Home Bias and High Turnover in an Overlapping Generations Model with Learning

This paper develops a two-country OLG model under the assumption that investors are on a Bayesian learning path. While investors from both countries receive identical information ßows, domestic investors start off with less precise prior beliefs concerning foreign fundamentals. On a learning path, differences in beliefs and estimation risk generate portfolio biases similar to those observed empirically: home bias in equity portfolios and trend-chasing in international ßows. In addition, due to the higher volatility of the estimates of foreign state variables, our model produces excessive turnover in foreign securities as reported by Tesar and Werner (1995). We use real GDP data for the US and Europe to calibrate the model and produce simulations that show that under the assumption of a Þnancial liberalization during the 1970s, substantial home bias and excess turnover should have been observed in the subsequent years.