The Determinants of Japanese Capital Flows in the 1980s

We saw that the movement of Japanese long-term capital flows in the 1980s and early 1990s has remained a mystery according to standard explanations.1 Indeed, very few researchers even tackled the problem. One of the few is Ueda (1990), whose regression has a disappointing fit and is not subjected to the standard tests. The only paper that successfully explains Japanese long-term capital flows in the 1980s and early 1990s has done so by moving beyond the traditional portfolio models by incorporating, more or less ad hoc, a variable that represents the Japanese asset price bubble: in a portfolio model of capital flows, of the Kouri and Porter (1974) type, price variables, such as interest rates, had little explanatory power, while land-related credit creation, which was suspected to be fuelling the land price boom in Japan, was strongly significant (Werner, 1994a). However, this attempt to reconcile traditional models with reality did so with great difficulty: the key variable of the portfolio model is not a price, but a quantity. Moreover, this quantity — land-related credit — is not a private sector asset, but a liability. This raises questions about the applicability of the portfolio model approach altogether, which so far has been the theoretical underpinning of capital flow studies.