Currency Substitution, Flexible Exchange Rates, and the Case for International Monetary Cooperation: Discussion of a Recent Proposal

Over the past few years it has been suggested repeatedly that currency substitution makes the demand for single currencies unstable. Furthermore, because of asymmetric nonsterilized intervention, 1 currency substitution was held to contribute importantly to swings in the world money stock, which could ultimately produce world inflation or deflation. A number of proposals have been made to reduce destabilizing effects associated with currency substitution. One of these would amount to establishing a target for world money growth to assure that currency substitution affects only the composition of the world money stock and not its level. Under the proposed scheme, individual central banks would not pursue a fixed rate of monetary growth individually; rather, they would stand ready to accommodate any swing in the demand for domestic money by nonsterilized intervention designed to protect fixed exchange rates.

Abstract

Over the past few years it has been suggested repeatedly that currency substitution makes the demand for single currencies unstable. Furthermore, because of asymmetric nonsterilized intervention, 1 currency substitution was held to contribute importantly to swings in the world money stock, which could ultimately produce world inflation or deflation. A number of proposals have been made to reduce destabilizing effects associated with currency substitution. One of these would amount to establishing a target for world money growth to assure that currency substitution affects only the composition of the world money stock and not its level. Under the proposed scheme, individual central banks would not pursue a fixed rate of monetary growth individually; rather, they would stand ready to accommodate any swing in the demand for domestic money by nonsterilized intervention designed to protect fixed exchange rates.

Over the past few years it has been suggested repeatedly that currency substitution makes the demand for single currencies unstable. Furthermore, because of asymmetric nonsterilized intervention, 1 currency substitution was held to contribute importantly to swings in the world money stock, which could ultimately produce world inflation or deflation. A number of proposals have been made to reduce destabilizing effects associated with currency substitution. One of these would amount to establishing a target for world money growth to assure that currency substitution affects only the composition of the world money stock and not its level. Under the proposed scheme, individual central banks would not pursue a fixed rate of monetary growth individually; rather, they would stand ready to accommodate any swing in the demand for domestic money by nonsterilized intervention designed to protect fixed exchange rates.

The purpose of the paper is to evaluate the analysis as well as the empirical evidence from which the preceding proposal follows; the focus is on the work of Ronald McKinnon. Section I presents an overview of such work. Section II recalls the main conclusion of the empirical literature on currency substitution. It shows that the hypothesis that the demand for individual currencies is quite unstable because of currency substitution has received little support in past research. However, the case for international monetary cooperation does not rest solely on the hypothesis of currency substitution. Indeed, Section III demonstrates that complete monetary independence with flexible exchange rates occurs only under a restrictive set of assumptions. Whenever monetary dependence results from currency substitution or other factors, both domestic and world money turn out to enter as determinants of domestic output and prices. The implication of these theoretical deductions and of the empirical evidence presented in Section IV is that central banks should not subordinate control of domestic money to control of a world monetary aggregate, although external effects generally arise from the policies pursued by each country. Specifically, Section V stresses that the world demand for M1 does not appear to be stable enough to justify the move toward world monetary targets. There are alternative ways of achieving greater stability, even for countries that experience instability in the demand for their currency under fixed exchange rates.

I. An Overview

The purpose of this paper is to look into the analysis as well as the empirical evidence that lies behind the policy suggestions that McKinnon has put forward repeatedly over the past few years (McKinnon (1981; 1982; 1983)). 2 The three major building blocks of his model are as follows:

(1) The demand for domestic currency is highly sensitive to expected changes in the exchange rate. Since expectations about the future behavior of exchange rates have become more volatile with the adoption of flexible rates, and since, to a great extent at least, they are exogenous to the money market, the demand for any single currency is also subject to significant and unpredictable shifts.

(2) Central banks (of countries other than the United States) are not prepared to let their exchange rates go as currency substitution takes place; rather, they intervene without sterilizing so that their monetary base is often affected by the expectations of economic agents about the behavior of the U.S. dollar over the relevant future period. Furthermore, since those banks keep their foreign exchange reserves in short-term U.S. Government bonds, the induced changes in their supplies of base money are not matched by changes in the U.S. monetary base in the opposite direction. All in all, currency substitution translates into a loss of control over the world money stock.

(3) The demand for world money is stable, and world (as opposed to domestic) money is a better predictor of domestic inflation.

With these building blocks, McKinnon derives certain money supply and intervention rules designed to stabilize prices in individual countries. These rules involve (1) advance determination of the rates of growth of the national money supply that would be compatible with low or zero inflation in each country in the absence of money demand shocks, and (2) incourse modification of these rates through symmetric nonsterilized intervention designed to counter any unexpected exchange rate pressures that may arise.

Empirically these rules would be applied as follows:

Given the trend rate of change in output, in the income velocity of money, and in the money multiplier, each country will be able to ascertain what rate of growth of high-powered money should be consistent with approximate stability in the domestic price level. Each country could agree to fix the rate of growth of base money tentatively at that rate. Then if exchange rate pressures occur, nonsterilized symmetric intervention would be used to counter them. For instance, if there is upward pressure on the deutsche mark because of incipient inflationary tendencies in the United States, the authorities in the Federal Republic of Germany would buy dollars for deutsche mark claims on the Bundesbank. The dollars so acquired by the Bundesbank would be deposited with the U.S. Federal Reserve System so that the monetary base would be reduced in the United States by as much as it increased in Germany. If the money multipliers, specifically with respect to M1, are identical, the world money supply would be unchanged. If M1 velocity is also the same in Germany and the United States, the incipient percentage fall in money and prices produced in the United States by the aforementioned nonsterilized intervention operation multiplied by U.S. gross national product (GNP) would be equal to the (larger) incipient percentage rise in money and prices multiplied by the (smaller) GNP of Germany. If the incipient inflationary tendencies in the United States that caused the intervention in the first place were due to a shift in money demand from the United States to the Federal Republic of Germany, that is, to currency substitution, so that there was incipient deflation in the latter country, these tendencies would be forestalled through nonsterilized symmetric intervention. Each country would achieve its price objective precisely because both deviated from their tentative money supply growth targets to counter shifts in the distribution of money demand that left the demand for world money unaffected. 3

The major and striking implication that follows is that central banks should cease to aim at fixed rates of monetary growth. What is needed is a de facto world central bank that makes sure that, despite currency substitution, the “world” monetary base grows at an agreed-upon fixed rate. This situation could be achieved by having the central banks of the major industrial countries shift their main foreign exchange reserve assets from U.S. Government securities to deposits with the U.S. Federal Reserve System and begin to systematically and passively accommodate any swing in the demand for domestic money by nonsterilized intervention designed to protect fixed exchange rates.

Such analyses and proposals, which, at least in part, are also shared by others (Laffer (1976), Miles (1978; 1981), and Brittain (1981)) raise a number of analytical and empirical problems. For instance, a short-run partial equilibrium approach is usually taken, 4 some of the key empirical assumptions are not tested, 5 and, in some studies, no distinction is drawn between currency substitution and capital movements. 6 Rather than dwell on these points, we will examine the empirical evidence on currency substitution and ask whether phenomena other than currency substitution also call for international monetary cooperation with or without surrender of domestic monetary autonomy.

First, however, three prior findings should be brought to mind. First, the hundreds of empirical papers on the demand for money that have become available over the past 25 years tend to show that, by and large, once a scale variable and a domestic interest rate are employed as explanatory variables, very little unexplained variance is left over. Second, that same literature suggests that the elasticity of money demand with respect to a domestic interest rate is already on the low side (Laidler (1977 b)). Third, even those studies that look into the specific problem of the instability in money demand over the 1970s do not find it particularly helpful to turn to foreign interest rates or expected changes in the exchange rates as further regressors; indeed, it is interesting that in their recent survey of that literature, Judd and Scadding (1982) do not even mention the open economy.

II. Empirical Evidence on Currency Substitution

Turning now to the specific empirical literature, the questions that this section seeks to answer are the following: Is a statistically significant cross-elasticity effect easy to detect? Are cross elasticities very high in absolute value, and, at any rate, how do they compare with the estimated own elasticities? Is the hypothesis of a stable world demand for money supported by the empirical evidence on currency substitution? What is the pattern of the signs of cross elasticities that can be detected, and does it clearly indicate which countries could benefit most from the creation of a monetary union? Do such countries happen to be the United States, the Federal Republic of Germany, and Japan?

From the point of view of this discussion, it is perhaps more convenient to distinguish between, and deal separately with, two sets of empirical papers according to whether they simply estimate the effects of currency substitution on the demand for one or more currencies, or whether they go further and purport to show that the world demand for money is stable. The former set of studies includes both single-equation estimates 7 and portfolio models; 8 furthermore, it covers quite a few countries9 under both fixed and flexible exchange rates. A fair description of this literature would be as follows:

(1) Currency substitution is never found to be close to infinite, or just high, not even for two highly integrated economies, such as the United States and Canada (Alexander (1980), Bordo and Choudhri (1982 b)), or of countries with no constraint at all on capital movements, such as Switzerland (Vaubel (1980), Howard and Johnson (1982)).

(2) Statistically significant cross elasticities are not at all easy to detect. To quote from Brillembourg and Schadler (1979, p. 527), “Of the off-diagonal terms of the matrix [of elasticities], only about one fifth are significantly different from zero at the 95 per cent confidence level….”

(3) Generally speaking (Chrystal (1977) gives a few exceptions), the estimated values for the cross elasticities are so low that they tend to fall between ¼ and 110 of those for the own elasticities.

(4) It is not quite clear what the pattern of currency substitution is. However, if the most weight is attached to the results by Brillembourg and Schadler (1979), one could infer that the demand for U.S. dollars and Japanese yen is relatively unaffected by changes in the rates of return on foreign currencies, while there appears to be some degree of complementarity between the U.S. dollar and the deutsche mark. 10 Hence, the case for a monetary union between the United States, Japan, and the Federal Republic of Germany is far from obvious.

The second set of studies is particularly interesting in that it includes papers by Miles (1978; 1981) and Brittain (1981)—both strong advocates of the need to switch to world monetary targets. The Appendix deals fairly thoroughly with such papers. Here, it is simply stated that none of the two hypotheses that the papers were meant to support—that money demand is quite sensitive to open-economy variables and that its instability disappears with aggregation—can be said to have been tested in a meaningful way or to have survived whatever tests were pertinent.

The upshot of the discussion so far then is that perhaps all that a central bank should do under the present circumstances is to pay attention to possible international shifts in the demand for money when formulating its monetary target (Vaubel (1980)). Those who disagree with this conclusion and insist on the call for international monetary cooperation are nevertheless still faced with the difficulties that were raised in Section I. International monetary cooperation is certainly not easy to implement; the form that it will take must depend ultimately upon what the most pressing problem is seen to be. Even if one believes that currency substitution is menacing, there are still other problems that could also call for monetary cooperation. Second, no matter what the problem is, what is the rationale behind the suggestion that domestic monetary control be given up entirely? More explicitly, why should central banks have to choose between domestic and world money? The remainder of this paper deals mainly with these issues.

III. Insulation Properties of Flexible Exchange Rates and the Case for International Monetary Cooperation

The analytical tool that is employed is the rational expectations version of the Mundell (1961), Fleming (1962), and Laidler (1977 a) model of an open economy. The aggregate demand function has the ratio of actual to permanent (equilibrium) output depend upon domestic monetary disequilibrium and the ratio of world to domestic prices. Monetary disequilibrium is defined to be the ratio of nominal money supply and demand that, in its turn, is homogeneous of degree one in permanent income and prices. The aggregate supply curve is of the Lucas (1973) type. The balance of payments equation makes the change in the stock of international reserves (equal to zero under flexible exchange rates) depend on the ratio of world to domestic prices and monetary disequilibrium. The money supply is supposed to follow a steady, known path. Finally, two more functions are needed to describe the equilibrium in the goods market and the expectations formation process. All variables are in natural logarithms, permanent income is assumed to be constant and is set equal to one so that its logarithm goes to zero, and the demand for money becomes identical to the domestic price level.

yd =α1(msp) +α2(π+ϵp)(1)
ys =(1/η)(ppe)(2)
yd=ys=y(3)
0=γ1(msp)+γ2(π+ϵp)(4)
Xe=E(X|I)(5)

where yd and ys indicate the deviations of real output from its permanent (zero) level, ms the nominal money stock, p the domestic price level and the demand for money, π world prices, ϵ the exchange rate, and Xe the expected value of variable X, which depends on all the available information (I) at time t; for the time being, the stochastic terms are ignored, for their presence would not add in any meaningful way to the results.

What is shown is that (a) it is only under a fairly extreme set of assumptions that an open economy can insulate itself from foreign nominal shocks and thus can obviate the need for international monetary cooperation; (b) currency substitution is only one of the hypotheses that, despite flexible exchange rates, translate into a link between domestic and world prices; and (c) domestic money turns out to affect domestic prices even when complete monetary independence is not achieved, so that there is little justification for giving up its control entirely.

The solution of the whole system in terms of deviation from expected values is as follows: 11

y=0(6)
Ppe=0(7)
ϵϵe=πeπ(8)

In words, the domestic economy appears to be completely insulated with respect to any monetary shock that takes place in the rest of the world. This result is the joint product of three hypotheses, namely, that there is no currency substitution, purchasing power parity (PPP) holds all the time, and economic agents are endowed with perfect foresight. The first can be found in the absence of, say, abϵ˙e term from the demand for money function, the others in the ϵ – ϵe = πe – π (the exchange rate acts as a perfect shock absorber with respect to any changes in world prices) and pe = m results.

What deviations from PPP do is easy to see. If one begins with a system that is the same as the one that has been used thus far, except that the exchange rate equation (4) is now replaced by

ϵ=pπ+c(p˙π˙)(4)

the solution becomes

y=[α2c/(1+ηα2cηα1)](ππe)0(9)
ppe=[ηα2c/(1+ηα2cηα1)](ππe)0(10)
(ϵϵe)/(ππe)=(α1η+cα1η+1+c)/(1+ηα2cηα1)1(11)

which, of course, reduce to equations (6), (7), and (8) as c goes to zero.

The point then is that, even with perfect foresight and no currency substitution, any deviation from PPP implies a loss of monetary independence; the economics of this result is so evident that no comment is needed.

Next, an asymmetry is introduced into the information-gathering process with the assumption that at time t the exchange rate is the only known variable, or, in other words, that the speed with which expectations about exchange rates are revised is greatest. To deal with this case, one is forced to bring the disturbances into the picture and to write

yd=α1(msp)+α2(π+ϵp)+u1(12)
md=p+u2(13)
ys=(1/η)(ppe)+u3(3)

where u1,2,3, are random variables with zero mean and σi2 variance. Once again, no currency substitution is considered; as for the exchange rate, the absolute version of the PPP is imposed so that if monetary dependence eventually results, it will have to be attributed entirely to the particular hypothesis about the information-gathering process that is being made here. So, ϵ = p – π. Formally, the assumption about the information-gathering process is as follows:

Xe = E(X |I–1, ϵ)

which says that at time t the only current information with which economic agents are endowed consists of observations of the exchange rate. The expected level of world prices at time t is given by

πe = E(π|It–1)

while its actual value may be written as

π=πe+u4u4N(0,σ42)

The same is true of domestic prices; here, one has

pe = E(p|It–1)

and

p=pe+u5u5N(0,σ52)

where u4 and u5 are independently distributed random variables that also satisfy the conditions of orthogonality. 12

The solution for output and prices is as follows:

y=(1/η)[(1db)u5+dbu4]+u3(15)
p=m(1/α1){(1/η)[(1db)u5+dbu4]+u3u1}(16)

Once again then, any shock to world prices enters the expressions for domestic prices and output. Furthermore, the db value lies between zero and one, and tends to one as σ52 rises relative to σ42; this result means that the greater the variance of world prices, the lower the impact of a foreign nominal shock on domestic output and prices, and vice versa—a fairly uncontroversial proposition in rational expectations theory. 13 The economics of this result, which, incidentally, was also put forward by both Laidler (1977 a) and Turnovsky (1979) in the context of an analysis of adaptive expectations, is as follows: As long as the exchange rate is the only variable to be forecast perfectly, any unexpected increase in world prices will translate into a drop in domestic prices, and hence into excess real money balances, which, in its turn, will trigger output and (further) price changes.

Finally, let us consider currency substitution by writing

md=pbϵ˙e(17)

Solving the system in terms of deviations from expected values yields

y=(1/|D|)[(1/η)b(α1γ2γ1α2)](ππe)014(18)
ppe=(1/|D|)[b(α1γ2γ1α2)](ππe)0(19)
(ϵϵe)/(ππe)=(1/|D|)[(1/η)γ2+(α1γ2γ1α2)]1(20)

For the third time in a row, then, one encounters a situation in which flexible exchange rates do not insulate the domestic economy completely from foreign nominal shocks, the reason here being that every time the exchange rate moves to counter the foreign shock, the demand for money shifts, thereby bringing about changes in domestic output and prices. 15

Of course, one could go on along these lines by considering capital movements, wealth effects, etc. 16 What has been done so far, however, suffices to support a few considerations that are relevant to the present discussion. The expressions that were just arrived at, plus the fact that (a) deviations from PPP, no matter what their origin, (b) differences in the costs of gathering information about exchange rates, on one hand, and domestic and world prices, on the other hand, and (c) currency substitution (according to some) are all undisputed features of the world around us (Genberg (1978) and Frenkel (1981)), suggest that, notwithstanding the presence of flexible exchange rates, one single country can hardly expect to enjoy complete monetary independence and hence be able to fully control its domestic price level. The consideration of capital mobility, wealth effects, etc., reinforces such conclusions and adds to the case for international monetary cooperation. From this point of view, McKinnon’s implicit criticism of much of the (theoretical) literature on the open economy under flexible exchange rates is correct.

At the same time, however, just because there are several potential sources of a link between domestic and foreign prices that cannot all be taken care of by one single form of international monetary cooperation, the question arises as to which of them is the most relevant. From this point of view, for instance, many would be prepared to argue that such phenomena as deviations from PPP or poor and costly information-gathering processes across countries are more disturbing than currency substitution. If that were indeed true, then it would be difficult to accept the idea of an international monetary cooperation that is restricted to a few countries (the currencies of which happen to be substitutes in demand) and that is based on the adoption of fixed exchange rates and world monetary targets. Finally, no matter what the rationale is for having some sort of international monetary cooperation, and even when currency substitution is the source of the loss of monetary independence, the analysis does show that both domestic and world monetary variables always appear in the price equation at the same time. Therefore, the alternative is not between controlling domestic or world money but between controlling domestic or domestic plus rest of the world money. From this point of view, McKinnon’s argument that world money is a better predictor of domestic inflation and, hence, is the only variable that is worth controlling does not appear to be supported by the theory. Next, it will be shown that its empirical foundations are lacking also.

IV. A Reassessment of McKinnon’s Empirical Evidence

Since the beginning of the flexible exchange rate period and the first oil shock, a considerable amount of empirical evidence has become available that indicates that, regardless of the fitted equations, the sample periods, the estimation techniques, or the countries involved, world nominal shocks always seem to affect the domestic rate of inflation in a significant fashion; 17 on average, the estimated elasticities (of domestic to foreign prices) tend to fall in the 0.2–0.5 range.

Because of this finding, and given the purposes of the paper, there is no need to specify and estimate a set of structural or even semireduced form equations for various countries; rather, domestic inflation rates are regressed on distributed lags of domestic and/or world money growth. There are two major gains from doing so. First, as a by-product of the exercise, the view that world money is a better predictor of domestic inflation is tested directly in the same way as in McKinnon (1983); second, since the ultimate cause of world price shocks is world money, and world money could affect domestic money via the reaction function of the domestic monetary authorities—see Gordon (1977), Gandolfi and Lothian (1980), Cassese and Lothian (1982), Bordo and Choudhri (1982 a)—having world money rather than world prices appear in the domestic price equation should eliminate part of the bias that may be present in some of the standard equations.

The countries in the Group of Ten are considered, and three equations are fitted for each of them. The first is the “closed economy” version, in which domestic prices are regressed on domestic money. The second has the rest of the world money appear as a second regressor, as theory suggests; the third, in which domestic prices are regressed on world money, represents McKinnon’s hypothesis.

Quarterly data (deseasonalized and centered on midquarter) are used; rates of change are taken to be the first difference in natural logarithms. Two alternative money definitions—M1 and M2—are tried out; the world and the rest of the world money stock is computed on the basis of the current exchange rate and is expressed in terms of domestic currency units. As for prices, both wholesale prices 18 and GNP or GDP (gross domestic product) deflators are employed; because of lack of data on quarterly GNP, no regression with the GNP deflator as a dependent variable could be run for Belgium, Sweden, or the Netherlands.

A third-degree Almon polynomial with both end-period and beginning-period constraints imposed is employed in each case, and allowance is made for 12 lags, starting at t–2; the constant term is systematically suppressed and the Cochrane-Orcutt technique is utilized. The sample is restricted to the period of floating (the first quarter of 1973 through the fourth quarter of 1980), with the crucial period 1973–74 included in the sample. 19

Because dealing with ten countries at the same time and with different combinations of dependent and independent variables translates into just over 100 regressions and because the overall explanatory power of the three alternative hypotheses is of most interest, the most effective and least space-consuming procedure is to give the R2 generated by each of those regressions; they appear in Table 1. 20

Table 1.

Ten Industrial Countries: Explanatory Power (R2 Around the Origin) of (1) Domestic Money, (2) Domestic and Rest of World Money, and (3) World Money over Wholesale Prices and Gross National Product (GNP) Deflators

article image

There is little need for a detailed comment on the results on a country-by-country or a variable-by-variable basis; the general message is that (a) except for France, domestic money performs better than world money alone, and (b) the inclusion of the rest of the world money translates into an increase in the R2 for almost all countries. Therefore, the empirical evidence seems to be in line with what theory suggests, namely, when explaining the inflation rate under flexible exchange rates, first, look at the growth of the domestic money stock and then supplement the model with some proxy for world nominal shocks; under no circumstances should one use world money only.

One further point is that McKinnon argues that, at least as far as the U.S. economy is concerned, there are well-defined periods (1973–74 and 1979–80) over which domestic money is expected to generate a particularly poor fit and to underpredict the actual inflation rate considerably. In Chart 1 the residuals generated by the equations with U.S. wholesale prices and, alternatively, U.S. M1 and world M1 have been plotted. If anything, up to the end of 1974, world money performs worse. Subsequently, domestic and world money overpredict and underpredict, respectively; in 1978–79 that situation reverses itself, but once again world money is not found to fare any better than domestic money.

Chart 1.
Chart 1.

Residuals from Regression of U.S. Wholesale Prices on U.S. M1 (——) and WorldM1 (– – –), 1973–80

Citation: IMF Staff Papers 1983, 004; 10.5089/9781451930610.024.A004

The last step is to take a close look at the exact kind of empirical evidence that McKinnon (1983) provides. Basically, the reader is presented with a set of regressions of annual changes in the U.S. wholesale price index (XWPUS) on a constant and on changes in U.S. M1 (XM1 U US) or world M1 (XM1 W WORLD), both lagged once and twice. The regressions show that, over the sample period 1970–81, world M1 generates a higher R2, as is confirmed by the first two equations in Table 2, in which the experiment is replicated.

Table 2.

Further Results on Relationship Between Prices and Domestic and World Money, 1970–81 1

article image

The t-statistics are shown in parentheses.

Two basic reasons suggest that this finding is not robust. The first question that arises when viewing the residuals generated by equations (1) and (2) is whether world money performs systematically better than domestic money over the whole decade. This point is crucial, since the debate over cost-push in the late 1960s and early 1970s and, of course, the development of the literature on rational expectations, have taught that an effort must be made to distinguish between systematic and nonsystematic impulses in studying inflation. In the present instance, the relative performance of the two equations appears to depend to a great extent on just one observation—that for 1974. Therefore, it would be interesting to find out what happens when a dummy variable is introduced to take out that particular observation, as is done in equations (3) and (4). The Durbin-Watson statistic of the equation with domestic money now becomes acceptable, and both money lagged once and money lagged twice are significant and have reasonable coefficients. The explanatory power of the equation is such that one can no longer say that world money is a better predictor of domestic inflation.

A fairly dramatic change in the picture also occurs as the GNP deflator replaces wholesale prices as the dependent variable, a perfectly legitimate substitution, of course. Equations (1’)(4’) suggest that whether correction for first-order autocorrelation is made or not, domestic money always wins. 21

The message of this section is twofold. First, the view that, despite flexible exchange rates, individual countries do not fully control their domestic price level stands up to the data. Hence, countries could benefit from an internationally concerted action against either inflation or the problems that generate monetary dependence. Second, it does not seem to make much sense to say that under present circumstances no monetary independence at all is enjoyed by individual countries, that domestic inflation is driven just by world money, and, hence, that world money is the only variable that ought to be controlled.

V. World Demand for M1 and Targets for Domestic Credit Expansion

As pointed out in Section I, McKinnon’s reader is never presented with any direct empirical evidence pertaining to the nature of the world demand for money function. Rather, the stability of such function is asserted in the ex post interpretation of past developments. Since it is also a key prerequisite for the successful working of the plan, this assumption has been scrutinized in a separate paper (Spinelli (1983 a)), which suggests that results are sensitive to aggregation procedures. In particular, world (defined to be the United States, the Federal Republic of Germany, and Japan) demand for M1 turns out to be relatively stable when both world money and income are computed on the basis of current exchange rates. If, however, a base-period exchange rate is employed for income and, hence, any spurious correlation between the two variables is removed, world money demand appears to be quite un stable. Therefore, both the specific empirical evidence and a careful look at McKinnon’s own analysis seem to imply, once again, that the suggestion that individual countries give up monetary independence entirely and switch to a world monetary target is not justified. 22

The problem, however, could be even more difficult in that it is not entirely clear whether stability in the world demand for money is a sufficient condition for the successful working of a plan based on a rule for world monetary growth that is simple to define and to apply. The way in which McKinnon describes the past as well as the future reveals that somehow he happens to believe that it does not make any difference as to which component of the world money stock is shrinking and which is growing; as long as their variations match and total demand is stable, the world price level should be unaffected. Other observers would maintain that that, too, has to be proved. 23

Finally, the question arises as to whether fixed exchange rates should be seen as the only way out of the present problem, and what would then be the appropriate monetary target. McKinnon does not make clear why cleanly floating rates, as an alternative to the current system of managed floating, would not take care of much of the problem by severing the link between currency substitution and swings in the world money stock. He also does not address the more general and fundamental question of whether or not currency substitution necessarily forces a move back to fixed exchange rates, or whether it simply weakens the case against fixed exchange rates, as many would argue (Vaubel (1977)). These are perfectly legitimate questions, because the case for flexible exchange rates has never been predicated on the absence of any measurable degree of currency substitution.

All that having been said, however, suppose that a good case has been made for a return to fixed exchange rates. Here, one would find a dog chasing its tail, in the sense that it is well known that since the money demand of a small open economy is determined exogenously (permanent income, prices, and interest rates are given), the monetary authorities must not pursue a monetary target that is defined in terms of rates of change of the domestic monetary base. If they did, then any shift in money demand would translate into a permanent disequilibrium in the money market as well as in swings in the world money stock, that is, it would give rise to the same problem that has been discussed. A solution to such a problem was found a fairly long time ago in the definition of domestic credit expansion (DCE), which can be seen as an alternative to McKinnon’s plan. 24

The implementation of such an alternative is not without problems, particularly in view of the asymmetry that it introduces into the treatment of the reserve and nonreserve currency countries. Those in the latter group, unable to finance deficits forever, have to react to any monetary shock, whether of domestic or foreign origin, and therefore have no choice but to bear part of the cost of adjustment. On the contrary, the reserve currency country, by taking the acquisitions or sales of the domestic reserve currency by foreign monetary authorities above or below the line, that is, by defining DCE in one way or another, can either insulate from or make the domestic money stock depend on foreign developments. Therefore, there could be situations in which the whole burden of the adjustment process is forced on the nonreserve currency countries (Day (1979)). The problem of determining the origin of the money shock is also quite basic. A touchy question could be: Is the domestic interest rate too low or is the foreign rate too high? So there certainly are problems with DCE; however, one can hardly ignore this solution as a possible alternative to other proposals.

VI. Conclusions

A number of conclusions emerge from this paper. First, it is certainly true that under flexible exchange rates an individual country still does not enjoy complete monetary independence, and it is not in full control of the domestic price level. Once unrealistic assumptions are put aside, models of the open economy have been shown to predict that domestic prices are also affected not only by domestic but also by world variables. This result, in turn, implies that there certainly exists a good case for having some form of international monetary cooperation.

However, what kind of monetary cooperation should be provided? Here, McKinnon’s analysis is found to be weak in that it ignores all but one of the several causes of a link between domestic and world prices without ever showing that it is the most relevant one; his analysis misjudges the relative roles of domestic and world variables in the explanation of domestic inflation. In particular, the suggestion that a world central bank be created and that national policy independence be given up is the product of several hypotheses: (a) monetary interdependence is caused mainly by currency substitution, (b) the world demand for money is stable, and (c) world money is all that matters. Each of them has been shown not to stand up to the data or the scrutiny of the already available empirical evidence. The paper also points to the existence of alternative solutions to the problem of currency substitution—cleanly floating rates, domestic monetary targets net of expected shifts in the demand for money, and fixed exchange rates combined with DCE targets—all of which, by and large, have been ignored in recent discussions.

APPENDIX

Discussion of Miles (1978; 1981) and Brittain (1981) and Additional Results on Currency Substitution

Previous studies

Miles and Brittain are the only ones, among the advocates of the need to switch to world monetary targets, to present some empirical evidence not only on the relationship between currency substitution and instability in the demand for single currencies but also on the alleged stability of the world money demand functions. Since their results appear to be mentioned fairly often, detailed discussion may be useful.

Miles (1978; 1981) regresses the relative demand for domestic and foreign currency by economic agents in Canada, the United States, and the Federal Republic of Germany on the ratio of foreign to domestic interest rates. He finds that the two variables are positively and significantly related to each other, and winds up arguing in favor of the hypothesis of a stable world money demand and calling for an end to the adoption of monetary targets by individual countries. Two studies (Bordo and Choudhri (1982 b) and Laney, Radcliffe, and Willett (1982)) examine these results and show that while for Canada they do not survive further tests, for the United States, even when taken at face value, they point to a currency substitution that has an influence on the order of 150 of the impact of the domestic interest rates, which is much in line with some of the conclusions by Brillembourg and Schadler (1979) mentioned in Section I in the text.

Apart from all this, however, there is one more general and fundamental criticism of the way in which Miles interprets his results. The positive coefficient on the interest rate ratio is taken to imply that “if the opportunity cost of holding real balances denominated in currency A rises relative to the opportunity cost of holding those denominated in currency B, all of these individuals will be assumed to reduce their real balances denominated in currency A and to increase their holdings denominated in currency B” (Miles (1978, p. 429)). Considering the implied relationship between relative interest rates and exchange rates, it is not difficult to find out what is wrong with this analysis. Basically, Miles overlooks the fact that the way in which he extends Christ’s (1963) analysis is not correct. Christ regresses the ratio of the overall demand for currency to bonds on the interest rate differential, obtains a negative coefficient, and (correctly) infers that, as interest rates move, economic agents switch from currency into bonds and vice versa. In that case, there is no reason to expect a change in the portfolio size (i.e., in the sum of currency plus bonds). Miles, on the other hand, regresses the ratio of two components of money demand on the domestic/foreign interest rate (on bonds) differential. The negative relation between these variables cannot be identified as being due solely to the changing composition of a given total demand for currency; in this case, it is less reasonable to assume that the portfolio size (i.e., currency i plus currency j) is left unchanged. 25 In other words, if domestic and foreign bonds are brought into the picture, a rise in the foreign interest rate would cause residents to switch out of both domestic and foreign currency. It will therefore take no more than a greater elasticity in the demand for foreign currency with respect to foreign interest rates to generate a positive coefficient on the interest rate ratio. This fact represents both the economics that lies behind some of the formal points raised by Bordo and Choudhri (1982 b) in their reappraisal of Miles’s analysis and a criticism of the way in which some of the literature on the determinants of exchange rates also tends to be interpreted vis-á-vis the currency substitution problem. 26 Taken on their own, the two papers by Miles provide only indirect tests of currency substitution that suggest that world money demand is unstable.

Brittain’s (1981) study is specifically aimed at showing that for several countries the inclusion of a foreign portfolio variable adds significantly to the overall performance of a standard demand for money function, and this is seen as an intermediate step toward stressing the stability of world demand. The results presented in his text show, however, that in five out of ten instances the coefficient on the domestic interest rate variable is positively signed; in nine, the coefficient on the portfolio variable is no greater than 0.01. Furthermore, the equations for the United States do not look any better than the ones that have been used for quite some time. 27

On the whole, the papers that have been discussed here do not appear to mark a turning point in the debate over the empirical relevance of the currency substitution phenomenon.

Extensions

In the existing empirical literature on currency substitution, there is no study that combines the following four features: (a) M1 is the dependent variable, (b) the expected change in the exchange rate appears explicitly as an independent regressor, 28 (c) the data sample covers only the flexible exchange rate period, and (d) a portfolio approach is taken. The first three characteristics can be appreciated with McKinnon’s work in mind.

To fill the gap, a small portfolio model is set up and fitted to the period (fourth quarter of 1971 through fourth quarter of 1980) by full-information maximum likelihood. Four currencies are considered: the U.S. dollar (USD), deutsche mark (DM), Japanese yen (JY), and British pound (BP); their desired stocks (denoted by Mi*, i = USD, DM, JY, BP) depend on just a scale variable (W), defined to be the sum of the four actual stocks, and a vector of expected changes in the exchange rate that are proxied by the 90-day premium on the U.S. dollar. Three further equations make the adjustment in the actual stocks of DM, JY, and BP a function of the gaps between both the own and the USD desired and actual stocks. To close the system, an identity is added that defines the actual change in the stock of USD to be equal to itself, or to the difference between the total change in W and the actual change in the other three currencies. The U.S. dollar is included in the three adjustment equations to make sure that, via the demand for it, any disequilibrium in the market for one currency is transmitted to the rest of the portfolio. The use of an identity for the adjustment function for the U.S. dollar reveals that basically the world of currencies tends to be considered to be the sum of two subsets—the U.S. dollar, which takes up all the slack, and the remaining currencies. Formally,

Mi*=Aϵ˙e+BWfor all isMi=Mi1=λi,i(Mi*Mi1)+λi.USD(MUSD*MUSD1),iUSDMUSDMUSD1ΔWΣ(MiMi1),iUSD

where

ϵ˙e = vector of expected changes in exchange rate

W = ∑Mi, all i

A = 4 × 4 symmetric matrix with all columns adding to zero

B = vector of weights adding to one

λi,i and λi, USD = speed of adjustment with respect to difference between actual and desired values of own and USD stocks

Results are as follows (t =statistics in parentheses): 29

|MUSD*MDM*MJY*MBP*|=|0.6970.1660.4890.041(1.22)(1.17)(1.17)(1.66)0.0160.1600.009(0.33)(1.51)(0.43)(0.351)0.021(1.13)(0.80)0.072(2.83)||030ϵ˙DM.USDeϵ˙JY.USDeϵ˙BP.USDe|+|0.370(3.12)0.156(5.23)0.401(4.60)0.073(13.19)|W
λDM.DM=0.482λJY.JY=0.721λBP.BP=0.151(9.64)(7.63)(3.37)λDM.USD=0.118λJY.USD=0.506λBP.USD=0.002(7.04)(5.93)(0.37)

By focusing on the A matrix, one can see that one elasticity out of ten is significant by conventional standards and three t-statistics are less than 1.0, while the remaining ones tend to fall into the 1.5–1.65 interval. Own and cross elasticities are all positive and negative, as theory suggests, except for the DM-JY case, in which complementarity seems to dominate. 31 The B and λ values seem to be reasonable and significant; it is only for the British pound that the gap between desired and actual U.S. dollars does not seem to affect the adjustment process, and the own-adjustment parameter appears to be on the low side. 32

This effort has yielded another piece of evidence to support the conclusion, given in the text, that while there is some currency substitution, it is also true that its quantitative consequences on the demand for individual currencies should not be overstated. To stress this point, some elasticity values that are implied by the results have been computed: 33

article image

These values are so low that they speak for themselves.

REFERENCES

  • Alexander, W.E., Foreign Influences on the Demand for Money in an Open Economy: The Canadian Case,paper presented at the 17th Meeting of Technicians of Central Banks of the American Continent, Bogotá, Colombia (November 26–29, 1980).

    • Search Google Scholar
    • Export Citation
  • Bilson, John F.O., The Deutsche Mark/Dollar Rate: A Monetary Analysis,in Policies for Employment, Prices, and Exchange Rates, ed. by Karl Brunner and Allan H. Meltzer, Carnegie-Rochester Conference Series on Public Policy, A Supplementary Series to the Journal of Monetary Policy, Vol. 11 (Amsterdam, 1979), pp. 59101.

    • Search Google Scholar
    • Export Citation
  • Bordo, Michael D., and Ehsan U. Choudhri (1982 a), “The Link Between Money and Prices in an Open Economy: The Canadian Evidence from 1971 to 1980,Federal Reserve Bank of St. Louis, Review, Vol. 64 (August/September 1982), pp. 1323.

    • Search Google Scholar
    • Export Citation
  • Bordo, Michael D., and Ehsan U. Choudhri (1982 b), “Currency Substitution and the Demand for Money: Some Evidence for Canada,Journal of Money, Credit and Banking, Vol. 14 (February 1982), pp. 4857.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Boughton, James M., Demand for Money in Major OECD Countries,OECD Economic Outlook: Occasional Studies (Paris, January 1979), pp. 3557.

    • Search Google Scholar
    • Export Citation
  • Boyer, Russell S., The International Transmission of Inflation,” Chap. 9 in Essays in Economic Analysis: Proceedings of the Association of University Teachers of Economics, Sheffield, 1975, ed. by M.J. Artis and A.R. Nobay (Cambridge University Press, 1976), pp. 17388.

    • Search Google Scholar
    • Export Citation
  • Brillembourg, Arturo, and Susan M. Schadler, A Model of Currency Substitution in Exchange Rate Determination, 1973–78,Staff Papers, Vol. 26 (September 1979), pp. 51342.

    • Search Google Scholar
    • Export Citation
  • Brittain, Bruce, International Currency Substitution and the Apparent Instability of Velocity in Some Western European Economies and in the United States,Journal of Money, Credit and Banking, Vol. 13 (May 1981), pp. 13555.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bruno, Michael, An Analysis of Stagflation in the Industrial Countries: Some Preliminary Results, The Maurice Falk Institute for Economic Research in Israel, Discussion Paper No. 788 (Jerusalem, July 1978).

    • Search Google Scholar
    • Export Citation
  • Cagan, Phillip, Imported Inflation, 1973–74, and the Accommodation Issue,Journal of Money, Credit and Banking, Vol. 12 (February 1980), pp. 116.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Cassese, Anthony, and James R. Lothian, The Timing of Monetary and Price Changes and the International Transmission of Inflation,Journal of Monetary Economics, Vol. 10 (July 1982), pp. 123.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Chetty, V. Karuppan, Measuring the Nearness of Near-Moneys,American Economic Review, Vol. 59 (June 1969), pp. 27081.

  • Christ, Carl F., Interest Rates and ‘Portfolio Selection’ Among Liquid Assets in the U.S.,” Chap. 8 in Measurement in Economics: Studies in Mathematical Economics and Econometrics in Memory of Yehuda Grunfeld (Stanford University Press, 1963), pp. 20118.

    • Search Google Scholar
    • Export Citation
  • Chrystal, K. Alec, Demand for International Media of Exchange,American Economic Review, Vol. 67 (December 1977), pp. 84050.

  • Day, William H.L., Domestic Credit and Money Ceilings Under Alternative Exchange Rate Regimes,Staff Papers, Vol. 26 (September 1979), pp. 490512.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dornbusch, Rudiger, Devaluation, Money, and Nontraded Goods,American Economic Review, Vol. 63 (December 1973), pp. 87180.

  • Fleming, J. Marcus, Domestic Financial Policies Under Fixed and Under Floating Exchange Rates,Staff Papers, Vol. 9 (November 1962), pp. 36980.

  • Frenkel, Jacob A., The Collapse of Purchasing Power Parities During the 1970’sEuropean Economic Review, Vol. 16 (May 1981), pp. 14565.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gandolfi, Arthur E., and James R. Lothian, International Price Behavior and the Demand for Money,National Bureau of Economic Research, Working Paper No. 602 (Cambridge, Massachusetts, December 1980).

    • Search Google Scholar
    • Export Citation
  • Genberg, Hans, Purchasing Power Parity Under Fixed and Flexible Exchange Rates,Journal of International Economics, Vol. 8 (May 1978), pp. 24776.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gordon, Robert J. (1977), “World Inflation and Monetary Accommodation in Eight Countries,Brookings Papers on Economic Activity: 2 (1977), pp. 40968.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gordon, Robert J. (1980), “A Consistent Characterization of a Near-Century of Price Behavior,American Economic Review, Papers and Proceedings, Vol. 70 (May 1980), pp. 24349.

    • Search Google Scholar
    • Export Citation
  • Gramlich, Edward M., Macro Policy Responses to Price Shocks,Brookings Papers on Economic Activity: 1 (1979), pp. 12466.

  • Hamburger, Michael J., The Demand for Money in an Open Economy: Germany and the United Kingdom,Journal of Monetary Economics, Vol. 3 (January 1977), pp. 2540.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hamburger, Michael J., and Geoffrey E. Wood, Interest Rates and Monetary Policy in Open Economies” (mimeographed, November 1978).

  • Howard, David H., and Karen H. Johnson, The Behavior of Monetary Aggregates in Major Industrialized Countries,International Finance Division, Board of Governors of the Federal Reserve System (mimeographed, Washington, May 1982).

    • Search Google Scholar
    • Export Citation
  • Johnson, Karen H., The Demand for Swiss Monetary Assets,International Finance Discussion Papers, No. 210, Board of Governors of the Federal Reserve System (Washington, June 1982).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Judd, John P., and John L. Scadding, The Search for a Stable Money Demand Function: A Survey of the Post-1973 Literature,Journal of Economic Literature, Vol. 20 (September 1982), pp. 9931023.

    • Search Google Scholar
    • Export Citation
  • Keran, Michael, and Michael Riordan, Stabilization Policy in World Context,Federal Reserve Bank of San Francisco, Economic Review (Fall 1976), pp. 519.

    • Search Google Scholar
    • Export Citation
  • Kouri, Pentti J.K., and Jorge Braga de Macedo, Exchange Rates and the International Adjustment Process,Brookings Papers on Economic Activity: 1 (1978), pp. 11150.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Laffer, Arthur B., Optimal Exchange Rate Systems,paper presented at the Eighty-Ninth Annual Meeting of the American Economic Association, Atlantic City, New Jersey, September 1618, 1976.

    • Search Google Scholar
    • Export Citation
  • Laidler, David E.W. (1977 a), “Expectations and the Behaviour of Prices and Output Under Flexible Exchange Rates,Economica, Vol. 44 (November 1977), pp. 32735.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Laidler, David E.W. (1977 b), The Demand for Money: Theories and Evidence (New York, 2nd ed., 1977).

  • Laney, Leroy O., Chris D. Radcliffe, and Thomas D. Willett, International Currency Substitution by Americans Is Not High: A Comment on Miles” (mimeographed, Dallas, September 1982).

    • Search Google Scholar
    • Export Citation
  • Levy, Haim, and Marshall Sarnat, Exchange Rate Risk and the Optimal Diversification of Foreign Currency Holdings,Journal of Money, Credit and Banking, Vol. 10 (November 1978), pp. 45363.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lucas, Robert E., Jr., Some International Evidence on Output-Inflation Tradeoffs,American Economic Review, Vol. 63 (June 1973), pp. 32634.

    • Search Google Scholar
    • Export Citation
  • McKinnon, Ronald I. (1973), “On Securing a Common Monetary Policy in Europe,Banca Nazionale del Lavoro, Quarterly Review, No. 104 (March 1973), pp. 320.

    • Search Google Scholar
    • Export Citation
  • McKinnon, Ronald I. (1981), “The Exchange Rate and Macroeconomic Policy: Changing Postwar Perceptions,Journal of Economic Literature, Vol. 19 (June 1981), pp. 53157.

    • Search Google Scholar
    • Export Citation
  • McKinnon, Ronald I. (1982), “Currency Substitution and Instability in the World Dollar Standard,American Economic Review, Vol. 72 (June 1982), pp. 32033.

    • Search Google Scholar
    • Export Citation
  • McKinnon, Ronald I. (1983), “A Program for International Monetary Stability,Discussion Paper No. 3, Center for Economic Policy Research, Stanford University (January 1983).

    • Search Google Scholar
    • Export Citation
  • Miles, Marc A. (1978), “Currency Substitution, Flexible Exchange Rates, and Monetary Independence,American Economic Review, Vol. 68 (June 1978), pp. 42836.

    • Search Google Scholar
    • Export Citation
  • Miles, Marc A. (1981), “Currency Substitution: Some Further Results and Conclusions,Southern Economic Journal, Vol. 48 (July 1981), pp. 7886.

  • Modigliani, Franco, and Lucas Papademos, Monetary Policy for the Coming Quarters: The Conflicting Views,Federal Reserve Bank of Boston, New England Economic Review (March-April 1976), pp. 235.

    • Search Google Scholar
    • Export Citation
  • Mundell, Robert, Flexible Exchange Rates and Employment Policy,Canadian Journal of Economics, Vol. 27 (November 1961), pp. 50917.

    • Search Google Scholar
    • Export Citation
  • Mussa, Michael, Macroeconomic Interdependence and Exchange Rate Regime,” in International Economic Policy: Theory and Evidence, ed. by Rudiger Dornbusch and Jacob A. Frenkel (Johns Hopkins University Press, 1979), pp. 160204.

    • Search Google Scholar
    • Export Citation
  • Spinelli, Franco (1983 a), “Should the Hypothesis of a Well-Defined and Stable World Demand for M1 Be Reinstated? Simple Considerations and Tests” (unpublished, International Monetary Fund, March 1983).

    • Search Google Scholar
    • Export Citation
  • Spinelli, Franco (1983 b), “On the Case for International Monetary Cooperation” (unpublished, International Monetary Fund, May 12, 1983).

    • Search Google Scholar
    • Export Citation
  • Spitäller, Erich, A Model of Inflation and Its Performance in the Seven Main Industrial Countries, 1958–76,Staff Papers, Vol. 25 (June 1978), pp. 25477.

    • Search Google Scholar
    • Export Citation
  • Turnovsky, Stephen J., On the Insulation Properties of Flexible Exchange Rates,Revue Economique, Vol. 30 (July 1979), pp. 71946.

    • Search Google Scholar
    • Export Citation
  • Turnovsky, Stephen J., and Jagdeep S. Bhandari, The Degree of Capital Mobility and the Stability of an Open Economy Under Rational Expectations,Journal of Money, Credit and Banking, Vol. 14 (August 1982), pp. 30326.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vaubel, Roland (1977), “Free Currency Competition,Weltwirtschaftliches Archiv, Vol. 113 (No. 3, 1977), pp. 43559.

  • Vaubel, Roland (1980), “International Shifts in the Demand for Money, Their Effects on Exchange Rates and Price Levels, and Their Implications for the Preannouncements of Monetary Expansion,Weltwirtschaftliches Archiv, Vol. 116 (No. 1, 1980), pp. 144.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • von Furstenberg, George M., and William H. White, The Inflation Process in Industrial Countries Individually and Combined,Kyklos, Vol. 33 (No. 2, 1980), pp. 26186.

    • Crossref
    • Search Google Scholar
    • Export Citation
*

Mr. Spinelli, economist in the Research Department, holds degrees from the Catholic University in Milan and the University of Manchester. He has been on the faculty of the University of Western Ontario and of the Catholic University in Milan.

This paper has benefited from comments by Thomas Mayer, Kenneth Rogoff, and George M. von Furstenberg, as well as colleagues in the Fund.

1

Nonsterilized intervention is asymmetric if either the country gaining reserves lets its money supply expand or the country losing reserves (or acquiring official external liabilities) lets its money supply contract, but both countries do not undertake such mutually reinforcing action simultaneously.

2

For an earlier exposition of many of the same ideas, see McKinnon (1973).

3

Of course, if the money multiplier and/or velocity are greater in the Federal Republic of Germany than in the United States, the shift in the distribution of base money could be inflationary. Although the growth of base money would be unchanged in the aggregate, the world money supply could then grow faster than initially planned.

Shocks to money demand that arise for reasons other than a desire for currency substitution are less readily countered under such rules. For instance, if the demand for money increased in the Federal Republic of Germany at a given level of income and interest rates, without falling spontaneously in the United States, there would be a deflationary tendency that would be shared by both countries after a smaller amount of nonsterilized intervention had occurred to support the dollar than in the previous case. Similarly, if the demand for money fell in the United States at a given level of income and interest rates without rising spontaneously in Germany, there would be an inflationary tendency in both countries after nonsterilized intervention had led to an increase in the money supply in Germany and to a reduction in the U.S. money supply from the level that was tentatively targeted, but by less than the fall in demand. In each instance, shocks that affect only the money demand in one country would be offset only partly and in a fashion that would interfere with the price stability objectives of the other country. Disturbances to aggregate supply that do not change money demand and prices equally in both countries would also lead to exchange market pressures, calling for symmetric nonsterilized intervention.

4

The following quotations from McKinnon (1982, p. 325) may illustrate this point. “A complete picture of international inflation would link money creation to realized price and possibly output increases—with differing variable lags. Such a complex process cannot be captured within a simple analytical framework. Focus instead on the much narrower problem of how changing exchange-rate expectations immediately influence the demand for rowa [rest of world money] relative to dollars and the total supply of world money.” Under this approach, “fluctuations in s [the expected change in the exchange rate] are given exogenously to the model.”

5

One example is given by the assumption of a stable world demand for money. All the reader comes across is the following passage, “The world demand for money seems relatively stable. By considering a crude index of a ‘world’ money supply… the two great outbreaks of international price inflation in the 1970’s become explicable” (McKinnon (1982, p. 320)).

6

On this, see the discussion in the Appendix of the studies by Miles (1978; 1981).

9

Canada, Denmark, France, the Federal Republic of Germany, Italy, Japan, Switzerland, the United Kingdom, and the United States.

10

This result is also obtained by Kouri and de Macedo (1978) in a study of the demand for securities in five major countries.

11

All the mathematical proofs can be found in Spinelli (1983 b).

12

Orthogonality means that the vector of the error term is perpendicular to the vector of the independent variables, so that πe and pe are the true least-squares solutions to the problem.

13

As σ52 goes to infinity, u5 becomes perfectly predicted by u5u4, hence, by u5. This result means that pe acts as a perfect shock absorber with respect to domestic price shocks, which then cannot affect output via the aggregate supply curve. At the same time, however, the fact that u4 goes “unnoticed” means that any shock to world prices (that is not captured by pe is bound to hit the domestic economy. Equations (15) and (16) say just that.

14
D” indicates the determinant of the system and is equal to

–(1–η)(γ2 + γ1b) + (1–b) (γ1α2 – γ2α1)

15

At first sight, this result is hard to understand in the sense that any unexpected shock to world prices affects only the actual exchange rate and, hence, leaves money demand unchanged. However, it is also true that such change in the actual exchange rate will show up in the definition of the expected exchange rate at time t + 1, and money demand will have to react.

16

It is clear that by not having capital flows, one of the channels through which foreign nominal shocks might affect the domestic economy is closed. This fact should be kept in mind, particularly in the light of the results put forward in studies such as Mussa (1979), which shows that in Keynesian models, flexible exchange rates provide complete insulation from foreign shocks when capital mobility is ruled out, or Turnovsky and Bhandari (1982, p. 315), where one reads, “It is clear from the purchasing power parity relationship that if the spot rate were to appreciate exactly in proportion to the foreign price level increase then the domestic price and income levels would be unaffected by foreign price movements; … this will occur… if either the domestic demand for money is independent of the nominal interest rate, or if there is zero capital mobility.” Also, by assumption, domestic economic agents do not hold interest and/or noninterest-bearing assets denominated in foreign currency; this, too, by ruling out any wealth effect of changes in the exchange rate, helps to strengthen the insulation properties of the flexible rate system (Dornbusch (1973), Boyer (1976), Turnovsky (1979)).

18

For Sweden, consumer prices had to be used.

19

To estimate the equations, some observations were needed prior to the first quarter of 1973 for constructing the appropriate lags.

20

Besides, the R2 criterion has been adopted in McKinnon (1983).

21

To stress the peculiarity of the combination of annual data and wholesale prices, it should also be added that in this case the dummy variable does not turn out to be significant.

22

In fact, we are told that there are two channels through which currency substitution manifests itself: as the expected exchange rate varies, economic agents switch directly from one currency to another and from domestic currency into domestic bonds (and vice versa in the other country) because of the changes in interest rates induced by exchange rate expectations. To the extent that the latter channel is the major one (McKinnon, 1982, p. 326) and, as all the available empirical evidence seems to suggest, the interest elasticities of the national demands for money differ from one another, any change in the expected exchange rate is bound to translate into a shift in the world demand for money.

23

Here, both the developments in the field of index number theory and the problem of the different behavior of velocity in different countries and of different components of money within countries are being considered.

24

For instance, suppose that once the domestic monetary target has been set, money demand shifts down. The country will lose money to the rest of the world, and the process will continue unless the monetary target is brought into line with demand. But this fact means that the variable that should be targeted is not money but money plus the absolute value of the balance of payments deficit, that is, DCE. Therefore, if under fixed exchange rates instability in money demand comes into the picture, one is going to face much the same problem that McKinnon has in mind, making DCE targeting another alternative to McKinnon’s plan.

25

See also Chetty (1969) on this. In the context of this discussion, it is worth noting that when M3 is employed, no currency substitution is detected (Miles (1981)). Much the same criticism applies to Levy and Sarnat (1978). What they do is to compute mean and variances of the exchange rate changes of a basket of currencies vis-á-vis the U.S. dollar, and then ask the question as to what a U.S. investor would do to maximize the return on his investment, given the level of risk. The problem with this approach is that, as the authors themselves recognize, it does not explain the total demand for any currency over time; what it does explain is the portfolio portion of total demand. A second and major point is that by ignoring the alternative represented by bonds denominated in domestic currency, which gives a positive yield at no risk, even the portfolio component of the demand for the domestic currency (and, hence, for all currencies) is biased upward.

26

See, for instance, Bilson (1979), where the coefficient on the interest rate differential in the exchange rate equation is seen as a proxy for currency substitution.

27

For instance, the coefficient on the lagged dependent variable continues to be equal to one. The reader might also want to take a look at Boughton (1979) and Vaubel (1980), in which the inclusion of open-economy variables does not add much to the results.

28

Usually ϵ˙e is assumed to be given by the interest rate differential.

29

Money is expressed in real U.S. dollars; a weighted average of the four GNP deflators is used to deflate nominal money balances, which are also centered on midperiod. The reader will notice that the program that is used here estimates both the parameters and the constraints.

30

Here, one should be careful to distinguish between mathematics and economics; 0 replaces the variable ϵ˙USD.USDe

31

This statement partially contradicts previous results by Brillembourg and Schadler (1979), who obtained a positive cross elasticity between the U.S. dollar and the deutsche mark.

32

The reader might wonder about the relative values of bUSD and bJY. The fact that they appear to be off the mark is due to the particular behavior of the two stocks over time; the real per capita stock of U.S. dollars is a declining function of time (297,565 in the first quarter of 1967; 281,451 in the fourth quarter of 1981), while the stock of yen has been growing rapidly (54,384 and 212,564, respectively). Given the purpose of the exercise, there seemed to be no need to take care of the problem.

33

Computation was limited to the instances in which a t-statistic greater than one was obtained. Elasticities are defined as Mi*/dϵ˙i,je.

IMF Staff papers: Volume 30 No. 4
Author: International Monetary Fund. Research Dept.