Mislearning and (Poor) Performance of Individual Investors

We contribute to the literature by developing a model that studies the incentives of individuals to engage in market timing when their ability to trade is unknown. We characterize the optimal investment decision process and provide empirical predictions. Using administrative data from the Chilean Defined Contribution pension system, we test our predictions and provide evidence regarding the performance achieved by members that make voluntary fund changes. Regarding individuals' performance, our main finding is that performance is generally poor, although this depends on the exact time period under analysis. However, performance is negatively related to the number of fund changes, regardless of the period considered. These results cast doubt on the presence of market timing skills among affiliates. Simulation exercises further provide support to the claim that luck has played a major role in explaining members' performance. We find evidence of the existence of learning. Indeed, individuals tend to make more (less) fund changes when they have made more (less) successful fund changes in the past. However, this result holds when using a naive definition for a successful change. If members of the pension system are overestimating the performance of their investment decisions and engaging in further active behavior (i.e. pension fund changes) as a result, our findings suggest that this is detrimental for the expected return of their pension savings. These results stress the need for serious considerations regarding the trade-offs between having a large degree of freedom in the ways in which pension savings can be invested against the actual outcomes obtained by affiliates who are making frequent use of these possibilities. Also, further research about the way in which individuals evaluate their performance and understand the consequences of their investment decisions seems necessary