مشارکت های عمومی - خصوصی مقابل خرید سنتی: یک تحقیق تجربی

A government agency wants an infrastructure-based public service to be provided. Our experimental study compares two different modes of provision. In a public–private partnership, the two tasks of building the infrastructure and operating it are delegated to one private contractor (a consortium), while under traditional procurement, these tasks are delegated to separate contractors. We find support for the theoretical prediction that, compared to traditional procurement, a public–private partnership provides stronger incentives to make cost-reducing investments (which may increase or decrease service quality). In two additional treatments, we study governance structures which explicitly take subcontracting within private consortia into account.

Over the last two decades, governments in a growing number of countries initiated public–private partnerships to let the private sector take over the responsibility for building an infrastructure and subsequently operating it to provide public goods or services. In industrialized countries as well as in emerging economies, public–private partnerships have been set up for large-scale projects in various sectors such as public transportation, health care, and education.1 A key characteristic of public–private partnerships is that the two tasks of building a facility and subsequently operating it are bundled and delegated to a single private contractor, while under traditional procurement, separate contractors are in charge of these two tasks.2 An argument often put forward in favor of public–private partnerships is that when the same private contractor is responsible for construction as well as operation of a public facility, then he will be inclined to invest more during the construction phase in order to reduce the costs incurred in the subsequent operating stage.3 Hart (2003) demonstrates that the incomplete contracting approach offers a very useful framework to theoretically investigate how the incentives to make cost-reducing investments differ between public–private partnerships and traditional procurement. In his model, there are two stages. In the first stage, a public infrastructure is built, while in the second stage, the infrastructure is operated to provide a public service. In the first stage, the builder can make investments that reduce the operating costs in the second stage. In line with the above-mentioned argument, Hart (2003) finds that given a public–private partnership, the private contractor has strong incentives to make investments, since they reduce the operating costs that he will have to incur in the operating stage. In contrast, under traditional procurement, the builder has no incentives to invest in cutting the operating costs, since another private party will have to bear these costs. Whether a public–private partnership or traditional procurement is preferable depends on the effects that the cost-reducing investments have on the service quality. In particular, Hart (2003) assumes that two different kinds of investments can be made. Investment i not only reduces the operating costs, but it also increases the service quality. In contrast, while investment e also reduces the operating costs, it does so at the expense of a reduced service quality. Hence, investment i is socially desirable, while investment e might be socially undesirable if the negative side effect on the service quality is sufficiently strong. In line with Hart (2003), we consider a situation in which in a first-best world (i.e., if the investments were contractible), a high level of investment i, but a low level of investment e would be chosen. In a second-best world (i.e., if the investments are non-contractible), we are then confronted with the following trade-off. In a public–private partnership, high levels of both kinds of investments are induced. Hence, there is overinvestment with regard to e, while the first-best level of investment i is chosen. In contrast, under traditional procurement, there are no incentives to make high investments. Thus, there is underinvestment regarding i, while the first-best level of investment e is chosen. It is an important research question to investigate whether the trade-off between strong investment incentives in a public–private partnership and weak investment incentives under traditional procurement as identified by Hart (2003) is of empirical relevance. As a first step in that direction, we have conducted a large-scale public procurement experiment in the laboratory. Specifically, we conducted two main treatments, a public–private partnership (PPP) treatment and a traditional procurement (TP) treatment. We have implemented a parameter constellation where encouraging the desirable investment i is more important than preventing the undesirable investment e, so that according to the theoretical analysis, a public–private partnership is preferable to traditional procurement. The experimental data largely corroborates the theoretical analysis. In the PPP treatment, subjects chose the high levels of both kinds of investments significantly more often than in the TP treatment. As a consequence, also the total surplus generated in the PPP treatment was significantly larger than the total surplus in the TP treatment. However, modelling the private contractor in a public–private partnership as a single decision maker might be seen as an analytical shortcut. In practice, different skills are needed in the building and operating stages. Thus, it is important to take a closer look at different subcontracting arrangements. For this reason, we have conducted two further treatments. In one treatment (Sub I), the builder is the main contractor and subcontracts with an operator. As has already been pointed out by Hart (2003), in theory this setting induces the same investment behavior as the simple PPP setting (since the main contractor must reimburse the subcontractor for his operating costs, the main contractor internalizes these costs). In another treatment (Sub II), the operator is the main contractor and subcontracts with a builder. In theory, this setting leads to the same investment behavior as traditional procurement (since the subcontractor disregards the operating costs, he has no incentives to invest). Also in the subcontracting treatments, it turns out that the observed behavior in the laboratory is mostly in line with the theoretical predictions. In recent years, the theoretical literature on public–private partnerships has grown steadily. Building on Hart (2003), several contributions have investigated the implications of bundling the building and operating stages in public procurement projects.4Bennett and Iossa, 2006a and Bennett and Iossa, 2006b and Chen and Chiu (2010) explore how different ownership structures interact with the choice between a public–private partnership and traditional procurement.5Martimort and Pouyet (2008) analyze a model that includes both traditional agency problems and property rights and they find that the most relevant question is not who owns the assets, but instead whether the tasks are bundled or not. Iossa and Martimort, 2008 and Iossa and Martimort, 2009 discuss extensions and applications of this framework. Also focusing on the externalities between the tasks of building and operating a public project, Li and Yu (2010) investigate whether these tasks should be auctioned off separately or bundled. Nishimura (2011) discusses the pros and cons of bundling in the presence of risk-aversion. Hoppe and Schmitz (2010b) study how the decision to bundle the building and operating stages affects the incentives to gather information about future costs of adapting the service provision to changing circumstances. While public–private partnerships have received growing attention in the theoretical literature, so far empirical research is scarce.6 In particular, to the best of our knowledge, our study is the first experimental contribution that compares the performance of public–private partnerships and traditional procurement in the laboratory.7 Following Hart’s (2003) initial contribution, we consider a very stylized framework, solely focused on the investment incentives generated by the two different modes of provision. While the theoretical literature by now has considered many extensions that reflect particular institutional details,8 conducting experiments that take all these specific aspects into account might be a promising task for future research. Yet, before tackling such a daunting task, it is important to first gain a clearer picture of whether the basic forces underlying Hart’s (2003) work actually show up in the laboratory. A priori, this is not obvious, taking into consideration that many simple games are played quite differently by real players than predicted by standard theory. Given our focus on basic economic principles (that are of a fundamental nature and thus also relevant in other contexts), one has to be very careful in making specific policy recommendations based on the experiment, which clearly abstracts from other aspects that may be important in practice. However, experimental work testing basic economic principles can also be important to bring up relevant aspects that should then be incorporated into future theoretical models. For instance, our results suggest the need for a careful formalization of the potential effects of reputation and reciprocity, which may be particularly useful when deciding between traditional procurement and subcontracting arrangements that would yield the same outcome under conventional theory. The remainder of the paper is organized as follows. In Section 2, public–private partnerships and traditional procurement are compared, while in Section 3, different ways of subcontracting are considered. Each of these two sections consists of subsections in which we describe the theoretical framework, present the experimental design, derive predictions, and report the results. Concluding remarks follow in Section 4.

Our two main treatments provide strong evidence for the fundamental trade-off identified by Hart (2003). A public–private partnership induces very strong incentives to invest in cost reductions, which is desirable if the investments are also quality-enhancing, but may well be undesirable if the investments have a negative side-effect on quality. In contrast, under traditional procurement incentives to invest are weak, both with regard to desirable as well as undesirable investments. In the experiment, we considered a parameter constellation where inducing the desirable investment was relatively more important, such that a public–private partnership would be preferable according to the theoretical analysis. Indeed, in the experiment both kinds of investments were made much more often in the PPP treatment (in line with Prediction 1) and the principal was better off under this governance structure (in line with Prediction 2). While in the last round, almost all investment decisions were as theoretically predicted, the only noticeable deviation from the theoretical analysis was the fact that in the TP treatment, in a relevant number of cases the payments from the principals to the selected builders were relatively large, which was reciprocated by choosing high levels of the desirable investment i.33 In addition, we have considered two subcontracting treatments. The investment behavior and the principals’ payoffs again differed between these two treatments as suggested by the theoretical analysis (supporting Prediction 3 and Prediction 4). According to the theoretical analysis, moreover there should be no differences between the PPP treatment and the Sub I treatment (where the builder is the main contractor), and similarly, there should be no differences between the TP treatment and the Sub II treatment (where the operator is the main contractor). Fig. 16 illustrates the average total surplus levels achieved in all four treatments. Note that in the final period, as predicted, neither PPP and Sub I nor TP and Sub II differ much from each other. In the somewhat more complex Sub I treatment, in the early rounds the average surplus is smaller than in the PPP treatment. Yet, in practice there might be no real choice between PPP and Sub I, since modelling the consortium as a single decision maker might best be seen as an analytical shortcut. Hence, Sub I may be considered to be the relevant alternative if (as in the parameter constellation that we have chosen in our experiment) a high level of the quality-reducing investment e is less harmful than underinvestment in i. Our experiment hence illustrates that frictions within the consortium might make a public–private partnership slightly less attractive than it appears when modelled as a monolithic entity. This result might encourage further theoretical studies of public–private partnerships to open the black box of contracting arrangements within the private consortia. Moreover, if it is more important to avoid overinvestment in e, our experiment leads to another important insight. If the parties involved are interested in establishing reputations by acting in reciprocal ways, then traditional procurement might be superior compared to a public–private arrangement with the operator as main contractor. The reason is that if the investing party reciprocates generous payments, then it tends to do so by taking the investment decisions that are best for the main contractor in the Sub II treatment, while it takes the first-best decisions in the TP treatment. This finding suggests that paying more attention to reputation and reciprocal behavior might be an interesting avenue for future theoretical research on the organization of public procurement.