Developed                                     37947
Countries? Yes
WPS1 749 Applying Economic Analysis to    Gary McMahon            April 1997         C. Bernardo
Technical Assistance Projects                                               37699
WPS1750 Regional Integration and Foreign  Magnus Bl6mstrom        April 1997         J. Ngaine
Direct Investment: A Conceptual  Ari Kokko                                  37947
Framework and Three Cases
WPS1751 Using Tariff Indices to Evaluate  Eric Bond               April 1997         J. Ngaine
Preferential Trading Arrangements:                                          37947
An Application to Chile
WPS1752 Ghana's Labor Market (1987-92)    Sudharshan Canagarajah  April 1997         B. Casely-Hayford
Saji Thomas                                34672
WPS1753 Can Capital Markets Create        Paul Lanoie             April 1997         R. Yazigi
Incentives for Pollution Control?  Benoit Laplante                          37176
WPS1754 Research on Land Markets in South  Rashid Faruqee         April 1997         C. Anbiah
Asia: What Have We Learned?      Kevin Carey                                81275
WPS1 755 Survey Responses from Women      Mari Pangestu           April 1997         J. Israel
Workers in Indonesia's Textile,  Medelina K. Hendytio                       85117
Garment, and Footwear Industries
WPS1756 World Crude Oil Resources:        G. C. Watkins           April 1997         J. Jacobson
Evidence from Estimating Supply    Shane Streifel                           33710
Functions for 41 Countries
WPS1757 Using Economic Policy to Improve   Rashid Faruqee         April 1997         C. Anbiah
Environmental protection in Pakistan                                        81275
WPS1758 The Restructuring of Large Firms    Simeon Djankov        April 1997         F. Hatab
in Slovakia                      Gernard Pohl                               35835
WPS1759 Institutional Obstacles to Doing  Aymo Brunetti           April 1997         M. Geller
Business: Region-by-Region Results Gregory Kisunko                          31393
from a Worldwide Survey of the   Beatrice Weder
Private Sector



Policy Research Working Paper Series
Contact
Title                            Author                   Date               for paper
WPS1760 Credibiiity of Rules and Economic    Aymo Brunetti          April 1997          M. Geller
Growth: Evidence from a Worldwide  Gregory Kisunko                            31393
Survey of the Private Sector      Beatrice Weder
WPS1761 Bending the Rules; Discretionary   Susmita Dasgupta         May 1997            E. de Castro
Pollution Control in China       Mainul Huq                                   89121
David Wheeler
WPS1 762 Strategies for Pricing Publicly    Paul J. Gertler         May 1997            C. Bernardo
Provided Health Services         Jeffrey S. Hammer                            31148
WPS1763 China and the Multilateral investment Yushu Feng            May 1997            E. Beers
Guarantee Agency                                                              36175
WPS1764 A Reversal of Fortune for Korean   Yaba vab der Neykeb      May 1997            D. Ballantyne
Women: Explaining 1983's Upward    Rodgers                                    87198
Turn in Relative Earnings
WPS1765 Explaining Agricultural and Agrarian  Hans P. Binswanger    May 1997            D. Housden
Policies in Developing Countries  Klaus Deininger                             36637
WPS1766 New Systems for Old Age Security:  Estelle James            May 1997            S. Khan
Theory, Practice, and Empirical                                               33651
Evidence
WPS1767 Pension Reform: Is There a Tradeoff Estelle James           May 1997           S. Khan
between Efficiency and Equity?                                                33651
WPS1768 Is There a Quantity-Quality Tradeoff  P. Durassamy          May 1997            S. Khan
as Enrollments Increase? Evidence   Estelle James                             33651
from Tamil Nadu, India           Julia Lane
Jee-Peng Tan
WPS1769 Information, Incentives, and       Mary M. Shirley          May 1997           P. Sintim-Aboagye
Commitment: An Empirical Analysis  L. Colin Xu                                38526
of Contracts between Government
and State Enterprises
WPS1770 Poverty and Social Transfers       Christiaan Grootaert     May 1997            G. Ochieng
in Hungary                                                                    31123