Direkt zum Inhalt

Essays on Time Preferences and Expectations in Dynamic Decision Making


Ulrich Schneider



The three chapters of this dissertation provide new insights in modeling and estimating dynamic discrete choice models. Building on previous identification results, several new strategies are presented to estimate important aspects of dynamic decision processes. A focus lies on hyperbolic discounting and biased expectations, two elements that the vast majority of the literature on female labor supply ignores. Furthermore, the empirical analyses demonstrate that the proposed models describe behavior better than the models dominantly used in the literature. Estimations show that these elements have an economic meaning, as they increase the costs of child-related career breaks. The first chapter provides new exclusion restrictions for identifying the exponential discount factor in dynamic discrete choice models. It demonstrates how exogenous changes in restriction probabilities can be used to recover time preferences from choice data. The second chapter further extends this strategy and analytically shows the identification of hyperbolic discounting parameters within a three-period model of dynamic discrete choice. Furthermore, the chapter discusses how the identification carries over to models with more than three periods. Focusing on women's child-related career breaks, a dynamic discrete choice model of female labor supply, human capital accumulation and labor market frictions is developed. Data from the German Socio-Economic Panel is used to estimate the model, finding a present-bias of 0.77 that significantly deviates from an exponential discounting model. Interestingly, the yearly discount factor for two subsequent future periods is close to values usually assumed for exponential discounters in the literature, with a value of 0.92. This result implies that women do not tend to be myopic per se since even utilities 20 years away receive an overall weight of 0.14 in the lifetime utility. Instead, it is that instantaneous utility is valued much more than any future utility, thus providing evidence that women postpone their re-entry into the labor market, by overvaluing current leisure over future career advancements. The third chapter concentrates on another crucial aspect of dynamic choice, expectations about future states of the world. It represents one of the first analyses in the literature on female labor supply that deviates from the rational expectations hypothesis. This hypothesis assumes that individuals, on average, correctly predict future events. After presenting some suggestive evidence that economic agents systematically mispredict their future employment opportunities, I develop a structural life-cycle model of labor supply, human capital accumulation, and labor market frictions. Nesting the rational expectations framework, the model allows job offer expectations to deviate from their real probability. In this chapter, I develop an identification strategy to recover expectations within the model that allows estimating the key parameters from observed labor supply choices. Estimations provide evidence that mothers strongly overestimate their chances on the labor market since they expect the half-yearly job arrival rate to be 66% higher than the actual rate. The structural framework allows for quantifying the costs stemming from these biased beliefs. The results are striking. Overestimating future employment opportunities prolongs career breaks on average between five and eight months, depending on the length of employment protection. In total, the bias causes a reduction of lifetime earnings from employment, as measured at the first birth of a child, between 12% and 18%. Various benefits and the earnings of a potential husband mitigate some of these losses, causing consumption losses to only range from 3.1 % to 4.1 %.