Understanding the effects of institutions on labor market outcomes has been a crucial research agenda in economics. Which institutional factors allow labor markets with incomplete contracts to achieve economic efficiency, and do these factors affect how a given surplus is split? More specifically, we want to know how the flexibility of a workers contract - i.e. wage flexibility and the possibility of firing - and the ability to replace a worker determine market outcomes. Despite strongly held beliefs and common intuitions on the issue, the empirical evidence is not conclusive.
We present an experimental design to investigate how efficiency and distribution of production surplus change in a worker-firm relationship when key institutional factors are varied, in particular employment protection and substitutability of labor. Methodologically, we argue in the proposal that firm-worker relations are best modeled using an indefinite horizon. Consequently, we set up the interaction as a repeated gift exchange game with random termination. Moreover, we elicit beliefs to investigate whether any differences in behavior can be explained by agents' perceptions about their alternatives to a given relationship. The failure of standard equilibrium concepts to discriminate between treatments motivates a discussion of equilibrium selection mechanisms and alternative theories to derive sharper predictions, with our collected data on beliefs and choices providing guidance.
Our project's contribution to the field can be summarized as follows: First, this project goes beyond existing worker-firm gift exchange experiments in that we model worker-firm relationship as a repeated game with indefinite horizon. This design choice is motivated by our focus in understanding institutional effects, and allows cooperation to be sustained theoretically, without reference to behavioral (fair/reciprocal) types. We also take effects of employment protection and substitutability of labor to be most salient when the firm-worker relation is of indefinite horizon.
Second, we take a step ahead in exploring how agents' beliefs affect their actions in a firm-worker relationship. We hope that our findings will shed light on how beliefs on "outside options" (such as firm's beliefs on expected payoffs if they replaced the current worker with another) affect strategy formation and equilibrium selection in the context of infinitely repeated games.
Third, we investigate how the ability to replace a worker with a new one as opposed to the ability to dismiss her without replacement affects efficiency, distribution of production surplus and unemployment differentially in an economy. Again, to our knowledge, our project is the first to make this distinction experimentally. We take this to be crucial for understanding effects of hiring frictions on market outcomes in terms of efficiency and distribution.
Our project has immediate relevance for labor market policies since we look precisely at efficiency and distribution, which are key target variables for policy. Moreover, our design allows conclusions about the link between institutional structure and unemployment. Although our experiment is limited to 3 treatments in consideration of feasibility and simplicity, we intend for it to become the basis of a broader research agenda investigating institutional structure on market outcomes. The extent of public information in the market and individual versus collective bargaining are but two examples.
Our experiment set out to contribute to the understanding of labor markets, and in particular, to the ongoing debate about employment protection laws. A key contribution of our experiment is to separate the effects of being able to fire from those of being able to replace. We believe this distinction to be crucial for understanding why employment protection laws are costly. Our results show that both firing and replacement strongly and separately affect market outcomes. Restrictions on firing have a negative effect on efficiency as predicted by standard economic theory, and we do identify incentive effects as a key contributor to this fact. However, we also find that the ability to replace workers, as opposed to merely firing them, matters for efficiency, despite it being theoretically possible to induce maximal effort even without replacement. This suggests that incentives are only part of the story in explaining why employment protection may hurt efficiency. We identify several factors contributing to the result, notably the sorting of more cooperative workers into jobs. When workers differ in their inherent productivity, the sorting effects may be expected to be even more severe. In policy terms, this suggests that attempts to fix only the incentive problem of such laws through mechanisms such as bonus payments, are unlikely to eliminate their efficiency cost. On distributional issues our main finding is that the ability to replace workers does not increase the share of surplus firms obtain: equal splits are predominant in both markets. This does not contradict non-cooperative game theory, but is in stark contrast to the logic of bargaining with (endogenous) outside options, which suggests that replacement should allow firms to take home a much larger share of the surplus. Thus, in relation to the debate on employment protection laws, we find no evidence that that firms use the ability to replace as a bargaining tool to extract higher surplus. Higher firm profits when replacement is possible are primarily due to efficiency effects. More generally, beyond that debate, the equal split finding suggests an important role for fairness norms in equilibrium selection for markets with incomplete contracts. The difficulty in identifying the exogenous effects of restrictions on firing and replacement using field data made an experimental approach particularly appropriate to answering our research questions. Our experiment is also well suited to inform the theoretical relational contracts literature both with respect to equilibrium selection and forces that are not captured by conventional models. In that regard, we believe that our design choice of random termination rather than finite repetition, consistent with theory, strengthens the relevance of our conclusions. Allowing for an indefinite horizon may also have contributed to another important feature of subject behavior, which has not previously been observed in the literature on experimental labor markets: promotion. In all treatments, firms used promotion contracts, which enabled them to screen for uncooperative workers early in a relationship when the stakes were low. Promotion was especially important for firms under employment protection, and can help explain persistent positive levels of efficiency in that treatment. Promotion is ubiquitous in real-world labor markets. The fact that it emerged endogenously in our experiment, without an explicit formal mechanism that was given or pointed out to subjects, makes us optimistic about the ability of our experiment to capture essential features of real-world employment relationships. Beyond it's immediate relevance for labor market policies, our work also contributes to a long literature on behavioral patterns in repeated interactions, which is a multidisciplinary research area, including psychology, sociology, political science, biology and computer science. Our results show how the ability to switch partners and the ability to endogenously terminate a relationship can affect the level and nature of cooperation in repeated interactions.
