Identifying and Estimating Beliefs from Choice Data - An Application to Female Labor Supply

This paper investigates the life cycle costs of being too overconfident about future employment possibilities. Focusing on women who experience child-related career breaks, a dynamic discrete choice model of female labor supply is estimated. A novel strategy is developed to identify expectations about future employment possibilities from labor supply choices. The strategy exploits reactions at the end of employment protection for mothers who have been regularly employed before the birth of a child. Reforms that change the length of this protection period allow to separately identify each of expectations, job-arrival rates, and preferences. Validated by suggestive evidence, the estimated life cycle model indicates that expectations are substantially upwards biased: on average women expect the half-yearly job arrival rate to be 1.6 times as high as the actual rate. This overconfidence significantly prolongs child-related career breaks and increases the number of mothers who never return to employment. Although lifetime labor market earnings drop by 14% when overconfident, individual consumption decreases only by around 4%. This difference is due to joint taxation, and most husbands are continuously working full-time while mothers tend to work part-time.

Identification of Time Preferences in Dynamic Discrete Choice Models - Exploiting Choice Restrictions

I study the identification of time preferences in dynamic discrete choice models. Time preferences play a crucial role in these models, as they affect inference and counterfactual analysis. Previous literature has shown that observed choice probabilities do not identify the exponential discount factor in general. Recent identification results rely on specific forms of exogenous variations that impact transition probabilities, but not instantaneous utilities. Although these variations allow for set identification of the respective parameter, point identification is only achieved in limited cases. To circumvent this shortcoming, I focus on models, in which economic decision-makers might be restricted in their choice set. I show that time preferences can be identified provided there is variation in the probability of being restricted that does not affect utilities nor transition probabilities. I provide many examples of choice restrictions in different contexts, such as labor economics, industrial organization, environmental economics, and marketing. The derived exclusion restrictions are easy to interpret and potentially fulfilled in many empirical applications. Identification is studied for a broad class of models, including both finite and infinite horizon models.

Time Preferences and Female Labor Supply

(w/ Peter Haan, Luke Haywood, draft available on request)

We estimate a dynamic life-cycle model of labor supply with a focus on women's time preferences. We extend the dynamic discrete choice model of female labor supply to accommodate potentially time-inconsistent behavior. To identify time preferences, we exploit natural experiments: we use several parental leave reforms in Germany which extend parental leave from one to three years. Preliminary results provide evidence for significant time-inconsistency. Child-related career breaks are, on average longer when women exhibit time-inconsistent behavior. Our approach allows us to shed light on the importance of time preferences in explaining important labor supply choices.