Determinants of exchange rates between two major currencies, The

determinants of exchange rates between two major currencies, The

Murphy, A

This research examines over three decades of data on mark-dollar exchange rates and provides evidence that long-term exchange rates between dollars and marks are largely a function of relative wholesale prices in the U.S. and Germany. However, interest rate changes and the balance of trade are found to have an important transitory impact on exchange rates as well. The findings are important for demonstrating the significance of Purchasing Power Parity, interest rates, and trade flows on exchange rates over a long time interval that spanned systems of both fixed and flexible exchange rates.

This research investigates the determinants of the exchange rate between the two of the most important currencies in the world over the last 30 years: the U.S. dollar ($) and the German Mark (DM). The investigation is important not only for indicating what drives the relative value of these two important currencies but also for providing evidence on the factors that affect exchange rates in general.

In Section I, prior theories, hypothesis, and empirical studies of exchange rates are reviewed. In Section II, the empirical examination procedures and data are described. The results are evaluated in Section III, and the conclusion is written in Section IV.


As summarized by Murphy (1992), the exchange rate between two currencies is determined by the supply and demand for those two currencies. The supply and demand for currencies on the international markets is, in turn, determined by the buying power of the currencies in terms of both current consumption opportunities and investment for future consumption.

With the demand for currency to satisfy current consumption being a function of relative prices in different countries, exchange rates would be determined by the relative prices of some basket of goods in different countries if there were perfect markets and perfect elasticity of demand relative to price (Melitz, 1982). This latter theory, often called the Purchasing Power Parity (PPP) Theory, is supported by some empirical evidence indicating a positive relationship between price indexes and exchange rates, especially over long time horizons (Gaillot, 1970).

However, empirical studies have also uncovered evidence of a non-perfect relationship between relative prices and exchange rates (Daniel, 1986). The existence of heterogeneous consumer preferences and of market imperfections like taxes, tariffs, transactions costs, and asymmetric information may cause these significant deviations from PPP (Adler and Dumas, 1983).

In a world with imperfect markets, a number of factors other than the relative current prices of goods and services can affect exchange rates. In particular, the supply and demand for currencies for investment for future consumption may have an important effect on the exchange rates between currencies. The relative demand for currencies from investors will be affected by the relative interest rates that can be earned on invested funds in different countries. In addition, investors’ supply and demand for currencies will also be a function of the future expected purchasing power of the currencies, which in turn may depend on future inflationary expectations as well as on the current buying power of the currencies in the international markets. The general psychological outlook of investors may also affect their investment decisions and exchange rates (Bergstrand, 1983).


The prior section indicated that exchange rates between currencies in two countries could be hypothesized to be a function of relative prices, relative interest rates, and relative inflationary expectations in the two countries. To test the effect of each of these factors on exchange rates, a log-linear model can be specified. A log-linear form is appropriate because it permits the interrelationship between the independent variables to have a multiplicative effect (Murphy, 1992). For instance, investors’ demand for currencies is an interrelated function of current purchasing power, future inflation, and interest rates.

The dependent variable may be specified to be the log of the DM/$ exchange rate at the end of each month. The independent variables may be specified to be the U.S. balance of trade, the German balance of trade, the log of relative short-term interest rates in the two countries (i.e., the log of the short-term interest rate in the U.S. divided by the short-term interest rate in Germamy), the log of relative long-term interest rates, a separate variable for measuring the change in each interest rate in each country from the previous month (for a total of 4 separate variables), the log of the forward interest rate in the U.S., the log of the forward interest rate in Germany, the 1-month change in the forward interest rate in the U.S., the 1-month change in the forward interest rate in Germany, the log of the relative industrial production indexes in the two countries, the log of the relative share price indexes in the two countries, the log of the forecasted wholesale price index in the U.S. divided by the forecasted wholesale price index in Germany, and the actual deviation from expectations in this log ratio of forecasted wholesale prices (i.e., the difference between the forecasted log ratio and the actual log ratio). All data are obtained over the time interval May 1963 through June 1994 from the IMF’s International Financial Statistics (with the index data being adjusted for periodic resettings of the index values to 100).

Balance of trade variables are included in the regression to proxy for the relative demand for currencies relating to current consumption. Because of the effect on demand of factors such as market imperfections and non-perfect elasticity of demand for goods relative to price, balance of payments data may provide a better indication of the effect of relative consumption demand than price indexes that do not incorporate the effect of such factors (and which may not adequately adjust for quality differentials).

Because investor demand for currencies are a function of both short-term and long-term interest rates, independent variables are included for both. However, since investor demand for currency may be affected only transitorily by changes in interest rates (as opposed to the level of interest rates), changes in the interest rates variables are also included as independent variables. In addition, because investor demand for currency may also be affected by expected changes in interest rates, separate variables are also included for the level of and changes in forward interest rates. Forward interest rates are estimated using the formula

k = [(1+i^sub LT^)^sup 10^/(1+i^sub ST^)]^sup 1/9^ – 1,

where i denotes the interest rate on bonds with the subscripted long-term (LT) or short-term (ST) maturity (where it is being assumed that the maturity of the long-term bonds is ten years, while the maturity of the short-term bonds is one year).

Since investor demand for a currency may also be a function of the relative optimism about the state of the economy or the stock market, two additional variables are included to incorporate these factors. The log ratio of the two countries’ industrial production indexes and stock market indexes provide some indication of how the relative current (and forecasted) strength of the two economies affects investor demand for currencies.

Independent variables for relative wholesale price indexes are included in the regression because investor demand for currency is a function not only of interest rates but also of the relative future buying power of the currencies, which in turn is determined by the current price level and future inflationary forecasts. Since the supply and demand for currencies has been shown empirically to be related to the prices of internationally traded goods, wholesale price indexes are used, as opposed to retail prices that include many goods such as housing that are not traded across borders (Kravis and Lipsey, 1978). In order to incorporate the effect of both forecasted and unanticipated changes in wholesale prices (which affect investor forecasts), the relative wholesale prices are decomposed into their expected and unexpected components. Expected relative wholesale prices are estimated using a Box-Jenkins (1976) autoregressive integrated moving-average (ARIMA) model.

Because of the likely existence of autocorrelation in a regression using exchange rate levels as the dependent variable (without taking first differences of this variable), a Durbin-Watson (DW) test must be conducted (Judge et al., 1980). If the DW statistic is significant from 2 at the .10 level, a Generalized Least Squares (GLS) regression will be conducted using a Harvey (1981) two-step full transform method to estimate the autocovariances.

To eliminate the effect of insignificant variables, which can cause multicollinearity and distort paramater estimates and significance levels (Judge et al., 1980), a backward elimination iterative technique is used to purge statistically insignificant variables (Kennedy, 1979). This procedure can be set up to remove all variables from the regression whose significance level is less than .10.


The results of the test are shown in Table 1. Because the DW statistic of 0.32 indicated the existence of statistically significant autocorrelation, GLS was used. The findings in Table 1 indicate that the exchange rate between dollars and marks is largely a function of PPP and German variables. In particular, German interest rates and balance of trade data are found to have a significant effect on exchange rates, as do relative wholesale prices (both expected wholesale prices and deviations from expectations).

As expected, evidence is found that higher German short-term interest rates do drive up the value of the German currency. Evidence is also found that expectations of higher interest rates in Germany in the future (as estimated using the forward interest rate) also drive up the demand for the German mark. Thus, changes in both interest rates and interest-rate expectations in Germany do seem to have the predicted effect on the value of the German mark.

However, higher German long-term rates are associated with declines in the value of the German currency. One reason for the latter relationship may be due to the psychological effects of long-term bondholders suffering losses as a result of rising long-term interest rates and liquidating their holdings of German mark investments as a a consequence (Barron’s, 1994). In addition, higher long-term interest rates may also proxy for long-term inflation forecasts (Frankel, 1982) and may therefore reflect an expectation of a declining purchasing power for the mark in the future (and may reduce investor demand for marks as a result).

The regression results reported in Table 1 also provide direct evidence for the hypothesis that higher expected wholesale prices in Germany (relative to the U.S.) do result in a decline in the value of the German mark. Although it is difficult to arbitrage relative good price discrepancies from month to month, capital investors must consider the relative purchasing power of currencies when making their investment decisions, and so it is not surprising to find that expected real prices do affect exchange rates. In addition, because deviations from expectations with respect to real good prices can affect future inflationary expectations (e.g., higher inflation in Germany than expected can cause investors to increase their forecasts of inflation in the future also), the significantly negative relationship between the value of the mark and deviations from inflationary expectations is consistent with rational capital flows (e.g., higher than expected inflation in Germany can cause investors to believe the mark to have less purchasing power in the future and to therefore value marks less highly now).

The finding that the contemporary balance of trade surplus for Germany also affects exchange rates provides evidence that not just capital flows but also trade flows affect exchange rates. This result is important as the short-term demand for currency to finance capital flows dwarfs the demand to finance trade flows (Sesit, 1986). This variable may, however, also be proxying for investor sentiment concerning the relative future purchasing power of the currencies on the international markets.

The lack of significance for any of the U.S. variables provides some evidence that the relative demand for the DM is not signficantly affected by U.S. interest rates and trade flows. Thus, although the value of the dollar against other currencies may be influenced by the economic environment in the U.S., deviations from PPP with respect to the value of the dollar vis-a-vis the mark seem to be largely a function of the economic environment in Germany. However, multicollinearity may have contributed to the failure to find a statistically significant relationship here.

It is also important to note that the relative level of interest rates in the 2 countries seems to have no effect on currency values. This finding implies that the effect of interest rate changes is a transitory one that has no long-term direct impact on currency values.


The empirical result of this research provide evidence that rises in short-term interest rates can attract capital and temporarily raise currency values. Evidence is also found that expectations of higher interest rates in the future, as well as forecasts of relatively low inflation, can also draw investment and increase demand for currency. However, rising long-term interest rates are found to be associated with falling currency values, perhaps because long-term interest rates proxy for long-term future inflation forecasts and/or because rises in long-term interest rates result in losses to bond investors who send the money out of the country as a result of becoming psychologically discouraged. Some evidence is also found that contemporary trade flows also affect currency values.

One important implication of the study is that central banks can manipulate exchange rates with interest rate policies over short monthly intervals. However, longer-term, exchange rates are largely found to be a function of the relative prices of real goods that are forecast by capital market investors.


M. Adler and B. Dumas. “International Portfolio Choice and Corporation Finance: A Synthesis.” Journal of Finance 38 (1983), 925984.

Barron’s. “Mixed Grill. ” June 27 (1994), 33 50.

J. Bergstrand. “Selected Vies of Exchange Rate Determination After a Decade of ‘Floating’.” New England Economic Review (May/June 1983), 14-29.

G. Box and G. Jenkins. Time Series Analysis: Forecasting and Control. San Francisco: Holden-Day (1976).

B. Daniel. “Empirical Determinants of Purchasing Power Parity Deviations.” Journal of International Economics 21 (1986), 313-326.

R. Engle and C. Grnger. “CoIntegration and Error Correction: Representation, Estimation, and Testing.” Econometrica 55 (1987), 251-276.

J. Frankel. “The Mystery of the Multiplying Marks: A Modification of the

Monetary Model.” Review of Economics and Statistics 64 (1982), 515-519.

H. Gaillot. “Purchasing Power Parity as an Explanation of Long Term Changes in Exchange Rates.” Journal of Money, Credit and Banking 2 (1974), 348-357.

A. Harvey. The Economictric Analysis of Time Series. New York: Wiley (1981).

L. Hoskins. “A European System of Central Banks: Observations from Abroad.” Economic Commentary (November 15, 1990).

J. Johnston. Econometric Methods. McGraw-Hill: New York (1984), Third Edition.

G. Judge, W. Griffiths, R. Hill, and T. Lee. The Theory and Practice of Econometrics. New York: Wiley (1980).

P. Kennedy. A Guide to Econometrics. Cambridge: MIT (1979).

I. Kravis and R. Lipsey. “Price Behavior in the Light of Balance of Payments Theories.” Journal of International Economics 8 (1978), 193-246.

J. Melitz. “More About the Law of One Price.” European Economic Review 18 (1982), 325-344.

A. Murphy. “An Empirical Analysis of the Market for Inconvertible Currency.” Journal of International Financial Markets, Institutions & Money 2 (1992), 51-75.

M. Sesit. “Nations’ Devaluations of Currencies Spark a Global Trade War.” Wall Street Journal (December 22, 1986), 1.

The very useful research assistance of Ulrike Moeller is gratefully acknowledged.

Copyright College of Business Administration. University of Detroit Mercy Spring 1996

Provided by ProQuest Information and Learning Company. All rights Reserved