Learning and investment in human capital: Empirical identification of social learning in school investments

I examine human capital investments under imperfect information by empirically identifying the role of localized social learning in schooling investments. This paper provides a framework for explaining variations in the speed of human capital accumulation after a change in schooling returns, as well as for quantifying the cost of imperfect information in transition. For this purpose, a Baysian learning model is formulated to identify the effects of some salient features of low-income rural economies on the speed of agents' learning. The estimates of learning-investment rule, using farm household panel data from India at the onset of the return-augmenting Green Revolution, are consistent with our framework. More specifically, this paper shows that (i) agents learn about schooling returns from the income realizations of their neighbors and about their ability from their parents', which reinforces the actual return differential, (ii) the schooling distribution of the parents' generation in a community affects agents' learning speed and schooling investments, and (iii) income volatility slows learning speed and reduces investments. Simulations based on the parameter estimates of the decision rule illustrate the dynamic changes in enrollment rate and average household income, and uncover the quantitative importance of the initial schooling distribution and income volatility as the determinants of transition dynamics of human capital accumulation.