Kremers and lane (1990) analyze the money demand in the European Monetary System (EMS) over the period 1978–87. They conclude that aggregate demand for Ml in countries participating in the Exchange Rate Mechanism (ERM) is a stable function of income, inflation, interest rates, and the ECU-U.S. dollar exchange rate. Therefore, a European central bank might be able to implement monetary policy more effectively than the individual central banks.
In this comment, I claim that Kremers and Lane made two errors in the derivation of their aggregate data. The first concerns the use of purchasing power parity (PPP) exchange rates. The second has to do with the procedure that has to be followed in aggregating and deflating data. I then assess the sensitivity of Kremers and Lane’s cointegration results to the questions I raise.
Kremers, Jeroen J.M., and Timothy D. Lane, “Economic and Monetary Integration and the Aggregate Demand for Money in the EMS,” Staff Papers, International Monetary Fund. Vol. 37 (December 1990), pp. 777–805.
Ivo J.M. Arnold is an Assistant Professor in the Department of Monetary Economics at Erasmus University, Rotterdam.
The author wishes to thank Jeroen J.M. Kremers for providing the aggregate series used in this paper, and Eduard Bomhoff and Dave Smant for useful comments.
I use the nominal money series as an example. The money series are taken from the International Financial Statistics (IFS) datatape (line 34b). I obtained quarterly PPP exchange rates by linear interpolation of the yearly OECD (1991) data. All other data are as in Kremers and Lane.
This error should be distinguished from the other error in the use of PPP rates, discussed in Section I.
In footnote 18 in Kremers and Lane (1990, p. 794), the aggregate price level is defined as a four-quarter moving geometric average of ERM-wide CPI. I assume, in conformity with their Figure 3 (p. 788), that this definition applies only to the inflation variable in their equation (1) and not to the price level appearing in (1) as a deflator of nominal money.
The regressions with MPY and DMP as endogenous variables are Kremers and Lane’s original estimates. The preferred regressions are those with MPYN and DMPN as endogenous variables.
I followed Kremers and Lane in taking the growth rate of real money as an endogenous variable, although the growth rate of the liquidity ratio would be a more logical choice.