ESSAYS ON LIFE CYCLE DYNASTIC DISCRETECHOICE MODELS

Models of dynastic households have been traditionally used to analyze persistence in earnings and wealth across generations, more recently to study patterns of wealth and fertility, transfers to children and education choices. However most of those models have looked at the theoretical outcomes and there are some limited calibration studies. Some other literature follows the regression based techniques to answer empirical questions regarding the generational transfers. In Chapter 2 -co-authored with Gayle and Golan- we develop an estimator that makes the structural estimation of dynastic models feasible. We propose an estimation framework for dynastic models which allows the estimation of the problem in several steps. Our estimator compared to the full solution structural estimation known as the Nested Fixed Point Algorithm (NFXP) performs comparable in small samples while reducing the computation time considerably. A Monte Carlo exercise compares our estimator to the NFXP. We show that the alternative representation of the continuation value of the problem enables us to apply the Hotz and Miller (1993) estimation to the dynastic problem.Using data of two generations from the PSID, Chapter 3 estimates a dynastic life-cycle model with endogenous fertility, labor supply and inter-generational transfers. This chapter uses data on time spent with children and measures outcomes in terms of education. Education and skills both affect the children's earnings and marriage market outcomes stochastically. We contribute to the literature by measuring the returns in a life-cycle dynastic model in which fertility and time spent with children is endogenous and the different aspects of returns to investment (i.e. education and skill) in children are aggregated and measured in terms of their life-time utility. We model couples decisions as a noncooperative game and solve for a Markov Perfect Equilibrium (MPE) in pure strategies. Therefore the valuation functions of the dynastic model are not only the optimal solution to the problem given the state variables for the individual, but they are the best response valuation functions given the spouse's choice. This requires an equilibrium choice which we assume as MPE in pure strategies.