POLICY RESEARCH WORKING PAPER 3 095 The Emerging Project Bond Market Covenant Provisions and Credit Spreads Mansoor Dailami Robert Hauswald The World Bank Development Prospects Group July 2003 | POLICY RESEARCH WORKING PAPER 3095 Abstract The emergence in the 1990s of a nascent project bond Much of the recent work relating to the role of market to fund long-term infrastructure projects in contractual covenants to the determination of bond developing countries merits attention. Dailami and prices has focused on the U.S. corporate bond market Hauswald compile .tailed information on a sample of with its unique bankruptcy code (Chapter 11) and well 105 bonds issued between January 1993 and March developed legal framework, recognizing the bond 2002 for financing infr --'lcrr.. projects in developing contract as the sole instrument of defining the rights and countries, document their contractual covenants, and duties of various parties. In circumstances in which the analyze their pricing determinants. They find that on underpinning legal and institutional frameworks average, project bonds are issued at approximately 300 governing contract formation and enforcement are not basis points above U.S. Treasury securities, have a well developed, the link between bond pricing and legal surprisingly high issue size of US$278 million, a maturity framework becomes important. This finding is confirmed of slightly under 12 years, and are rated slightly below by the authors' econometric analysis of project bond investment grade. In terms of geographic origin, projects pricing model. So, investors take into account the quality in Asia and Latin America have issued more bonds than of the host country's legal framework and reward those located in other regions. projects located in countries that adhere to the rule of law with tighter credit spreads and lower funding costs. This paper-a product of the Development Prospects Group-is part of a larger effort in the group to promote a healthy flow of investment capital to developing countries' infrastructure. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Sarah Crow, room MC2-358, telephone 202-473-0763, fax 202-522-2578, email address scrow@worldbank.org. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at mdailami@worldbank.org or hauswald@american.edu. July 2003. (35 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by Partnerships, Capacity Building, and Outreach World Bank Policy Research Working Paper The Emerging Project Bond Market: Covenant Provisions and Credit Spreads Mansoor Dailami Development Prospects Group World Bank Washington, DC 20433 and Robert Hauswald Kogod School of Business American University Washington, DC 20016-8044 Abstract The emergence in the 1990s of a nascent project bond market to fund long-term infrastructure projects in developing countries merits attention. This paper complies detail information on a sample of 105 bonds issued between January 1993 and March 2002 forfinancing infrastructure projects in developing, documents their contractual covenants ,and analyses their pricing de- terminants. It is found that on average, project bonds are issued at approximately 300 basis points above US Treasury securities, have a surprisingly high issue size of US$278 million, a ma- turity of slightly under 12 years and are rated slightly below investment grade .In terms of geo- graphic origin, projects in Latin America and Asia have issued more bonds than those located in other regions. Much of the recent work relating the role of contractual covenants to the determination of bond prices has focused on the US corporate bond market with unique bankruptcy code ( Chap- ter 11) and well developed legalframework, recognizing the bond contract as the sole instru- ment of defining the rights and duties of various parties. In circumstances in which the underpin- ning legal and institutional frameworks governing contract formation and enforcement are not well developed, the link between bond pricing and legalframework becomes important, a finding confirmed by our econometric analysis ofproject bond pricing model. Hence, investors take into account the quality of host country legal framework and reward projects located in host coun- tries that adhere to the rule of law with tighter credit spreads and lowerfunding costs I Emerging Project Bond Market: Covenant Provisions and Credit Spreads I. Introduction The emergence in the 1990s of a nascent project bond market to fund long-term infrastructure pro- jects in developing countries, such as electric power plants, roads, ports, airports, telecommunica- tions networks, and water and waste water facilities, merits attention for several reasons. First, they highlight the attractiveness of such investment opportunities that are traditionally the pre- serve of the public sector for private sources of capital. Second, project bonds are potentially a major source of long-term private debt capital linked directly to economic growth and competi- tiveness. Third, they are a new asset class in the emerging market debt spectrum, offering asset diversification and investment opportunities particularly to institutional investors, such as insur- ance companies and pension funds whose long-term liabilities match the long-term tenor of pro- ject bonds. Finally, they mirror the shift in the pattern of capital flows from bank loans to publicly issued bonds.' Although the volume of capital raised through international project bonds remains rela- tively small, the market has gained maturity in a very short time span, delivering a series of high profile transactions such as US$1.2 billion issued by the Ras Laffan Liquefied Natural Gas project in Qatar, US$1 billion issued by the Petrozuata heavy oil project in Venezuela,2 and US$125 mil- lion issued by the Quezon power project in the Philippines. Today, the market encompasses a broad range of project types, issue sizes, seniorities, and maturities. The total issuance volume worldwide has been on the order of US$25 billion (2000-2001) of which about one third is attrib- uted to bonds issued by projects located in developing countries. The market's long-term pros- pects, driven by the massive infrastructure needs in developing countries, look very promising. For more on global capital flows see the World Bank Global Development Finance 2003; capital flows to infra- structure development are discussed in Dailami and Leipziger (1998). 2 See Dailami and Hauswald (2001) for an in-depth analysis of the Ras Laffan project and the role that international bond finance played in its successful design and completion. Esty (1999) describes the Petrozuata project, a heavy- crude oil project in Venezuela that provides a complementary example to Dailami and Hauswald (2001) and shows how international bond finance could be accessed despite complex legal, contractual, and political risks. 2 Projet Bond, Emerging Sovereign, and IOY US Treasury Yields 20.00% - 16.00%/o 14.00%, 12.00%/ - S 10.00% _ 8.00%o. 6.00Yo - 4.00% 200%/ 1995 1996 1997 1998 1999 200D 2001 2002.06 Year -Pro*ct Bonds --*- EMBI Gbbal Index USTOY Note: All yields are yields to maturity; the Emerging Market Sovereign yields are from the JP Morgan EMBI Global index. This note examines the emergence and growth of this market as a new asset class within the emerging-market debt spectrum. The evolution of project bonds is benchmarked against the more established fixed income markets in terms of pricing (at-issue spreads) as well as legal struc- tures and covenant provisions. An examination of a sample of such bonds issued between January 1993 and March 2002 reveals that project indentures contain the standard covenant provisions aimed at mitigating conflicts of interest between bondholders and shareholders that manifest themselves through asset substitution, dividend policies, claim dilution, and under investment (Wamer and Smith 1997). In addition, project-bond indentures contain clauses that serve as com- mitment and incentive devices for host governments and other contracting parties to the project. In terms of borrowing cost, we find that project bonds are priced at a considerable markup (average 300bps spread) over comparable US Treasury securities, but with a high degree of varia- tion across bonds that depends on project-specific characteristics, bond features, and the quality of host-countries' legal institutions in determining investor rights and the degree of their protection. However, the preceding graph also shows that project bonds, despite wide variations in number of 3 issues and their size, have consistently carried issue yields below comparable emerging-market sovereign yields (JP Morgan's EMBI Global Index). Two factors are at work. First and foremost, only the most creditworthy projects can tap the markets and, therefore, often carry a comparatively higher credit rating. Second, issuers take particular care in designing their projects' organizational, legal, and financial structure when they wish to fund them in public debt markets. Taken together, both forces suggest that the at-issue spread evolution depicted above is largely due to self- selection by borrowers: only high-quality and well-designed projects and their bonds come to market which then carry credit ratings and issue-yields below a much larger and diverse group of sovereign borrowers. This note is organized as follows. We next discuss the key economic and financial issues in the international project-bond market before turning to the legal design of typical project bonds in Section Im. Section IV summarizes our analysis of credit-spread determinants that highlight the importance of the ambient institutional development, and Section V concludes. II. Key Characteristics of the Project-Bond Market Access to the international bond markets by infrastructure projects in emerging economies is a relatively new phenomenon, borne of the economic reforms, market liberalization, and financial innovations in the early 1990s. The world-wide move towards private participation in infrastruc- ture (PPI) schemes that gained momentum in the early 1 990s brought about fundamental changes in the traditional fiscal financing of infrastructure facilities.3 It also ushered in structural changes in the way in which infrastructure was operated and managed as a pre-requisite for successful pri- vate funding or projects. For instance, the development of structured credit techniques, most prominently limited recourse project financing methods, along with various risk sharing and hedge devices (multilateral and export credit agencies (ECA) guarantees, private political risk in- surance), were instrumental in containing projects' credit risk sufficiently to make them of interest to bond investors. At the same time, privatization, market liberalization, and regulatory reforms created an economic environment that could provide private investors with return potentials that could justify the considerable risks associated with debt investments in emerging market infrastructure. 3 Brealey, Cooper and Habib (1996) contains an excellent survey of the economic issues involved in project finance. 4 An important factor contributing to investor interest has been the creative design of the debt securities' legal structures such as indenture, trust structure, selective guarantees, and cove- nant provisions to mitigate risks and provide contractual protection to bond holders. Financial economists have long recognized the adverse incentives that debt finance'provides to shareholders and managers and the agency costs that those entail. Smith and Warner (1979) discuss how bond covenants typically attempt to address various conflicts of interest between different classes of claim holders while Green (1984) and John (1987) forrnally analyze the incentives that leverage creates for shareholders (project sponsors) to enhance their own returns by shifting risk to deb- tholders through project attribute selection. While most project bonds are corporate bonds, the reverse is not true. There are subtle fi- nancial, economic, and analytical differences between the two segments that merit further atten- tion in the context of an institutional analysis of the market. The dissimilarities primarily stem from the underlying economics of the borrower. In the case of a project, the issuer raises funds to finance a single indivisible large-scale capital investment project whose cash flows are the sole source to meet financial obligations and to provide returns to investors.4 In the case of a typical corporate borrower, the security is typically issued against the firm's general credit and the under- lying assets consist of multiple sources of cash flows. Hence, typical corporate bonds are secured by all the firm's various assets and cash flows that offer in themselves risk diversification and an important cross-insurance mechanism. If a certain set of cash flows becomes unavailable for debt service, firms typically have other sources of cash that might tide the issuer over the liquidity cri- sis. No such cross-insurance exists in the case of project bonds: the moment the single source of cash flows ceases to exist, the issuer experiences a liquidity crisis that might force it to default on its bonds. In addition, projects suffer from asset-specificity (location and/or use of the assets), often ill-defined or ill-enforced property rights, and bilateral monopoly settings (dominant output buyer) that render them vulnerable to opportunistic behavior and unilateral contract renegotiation. Indeed, such opportunistic behavior coupled with shortcomings in the ambient legal institutions is often at the root of project's economic distress and, ultimately, financial distress. 4 On the other hand, the single source of cash flows and limited number of contractual relations facilitated the analy- sis of project bonds. 5 These often overlooked, but crucial differences between project and general corporate bonds subtly affect investors' risk perceptions, the pricing of the bonds, and their legal structure. In particular, investors do not tend to view the underlying assets as "true security" even if they are pledged as such, but take into account and price factors, such as the creditworthiness of off-takers, third-party guarantees, the legal and institutional environment, and, ultimately, the quality of the cash flows. Put differently, investors in project bonds are much more cash-flow quality oriented than buyers of typical corporate bonds and tend to price factors that determine the underlying eco- nomics of the project. However, since projects and their securities demand much more careful analysis of the issuer's economic and legal structures, buyers of project bonds are mostly sophisti- cated institutional investors that have the requisite analytical expertise, rather than retail investors. In order to document current trends and best practices in the international project-bond market, we collected a representative sample of 105 emerging market project bonds issued be- tween January 1993 and March 2002. The issue information that we cross-checked with other data sources comes mainly from Bloomberg and Interactive Data Corporation (IDC). If the spread-at- issue over comparable US Treasury securities is not provided, we calculate it from the bond's is- sue yield and the yield of an interpolated maturity-matched Treasury security. Bond prospectuses and ratings studies from Moody's and Standard & Poors provide the necessary information on the projects' contractual structure, its off-take (output supply) agreement, the bond covenant, and le- gal terms and conditions. Table 6 in the Appendix lists all our bonds by country and provides specific information on the terms and structure of each issue. Our sample reflects a broad cross-section of countries, project types, and sectors. International project bonds differ widely in their issue size, maturity, issue spread, host country sovereign spread, underlying project structure, legal characteristics, and covenants. Issue size ranges from US$23 million (LIGHT, Brazil) to US$1 billion (Kowloon- Canton Railway Corp., China, and Pemex Mexico) their rating by Moody's from AAA to B2, their maturity from less than three years (Transportadora de gas del Sur, Argentina) to 100 years (albeit callable after 30 years), and the yield at issue over US Treasuries from 10 basis points for a convertible bond by a Chinese issuer to 802 basis points for a South-African one. The following table summarizes typical characteristics of project bonds on the basis of our sample. 6 Characteristics Mean Std. Dev. Min Max Spread over US Treasuries 297.80 173.81 10 802.17 Amount 278.07 201.62 23 1000 Maturity (years) 11.82 10.50 2.97 100 Rating classification (average of Moody's and S&P) BBB/BBB- 3 notches B AAA Based on a sample of 105 infrastructure-related, US dollar-denominated international bonds is- sued by projects in 20 emerging economies (Argentina, Brazil, Chile, China, Colombia, Czech Rep., Dominican Rep., Hong Kong, India, Indonesia, Malaysia, Mexico, Panama, Philippines, Qatar, Russia, South Africa, South Korea, Thailand, and Venezuela). On average, emerging-economy project bonds are issued at approximately 300 basis points above US Treasury securities of comparable maturities, have a surprisingly high issue size of US$278 million, a maturity of just under 12 years and are rated slightly below investment grade (exactly between BBB- and BBB). Most project bonds are senior debt or issued against a collec- tion of project receivables as asset-backed securities (ABS). The latter type of debt, while not ex- plicitly senior obligations of the project, are issued as pari passu instruments that will become de facto senior once other unsubordinated debt comes into existence. Unsecured debt tends to be rated higher than secured debt, possibly reflecting the fact that higher rated projects can afford to provide less security to their investors. In terms of geographic origin, intemational project bonds from Latin America and Asia are more numerous than from Eastern Europe, the Middle East, and Africa. All major project types are represented albeit with a particular concentration of issues in the energy, power, telecom, and transport sectors. III. Covenant Provisions A fruitful conceptual framework for analyzing projects and, in particular, their organizational, contractual, and financial design relies on the view of the firm as a nexus of contracts. First for- mulated in the seminal papers by Alchian and Demsetz (1972) and Jensen and Meckling (1976), it underlies much of modem corporate finance. The allocation of control rights and interaction of all constituent contracts of a firm (infrastructure project) motivate financing choices (Fama, 1990), determine corporate governance arrangements (Jensen and Meckling, 1976), and even provide a framework for project valuation (see Kaplan and Ruback, 1995 for an application in terms of dis- counted cash flows). Hence, we would expect bond covenants of projects that, by their very na- ture, most closely correspond to the stylized view of the firm as a nexus of contracts, to reflect and 7 address conflicts of interests not only between different claimholders but also other stakeholders, such as host governments and customers, in the project. Projects suffer from typical contracting problems arising from relationship specificity, sunk costs, and the associated "hold-up" problem that were first described in other areas of eco- nomics by Klein, Crawford, and Alchian (1978), and Williamson (1979, 1983). Three types of solutions have been proposed in the literature that balance incentives for ex-ante efficient invest- ments and ex-post trade efficient: (i) writing contracts with proper legal remedies in case of breach of contract (Shavell, 1980 and 1984; Rogerson, 1984), (ii) agreeing on a rule for the re-negotiation of contracts (Aghion, Dewatripont and Rey, 1994), and (iii) writing option contracts (Noldeke and Schmidt, 1995). In addition, the parties can always attempt to write a self-enforcing contract so that, as Jensen and Meckling (1976) have argued, conflicts of interest between bondholder and stockholder are resolved through the contractual and financial design of firms. This insight under- lies much of our analysis of project bond covenants that we can take to be the contractual re- sponses to the afore-mentioned contracting problems. Since the presence of risky debt in a firm's capital structure can lead to expost conflicts of interests between the firn's equity holders and bondholders, contractual devices such as debt covenants have evolved to mitigate their adverse consequences. For instance, companies that issue bonds either on a project (non-recourse) or corporate (on-balance sheet) basis, be it in domestic or in international markets, generally agree to a set of contractual covenants requiring them to take or to refrain from taking certain specified actions. Such actions are designed fundamentally to protect the interest of bondholders-safety and seniority of their claims, repayment, and legal remedies in the event of default-after the bonds have been issued. Covenant provisions contained in bond indentures typically take the form of restrictions on dividend, M&A transactions, and asset dis- posals, limitations on indebtedness, requirements of third party guarantees, maintenance of good regulatory standing and, in certain circumstances, the establishment of offshore and debt service reserve accounts. Violations of such provisions usually trigger contractual penalties or renegotia- tion and might ultimately lead to default and court-supervised bankruptcy proceedings. The ability to design and enforce solid bond covenants to protect the interest of bondhold- ers is a critical factor for infrastructure projects located in developing countries in tapping off- shore markets for financing. The complexity of infrastructure project finance transactions- 8 involving multi-source financing structures, numerous public and private contracting parties, and intricate contractual arrangements and legal documentations, compounded by the weakness in the legal and institutional framework to protect investors interests-makes this task a challenging one. The specific covenants included in a particular debt agreement and the extent to which such covenants effectively serve to protect the interests of creditors depend inter alia on the nature of the debt instruments, governing law, and the .underpinning legal and institutional frameworks governing contract formation and enforcement. Given that the writing, negotiating, and monitor- ing of specific provisions are costly, two sets of considerations become relevant: the ease with which the stipulated covenants can be monitored, and the scope for potential opportunistic behav- ior that could lead to transfer of wealth from bondholders to shareholders. More generally, investors are concerned about the availability of legal recourse that de- pends on the bond's terms and the quality of the legal and institutional environment in the host country. An examination of the project bond covenants in our sample reveals that project bond indentures contain the usual covenant provisions aimed at mitigating typical shareholder- bondholder conflicts such as asset substitution, dividend policies, claim dilution, and underin- vestment (Wamer and Smith, 1997). In the absence of sufficient contractual protections, the out- come is likely to be an inefficiently low investment, often referred to as the under-investment phenomenon (Hart and Moore, 1988). In addition, they contain two further categories of clauses that arise from the very specific nature of project finance. Project debt covenants include incentive provisions for the contractors, operators, and sponsors such as performance targets, mandatory penalties, and minimal equity par- ticipation in the project. They also contain institutional environment provisions that, in case of changes in the ambient regulatory, legal, or tax environment, trigger change of control and/or mandatory redemption of the debt that would assure bankruptcy and operating disruptions of the project. Akin to poison pills, such provisions strengthen the position of (foreign) creditors vis-a- vis the host country and its policies. From our 105 project bonds, we extract a subsample of 27 bonds for which we have de- tailed covenant information from offering circulars, regulatory filings, and rating analyses. As the 9 following table shows, the summary statistics for the subsample mirror the financial characteris- tics of the full sample: Characteristics Mean Std. Dev. Min. Max. Spread over US Treasuries 182 117 10 375 Amount* 319.49 193.51 180.00 800.00 Maturity (years) ** 10 3 5 18 Rating classification (average of Moody's and S&P)*** BBB+/BBB 2 notches BB A+ (*) Ras Laffan issued the largest and AES China the smallest amount (**) Ras Laffan has the longest and China Telecom the shortest maturity (***) CEZ Finance has the highest and Fideicomiso Petacalco with the lowest credit rating Based on a subsample of infrastructure-related, USD-denominated international bonds issued by projects in 5 emerging economies (Chile, China, the Czech Republic, Mexico, the Philippines, and Qatar) for which full covenant information is available. All project bonds in our sample are issued under New York Law. This particular segment of emerging market debt has often acted as an innovating force and contractual catalyst as the fol- lowing example shows. While project bonds often contain collective action clauses such as quali- fied majority rules to limit inter-creditor conflicts of interests, comparable sovereign-bond inden- tures issued under New York law typically did not have such provisions until recently. However, established market practices seem to be changing as the Republic of Mexico recently offered a global bond with a qualified majority clause (February 2003) followed by Brazil (April 2003).5 In some sense, sovereign borrowers from emerging economies follow more established corporate precedent in order to insure a better reception of the issue by investors. A preliminary analysis shows that project covenant provisions differ widely in their strin- gency. The more projects are removed from their sponsors as measured by specific references to their limited-recourse status (about 52 percent), the more restrictive their covenants tend to be. As stand-alone investments representing a single source for cash flows, debtholders require additional assurance that cash flows (and operations) are not used to enhance shareholder value to their det- riment. S For more on the Mexican issue see Dailani and Kim, "Mexico's Collective Action Clause Bond," International Finance Briefing Note 24, March 7, 2003, The World Bank; the Brazilian issue is discussed in the International Finance Review, May 2003. 10 Category Provision/Restriction Type Frequency Project driven Limited recourse status 52.31% Limited recourse definition 28.57% Collateral 33.33% Fixed asset 14.28% Receivables 38.09% Off-shore trust account 28.57% Intercreditor Agreement 9.52% Stakeholder incentives Capitalization requirements 19.05% Party-specific equity stakes 19.05% Performance-contingent put provision 17.64% Performance targets, penalties 4.76% Govemment incentives Mandatory redemption for concession cancellation 23.80% Redemption for change in tax law or regulation 66.67% Maintenance of govemment approval, regulatory compliance 19.04% Asset substitution Put provision 9.52% Contingent put provision 17.64% Cost overrun 9.52% Asset sale, lease-back 85.71% Transactions with affiliated firms 23.80% Counter-party restrictions 9.52% Nature of business 42.85% Use of funds 23.80% Claim dilution Additional indebtedness 73.68% Lien limitations 100% M&A restrictions 95.23% Collateral value preservation 19.04% Modification of indenture 85.71% Reporting requirements 80.95% Maintenance of insurance 33.33% Equity conversion 5.00% Permission of highly-leveraged transactions 28.57% Payments Dividends, debt-service coverage ratio restrictions 47.62% Sinking fund 35.00% Third-party guarantees, debt service reserves 26.31% Default definition 85.71% Underinvestment Call provisions 26.67% Investment limitations 33.33% Based on information extracted from the offering documents (registration filings, offering circulars, rating studies) avail- ablefor a subsample of 27 infrastructure-related, USD-denominated international bonds. The Frequency column records thefrequency of occurrence of the respective provision types in the bond covenants. In particular, we classified project- bond covenant provisions into 45 broad categories and seven instrument-specific classes and attributed for each bond in- denture containing a particular clause orfeature a 1, and 0 otherwise. While all indentures contain sensibly the same standard provisions aimed at preventing as- set substitution, claim dilution, cash payments, and underinvestment, two thirds of the bonds also include project-specific stipulations. Most telling are minimal ownership requirements (19 percent of covenants) for sponsors, operators, contractors, and off-takers. Clearly, such provisions are meant to align the interests of certain stakeholders crucial to the project's commercial success 11 with debtholders. Equity stakes act as commitment and incentive devices for key players. Provi- sions specifying remedies in case of cost overruns (24 percent of indentures), performance targets (5 percent), capitalization requirements (19 percent), and restrictions on counter-parties (10 per- cent) further protect the interests of debtholders. A second set of covenant provisions address the institutional environment and possible opportunistic behavior by regulators and host governments. Roughly one quarter of indentures (24 percent) provide for mandatory debt redemption in case of concession cancellations (a further 17 percent offer optional redemption at the discretion of the bondholders in case of completion, fi- nancing, or operating problems). Conversely, 22 percent of covenants stipulate that the projects are to maintain government approval and comply with all laws, rules, and regulations applicable to the project. The objective is clear: on the one hand, the project is not to give the host country any reason to intervene. On the other, mandatory redemption in case of concession cancellations forces the project into bankruptcy so that the ensuing disruption of service is meant to dissuade the host country from unilateral regulatory actions. In the same vein, 77 percent of all indentures require mandatory or optional early redemption in case of changes in tax regulation. IV. Determinants of Credit Spreads The cost of international bond financing for infrastructure projects in emerging economies is a key determinant of their tariff structure and, hence, economic viability. Our analysis of at-issue credit spreads of emerging market project bonds over US Treasuries reveals how legal, regulatory, eco- nomic, and financial institutions in host countries influence risk perceptions and, hence, the cost of debt for infrastructure development. We find that market risk perception in terms of at-issue spreads over US Treasury bonds are a function of a project's contractual structure and its ambient institutions. Since it is nearly impossible to anticipate on all contingencies in writing the contract, and since parties might have an incentive for opportunistic behavior, contracts are always incom- plete by their very nature and need to rely on other institutions for their execution.6 It emerges that the quality of the ambient institutional environment is an important factor for market risk percep- tions and the initial pricing of project bonds. 6 According to the transaction-cost approach, contract incompleteness is attributed to high transaction costs of writ- ing, negotiating of contracts, and costs associated with monitoring contractual performance; see, for instance, Joskow (1987, 1988), Hart (1988), and Aghion and Bolton (1992). 12 In theory, the (second-best) optimal choice of debt contracts can mitigate some of these risks as long as investors can threaten the firm with a future cost that one could interpret as collat- eral realization (Diamond, 1984; Gale and Hellwig, 1985; or Bolton and Scharfstein, 1990), with- holding of new financing (Gromb, 1999), or liquidation (Hart and Moore, 1994 and 1998). In practice, the effectiveness of such covenant provisions critically hinges on the quality of the ambi- ent legal institutions required to make the investors' threat credible and, thus, the contract self- enforcing. Hence, we would expect pricing to be a function of the institutional environment and project attributes bonds in addition to the nature of their covenants. The institutional, political, and economic environment feeds through to project-bond pric- ing through the market's collective assessment of the issue's systematic and idiosyncratic risks and, hence, the premium that bondholders demand over comparable default-free sovereign bonds. First and foremost, investors take into account the likelihood of debtor default and recovery in bankruptcy. In the context of project bonds, counter-party (off-take), price, and demand risk drive idiosyncratic risk perceptions, while political, macroeconomic, and institutional factors such as definition and enforcement of property rights determine the systematic ones. The institutional en- vironment, often overlooked or taken for granted by researchers and practitioners alike, is of par- ticular importance as it can mitigate or amplify the degree to which counter-party and political risks feed through to creditors. Put differently, deficiencies in the institutional development of a host country might exacerbate the market's perception of counter-party and other risk factors in pricing bonds. It also explains why project-rating analyses pay particular attention not only to the project's contractual structure but also to its ambient legal, regulatory, and political environment. Traditionally, empirical studies of credit spreads have analyzed the dynamic aspects rather than cross-sectional ones such as the legal and institutional factors affecting corporate bonds (Longstaff and Schwartz, 1995 or Duffee, 1998), reflecting the focus of much of the theoretical work in this area. However, recent work by Madan and Unal (2000) linking default rates to struc- tural factors implies that credit spreads are linearly related to firm-specific and exogenous vari- ables. In contrast the recent theoretical literature (e.g., Duffie and Singleton, 1999), this approach provides a solid theoretical foundation for the nascent empirical literature on the cross-sectional determinants of credit spreads. Closest to our analysis are Elton et al. (2001) who relate the cross- sectional variation of US corporate yield spreads to factors other than default expectations such as taxes and equity risk factors. However, they do not study the impact of issuer-specific contractual 13 and organizational design factors on credit spreads, nor the impact of institutional factors such as the quality of legal, regulatory, and political institutions. In analyzing the determinants of (at-issue) spreads of project bonds, we relate project credit spreads over US Treasuries to the relevant issue information (amount, maturity, rating), a set of variables extracted from the bond's covenant provisions (seniority, collateral, etc.), industry indicator variables (energy, power, water, etc.), factors capturing financial and economic aspects of the underlying project, a set of host country economic indicators (GDP, growth, etc.), and a set of indices measuring a host country's quality of financial, legal, and political institutions in a cross-sectional random-effect regression framework. Tables 2 and 3 contain detailed summary statistics on our explanatory variables while table 5 describes the institutional variables in more detail. More precisely, we estimate the following linear cross-sectional model of project credit spreads by random country-effects regressions: SPREAD; =flo + ZI3k1SSUEki + E )kBCOVki + EflkljDS, + E jkPROJ,. I k!K, K,/J PROJA, I5k:K, K,