تحرک سرمایه در صرفه جویی و سرمایه گذاری: از روش ضرایب متغیر با زمان

This paper uses a model with time-varying coefficients in order to track changes in Feldstein–Horioka saving-retention coefficients over time. To the extent that such coefficients measure international capital mobility, the main empirical findings are as follows. First, the stability of the saving-retention coefficient is strongly rejected. Second, capital has long been perfectly mobile in Canada. Third, capital mobility has never been high in the United States. Fourth, capital was more mobile in Japan and the United Kingdom at the turn of the 20th century than it has been during the postwar period. Capital mobility has risen in Argentina, Italy and Sweden since around 1970. Finally, capital mobility for most of the countries considered has not monotonically increased during the postwar period.

A large empirical literature has investigated how mobile capital is between countries. Several methods have been proposed to measure the degree of capital mobility. One such way – suggested by Feldstein and Horioka (1980) (hereafter, FH) – is to estimate how strongly saving and investment are related either across countries at a given time or over time in a given country. According to FH, if capital is perfectly mobile, investors care only about the rate of return on their investments and not about which country they invest in. This means that domestic saving need not be related to domestic investment under perfect international capital mobility. Indeed, each country's domestic saving responds to and seeks out opportunities for investment everywhere in the world while its domestic investment is financed by the worldwide pool of capital. By contrast, if incremental saving does tend to be invested in the country of origin, FH argued that domestic savers must not be able to readily avail themselves of all investment opportunities in the rest of the world; i.e., capital must be imperfectly mobile internationally. Moreover, the greater is this tendency of domestic saving to flow only into domestic investment, the less mobile capital is likely to be. On the basis of this rationale, FH regressed domestic investment on domestic saving for cross-sectional samples of OECD countries in order to assess how mobile capital was among them. They found that the estimated regression coefficients, which they termed “saving-retention coefficients,” were all close to one. This finding, that most of incremental saving tended to remain in the country of origin, was surprising since it suggested that capital was closer to being completely immobile than perfectly mobile internationally. FH's seminal study stimulated a large literature confirming high saving-retention coefficients; see Coakley et al. (1998) for a review of this literature. This paper reexamines the relationship between domestic investment and domestic saving by investigating how it has evolved over time and by paying special attention to the time-series properties of the data. If they are nonstationary as one would expect, FH's conventional approach could yield misleading estimates because of the spurious-regression problem. We formulate an appropriate way to perform statistical inference on saving-retention coefficients when the data are nonstationary. In order to test the stability of the coefficients, we apply two tests: Hansen's (1992) LM-type tests and Park's (1990) variable-addition approach. We estimate the Time-Varying Coefficients Cointegration Model (hereafter, TVC) with polynomial and trigonometric functions. This recently developed estimation method can provide an especially fruitful way to measure the variation in international capital mobility over time. Our main empirical findings are as follows. First, domestic saving and investment rates are well approximated as difference stationary. Second, the stability of savings-retention coefficients is strongly rejected. Third, capital appears to have long been perfectly mobile in Canada. Fourth, capital was more mobile before the World War I than it has been in recent years. Fifth, the degree of capital mobility does not show a monotonically increasing trend during the postwar period. Section 2 briefly surveys the FH literature, focusing on cointegration approaches. Section 3 introduces statistical inference methods for time-varying coefficient models with I(1) variables, and Section 4 reports the findings of our empirical analysis. It also compares our findings with those reported in the literature.

This paper estimates Feldstein–Horioka saving-retention coefficients by using methods that allow domestic saving and investment rates to be cointegrated and the saving-retention coefficient to vary over time. The changes in the saving-retention coefficients indicate how international capital mobility has changed. We obtained the following empirical findings. First, the parameter stability of saving-retention coefficient is strongly rejected. Second, capital mobility from Canada to the rest of the world appears to have long been perfect: its path for domestic investment is basically disconnected from its path for domestic saving. The result that cointegration is rejected for Canada does not indicate that Canada violates her long-run budget constraint, however. When we apply a unit-root test that permits nonlinear mean reversion in the ratio of its current-account surplus to its GDP, we find evidence of nonlinear mean reversion. Nonlinear mean reversion implies that the Canada's long-run budget constraint is indeed satisfied. Third, at the turn of the 20th century, capital appears to have been much more mobile between both Japan and the United Kingdom and the rest of world than it has been in the postwar period. Fourth, capital appears never to have been especially mobile between the United States and the rest of the world. Fifth, the capital mobility into and out of Argentina, Italy and Sweden has risen since around 1970. Finally, the capital mobility of the countries that we consider appears not to have increased monotonically in the postwar period. Our findings therefore confirm those of Taylor (1996).