|Forscher:||Matthias Blonski, Paul Gortner|
|Kategorie:||Financial Markets, Experiment Center|
Based on non-standard preferences, we construct a general equilibrium model, where agents trade state contingent securities. This model exhibits multiple equilibria, which are not necessarily Pareto-efficient. Social concerns like conformity might lead to herding behavior, and agents might herd towards inefficient portfolios. In contrast, the same model with standard preferences has a unique, Pareto-efficient equilibrium.
In the experimental part, we study an asset-market, where agents trade state-contingent securities in a continuous double auction. We use three conditions to cleanly identify how social concerns matter for market outcomes. First, in a control condition agents trade anonymously in the asset market. Building upon the previous literature, we expect markets to be efficient in this condition. Second, in a treatment condition, we induce social concerns. Agents now can observe each others’ portfolios. Relative payoff concerns could now lead to deviations from Pareto-efficient allocations. However, agents could draw conclusions how to structure their portfolio, when observing others. So social-learning is a possible confound for the identification of social-concerns. Therefore, in a third condition, some agents receive a random portfolio and are restricted from trade. The other agents are still able to trade and only observe those with a restricted portfolio. Since their portfolios are random, learning is ruled out. If these portfolios influence the non-restricted agents, it can only be due to social-concerns.
It is clear from the household finance literature, that social-concerns play an important role in financial decision making. However, the general-equilibrium implications are only explored theoretically. This project bridges the gap between the empirical literature on peer-effects in financial decision-making and the theoretical literature on the general-equilibrium effects of social-concerns.
In March this year, we conducted three sessions of the experiment. A fourth session is to follow at the beginning of June. The results of the first session were already analyzed. We find significant effects of our treatments. The fourth session can be analyzed using the same framework, so we expect to finish the data analysis soon after. Some building blocks of a working paper are already finished, the rest awaits the final results of our analysis. We hope to produce a first draft of the working paper by the end of August.