IN COMPARISON with the analysis of trade flows, advances in the quantitative study of capital movements have been modest. Empirical treatment of the foreign sector in national econometric models has generally been confined to the current account, and capital movements, if they are not completely disregarded in these models, have been considered exogenous. What little empirical work has been done on capital movements has centered on the United States and Canada and has been largely exploratory. The difficulties of a weak theoretical framework are aggravated by the econometric and statistical problems that are often encountered in the testing of hypotheses relating to capital movements.
Empirical analyses of long-term capital movements are concerned mainly with factors that determine the demand for foreign fixed and financial assets and assume that there are unlimited supplies both of funds financing the investment and of financial assets. Since there is, however, evidence of influences on supply, capital movements may reflect the combined effect of demand and supply factors. To the extent that these factors are mutually reinforcing or offsetting in their combined effect and change independently of one another, estimates from either demand or supply models are likely to be unreliable and forecasts from such models may be subject to substantial error.
Despite the uncertainty surrounding the results of quantitative analysis, recent empirical work has shown that some factors are generally significant determinants of capital movements. This study attempts to summarize these findings, dealing first with foreign direct investment and thereafter with long-term portfolio flows.
I. Foreign Direct Investment
The theory of foreign direct investment has closely followed developments in the theory of domestic investment. The traditional view that investment would be forthcoming as long as the return on capital outlays exceeded the interest rate was also applied to capital movements between countries. Differences in interest rates were assumed to reflect differences in the marginal efficiency of capital and were supposed to lead to the flow of both portfolio and direct investment capital.1 More recently, the traditional rationale for domestic investment has been qualified by the notion that investment outlays are related to output and sales. This notion has also entered the current theory of foreign direct investment; thus, the relationship between direct investment and output in the country receiving the investment has been examined within the framework of accelerator models and neoclassical models as developed by Jorgenson . In contrast to simple models relating direct investment to output, these models specify the relationship between direct investment and the foreign market in such a way as to take account both of lags in the reaction of capacity to changes in output2 and of the possibility of a correction in the current level of direct investment to compensate for past deviations from a desired level.3
Its determinants in summary
The descriptive literature on the determinants of direct investment maintains that direct investment is affected by such factors as the yield on investment, market and political situations, tariffs, taxation, and exports to the country receiving the investment.4 The general conclusion is that the traditional theory based exclusively on yields is inadequate and that the potential demand for the output of the sector receiving the direct investment, as reflected in the size of the host country’s market, is its principal determinant.5
Results of quantitative analyses support this conclusion by consistently pointing to scale factors, such as income and expenditure on fixed investment in the host country, as significant determinants of direct investment. The quantitative evidence as to the influence of interest rates on direct investment appears contradictory. It seems, however, that while differences between domestic and foreign rates of return may have an uncertain effect on direct investment in individual foreign countries, they do contribute to an explanation of the total of direct investment abroad. The contention found in the descriptive literature—that direct investment is insensitive to differentials in interest rates—cannot, therefore, be upheld when the analysis is highly aggregative, as, for example, in capital account models of the balance of payments.
Quantitative analysis has produced no evidence that factors other than the size of the market and the rate of return—such as tariffs, exports to the country receiving the investment, and differences in rates of inflation—affect investment from the demand side.6 Some studies that examined direct investment from the supply side have found the supply of funds financing direct investment to be related to economic activity in the investing country. There is, in addition, some evidence that firms are not indifferent to the form in which direct investment is financed, i.e., debt or equity.
Governments of both investor and host countries occasionally adopt measures to influence the flow of direct investment. Evidence on the effectiveness of such action is very scarce and relates almost exclusively to the U.S. balance of payments programs of the late 1960s. While these programs have had no apparent effect on the total of U.S. direct investment abroad, if one includes the retention of profits by subsidiary companies abroad, they do seem to have reduced the amount of direct investment outflow that enters the balance of payments.
The dependent variable in direct investment models
Direct investment, owing to its complex nature, is not always specified in the same way in different empirical studies. It is either new or expansionary and does not refer to a single homogeneous flow but represents the sum of different capital movements, which are composed of flows entering the balance of payments and of the retention of earnings by subsidiary companies abroad. It includes purchases of fixed capital, imports of investment goods, the acquisition of existing assets, and funds for other business transactions. Definitions of direct investment refer to equity in certain kinds of enterprises abroad.7 Different studies have dealt with different components and different types of investment, examining its behavior, for example, only in the context of the balance of payments8 or only in the context of capital formation in certain sectors of the host economy.9 These differences in specification limit the usefulness of comparisons of these studies.10
The market-size hypothesis
Irrespective of differences in specification of the dependent variable and of differences in the level of aggregation,11 the proxy for market size is found to be a consistently significant independent variable. This is not true for direct investment in those developing countries where the output produced with the help of foreign capital is exported rather than sold in the local market. In these instances foreign investment is more likely to be drawn by the opportunity to exploit resources.12 Where the exploitation of raw materials predominates, direct investment is supply oriented in developing countries as opposed to demand oriented in developed countries. Thus it can be assumed that, in a sample containing both developing and developed countries, a statistically significant relationship between scale factors and foreign investment indicates a preponderance of “demand cases” over “supply cases.” 13
The rate-of-return hypothesis
The consistent significance of the scale factors in “demand” cases is explained in terms of the investing firm’s aim of profit maximization. Yet profit maximization has been considered a determinant of direct investment throughout the course of its theoretical interpretation, so that the shift in the stress of econometric work from rates of return to scale factors does not reflect a fundamental change in theory but rather an improved statistical test of the profit motive. This may mean that profit maximization—particularly over the long run—is more concerned with the volume of sales and expectations of market growth than with rates of return. But it may also be that the wrong rates were chosen to test responses. Thus, it seems unlikely that the total of foreign direct investment decisions of U.S.-owned international firms would respond to rates of return in individual foreign countries or to rate differentials between individual foreign countries and the United States. Such a specification would assume that the only alternative to investing in an individual foreign country would be domestic investment, but the fact that international firms invest in a number of foreign countries makes this assumption implausible. A high rate of return in one of these countries or a high positive rate differential between one of them and the United States may, depending on the elasticity of substitution between alternative foreign investment, induce little or no direct investment flows to that country if rates of return in other countries receiving investment are as high or even higher. The substitution elasticity is likely to be higher, the larger these other countries’ markets are for the goods produced from the investment. Insofar as direct investment in any country responds to rates of return, the response should be determined by the rate of return in the investing country and the rates of return in all countries that are potential recipients. One cannot conclude from the apparent absence of any relationship between direct investment in individual foreign countries and rates of return in these countries,14 or rate differentials between the investing country and the individual countries receiving investment,15 that direct investment is generally insensitive to rates of return. Indeed, there is evidence to the contrary. Kwack  has shown that total U.S. direct investment abroad responds to the differential between the U.S. long-term interest rate and a weighted average of foreign rates, where the weights relate to the size of markets in the countries receiving investment. Such a weighted rate takes account, indirectly, of differences in the elasticity of substitution of U.S. direct investment with respect to rate differentials between individual countries receiving investment.
Prachowny  has obtained evidence suggesting that the ratio of foreign to domestic rates of return is a highly significant determinant of the ratio of the stock of U.S. direct investment abroad to the total of U.S. corporate assets; in this case, rates of return are defined as the ratio of after-tax earnings to the stock of investment.
Apart from disregarding possibilities of substitution, the apparent insensitivity of direct investment to rates of return may be due to imperfect methods of measuring rates of return. There is evidence that rates of return and their most frequently used proxy, interest rates, do not move together over time as closely as might be assumed.16 Thus, while studies that take account of the possibilities of substitution and use representative measures of the rate of return show direct investment to be responsive to rates of return, findings to the contrary are less convincing when the possibility of substitution between assets of different countries receiving investment is not considered or when the rate of return is inaccurately measured.
Hypotheses concerning other determinants of direct investment on the “demand” side
Measures of the profitability of direct investment other than rates of return and the market size in the country receiving investment have included the earnings of companies investing abroad. Attempts have also been made to trace the impact on direct investment of such factors as wage differentials and differential rates of inflation, which have been specifically linked to the profitability of direct investment. It has not been shown that direct investment flows are affected by earnings, or wage differentials, or differential rates of inflation. Popkin  found no evidence that relatively lower foreign wages induced U.S. companies to invest abroad. Bandera and White  and Moose  found no apparent association between earnings and direct investment, and Weintraub  concluded that differences between the rates of inflation in investing countries and countries receiving investment had no effect on the flow of direct investment.17 Since Popkin  has found a highly significant relationship to exist between the ratio of foreign to domestic investment by U.S. manufacturing companies and the ratio of foreign to domestic profits, and since the apparent dependence of variations in direct investment on the size of the market and the rates of return implies a significant association between direct investment and profit opportunities, the failure of these other factors to explain direct investment would suggest that they inadequately measure profit opportunities rather than provide evidence that the maximization of profits is not a determinant of investment.
A number of studies have attempted to assess the influence of factors that have frequently been associated with variations in direct investment but are not directly related to profits, namely, such factors as tariff discrimination, exports to the country receiving investment, and the level of international reserves of the country receiving investment.
High tariff walls have been expected to encourage direct investment, as companies trying to circumvent them would produce abroad goods that were previously exported. Exports can lead to direct investment by familiarizing companies with certain foreign markets and revealing opportunities for investment in these markets; and the level of international reserves in the country receiving investment could affect direct investment to the extent that investors consider it a measure of the probability that profits may be repatriated without significant loss in the exchanges. While such factors may occasionally help to explain individual investment decisions, there is little evidence that they play a general role in the movement of direct investment capital. Scaperlanda and Mauer  found no evidence that the common tariff of the European Economic Community and removal of intra-Community tariff barriers had led to U.S. direct investment within the Community.18Popkin  found no statistical support for the export hypothesis and Bandera and White  found only two exceptions to the general inability of the reserves hypothesis to explain variations in direct investment. These two exceptions refer to U.S. investment in the French manufacturing and trade sectors and in the Italian petroleum and trade sectors. The authors attribute the statistical significance of the international liquidity variable in these cases to the fall in reserves experienced by France in the mid-1950s and by Italy in 1963, and argue that foreign investors tend to react to a decline rather than to a rise in the reserves of countries receiving investment.
Direct investment explained by the supply of funds
The bulk of the literature on direct investment deals with factors affecting direct investment from the demand side with the implication that the supply of funds financing the investment is infinitely elastic. However, in instances where an explanation of direct investment from the supply side is attempted, it appears that the amount of funds supplied is related to economic activity in the country investing abroad. Depending on whether investment abroad complements domestic investment or substitutes for it, this supply relationship is positive or negative. The United Kingdom and the United States provide an example of such contrasting supply relationships. Krainer  has demonstrated that British investment abroad, which is predominantly investment in the exploitation of raw materials, rises and falls with domestic output, while U.S. investment abroad, which is mostly in manufacturing, rises when domestic economic activity slows down and falls when the domestic economy expands rapidly. Apart from a decline in domestic investment opportunities because of spare capacity, a slowdown in the domestic economy is likely to cause a decline in the rate of return of domestic investment. In such a situation, foreign investment that substitutes for domestic investment will increase for two reasons. First, the decline in domestic investment opportunities will initially raise the amount of funds available for investment abroad. Second, a decline in the rate of return on domestic investment will raise the relative profitability of foreign investment.19 Yet, as the economy continues to slow down, the total amount of funds available for both domestic and foreign investment will decline—initially further reducing domestic investment but eventually also reducing investment abroad. At some point, the increase in the supply of funds for investment abroad owing to the decline in domestic investment opportunities and the drop in the relative rate of return domestically will be more than offset by the decrease in the supply of funds for investment abroad owing to the reduction in the total amount of funds available for both domestic and foreign investment.20 At this point, therefore, direct investment will become positively related to economic activity in the country investing abroad. It appears from an analysis of U.S. capital exports carried out by Rhomberg  that a slowdown in the economic activity in the United States does in fact cause a reduction in U.S. direct investment about a year later, and that, similarly, a rise in the economic activity of the United States will, after about a year, lead to an increase in U.S. direct investment.21 The relationship between domestic economic activity and direct investment abroad that complements domestic investment, instead of substituting for it, is less complex; it will always be positive.
The differences in response of direct investment to domestic economic activity, as exemplified by the U.S. and the British experience, will also be reflected in the balance of payments. An increase in British investment abroad owing to a rise in domestic economic activity will tend to increase a deficit, or reduce a surplus, in the balance of payments of the United Kingdom; a decrease in foreign investment owing to a decline in economic activity will tend to increase a surplus, or reduce a deficit. In the United States, the initially negative relationship between output and investment abroad will tend to put the balance of payments into surplus at the beginning of recovery and into deficit at the beginning of a recession. Since, however, the relationship between output and U.S. investment abroad changes over time, an upturn in activity will eventually increase a deficit, or reduce a surplus, in the balance of payments, while the opposite would apply for a downturn.
At any point in time, total funds available for investment abroad are composed of retained earnings, funds obtained from issuing stock, and funds obtained from borrowing domestically and/or abroad. If companies investing abroad prefer to finance their foreign investments with certain funds, the investments will depend on the availability of funds in the preferred categories.
It has been argued that companies are indifferent to alternative sources of investment funds because costs would be independent whether financing is secured by debt or by equity.22 Empirical results have not sustained this argument. Severn  found a negative relationship between U.S. expenditure on foreign fixed assets and a lagged ratio of debt to equity. Such a finding seems to suggest that companies are not willing to resort increasingly to debt financing of foreign investment expenditures.
The effect of government intervention
The total flow of direct investment capital, as determined by factors that attract investment and constraints on the supply and composition of funds available for investment, can be affected by government policies. Empirical evidence on the effectiveness of such policies as have been applied is particularly scarce. Kwack  has shown that the recent balance of payments programs of the United States have reduced the outflow of U.S. direct investment that enters the U.S. balance of payments. Results obtained by Stevens  support Kwack’s findings but show in addition that, because of the heavy foreign borrowings of parent firms and foreign subsidiaries, the balance of payments programs left the total of U.S. direct investment abroad unaffected. The Canadian Government’s policy of increased taxation on investment income has reduced the rate of return on foreign direct investment in Canada. Yet no evidence could be found in an econometric study of the Canadian economy conducted by Officer  that this tax measure had had an effect on the balance of payments component of the flow of direct investment into Canada.
The review of the quantitative evidence on the behavior of foreign direct investment suggests that the desire to invest abroad is primarily affected by the size of the foreign market. While this merely corroborates the conclusions of descriptive literature, further quantitative evidence has revived traditional views stressing the impact on direct investment of differential rates of return, which recent descriptive literature has considered to be of lesser importance. The significance of differential rates of return is inferred from evidence obtained from highly aggregative analyses of direct investment and not from studies examining direct investment on the basis of geographical distribution, which does not take account of possibilities of substitution. It may therefore be preferable to base any investigation into the causes of direct investment flows on a variable that is specified in a way that reflects the unity of the investment decision of the international firm. This unity implies that the foreign investment decisions of a firm are interdependent, a view which disaggregation by countries receiving investment implicitly rejects. Apart from the size of the foreign market and differential rates of return, no other hypotheses relating to the desire to invest abroad could be confirmed empirically. There is considerable evidence that the actual flow of direct investment that is caused by factors affecting the desire to invest abroad is subject to financial constraints as to the absolute amount of funds available for investment as well as the composition of this amount by debt and equity capital.
II. Long-Term Portfolio Capital
In the analysis of portfolio capital a distinction is made between long-term and short-term assets. Any asset with a maturity of less than one year is considered short term, while any asset with a maturity equal to or exceeding one year is considered long term. On the assumption that long-term assets are held until maturity—or at least for one year—developments in the foreign exchange market can be expected to influence the return on long-term capital less than the return on short-term capital. It is this difference in the importance of the foreign exchange market that leads in both theoretical and empirical analysis to the distinction between the two types of asset.
The holding of foreign assets can be explained alternatively in terms of the traditional yield theory and the theory of portfolio choice as developed by Markowitz  and Tobin .23 Since the traditional yield theory requires that, for any foreign assets to be held, foreign interest rates should be higher than domestic interest rates, it cannot account for foreign asset holdings when the differential between foreign and domestic rates is zero or negative. The theory of portfolio choice is instead based on the preferences of the asset holders with respect to return and risk of return, and such zero or negative rate differentials are compatible with foreign asset holdings.
The residents of any country hold wealth in the form of money balances and nonmonetary assets. There is a set of equilibrium interest rates at which asset holders have no desire to change the distribution of wealth between these two components. The community’s portfolio of nonmonetary assets is in equilibrium in the sense that, at these interest rates, the risk from the ownership of wealth is minimized. The reduction of risk to this minimum level is achieved by diversification of asset holdings. In an open economy, such diversification implies that part of the portfolio is held in the form of foreign financial assets.
The theory implies that owners of wealth will hold assets bearing relatively low interest if the inclusion of such assets in their portfolio causes a reduction in total portfolio risk that more than compensates for the sacrifice in gains involved. The size of the reduction in total portfolio risk depends on the degree to which the return on the assets in question is independent of the return on the other assets in the portfolio.24
Since foreign interest rates are independent of domestic interest rates or at least less dependent on them than domestic rates are among themselves, the inclusion of foreign assets in the portfolio will considerably reduce total risk. Thus, if foreign interest rates are higher than domestic interest rates, owners of wealth who hold both foreign and domestic assets will be able to both reduce total portfolio risk and increase return. If foreign interest rates are lower, total portfolio risk will be reduced at the expense of some return. Whether, in this case, owners of wealth will experience a gain in utility owing to a reduction in risk that exceeds their loss owing to a reduction in return, thus raising their total utility, can be inferred from the properties of their indifference curve relating return to risk.
In the state of equilibrium where owners of wealth are satisfied with both the share of money balances in total wealth and the composition of nonmonetary assets in their portfolio, capital movements may result from two elements—one relating to the growth of income and the money supply, and the other to changes in the relative returns on assets held. While these two elements are interrelated, the return on assets can be assumed to be constant if money supply and income grow at the same rate. In such a state of dynamic equilibrium, asset holders will, provided that foreign interest rates remain unchanged, distribute increments in wealth on the basis of the existing portfolio composition of monetary and nonmonetary assets. The movement of capital resulting from such distribution is referred to as flow adjustment, and it continues as the stocks of wealth and money grow. The magnitude of this continuous flow adjustment will depend on the rate of growth of wealth and money and the original share of foreign financial assets in the portfolio of nonmonetary assets of the owners of wealth, which in turn is determined by the level of the differential between foreign and domestic interest rates and the reduction in total portfolio risk achieved by the inclusion of foreign assets in the portfolio.
A movement of capital in response to changes in the differential between foreign and domestic interest rates is referred to as stock adjustment. A change in this interest rate differential will cause portfolio redistribution as owners of wealth move to a new portfolio equilibrium. The consequent flow of capital, which the redistribution process implies, will end as soon as the new equilibrium is reached. The magnitude of the stock adjustment will vary with the size of the portfolio and the amount of change in the interest differential.
The identification of observed capital flows as pure stock adjustment presupposes constant portfolio size. Since, however, with the passage of time the assumption of zero portfolio growth becomes increasingly unrealistic, only those capital flows that immediately follow interest rate changes can be considered predominantly to reflect stock adjustment. Continued capital movements some time after the change in interest rates may still reflect a lagged stock adjustment, but the greater part of the movements will increasingly be ascribable to portfolio growth. Finally, when the new equilibrium has been achieved and the process of portfolio redistribution is completed, observed capital movements will reflect portfolio growth alone.
The relationship between capital movements and both the level of and changes in wealth, return, and risk suggests that, in a world of changing interest rates and growing portfolios, observed capital movements reflect the combined effect of stock and flow adjustments. Since stock adjustment occurs over time at an initially increasing and eventually declining rate—as a new equilibrium is being approached—the share of the stock adjustment component in the total of the observed portfolio capital movement will, independent of the absolute magnitude of the stock adjustment, change over time. If changes in interest rates occur at frequent intervals, inducing repeated portfolio redistribution, observed capital movements may well, even through time, reflect more the effect of stock adjustments than the effect of continuous flows.
A rise in foreign interest rates will lead to an increase in foreign claims—owing to redistribution of the portfolio in favor of foreign assets—at the same time that such claims are also rising with lenders’ portfolios. But stock and flow adjustments need not necessarily be reinforcing in their effects on observed capital movements: a decline in foreign claims, because of a redistribution of the portfolio in favor of domestic assets after a fall in foreign interest rates, may be partly offset by an increase in foreign claims owing to the growth in the lenders’ portfolio.25
Its determinants in summary
Recognition of the fact that observed capital movements may represent the combined effect of stock and flow adjustments has resulted in models that attempt to distinguish between these two types of adjustment. Accordingly, portfolio flows are associated with both levels and changes in interest rates, together with a scale factor. Foreign and domestic interest rates enter current models either individually or as a differential. While interest rates prove statistically significant in either form, there is evidence from models in which portfolio movements are dependent on both foreign and domestic interest rates that results from such models are superior to results from models that use the differential. Depending on whether capital movements are explained in terms of the behavior of lenders or borrowers, the scale factor is a proxy for the volume of the supply of, or demand for, funds. A number of models have tried to assess the influence of factors that, along with nominal interest rates, determine the effective rate of return and cost on foreign long-term financial claims and liabilities, respectively. These factors include expectations with respect to interest and exchange rates, changes in the risk involved in asset holding, and government intervention. Finally, some studies have analyzed long-term capital flows in the framework of a partial adjustment model, i.e., one in which observed capital movements are taken to represent the desire of asset holders to achieve—or at least to approach—within a given time period some desired stock of foreign asset holdings.
Results of quantitative analysis demonstrate that all these factors exert a statistically significant influence on long-term portfolio flows. The evidence on the response to interest rates, scale variables, and government intervention is, with few exceptions, quite consistent. The evidence on the effect of expectations regarding exchange and interest rates, while very scarce and less consistent, does on balance suggest the existence of such an effect. Statistical evidence is also consistent in supporting the partial adjustment hypothesis.
Portfolio capital models in the literature
The quantitative literature on long-term portfolio capital flows can be divided into three main categories, of which the first and the second explain capital flows in the framework of continuous flows 26 and stock adjustment27 models, respectively, and the third tries to assess the joint effects of both flow and stock adjustment.28 Within each of these three categories the behavior of either borrowers 29 or lenders,30 or both,31 is examined. Studies dealing with the behavior of borrower or lender can, in turn, be subdivided into those in which the differential between foreign and domestic rates 32 is taken as the interest variable and those that include both foreign and domestic rates as individual explanatory variables.33 Finally, studies that test the performance of partial-adjustment models may be specifically mentioned.34
The foregoing classification does not exhaustively enumerate differences in existing models of capital movements; in particular, there is substantial differentiation with respect to the dependent variable.
The dependent variable
The flow of long-term portfolio capital is not homogeneous but rather represents the sum of various financial flows that show considerable differences as to source, destination, and behavior. Accordingly, existing models of long-term financial flows specify the dependent variable in a number of different ways. Some consider the net flow of foreign claims and liabilities, while others separate claims from liabilities. Some models are concerned with the total of long-term securities, while others distinguish between them by type and issuer. With respect to type, distinctions are made, on the one hand, between outstanding and newly issued securities and, on the other, between securities denominated in domestic currency and those in foreign currency.35 With respect to issuer, distinctions are made between regional36 and corporate securities.37 Proponents of distinctions by type of security argue that the issue of new securities depends on the initiative of the issuer, while transactions in outstanding securities depend on the initiative of investors. The aggregate flow of newly issued and outstanding securities reflects therefore the influence of factors that determine, on the one hand, the supply of securities and, on the other hand, the demand for them. In this sense, aggregative models may contain both supply and demand elements. The case for distinction by issuer is argued, for example, by Learner and Stern , who maintain that certain issuers, since they operate under short-run constraints, do not—at least in the short run—conform to the hypothesis of equilibrating portfolio adjustment.
Contribution of the interest-rate variable
Interest rates have traditionally been the main independent variable in the explanation of capital movements. They represent the intention of lenders and borrowers, subject to an acceptable degree of risk, to maximize returns or minimize cost, respectively. Their importance as determinants of long-term capital movements has been confirmed by all empirical work, except for the studies of Bell , Cohen , and Zumeta .
While Bell and Cohen found some evidence of sensitivity of U.S. long-term capital flows to the yield on long-term U.S. Government bonds, they concluded that U.S. interest rates determined the timing rather than the volume of transactions in foreign securities. Foreign borrowers postpone their new security issues in the United States for some three to six months, in response to a substantial rise in the level of U.S. rates. The length of the postponement depends on the degree to which the availability of funds in alternative markets is limited. In any event the new issues do take place eventually, and the annual volume of the U.S. portfolio investment is unaffected. These conclusions imply complete inelasticity of substitution of other markets for the U.S. market in response to a rise in the U.S. rate. Such inelasticity may have been plausible during the period covered in the Bell and Cohen analyses—the first quarter of 1957 through the first quarter of 1962—predating the Euro-dollar market. Its subsequent development may be one reason for contradictory findings in later studies. A further reason may be the greater degree of currency convertibility in recent years. Finally, the inclusion (in addition to interest rates) of a scale variable representing supply or demand of loanable funds may improve results,38 and neither Bell nor Cohen has included such a variable in their models. All the points raised in connection with the findings of Bell and Cohen apply also to the results obtained by Zumeta .39
In empirical analysis, portfolio flows are made dependent either on both foreign and domestic rates or on the interest differential alone. In theory, it could be argued that, ceteris paribus, the differential would be the relevant measure of the influence of interest rates. It appears, however, from results reported in the literature that the separate inclusion of foreign and domestic rates is superior to the use of the rate differential, for the following reasons. First, findings of interest sensitivity in U.S. portfolio flows consistently reveal the U.S. rate as a more important determinant of these flows than the foreign rates. (See Branson , Bagueley , Kwack , and Ripley . Branson’s main conclusion is the “powerful impact of changes in the U.S. long-term rate on portfolio capital movements.”) The differential does not capture this aspect because it would show the same change whether the foreign rate fell by a certain amount or the U.S. rate rose by the same amount. Yet if the U.S. rate rose, the decline in U.S. purchases of foreign securities should be larger than if the foreign rate fell.
Thus, the usefulness of the differential to forecasting portfolio flows is quite limited, differentials of the same amount being likely to have different effects. Policymakers should not deduce, from an estimated relationship between portfolio flows and interest rate differentials, that a given rise in the U.S. rate will result in a certain reduction in the outflow of U.S. long-term capital. Such a deduction40 would be valid only if the elasticities of U.S. capital flows with respect to foreign and domestic interest rates were the same, which, on the basis of available evidence, does not appear to be true.
Second, the use of changes in rate differentials does not allow for the possibility of greater inelasticity in capital flows at higher interest rates. In their discussion of the costs of a high U.S. interest rate policy, Willett and Forte  point out that, as a balance of payments instrument, such a policy appears subject to sharply diminishing returns because of the greater inelasticity of stock adjustment at higher interest rates. Ripley  reports a conversation with an executive of a major insurance firm that supports this supposition.41
In aggregative analysis, the use of composite rates to represent the foreign interest rate produces better results than the use of individual rates, which are less representative of the foreign rate structure and do not allow for the possibility of substitution among assets of different origin. Kwack  demonstrates that a weighted foreign rate is a significant determinant of total U.S. long-term claims on foreigners, while Branson , using individual foreign rates as separate explanatory variables, discards most of them as insignificant. In the disaggregation of transactions in foreign securities by type, issuer, and country, the rate on any security is probably best expressed relative to other foreign rates, as long as the elasticity of substitution between such transactions with respect to different foreign rates is high. While transactions in some securities may be so elastic, it appears that such an assumption is not always warranted. Lee , finding evidence that U.S. demand for Canadian securities is independent of the rate of return on U.K. securities, suggests that capital markets are compartmentalized into a number of submarkets, each of which specializes in securities of certain countries. Kwack  has found that the long-term claims of U.S. nonbank institutions on foreigners can be explained by separate dependence on the differential between the U.S. rate and a weighted average of the long-term government bond yields of the United Kingdom, Canada, and Germany and the differential between the U.S. rate and an average of Japanese loan and discount rates. These results also indicate that some institutions react more to yields in the United Kingdom, Canada, and Germany, while others react more to those in Japan.
All studies with the exception of Freedman  examine the behavior of borrowers and lenders separately, and present single-equation estimates of the determinants of either long-term foreign claims or liabilities. Freedman estimates a model of simultaneous equations for both the Canadian supply of external securities to the United States and U.S. holdings of these securities, thus taking into account the behavior of both the Canadian borrower and the U.S. lender. Simultaneous estimation techniques would be appropriate in a situation where capital movements affect interest rates,42 since, for instance—as in Freedman’s model—the rate on Canadian external securities could be expressed in terms of U.S. demand for these securities and the U.S. long-term interest rate.
Contribution of the scale variable
While the interest rate mechanism helps to explain capital movements, an explanation in terms of interest rates alone presupposes an infinitely elastic supply of, and demand for, loanable funds. The additional inclusion of a scale variable avoids this assumption. Its coefficient is likely to affect the coefficient of the interest rate variable and provides information about the magnitudes of capital movements involved. In accordance with flow and stock adjustment hypotheses, the scale variable represents additions to the supply of, or demand for, funds in flow models and the stock of supply or demand in stock adjustment models. For example, the magnitude of lenders’ purchases of foreign securities in response to a change in interest rates will depend on the size of their portfolio, while the magnitude of flows induced by a differential level of foreign and domestic interest rates will depend on the increments to their portfolio. However, because of some overlap, the difference between continuous flows and stock adjustment may not be quite so distinct. If a change in interest rates induces lenders that have never before engaged in transactions in foreign securities to do so, this would be tantamount, in its effect on capital movements, to an increase in the portfolio of lenders that commonly engage in such transactions. Growth in demand and supply of funds is measured by such variables as borrowers’ income, total new securities issues by borrowers, and the portfolio investment or wealth of lenders.43 The stock of supply of or demand for funds is either represented by income, or, in models that relate the foreign share in total long-term claims or liabilities to the interest rate variable,44 implicitly contained in the dependent variable.
The risk factor
Since portfolio selection implies a trade-off of return against risk, a change in the risk attached to the holding of domestic assets will, like a change in interest rates, lead to capital flows. While this is generally recognized, there is only one study (see Miller and Whitman ) that makes explicit allowance for possible effects of a change in the risk structure. The reason for this scarcity of work may be twofold: it is not obvious how best to measure risk, and a given fall in the risk attached to domestic assets will not necessarily lead to the substitution of domestic for foreign assets in lenders’ portfolios. Such substitution might cause an increase in total portfolio risk exceeding the reduction implied by the fall in the risk attached to domestic assets, because total portfolio risk depends not only on the risk attached to individual assets but also on the degree of risk independence among assets.
Miller and Whitman  relate changes in risk to the deviations of income from its long-run trend. They argue that U.S. investors will find the risk attached to the holding of domestic assets to be lower in times of boom than in times of a domestic slowdown, because domestic corporate bonds and loans then have a lower default risk and the share of total loans going to “safe” customers is larger during the boom than during a slowdown. The authors’ hypothesis appears to be confirmed by the significant contribution made by both the lagged value of income and a linear trend variable to the explanation of U.S. long-term foreign claims.
Expectations regarding exchange and interest rates
Portfolio investment decisions concerning foreign long-term securities denominated in foreign currency may be affected by expectations of exchange rate changes. The possibility of devaluation for a certain foreign currency would increase the risk implied in holding assets denominated in that currency, and actual devaluation would lower the effective return on the asset. As a result, lenders might invest in assets denominated in other currencies. Two of the studies that include a measure of exchange rate expectations in their models find a statistically significant relationship between such a measure and portfolio flows. Lee  found significant performance on the part of a dummy variable, included to capture any increase in U.S. investors’ holdings of Canadian securities owing to speculation on an appreciation of the Canadian dollar after 1958. Miller and Whitman  obtained statistical evidence to support the hypothesis that a deficit in the U.S. balance of payments would raise U.S. purchases of foreign securities while a surplus would lower them.45 In contrast, Branson  and Helleiner  found no evidence that exchange rate expectations represented by actual spot rates had any effect on the behavior of either U.S. lenders or Canadian borrowers, and Freedman  found that between 1961 and 1962 Canadian borrowers also appeared unaffected by their Government’s declared intention to depreciate.
Expectations regarding interest rates enter existing capital models to the extent that actual rates are considered to be proxies for expected rates. Beyond this, explicit allowance for the effect of interest rate expectations has been made by Bagueley . The differential between Canadian long-term and short-term rates that he used as the measure of interest rate expectations showed a negative sign as hypothesized and proved a significant determinant of net portfolio inflows into Canada, implying that Canadian net long-term liabilities to foreigners would rise whenever the domestic short-term rates were higher than the long-term rate. The successful performance of the variable may be interpreted to confirm the hypothesis that long-term rates are a positive function of short-term rates.46
The effects of government intervention
In the context of long-term capital models for the United States and Canada, effects of government intervention relate, on the one hand, to the Interest Equalization Tax (IET) and subsequent capital controls imposed by the United States and, on the other hand, to Canadian taxation of interest payments on foreign liabilities. In the assessment of the effectiveness of the measures taken by the United States, two phases can be distinguished. The first starts with the imposition of the IET in the latter part of 1963, and the second with the extension of the tax to long-term nonbank claims on foreigners and the initiation of the voluntary restraint program in the first half of 1965. Empirical evidence on the effectiveness of the measures 47 demonstrates that since the end of 1963 total U.S. purchases of foreign long-term securities have indeed been lower than they would have been in the absence of the measures. During the third quarter of 1963 and the first quarter of 1964, there is even some indication of an initial overreaction on the part of U.S. investors to the introduction of the IET (see Prachowny . The particular situation with respect to Canadian securities is, however, different from the aggregate. While the restrictions of 1965 have apparently also reduced U.S. purchases of Canadian securities (see Helliwell, Officer, Shapiro, and Stewart ), the measures of 1963 seem to have led to a rise in U.S. holdings of Canadian securities. Lee  explains this by the exemption from the IET of new Canadian security issues, in the United States, and by the additional exemption of these securities from the Canadian withholding tax. No evidence of a reduction of U.S. claims on Canada in 1965 was found, in spite of the 1965 agreement between Canada and the United States to defer deliveries of new issues of securities sold to U.S. residents.
Capital models and the partial-adjustment hypothesis
Some existing capital models rely on the assumption that the flow of long-term capital during any period of time represents the adjustment, toward some desired level, of the capital stock recorded at the beginning of the period. Since the desired capital stock is in turn taken to be a function of the explanatory variables discussed above, the partial-adjustment hypothesis is econometrically represented by adding to these variables the lagged value of the actual capital stock. The size of the coefficient of the lagged stock variable indicates the speed of adjustment toward the desired level. A coefficient of 1.0 implies complete adjustment within one period of observation, while a coefficient of zero implies no adjustment whatsoever. Statistical evidence on the performance of the lagged stock variable verifies the value of the partial adjustment hypothesis and suggests that the actual speed of adjustment is rather low. Miller and Whitman  and Kwack  find that only one third of the adjustment of actual to desired level occurs within one quarter of a year, regarding both the share of foreign in total U.S. assets and the stock of foreign assets owned by U.S. nonbank residents. Freedman’s findings  point to even slower adjustment of the share of foreign in total Canadian claims or liabilities, consistently about one fifth per quarter for the provincial, municipal, and corporate sectors.
Empirical findings regarding the principal determinants of the behavior of long-term portfolio capital are in general agreement, and there is evidence to support the hypothesis that observed movements of long-term capital comprise both a continuous-flow and a transitory-stock component of adjustment. All the same, the parameter estimates of existing capital models are subject to considerable contrast and uncertainty. This is in part due to the derivation of most of the estimates from either pure flow or pure stock adjustment models.48 It is also due to such factors as differences in the time-path of response to the explanatory variables and the apparent instability of parameters estimated over different periods of time. The main reason, however, for the lack in consistency of parameter estimates, and their inability to lend themselves well to comparison, resides in substantial differences in specification of dependent and independent variables.
The one exception to the instability of parameter estimates appears to be the behavior of the lagged stock variable (representing the partial adjustment hypothesis): while this variable provides no information on the determinants of portfolio flows, its inclusion improves the models’ forecasting ability, and its coefficient is highly significant. The findings of most of the empirical work on portfolio flows rest on the often implicit assumption that domestic and foreign interest rates are unaffected by capital flows. While this may be a proper assumption in an analysis of the total of U.S. foreign claims, there may be instances where, given a certain transactor in a certain market, capital flows affect interest rates. In such instances of interdependence, the use of simultaneous estimation techniques rather than single-equation estimates is required.
A. Foreign Direct Investment
I. Quantitative Analysis
Amano, Akihiro,Japan’s Balance of Payments and Its Adjustment: A Quarterly Econometric Model (Institute of Social and Economic Research, Osaka University, 1969).
Bandera, V.N., and J.T.White,“U.S. Direct Investments and Domestic Markets in Europe,”Economia Internazionale, Vol. XXI (1968), pp. 117–33.
Choudhry, Nanda K., YehudaKotowitz, JohnA. Sawyer, and JohnW.L. Winder,An Annual Econometric Model of the Canadian Economy, 1928-66 (Institute for the Quantitative Analysis of Social and Economic Policy, University of Toronto, November1968).
Helliwell, John F., LawrenceH. Officer, HaroldT. Shapiro, and IanA. Stewart,The Structure of RDX 1, a quarterly econometric model of the Canadian economy (Bank of Canada, Staff Research Studies, No. 3), 1969.
Krainer, Robert E.,“Resource Endowment and the Structure of Foreign Investment,”The Journal of Finance, Vol. XXII (1967), pp. 49–57.
Kwack, Sung Y.,A Model for U.S. Direct Foreign Investments (preliminary results of a study undertaken at The Brookings Institution, Washington, 1969).
Moose, James,“A Study of U.S. Direct Investment Overseas in Manufacturing and Petroleum Industries” (doctoral thesis presented at Harvard University, 1968).
Officer, Lawrence H.,An Econometric Model of Canada under the Fluctuating Exchange Rate (Harvard University Press, 1968).
Popkin, Joel,“Interfirm Differences in Direct Investment Behavior of U.S. Manufacturers” (doctoral thesis presented at University of Pennsylvania, 1965).
Prachowny, Martin F.J.,“Direct Investment and the Balance of Payments of the United States: A Portfolio Approach” (paper presented at the Conference on International Mobility and Movement of Capital, The Brookings Institution, Washington, January30-February1, 1970).
Rhomberg, Rudolf R.,“Transmission of Business Fluctuations from Developed to Developing Countries,”Staff Papers, Vol. XV (1968), pp. 1–29.
Scaperlanda, Anthony,“The E.E.C. and U.S. Foreign Investment: Some Empirical Evidence,”The Economic Journal, Vol. LXXVII (1967), pp. 22–26.
Scaperlanda, Anthony E., and LawrenceJ. Mauer,“The Determinants of U.S. Direct Investment in the E.E.C.,”The American Economic Review, Vol. LIX (September1969).
Severn, Alan K.,“Investment and Financial Behavior of American Direct Investors in Manufacturing” (paper presented at the Conference on International Mobility and Movement of Capital, The Brookings Institution, Washington, January30-February1, 1970).
Stevens, Guy V.G.,“Fixed Investment Expenditures of Foreign Manufacturing Affiliates of U.S. Firms: Theoretical Models and Empirical Evidence,”Yale Economic Essays, Vol. 9 (Spring1969).
Stevens, Guy V.G.,“Capital Mobility and the International Firm” (paper presented at the Conference on International Mobility and Movement of Capital, The Brookings Institution, Washington, January30-February1, 1970).
Weintraub, Robert,“Studio empirico sulle relazioni di lungo andare tra movimenti di capitali e rendimenti differenziali,”Rivista Internazionale di Scienze Economiche e Commerciali, Vol. 14 (1967), pp. 401–15.
II. Theoretical and Descriptive Work
Aharoni, Yair,The Foreign Investment Decision Process (Boston, 1966).
Barlow, Edward Robert, and Ira TensardWender,Foreign Investment and Taxation (Englewood Cliffs, N.J., 1955).
Behrman, J.N.,“Foreign Associates and Their Financing,”in U.S. Private and Government Investment Abroad, ed. byRaymondF. Mikesell (Eugene, Oregon, 1962).
Bouter, Arie C.,“Direct Investment: What It Really Is” (unpublished paper, International Monetary Fund, August27, 1970).
Hymer, Stephen,“International Operations of National Firms” (doctoral thesis presented at Massachusetts Institute of Technology, 1960).
Iversen, Carl,Aspects of the Theory of International Capital Movements (Copenhagen, 1935).
Jorgenson, Dale W.,“Capital Theory and Investment Behavior,”American Economic Association, Papers and Proceedings of the Seventy-fifth Annual Meeting (The American Economic Review, Vol. LIII, May1963), pp. 247–59.
Koyck, L.M.,Distributed Lags and Investment Analysis (Amsterdam, 1954).
Kreinin, Mordechai E.,“Freedom of Trade and Capital Movement: Some Empirical Evidence,”The Economic Journal, Vol. LXXV (1965), pp. 748–58.
Meyer, John R., and RobertR. Glauber,Investment Decisions, Economic Forecasting, and Public Policy (Boston, 1964).
Mikesell, Raymond F.,“Decisive Factors in the Flow of American Direct Investment to Europe,”Economia Internazionale, Vol. XX (1967), pp. 431–56.
Modigliani, Franco, and MertonH. Miller,“The Cost of Capital, Corporation Finance and the Theory of Investment,”The American Economic Review, Vol. XLVIII (June1958), pp. 261–97.
Nurkse, Ragnar,“Ursachen und Wirkungen der Kapitalbewegungen,”Zeitschrift für Nationalökonomie, Band V (1934), pp. 78–96.
Ohlin, Bertil,Interregional and International Trade (Harvard University Press, 1933).
Pizer, Samuel, and FrederickCutler,U.S. Business Investment in Foreign Countries (Washington, 1960).
Polk, Judd, IreneW. Meister, and LawrenceA. Veit,U.S. Production Abroad and the Balance of Payments: A Survey of Corporate Investment Experience (National Industrial Conference Board, New York, 1966).
Resek, Robert W.,“Multidimensional Risk and the Modigliani-Miller Hypothesis,”The Journal of Finance, Vol. XXV (1970), pp. 47–51.
Richardson, J. David,“Theoretical Considerations in the Analysis of Foreign Direct Investment” (University of Michigan, Seminar Discussion Paper, No. 18, April1970).
Robinson, Harry J.,The Motivation and Flow of Private Foreign Investment (International Industrial Development Center, Menlo Park, California, 1961).
Standke, Klaus-Heinrich,Amerikanische Investitionspolitik in der EWG (Berlin, 1965).
Stiglitz, Joseph E.,“A Re-Examination of the Modigliani-Miller Theorem,”The American Economic Review, Vol. LIX (December1969), pp. 784–93.
B. Long-Term Portfolio Capital
I. Quantitative Analysis
Bagueley, Robert,“International Capital Flows and Canadian Fiscal and Monetary Policy” (doctoral thesis presented at Harvard University, 1969).
Bell, Phillip W.,“Private Capital Movements and the U.S. Balance-of-Payments Position,”in Factors Affecting the U.S. Balance of Payments, a compilation of studies prepared for the Subcommittee on International Exchange and Payments of the Joint Economic Committee, U.S. Congress (Washington, 1962), pp. 395–481.
Branson, William H.,Financial Capital Flows in the U.S. Balance of Payments (Amsterdam, 1968).
Cohen, Benjamin J.,“A Survey of Capital Movements and Findings Regarding Their Interest Sensitivity,”The United States Balance of Payments: Part I—Current Problems and Policies, Hearings Before the Joint Economic Committee, U.S. Congress (Washington, 1963), pp. 192–208.
Choudhry, Nanda K., YehudaKotowitz, JohnA. Sawyer, and JohnW.L. Winder,An Annual Econometric Model of the Canadian Economy, 1928-66 (Institute for the Quantitative Analysis of Social and Economic Policy, University of Toronto, November1968).
Freedman, Charles,“Long-Term Capital Flows Between the United States and Canada” (doctoral thesis presented at Massachusetts Institute of Technology, February1970).
Helleiner, Gerald K.,“Connection Between United States’ and Canadian Capital Markets, 1952-1960,”Yale Economic Essays, Vol. 2 (Fall1962), pp. 351–400.
Helliwell, John F., LawrenceH. Officer, HaroldT. Shapiro, and IanA. Stewart,The Structure of RDX 1, a quarterly econometric model of the Canadian economy (Bank of Canada, Staff Research Studies, No. 3), 1969.
Kwack, Sung Y.,U.S. Investments in Foreign Securities (preliminary results of a study undertaken at The Brookings Institution, Washington, 1969).
Kwack, Sung Y.,The Determinants of U.S. Non-Bank Financial Capital Outflows (preliminary results of a study undertaken at The Brookings Institution, Washington, 1969).
Lee, C.H.,“A Stock-Adjustment Analysis of Capital Movements: The United States—Canadian Case,”Journal of Political Economy, Vol. 77 (Part I, July-August1969), pp. 512–23.
Miller, Norman C., and Marinav.N. Whitman,“A Mean-Variance Analysis of United States Long-Term Portfolio Foreign Investment,”The Quarterly Journal of Economics, Vol. LXXXIV (1970), pp. 175–96.
Officer, Lawrence H.,An Econometric Model of Canada under the Fluctuating Exchange Rate (Harvard University Press, 1968).
Prachowny, Martin F.J.,A Structural Model of the U.S. Balance of Payments (Amsterdam, 1969).
Rhomberg, Rudolf R.,“Canada’s Foreign Exchange Market: A Quarterly Model,”Staff Papers, Vol. VII (1960), pp. 439–56.
Rhomberg, Rudolf R.,“A Model of the Canadian Economy under Fixed and Fluctuating Exchange Rates,”The Journal of Political Economy, Vol. LXXII (1964), pp. 1–31.
Ripley, Duncan,“United States Investment in Canadian Securities, 1958-65” (doctoral thesis presented at Harvard University, 1969).
II. Theoretical and Descriptive Work
Moore, Basil J.,An Introduction to the Theory of Finance (New York, 1968).
Floyd, J.E.,“International Capital Movements and Monetary Equilibrium,”The American Economic Review, Vol. LIX (September1969).
Grubel, Herbert G.,“Internationally Diversified Portfolios: Welfare Gains and Capital Flows,”The American Economic Review, Vol. LVIII (December1968), pp. 1299–1314.
Learner, Edward E., and RobertM. Stern,“Problems in the Theory and Empirical Estimation of International Capital Movements” (paper presented at the Conference on International Mobility and Movement of Capital, The Brookings Institution, Washington, January30-February1, 1970).
Markowitz, Harry Max,Portfolio Selection: Efficient Diversification of Investments (New York, 1965).
Tobin, J.,“Liquidity Preference as Behavior Towards Risk,”The Review of Economic Studies, Vol. 25 (1958), pp. 65–86.
Examen des études quantitatives récentes sur les mouvements de capitaux à long terme
Dans cette étude, l’auteur s’efforce de résumer et d’analyser les conclusions des travaux empiriques effectués récemment sur les mouvements de capitaux à long terme, s’attaquant d’abord aux investissements directs à l’étranger et ensuite aux mouvements des placements à long terme en valeurs de portefeuille. Elle couvre une quarantaine de contributions apportées pendant les années soixante aux travaux empiriques sur les mouvements de capitaux, qui présentent un caractère en grande partie exploratoire et sont concentrés sur les Etats-Unis et le Canada.
Il ressort de l’examen des données quantitatives sur le comportement des investissements directs à l’étranger que le désir d’investir à l’étranger est affecté principalement par la dimension du marché étranger. Alors que ceci confirme simplement les conclusions des travaux descriptifs en la matière, des données quantitatives nouvelles sont venues renforcer les opinions traditionnelles mettant en vedette l’effet, sur les investissements directs, des différences entre les taux de rendement, auxquelles des travaux descriptifs récents avaient attribué une importance moindre. Outre la dimension du marché étranger et les différences entre les taux de rendement, aucune autre hypothèse relative au désir d’investir à l’étranger n’a pu être confirmée empiriquement. Il existe de nombreuses preuves que le flux effectif des investissements directs qui est dû aux facteurs agissant sur le désir d’investir à l’étranger, est soumis à des contraintes financières en ce qui concerne le montant absolu des fonds disponibles aux fins d’investissement ainsi que la répartition de ce montant entre les prêts et les investissements sous forme de prises de participation au capital actions de sociétés.
Les conclusions empiriques relatives aux principaux éléments qui déterminent le comportement des valeurs de portefeuille à long terme confirment l’hypothèse selon laquelle les mouvements observés de capitaux à long terme s’expliquent par un processus d’ajustement comprenant à la fois un élément de flux continu et un élément de stock transitoire. En ce qui concerne la sensibilité aux taux d’intérêt des mouvements de valeurs de portefeuille, on a constaté qu’il est préférable d’inclure séparément, dans les modèles de capitaux, les taux d’intérêt en vigueur à l’étranger et ceux en usage dans le pays de l’investisseur, au lieu d’utiliser la différence entre ces taux. Bien que les analyses empiriques soient généralement d’accord sur les déterminants principaux des mouvements des valeurs de portefeuille, les estimations de paramètres ne se prêtent pas bien à des comparaisons en raison de leur instabilité apparente dans le temps, du fait qu’elles sont établies à partir de simples modèles d’ajustement de flux ou de stocks, et aussi des différences importantes qui existent dans la définition des variables dépendantes et indépendantes. La plupart des conclusions empiriques reposent sur l’hypothèse selon laquelle le flux de capitaux de portefeuille ne modifie pas les taux d’intérêt. Dans des cas où il existe une interdépendance entre les mouvements de capitaux et les taux d’intérêt, une telle hypothèse ne serait pas justifiée, et il faut alors utiliser des techniques d’estimation simultanée plutôt que des estimations tirées d’une seule équation.
Examen de los estudios cuantitativos recientes sobre movimientos de capital a largo plazo
En este estudio se intenta resumir y analizar las conclusiones de los trabajos empíricos realizados recientemente sobre movimientos de capital a largo plazo, ocupándose primero de la inversión directa en el exterior y luego de los flujos de inversión en cartera a largo plazo. Trata de unas 40 obras añadidas en la década de 1960 a la literatura empírica sobre movimientos de capital, que en su mayor parte es exploratoria y se concentra en Estados Unidos y Canadá.
El examen de los estudios cuantitativos sobre el comportamiento de la inversión directa en el exterior indica que lo que influye primordialmente en el deseo de invertir en el exterior es la magnitud del mercado extranjero. Aunque esto simplemente corrobora las conclusiones de la literatura descriptiva, nueva evidencia cuantitativa ha venido a reforzar la creencia tradicional en el impacto de las tasas diferenciales de rendimiento sobre la inversión directa, impacto que en la literatura descriptiva reciente se ha considerado de importancia menor. Aparte de la magnitud del mercado exterior y de las tasas diferenciales de rendimiento, no se ha podido comprobar empíricamente ninguna otra hipótesis relativa al deseo de invertir en el exterior. Hay abundante prueba de que el flujo efectivo de inversión directa causado por factores que afectan al deseo de invertir en el exterior está sujeto a restricciones financieras atinentes al volumen absoluto de fondos disponibles para la inversión, así como a la proporción que pueda dirigirse a créditos o a capital social.
Los resultados empíricos relativos a los principales factores que determinan el comportamiento del capital en cartera a largo plazo confirman la hipótesis de que los movimientos observados del capital a largo plazo comprenden dos componentes de ajuste, uno continuo de flujos y uno transitorio del capital en existencia. Respecto a la sensibilidad de los flujos de capital en cartera ante las variaciones del tipo de interés, parece que es mejor incluir en los modelos de capital las tasas de interés externas e internas separadamente en vez de guiarse por la diferencia entre ambas. Aunque los análisis empíricos concuerdan en general en cuanto a los principales determinantes de los flujos de capital de cartera, las estimaciones de los parámetros no se prestan bien a la comparación, a causa de su aparente inestabilidad respecto al factor tiempo, a causa de que se les obtiene de modelos de ajustes que son, o bien simplemente de flujos o simplemente del capital en existencia, y a causa de las diferencias sustanciales en la inclusión de variables dependientes e independientes. La mayoría de los resultados empíricos se basan en el supuesto de que las tasas de interés no están afectadas por el flujo de capital de cartera. En los casos en que haya interdependencia entre los flujos de capital y las tasas de interés no se justificaría tal supuesto, y habría que recurrir a técnicas de cálculo simultáneo más bien que a cálculos de una sola ecuación.
Mr. Spitaller, an economist in the Research Department, is a graduate of the University of Graz, Austria, and of the School of Advanced International Studies of the Johns Hopkins University, Washington, D.C.
For the lagged reaction of capacity to output in general, see Koyck .
For a general discussion of investment decisions and the role of corrective adjustment, see, for example, Meyer and Glauber .
See Behrman , Robinson , Polk, Meister, and Veit , Aharoni , and Hymer . In a discussion of U.S. investment in the European Common Market, Standke  points to the “defensive” character of this investment aimed at forestalling the installation of plants by local or foreign firms.
See, e.g., Raymond Vernon . Mikesell  argues that the main determinant of U.S. direct investment in Europe is the market size and its long-term growth potential. The desire to skirt tariff walls, differences between domestic profits and profits abroad, and differences in expected rates of inflation are considered factors with little explanatory power. In an earlier study by Barlow and Wender , the market size is considered a main determinant of U.S. foreign direct investment in addition to trade barriers, freedom of convertibility, attitude of host country toward investor, accessibility to foreign labor, and U.S. and foreign tax structures. The authors group these influences into potential profitability, repatriation of profits, and protection of assets, and conclude that opportunity for profit is the basic factor in an investment decision, while the other factors relate to the ability of the company to realize its profit.
Absence of statistical support for hypotheses relating direct investment to these factors does not exclude the possibility that they may influence the investment decisions of individual companies in individual instances.
For references to transactions included in direct investment, see, for example, Officer , p. 73. For definitions of direct investment in terms of equity in enterprises aboard, see the U.S. Department of Commerce in Pizer and Cutler , p. 24. Also, International Monetary Fund, Balance of Payments Manual (Washington, Third Edition, July 1961), para. 373, p. 120, and Bouter .
Richardson  points out that differences in the specification of the dependent variable in the analysis of direct investment also limit the usefulness of the analogy to the theory of domestic investment, since the adaptation of the domestic investment theory to direct investment does not take account of these differences.
Scaperlanda and Mauer , for example, relate total U.S. direct investment to total income in the European Economic Community, while Bandera and White  examine U.S. direct investment in Europe by country and sector.
See, for example, White .
Prachowny  found little evidence that the rate of return expressed as the ratio of after-tax earnings to the stock of investment moves together with the long-term interest rate over time.
In contrast to Popkin, who examines the total foreign investment activities of a number of companies, Bandera and White , Moose , and Weintraub  derive their conclusions from an analysis of U.S. direct investment in selected countries. As already pointed out in connection with the rate-of-return hypothesis, such studies do not consider possibilities of substitution. The international firm is more likely to react to earnings from the total of its foreign direct investments rather than to earnings from investment in individual countries.
The contention that tariff discrimination induces foreign investment was also rejected in an earlier study by Scaperlanda . An inquiry into the possible effects of an Atlantic Free Trade Area on U.S. direct investment, addressed to 2,000 U.S. firms, found that tariff reductions abroad would not affect the direct investment decisions of U.S. firms. See Kreinin .
In the initial phase of a recession, supply and demand factors are therefore reinforcing in their effects on direct investment.
At this later phase of the recession, supply and demand factors are offsetting in their effects on direct investment, since the decrease in the supply of funds for investment abroad owing to the reduction in the total amount of funds available for both domestic and foreign investment coincides with a continued rise in the relative profitability of foreign investment.
Rhomberg  argues that at the beginning of a recession high earnings accumulated over previous periods coincide with a relative decline in domestic investment opportunities and that at the beginning of recovery low earnings coincide with an improvement in the domestic investment situation. In both instances direct investment is expected initially to be negatively related to domestic activity, and the findings of Krainer  seem to confirm these expectations. Rhomberg’s results show that as recession or recovery continues, U.S. foreign direct investment will eventually become positively related to domestic activity.
For a suggestion to apply the theory of portfolio choice to international capital movements, see Grubel .
See Basil J. Moore  for a discussion of portfolio risk. The risk attached to the holding of any asset can be measured by the variance of the probability distribution of the return on the asset around its mean, which is defined as the average of the squared deviation of all observations from the mean. A measure of the extent to which the returns on any two assets move together is the covariance, which is defined as the expected value of the product of the deviations of the returns on the two assets from their respective means. If the returns on any two assets are independent of one another, their covariance will be zero. Total portfolio risk, as measured by the variance of the sum of all assets contained in the portfolio, is equal to the sum of the individual variances of all assets plus twice the sum of the covariances of all pairs of assets. In the process of summing variances and covariances, the share of each asset in the total portfolio must be considered.
The effect on observed capital movements of a stock adjustment in response to current changes in interest rates may not only offset the effect of a concurrent flow adjustment but also, since total stock adjustment is not immediate, offset a stock adjustment induced by past changes in interest rates. If, for example, in two consecutive quarters, foreign interest rates rise in the first quarter and domestic interest rates in the second, owners of wealth will start by raising the share of foreign assets in their portfolio in the first quarter and follow by lowering it in the second, when in the absence of an increase in domestic interest rates the share of foreign assets in the total portfolio would have risen further on account of lagged reactions to the first quarter.
See Bagueley , Bell , Cohen , Choudhry, Kotowitz, Sawyer, and Winder , Helleiner , Helliwell, Officer, Shapiro, and Stewart , Kwack  and , Officer , Prachowny , Rhomberg  and , and Ripley .
See Helleiner  and Zumeta . Foreign borrowing is net of foreign lending in Bagueley , Choudhry, Kotowitz, Sawyer, and Winder , Helliwell, Officer, Shapiro, and Stewart , Officer , Rhomberg  and , and Ripley .
Choudhry, Kotowitz, Sawyer, and Winder , Freedman , Helleiner , Helliwell, Officer, Shapiro, and Stewart , Kwack , Miller and Whitman , Officer , Prachowny , Rhomberg  and , and Zumeta .
Distinctions by type of security can be found in Bagueley , Officer , and Ripley . Officer  indicates the possibility of a further breakdown of securities into bonds, debentures, and common or preferred stock.
The only distinction of securities by region relates to Canada, where a distinction is made between provincial and municipal securities.
The fact that Bell’s results improved when a linear trend variable was added may mean that some of the influence of the scale variable was captured, since the trend variable may have reflected the growth in the supply of loanable funds over time.
Since Zumeta’s study spans the time period 1947 to 1961, the absence of currency convertibility is an even more important element in the inelasticity of capital movements to interest rates than during the periods covered by the Bell and Cohen studies.
See, for example, the tabulation of the magnitude of interest rate effects, on U.S. and Canadian portfolio flows, in Learner and Stern , Tables 1 and 2, pp. 32–33.
An executive of the John Hancock Life Insurance Company declared that if the returns in the United States were relatively high, institutions would no longer be willing to incur the risk inherent in foreign investment, even though the interest rate differential between Canada and the United States had not decreased. See Ripley .
For the argument that interest rates affect capital movements and vice versa, see J.E. Floyd .
Growth in demand of funds is measured by borrowers’ income in Kwack  and by total new securities issues in Bagueley , Choudhry, Kotowitz, Sawyer, and Winder , Officer , Rhomberg , and Ripley . Growth in supply of funds is measured by lenders’ portfolio investments in Rhomberg  and by wealth in Kwack . Conceptually, the growth in the supply of funds of lenders would be better represented by income than by wealth.
The stock of supply of funds is measured by income in Branson  and implicitly contained in the dependent variable in Freedman , Lee , and Miller and Whitman . While income may be a good proxy for wealth it really represents additions to wealth, so that the stock effect may conceptually be better represented by some average of past and present values of income.
Since their analysis covered the period from 1957 to 1966, when the U.S. balance of payments was persistently in deficit, the results confirm only the hypothesis that a U.S. deficit raises U.S. purchases of foreign securities.
For the dependence of Canadian long-term interest rates on short-term interest rates, see Rhomberg .
See Branson , Kwack , and Prachowny  for the effectiveness of the restrictions imposed in 1963, and Kwack  and  and Miller and Whitman  for the effectiveness of the 1965 measures.
While these models are of some usefulness for purposes of forecasting total portfolio flows, they cannot be expected to isolate effectively flow and stock components, because the stock of wealth, representing portfolio size, and the change in the stock of wealth, representing portfolio growth, are interrelated.