شراکت های دولتی-خصوصی و محدودیتهای هزینه ای دولت

We consider public–private partnerships, in which a public official selects a project that is then developed and operated by a private contractor. We derive optimal public accounting rules when the official's choice among projects is biased by ideology or social ties or because of pandering to special interests. We give particular emphasis to how the rules should constrain the official's incentive to understate the costs of her pet projects. In the basic model, we show that the optimal accounting rule takes the form of a budget cap, with a project's expected cost modified to reflect the official's distortionary incentives. If the project can be partially financed privately, then “fixed-price” contracts can serve to curb political misbehavior by “securitizing” public sector liabilities. We also consider the possibility that development and operations are each handled by different contractors. Such “unbundling” deprives public accountants of forward information about future costs, but can prevent the official from funneling hidden future rents to contractors.

Public procurement accounts for a sizeable share of economic activity in most countries. Thus, how to deliver high-quality public services at low cost to the taxpayer and user is an important problem. An interesting recent development in the effort to find solutions is the growth of public–private partnerships (PPPs), both in industrialized countries (e.g., the United Kingdom, as in its Private Finance Initiative launched in 1992) and in emerging economies (e.g., Latin America, Eastern Europe, and China during the 1990s). PPPs have been created for large-scale projects in transportation (rail systems, highways, subways), medical care, telecommunications, energy, water systems, and even orphan drugs. Although the variety of risk-sharing arrangements and governance structures makes a precise characterization difficult, a PPP is usually defined as a long-term development and service contract between government and a private partner. The government typically engages its partner both to develop the project and to operate and service it. The partner may bear substantial risk and even raise private finance. Its revenue derives from some combination of government payments and user fees. In comparing PPPs to more traditional procurement (in which project development on the one hand and operations and maintenance on the other are generally arranged under separate contracts), the literature has generally focused on the incentives of the private partner. For example, one much-discussed potential advantage of PPPs is that, by “bundling” construction and operations, they induce the developer to internalize cost reductions at the operations stage that are brought about by investment at the development stage.1 But, by the same token, bundling may lead to a loss in operational efficiency because the best developer might not also be the best operator.2 Moreover, it may encourage choices that reduce future costs at the expense of service quality.3,4 The literature's focus on the private agent is understandable in view of the standard presumption in academic and policy work on public procurement that the government acts to maximize social welfare. Assuming governmental benevolence is a reasonable first step in the analysis of PPPs, but, of course, over-simplifies reality. Accordingly, a fair number of recent studies have departed from the benevolence assumption by supposing that the private partner or other parties may capture the procurement process by side-contracting (colluding) with the government.5 In this paper we consider a less-explored reason for why procurement projects may not align with the public's best interest: government officials may have preferences that differ from those of a social welfare maximizer. More specifically, ideology, social or political ties, or the incentive to pander may induce an official to favor the pet projects of particular interest groups—i.e., to practice “pork-barrel” politics—even though these projects may not be justifiable from the standpoint of social welfare. We are particularly interested in how spending caps can mitigate the official's biases. There is substantial evidence that politicians' project choices are influenced significantly by the desire to please constituencies and by budgetary constraints. Levin and Tadelis (2006) document that local political institutions in the U.S. have a profound impact on such choices. Less formal evidence in France suggests that efficiency considerations in the production of public goods are often secondary to the government's determination to deliver visible private benefits to particular interest groups, with costs hidden or delayed as much as possible. For that matter, the very fact that governments in many countries are made to face budgetary constraints at all would be quite mysterious if their goal were truly to maximize social welfare. Indeed, the marked increase in PPP contracts worldwide is often attributed less to the intrinsic qualities of such contracts than to governments' attempts to evade budget constraints by taking liabilities off the balance sheet.6 For this reason, some commentators worry that accounting gimmickry may become the primary motive behind PPPs, so that “governments may not take the care to properly design contracts to ensure that appropriate incentives are in place” (Mintz and Smart, 2005, page 17; see also IMF, 2005, p. 27).7 Our paper builds on Maskin and Tirole, 2004 and Maskin and Tirole, 2007 to examine PPPs as instruments in pork-barrel politics. To keep the analysis simple, we limit our focus to the constraining role of public accounting systems, and, unlike our earlier papers, ignore the restraints imposed by electoral accountability.8 However, as explained in Section 3 (see footnote 16), the most straightforward way of incorporating accountability in our model changes none of our qualitative conclusions. We lay out our benchmark model in Section 2. A public official is in charge of choosing projects and a contractor of carrying them out. Each project comprises two stages, with a commonly known first-period cost and an (a priori) uncertain second-period cost (which can be high or low). In the benchmark model, the two stages are “bundled”: the same contractor is there for both periods. The public official and her contractor have the same information about the project's second period cost: with probability x, they learn (privately) the magnitude of this cost (i.e., whether it is high or low); with probability 1 − x, they, like the public, remain uninformed. There is a continuum of interest groups, and the public official “favors” a fraction of them in the sense that she prefers a project she knows is high-cost and that benefits a favored group to one that benefits some unfavored group and whose cost is not yet known. This preference give rise to the central inefficiency of the model: the official has the incentive to “pass off” high-cost projects she favors as projects with still unknown costs. In Section 3, we study PPPs when contractors are “cashless,” i.e., they can bear no risk in their costs. This set up is particularly simple, as we can focus without loss of generality on just fixed-price contracts (which can be used for projects with known costs) and cost-plus contracts (appropriate for projects with unknown costs).9 The public official can pass off a favored high-cost project as one with unknown cost by awarding the contractor a cost-plus contract. That is, cost-plus contracts are vulnerable to adverse selection: the official will use them not only for the projects for which they were designed (those with unknown costs), but also for her inefficient pet projects.10 We show that the public official can be induced to behave more in line with social welfare if subjected to a spending limit and a public accounting system. Moreover, the public accounting system can be chosen to be “linear,” and we derive its optimal form. The accounting costs will, in general, differ from true costs to reflect the adverse selection problem described above. The optimal accounting system induces a public spending rule that takes one of two forms. Either it is “tight,” so that only favored projects of uncertain cost and projects known to be low-cost are undertaken. Or else it is “lax,” in which case all projects are undertaken except those that are high cost and do not benefit a favored group. Ceteris paribus, lax spending limits will pertain when the fraction of interest groups that are favored is small and the probability x that the public official learns the second period cost is low. In Section 4, we allow contractors to be privately financed, a possibility that can raise welfare. Private finance allows cost-plus PPP arrangements to be replaced by more efficient fixed-price contracts; hence, the theory predicts that private finance will be associated with a higher use of fixed-price contracts. Intuitively, private finance shifts risk to the private sector and attenuates (indeed—in our simple model—altogether eliminates) the adverse selection problem. It thereby enables the “securitization” of public sector liabilities. In 5 and 6, we compare PPPs with the more conventional arrangement in which development and operations are “unbundled,” i.e., there is a separate contractor for each. We show that PPPs offer the potential advantage that projects' true costs can be assessed earlier, making it hard for the official to push through her favorite project. However, PPPs also introduce the countervailing danger that contractors may be able to mask high costs by accepting low initial payments in exchange for high rents later on. Specifically, if the contractor obtains second-period rents that are not observable to accountants at stage 1, then the contractor will be willing to undertake high-cost projects at an initially low-cost rate, providing extra scope for the public official to sidestep the financial constraints in period 1. Section 7 concludes by suggesting a few avenues for further work.