On the fiscal policy implications of low capital mobility: some further evidence from cross-country, time-series data.

• Author: Dar, Atul A.

1. Introduction

   In an attempt to determine whether an increase in a nation's private saving, stimulated (say) by tax policies, raises domestic investment or flows abroad, Feldstein and Horioka [4] and Feldstein [2] examined the relationship between the saving and investment rates of a sample of industrialized countries. Their results pointed to the existence of a strong, almost one-to-one relationship between investment and saving rates. They interpreted that finding as implying that the assumption of perfect capital mobility is invalid in that ". . . sustained increases in domestic saving rates induce approximately equal increases in domestic investment rates" [2, 150]. The finding has far-reaching implications. For instance, with low international capital mobility, a country's growth prospects are tied to its saving effort. If the saving effort is weak relative to investment opportunities, realized investment rates could fall short of their potential, and thus impact adversely on growth. It also implies that fiscal deficits lead to neoclassical crowding out of investment, so that both the Keynesian view as well as the Ricardian Equivalence proposition do not hold.

   Since the Feldstein and Horioka (henceforth denoted by FH) article, numerous papers have offered alternative explanations for the strong saving-investment correlations. Many of those explanations center on supposed weaknesses of the econometric methodology employed by FH and Feldstein. One of the arguments, initially proposed by Fieleke [5] and Tobin [19], is the policy-response argument. According to that argument, if the government adjusts its policy instruments to offset current account imbalances, a strong positive correlation between saving and investment would result.(1) However, that correlation reflects the policy response and not low capital mobility. In a recent paper, Summers [16] proposed a direct test of that argument and his empirical results appeared to validate it. On the other hand, Feldstein and Bacchetta [3], henceforth denoted by FB, have questioned the Summers test and his results. They propose an alternative model and their results indicate that the FH finding is quite robust and the policy-response argument is not supported by the data.

   In this study, our aim is to re-assess the FB model, using cross-country time series data for the G-7 countries. Our re-assessment involves two major innovations. First, we make a more efficient use of the data than attempted by both FB and Summers. Second, we adopt a varying-coefficients approach which is more general and deals properly with the problem of econometric endogeneity, and can be viewed as a more accurate statement of stochastic laws by which saving, investment and deficits might be related. That approach is also appropriate when there are outlying observations, arising due to substantial cross-country differences in macroeconomic experience. Section II reviews, briefly, the empirical literature. In section III, we present the model, the data set and the empirical results and discuss them in relation to previous findings. Section IV presents a summary of our major findings.

2. Saving-Investment Correlations and Capital Mobility

   FH initially proposed a regression of the investment rate on the saving rate to test for the mobility of international capital. That is,

   [I.sub.it] = [[Beta].sub.1] + [[Beta].sub.2][S.sub.it] + [u.sub.it] (1)

   where I and S denote five-year average ratios of the investment and saving to gross domestic product (GDP) for the ith country in time period t, and the u's are the random disturbances. They found an estimate of [[Beta].sub.2] not significantly different from unity, implying that all incremental saving within the country would remain there to finance its own investment. Capital, in this case would be perfectly immobile.

   Most studies on the capital mobility issue use equation (1) or some variant of it. The FH finding was supported by a subsequent study by Feldstein [2] but challenged by a pair of studies by Sachs [13; 14]. In his studies, Sachs examined the behavior of OECD current account balances during the 1970s and concluded that the long term behavior of current account balances could largely be explained in terms of investment shocks. Penati and Dooley [10] reviewed Sachs' findings by using more comprehensive data sets and concluded they were not robust. Further, their study, along with a subsequent study by Dooley, Frankel and Mathieson [1] supported the FH and Feldstein findings that long-term variations in the current account ratios are independent of variations in saving ratios. In other words, investment ratios are largely explained by saving ratios. It needs to be emphasized that the question of how mobile capital is internationally is primarily an issue of how close rates of return on physical capital are, across countries. That is, are net flows of capital (both in the form of portfolio and direct investment) sufficiently interest-sensitive so as to equalize international returns to physical capital (the case of perfect capital mobility)? Thus, it does not really matter too much whether capital inflows involve purchases of bonds and other debt instruments or of alternative financial instruments (equities) as a result of portfolio or direct investment decisions. Capital would be highly mobile internationally (that is, [[Beta].sub.2] in equation (1) would be close to zero) if such flows were sensitive to international differences in the returns to physical capital. At the same time, it must be cautioned that the apparent perfect mobility of financial capital (given the high level of integration of financial markets for a range of financial instruments, at least among OECD countries) is necessary but not sufficient for the equalization of international returns on physical capital, as long as some non-tradeable assets are not perfect substitutes for those that are. (See Penati and Dooley [10], Dooley, Frankel and Mathieson [1] and Frankel [6].) Political and exchange risk as well as non zero expectations of exchange rate depreciation/appreciation can result in less than perfect capital mobility in the sense defined above. Thus, as argued by Frankel, if capital was less than perfectly mobile across nations, it is difficult to see how truly exogenous, long-run shifts in saving ratios would not impact on investment ratios, whatever the econometric problems associated with measuring those impacts.

   In a large number of studies, much attention has focused on the econometric problem of endogeneity of the saving rate. Thus, investment and saving rates could be correlated, even in the presence of high capital mobility if there were common factors affecting both investment and saving rates. Obstfeld [9] argued that the FH results could well reflect the correlation between both the investment and saving rates and the rate of economic growth and income distribution variables, neither of which were considered by FH. An alternative form of the endogeneity argument, attempting to explain the FH and Feldstein results, is the country size arguments of Harberger [7] and Murphy [8]. Essentially, if, for instance, a large country's saving falls, world interest rates rise and depress investment in all countries. This produces a positive correlation between saving and investment rates. However, that correlation cannot be attributed to low capital mobility. These criticisms notwithstanding, it is interesting to note that FH did, in fact, attempt to deal with endogeneity by using the instrumental variables (IV) method of estimation. Indeed, as Frankel [6] notes, the original FH study is more sound in its choice of instruments than some of its critiques. In the FH study, the use of IV method of estimation, does not reduce the strong positive saving-investment relationship found by standard estimation methods.

   The Fieleke and Tobin policy response argument, alluded to above, is that the observed investment-saving correlations actually reflect the policy reactions of the government. This...