Information, institutions and efficiency: essays in experimental economics.

Summary This thesis has 3 main chapters. Chapters 1 and 2 present experimental results from a coordination game proposed by Rosenthal (1981) and Beard and Beil (1994). This game has two Nash equilibria: the first is efficient, the second relies on the use of weakly dominated strategies. In laboratory experiments based on this game, players very often fail to make decisions that simultaneously maximize the payoffs of all parties. These failures of efficient coordination arise from two behaviors: (i) subjects doubt that other players will seek to maximize their own payoff, and (ii) these doubts are, in some cases, justified. In Chapter 1, we present a new experiment that tests whether this behavior is due to the inequality of payoffs between players (which persists in most laboratory implementations conducted so far). Our data clearly show that the failure to maximize personal payoffs, as well as the fear that others might behave in this way, does not stem from inequality aversion. This result is robust to variations in decision salience, learning by repetition, and cultural differences between France and Poland. We then study the impact of information on strategic behavior in this game. The experimental treatments introduce three mechanisms that improve the level of information in the game: simple repetition, cheap-talk messages, and observation of the partner's past actions. Repetition learning increases the frequencies of the most efficient outcome, as well as the risk of the most costly strategic mismatch. Furthermore, this type of learning is replaced by individual cues. Similar to previous studies, we show that signals help predict partners' intentions, which reduces the frequency of coordination failures. Nevertheless, in contrast to these studies, we find that the transmission of information between partners, whether using messages or observation, is not sufficient to significantly increase the overall efficiency of outcomes. This occurs primarily because information transmission does not restrict the use of dominated strategies. In Chapter 2, we propose an experiment that applies commitment theory, established in social psychology, in the economic context of the coordination game. In this environment, the coordination game, which takes place with communication, is preceded by the oath stage where subjects have the opportunity to solemnly commit to telling the truth. Three main results emerge. First, in the presence of the oath, coordination on the most effective balance increases by almost 50% to a level of 75%. Second, with the oath procedure, players become more honest: they send messages that more often match what they actually do in the game. Moreover, the actions they choose are also more effective. In one, the players who receive the messages become more confident and they more often choose an action that is in line with the intentions that are sent to them.