A Dynamic Theory of the Feldstein–Horioka Puzzle: A New Measure of Home Country Bias

Abstract

Abstract The recent home country bias is primarily related to the home bias of equities as measured by the international Capital Asset Pricing Model (CAPM). This measure has declined over the past two decades amid financial globalization but remains high in most developed countries. The key to understanding this puzzling phenomenon lies in how accurately this indicator can be predicted. For example, a major drawback of the equity home bias indicator is that it cannot be applied to portfolios that span a broad menu of assets, including stocks, bonds, and bank loans. We propose an aggregate home bias indicator associated with savings retention rate (called "beta"), which is estimated by a Feldstein-Horioka regression. Note that domestic savings consist of a variety of assets. However, a drawback of the beta is that it is subject to variety of estimation biases. To avoid this problem, we re-estimate beta based on the saddle-path dynamics of investment under convex adjustment costs based on Tobin's q theory, an approach quite different from previous studies related to the Feldstein-Horioka puzzle (FHP). In doing so, we applied dynamic panel estimation to measure the impact of changes in domestic savings on investment behavior. The main empirical results are as follows. Home bias declined steadily from 0.57 to 0.32 through 2000, and the correlation between investment and savings disappeared from 2000 to 2008, but recovered to 0.5, the average of the 40-year home bias measure, after the 2008 financial crisis. Interestingly, after the financial crisis, people expected the home bias to be very high, but in fact it simply returned to the average level. The paper will focus only on the precise estimation of β and leave the causes of this phenomenon to other studies. JEL: C23, F21, F32