Exchange Rate Stability and Managed Floating: The Experience of the Federal Republic of Germany
Author:
Jacques R. Artus
Search for other papers by Jacques R. Artus in
Current site
Google Scholar
Close

The recent experience with floating exchange rates provides economists with what they have been waiting for for years: the opportunity to test their theories on the working of a floating rate regime for major industrial countries that have well-developed capital markets, relatively few restrictions on capital flows, and reliable statistical data, and that make intensive use of monetary policy, mainly to neutralize external inflationary and recessionary pressures. The Federal Republic of Germany is singled out here as the subject for analysis because it exhibits those characteristics to the fullest extent: namely, it has one of the most active monetary policies, detailed monthly data on foreign capital flows, and no official compensatory borrowing disguised as private capital flows. Its experience is a crucial source of empirical observation on how investors behave under current circumstances, and on whether financial capital flows have the benign effects that advocates of floating rates have foretold, or whether destabilizing speculation results in large erratic movements in the rate. Frequent recourse to intervention by the central bank also provides the opportunity to examine whether such intervention can attenuate certain defects of a floating regime.

Abstract

The recent experience with floating exchange rates provides economists with what they have been waiting for for years: the opportunity to test their theories on the working of a floating rate regime for major industrial countries that have well-developed capital markets, relatively few restrictions on capital flows, and reliable statistical data, and that make intensive use of monetary policy, mainly to neutralize external inflationary and recessionary pressures. The Federal Republic of Germany is singled out here as the subject for analysis because it exhibits those characteristics to the fullest extent: namely, it has one of the most active monetary policies, detailed monthly data on foreign capital flows, and no official compensatory borrowing disguised as private capital flows. Its experience is a crucial source of empirical observation on how investors behave under current circumstances, and on whether financial capital flows have the benign effects that advocates of floating rates have foretold, or whether destabilizing speculation results in large erratic movements in the rate. Frequent recourse to intervention by the central bank also provides the opportunity to examine whether such intervention can attenuate certain defects of a floating regime.

The recent experience with floating exchange rates provides economists with what they have been waiting for for years: the opportunity to test their theories on the working of a floating rate regime for major industrial countries that have well-developed capital markets, relatively few restrictions on capital flows, and reliable statistical data, and that make intensive use of monetary policy, mainly to neutralize external inflationary and recessionary pressures. The Federal Republic of Germany is singled out here as the subject for analysis because it exhibits those characteristics to the fullest extent: namely, it has one of the most active monetary policies, detailed monthly data on foreign capital flows, and no official compensatory borrowing disguised as private capital flows. Its experience is a crucial source of empirical observation on how investors behave under current circumstances, and on whether financial capital flows have the benign effects that advocates of floating rates have foretold, or whether destabilizing speculation results in large erratic movements in the rate. Frequent recourse to intervention by the central bank also provides the opportunity to examine whether such intervention can attenuate certain defects of a floating regime.

The present study considers only the very short-run adjustment process; price and economic activity levels in the goods markets as well as current balances, long-term capital flows, and trade-related capital flows are taken as given. It is an analysis of the nexus between short-term interest rates; private monetary capital flows, hereafter defined as those short-term private capital flows that are not linked to trade transactions (i.e., trade credits are excluded); and the exchange rate. The approach employed here is to specify a relatively small model of the balance of payments, including the demand for and supply of base money as well as policy reaction functions explaining the central bank net demand for domestic assets (monetary policy) and foreign assets (intervention policy). The core of the model consists of equations that explain private monetary capital flows. The model covers the period of floating rates from March 1973 to July 1975. Its parameters are estimated from monthly observations. Lessons as to the working of the floating regime are derived from the estimates obtained for certain parameters, in particular, those that determine how exchange rate movements are affected by changes in monetary policy and various external shocks.

Theoretical controversies as to the advantages and disadvantages of a floating regime are reviewed briefly. The focus in this review is on the empirical questions at issue—questions that can be settled only by observing empirical facts. Then, the model of the balance of payments of the Federal Republic of Germany is presented. Results of the econometric estimation of the parameters of the model, and simulations of the effects that certain monetary policies, intervention policies, and exogenous shocks have under the floating rate regime, are given next. Finally, lessons from the experience of the Federal Republic of Germany are summarized.

I. Theoretical Controversies and Empirical Issues 1

The relative merits of pegged and floating exchange rates have been discussed so often in the economic literature that an nth presentation of the various arguments is hardly called for. The only purpose of the following analysis is to point out the main issues under debate, issues on which the recent experience with floating might throw light.

National monetary independence is the most-often-quoted advantage of a floating rate system. Few, if any, economists have serious doubts as to the ability of the monetary authorities to control the money supply better in a floating rate system than in a pegged rate system. The real debate is about the efficiency of monetary policy in a floating rate system and, in particular, its cost in terms of the degree of variation in the exchange rate and its effects on other countries. The traditional Nurksian argument is that an increase in the domestic interest rate will cause the spot price of the domestic currency to rise in terms of foreign exchange. The higher interest rate is assumed to increase the foreign demand for interest-bearing assets issued by the country considered and to cut the demand by its residents for foreign assets. The resulting excess demand for the currency of that country is eliminated by an increase in its price. In turn, the rise in the spot exchange rate causes: (i) a fall in the local currency price of imports (and possibly of exports); (ii) a fall in the volume of exports; and (iii) an increase in the volume of imports. Thus, the efficiency of monetary policy in reducing demand pressures on the production system, and possibly inflation, is reinforced by exchange rate movements.2

This increased efficiency, however, is not itself without risks. More forcefully than anyone else, Nurkse (1944) has pointed out that this movement in the spot rate could become self-perpetuating, at least for a while, if market participants’ expectations as to the future value of the rate were revised in the same direction as the actual change in the rate. Ultimately, investors would realize that the movement of the rate had gone too far and a corrective movement would begin—a corrective movement that might itself push the rate too far in the opposite direction. Speculation would become destabilizing. Sharp variations in the exchange rate could discourage international transactions and have detrimental effects on prices and economic activity both at home and in trading partner countries. This is, in a nutshell, the present state of the argument about monetary policy and the exchange rate regime.

Most other arguments in favor of floating are based on the same assumptions as the one related to monetary policy. For example, floating has been presented as an efficient way to protect a country from foreign disturbances. Recently, this argument has been developed mainly with reference to imported inflation; however, it would apply equally well to various other kinds of disturbance. Floating has also been advanced as the best solution to the problem of adjustment where large exogenous shocks have a somewhat unpredictable effect on the relative position of the various countries, for example, the oil price increases. What is common to all those arguments are the assumptions, sometimes left implicit, about the substitutability between foreign and domestic financial assets and about the expectation behavior of investors. In all cases, the advantages claimed for the floating regime rest crucially on the assumptions that investors are willing to speculate and that in doing so they are guided by some stable expectations as to future exchange rate values, for example, expectations are based on the anticipated future purchasing power of the various currencies.

The role played by investors’ expectation behavior can be seen quite clearly by considering the factors that affect the additional yield, Π, derived from holding domestic bonds rather than foreign bonds. The yield, Π, is defined as 3

Π = R RF 100 [ S F S e F S e ] ( 1 )

where

article image

If the risk element is ignored, it can be assumed both that the yield, n, is zero in equilibrium and that the equilibrium is always maintained. Assuming first a change in the domestic interest rate, ΔR, the equilibrium condition will require (at the margin) that4

Δ % S = Δ % FS e + Δ R Δ R F ( 2 )

Assuming, for simplicity, a fixed foreign interest rate, the magnitude of the change in the spot rate will depend only on ΔR and on the extent of the change, if any, in the expected future value of the spot rate, Δ%FSe. Two major hypotheses are in contention to determine the change in the expected future spot rate. The first hypothesis, advanced by advocates of the floating regime, is that FSe is determined largely by basic factors, such as the expected relative purchasing power of domestic and foreign currencies. The second hypothesis, mentioned earlier, is that investors extrapolate to some extent current exchange rate changes into the future, that is, there is a direct positive causal relationship from Δ%S to Δ%FSe. The change in the spot rate, Δ%S, will be equal to ΔR under the first hypothesis, and to more than ΔR in the second one. Further, the second hypothesis is likely to lead to erratic movements in the exchange rate, since, ultimately, S will reach an obviously “absurd” level and corrective movements will take place.

The two expectation hypotheses can be considered to be particular cases of the following general expectation formula that allows for direct effects of “exogenous events”

Δ % F S e = d 1 L ( Δ % S ) + d 2 L ( Δ % FP R e ) + d 3 SPE ( 3 )

where FPRe is the expected future domestic price level relative to foreign prices, SPE is a vector of dummy variables for exogenous events, and the notation L(—) indicates a lag operator.

The first hypothesis assumes that d1 = 0 and d2 ≃ 1, and the second, that d1 > 0. The speed of the dynamic adjustment determined by the lag operators could accentuate or attenuate the stability of the adjustment.

Since speculation involves risks and most investors can be assumed to be risk averters, domestic and foreign securities are less than perfectly substitutable, so that the change in the shares of domestic and foreign securities in investors’ portfolios will be a function of the size of the change in the expected yield, Π. Equation (2) will not always hold. Thus, the efficient working of the adjustment mechanism depends on the degree of substitutability between domestic and foreign securities. At the limit, if the exchange rate has been fluctuating widely, the degree of risk (uncertainty) may be so high that an increase in the expected relative yield on domestic bonds caused by a rise in the domestic interest rate will have only a small effect on foreigners’ willingness to buy such bonds. High risk could isolate the various financial markets more efficiently than would capital controls.5 In such a case, an increase in the domestic interest rate may fail to have much effect on the spot exchange rate, whether or not it has any effect on the expected future rate.

Expectation formation and the substitutability between domestic and foreign securities determine not only the effect of monetary policy but more generally the effect of any “shocks” on S, R, RF, or FSe. In particular, any shock on S will cause a whole series of adjustments that push S farther away from its initial value if the expectation scheme is closer to the second hypothesis than to the first one. If the shock was random and temporary, the economic cost of the variation in S may be substantial. On the other hand, under the same hypothesis, intervention by the central bank in the foreign exchange market may be quite effective in moving the rate in the desired direction. Intervention would be inefficient if the degree of substitutability between domestic and foreign securities is high and exchange rate expectations are based on anticipated relative prices for domestic and foreign goods. At the limit, intervention would only finance speculation without affecting the rate. Of course, under these circumstances there would be no need for intervention either, since the rate would be stable and in “equilibrium” in the sense of reflecting relative prices in the goods markets.

Most of the arguments in favor of or against the expectation hypotheses under consideration, or relating to the influence of floating on the substitutability between domestic and foreign assets, are based on a priori reasoning. Mainly, as to expectation schemes, it has been argued that only schemes that do not consistently lead to unprofitable speculation should be considered plausible.6 A more recent argument is that investors should be assumed to be “rational.”7 This is a rather innocuous assumption; however, the problem is to define exactly what is meant by being rational. Being rational has often been shamelessly defined in the recent studies as accepting the simplified view of the world depicted by a given theoretical model. This may close the circle and insure the internal consistency of the model in question. It does nothing as far as determining how investors will actually behave under floating. While more useful, empirical studies of past experiences with floating have also been rather inconclusive. Basically, the interwar experience is too far behind us to be of much relevance to present-day conditions, and the Canadian experience is relevant only to the case of a relatively small country floating against its dominant partner.8 The recent experience of the Federal Republic of Germany should be richer in lessons on investors’ expectation behavior, the substitutability between foreign and domestic securities, and, more generally, on the interdependence between monetary policy, financial capital flows, and exchange rate variations in a floating rate regime.

II. A Model of the Balance of Payments

The model presented here for the Federal Republic of Germany draws upon previous work by Kouri and Porter (1972 and 1974), Argy and Kouri (1974), and Kouri (1975). However, it differs from the model developed by those authors in several ways: (1) the specification of investors’ expectations is the core of the present model, while expectations were either neglected in Kouri and Porter (1974) or arbitrarily based on the purchasing power parity in Kouri and Porter (1972); (2) the domestic interest rate in the money market is related to the marginal rate of interest for central bank credit (Lombard rate) and the level of banks’ free liquid reserves instead of being determined directly by the private sector’s demand for base money; and (3) policy reaction functions are specified not only for monetary policy, as in Argy and Kouri (1974), but also for intervention policy. Further, the model is designed to be estimated in its structural form, so that there is not the same need to neglect certain institutional facts and to simplify functional forms as in the reduced form approach employed in the studies mentioned earlier.9

The model is presented in Table 1, with the notation in Table 2, and a simplified flow chart as Chart 1. The model is composed of three parts describing, respectively, the money market, the balance of payments per se, and policy reaction functions. Schematically, the model determines simultaneously the value of the deutsche mark in terms of U. S. dollars and the interest rate on three-month deposits in the money market in Frankfurt, given various income and price variables; the interest rate on three-month Euro-dollar deposits in London; and certain assumptions as to factors that influence the monetary policy and the foreign exchange intervention policy of the Deutsche Bundesbank. The main characteristic of the model is its simplicity. Clearly, its purpose is to elucidate the empirical relevance of certain theoretical arguments, and to provide a framework for analyzing the recent experience of the Federal Republic of Germany with floating. It is not intended to be an “operational” model for elaborating policies or for forecasting exchange rates.

Table 1.

Model of the Balance of Payments

article image
Table 2.

Notation for the Balance of Payments Model1

article image

Prices, exchange rates, interest rates, and the various stock variables (FLR, MD, Ms, M*, NDAD, NFAD, NFAS) are measured at end of month. [Data on most stock variables at end of month are not available. In the analysis, monthly averages of daily data will be used for FLR in equation (4), MD in equation (7), and NDAD and NFAD in equation (12).] Flow variables (K, MF, and NMF) are measured over monthly periods. Here Δ indicates that the end-of-month to end-of-month change in the value of a variable is considered; Δ% indicates that the change is expressed in per cent at an annual rate. Interest rates and the inflation rate are also expressed at an annual rate. Stock and flow variables are expressed in billions of deutsche mark. Following the definition of a variable, † indicates that the variable is seasonally adjusted.

Monthly series were obtained by interpolation of quarterly series.

Chart 1.
Chart 1.

Simplified Flow Chart of the Balance of Payments Model

Citation: IMF Staff Papers 1976, 002; 10.5089/9781451947496.024.A002

The specification of the money market (equations (4)-(7)) is similar to the one used in the Deutsche Bundesbank’s model.10 It reflects the particular monetary policy setup that exists presently in the Federal Republic of Germany.11 The interest rate in the money market is determined to a large extent by the Bundesbank, at least on a short-run basis. The Bundesbank exercises its influence directly through its control of the Lombard rate (rate for advances on securities), LR, and indirectly by using its control on the supply of central bank money to affect the level of banks’ free liquid reserves. Monthly changes in the interest rate on three-month deposits in the money market in Frankfurt are explained by variations in the Lombard rate and the level of banks’ free liquid reserves—equation (4). Banks’ free liquid reserves represent the potential for expansion in the hands of the banks—a potential for expansion that makes them less dependent on the creation of central bank money by the Bundesbank.12 Bank’s free liquid reserves are equal to the difference between what are referred to hereafter as the supply of central bank money (or base money), Ms, and the demand for central bank money, MDequation (5). The supply of central bank money is equal to the sum of the central bank money stock and banks’ free liquid reserves. It could also be called the potential central bank money stock. As a balance sheet identity, the supply of central bank money is also equal to the sum of the (net) holding of foreign assets by the central bank, NFAD, and its (net) potential holding of domestic assets, NDAD13equation (6). The variable NDAD refers to the (net) amount of domestic assets that the central bank would need to acquire if banks were to convert all their free liquid reserves into base money; it is referred to hereafter as the central bank (net) demand for domestic assets. The demand for central bank money, MD, is proxied by the central bank money stock. It is explained by nominal income, the interest rate, and the rate of inflation with the usual partial adjustment of actual to desired values—equation (7).

The balance of payments part of the model is composed of the two behavioral equations (8) and (9) and the two identities (10) and (11). Equation (8), which explains private monetary capital flows, MF, is based on the portfolio analysis: capital flows reflect a modification in the composition of investors’ portfolios caused by a change in the relative yield-risk attributes of foreign and domestic assets. No variable is introduced as a proxy for the risk element. The period considered here is so short that it would not be possible to identify separately the effect of any change in the exchange rate risk that would have taken place over the period. However, the coefficient c1 for the yield variable in equation (8) will reflect the effect of the average degree of risk during the period. For the same reason, no attempt was made to introduce variables measuring domestic or foreign wealth.14 Equation (9) explains the change in the expected future value of the spot rate, Δ%FSe, which is an important explanatory variable in the preceding equation. It is identical to equation (3). The identity (10) defines the change in the net supply of foreign assets by the private sector and the federal government, ΔNFAS, as the sum of private monetary flows, MF, and nonmonetary flows, NMF, in the balance of payments. The equilibrium in the foreign exchange market is reached when the change in the central bank demand for net foreign assets, ΔNFAD, is equal to ΔNFAS (equation (11)).15

The policy reaction functions (12)-(15) need to be included to avoid the estimation biases that would be introduced in the rest of the model if changes in central bank demand for domestic and foreign assets were assumed to be exogenous in the context of the present model.16 Equations (12) and (13) describe the monetary policy. The central bank target for the change in the base money supply, ΔM*, depends upon the unemployment rate, U, and the inflation rate, Ṗ. Changes in the central bank demand for net domestic assets, ΔNDAD, are determined by the Bundesbank target for the change in the base money supply, ΔM*, given the intervention policy of the central bank (i.e., given ΔNFAD). The parameter e2 in equation (12) measures the ability of the central bank to neutralize unwanted changes in the base money supply resulting from its interventions in the foreign exchange market; a value of -1 for e2 would indicate complete neutralization. Equations (14) and (15) describe an intervention policy that aims at reducing deviations of the spot rate, S, from a target rate, S*, based on relative prices in the Federal Republic of Germany and the United States, and at minimizing the rate of change in the spot rate. Various alternative specifications of the intervention policy that were intended to take into account the European joint float were markedly unsuccessful. A likely explanation is that, over that period, the deutsche mark dominated other snake currencies.17

III. Empirical Results

For purposes of econometric estimation, equations (9), (13), and (15), which cannot be estimated directly, have been substituted, respectively, in equations (8), (12), and (14) to yield:

MF = c 0 + c 1 ( d 1 1 ) Δ % S + c 1 ( ΔR ΔR$ ) + c 1 d 2 Δ % FP R e + c 1 d 3 SPE ( 16 )
ΔND A D = e 1 h 0 + e 1 h 1 U + e 1 h 2 P ˙ + e 2 ΔNF A D ( 17 )
ΔNF A D = f 1 g 0 + f 1 S f 1 g 1 ( P / PUS 1 ) + f 2 Δ % S ( 18 )

Further, equation (16) was solved for the change in the exchange rate, Δ%S, rather than for MF. The reason for the change is that, in a floating regime, proximate determinants of the monetary flow, MF, are the nonmonetary flows, NMF, and the amount of intervention, ΔNFAD. Equation (16) explains the-exchange rate, given the amount of monetary flow determined elsewhere in the model; that is, the error term is in Δ%S rather than in MF. Taking into account the lag operators, the rearranged equation is

Δ % S = c 0 1 c 1 0 ( 1 d 1 0 ) M F + c 1 c 1 0 ( 1 d 1 0 ) L ( Δ R Δ R $ ) + c 1 d 2 c 1 0 ( 1 d 1 0 ) L ( Δ % FPR e ) + c 1 d 3 c 1 0 ( 1 d 1 0 ) L ( S P E ) + c 1 ( d 1 1 ) c 1 0 ( 1 d 1 0 ) L 1 ( Δ % S ) ( 19 )

where c10 and d10 indicate, respectively, the value of the parameter c1 and d1 that applies to the current value of a variable, and where the notation L-1 for a lag operator indicates that the weight of the current period is zero.

The equilibrium form of equation (19) is

Δ % S = c 0 1 c 1 ( 1 d 1 ) MF + 1 1 d 1 L ( ΔR ΔR$ ) + d 2 ( 1 d 1 ) L ( Δ % FP R e ) + d 3 ( 1 d 1 ) L ( SPE ) ( 20 )

In the empirical analysis, two dummy variables have been used as proxies for exogenous changes in the expected future spot rate. The first, REV, takes the value of one in June 1973 for the revaluation of the deutsche mark by 5.5 per cent against other snake currencies, and zero otherwise. The second, OIL, takes the value of one in October and November 1973 for the announcement of the oil embargo and cuts in oil production in the Middle East, minus two thirds in February, March, and April 1974 for the end of the oil shortage and the gradual change in what market participants considered to be the relative impact of the oil crisis on the Federal Republic of Germany and the United States; and zero otherwise.18 A case could be made for the introduction of dummy variables for the collapse of the Herstatt Bank, political crises in the Federal Republic of Germany and the United States, changes in capital controls, etc.; however, it was thought preferable to keep the number of dummy variables to a minimum.19 The current change in relative consumer prices, Δ%(P/PUS), with various lags, was used as a proxy for the change in expected relative prices, Δ%FPRe. The assumption here is that the expected value of the relative price level n months ahead (say, 3 months) is based on current and past relative price levels. Thus, the change in expected relative prices can be proxied by current and past relative inflation rates.

The econometric model is composed of the four identities (5), (6), (10), and (11), and the five regression equations (4), (7), (17), (18), and (19). All equations were estimated by employing two-stage least-squares regression methods. Econometric results are presented in Table 3.

Table 3.

Balance of Payments Model: Econometric Results 1

article image

The t statistics are shown in parentheses; coefficients significantly different from zero at the 5 per cent level are indicated by a star; R¯2 is the coefficient of determination adjusted for degrees of freedom; D-W is the Durbin-Watson statistic; SEE is the standard error of estimate; 3 and RHO is the estimated value of the autoregressive coefficient in equations where a first-order serial autocorrelation scheme was specified for the error term.

The three-month interest rate in Frankfurt is explained relatively well by equation (4). Variations in banks’ free liquid reserves play the crucial role. The Lombard rate is also an important explanatory factor, however; an increase of 1 per cent in the Lombard rate results in an increase of about 0.8 per cent in the money market rate.20 The period of observation starts in August 1971; the new series on free liquid reserves are available only since May 1971, and the months of May and June 1971 were excluded because of erratic capital flows preceding and following the beginning of the first floating period that ended in December 1971.

The demand for base money, equation (7), was estimated on a longer period, starting in January 1969. The various parameters have the right sign and plausible values. The long-run elasticity of the demand for base money with respect to nominal income is 1.10 when calculated at the mean values of the variables. Ultimately, increases of 1 percentage point in the inflation rate and in the nominal interest rate lead, respectively, to a cut in the demand for base money by 0.01 per cent and 1.56 per cent. The speed of adjustment is relatively slow; only 5 per cent of the total effect takes place during the first month. Thus, in the short run, the largest source of variations in banks’ free liquid reserves is the change in base money supply, not the change in base money demand.

Equation (19) explains more than 80 per cent of the monthly variance in the exchange rate.21 The change in the interest rate differential is the major explanatory variable. An increase of 1 percentage point in the interest rate differential in favor of Frankfurt (with interest rates expressed at an annual rate) will lead within a month to an increase of 2.4 per cent in the value of the deutsche mark in terms of U. S. dollars, other things assumed to be equal. The long-run effect will be 3.4 per cent.22 In the short run, a further fall of 1.1 per cent in the value of the deutsche mark is needed to give investors the incentive to buy an additional DM 1 billion worth of short-term financial assets of the Federal Republic of Germany. Taking into account the lagged exchange rate adjustment, the decline in the exchange rate will need to be close to 1.6 per cent. All effects were found to take place extremely rapidly. No explanatory variable having a lag exceeding one month was found to play any role in this equation. The coefficient of the lagged exchange rate change is also relatively small, 0.30; however, it is statistically significant. The relative price term does not play any role.23

The values of the structural parameters c1, d1, and d2—in equations (8) and (9)—can be calculated from the empirical results obtained for equation (19). The estimated value of c1 is 2.10; that is, the direct effect of an increase of 1 percentage point in the interest rate differential in favor of Frankfurt, adjusted for expected exchange rate changes, is to cause an inflow of funds in the Federal Republic of Germany amounting to DM 2.10 billion. The estimated value of d1 and d2, respectively, are 0.70 and 0.03. Investors revised their expectations in the same direction as, but by less than, the current exchange rate change (i.e., 1 > d1 > 0) and were not influenced by relative price developments (i.e., d2 ≃ 0). The Nurksian expectation hypothesis is the one that is validated by market developments during the past two years.24 These results are difficult to reconcile with analyses presented by advocates of the “random walk” view of exchange rate movements, such as Dooley, Shafer, and Smith (1974), Giddy and Dufey (1974), and Levich (1975). The significant value of the lagged exchange rate change in equation (19), in particular, implies either a certain bandwagon effect or the existence of an adjustment lag in the capital markets that should not be there if, in the short run, exchange rates were moving at random.

Equation (17) explains only about half of the variance in the central bank demand for net domestic assets. The unemployment rate, UH, and the rate of inflation, , have the expected signs; however, they are not statistically significant. Those results may be due to the shortness of the period of time considered, only 27 months; or they may reflect the inadequacy of the equation as a description of the monetary policy targets.25 The interesting result is the estimate of the sterilization coefficient e2, 0.80. The central bank was successful in sterilizing four fifths of the variations in the supply of base money caused by its intervention policy in the foreign exchange market. This result is markedly different from the estimate of about −0.2 obtained by Argy and Kouri (1974) for the Federal Republic of Germany from the quarterly series covering the period including the first quarter of 1963 through the fourth quarter of 1970, but it is similar to the estimate of −0.86 obtained by Willms (1971) for the period including the first quarter of 1958 through the second quarter of 1970.26 It seems that whenever the intervention in the foreign exchange market is not extremely large and continuously in the same direction, as it was from mid-1970 to March 1973, the Bundesbank has no difficulty in sterilizing its effects on the supply of base money, as it is defined here.

The intervention policy of the Bundesbank in the foreign exchange market is traced remarkably well by equation (18). The results can be unscrambled to obtain the parameter of the structural equations (14) and (15)

Δ NFA D = 0.463 [ 0.093 ] ( S S * ) + 0.359 [ 0.057 ] Δ % S ( ( 14 ) estimated )

where the bracketed numbers are the standard errors, and

S * = 40.2 54.8 ( P / PUS 1 ) ( ( 15 ) estimated )

Equation (14) indicates that, on average, the Bundesbank has been intervening at a monthly rate of about ½ billion deutsche mark (on the order of $200 million) for each US$0.01 of discrepancy between the current value of the deutsche mark in U. S. cents and its target value, that is, for a deviation of about 2½ percentage points. Further, it would appear that a significant effort was made to “lean against the wind”; for each percentage point of rise (fall) in the value of the deutsche mark over a one-month period, the Bundesbank has bought (sold) about ⅓ billion deutsche mark (approximately $150 million) of foreign exchange.

Equation (15) presents a plausible description of the target exchange rate, S*. The target rate indicated by the equation is DM 1 = US$0.40 whenever the relative price of consumer goods in the Federal Republic of Germany and in the United States is equal to the 1970 base-period value; and it depreciates by US$0,005 for each percentage point increase in the relative price of consumer goods in the Federal Republic of Germany (the mean elasticity is 1.3). In terms of the period considered here, S* was US$0.38 at the end of April 1973 (S = US$0,352) and US$0,413 at the end of April 1975 (S = US$0,421). Clearly, such results must be interpreted with a great deal of caution. The fact that certain policy reaction functions have been stable in the past does not imply that they will be stable in the future. As noted earlier, the main purpose of the policy reaction functions in the present model is to avoid certain estimation bias in the rest of the model.

Various simulations can be made with the model, but only three are presented here. The first traces the effect of a sudden deterioration of DM 1 billion in the balance of all nonmonetary external transactions of the Federal Republic of Germany, NMF, at a monthly rate. The second traces the effect of a one-shot increase of DM 10 billion in the central bank demand for net domestic assets, NDAD; in this simulation, equations (12) and (13) are excluded and NDAD is considered to be exogenous.27 In the third, the amount of intervention is taken to be exogenous; the simulation traces the effect of a one-shot increase of DM 10 billion in the central bank demand for net foreign assets, NFAD. All three simulations assume that the exchange rate S is equal to the target S* in the base period.

Results of the simulations are presented in Chart 2. Effects are calculated in cumulative form. Long-run effects are depicted only to show the equilibrium properties of the model; the model is specified to study the adjustment in the short term, say, within three months, and it should not be used to derive the long-run implications of various policy measures or other exogenous changes. Even for the short-term period, the results must be regarded as tentative.

Chart 2.
Chart 2.

Cumulative Effects of Various Shocks as Observed in the Balance of Payments Model

Citation: IMF Staff Papers 1976, 002; 10.5089/9781451947496.024.A002

1 The variable CMF is the cumulated amount of private monetary capital flows from period 0 to period t. In simulations (B) and (C), CMF is always equal to NFA.2 In months.

The deterioration in the balance of nonmonetary external transactions (Chart 2) leads to a moderate but persisting fall in the value of the deutsche mark, −2.0 per cent within the first three months. Gradually, the exchange rate diverges from its target value; and the central bank sells larger and larger amounts of foreign exchange. Effects of this intervention policy on the supply of central bank money are largely offset by an increase in the net domestic assets of the central bank, so that the domestic interest rate is not significantly affected.

The one-shot increase in the central bank demand for net domestic assets causes a once and for all drop in the domestic interest rate, −2.6 per cent within two months. There is also a sharp drop in the value of the deutsche mark during the first three months, −3.6 per cent. However, the exchange rate stops falling after three months and starts to rise gradually because of sales of foreign exchange by the central bank that aim at bringing the exchange rate back to its target level. The one-shot increase in the central bank demand for net foreign assets leads only to a sharp once and for all drop in the value of the deutsche mark, by 11.7 per cent the first month and a further 4.3 per cent the following month. Without further intervention by the central bank, the exchange rate will not move back to its target level. This, of course, reflects the fact that, as far as the short-run period considered here is concerned, investors were not found to base their expectations as to the future value of the deutsche mark on relative prices.

IV. Lessons from the Experience of the Federal Republic of Germany

Since the beginning of the floating regime in March 1973, fluctuations in the dollar rate of the deutsche mark have had a large amplitude and a short duration (three complete cycles in about two years). The oil crisis explains some of these fluctuations. However, this paper suggests an additional factor, namely, that short-term capital movements do not seem to have been stabilizing. Relatively small variations in the interest rate differential between Frankfurt and the Euro-dollar market have led to large fluctuations in the exchange rate. An increase of 1 per cent in the interest rate differential (at an annual rate) in favor of Frankfurt has caused, on average, an increase of 2.4 per cent in the dollar rate of the deutsche mark within a month and an increase of 3.4 per cent by the end of the adjustment period. There are also some signs that either the exchange rate adjustment to exogenous shocks is not instantaneous because of information lags and adjustment costs or that a bandwagon effect has been present: other things being equal, an increase of 1 per cent in the exchange rate in month t has led, on average, to a further increase of 0.30 per cent in the rate in month t + 1.

Up to the present time, the experience of the Federal Republic of Germany with floating does not bear out the faith that many economists have had in the stabilizing role of short-term capital flows. Obviously, the period is far too short to allow definitive conclusions to be drawn as regards the behavior of speculators in a floating regime. Possibly speculators need time to adjust to the new regime and will gradually learn from their past mistakes. Possibly there were so many destabilizing exogenous events during the past two years that the new regime has not yet had a fair chance. However, signs that a floating regime may lead to large fluctuations in exchange rates in the short run even for major industrial countries with relatively efficient capital markets are now too obvious to be ignored.

These findings raise serious questions about the use of monetary policy. Floating has increased the efficiency of monetary policy but also the risk of potentially harmful conflicts between the monetary policy targets of the various countries. Because of its large effect on the exchange rate, monetary policy has become an extremely useful instrument whenever it is appropriate from the point of view of domestic and external adjustment to have the domestic interest rate and the value of the national currency in terms of foreign exchange moving in the same direction. In other cases, it has become a potentially powerful beggar-my-neighbor policy because of its large exchange rate effect. Thus, the need for the major countries to harmonize their monetary policies is at least as great now as before March 1973. Intervention in the foreign exchange market by the central bank seems also to be an efficient means of influencing the exchange rate; and therefore it presents the same advantages and risks as monetary policy.

REFERENCES

  • Aliber, Robert Z. (1962), “Speculation in the Foreign Exchanges: The European Experience, 1919-1926,” Yale Economic Essays, Vol. 2 (Spring 1962), pp. 171245.

    • Search Google Scholar
    • Export Citation
  • Aliber, Robert Z. (1975 a), “Monetary Independence Under Floating Exchange Rates,” Journal of Finance, Vol. 30 (May 1975), pp. 36576.

    • Search Google Scholar
    • Export Citation
  • Aliber, Robert Z. (1975 b), “Monetary Independence Under Fixed and Floating Exchange Rates” (mimeographed, 1975).

  • Argy, Victor, and Pentti J. K. Kouri, “Stabilization Policies and the Volatility in International Reserves,” in National Monetary Policies and the International Financial System, ed. by Robert Z. Aliber (University of Chicago Press, 1974), pp. 20930.

    • Search Google Scholar
    • Export Citation
  • Black, Stanley Warren, International Money Markets and Flexible Exchange Rates, Princeton Studies in International Finance, No. 32, International Finance Section, Princeton University (1973).

    • Search Google Scholar
    • Export Citation
  • Caves, Richard E., and Grant L. Reuber, Capital Transfers and Economic Policy: Canada, 1951-1962 (Harvard University Press, 1971).

  • Dooley, Michael P., Jeffrey R. Shafer, and Ralph W. Smith, “Lessons from Exchange Market Experience: March 1973—Present” (mimeographed, July 1974).

    • Search Google Scholar
    • Export Citation
  • Fleming, J. Marcus, “Domestic Financial Policies Under Fixed and Under Floating Exchange Rates,” Staff Papers, Vol. 9 (November 1962), pp. 36980.

    • Search Google Scholar
    • Export Citation
  • Friedman, Milton, “The Case for Flexible Exchange Rates,” Essays in Positive Economics (University of Chicago Press, 1953), pp. 157203.

    • Search Google Scholar
    • Export Citation
  • Giddy, Ian H., and Gunter Dufey, “The Random Behavior of Flexible Exchange Rates: Implications for Forecasting” (mimeographed, 1974).

    • Search Google Scholar
    • Export Citation
  • Hodjera, Zoran, “Some Problems in Analyzing International Capital Movements” (mimeographed, 1975).

  • Kindleberger, Charles P., “Lessons of Floating Exchange Rates” (mimeographed, 1974).

  • Kouri, Pentti J. K., “The Hypothesis of Offsetting Capital Flows: A Case Study of Germany,” Journal of Monetary Economics, Vol. 1 (January 1975), pp. 2139.

    • Search Google Scholar
    • Export Citation
  • Kouri, Pentti J. K., and Michael G. Porter (1972), “Interest Rates, Capital Flows, and Exchange Rate Expectations” (unpublished, International Monetary Fund, October 24, 1972).

    • Search Google Scholar
    • Export Citation
  • Kouri, Pentti J. K., and Michael G. Porter (1974), “International Capital Flows and Portfolio Equilibrium,” Journal of Political Economy, Vol. 82 (May/June 1974), pp. 44367.

    • Search Google Scholar
    • Export Citation
  • Levich, Richard M., “Efficient Markets and International Financial Management” (mimeographed, 1975).

  • McKinnon, Ronald I., “Floating Foreign Exchange Rates, 1973-1974: The Emperor’s New Clothes” (mimeographed, 1975).

  • Mundell, Robert A., “The Monetary Dynamics of International Adjustment Under Fixed and Flexible Exchange Rates,” Quarterly Journal of Economics, Vol. 74 (May 1960), pp. 22757.

    • Search Google Scholar
    • Export Citation
  • Nurkse, Ragnar, International Currency Experience: Lessons of the Inter-War Period, League of Nations (Geneva, 1944).

  • Officer, Lawrence H., An Econometric Model of Canada Under the Fluctuating Exchange Rate (Harvard University Press, 1968).

  • Rhomberg, Rudolf R., “Canada’s Foreign Exchange Market: A Quarterly Model,” Staff Papers, Vol. 7 (April 1960), pp. 43956.

  • Riechel, Klaus, “The Hypothesis of Offsetting Capital Flows: A Case Study of Germany” (mimeographed, 1975).

  • Shiller, Robert J., “Rational Expectations and the Dynamic Structure of Macro-economic Models: A Critical Review” (mimeographed, 1975).

    • Search Google Scholar
    • Export Citation
  • Sohmen, Egon, Flexible Exchange Rates: Theory and Controversy (University of Chicago Press, 1961).

  • Tsiang, S. C., “Fluctuating Exchange Rates in Countries with Relatively Stable Economies: Some European Experiences After World War I,” Staff Papers, Vol. 7 (October 1959), pp. 24473.

    • Search Google Scholar
    • Export Citation
  • Willms, Manfred, “Controlling Money in an Open Economy: The German Case,” Federal Reserve Bank of St. Louis, Review, Vol. 53 (April 1971), pp. 1027.

    • Search Google Scholar
    • Export Citation
*

Mr. Artus, Assistant Chief of the Special Studies Division of the Research Department, holds degrees from the Faculty of Law and Economics in Paris and from the University of California at Berkeley.

1

This section draws upon the discussion of the arguments for floating rates made in Kindleberger (1974) and Aliber (1975 a).

2

Recent advocates of this argument include Mundell (1960), Sohmen (1961), and Fleming (1962).

3

More precisely, Π = R/100 − [S(1 + RF/100)/FSe − 1]—a formula that can be simplified into equation (1) by assuming that the exchange rate gain on the interest payment (i.e., the term RF(S − FSe)/FSe) is negligible.

4

Here Δ indicates increments in variables; Δ% indicates that increments are expressed in per cent. Formula (2) is a numerical approximation for Δ%S = (R − RF) (FSe/S−1) − (R−1 −RF−1) (FSe−1 − S−1) + 100 (FSe − FSe−1)/S−1. The approximation is based on the assumption that the scalars FSe/S−1 and FSe−1/S−1 are relatively close to one.

5

Some signs that such an effect was at work in 1973–74 have been noted by various authors; see Kindleberger (1974, pp. 12–13) on the decline in Euro-bond lending and McKinnon (1975) on the lack of speculative flows. However, most authors, including Kindleberger, have been rather skeptical about this danger.

6

See Friedman (1953), among others.

7

For a critical analysis of the rationality assumption applied to the specification of expectation schemes, see Shiller (1975). For an application to exchange rate expectations, see Black (1973).

8

See Tsiang (1959) and Aliber (1962) for a study of the interwar experience, and Rhomberg (1960), Officer (1968), and Caves and Reuber (1970) for a study of the Canadian experience.

9

Critical analyses of the Argy-Kouri-Porter approach by Riechel (1975) and Hodjera (1975) deal at length with this weakness in their work.

10

See “Structure and Results of the Econometric Model of the Deutsche Bundesbank,” Deutsche Bundesbank, Monthly Report, Vol. 27 (May 1975), pp. 26–32.

11

Descriptions of that country’s monetary policy can be found in the Deutsche Bundesbank’s Monthly Report, “Redefinition of Banks’ ‘Free Liquid Reserves,’” Vol. 25 (June 1973), pp. 43-44, and in “Central Bank Money Stock and Banks’ Free Liquid Reserves—Notes on the Bundesbanks Liquidity Calculation,” Vol. 26 (July 1974), pp. 14-23.

12

Banks’ free liquid reserves include banks’ excess balances, open market paper that the Deutsche Bundesbank has promised to purchase, unused rediscount quotas, and—up to May 1973—scope for raising Lombard loans. Banks’ free liquid reserves play the role of a buffer stock, or cushion, between the quantitative measures of the Bundesbank and their effects on the central bank money stock. Whenever this cushion is relatively thin, as it was over the period considered here, the Bundesbank enjoys nearly direct control over the creation of central bank money. However, if banks’ free liquid reserves are extremely large, quantitative measures by the Bundesbank become ineffective.

13

No series are available on NDAD, as defined here. Series on NDAD were calculated by subtracting NFAD from MS.

14

A dummy variable of the zero-one type was initially introduced to take into account the change in the rate of growth of foreign liquid wealth resulting from the increase in oil prices at the end of 1973. However, it was excluded later because its estimated coefficient was small, not statistically significant, and had the wrong sign.

15

Both NFAD and NFAS are valued here at constant exchange rate parity.

16

See Argy and Kouri (1974) and Hodjera (1975) for a discussion of the estimation biases introduced in any model of international capital flows by the assumptions that the change in the net domestic assets of the central banks is an exogenous variable.

17

During the period under consideration, the currencies of the countries participating in the joint float were the Belgian franc, the Danish krone, the Netherlands guilder, the Norwegian krone, and the Swedish krona, as well as the deutsche mark. France temporarily withdrew from the European joint float on January 21, 1974, and rejoined it on July 10, 1975.

18

A certain element of arbitrariness is necessarily involved in the choice of such a dummy variable. In the present case, the net effect of the dummy variable is constrained to be nil, not because the oil shortage had no net effects but because ultimately whatever net effects it had should take place through other variables that were already included. Further, the beginning and end of the period of the oil shortage were assumed to be more important in the present case than the timing of the oil price increases (October and December 1973).

19

The existence of capital controls, mainly the controls on capital inflows into the Federal Republic of Germany in 1973, may have led to a downward bias in the estimated value of certain coefficients, such as c1.

20

The Lombard rate is an independent instrument of monetary policy only from a short-term point of view. It is used “to stabilize” the money market on a month-to-month basis. In the long run, the level of the Lombard rate reflects mainly the inflation rate.

21

It still explains about 75 per cent of the variations in the exchange rate when one excludes the six observations that are explained with the help of the two dummy variables.

22

The multiplier is equal to 1/(1−0.296), or 1.420.

23

Averages of relative prices over preceding periods of three and six months were also introduced in the equation without any success.

24

A similar conclusion was reached by Aliber (1975 b) from a comparison of the amplitude of changes in exchange rates and interest rate differentials.

25

Most other studies have also found it difficult to capture monetary policy targets in a simple mechanical formula. For example, see Argy and Kouri (1974).

26

Differences in the definition of the relevant variables may account for part of the differences in the results of those various studies. In particular, NDAD is defined in the present study as the difference between the sum of the central bank money stock and banks’ free liquid reserves (i.e., MD + FLR) and the (net) demand by the central bank for foreign assets (NFAD), rather than as the amount of net domestic assets of the central bank, as in the studies by Willms (1971) and by Argy and Kouri (1974). The variable NDAD, as defined in the present study, is under the direct control of the Bundesbank, so that it is not surprising to find that it varies so as to sterilize unwanted effects of the intervention of the Bundesbank in the exchange market on the supply of base money.

27

Implicitly, this simulation exercise assumes that banks’ free liquid reserves were not abundant before the increase in NDAD. As indicated in footnote 12, once free liquid reserves are reaching a high level, further increases may have smaller and smaller effects on the interest rate in the money market.

  • Collapse
  • Expand
IMF Staff papers: Volume 23 No. 2
Author:
International Monetary Fund. Research Dept.