Current Substitition and Money Demand in the United States, West Germany and Japan

Baade, Robert A.; Nazmi, Nader Article Current Substitition and Money Demand in the United States, West Germany and Japan Kredit und Kapital Provided in Cooperation with: Duncker & Humblot, Berlin Suggested Citation: Baade, Robert A.; Nazmi, Nader (1989) : Current Substitition and Money Demand in the United States, West Germany and Japan, Kredit und Kapital, ISSN 0023-4591, Duncker & Humblot, Berlin, Vol. 22, Iss. 3, pp. 363-374, https://doi.org/10.3790/ccm.22.3.363 This Version is available at: https://hdl.handle.net/10419/293147 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Currency Substitution and Money Demand in the United States, West Germany and Japan By Robert A. Baade and Nader Nazmi*, Lake Forest/Ill. In a 1973 article, Stephen Goldfeld concluded that a standard Keynesian money demand function specified as linear in the logs of the variables was a powerful predictive tool (Goldfeld 1973). Less than a year later, the Goldfeld equation began to seriously overpredict money demand. This shortfall of money demand from its expected levels has inspired explanations that run the gamut of economic science. Among the explanations offered for the overprediction is currency substitution. Some scholars have suggested that the currency composition of financial portfolios varies with alterations in currency prices. The purpose of this paper is to examine this thesis using more direct techniques than have been employed in the past. In explaining the failure of the Goldfeld model, some scholars have indicated that the Goldfeld experience is but another manifestation of limitations inherent in traditional econometric analysis (Hafer and Hein 1982; Rose 1984; Laidler 1985). Other analysts have argued that the "missing money" reflected an error in the specification of the Goldfeld model (Hamberger 1977). Then, too, there is a legion of researchers who believe that the missing money can be found by including independent variables that somehow capture the financial innovation spawned by structural changes in the global economy that occurred after 1973. This group basically subscribes to the idea that econometric technique has been underemployed in locating the lost money. A subset of this group of economists hypothesizes that developments in the internatonal monetary system in the past decade and one-half have been responsible for changes in the currency composition of cash balances (Miles 1978; Brittain 1981; McKinnon 1982; 1984; Baade 1985; Akhtar and Putnam 1979; Hetzel and Mehra 1986). Other scholars have concluded that currency substitution is not important in accounting for the missing money {Batten and Hafer 1984). * Vail Professor of Economics and Assistant Professor of Economics, respectively, at Lake Forest College. 24' OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.22.3.363 | Generated on 2023-01-16 12:58:09 364 Robert A. Baade and Nader Nazmi This paper draws its inspiration from the debate on currency substitution and has a two-fold purpose. First, evidence will be analyzed for the United States, West Germany, and Japan on whether currency substitution has occurred since the standard money demand equation began to yield unsatisfactory results, a time that roughly coincides with a global commitment to a system of flexible exchange rates. Second, evidence will be examined on the extent to which currency substitution explains errors in money demand predictions for the United States, West Germany, and Japan. The paper is organized as follows. Section one provides a more detailed review on the theory and evidence relating to the currency substitution thesis, and explains the particular contribution of this study. Sections two and three present the analysis corresponding to the twin objectives of the study. A summary and concluding remarks are provided in part four. I. Theory and Evidence Relating to the Currency Substitution Hypothesis Money demand in the U. S. was stable in the 1950s and 1960s, and the U. S. was the reserve currency country in a fixed exchange rate regime. Dollar demand ceased to be stable in 1973 and 1974, a time that coincides with a global adoption of a system of flexible exchange rates. Was this coincidence? Perhaps, but there is a theoretical basis for arguing that this development could have been anticipated. The only question seems to be one of degree. Are money holdings sufficiently transnational in character to help explain the shortfall in dollar demand? Casual empiricism supports a dollar substitution thesis for the mid 1970s. As economic events promised a reduction in the dollar's foreign currency values, actual dollar demand would be expected to fall. Thus it would follow that models for dollar demand that ignored the flexible exchange rate influence would overpredict dollar demand. The argument should be symmetric. Models that ignore the influence that alterations in foreign currency prices of dollars have on dollar demand, should underpredict dollar demand when the foreign currency price of dollars is expected to increase. Of course, it could be argued that the adoption of a system of flexible exchange rates induces a once and for all downward shift in money demand schedules reflecting the risk of greater exchange rate variability. Literature on the currency substitution question does not have a long history. Chow-Nan Chen (1973) may have been the first to argue that currency substitution undermines the much-heralded insulating power of a flexible exchange rate regime. However, Chen's conclusions qualify as conjecture since most of our experience with flexible exchange rates followed his work. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.22.3.363 | Generated on 2023-01-16 12:58:09 Currency Substitution and Money Demand 365 Marc Miles (1978) performed some early empirical tests on currency substitution between the U.S. and Canada. He concluded that currency substitution by the private sector has characterized the flexible exchange rate regime, and in effect replaced the "substitution service" provided by the public sector during the Bretton Woods era. While Miles' methodological approach based on a CES production function for money services was impressive, his findings were limited by both the number of countries studied and the number of observations. M. A. Akhtar and Bluford H. Putnam (1979) analyzed the relationship between exchange risk and the demand for money in Germany between 1972 - 76. Although their findings did not prove currency substitution, they concluded that the German demand for money schedule shifted downward with the advent of a system of flexible exchange rates. Like the Miles research, the Akhtar and Putnam study was limited by the data available relating to a flexible exchange rate system. One of the more complete and direct tests of the currency substitution phenomenon was undertaken by Dallas S. Batten and R. W. Hafer (1984). They included a variable that captured the expected return from holding foreign money balances (the threemonth forward premium /discount) in Goldfeld-type money demand equations for Canada, France, Germany, the Netherlands, and the United Kingdom for the 1966 - 1983 period. They found that only for Canada and Germany during the flexible exchange rate years did the expected return from holding money balances exercise a significant impact on money demand. Interestingly these findings on Canada and West Germany conform to the earlier work supporting the currency substitution thesis. Other research has explored the currency substitution question through less direct techniques than that employed by Batten and Hafer. Ronald I. McKinnon (1982, 1984) has advanced the currency substitution hypothesis through the development of a model which leads him to conclude that the global money supply is more closely correlated with U. S. prices than is the U.S. money supply. Henry Goldstein and Stephen Hayes (1984) and Christopher Radcliffe, Arthur D. Warga, and Thomas O. Willet (1984) reject McKinnon's world money supply hypothesis, and, in the process, take issue with his position on currency substitution. Robert L. Hetzel and Yash Mehra (1986) studied the relationship between U. S. inflation and the real exchange rate, and offered some insight based on a quantity theory perspective. They concluded that the relationship they observed between inflation and the real exchange rate very likely reflected one of three phenomena. First, the real exchange rate affects the rate of OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.22.3.363 | Generated on 2023-01-16 12:58:09 366 Robert A. Baade and Nader Nazmi inflation through a wealth effect that operates through the demand for money. Second, real exchange rate changes influence the currency composition of cash holdings, and, in turn, alter the rate of inflation through an alteration in interest rates. Third, changes in the real exchange rate directly influence the money supply and prices. Hetzel and Mehra tentatively endorse the second explanation, and the tenuousness of their currency substitution conclusion reflects the indirect test of the thesis. In summary, there is a paucity of evidence that directly addresses the currency substitution question. The issue is of vital interest since shifts in money demand schedules can influence interest rates, employment, and prices. Furthermore, evidence offered on the currency substitution question to date has been ambiguous. The empirical work summarized in the next two sections of the paper attempts to provide more conclusive evidence on the issue. Before the direct tests on the currency substitution thesis were conducted, it seemed logical to define money demand functions for the three countries we believed would be most significantly involved in any currency substitution. The estimated money demand equations for Japan, Germany, and the United States are reported in the next portion of the paper based on the premise that currency substitution can be captured and tested by analysing the residuals of the money demand equations for the three above mentioned countries. Tests on the currency substitution thesis are reported in Section III. Section IV offers some concluding remarks. II. Test Results Relating to Money Demand in Germany, Japan and the United Stated The data used to estimate money demand equations for Germany, Japan, and the United States1 were quarterly observations that were seasonally adjusted either at the source or by the authors using the Census X-ll program.2 Specifically, the following equations were estimated: (2.1) In (M/P)t>j = p0j + ft,, In (M/PJt.u + ftfjln(Ytij) + + ftjln (r\j) + ft.¿In (r2tij) + etJ , 1 The data for the U. S. were directly obtained from the Fed while those of Germany and Japan were found in various issues of International Financial Statistics. 2 Controversy on the advantages and disadvantages of using seasonally adjusted data is by no means resolved. Here we used seasonally adjusted data since almost all observations were seasonally adjusted at the source. (See Bell and Hillmer 1984 for a discussion of seasonal adjustment of economic time series). OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.22.3.363 | Generated on 2023-01-16 12:58:09 Currency Substitution and Money Demand 367 where, j =1,2,3 representing the United States, Japan, and Germany, respectively, M = Ml domestic nominal money stock, P = the consumer price index,3 Y = Real GNP (1972 dollars) r1 = passbook rate in the U. S. and the discount rate in Germany and Japan, r2 = commercial paper rate used in the case of the U. S. only e = stochastic error. The application of Durbin1 s (two-sided) M-Test4 to equation 2.1 indicated the presence of first-order autocorrelation for the United States and Japan, but no such problem for Germany. In particular, the regression of the error terms on all independent and dependent variables and on the lagged values of the residuals resulted in the ¿-statistics for the coefficients of the lagged error terms recorded in Table 2.1. Table 2.1 The t-statistic for the Durbin M-test Country t-statistics U.S. 1.7011 Japan -4.4523 Germany 1.1844 Given the results presented in Table 2.1, ordinary least squares (OLS) was used to estimate Germany's money demand equation and the Hildreth-Lu (H-L) (1960) procedure was employed to estimate money demand models for the U. S. and Japan. A refined grid value of 1 x 10~4 was adopted to guarantee the convergence of the H-L procedure to a global optimum. The complete estimation results are recorded in Table 2.2. It is observed that such standard money demand functions specified as linear in the logs of the variables has high explanatory power for all three countries. The adjusted coefficients of determination all exceed 96% indi3 The GNP deflator was used as a measure of this variable for the U. S. while in the case of Japan and Germany we employed the wholesale price to construct the CPI that was used instead of the GNP deflator. 4 Almost all other empirical studies that we know of in this field use the Durbin h-test. Our preference for the Durbin M-test stems from the Monte Carlo results of Spencer, 1975. OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.22.3.363 | Generated on 2023-01-16 12:58:09 368 Robert A. Baade and Nader Nazmi OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.22.3.363 | Generated on 2023-01-16 12:58:09 Currency Substitution and Money Demand 369 eating that at least 96% of the variation in money demand is explained through the specified equations. Furthermore, all variables have the expected signs and all are significant except for the commercial paper rate coefficient for the United States. The Chow test was then used to discern whether or not a structural shift that coincided with the adoption of a system of flexible exchange rates, had occured in the money demand equations. This test was performed for each quarter of 1974 separately. The null hypothesis of no structural change had to be accepted at 1 % level of significance for the first three quarters of 1974 and rejected for the fourth quarter of that year. The same test did not indicate any support for the hypothesis of structural change in 1975. The same behavior was detected for the Japanese and the German money demand equations. Table 2.3 provides the /-statistics for the Chow test for all three countries and for all four quarters of 1974. As can be seen, the evidence supports the occurrence of a structural shift in the money demand equations for the fourth quarter of 1974. This evidence pertains to all three countries and indicates a simultaneous change in demand for money in the United States, Japan and Germany which coincided with the advent of flexible exchange rates. Therefore it may be hypothesized that currency substitution played an important role in effecting the demand for money in the post-1974 period. In the next section such a possibility is examined and tested. Table 2.3 F-statistics for Test for Structural Change U.S. W. Germany Japan 1974.1 2.1 3.4 1.9 1974.2 1.8 3.2 1.6 1974.3 3.8 3.6 2.7 1974.4 8.5 9.2 7.1 III. Testing For Currency Substitution In order to examine the possible effect of currency substitution on the U.S. money demand, we regressed the residuals of the U.S. money demand equation on the error terms generated by the money demand equations for Germany and Japan. This test could help to detect a possible link between the overestimation of the United States' money demand equation and a posOPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.22.3.363 | Generated on 2023-01-16 12:58:09 370 Robert A. Baade and Nader Nazmi sible underestimation of the German and Japanese money demand models. The regression resulted in a very low coefficient of determination (R2 = .002) suggesting that the residuals (overestimation) of the money demand equation for the U. S. dollar cannot be significantly explained by the residuals (underestimation) of the money demand equations for the deutschmark and the Japanese yen. The same experiment was repeated by regressing the German money demand error terms on those of Japan and the U.S., and by regressing the Japanese residuals on the error terms of the money demand equations for the U.S. and Germany. For both cases low coefficients of determination were obtained (R2 = .031 and .032, respectively) and none of the independent variables (residuals from the money demand equations of the other two countries) were significant. However, these regression models performed much better when only the Post-1974.4 data was employed, with R2,s of .055, .108, and .116 for the three cases, respectively. While such tests are not particularly appropriate for direct examination of the currency substitution hypothesis, they shed some light on the problem. For example, the evidence suggests that the error terms from the Japanese money demand seem to be best explained by the error terms of the other two countries, while that of the U.S. is least explained by the error terms of Japanese and German money demand error terms. Therefore it seems plausible that the Japanese money demand could be more strongly influenced by currency substitution and that the U.S. money demand is perhaps least affected. An explanation for this may lie in the analysis of the size of a country's international sector relative to the size of its economy. In other words, since in the case of Japan the international sector constitutes a relatively large part of the economy, currency substitution will exhibit more of an effect on the money demand, while in the case of the United States the substitution effect is nominal given the relatively small size of its international sector. This possibility is further confirmed by examining the cross correlations for the residuals of the three money demand equations. Results of such a comparison are given in Table 3.1 where the numbers in parentheses represent relevant ¿-statistics. As can be seen from the Table, all correlations are insignificant prior to the advent of flexible exchange rates and are all significant for the post 1974.4 period. This evidence can provide further support for the currency substitution thesis. In order to present a more formal and direct test of currency substitution we added the residuals obtained from the Japanese and the West German money demand equations as independent variables for the money demand OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/ccm.22.3.363 | Generated on 2023-01-16 12:58:09