Financial Integration and Interest Rate Linkages in Industrial Countries, 1958–71
  • 1 https://isni.org/isni/0000000404811396, International Monetary Fund

Beginning late in the 1950s a number of major developments had a bearing on money market integration in the industrial countries. On the one hand, controls over capital movements were liberalized in the second half of the 1950s. There is little doubt that this development increased the sensitivity of capital movements to differences in monetary conditions among these countries. During the 1960s, too, certain institutional developments—in particular, the rapid development of the Euro-dollar market and the growth of the multinational company—may have contributed further to the increase in the degree of money market integration. On the other hand, during the 1960s the monetary authorities of the industrial countries made various efforts to counteract the effects of increased capital mobility; they implemented various devices and controls to influence the movement of capital so as to give them increased leverage over their monetary policies. While there is little doubt that money markets are more highly integrated than in the years before convertibility, it is not at all clear that during the 1960s, on balance, the process of financial integration actually accelerated. This paper focuses on policy and theoretical issues posed by financial integration, but at the same time it explores measures of, and possible trends in, financial integration among the industrial countries.

Abstract

Beginning late in the 1950s a number of major developments had a bearing on money market integration in the industrial countries. On the one hand, controls over capital movements were liberalized in the second half of the 1950s. There is little doubt that this development increased the sensitivity of capital movements to differences in monetary conditions among these countries. During the 1960s, too, certain institutional developments—in particular, the rapid development of the Euro-dollar market and the growth of the multinational company—may have contributed further to the increase in the degree of money market integration. On the other hand, during the 1960s the monetary authorities of the industrial countries made various efforts to counteract the effects of increased capital mobility; they implemented various devices and controls to influence the movement of capital so as to give them increased leverage over their monetary policies. While there is little doubt that money markets are more highly integrated than in the years before convertibility, it is not at all clear that during the 1960s, on balance, the process of financial integration actually accelerated. This paper focuses on policy and theoretical issues posed by financial integration, but at the same time it explores measures of, and possible trends in, financial integration among the industrial countries.

Beginning late in the 1950s a number of major developments had a bearing on money market integration in the industrial countries. On the one hand, controls over capital movements were liberalized in the second half of the 1950s. There is little doubt that this development increased the sensitivity of capital movements to differences in monetary conditions among these countries. During the 1960s, too, certain institutional developments—in particular, the rapid development of the Euro-dollar market and the growth of the multinational company—may have contributed further to the increase in the degree of money market integration. On the other hand, during the 1960s the monetary authorities of the industrial countries made various efforts to counteract the effects of increased capital mobility; they implemented various devices and controls to influence the movement of capital so as to give them increased leverage over their monetary policies. While there is little doubt that money markets are more highly integrated than in the years before convertibility, it is not at all clear that during the 1960s, on balance, the process of financial integration actually accelerated. This paper focuses on policy and theoretical issues posed by financial integration, but at the same time it explores measures of, and possible trends in, financial integration among the industrial countries.

Section I examines the role of the Euro-dollar market as a major financial intermediary that channels short-term funds between the money markets of the industrial countries. It develops a simple model of the Euro-dollar market aimed mainly at explaining the movement in the Euro-dollar interest rate. This model is then tested by econometric analysis. Section II explores the relationship between financial integration and interest rate harmonization and looks at some statistical evidence on interest rate harmonization in the industrial countries. Section III develops a fairly simple model of the forward exchange market. The model is solved for the forward premium, which is explained by a number of variables, including the interest rate differential. The solution is then used as a basis for a regression analysis of the forward premium for selected industrial countries, focusing particularly on the responsiveness of the forward premium/discount to changes in interest differentials as a possible way of evaluating the degree of financial integration and possible changes over time. Finally, Appendix I looks in some detail at the extent to which Germany’s monetary policies may have been counteracted by capital movements.

I. The Euro-Dollar Market and the Determination of the Euro-Dollar Rate

This section explores the major factors that enter into the determination of the Euro-dollar interest rate. With this objective, it develops a model of the Euro-dollar market and later seeks to verify it empirically. In developing the model, a number of simplifying assumptions are made. Only three markets are identified—the U.S., the Euro-dollar, and the European markets, with the Euro-dollar market acting as an intermediary that channels funds between Europe and the United States. Corresponding to these three markets are three short-term interest rates—the U.S., the European, and the Euro-dollar rates. The U.S. interest rate is treated as exogenous1 in the sense that while it is allowed to influence the Euro-dollar interest rate, it is itself insensitive to movements in that rate. In this section, additionally, the same assumption is made about the European rate; however, the ways in which the Euro-dollar rate may influence the European rate are developed in some detail in the next section.

The Euro-dollar market is very competitive, and interest rates in that market are highly responsive to influences of demand and supply. Hence, a simple approach to the determination of the Euro-dollar rate is to examine the elements that influence the demand for and the supply of Euro-dollars. On the demand side we try to identify the variables influencing the stock of indebtedness (borrowing) in the market, while on the supply side (lending) we look at the influences determining portfolio choice between alternative financial assets.

Symbols used

Ius = constellation of short-term interest rates in the United States

Ie = the Euro-dollar rate

Io = constellation of short-term interest rates in Europe

Re = European interventions influencing the use of the Euro-dollar market

Qus = reserve requirements imposed on U.S. banks against Eurodollar borrowings

Iq = interest rate ceiling on U.S. certificates of deposit (Regulation Q)

Sp = speculation index (larger Sp implies greater expectation of a revaluation of the European currency)

Si = stock of indebtedness in Euro-dollar market

Sa = stock of assets in Euro-dollar market

A = net autonomous demand for Euro-dollars

The stock of indebtedness in the Euro-dollar market (borrowings in the market)

It is convenient to examine separately Euro-dollar borrowings by the United States and by Europe. Since U.S. borrowings have been undertaken predominantly by U.S. banks, we begin by looking at the determinants of these borrowings.

Consider first a situation in which Regulation Q is not in effect or is not operative.2 Then, banks will normally determine their borrowings by comparing the cost of borrowings in the Euro-dollar market with the cost of obtaining domestic funds. Domestic funds may be obtained by sales of commercial bills or certificates of deposit, by borrowing in the federal funds market, or by liquidating domestic assets, for example, treasury bills. The interest rate on these sources of funds (Ius) will represent one element in the cost of resort to domestic sources. Reserve requirements on different deposits will also be relevant in determining the effective cost of alternative sources of funds: for example, if Euro-dollar borrowings carry no reserve requirement, then, other things being equal, they will be a cheaper form of borrowing than domestic deposits that carry some reserve requirement.3

When Regulation Q is effective (e.g., as in 1966 and 1969), the analysis is somewhat more complicated. As market rates rise, large U.S. banks will find themselves in a position where the public is running down on a large scale its holdings of certificates of deposit. These banks will tend to lose reserves and will be seeking alternative methods of replenishing their liquidity. Alternative sources in the form of commercial paper4 or federal funds may be inadequate; the banks may have liquidated their treasury bills to near-minimum levels or, alternatively, they may be constrained in their sales of treasury bills by the knowledge that the interest rate on bills will rise still further, hence encouraging an additional movement out of certificates of deposit. In these conditions, the effective domestic interest rates may well be the interest rate on long-term bonds or the lending rate to customers. Indeed, in order to maintain customer good will, banks may well be prepared, at the margin, to pay a significantly higher interest rate on their Eurodollar borrowings than they earn on lending to their customers. Regulation Q, then, when it is effective, must be treated as an independent element in the demand for Euro-dollars by U.S. banks. The role of Regulation Q in these conditions might be represented by the difference between market rates in the United States and the ceiling on certificates of deposit (Ius-Iq); the larger this difference, the greater the depletion of certificates of deposit and hence the greater (other things being equal) the U.S. banks’ indebtedness in the Euro-dollar market.

To the extent that nonbanks in the United States borrow in the Euro-dollar market, their demand will also be determined by relative interest rates in the two markets, so that no additional variable needs to be introduced here.

Banks and nonbanks in Europe will determine their borrowings on the Euro-dollar market on the basis of a number of considerations, including relative interest rates (Ie, Io), expectations of changes in exchange rates, and various regulations, directives, and interventions that influence the extent to which the market will be used. For example, a rise in the Euro-dollar rate in relation to interest rates in Europe will tend to reduce the stock of European indebtedness in the market. An expectation of a European revaluation in relation to the dollar will generate increased borrowing in the Euro-dollar market. A directive by a central bank to the commercial banks to reduce their net foreign asset positions will tend to reduce the net demand for Euro-dollar borrowings. An increase in reserve requirements on foreign liabilities will have the effect of reducing the indebtedness in the Euro-dollar market. Forward market intervention by central banks may also influence incentives to borrow on the Euro-dollar market.

Additionally, it is useful to allow explicitly for the possibility of net autonomous demands being made on the market, for example, by international agencies, or by central banks or government departments. A particular instance of this was the U.S. Treasury borrowing on the Euro-dollar market during 1970, with an objective of maintaining relatively higher interest rates in the Euro-dollar market. These autonomous demands will be represented by the symbol A.

If we consolidate all these considerations that determine the stock of indebtedness in the market, then, assuming a simple linear function, we have

(1.1)Si=a(IusIe)+b(IusIq)cQus+dSp+A+eRe+f(IoIe)

where Re represents all European interventions in the free use of the market (an increase in Re increasing net borrowings on the market).

The stock of assets in the Euro-dollar market (lending in the market)

The supply of funds to the market from the United States will be influenced by the differential between the Euro-dollar and U.S. rates, while the supply of funds from Europe will be sensitive to the differential between the Euro-dollar and European rates. Additionally, speculation, interventions, and autonomous factors will, equally, affect the supply of funds. If we suppose, however, that the net effects of these variables have already been allowed for in the equation representing the demand for Euro-dollar borrowing, the function for the stock of assets may be written simply as

(1.2)Sa=g(IeIus)+h(IeIo)

Consolidating demand and supply

In equilibrium, the stock of indebtedness will be equal to the stock of assets. These two equations may be consolidated and solved for the Euro-dollar rate.5

(1.3)Ie=mIus+(1m)Io+bk(IusIq)ckQus+dkSp+1kA+ekRe,

where

k=(a+g)+(f+h)m=a+gk

This equation explains the Euro-dollar rate in terms of seven exogenous variables: the U.S. rate, the European rate, the differential between the U.S. market rate and the ceiling rate on certificates of deposit, reserve requirements in the United States, an index of speculation, autonomous demands, and regulations in Europe.

An increase in the U.S. rate will tend to raise the Euro-dollar rate, given all the other variables in the equation. The rise in the U.S. rate will encourage increased borrowings in the Euro-dollar market (in particular by the U.S. banks), at the same time encouraging a movement out of Euro-dollar assets and into U.S. assets. This will pull the Euro-dollar rate up, which in turn will induce, simultaneously, a move out of European assets and into Euro-dollars and a decrease in borrowings by Europeans. By analogous reasoning, an increase in the European rate will tend to push up the Euro-dollar rate. When Regulation Q is effective this will generate an additional demand for Eurodollars, which will tend to raise rates in the market. The more effective is Regulation Q, the greater is the gap between market and ceiling rates and the stronger will tend to be the demand for Euro-dollars. An increase in reserve requirements on Euro-dollar borrowings in the United States, given reserve requirements on other deposits, will reduce the demand for Euro-dollars, which will tend, other things being equal, to lower the interest rate on Euro-dollars. Given the U.S. and European interest rates, an expectation of a revaluation of the European currency in relation to the dollar will generate an increase in Euro-dollar borrowings, which in turn will pull up Euro-dollar rates, in this way encouraging a movement on the supply side out of the U.S. market and into the Euro-dollar market. An increase in autonomous demands for Euro-dollar funds will tend to raise the interest rate. Finally, increased interventions in Europe will reduce the supply of and increase the demand for Euro-dollars, which will tend to pull up Euro-dollar rates.

The most critical coefficients in equation (1.3) are a, f, g, and h; a and g are a measure of the flows between the U.S. and Euro-dollar markets in response to the interest rate differential. Since arbitrage between these two markets has been effected chiefly through the intermediary of U.S. banks, it is reasonable to assume that a > g; f and h, on the other hand, represent the sensitivity of the flows between Europe and the Euro-dollar market in response to the interest differential between these two markets. Given that the Euro-dollar is a closer substitute for U.S. instruments than for European instruments, one would expect changes in the U.S. rate to generate more sympathetic movements in the Euro-dollar rate than equivalent changes in the European rate. This amounts to saying that there is a strong presumption that m > 1 – m.

Where the Euro-dollar and U.S. markets are perfect substitutes (e.g., where a + g approaches infinity), m will approximate 1. In this case, financial conditions in the United States will completely determine the Euro-dollar rate. To the extent that U.S. banks dominate arbitrage movements between the U.S. and the Euro-dollar markets, the interest rates in these two markets will settle at levels that reflect relative cost or availability considerations to these banks. On the other hand, changes originating in Europe will make no impact on the Euro-dollar rate. For example, an increase in interest rates in Europe will reduce the supply of funds available to the Euro-dollar market and at the same time increase the demand for Euro-dollar funds. The increased demand will be fully met, at unchanged Euro-dollar interest rates, by increased supplies from the United States, in part to offset the drop in supply from European sources and in part to meet the additional demand. Again, an increased demand for Euro-dollar borrowing triggered by an expected revaluation of a European currency in relation to the dollar will be fully met by supplies from the United States without making any impact on the Euro-dollar rate.

An interesting policy issue that arises from this discussion is the extent to which it might be feasible to generate effects on capital flows (and the balance of payments) between Europe and the United States without changing either U.S. or European interest rates. For example, it may be felt that in the interest of a more optimal distribution of reserves there should be a movement of capital funds out of Europe and into the United States. At the same time, both Europe and the United States may consider their respective interest rates as fully appropriate to their particular cyclical phases. Suppose in this instance that there was some official borrowing in the Euro-dollar market by the United States. This official demand may be accommodated from two sources—the United States and Europe. From equation (2.1), it may be seen that the effect of a unit increase in autonomous borrowing on the Euro-dollar interest rate is 1k. This demand must equal the sum of net supplies from the United States and from Europe, which are represented, respectively, by m and (1 – m). Hence, we have the result that

A=1=m+(1m)

This shows the proportions in which the funds are supplied from the United States and Europe. Where a + g → ∞, all the funds are supplied from the United States, and the Euro-dollar rate cannot be altered at all. At the other extreme, where f + h → ∞, that is, the Euro-dollar and European rates are perfectly integrated, all the funds will come from Europe, and again the Euro-dollar rate will not change. Any intermediate case will provoke some rise in the Euro-dollar rate and some flows from the two sources, with the more integrated source providing a larger proportion of the funds. Provided therefore that a + g does not approach infinity, and provided that there is some substitution between the Euro-dollar market and Europe (i.e., f + h > 0), it would be possible to provoke some changes in the balance of payments between the United States and Europe by operating directly on the Euro-dollar market.6

Although simplifying assumptions have been made in the analysis, it is felt that the influences identified tend to capture the main elements that enter into the determination of the Euro-dollar rate. One obvious limitation of the analysis is the assumption of a single European currency and a single European rate. If that assumption were relaxed, one would need to spell out demand and supply functions for each country. In these conditions one would need to allow additionally for the possibility of flows within Europe through the intermediary of the Eurodollar market. This would make the theoretical framework much more cumbersome, while adding little new analytical insight.

Empirical evidence

A market characterized by rapid growth and by frequent institutional and legal changes impinging on its workings is difficult to study econometrically. Nevertheless, an attempt will now be made to test the model developed in the previous section and to try to arrive at some rough estimate of the significance of some of the variables identified earlier. Since we are concerned mainly with explaining the trends in the Eurodollar interest rate, rather than in identifying individual supply and demand equations, we may attack the problem by attempting a direct estimate of equation (1.3). Of the variables in the equation, a number may be dropped either because they are considered to be of little practical significance or because they cannot be quantified. The variables dropped for these reasons were Qus, A,7 and Re. This leaves us with the truncated equation

Ie=mIus+(1m)Io+bk(IusIq)+dkSp+u

The data used in estimating a general equation in this form on monthly data for the period January 1961 to March 1971 were

Ie = three-month Euro-dollar deposit rate

Ius = three-month U.S. Treasury bill rate

Io = three-month German interbank rate (Ig) and three-month U. K. Treasury bill rate (Iuk)

Ds1g = a dummy for speculation in favor of the deutsche mark8

Ds2g = a dummy representing the unwinding of speculative inflows into Germany8

Dsuk = a dummy for speculation against sterling8

Ius – Iq = three-month U.S. commercial bill rate less Regulation Q ceiling rate on time deposits. Effective during July 1966 to January 1967 and again during May 1968 to June 1970. Set at zero in all other months.

Before any results are reported, certain difficulties in estimating this equation should be noted. The effects of Regulation Q may not have been the same during 1966-67 and 1968-70. The U.S. banks by 1968-70 had developed closer links with the Euro-dollar market and had learned something from their experience in 1966-67. The gap between market and ceiling rates not only was larger in the later episode but also persisted over a much longer period. Also, as already indicated, the incentives to use the Euro-dollar market were very much a function of the cost and availability of domestic sources of funds. As banks discover new domestic sources of funds (e.g., in 1970, when they resorted to issues of commercial paper through their subsidiaries), they need to rely less on Euro-dollar borrowings; on the other hand, if domestic sources of funds run down (e.g., if treasury bills are depleted to minimum levels), there will be greater resort to the Euro-dollar market.9 All of this suggests that there is no simple relationship between Regulation Q interest rate gaps and the Euro-dollar interest rate. Also, a simple dummy to represent speculative periods cannot capture the intensity of speculation. Again, there is bound to be some simultaneity between the Euro-dollar rate and the European rates, since causation runs not only from the European rate to the Euro-dollar rate but also in reverse. Nor, of course, are the German and U. K. rates necessarily representative of all relevant rates in Europe. Finally, by excluding certain variables, especially if these made their impact over a period of several successive months, there is a serious danger of serial correlation in the residuals.

Some of the results of estimating equation (1.3) in levels and first differences are given below; t-ratios are shown in parentheses below the coefficients.

(1.3.1)Ie=0.030(0.15)+0.892Ius(13.81)+0.416(7.29)(IusIq)+0.134(4.36)Ig+0.471(4.11)Ds1g+0.103(0.56)Ds2g+0.137(2.73)Iuk0.081(0.86)DsukR¯2=0.962SE¯=0.391DW=0.61(123observations,Jan.1961Mar.1971)
(1.3.2)Ie=0.334(1.97)+1.000(20.07)Ius+0.417(7.44)(IusIq)+0.132(4.72)Ig+0.503(4.92)Ds1gR¯2=0.960SE¯=0.406DW=0.60(123observations)
(1.3.3)Ie=0.358(2.19)+0.847(13.06)Ius+0.518(4.20)(IusIq)+0.124(3.24)Ig+0.111(3.70)Iuk0.078(1.16)DsukR¯2=0.958SE¯=0.204DW=0.95(71observations,Aug.1961June1967)
(1.3.4)ΔIe=0.0005(0.02)+0.622(5.09)ΔIus+0.231(2.92)Δ(IusIq)+0.216(4.27)ΔIg+0.125(1.50)ΔIuk+0.218(2.61)ΔDs1gR¯2=0.414SE¯=0.287DW=2.08(123observations)

On the face of it, the results of equations (1.3.1) and (1.3.2) for levels of the variables would appear to be good. Nearly all the variables are significant, and the explanatory power of the equations is high; however, there is strong evidence of serial correlation.10 This suggests the omission of systematic variables, which may have influenced the Euro-dollar rate.11 Despite this weakness, the results of the estimations are revealing. Equations (1.3.1) and (1.3.2) suggest that the coefficient for the United States may be close to 1, which, as suggested in the earlier discussion, implies nearly perfect arbitrage between the U.S. and the Euro-dollar markets. The coefficients for interest rates in Germany and the United Kingdom in equation (1.3.1) are both significant but very low. Speculation in favor of the deutsche mark (Ds1g) proves to be significant, with a fairly large coefficient, but neither of the other two speculative dummies is significant. Regulation Q also turns out to be very significant. The results suggest that for every 1 per cent that the market rate is over the ceiling rate, the Euro-dollar rate will tend to rise by 40 basic points. As an illustration, during the second half of 1969 and the early part of 1970, market rates were more than 3 percentage points above ceiling rates. This would imply that in those months Regulation Q was responsible for pushing up the Euro-dollar rate, additionally, by about 120 basic points.

Equation (1.3.3) is estimated for a shorter period, from August 1961 through June 1967. This period excludes all the early months of 1961, which were influenced by the revaluations in favor of Germany and the Netherlands, and all the months after June 1967, which were interspersed with bouts of speculation starting with the attacks on sterling later in that year. There is little change in the estimated coefficients for the major variables. Serial correlation, while still present, is somewhat reduced.

Equation (1.3.4) is an estimate of our basic equation (1.3) in first-difference form. An interesting feature of this result is that serial correlation disappears. The explanatory power is not very high, but most of the variables (the only exception being the U. K. rate) continue to be significant. The coefficients for the U.S. rate, speculation, and Regulation Q all drop quite sharply. For the U.S. rate, the coefficient is now significantly below 1. Nevertheless, it is safe to conclude, on the basis of a priori reasoning and on the basis of some of our own and other results,12 that the coefficient of the U.S. rate would probably lie between 0.8 and 1.0.

An attempt was also made to determine whether there were any lags in the effects of some of the independent variables on the Euro-dollar rate. In particular, the lagged U.S. interest rate and the lagged Regulation Q gap were also included as additional variables. These, however, were not significant, suggesting that the adjustment to changes in conditions affecting the Euro-dollar market are rapid.

Chart 1 shows the trends between 1960 and 1971 in the Euro-dollar and U.S. rates as well as the trends in the differentials between these two rates (the Euro-dollar rate less the U.S. Treasury bill rate). It is evident that the two rates have moved very closely together, with turning points coinciding fairly well. This is consistent with the finding that the U.S. interest rate is a dominant variable in explaining the behavior of the Euro-dollar rate.

Chart 1.
Chart 1.

Euro-Dollar Rate and U.S. Treasury Bill Rate, 1960–71

(In per cent per annum)

Citation: IMF Staff Papers 1973, 001; 10.5089/9781451956351.024.A001

At the same time, the lower portion of the chart shows that the differential has not been constant over the same time period. The differential is consistently in favor of the Euro-dollar market, during most of the time fluctuating roughly around 1 percentage point.13 A striking feature of this differential is the fact that it increased sharply during 1966 and even more sharply during 1969. On the basis of the earlier discussion, we would expect the differential to be increased as European rates rise, to the extent that there is speculation against the dollar, and during periods when Regulation Q is effective. A simple regression explaining the differential along those lines was estimated as follows:

(1.3.5)IeIus=0.034(0.18)+0.369(8.12)(IusIq)+0.109(4.30)Ig+0.460(4.32)Ds1g+0.097(2.44)Iuk0.114(1.26)DsukR¯2=0.736SE¯=0.406DW=0.60(123observations)

Although there is again strong evidence of serial correlation, the explanatory power of the equation is fairly good. The coefficients are consistent with the results obtained in equations (1.3.1) and (1.3.2). The sharp rise in the differential during 1969 may then be explained, in part at least, by the presence of speculation and the operation of Regulation Q, while much of the rise during 1966 may be attributable to Regulation Q.14 The sharp rise in the differential, however, would have been possible only if there were constraints on the movement of arbitrage funds from the United States to the Euro-dollar market. A differential approaching nearly 4 percentage points can be maintained only if there is limited arbitrage on the supply side between the two markets. This is supported, first, by the fact that U.S. residents are legally constrained from placing funds in the market, and, second, by the evidence that foreigners had virtually run down their money market assets to near-minimum levels, so that there was in fact little scope for arbitrage from this source.

II. Financial Integration and Interest Rate Harmonization

The previous section examined the effects of changes in conditions in the United States—in particular, changes in U.S. interest rates on the Euro-dollar rate. It was assumed that while changes in interest rates (as well as conditions) in Europe might affect Euro-dollar rates, changes in Euro-dollar rates in turn did not influence rates in Europe. This section explores in some detail the ways in which changes in the Euro-dollar rate, assumed to be provoked by changes in monetary conditions in the United States, might make an impact on interest rates in Europe under a regime of fixed exchange rates.

We begin by evaluating the effects of financial integration on the volatility in reserves over a business cycle. We then examine the ways in which financial integration affects policy making, focusing in particular on the conditions in which integration might bring some interest rate harmonization. Later, we explore possibilities of obtaining evidence bearing on the degree of financial integration by looking at statistical indicators of interest rate harmonization among the industrial countries. Finally, in Appendix I we review in some detail the experience of Germany in trying to maintain independent monetary policies in the face of integrated capital markets.

Symbols used

X = exports

M = imports

K = net capital account

m = foreign propensity to import

Yf2 = foreign income in boom

Yf1 = foreign income in recession

Yd = domestic income (assumed to be fixed)

If2 = foreign (Euro-dollar) interest rate in boom

If1 = foreign (Euro-dollar) interest rate in recession

Ih = domestic interest rate (assumed to be fixed)

B2 = balance of payments (change in international reserves) in recession

B1 = balance of payments in boom

a = sensitivity of capital movements to interest rate differences

Financial integration and the volatility in reserves

In this part we are concerned with the effects of differing degrees of financial integration on the movement in reserves over a business cycle in a representative European economy. The analysis that follows is based on the following set of assumptions.

1. All business cycle fluctuations originate in the United States, with the European economy assumed to remain stationary. This means that Europe would be in a relative recession when the United States was experiencing a boom, and in a relative boom when the United States was experiencing a recession. While Europe’s own income, imports, and interest rates would remain fixed, they would be exposed to fluctuations in all these variables in the United States.

2. Interest rates move procyclically in the United States.15 This means that a boom in the United States would be characterized by relatively high income and interest rates, and a recession by relatively low income and interest rates.

3. The current accounts of the balance of payments of Europe and the United States are functionally related to the level of activity, improving in Europe as the United States enters a boom and deteriorating as the United States enters a recession. The capital accounts of the balance of payments of Europe and the United States are functionally related to the movement in interest rates, improving in Europe as the United States enters a recession and deteriorating as the United States enters a boom.

Then

(2.1)B=XM+K
(2.2)X=mYf
(2.3)M=M¯
(2.4)K=a(ΔIhΔIf)
(2.5)I=I¯
(2.6)B2=mYf2M¯+a(If1If2)
(2.7)B1=mYf1M¯+a(If2If1)

B, the balance of payments of the representative European country, is equal to exports less imports plus the net capital account. Exports are a function of the foreign country’s income (Yf) where m represents that country’s marginal propensity to import; the net capital account is explained by the change in the interest differential.16 Imports and interest rates are assumed to be fixed, by definition. B2 and B1 represent the balance of payments in the recession and the boom, respectively.

The volatility in reserves is measured by B2B1 which is the change in the balance of payments over the cycle.

(2.8)B2B1=m(Yf2Yf1)2a(If2If1)

Consider, to begin with, the simplest case, where B2 + B1 = 0, that is, where there is no cumulated change in reserves over the cycle. This is so where a deficit (surplus) in the recession is exactly matched by a surplus (deficit) in the boom. Since by definition the net capital outflow in the recession is equal to the net capital inflow in the boom, this must also mean that the current account surplus in the recession (mYf2M¯) must be equal to the current account deficit in the boom (M¯mYf1). A country meeting this condition might be described as a neutral currency country, that is, a country whose currency is neither strong nor weak.17

The coefficient α is a measure of the degree of financial integration, while the coefficient m is a measure of the degree of integration in the goods market. By definition, Yf2 > Yf1 and If2 > If1; hence, it follows that if m(Yf2Yf1) > 2α(If2If1), the balance of payments will improve from the boom to the recession, and if 2a(If2If1) > m(Yf2Yf1), the balance of payments will deteriorate from the boom to the recession.

As the economy moves from the boom to the recession, there are, then, two opposing influences at work: the current account improves while the capital account deteriorates. The greater the financial integration in relation to goods integration (i.e., α in relation to m), the more likely it is that the economy will experience a deterioration in its balance of payments as it moves from boom to recession, since the increased capital outflow will more than offset the improved current account.

Whether increased integration increases or reduces the volatility in reserves (as measured by B2B1) depends very much on the starting point. Increased volatility in this context means simply that the absolute size of equivalent deficits and surpluses is increased. Suppose that the starting point is the case where a is relatively low so that m(Yf2Yf1) > 2a(If2If1), then increasing a will reduce surpluses and deficits up to the point where they would shrink to zero. Suppose now that a is sufficiently large so that 2a(If2 - If1) > m(Yf2 - Yf1), then the starting point is a deficit in the recession and a surplus in the boom. As a increases beyond that point, the volatility also increases.

The large jump in financial integration starting late in the 1950s may have increased or reduced the volatility in reserves, since it may have transformed the economy from a relatively low to a relatively high a. At the same time, these countries would have experienced a qualitative change in the movement of their balance of payments over different phases of the cycle, relative booms now being characterized—on balance—by overall surpluses, and relative recessions by overall deficits. It is possible, too, that in the years following 1958, as financial integration took hold and possibly increased as a result of a number of institutional developments (e.g., the growth of the Euro-dollar market, the multinational company), the potential volatility in reserves may well have been accentuated. This particular conclusion, that integration in the years after 1958 may have increased the potential volatility in reserves, is important in setting the stage for some of the arguments advanced in the next section.

There is clearly a point, as financial integration increases, at which the capital account would become potentially dominant in accounting for reserve movements. This also means that a country becomes increasingly exposed to speculative attacks, since in these conditions it would experience a large accrual of reserves during a boom and a large loss of reserves during a recession. In the boom the increased reserves would tend to be buttressed by speculative inflows, while in the recession the loss of reserves would be aggravated by speculative outflows. If we retain the assumption of a neutral currency, this implies that speculation is basically “symmetrical” in the sense that “speculation” is not any more likely in the boom than in the recession.18

The notion of symmetrical speculation breaks down when the assumption of a neutral currency is relaxed. A neutral currency has been defined as one where B2 + B1 = 0. A strong currency (e.g., Germany, Japan) is one where B2 + B1 > 0, that is, the cumulated sum of reserves is greater than zero. Reserves, in other words, accumulate over a cycle, so that the country is effectively in fundamental disequilibrium. Suppose that a is sufficiently large so that 2a(If2If1) > m(Yf2Yf1), that is, the improvement in the capital account is greater than the deterioration in the current account; then it follows that whatever the initial balance of payments, that is, whether B2 is greater or less than zero, there will be an improvement in the balance of payments in the boom. If there was a surplus in the recession (B2 > 0), the surplus will be larger in the boom. If there was a deficit in the recession (B2 < 0), the deficit will be transformed into a surplus. Since by definition B2 + B1 > 0, the surplus would have to be larger than the deficit. Hence, in either case, speculation will tend to be asymmetrical, being more likely in the boom than in the recession. A strong currency, which is also financially integrated (e.g., in Germany), will tend to experience particular difficulties during a boom.19

Suppose that a, however, is sufficiently low so that m(Yf2Yf1) > 2a(If2If1); in these conditions, the balance of payments will deteriorate in the boom. Then there must be a surplus in the recession, and this surplus must either shrink or turn into a deficit. Given that B2 + B1 > 0, the surplus must be larger (absolutely) in the recession than either the surplus or deficit in the boom. Hence, speculation will be asymmetrical, but it will be more likely during a recession than during a boom. A strong currency, which is weakly integrated financially, will tend to experience particular difficulties during a recession.20

Much the same analysis might be applied to a weak currency, where B2 + B1 < 0. Where financial integration is high, speculation will be more likely during a recession; on the other hand, where integration is low, speculation is more likely during a boom. Thus, this case is the exact reverse of the strong currency case.

At this point a number of qualifications and additional comments are needed to supplement the foregoing analysis. First, the notion of strong, weak, and neutral currencies tends to collapse when integration is high, since the potential volatility in reserves becomes overwhelmingly dominated by short-term capital movements. Second, the assumption of symmetrical, homogeneous, and independent cycles, on which the analysis was based, is, of course, simplistic. Cycles, in relation to other countries, may vary in duration. Given a highly integrated economy, for example, the longer the duration of its relative boom, the more severe the reserve problem that it will have to face. Nor are the countries’ cycles independent, since cycles may be transmitted through the current or capital accounts from one country to another. Third, we were concerned with the effects of increasing financial integration, given some integration in the goods markets, on the volatility in reserves. It may be that increased financial integration is accompanied by increased integration in trade, so that m may also increase. This, too, may affect the results, since it means that the current account also becomes more sensitive to changes in economic activity. Increasing both a and m leaves it ambiguous as to whether m(Yf2Yf1) > 2a(If2If1), or the reverse. Again, since the effect of increased integration is a function of both m and a, this effect will differ between countries that have different propensities to import. Fourth, the marginal propensity to import is not likely to remain fixed over the cycle. When demand pressure is particularly intense, the marginal propensity to import tends to rise significantly. At that point it is possible that the adverse effects on the current account will tend to dominate. Fifth, the analysis becomes somewhat more complicated if account is taken of the possibility that capital flows—more particularly, direct investment and long-term portfolio investment—may respond positively to increases in income, an assumption that is frequently made in the literature.21 If increased financial integration also means greater sensitivity of long-term flows to fluctuations in income, this fact would reinforce the conclusion that increased integration, beyond a point, would entail greater volatility in reserves.

Financial integration and policy options

We concluded that in conditions of increased financial integration, countries that persisted in interest rate policies geared to domestic activity would be likely to be exposed to greater fluctuations in their reserve positions. We now want to evaluate the kinds of policy option that are open to a country, under narrowly fixed exchange rates, in the conditions assumed.

To begin with, a distinction needs to be made between policies that allow a country to continue to pursue independent interest rates and policies that entail some harmonization of interest rates with interest rate developments in relevant foreign markets. Each of these two types of policy may be pursued with a variety of policy instruments.

Consider first those policies that allow interest rate independence. Two general approaches are possible here. One approach is to directly attack the source of the increased fluctuations in reserves, while at the same time continuing to maintain independent interest rates. This entails attenuating the reserve movements by encouraging or discouraging capital flows when the country’s cycle is out of phase, effectively reversing trends toward financial integration in the interest of monetary independence. For example, when the country is in a boom with relatively higher interest rates, it would try to offset the additional inflows by measures to discourage the entry of capital. During a recession, on the other hand, when its interest rates are relatively lower, it would implement measures to discourage the loss of capital. Measures to influence capital movements may include tighter exchange controls, controls on the net foreign positions of the commercial banks, policies that effectively raise or lower the cost of foreign borrowing (such as special reserve requirements), the prohibition of interest payments to foreigners, and spot and/or forward exchange market interventions.22

Another approach is simply to accept the increased volatility in reserves23 and to continue to maintain independent interest rate policies by attempting to neutralize the domestic liquidity effects of movements in reserves. Even assuming that the monetary authorities were equipped with adequate instruments to offset reserve movements, one difficulty with this policy is that if these reserve movements are very large they may set off expectations of changes in par values, which in turn, as we have already seen, will aggravate the volatility in reserves. The extent to which this is likely will depend on the length of time during which a country’s monetary policies are out of phase and the degree to which it is financially integrated with other economies.24

Consider now those policies that entail some interest rate harmonization. One way of attacking the problem is to alter the monetary/fiscal mixes over the cycles. For example, during a relative boom the same degree of restriction could be maintained by attenuating the rise in the domestic interest rate (i.e., by following a relatively more expansionary monetary policy) and simultaneously implementing a tighter fiscal policy. On the other hand, during a relative recession the domestic interest rate would drop by less but fiscal policy would then be more expansionary. In effect, one would want to avert the additional reserve accumulation in the boom and the additional reserve decumulation in the recession without changing the degree of restrictiveness or ease of monetary/fiscal policies combined.

Conceivably, too, it may be possible to avert some of the fluctuation in reserves by manipulating monetary instruments more flexibly. One possibility is to use the short-term and long-term interest rates more aggressively over the cycle. If short-term interest rates had a comparative advantage over long-term interest rates for balance of payments purposes, and long-term interest rates had a comparative advantage over short-term rates for influencing domestic expenditure, then the long-term rate should be allowed to rise relative to the short-term rate during a relative boom and the short-term rate should be allowed to rise relative to the long-term rate during a relative recession. These manipulations, to the extent that they were feasible,25 might go some way toward attenuating the reserve volatility without affecting the degree of restriction or ease of interest rate policies.

Another possibility is to alter the mix of discount rate and credit-rationing policies over the cycle. Let us assume that discount rate changes, to which money market rates tend to be geared, have a comparative advantage over credit-rationing policies for balance of payments purposes, and that the degree of rationing of bank credit has a comparative advantage over discount rate policies for influencing domestic expenditure. Then, during a boom, to avert the additional inflow of capital, discount rates would have to be lowered. At the same time, however, to attenuate the possible expansionary effect on expenditure, a more restrictive credit policy would have to be followed. During a recession, the reverse policy would have to be implemented.

In these cases some interest rate harmonization was achieved while attempts were also being made to maintain the strength of stabilizing policies to counter cyclical disturbances. If greater interest rate harmonization is pursued without policies to ensure the same degree of restrictiveness or ease during cyclical phases, any attenuation in the fluctuation in reserves may turn out to be at the expense of greater volatility in domestic income. For example, if there were constraints on the range of fluctuation in the domestic interest rate and no other stabilizing policies were pursued, then both the acceleration in income in the boom and the slowdown in the recession may be aggravated.

Many of the points just made may be illustrated with the aid of a diagram.26 In Figure 1, XX represents equilibrium points in the goods market; LL represents equilibrium in the money market where the demand and supply of money are equated; EE represents equivalent states of the balance of payments (for simplicity, where the net accumulation of reserves is zero). The EE schedule has a positive slope, because as income increases and the current account deteriorates the rate of interest must rise to induce sufficient capital inflow to offset the potential drain in reserves. E0 E0 represents the case where the economy is more financially integrated, with the balance of payments effect of a given rise in income now being offset by a smaller rise in the interest rate.

Figure 1.
Figure 1.

General Equilibrium of an Open Economy

Citation: IMF Staff Papers 1973, 001; 10.5089/9781451956351.024.A001

Consider now three types of disturbance: an autonomous increase in private expenditures (shifting the XX schedule upward); a fall in the demand for money (shifting the LL schedule to the right); and a rise in the foreign interest rate (shifting the EE schedule to the left). All three disturbances are shown in Figure 1.

Suppose, to begin with, that the monetary authorities sterilize the balance of payments effects of those disturbances on the money supply. The new equilibrium points will be a, b, and c, respectively, for the three types of disturbance. At each of these points the money and goods markets will be in equilibrium but the balance of payments will have been thrown out of balance. In each case it is easy to demonstrate that the change in reserves is larger for the more highly integrated economy (E0 E0) than for the one less highly integrated (EE).27 The balance of payments disequilibrium may be measured by the distance from the equilibrium points a, b, and c to the EE schedules. For a, for example, the disequilibria are af and ag, respectively, for EE and E0 E0; af, ag measure, at the interest rate at a, the change in income, with a fixed marginal propensity to import, required to re-equilibrate the balance of payments.28

Figure 2A illustrates some of the policy options for a highly integrated economy faced with a domestic boom. The monetary authorities may try to counter the boom by implementing a restrictive monetary policy, as shown by the shift in the LL schedule to L1 L1. The economy then moves from a to d. This policy is stabilizing, since it attenuates the expansion in income; however, the reserve gain is now larger. The reserve gain may be weakened by a shift in the monetary/fiscal mix to point e when monetary policy is relatively expansionary (to L2 L2) and fiscal policy more contractionary (to X2 X2). At point e the gain in reserves is less than at d but the degree of policy restrictiveness is the same. An alternative policy discussed is to pursue an independent monetary policy and move to point d but to reduce the reserve gain by restricting the inflow of capital (with the E0 E0 schedule now shifting to E01 E01). Finally, the authorities, constrained by external considerations in pursuing stabilizing monetary policies, may hold the economy at a or even allow the surplus to increase the money supply, so that the economy moves to point f. In these last cases, the monetary authorities will have lost some control over domestic activity.

Figure 2.
Figure 2.

Policy Options in an Open Economy

Citation: IMF Staff Papers 1973, 001; 10.5089/9781451956351.024.A001

The same points may be illustrated briefly in Figure 2B for a rise in the foreign interest rate. The rise in the foreign interest rate will move the E0E0 schedule to E01 E01. The same policy restrictiveness may be maintained and the loss of reserves attenuated by changing the monetary/fiscal mix and shifting the economy from c to g. Alternatively, the economy may be allowed to remain at c and the outflow of capital restricted, so that the E01 E01 schedule shifts to E02 E02. Finally, to avert excessive reserve losses, monetary policies may be restrictive (the LL schedule shifts to L1L1) and the economy allowed to move to h, where the level of income is below that considered desirable. Here again, the monetary authorities will have lost some control over the domestic economy.

We may conclude with the following remarks. Sterilization policies to counter massive movements of capital over cyclical phases may well expose the economy to destabilizing speculative attacks. At the same time, greater refinement and flexibility in the use of monetary and fiscal policies to permit increased interest rate harmonization has not been feasible or practical as a means of countering the fluctuations in reserves. Hence, under narrowly fixed exchange rates and highly integrated economies, the real choice may well be between increased interventions in the movement of capital to maintain independent interest rate policies and some loss of control over the domestic economy, or some combination of the two.

It is revealing to illustrate these last points by looking briefly at the experiences of the United Kingdom, France, Italy, the Netherlands, and Belgium in the years 1969-71.29

The years 1969-71, it will be recalled, witnessed huge swings in the Euro-dollar interest rate, the sharp rise during 1969 being followed by an equally dramatic fall during 1970 and early 1971. In most industrial countries, domestic developments were such, during 1969, that restrictive monetary policies were appropriate; it was difficult, however, to match the large increases in the Euro-dollar rate without severely compromising control over domestic activity. In all cases, then, despite increases in domestic rates, the interest differential swung strongly in favor of the Euro-dollar market. Hence, to avert severe loss of reserves, many countries also took special measures to discourage outflows.

During 1970 and early 1971 the difficulties were reversed for these countries. Interest rates were allowed to drop, albeit with different lags. Nevertheless, the interest margin now swung sharply in favor of these countries, provoking an inflow of short-term capital. In many countries during 1970-71, measures were taken to resist the inflows.

The key question is whether, on balance, it is possible to form any judgment as to whether monetary policy was more restrictive during 1969 or more expansionary during 1970 than was appropriate in the light of domestic developments. For 1969 it is possible to make a case for the view that Italy and Belgium (and perhaps France) may have been forced into a more restrictive monetary policy than was considered justified by their cyclical phases. In Italy, monetary policy became restrictive during 1969, even though economic activity continued to be well below the full utilization of capacity.30 In Belgium, the rate of growth in the money supply slowed down sharply in the second and third quarters at a time when the recovery from the earlier recession was still mild and hesitant. In France, the rate of growth of the money supply also slowed down noticeably during 1969; this may have been related in some degree to contemporaneous developments in the external sector, but, on the other hand, it is also true that economic activity did accelerate sharply in the year.

In contrast, during 1970-71 there was concern in the United Kingdom, France, Belgium, and the Netherlands that the capital inflows were threatening the independence of their monetary policies. In France, Belgium, and the United Kingdom, the rate of growth of the money supply accelerated during 1970, but in all three countries economic activity slowed down noticeably during the year. In the Netherlands, with the economy overheated during 1970 and early 1971, monetary expansion actually accelerated quite sharply in that period, suggesting perhaps that independent monetary policies may have been frustrated.

Italy and Belgium in 1969 and the Netherlands (as well as Germany, as we shall see) in 1970-71 were probably fairly unambiguous instances of countries that found it particularly difficult to use monetary policy to control domestic activity in the face of massive movements of short-term capital. In these countries it became necessary at some point to direct monetary policy to external objectives at the expense of domestic targets.

Statistical evidence on interest rate harmonization

One might argue that since neither sterilization policies nor interventions in capital movements is likely to be entirely successful in rehabilitating the independence of interest rate policies, highly integrated economies would tend in part to determine their interest rates in the light of interest rate developments overseas; additionally, an increase in financial integration would be likely to manifest itself in closer links between domestic and foreign interest rates.

We turn, in this part, to the statistical evidence of interest rate harmonization among the industrial countries.31 This evidence can take three forms. First, one might observe the movement over time in some measure of interest rate dispersion for a group of industrial countries. At best this would provide clues as to possible trends in financial integration for a consolidated group of countries, but it would tell us nothing either about developments within any one country or about differences in integration between different countries. Second, one might look at the behavior over time for each country in the uncovered interest differential. Conceivably, this would tell us something about possible trends in integration in individual countries but clearly it could tell us nothing either about differences in integration between countries or about the degree of integration for any one country. Third, one could look at the extent of covariation between domestic and relevant foreign interest rates. A possible advantage of this test is that it might, at the same time, reveal the degree of integration within any one country and trends in integration in individual countries, as well as differences in integration between countries. The evidence from each of these sources will now be reviewed.

Table 1 provides data on means, standard deviations, and coefficients of variation32 for short-term interest rates for ten industrial countries for each year between 1958 and the first half of 1971. The results show that the two measures of dispersion33 have moved in quite similar ways over the period. There is a noticeable drop in dispersion between 1961-62 and 1964-66 but no evidence of any continuing decline in later years. These results, however, should be treated with considerable caution, since measures of dispersion are sensitive not only to movements in individual rates but also to the degree of synchronization in business cycles.

Table 1.

Ten Industrial Countries:1 Annual Dispersion of Short-Term Interest Rates, 1958-712

article image

Belgium, Canada, France, Germany, Italy, Japan, Netherlands, Switzerland, United Kingdom, and United States. Three-month treasury bill rates for Canada, United Kingdom, and United States. Call money rates for Belgium, France, Japan, Netherlands, and Switzerland. Interbank three-month rates for Germany. Medium-term bond rates for Italy. Three-month Euro-dollar deposit rate.

End of June 1971.

Standard deviation divided by mean.

First half.

Chart 2 shows the movement in the uncovered interest rate differentials over the period 1960 to the second half of 1971. A striking feature of the chart is the sharp increases in absolute differentials that occurred for most countries during 1968-71. In large part this was due to the sharp fluctuations in the Euro-dollar rate,34 noted in another section. As we have seen, the movements in the Euro-dollar rate in these years provoked, in part, changes in domestic rates; in part, resort to increased interventions in short-term capital movements; and, in part, massive movements in short-term funds across national boundaries. Interventions in recent years were intended to alleviate the effects of larger interest rate differentials; they actually may have had the effect of reducing the degree of financial integration.35

Chart 2.
Chart 2.

Nine Industrial Countries: Interest Rate Differentials (Euro-Dollar RateLess Domestic Rates), 1960-711

(In per cent per annum)

Citation: IMF Staff Papers 1973, 001; 10.5089/9781451956351.024.A001

1 End of June 1971.

An alternative way to look at trends in differentials is to observe the movement in these differentials without regard to sign as a proportion of the Euro-dollar rate. An increase in financial integration might then manifest itself in a decrease in the proportionate differential. Linear trends were fitted to these proportional differentials. These trends were significant and negative only for Italy, the Netherlands, and Japan. In the United Kingdom, the trend was significant and negative but only when the proportional differential was against the U.S. rate.

Finally, we look at the results of simple correlation analysis. High correlations are not in themselves unambiguous evidence that the economy is highly integrated. Economies may have experienced similar trends in inflation and/or their cycles may have synchronized. Nor is a low correlation necessarily evidence of poor integration, since there may be a lag in adjustment—even a quite variable lag. Also, since interest rates are determined by a number of variables in addition to foreign rates, only a more complete model (outside the scope of this paper) would be able to identify the contribution of the foreign rate in influencing the domestic rate.

Chart 3 shows the movement in interest rates for ten industrial countries over the period 1958-71. Looking at the correlations for these interest rates (shown in Table 2), the correlation coefficients are all quite high, with the exception of Japan. The correlations are particularly high for the United Kingdom, the Netherlands, France, and Canada. Japan’s low correlation is consistent with the poor integration of its money markets with overseas markets, but the correlation for Germany is not high, although Germany is known to be highly integrated with world money markets. The first-difference results, not unexpectedly, are much poorer, all being below 0.5. This means that on a quarter-to-quarter basis the relationship between domestic and U.S./ Euro-dollar rates is weak.

Chart 3.
Chart 3.

Ten Industrial Countries: Short-Term Interest Rates, 1958-711

(In per cent per annum)

Citation: IMF Staff Papers 1973, 001; 10.5089/9781451956351.024.A001

1 United States: three-month treasury bill rate; Germany: three-month money market rate; United Kingdom and Canada: three-month treasury bill rate; Italy: rate on medium-term government bonds; France and Belgium: call money rate; Japan, Netherlands, and Switzerland: call money rate.
Table 2.

Nine Industrial Countries: Quarterly Correlations of National Short-Term Interest Rates1with U.S.2and with Euro-Dollar (ED)3 Rates, January 1958-November 1971

article image

Call money rates for Belgium, France, Japan, Netherlands, and Switzerland. Three-month treasury bill rates for Canada and United Kingdom. Interbank three-month rates for Germany. Medium-term bond rates for Italy.

Three-month U.S. Treasury bill rate.

Three-month Euro-dollar deposit rate.

Correlations of first differences.

Comparing the earlier and later periods, there is evidence of possible increased interest rate harmonization in the later period in Belgium, Canada, France, Japan, the Netherlands, and the United Kingdom, but in Germany, Italy, and Switzerland, the results are less clear cut. However, any increase in interest rate coordination that has been evident in more recent years cannot be taken as conclusive evidence of an increase in financial integration. Rather, it may reflect the fact that Euro-dollar interest rates moved much more sharply than in the past (see Chart 1), and, hence, that increased coordination was necessary to avoid opening up intolerably large differentials.

Not surprisingly, it has been quite difficult to arrive at any definite conclusions. There are, it is admitted, severe limitations in the kinds of statistical test that were carried out. The evidence from dispersion suggests that between 1961 and 1965-66 there may have been some significant increase in financial integration. The evidence from uncovered differentials for individual countries reveals little about trends in integration. The evidence from correlation analyses shows that for most industrial countries, over the whole period, the correlation between national rates and U.S. or Euro-dollar rates has been fairly high but that it was very difficult to draw any inferences about trends in integration in individual countries.

In the following section we explore an alternative method of identifying the degree of financial integration and possible changes in integration. Instead of focusing on interest rate relationships, we focus instead on the effects on the forward premium/discount of changes in the interest rate differentials. Integration here is revealed not by the closeness in the link between interest rates but rather by the strength of the response of the forward premium/discount to changes in the interest rate differentials. Changes in integration manifest themselves not necessarily by closer interest rate linkages but by a more vigorous adjustment in the forward premium/discount. This test is more direct and is somewhat analogous to measuring the degree of sensitivity of capital movements to changes in interest rate differentials.

III. The Arbitrage Function

In this section we apply the theory of forward exchange to the analysis of international financial integration and concentrate in particular on interest arbitrage. A simple model of the forward exchange market is developed36 and is solved for the forward premium, which is explained by a number of variables, including the interest rate differential. The solution is then used as a basis for estimating regressions for the forward premium for selected industrial countries. By focusing on the interest sensitivity of the forward premium, we evaluate, first, the degree of financial integration within a country and possible changes in integration over time, and, second, differences in integration among some of the industrial countries. The countries included in the study are the United Kingdom, Germany, France, Belgium, the Netherlands, and Canada.37

The underlying model

Arbitrageurs’ net supply of three-month forward foreign currency (demand for forward foreign currency, if negative), Af, can be expressed as follows:

(3.1)Aft=Af[4FtRtRt(IhtIft),σt,Σi=t1t89Afi]

and ∂Af//∂[FP - (Ih - If)] > 0; ∂Af/∂σ, ∂Af/∂∑Afi < 0

where F and R are the three-month forward and current spot rates, and FP is the forward premium; Ih and If are the three-month domestic and foreign short-term interest rates taken as exogenous variables; σ is the opportunity cost of holding foreign short-term assets; and Σit1t89Afi stands for the size of past commitments by arbitrageurs.

Speculators’ and traders’ net demand (supply) function for forward exchange, Sf, can be expressed as

(3.2)Sft=Sf[(RePtFPt),Xt*,Mt*,ρt,Σj=t1t89Sfj]

and

Sf/∂(ReP - FP), ∂Sf/∂M*, > 0; ∂Sf/∂X*, ∂Sf/∂ρ, ∂Sf/∂∑Sfj < 0

where ReP is the premium of the expected spot rate in three months over the current spot rate;38 X* and M* are parts of exports and imports, which are financed by short-term capital39 and covered forward; ρ is the marginal opportunity cost of speculative forward commitments; and Σj=t1t89Sfj represents total past commitments by speculators.

The first argument on the right-hand side of equation (3.1)—the difference between the forward premium and the interest rate differential—is the key variable defining the motive for activity by the arbitrageurs. The other two arguments represent constraints on such activity. Under general conditions of uncertainty, large past commitments will reduce the willingness of arbitrageurs to increase further their short-term holdings of assets abroad and thus to supply additional forward cover at the same forward exchange rate. Risk of interference in the freedom of short-term capital movements is a major factor that influences arbitrageurs’ marginal opportunity costs,40 although transactions costs may also cause a discontinuity between arbitrageurs’ demand for and supply of forward exchange.41

The key variable in equation (3.2) is the difference between the expected spot premium, ReP, and the forward premium, FP. Speculators’ past forward commitments and the marginal opportunity cost of present commitments,42 consisting mainly of the foreign exchange risk, act as constraints on speculative commitments in each particular foreign exchange market. Traders’ activity on the forward exchange market is incorporated into equation (3.2), since it is assumed that traders are sensitive to the exchange rate risk. When the difference between the expected spot rate premium and the forward premium is considered to be significantly higher than the marginal opportunity cost, in terms of the risk involved, traders are willing to leave uncovered forward a portion of their short-term transactions.43 Since, however, traders’ participation on the forward market is limited by the size of their short-term trade contracts, variables X* and M* are shift parameters in the function.

The two functions, expressed by equations (3.1) and (3.2), are shown in Figure 3, in which the spot rate is assumed to be given by the monetary authorities. Given the interest rate differential between the two countries, increasing marginal opportunity costs to arbitrageurs (reflecting mainly risk other than exchange risk, which is covered on the forward market) will generate an upward slope of the arbitrageurs’ supply function.44 Likewise, given the premium of the expected spot rate over the current spot rate, increasing marginal opportunity costs to speculators (reflecting mainly the exchange risk) will generate a downward slope of their demand curve.

Figure 3.
Figure 3.

Equilibrium in the Forward Exchange Market

Citation: IMF Staff Papers 1973, 001; 10.5089/9781451956351.024.A001

The equilibrium on the forward exchange market is expressed by equalizing the demand for and the supply of forward foreign currency.

(3.3)Aft=Sft

Using the equilibrium condition (equation 3.3) to solve a linearized form of equations (3.1) and (3.2) for the forward premium, one obtains

(3.4)FP=a1a1+b1(IhtIft)+b1a1+b1RePt+b2a1+b1λ(XtMt)+b3a1+b1ρta2a1+b1σta3a1+b1Σi=t1t89Afi+b4a1+b1Σj=t1t89Sfi

where the interest rate differential, the trade balance, and arbitrageurs’ and speculators’ opportunity costs are exogenous variables, while the total past forward positions of arbitrageurs and speculators are pre-determined variables. Also, the spot rate is assumed to be fixed by monetary authorities.

When the arbitrageurs’ supply function for forward exchange is infinitely interest elastic, in equation (3.1), the coefficient a1 = ∞. In equation (3.4) the coefficient a1/(a1 + b1) = 1, and coefficients of all other variables equal zero. This implies that in such a case the speculators’ and traders’ demand curve determines only the size of forward commitments by arbitrageurs but that the forward premium is determined entirely by the arbitrageurs’ supply function. Changes in forward premium and in the interest rate differential will then have a one-to-one relationship, and no other constraint will operate on equation (3.4).45 When, on the other hand, the speculators’ and traders’ demand function is infinitely elastic, the coefficient b1/(a1 + b1) = 1, and all others equal zero. The forward premium is then independent of interest rate differences, and all other constraints are inoperative. The arbitrage function would serve only to determine the size of forward commitments.

In the intermediate case when 0 < a1 < ∞ and 0 < b1 < ∞, in equations (3.1), (3.2), and (3.4), all constraints that determine the shape and the position of curves Af0 and Sf0 in Figure 3, are operative.

The equilibrium forward premium will then reflect these constraints and will differ from both the interest rate differential, Id (= Ih - If), and the expected spot rate premium, ReP. In Figure 3 the equilibrium forward premium, FP, is between regions B and C on the vertical axis. The difference between B and C and the equilibrium forward premium is inversely related to the respective are elasticities of the two curves, when these are calculated between the vertical axis and the equilibrium point, F.

We could now drop the assumption that the spot exchange rate is fixed by the monetary authorities. With a flexible spot rate, the equilibrium on the forward exchange market (expressed by equation (3.3)) can be sustained only if the net supply of foreign exchange on the spot exchange market, resulting from transactions that are independent of the current activity on the forward market, is equal to the demand for spot currency by arbitrageurs to cover their forward commitments. When demand for spot foreign exchange for purposes other than the current cover of forward commitments by arbitrageurs is expressed in terms of its own independent variables, both spot and forward exchange rates are determined jointly as endogenous variables in the foreign exchange market. The arbitrageurs’ activity on both spot and forward exchange markets plays the key role in this process of determination.46

However, because of difficulties faced when basic balance components are introduced as explanatory variables, the econometric analysis of short-term capital movements generally ignores the relationships expressed in the demand (supply) function for spot exchange. We also followed this procedure and based our analysis on equations (3.1) through (3.4). However, in our discussion of various shift parameters in equation (3.4), we also touched on the effect of changes in the spot exchange rate on the equilibrium of the forward exchange market.

Econometric estimation and implication for financial integration

Equation (3.4) may be used as a basis for econometric work on determining the forward premium. Ideally, we would want to incorporate all the explanatory variables identified in the analysis, focusing particularly on the interest sensitivity of the forward premium as a measure of the degree of financial integration. Unfortunately, a majority of the variables included are not observable. In most countries there are no consistent data (in the form of monthly series) on the size of trade financed by short-term contracts and on the size of outstanding arbitrageurs’ and speculators’ forward contracts; nor is it possible to get any estimate of the opportunity costs of arbitrageurs and speculators (the major component of which is risk of government intervention, for the former, and exchange risk, for the latter).

A major difficulty also arises in the treatment of the expected spot rate. Attempts to formulate a function that would explain speculators’ exchange rate expectations and thus determine their behavior on the forward and spot exchange markets have not been successful.47 What is most important in dealing with this problem is to be able to identify the major shifts in expectations that are associated with periods of crisis, when confidence in the exchange rate parity collapses. During these crises, the relationship between the forward premium/discount and the interest rate differential can become quite complicated. An expected devaluation of the exchange rate will induce speculators to sell domestic currency forward, and this will tend to push the forward rate below the lower intervention point without any accompanying change in the interest differential. Speculators may bring about autonomous changes in the forward premium or discount, which may in turn provoke changes in the domestic interest rate. For example, if speculators attack the domestic currency, opening up a forward discount, the monetary authorities may try to avert the outflow by raising domestic interest rates. In this case we have causation running from the forward premium/discount to the interest differential, and not the reverse. It is possible, too, that a sharp change in the monetary policy stance in response to balance of payments developments, for example, by adopting a very restrictive monetary policy, may in itself create doubts about the viability of the parity, and so provoke speculators’ activity in the forward markets. In this case, the coefficient of response of the forward premium/discount to the interest rate differential will be biased upward by the operations of the speculators.

The relationship between the forward premium/discount and the interest differential in a strong currency country is particularly interesting. Here, a rise in the domestic rate relative to the foreign rate may create expectations of a revaluation of the domestic currency. Instead of the forward premium (discount) on the currency falling (rising) in response to the change in the interest differential, it may actually rise (fall) as a result of speculators’ operations in the forward market. Hence, the relationship will be weak and indeed may even have the wrong sign. An alternative possibility arises where an exogenous expectation of a revaluation opens up a forward premium provoking a capital flow, which in turn—through the liquidity effects—lowers the domestic rate. In this case the sign is right, but the magnitude of the coefficient may depend on the degree to which the domestic interest rate is allowed to adjust. For example, speculators may open up a large forward premium but the domestic rate may drop by only a little, in which case the coefficient will appear quite large, a small interest differential in favor of markets overseas accompanying a large adjustment in the forward premium.

In our work we will try to isolate these periods by the use of dummies, although it must be recognized that dummies cannot capture the intensity of speculation and, hence, serious biases in the estimations may remain. Whenever possible, proxy variables will also be used, reflecting to some degree the intensity of destabilizing speculation.

Other difficulties arise in attempting to explain the expected spot rate when confidence in the parity is not affected. As long as there are random disturbances around parity, the expected spot rate may be quite insensitive to current movements in the spot rate, which is controlled by the monetary authorities. Or, the expected spot rate may change in the same direction but by a smaller magnitude than the movement in the spot rate. Again, the expected spot rate may well depend on the strength of the currency and the length of time that the spot rate has settled at a particular point. For example, suppose that differences in monetary conditions favor the outflow of capital, and the spot rate is allowed to drop to its lower intervention point. After some time, firm expectations may develop that the future spot rate will stay at this lower point. These expectations will be reflected in an infinitely elastic speculative demand curve, with the forward rate maintained by speculation in the vicinity of the spot rate.48 In these conditions, the forward premium/discount would not adjust to the interest rate differential. The relationship between the forward premium/discount and the interest differential would be very weak, even though in effect there would have been no change in the arbitrage function and the degree of financial integration.

Therefore, an improperly specified speculative function, and, particularly, unexplained shifts in the function, may create serious difficulties in estimating the degree of international financial integration. However, these difficulties tend to decrease with an increase in the elasticity of the arbitrageurs’ supply curve for forward exchange. As was seen earlier, the higher the elasticity of arbitrage function, the more it will participate in determining the equilibrium forward premium, and the less will be the speculators’ role in this process. Furthermore, the smaller are unexplained shifts in the speculators’ function, the less chance there is that the short-term equilibrium will occur on the inelastic portion of the arbitrageurs’ supply curve (i.e., beyond point D in Figure 3). Thus, it can be expected that, in principle, better econometric results can be obtained during periods when large shifts in exchange rate expectations, induced by destabilizing expectations, are absent.

The main variables that affect the degree of international financial integration are changes in arbitrageurs’ transactions costs and in the risk of various government measures intervening in the freedom of capital movements. Any government intervention in the foreign exchange market, such as changes in the system of exchange controls or changes in institutional conditions and in monetary policies intended to interfere with short-term capital movements, will tend to affect the interest elasticity of the arbitrage function. Thus, a liberalization of short-term capital movements, which occurred in several industrial countries in the latter part of the 1950s and continued into the early 1960s, could be reflected in an increase in the interest elasticity, while increased government intervention in international capital movements in the United States and the United Kingdom in the middle of the 1960s and in some other countries in the latter part of that decade might be reflected in a decrease of elasticity over that particular period. While liberalization of capital movements late in the 1950s is consistent with increased financial integration, government intervention aimed at controlling short-term capital flows may have resulted from attempts by national authorities to develop instruments that would permit them to cope with the effect of increased financial integration on the mobility of short-term capital. Since such interventions often occur only during periods of exceptional pressure of persistent short-term capital flows on a country’s balance of payments position or on its domestic liquidity, these interventions may cause a substantial decrease in the interest elasticity of flows over that particular period, without affecting it greatly in other periods.

The basic approach in this paper is to use as explanatory variables the international differences in short-term interest rates, trade flows, and dummies for destabilizing speculation. Other variables reflecting specific monetary policies and speculative behavior in individual countries are added where necessary.

The basic regression equation is as follows:

(3.5)FPt=4FtRtRt=c0+c1(IhtIft)+c2(XtMt)+c3Dst+ut

where Ds is a dummy variable equal to 1 during periods with destabilizing speculation in which the forward rate exceeded the intervention points by 0.75 per cent on both sides of parity and was equal to zero in other periods (when the forward rate remained within the band), while u is the error term.

The coefficient c1 approaches unity when the elasticity of arbitrage approaches infinity, indicating perfect mobility of short-term capital movement. Thus, the size of the coefficient c1, which can vary between zero and 1, may be indicative of the degree of international financial integration of a given country’s short-term capital market.

The change in the degree of international financial integration is tested by examining the change over time in the coefficient of the interest rate differential. After introducing a trend t in the regression coefficient c1 in equation (3.5)

(3.6)c1=α0+α1t

the following equation is obtained:

(3.7)FPt=c0+α0Idt+α1Idtt+c2BTt+c3Dst+ut

where Id = Ih – If and BT = X – M. The coefficient α1 is an estimate of the change over time in the response of the forward premium to movements in the interest rate differential.49 If α1 is significant and of the same sign as the coefficient α0, it would imply that over the period covered the financial integration of the country in question had increased.

In the middle of the 1960s, U.S. restraints on capital outflows and the British balance of payments crises may have produced a substantial change in the trend in financial integration. Such a change is tested by exploring in each country the possibility of the break in the trend in about 1965 (and in 1967 for the United Kingdom), and then examining the sign of the new trend after the break.

Substitute in equation (3.7)

(3.8)α1=β0+β1Dt

where D is a dummy variable equaling zero for the period before the break and 1 for the period after the break in the trend.

(3.9)FPt=c0+α0Idt+β0Idtt+β1IdttDt+c2BTt+c3Dst+ut

If significant, β0 indicates the trend before the break, β1 indicates that there was an actual break in the trend, and the sum of the two coefficients indicates the new trend. If the sum of the two coefficients is significant and has a sign opposite to the sign of β0 the trend was reversed after the break.

A hypothesis that a structural change in the system, occurring in the mid-1960s, may have caused a shift in the coefficient of the interest differential without affecting its trend (if any) was also tested.

Substitute in equation (3.7)

(3.10)α0=γ0+γ1Dt

where D is a dummy variable equaling zero before the shift in the coefficients and 1 after the shift. Then it follows that

(3.11)FPt=c0+γ0Idt+γ1IdtDt+α1Idt+c2BTt+c3Dst+ut

where γ1 indicates the shift in the coefficient of the interest differential.

The statistical series used in the regression analysis start with 1958 for Canada, 1960 for the United Kingdom, and 1961 for the remaining four countries. All series ended in March 1971, one month before the outbreak of that year’s exchange rate crisis. It was particularly unfortunate that statistical series of European countries could not be extended further into the past, since it was between the late 1950s and the very early 1960s—the period of exchange and trade liberalization in those countries—that the most important developments in international financial integration occurred.50

As much as possible, we were careful to match the maturities between domestic and foreign interest rates. Usually, domestic interest rates were three-month treasury bill rates or other relevant interest rates. However, whenever three-month interest rates were not available, or did not reflect in their behavior the working of the free market forces, other rates that satisfied the latter condition had to be selected.

Government intervention in short-term capital movements was generally taken as structural shifts, left unexplained in the statistical analysis except as a factor affecting the coefficient of the interest differential. However, when this intervention could be translated in terms of an increase or decrease in the cost of foreign transactions, it was included in the regression analysis. This was particularly relevant for Germany, since that country’s monetary authorities have made extensive use of preferential swap facilities and varying reserve requirements of banks as instruments to influence capital movements in the desired direction.

In view of the suspected instability over time in the relationships expressed by the regression equations, because of possible shifts of the speculative demand function, several variants of the test are used for each country. In the first place, for each country the estimation of the size of the correlation coefficient of the interest differential, of the trend in that coefficient, and of the change in the trend was carried out for (a) the whole period for which data were available and (b) the sub-period presumed to be the least subject to violent shifts in the expectations function. Second, these results are compared with a series of regressions estimated for each successive but overlapping two-year period with 24 observations in each regression equation. Thus, for example, between 1960 and the end of 1970, ten overlapping two-year periods were estimated and the performance of these regression equations was examined. The rationale was that a good part of these two-year periods will be less subject to shifts in the speculative function. As was expected, the regression analysis covering the whole period showed a substantial degree of positive serial correlation of residuals51 in most countries, while in most regression equations, in which the specification was reasonably satisfactory, the fit was much better when two-year periods were considered.

Statistical estimates of these two-year series, for each country, are shown in Appendix II, Table 4.

Regression results

The united Kingdom

Regression equations indicating the response (and trends in the response) of the forward premium to movements in the interest rate differentials and other variables cover the period between January 1960 and either September 1967 or March 1971. The difference between the U. K. local authority temporary loan rate and the Euro-dollar rate was taken as the interest differential variable; both rates have a maturity of three months. As a result of previous investigations by one of the writers,52 the lagged trade balance was taken as a variable reflecting destabilizing speculation. Since the balance of payments crisis in the autumn of 1964, the impact of the U. K. trade balance performance has been reflected in changes in the spot and forward exchange rates for the pound; good results would inspire confidence in the pound, and vice versa. Since data on the U. K. trade balance are available with a lag, and since the formation of exchange rate expectations is also a process requiring time, a trade balance lagged one, two, and three months was initially introduced into the regression equations as destabilizing speculation variables. The time during which the lagged trade balance was assumed to influence exchange rate expectations was limited to a period between October 1964 and March 1971.53 A dummy variable, Ds, was introduced to isolate periods of serious destabilizing speculations in the summer of 1961, in the autumn of 1964, and in several months each year since 1965, in which the forward rate fell to or below the lower intervention point.

The regression equation covering the whole period is as follows:

(3.1.1)FP=0.263(1.09)1.213(7.44)Id+0.019(5.34)Idt0.010(3.50)IdtD(t*=4.32)0.0002(0.10)BT+0.004(1.87)BT1+0.007(3.53)BT2+0.004(1.68)BT30.979(4.06)DsR¯2=0.465SE¯=1.06DW=0.81(135observations,Jan.1960Mar.1971)

where, in addition to the trade balance and a dummy variable for speculation (already explained), the coefficient of Id indicates the initial sensitivity of the forward premium to movements in the interest rate differential. The coefficient of Id·t shows the trend in Id between 1960 and the fall of 1967, and the coefficient of Id·t·D indicates the break in the trend at the time of the devaluation of the pound (D = 0 during January 1960-September 1967 and 1 during October 1967-March 1971). The sum of the two coefficients indicates the new trend in the coefficient of the interest differential after September 1967.

As might be expected, the regression equation shows evidence of severe serial correlation. However, coefficients exhibiting a sufficiently high t-ratio can be accepted as significant with a somewhat wider margin of error. The equation suggests that the sensitivity of the forward premium on the pound to movements of the interest differential was high at the beginning of the period (with a coefficient of Id of approximately 1), but that during the following years this sensitivity showed a decreasing trend.54 There appears to have been a break in the trend in the autumn of 1967, but the coefficient continued to decrease, although at a significantly slower rate.55 The trade balance lagged by two months appears significant, an improvement in the trade balance seeming to have contributed to an increase in the forward premium on the pound. Periods of speculation against sterling’s parity contributed substantially toward an increase in the forward discount on the pound.

When the turbulent years after the devaluation of the pound in 1967 are eliminated, the fit improves considerably. Furthermore, while adjustment for seasonality in the long series tested in equation (3.1.1) did not yield significant results, adjustment of the shorter series showed a seasonal pattern that appeared to be significant in February, August, November, and December.56 Regression results with seasonally adjusted series are as follows:

(3.1.2)FP=0.314(3.56)0.997(19.16)Id+0.0021(1.86)dt0.001(1.70)BT+0.002(3.35)BT20.235(3.69)DsR¯2=0.955SE¯=0.196DW=1.38(93observations,Jan.1960Sept.1967)
(3.1.3)FP=0.290(3.27)0.909(40.92)Id0.001(1.90)BT+0.002(2.75)BT20.198(3.22)DsR¯2=0.954SE¯=0.199DW=1.28(93observations)

Positive serial correlation, although still present, is less severe in both equations. Both speculative variables are significant, but their lower coefficient indicates that during that somewhat calmer period they exerted less influence on the forward premium. Equation (3.1.2) suggests that the regression coefficient of the interest differential exceeded 0.9 at the beginning of the period. The sign of the trend coefficient suggests that, over the period, the coefficient has declined; however, in the seasonally adjusted series the trend coefficient is not significant at the 5 per cent level. When, on the other hand, the series are not seasonally adjusted, the coefficient appears to be significant.57 Thus, it would seem that between 1960 and 1967 the change in the interest arbitrage coefficient does not support the hypothesis that financial integration has increased. On the contrary, a weak case could be made that integration had actually decreased somewhat over the period.

The apparently unchanged degree of international financial integration of the London short-term capital market between 1960 and 1967, and even the possibility of a declining trend in integration, resulted in part from our inability to extend the statistical series back beyond 1960. By using differences in 90-day treasury bill rates as the U.S./U. K. interest differential, we were able to extend the series to the beginning of 1958. This particular interest differential is not a very good indicator of financial flows beyond the first quarter of the 1960s.58 However, when used in the seasonally adjusted short series (i.e., for the period January 1958-September 1967), the test based on this differential suggests an increase in the responsiveness of the forward premium to movements in international interest differences between 1958 and the middle of 1964. There is a break in the trend during the third quarter of 1964, and the new trend, from that date to the beginning of the sterling crisis in 1967, is not significantly different from zero.59

Thus, it would appear that international financial integration had occurred in the United Kingdom mainly during and immediately after the de facto external convertibility of the pound from the late 1950s into the early 1960s. Some increase in controls on capital movements during the crisis of 1961 and severe balance of payments constraints imposed by the rather low level of reserves may have produced a break in the integration process. Measures to prevent capital outflows were intensified during periods of increasing misgivings about the performance of sterling early in the 1960s. However, measures not affecting the free exchange market mechanism, such as the support of the forward pound by monetary authorities, do not interfere with the process of financial integration, either. U. K. monetary authorities intervened extensively on the forward market during the period 1964–67, but it was impossible to quantify the effect of these interventions and to incorporate it in our regression analysis.

Germany

As a result of Germany’s strong foreign exchange position, exchange restrictions and, more particularly, nearly all controls on capital exports were removed by the end of 1957. The remaining restrictions on current and capital transactions by nonresidents and residents were abolished in December 1958 with the declaration of external convertibility of the deutsche mark. Thus, the initial impetus toward international financial integration occurred in the last third of the 1950s, the period for which, unfortunately, no complete statistical data are available.

A salient feature of German monetary policy was the active use of various monetary instruments and controls aimed at influencing that country’s short-term capital movements. Preferential spot/forward swap arrangements to encourage short-term capital outflow were resorted to, particularly in the first half of the period covered. Variations in the reserve requirement ratios of banks for domestic lending and for foreign borrowing also had an effect on short-term capital flows. Finally, occasional interdiction of interest payments by banks on short term liabilities to foreigners served to stem excessive short-term capital flows. The use of some of these instruments was designed to allow a degree of independence for domestic monetary policy, while the use of others was intended to control destabilizing speculative movements. To the extent that they affected forward arbitrage, some of these instruments served as substitutes for changes in the domestic interest rates in affecting the forward premium on the deutsche mark. Therefore, it becomes important, where feasible, to allow explicitly for these instruments in the regression equation. For these reasons we included in our regression equations, as additional explanatory variables, preferential forward swap rates granted to commercial banks by the Deutsche Bundesbank and domestic and foreign reserve requirements.60

The two regression equations covering, respectively, the whole period and the period largely free of destabilizing speculation between August 1961 and October 1967 are as follows:

(3.2.1)FP=0.430(1.169)+0.790(14.84)Id+0.112(3.74)SWA0.057(2.04)RR0.833(5.48)Ds10.121(0.57)Ds2R¯2=0.895SE¯=0.509DW=1.11(123observations,Jan.1961Mar.1971)
(3.2.2)FP=1.665(2.00)+0.449(6.10)Id+0.694(3.25)SWA+0.128(1.94)RR0.016(2.37)FR+0.186(1.65)BTR¯2=0.553SE¯=0.328DW=0.98(75observations,Aug.1961Oct.1967)

where FP is the forward premium or discount on U.S. dollars in terms of deutsche mark; Id is the differential between the German interbank loan rate and the Euro-dollar rate, both of three months’ maturity; SWA is the three-month preferential forward rate on the dollar (expressed in the same terms as FP) granted to German banks by the Bundesbank; RR is the percentage of banks’ reserve requirements against domestic liabilities; FR is the percentage reserve requirement against liabilities to nonresidents; Ds1 is a dummy variable for destabilizing speculative inflows at the time of the revaluation of the deutsche mark in 1961 and 1969, and of the devaluation of the pound in 1967, during speculative pressures during 1968, and prior to the crisis in May 1971; and Ds2 is a dummy variable for speculative outflows after the devaluation of the pound in 1967 and after the revaluation of the deutsche mark in 1969–70.

Equation (3.2.1), covering the whole period 1961-71, suggests a fairly high interest elasticity of arbitrage. The coefficient of 0.8 indicates a rather close relationship between the German/Euro-dollar interest differential and the forward premium on the dollar. Forward swaps granted by the Bundesbank are also significant and have the right sign.61 However, when one compares equations (3.2.1) and (3.2.2)—the latter being limited to the nonspeculative period before the U. K. devaluation in 1967—the interest coefficient over the earlier period is considerably lower, while the coefficient reflecting the effect of Bundesbank swap activities on the dollar forward premium is substantially higher. These results are consistent with a much more active swap activity of the Bundesbank in the first half of the 1960s. A dummy variable for destabilizing speculative inflows is significant and shows that expectations of the deutsche mark revaluation tended to substantially depress the dollar forward rate. A dummy variable for speculative outflow, however, is not significant, perhaps because of the infrequency of such outflows. In view of sizable serial correlation of residuals in both equations, the level of significance of domestic and foreign reserve requirements is doubtful. The sign of the latter is as expected, while the negative sign of the coefficient of the domestic reserve requirements variable in equation (3.2.1) can be consistent only if one assumes that an increase in reserve requirements during speculative periods generates capital inflows and contributes to expectations of an exchange rate adjustment. The trade balance was not significant as an independent variable and was omitted from the first equation. Tests for seasonal adjustment of the series did not yield significant results.

A substantial increase in the coefficient of the interest rate differential, occurring when the period tested is extended to include the speculative years after the autumn of 1967, suggests a structural change in the regression equation. Attempts to find a trend in the interest coefficient did not yield significant results. A regression equation based on equation (3.11), which assumed a structural shift in the coefficient of the interest rate differential at the time of the British devaluation in 1967, yielded the following results:

(3.2.3)FP=0.658(1.88)+0.469(5.01)Id+0.413(4.04)IdD+0.094(3.28)SWA0.075(2.78)RR0.667(4.48)Ds1+0.074(0.36)Ds2R¯2=0.907SE¯=0.48DW=1.00(123observations)

where D is a dummy variable reflecting an upward shift of the interest differential coefficient, Id, in June 1967.

Equation (3.2.3) is broadly consistent with equations (3.2.1) and (3.2.2). It suggests a shift from an interest differential coefficient with an average level of 0.5 before 1967 to a coefficient with an average level of about 0.9 after 1967. Seasonal adjustment of the series was marginally significant; the adjusted interest differential coefficient was 0.6 before 1967 and remained 0.9 after 1967. Thus, a shift in the coefficient would be somewhat smaller when the series are seasonally adjusted. These results are also broadly consistent with the regression analysis based on consecutive but overlapping two-year series, presented in Appendix II, Table 4. All two-year equations before 1967 have interest differential coefficients that are considerably lower than the equations that cover the period between 1967 and 1970. The coefficient of the equation for 1966–67—a period during which the shift is supposed to have occurred—is not significant, and the equation has a poor fit.

It would be very hazardous to presume that the upward shift in the interest differential coefficient late in the 1960s reflected an actual increase in international financial integration of the German short-term capital market. After November 1967 the exchange markets of industrial countries were subjected to a series of destabilizing disturbances in which Germany was the major strong currency country. It has been pointed out on page 42 that destabilizing speculative inflows into Germany generating a substantial discount on the dollar (premium on the deutsche mark) may be followed by a liquidity induced decrease in the German interest rate. In a regression analysis the reverse causation between the forward premium and the interest rate differential may yield a high but spurious regression coefficient of the latter. It is possible that after 1967 the strength of the deutsche mark compared with other major currencies increased the arbitrageurs’ awareness of the attractiveness of the German short-term capital market. However, the fact that the coefficient of the interest rate differential in the two-year regression equations for Germany in 1967-68, 1968-69, and 1969-70 (Table 4 in Appendix II) was in all cases 1.00 or higher suggests a strong reverse causation induced by speculative pressures.

France

Over the period of the 1960s, French international short-term capital transactions were mostly subject to government controls. These controls have been modified several times. Interest payments on short-term deposits by foreigners were restricted during 1963 in order to discourage undesired capital inflows; substantial liberalization occurred in 1967 but was rescinded and replaced by stricter controls after massive outflows in the summer of 1968 and the devaluation in 1969. These interventions represented structural changes in the system and substantially affected the demand and supply functions for forward foreign exchange.

Nevertheless, it is interesting that the regression equation used to explain the forward premium on the dollar as a function mainly of the French/Euro-dollar interest rate differential has a relatively good fit.

(3.3.1)FP=0.282(6.79)+0.641(16.36)Id0.058(0.79)BT+0.406(2.85)Ds10.293(1.79)Ds2R¯2=0.737SE¯=0.336DW=1.36(123observations,Jan.1961Mar.1971)
(3.3.2)FP=0.378(5.42)+0.704(16.13)Id+0.074(0.74)BTR¯2=0.768SE¯=0.280DW=1.32(80observations,Jan.1961Aug.1967)

where FP is the forward premium on the U.S. dollar in terms of French francs; Id is the difference between the French call money rate and the three-month Euro-dollar rate; and BT is the trade balance. Equation (3.3.1), which covers the ten-year period, also contains two dummy variables for destabilizing speculation: Ds1 for the speculative outflow during the French crisis in 1968 and Ds1, for the speculative inflow that preceded the British devaluation in 1967 and immediately followed the French devaluation in 1969. Tests for seasonality in the series did not appear significant.

The French/Euro-dollar interest differential appears to explain about 65-70 per cent of the movement of the forward premium on the dollar. While the trade balance was not significant as an explanatory variable in the equations, a dummy variable (Ds1) showed a sizable upward pressure on the forward premium on the dollar by speculative outflows during the crisis in 1968. A dummy variable, reflecting downward pressure—generated by speculative inflows—on the dollar forward premium, did not prove significant. This reasonably good relationship in the 1960s between the forward premium and French/Euro-dollar interest differential is not surprising, despite the exchange control system and its modifications over the years. Although before the summer of 1968 France was not plagued with destabilizing speculative flows and had a good trade and foreign exchange performance, the potential for such destabilization, caused by the public’s long memories of recurrent currency crises in the 1940s and 1950s, required an economic and monetary policy within fairly rigid balance of payments constraints.62

Attempts to fit a trend in the correlation coefficient of the French/Euro-dollar interest rate differential did not yield significant results. However, it is interesting that when the French/U.S. interest differential was used as an independent variable, a regression equation was fitted that showed a significant increasing trend in the interest differential coefficient by the end of 1965, at which point the trend was reversed.63 Although the explanatory power of this regression equation is worse than the one in which the Euro-dollar rate is used, the analysis suggests evidence of increasing financial integration between France and the United States from 1961 to 1965. However, U.S. restraints on short-term capital outflow in 1965 and particularly since 1966 have had a negative effect on the integration process.

Regression equations, shown in Appendix II (Table 4), which cover successive two-year periods between 1961 and 1970, indicate substantial variations in the regression coefficient for the interest differential from period to period. A substantial decrease in the size of the coefficient and of its level of significance in the regression equations for the periods 1962-63 and 1963-64 is associated with measures, instituted by the French monetary authorities in August 1963, to discourage capital inflows. These measures limited the size and the maturity of French banks’ borrowing from nonresidents and the maximum interest that could be paid on such loans. The regression coefficient in the other years between 1961 and 1966 was about 0.7 or higher, and the level of significance was generally higher. On the other hand, between 1966 and 1970 the coefficient never reached 0.6.

Belgium and the Netherlands

Both countries were open to strong external influences on domestic liquidity. Heavy dependence on foreign trade and increased capital mobility, as a result of the external convertibility of their currencies since late in the 1950s, forced the extensive use of existing, and the development of new, monetary instruments to insulate in some degree their domestic economies from external flows.

Since 1960, Belgium has had a free fluctuating foreign exchange market for financial transactions. Since the links between the official and financial markets were not severed completely, the movements of the official and financial foreign exchange rates were close. However, for a certain period of time substantial divergence between the two rates would occur before the market forces or government intervention would act to close the gap again. Such divergences occurred in 1963, during part of 1965, throughout most of 1966, and the gap widened particularly after the second half of 1968. Since data on forward rates are available only for the official exchange market, regressing of the forward premium on the interest rate differential obviously yields biased results. This bias was particularly large during the years of substantial international speculative pressures following the devaluation of the pound in November 1967.

The most representative regression equations for Belgium are as follows:

(3.4.1)FP=0.941(1.23)+0.817(4.02)Id+0.008(0.83)Idt0.014(1.70)IdtD+0.014(0.46)BT0.677(3.14)DsR¯2=0.372SE¯=0.674DW=0.65(123observations,Jan.1961Mar.1971)
(3.4.2)FP=0.034(0.86)+0.697(8.78)Id+0.013(3.53)Idt0.16(4.96)IdtD(t*=1.70)+0.016(0.88)BTR¯2=0.803SE¯=0.251DW=1.42(80observations,Jan.1961Aug.1967)

where FP is the forward premium on the U.S. dollar; Id is the difference between the rate on Belgian four-month Fonds des Rentes and the 90-day Euro-dollar rate; D (0 during January 1961-December 1964 and 1 during January 1965-March 1971) is a dummy variable indicating the break of the trend; and Ds is a dummy variable for destabilizing speculation that generates capital inflows, mainly in relation to the U. K. devaluation in 1967 and to the currency crisis in May 1971. The equation (3.4.2) was adjusted for seasonality.

The fit of equation (3.4.1), which covers the whole period from 1961 to the spring of 1971, is poor. The explanatory power is low, and the only reasonably significant variables are the interest differential and a dummy for destabilizing speculative inflows. There is a serious problem of serial correlation of residuals. Equation (3.4.2), which covers the period prior to the devaluation of the pound in 1967, has a better fit. The coefficient of the interest rate differential was about 0.7 at the beginning of 1961 and was following an upward trend (variable Id·t) until the end of 1964. There was a significant break in that trend at the end of 1964 (variable Id·t·D), which probably reflects the beginning of the British balance of payments crisis and U.S. constraints on capital outflows. The new trend in the correlation coefficient (which is obtained by summing up the coefficient of the two variables just mentioned) appears to be declining at a slow rate, but it does not appear significant. The only other variable that is not significant is the trade balance.

It would therefore appear that there was an increase in the international financial integration of Belgium until the beginning of 1965, but—at least when it is explained by an ex post response of capital movements to interest rate differentials—it may have remained unchanged from that time on.

In the Netherlands, instruments used by the monetary authorities to prevent any undesired effect of short-term capital movements on domestic liquidity included exchange controls on foreign transactions by non-bank residents and on the liability and asset positions of the banking sector. While the controls on nonbank transactions were fairly strict, particularly on the acquisition of foreign short-term assets, control of the banks’ net and foreign position was more flexible and was often used in a countercyclical fashion. Furthermore, especially in the latter part of the 1960s, the monetary authorities intervened more or less heavily in the forward exchange markets. This policy of frequent government intervention in short-term capital movements partly accounts for the rather low explanatory power of the regression equations.

(3.5.1)FP=0.394(2.42)+0.564(7.39)Id0.002(2.66)Idt+0.0005(1.72)BT0.854(5.65)DsR¯2=0.555SE¯=0.498DW=1.11(123observations,Jan.1961Mar.1971)
(3.5.2)FP=0.414(2.89)+0.585(8.00)Id0.004(3.08)Idt+0.0006(1.47)BTR¯2=0.521SE¯=0.340DW=1.68(72observations,Sept.1961Aug.1967)

where Id is the difference between the Dutch three-month treasury bill rate and the three-month Euro-dollar rate, and Ds is a dummy variable reflecting speculative inflows at the time of the revaluation of the guilder and the deutsche mark in 1961, of the British devaluation in 1967, of the deutsche mark revaluation in 1969, and prior to the crisis in May 1971.

Equation (3.5.2), which is seasonally adjusted,64 has a better Durbin-Watson statistic, and the true standard errors of the coefficients are probably considerably higher in equation (3.5.1). In the latter, the speculative dummy variable, Ds, has a t-ratio that is sufficiently high to be significant even with serial correlation. It indicates a substantial effect on the forward premium on the dollar of destabilizing speculative inflows related to the revaluation of the guilder and the deutsche mark in 1961, the devaluation of the pound in 1967, the revaluation of the deutsche mark in 1969, and the currency crisis in the spring of 1971.

Both equations show a fairly low regression coefficient of the initial interest rate differential in 1961. Furthermore, the coefficient appears to have followed a declining trend, which would indicate a decrease in the international financial integration of the Netherlands during the 1960s, probably as a result of government intervention in short-term capital movements.

It is interesting that explorations with consecutive two-year regression equations, shown in Appendix II (Table 4), suggest that between 1968 and 1971 both the Netherlands and Belgium had a very poor relationship between interest differentials with the Euro-dollar rate and their respective forward premiums. This would indicate that they were particularly subject to vagaries of speculative flows during this turbulent period and that they took measures to minimize the effect of these flows on their domestic economies.

Canada

Any empirical analysis of the determinants of the Canadian forward rate65 is faced with substantial difficulties. First, the period for which data are available spans two flexible exchange rate regimes during 1958-62 and 1970-71 and a regime of fixed parities between 1962 and 1970. Since speculative behavior differs in an essential way in these two regimes, the regression analysis covering the whole period 1958-72 will yield biased results. Second, serial correlation of residuals, already observed in the statistical analysis of Canada by earlier students, cannot fail to be substantial under such conditions, and noise in the regression analysis, particularly during periods of various speculative pressures, is considerable.

These reasons prompted us to test separately for Canada the period of the flexible exchange rate regime (1958-61), in addition to testing the whole series and the series covering only the less turbulent period before the devaluation of the pound in 1967. Furthermore, because of the substantial effect of the U.S. program of capital restraints that was initiated in 1965, data from this year are eliminated from the series used.

There was always a fairly close financial relationship between Canada and the United Kingdom. Furthermore, Canada’s financial market during the 1960s had an arbitrage role between the United States and the Euro-dollar market in London. In order to capture this relationship, the Canada/U. K. interest differential and the dummy variable for U. K. destabilizing speculation were introduced into the regression equations.

The results obtained are as follows:

(3.6.1)FP=0.022(0.22)0.841(8.00)Idus+0.0021(1.28)Idust0.066(2.07)Idukl0.0001(0.19)BT0.182(1.79)DsukR¯2=0.578SE¯=0.454DW=0.46(147observations,Jan.1958Dec.1964andJan.1966Mar.1971)
(3.6.2)FP=0.709(1.09)0.858(11.50)Idus+0.003(2.26)Idust0.061(2.92)Idukl0.0004(0.78)BTR¯2=0.791SE¯=0.275DW0.90(102observations,Jan.1958Dec.1964andJan.1966June1967)
(3.6.3)FP=0.332(3.08)0.957(7.70)Idus+0.017(3.52)Idust0.116(3.64)Idukl0.002(1.86)BTR¯2=0.763SE¯=0.298DW=1.14(41observations,Jan.1958May1961)

In all three regression equations, FP is the forward premium on the Canadian dollar in terms of U.S. dollars, Idus is the Canada/U.S. differential of three-month treasury bill rates, Idukl is the differential between Canada’s treasury bill and the U. K. local authority three-month rates,66 and Dsuk is a dummy variable for U. K. speculative pressures. Since Canada’s exchange rate is expressed in terms of U.S. dollars, an increase in the interest rate in Canada would generate a forward discount, so that the coefficients of interest differentials are expected to have a negative sign.

As expected, the results of regression equation (3.6.1) covering the whole period 1958-71 (with the exception of excluded data for 1965) are poor. The explanatory power of the equation is low, and the Durbin-Watson statistics indicate very serious serial correlation of residuals. When the period of speculative pressures that started with the British devaluation in 1967 is omitted, the fit improves considerably. Equation (3.6.2) shows the results of the regression analysis covering the period from the beginning of 1958 to the middle of 1967 (again omitting data for 1965), while equation (3.6.3) covers only the period of flexible exchange rates between 1958 and the middle of 1961.67 Although still plagued with serial correlation, both equations suggest a high initial regression coefficient of 0.85-0.95 at the end of the 1950s, but they also suggest a decrease of such a coefficient over the following years. This would imply a decrease in international financial integration early in the 1960s. As regards the flexible exchange rate period tested in equation (3.6.3), such a decrease is sizable and probably indicates that, despite eliminating data since the latter part of 1961, the effect of government interference in the exchange market toward the close of the period was not eliminated completely. Equation (3.6.2), in which the period tested is extended to mid-1967, also has a trend that shows a somewhat smaller rate of decrease of the interest differential coefficient. This may have been influenced by U.S. measures to restrain capital outflow, introduced in 1963, and U. K. controls, introduced in 1964, which were followed by substantial speculative pressures. A possibility was explored that there was a break in the trend in 1962, when a fixed exchange rate system was adopted, and in 1965. However, statistical tests of these hypotheses did not prove to be significant.

In addition to the Canada/U.S. interest differential, the forward premium on the U.S. dollar also appears to be affected by the movements of the Canada/U. K. interest differential. The influence seems to be somewhat larger during the flexible exchange rate regime of 1958-61, but it also appeared significant in later years. The balance of trade was marginally significant during the flexible rate regime but was not significant for the period including later years.

Concluding remarks

Our regression analysis for the six industrial countries leads to several interesting, albeit tentative, conclusions.

1. As expected, decidedly the dominant variable in all our tests proved to be the interest differential. Trade balance was a significant explanatory variable in some of the countries,68 while in Germany some monetary instruments that were used to influence short-term capital movements also proved significant. However, the most important variables, other than the interest differential, were the proxy variables for speculation. Regression results for the period before the British devaluation—a period largely free of substantial speculative pressures—were generally much better than the results based on series that included the later turbulent years.

2. Table 3 summarizes the results for the coefficient of the interest differential69 and can also serve as a rough basis for an intercountry comparison of the degree of financial integration. The comparison is rough because, despite our efforts, speculation is not included in the regression analysis in an entirely satisfactory way. As expected, the United Kingdom and Canada appear to have the most highly integrated short-term capital markets, followed by Germany,70 Belgium, and France. Controls on capital movements in the Netherlands appear to have been the most effective, so that the degree of its integration seems to be the lowest of the six countries.

Table 3.

Six Industrial Countries: Summary of Results for the Coefficient of Interest Rate Differential1

article image

A t-ratio is given in parentheses.

U.K./U.S. Treasury bill differential.

Jan. 1958-Dec. 1964 and Jan. 1966-Mar. 1971.

Flexible rate period.

Another regression equation shows a shift in the coefficient in June 1967 with the average coefficient of 0.50-0.60 before that date and of 0.90 thereafter.

French/U.S. interest differential.

3. Table 3 also summarizes the tests that were used to evaluate changes in international financial integration during the 1960s. Although tests for the United Kingdom suggest a significant increase in financial integration of European countries during the late 1950s, no such general conclusion can be reached for the major part of the 1960s. Only the results for Belgium suggest an increase in the elasticity of arbitrage between 1961 and 1965, reflecting a probable increase in integration. Results for France suggest an increase before 1965 in integration of the French short-term capital market with the United States but not with the more meaningful Euro-dollar market. In Canada and the Netherlands, on the other hand, there is a suggestion of a decrease in financial integration over the 1960s. In the United Kingdom no change appears to have occurred over that period, although there may have been an actual decrease in integration starting with the second quarter of the decade. Finally, the results for Germany show no significant change until the British devaluation in 1967 and then a substantial upward shift during the last three years of the decade; this shift, however (as already pointed out), also may have resulted from an increase in speculative pressures.

These results may not be surprising when examined within the context of institutional developments in the 1960s. A de facto external convertibility of most European currencies during the latter part of the 1950s generated a major increase in financial integration among industrial countries. However, in response to the effects of the increased capital mobility, many countries adopted various measures to restrict capital movements—starting with controls by the United Kingdom and the United States at the beginning of the second quarter of the 1960s—and these measures were intensified over the years. The process of international financial integration, therefore, depended upon the extent to which controls were used to undo the increasing short-term capital mobility that resulted from external convertibility. Finally, large shifts of short-term capital assets among countries after the British devaluation in 1967 did not necessarily reflect a process of increasing integration, but it did reflect exchange rate expectations by speculators during this turbulent period.

IV. Summary and Conclusions

We began by developing a simple model of the Euro-dollar market and then attempted, with the aid of econometric analysis, to explain the movement in the Euro-dollar interest rate. Our finding was that the movement in the Euro-dollar rate tended to be dominated by conditions in the United States, most particularly by U.S. short-term interest rates and the effects of Regulation Q. Not surprisingly, we found that the U.S. and Euro-dollar markets are highly integrated, although our results also lend some support to the view that the Euro-dollar rate may also be influenced by conditions in Europe, particularly by bursts of speculation.

We explored the relationship between financial integration and interest rate harmonization. We began by examining the effects of financial integration on the volatility in reserves over a country’s business cycle. We showed that, initially, increases in financial integration tended to be stabilizing to reserves, since capital movements tend to offset trends in the current account. Beyond a point, however, integration will destabilize reserves over a cycle. In a fixed exchange rate regime, countries will then be faced with difficult policy options. They may continue to pursue independent interest rate policies by sterilizing the liquidity effects of reserve movements or by increasing their interventions in capital movements. Since neither sterilization policies nor interventions in capital movements is likely to be entirely successful, highly integrated economies might also tend in part to determine their interest rates in the light of interest rate developments overseas. We then looked at the statistical evidence on interest rate harmonization to see if it could reveal something about the degree of financial integration and possible changes in integration over time in individual countries. These statistical tests have severe limitations, and, not surprisingly, the evidence from them proved quite inconclusive.

We looked at the responsiveness of the premium and discount in the forward market to changes in interest differentials as an alternative way to evaluate the degree of financial integration and possible changes over time. With integrated money markets, a rise in the interest differential in a country’s favor will induce the movement of arbitrage funds, which will have the effect of increasing (reducing) the forward discount (premium) on the domestic currency. The greater this sensitivity, the more integrated may be the economy; additionally, changes in the degree of sensitivity may reflect changes in integration over time. To apply this particular test, it was necessary first to develop a simple model of the forward market, from which a reduced form solution for the forward premium was derived. The forward premium was explained by a number of variables, including the interest rate differential. The solution was then used as a basis for econometric estimations of the determinants of the forward premium. In our results for selected industrial countries, we focused particularly on the coefficient of the interest differential as a measure of the degree of financial integration and changes in that coefficient over time as possible indicators of trends in integration. The coefficient was highly significant in all the countries; its size also suggested a substantial degree of integration. There was also some evidence of possible increases in integration between 1961 and 1965 in Belgium and France, but in Canada and the Netherlands our results suggested that there may have been a decline in integration over the 1960s. In the United Kingdom, there was some increase in integration late in the 1950s but no apparent change thereafter, while in Germany no trends in integration were observable.

Again, another way of measuring the extent of financial integration and interest rate linkages among the industrial countries would be by econometric analysis based on a more or less refined macromodel of an open economy.71 Consider a macromodel that is solved for net capital movements and for the domestic interest rate. Net capital movements would then be explained by a number of variables, including an indicator of domestic monetary policy (e.g., the change in domestic assets adjusted, where necessary, for changes in reserve requirements). The unbiased coefficient for the monetary policy indicator72 would then be a measure of the extent to which acts of monetary policy were being offset by capital movements. This offset coefficient would be expected to be larger, the more integrated the economy. Tests of possible changes over time in the size of the coefficient could also be carried out. At the same time the domestic interest rate could be explained by a number of variables, including the foreign interest rate. The significance and size of the coefficient for the foreign interest rate could reveal the extent to which the domestic rate responded to the foreign rate.

APPENDICES

I. Germany: A Case Study

The point was made that a strong currency, in conditions of highly integrated capital markets, may be confronted with particular policy dilemmas during a boom. In this phase, monetary restraint—appropriate for domestic conditions—’ may at the same time provoke capital inflows, strengthening still further the reserve position.73 The accumulation of reserves in time may well generate expectations of exchange rate changes, which in turn will induce further capital inflows. In contrast, as was pointed out, when economic activity is slack, the current surplus will tend to be offset by capital outflows provoked by easier domestic monetary policies.

Some of these points may be illustrated briefly by referring to the German experience between 1958 and 1971. Germany is a particularly interesting case because over this period its money markets were integrated closely with those abroad; at the same time, the deutsche mark was a strong currency in the exchange markets for almost the whole of this period.

Tight liquidity periods (1958-71)

It is possible to identify three periods during which liquidity could be said to be tight. Two criteria were used in determining monetary tightness: one was the behavior of the three-month interbank rate, and the other was the seasonally adjusted quarterly rate of change in the money supply. The three periods of tightness, covering whole quarters, based on the two indicators are as follows:

article image

Turning points in economic activity corresponding to these three tight periods were as follows:75

March 1959 (trough) to March 1961 (peak)

September 1963 (trough) to March 1965 (peak)

March 1965 (peak) to June 1967 (trough)

June 1967 (trough) to March 1970 (peak)

Continuing high capacity to June 1971

First tight period (1959-60)

The two indicators synchronize perfectly, so that there is no problem in identifying the tight period. During this whole period the basic balance was very strong, with the current account showing a large surplus and the net long-term capital account recording a small outflow. The interest differential (the German interbank rate less the U.S. Treasury bill rate) was negligible in the quarter ended December 1959 but began to rise sharply during the second quarter of I960.76 During the second half of 1960, there was a premium on the forward deutsche mark in the open market, providing an additional inducement to inflows of short-term capital into Germany. This forward premium rose sharply early in 1961, indicating considerable speculation in this period.

Coinciding with the sharp improvement in the interest differential in favor of Germany, there was a heavy inflow of capital in the second quarter of 1960. In the middle of the year the monetary authorities took a number of measures to stem the strong inflow, but it continued into the third quarter. In the end, the authorities decided that the attempt to maintain a restrictive stance, superimposed on a strong payments position, was becoming self-defeating, since offsetting the domestic liquidity effects of reserve accruals only perpetuated the inducements to import capital. So, from November 1960, the stance of monetary policy changed; the discount rate was lowered, bringing other short-term rates down with it, and the rate of growth in the money supply was allowed to accelerate in due course.

It is clear from the foregoing discussion that, despite the fact that economic activity was accelerating and at a high level by the end of 1960, the monetary authorities switched to an expansionary policy. External considerations were allowed to dominate interest rate policy. It became possible to reassert the role of monetary policy only when the German authorities revalued the deutsche mark in March 1961. Although the peak in economic activity was reached in the quarter ended March 1961, economic activity continued relatively high throughout that year. Some time after the revaluation, interest rates rose again and the rate of growth in the money supply decelerated.

Second tight period (1964-66)

Although interest rates began to rise from the last quarter of 1964, the rate of increase in the money supply did not begin to decelerate until about the second quarter of 1965. Monetary conditions continued to be relatively tight until about the last quarter of 1966. Capacity utilization in manufacturing, however, had apparently reached a peak in the first quarter of 1965, after which it fell steadily and noticeably to a low level in the second quarter of 1967. The current account turned sharply into deficit late in 1964, and large deficits persisted until late in 1966.77 As domestic rates rose, the interest differential began to swing significantly in favor of Germany. Short-term capital movements appear to have responded strongly to these developments, strong inflows synchronizing well with quarters when the interest margin in favor of Germany was largest. No significant measures were taken during this period to discourage these inflows.

During this episode a restrictive monetary stance was facilitated by two factors. One is the fact that foreign rates were also rising, allowing the German authorities to push interest rates up fairly sharply. The apparent margin in favor of Germany was neither as high nor as persistent as in the first restrictive episode.78 Another is the fact that, with large deficits emerging in the basic balance, short-term capital movements were stabilizing for reserves rather than destabilizing as they had been in the earlier episode.

One may ask why the monetary authorities persisted with their restrictive monetary policy apparently well beyond the point when economic activity had begun to slow down. It is possible, of course, that they misjudged the extent of the downturn. There are, however, two other possibilities. One is that the rate of inflation was high and persistent right up to the middle of 1966, and the authorities may have been guided more by this development than by the levels of capacity utilization in industry. Another is that the sharp rise in interest rates was dictated at least in part and in its later stages by the deficits in the basic balance. If this is so, it is conceivable that interest rates later in 1966 remained higher than appropriate on domestic considerations.

Third tight period (1969-70)

Depending on the indicator used, monetary conditions tightened some time in the first half of 1969 and continued into middle or late 1970. Since most of 1969 was dominated completely by crises in foreign exchange markets, beginning with the French devaluation in August and ending with the German float in September, it is convenient to take up the story when markets became more settled early in 1970.

During 1970 economic activity reached high levels. Although the peak was probably passed after the first quarter,79 activity remained high throughout 1970 and into the first half of 1971. The current account, still in surplus, had weakened, however, during 1970 in contrast to 1969, and with outflows on the long-term account the basic balance remained roughly in balance through 1970 and early 1971. With Euro-dollar and U.S. rates beginning to drop early in 1970, the interest differential swung sharply in favor of Germany from the first quarter of the year and persisted into 1970 and the first half of 1971. The inducements to import short-term capital were aggravated in 1971 by speculation, as indicated by a premium on the forward deutsche mark coinciding with a favorable interest differential. The strong inflows persisted despite special measures to resist them. By July 1970 the monetary authorities had already decided to lower domestic interest rates, and by the third quarter of the year the rate of growth in the money supply had begun to accelerate. However, since interest rates in the U.S. and Euro-dollar markets were falling sharply, the net incentives in favor of Germany persisted. In May 1971 the German authorities ceased to maintain the exchange rates for the deutsche mark within the established margins. There is little doubt that from about the third quarter of 1970 interest rate policy in Germany was becoming dominated by external developments, despite signs that the domestic economy continued to be overheated.

The 1970-71 episode bears comparison with the 1960-61 experience. In both cases tight monetary conditions were superimposed on a strong reserve position; whereas in 1960 the basic balance account was in strong surplus, in 1970-71 it was roughly in balance. On both occasions, rather similar measures were taken to counter the inflows. A striking feature of both periods is the heavy resort by nonbank corporations to overseas sources of funds, although this was probably more intense in 1970-71. In the two episodes the interest differential was strongly in favor of Germany, and the inflows were fed by speculation; the margin in favor of Germany, however, was more persistent and possibly larger during 1970-71, and speculation was more intense. Although there had been some speculation against the dollar in 1961, this assumed more serious proportions in 1971. Indeed, whereas at the height of the inflows in 1960 (in the six months March to September) the amount involved was about $1.2 billion, this total was far exceeded in the later period, approximating nearly $2 billion a quarter, on average, during 1970 alone. After both episodes the monetary authorities allowed the exchange rate to move upward—in the earlier episode by a change in the par value, in the later episode (initially, at least) by floating.

The expansionary period (1966-67)

Toward the end of 1966, monetary policy became expansionary. During the expansionary phase, which continued until middle or late 1967,80 economic activity reached low levels. The current account had swung into a large surplus by late 1966, a surplus that persisted through this episode. Long-term outflows were substantive, but the basic balance remained in surplus. Interest rate differentials swung sharply in favor of the Euro-dollar market by 1967, and during this period there were large short-term capital outflows. These outflows were stabilizing insofar as reserves were concerned, moderating the accrual in reserves that originated in the current account. It would appear that in this phase interest rate policy was not seriously constrained by external considerations.

II. Statistical Table

Table 4.

Forward Premium: Two-Year Regression Equations

For meaning of symbols, see equation (3.1.1).

For meaning of symbols, see equations (3.2.1) and (3.2.2).

For meaning of symbols, see equations (3.3.1) and (3.3.2).

For meaning of symbols, see equations (3.4.1) and (3.4.2).

For meaning of symbols, see equations (3.5.1) and (3.5.2).

Idukt is the Canada/U. K. differential of three-month treasury bill rates. For meaning of the remaining symbols, see equations (3.6.1) through (3.6.3).

REFERENCES

  • Argy, Victor, Monetary Policy and Internal and External Balance,” Staff Papers, Vol. XVIII (1971), pp. 50827.

  • Argy, Victor, and Pentti J. K. Kouri, Sterilization Policies and the Volatility in International Reserves,” presented at the Conference on National Monetary Policies and the International System, Racine, Wisconsin, July 1972.

    • Search Google Scholar
    • Export Citation
  • Arndt, Sven W., International Short Term Capital Movements: A Distributed Lag Model of Speculation in Foreign Exchange,” Econometrica, Vol. 36 (1968), pp. 5970.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bank of England, “Overseas Sterling Holdings,” Quarterly Bulletin, Vol. III (1963), pp. 26478.

  • Black, Stanley W., An Econometric Study of Euro-Dollar Borrowing by New York Banks and the Rate of Interest on Euro-Dollars,” The Journal of Finance, Vol. XXVI (1971), pp. 8388.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Branson, William H., Financial Capital Flows in the U.S. Balance of Payments (Amsterdam, 1968).

  • Branson, William H., The Minimum Covered Interest Differential Needed for International Arbitrage Activity,” Journal of Political Economy, Vol. 77 (November/December 1969), pp. 102835.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Clark, Peter B., and Michael W. Keran, Predicting the Euro-Dollar Rate and the Impact of Government Intervention in the Euro-Dollar Market” (unpublished, Federal Reserve System, 1972).

    • Search Google Scholar
    • Export Citation
  • Cooper, Richard N., Towards an International Capital Market?” in North American and Western European Economic Policies: Proceedings of a Conference Held by the International Economic Association, ed. by Charles P. Kindleberger and Andrew Shonfield (New York, 1971), pp. 192208.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Feldstein, Martin, and Otto Eckstein, The Fundamental Determinants of the Interest Rate,” The Review of Economics and Statistics, Vol. LII (1970), pp. 36375.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gibson, William E., and George G. Kaufman, The Sensitivity of Interest Rates to Changes in Money and Income,” The Journal of Political Economy, Vol. 76 (May/June 1968), pp. 47278.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hamburger, Michael J., and William L. Silber, An Empirical Study of Interest Rate Determination,” The Review of Economics and Statistics, Vol. LI (1969), pp. 36973.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Helliwell, John F., Monetary and Fiscal Policies for an Open Economy,” Oxford Economic Papers, New Series, Vol. 21 (1969), pp. 3555.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hendershott, Patric H., The Structure of International Interest Rates: The U.S. Treasury Bill Rate and the Euro-Dollar Deposit Rate,” The Journal of Finance, Vol. XXII (1967), pp. 45565.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hodjera, Zoran, Basic Balances, Short-Term Capital Flows, and International Reserves of Industrial Countries,” Staff Papers, Vol. XVI (1969), pp. 582612.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hodjera, Zoran, Short-Term Capital Movements of the United Kingdom, 1963-1967,” Journal of Political Economy, Vol. 79 (July/August