مشارکت بخش خصوصی ـ دولتی و خصوصی سازی تامین مالی: رویکرد قراردادهای ناقص

Governments have begun to embrace public–private partnerships (P3s) as vehicles for providing public services. This paper considers the controversial question of when private financing of public projects is optimal. Private development can dominate public financing through more efficient termination decisions for bad projects, resolving soft budget constraint problems. Due to contractual incompleteness and externalities, on the other hand, private developers cannot commit to large debt repayments, and hence can finance only a subset of valuable projects. Public developers, who do not face the same commitment problems, can finance a larger set of projects.

Over the last fifteen years, governments around the world have embraced public–private partnerships (P3s) as vehicles for the delivery of a wide variety of public services in major areas such as education, transportation, health care and corrections. Pioneered by the United Kingdom with its Private Finance Initiative of the early 1990s, the P3 approach is being adopted in countries of all wealth levels and on all continents.1 To economists, P3s may be seen as a simple extension of vertical disintegration or contracting out by governments. Rather than simply contracting out the construction of a new bridge for a fixed price, for example, a government may contract for the provision of “bridge services” including the design, construction, operation, maintenance and even the financing of the bridge. The idea behind such projects is most often expressed in general language as harnessing the efficiencies and innovativeness associated with a competitive private sector to help government achieve its public service goals at lower cost. Our aim is to be more precise about some of the tradeoffs involved. In our view (de Bettignies and Ross, 2004), two features of modern P3s set them apart from simple contracting out. First, the number of tasks that are contracted out to the same party or consortium is larger, as in the bridge example just given. Second, the privatization of the finance function – i.e. the delegation of the financing responsibility to a private firm or consortium – at one time extremely rare, has more recently become a central feature of P3 projects. In this paper we focus on the second and most controversial feature of P3s — the privatization of the finance function.2 Incorporating elements of industrial organization and corporate finance theory in a normative public policy analysis, we analyze the conditions under which either public or private finance ought to be preferred over the other. The model also sheds light on the positive question of when governments will choose private over public finance. The model considers a particular project, the construction and operation of a bridge, for example, which can be financed and developed by a private firm/consortium, or by a government agency. Whoever undertakes the project, private or public developer, must secure the initial capital required from an investor. We draw from the incomplete contracts frameworks3 of Bolton and Scharfstein, 1990 and Bolton and Scharfstein, 1996, and Hart and Moore (1998) to determine the optimal (debt) contract between the developer and the investor; and derive several key results: With private development, two issues arise. The first issue is related to contractual incompleteness: the possibility of strategic default by the developer caps the debt repayment that she4 can commit to make, and limits the amount a lender is willing to provide to the developer in the first place. Accordingly, contractual incompleteness under private development leads to fewer projects being financed relative to the first-best benchmark. The second issue results from the private developer maximizing profits rather than social surplus, and thus ignoring the impact of her decisions on consumer surplus. This has two consequences here: 1) The private developer might make profit maximizing but socially inefficient decisions. This lowers the social surplus relative to the first-best benchmark. In turn, it has a negative impact on the debt repayments that can be made to the investor and on the number of projects that can be financed in the first place. 2) Of that social surplus generated, the private developer extracts profits, but does not internalize the consumer surplus, and this also lowers the size of the debt repayments she is willing to make and the number of projects financed. Thus, contractual incompleteness and externalities both make private development ex ante inefficient by limiting the number of projects being financed. Externalities also yield an ex post inefficiency in reducing surplus generated by the projects that are indeed financed. We show, however, that government intervention may help mitigate these concerns: Through simple contract design and co-financing, the government can eliminate all externality-related inefficiencies. With public development – when the government does the borrowing – the problem is different. To the extent that the electorate can use the public developer's observable actions to infer information about the government's underlying quality, the public developer may take actions that are socially inefficient, in an attempt to manipulate the electorate's beliefs about government quality, and improve reelection prospects. We show that these attempts also lead to both ex post and ex ante inefficiencies. Comparing private and public financing from both ex ante and ex post standpoints, we find that – when both types of financing are available – private development may be preferred, as it gets around the belief manipulation problem faced by the public developer. On the other hand, private developers can only commit to smaller debt repayments, and hence can only find lenders for a subset of socially valuable projects. Indeed, some projects can only be financed by public developers, who do not have the same commitment problems. From a methodological standpoint, our model is most closely related to the incomplete contracts papers mentioned above; and indeed the possibility of strategic default under private development was identified in that literature previously. However we do depart from that line of research in placing issues of social welfare, as well as the role of government, at the forefront of the analysis; and believe the other key results of our model to be novel. In the P3 literature, the focus so far has been on the trade-off between public and private provision, without particular attention to financing (Schmidt, 1996, Hart et al., 1997, Shleifer, 1998 and Besley and Ghatak, 2001), and on the “bundled” outsourcing of both construction and operation to a private consortium (King and Pitchford, 2000, Bentz et al., 2002, Bennett and Iossa, 2003, Hart, 2003 and Iossa and Martimort, 2008). Our contribution here is in examining a different characteristic of P3s – private financing – and in analyzing the trade-off between private and public development through a modern corporate finance lens. Our modeling of public development is also related to the literature on the “soft budget constraint” (SBC), pioneered by Kornai, 1979, Kornai, 1980 and Kornai, 1986, and formalized more recently by Dewatripont and Maskin (1995) and others.5 This literature attempts to explain why governments tend to bail-out or continue projects that should be terminated. This tendency is central to our modeling of public financing, and here reflects an attempt by governments to manipulate the electorate's belief about their intrinsic quality, and hence to increase their reelection probability.6 The hypothesized implications of such soft budget constraints for the continuation of weak projects have been documented by many researchers.7 The two papers on the efficiencies of government spending that come closest to ours are those by Dewatripont and Seabright (2006) and Coate and Morris (1995). Both papers model governments that make “inefficient” decisions in order to improve their chances at reelection. Like the present paper, both consider that governments can be either “good” or “bad” in terms of their talents or true objectives and that they will take actions consistent with making voters believe they are good. Coate and Morris focus on decisions to redistribute resources toward groups favoured by the government (but not voters), while Dewatripont and Seabright consider decisions governments may take to proceed with projects (even those that may be wasteful) so as to be seen by voters to be working hard. Importantly, however, neither paper examines how to deliver public projects as a choice – considered here – between private and public developers. The paper is organized as follows: Section 2 sets up the basic model. 3 and 4 examine private development. Section 5 focuses on public development. Section 6 compares the two types of financing from both normative and positive standpoints. Finally, Section 7 discusses key assumptions of the model and concludes. Proofs are in the Appendix A.

Our model provides conditions under which private development – in effect a P3 with private provision of financing – dominates public funding and vice versa. We have shown that private development may be ex post superior to public financing, because under certain reasonable conditions it leads to the efficient termination of bad projects, while public developers may sustain such projects for political reasons. When both types of financing are available, private development may thus be preferred, as it “hardens” the project budget constraint. On the other hand, private developers can only commit to smaller debt repayments, and hence can only find lenders for projects requiring smaller capital outlays, i.e. with higher expected returns. Projects with lower expected returns can only be financed by public developers, who do not have the same commitment problems. In order to provide clear, stark and tractable results, the model presented here is highly stylized, and some of the assumptions made implicitly or explicitly should be explained briefly. First, a question that naturally arises is as to what makes this kind of relationship a P3 as opposed to, for example, simple contracting out. Indeed, to some extent P3s are really just extensions of contracting out, so the differences are more of degree than kind. In our view, contracting out reaches the stage at which we can usefully call it a P3 when certain conditions are met, including: (i) the private sector partner is responsible for the financing; (ii) multiple tasks such as financing, construction and operation are assigned to the same private partner (or consortium); and (iii) the private sector partner provides the operation of the facility.23 These three conditions are not generally met in simple contracting out arrangements, but are present (at least implicitly) in the model we develop here. The second issue is related to our assumption of unverifiable cash flows. An alternative would be to assume profits to be verifiable, in which case under private development an equity contract would be optimal: At date 0, the developer relinquishes a fraction of the equity to the investor such that his expected share of expected profits exactly equals his initial capital investment. In this scenario, the inefficiency that was previously associated with contractual incompleteness disappears, but without government intervention the pie and slice effects associated with the externality, and its consequent impact on ex post and ex ante inefficiencies, remain. As in our base model, however, government intervention can help eliminate the inefficiencies associated with this externality, and hence with verifiable profits and government intervention, private development is both ex post and ex ante efficient. In contrast, under public development, the verifiability of profits has no impact: The outcome in terms of inefficiency is the same with both debt and equity contracts. Thus we conjecture that allowing for verifiable profits would tilt the tradeoff between public and private development in favor of the latter, which would strictly dominate the former. Third, we assumed that the quality of the private developer was independently distributed from that of government, and consequently under private development, the electorate does not infer anything about government from the private developer's actions. But the quality of government may actually be correlated with the quality of the private developer, in the sense that a bad government is more likely to select a bad private developer. An extreme example would be the case of perfect correlation between developer type and government type. In that case, the kind of signal manipulation that we analyzed under public development in the base model would also take place under private development: In the bad state, the bad government would have an incentive to bribe the bad private developer to continue a project which ought to be terminated, thus generating additional inefficiencies under private development. Accordingly, in such cases the tradeoff between public and private development would tilt in favor of the former. Fourth, while we do not incorporate any considerations of moral hazard in our model, some qualitative guidance as to the possible effects follows from the work of Dewatripont and Seabright (2006) cited earlier. In their paper, the authors assume that governments must exert effort to find good projects. In the context of our model (in which some project must go forward) this kind of effort decision would seem to reinforce the forces that already encourage governments to maintain bad projects they are developing. Whether a project is going badly because the government selected it poorly or developed it incompetently might not matter so much to voters at election time – they will be unhappy in either case. What is potentially different is the reaction of governments when they have chosen the private development alternative. If the government needs to undertake no effort in this case (i.e. if the developer is responsible for the effort decision) nothing need change in our model. However, suppose the government undertakes the effort before deciding whether to develop the project itself or permit private development. In this case a project terminated by the private developer could signal to voters that the government did not exert sufficient effort; and voters might respond by punishing the government at election time. This will strengthen the government's interest in bailing out failing private projects reducing the benefit from private development. Finally, we remind the reader that certain important features of P3s were not considered here at all. Importantly, the potential advantage of private funding we work with in this paper has nothing to do with the private sector being more innovative or having lower costs of production, two of the reasons frequently offered in support of private involvement in P3 projects.24 And the government's loss of control over managerial decisions, often used as an argument against P3s by opponents, is also not an element of our model. We look forward to incorporating such dimensions of P3s in future research.