We construct a dynamic model of sequential schooling decisions of a risk neutral income maximizing agent. We evaluate his optimal behavior and we show how this is related to the ex ante return of schooling choices. The model sheds light on how 'option' (continuation) values are key determinants of schooling acquisition and how they arise from learning and nonlinearity of reward functions at different stages of the life cycle. The sources of the 'option' values are the opportunities of accessing even greater schooling granted by achieving a given schooling level. We estimate the model using lifecycle data and we assess empirically the role of heterogeneity and state specific learning in affecting expected and realized true returns and the effective costs of schooling. We compare the true returns implied by the model with what is usually calculated in the literature such as the internal rate of return (IRR) and the `Mincer' coefficient and we show how those fail in capturing quantities that determine agents' schooling decisions. We formally establish the semiparametric identifiability of our model and generalize our proof to a class of structural dynamic discrete choice models for stopping times and associated outcomes in which agents sequentially update the information on which they act.
Key Personnel: Professor James Heckman of the University of Chicago, Philipp Eisenhauer of the University of Mannheim, and Stefano Mosso of the University of Chicago.