Abstract
David barr has raised some interesting issues concerning our paper on money demand in the European Monetary System (EMS). These issues can be divided into two groups: Barr’ s comments on our results, and his attempt to replicate these results both in and out of sample.
David barr has raised some interesting issues concerning our paper on money demand in the European Monetary System (EMS). These issues can be divided into two groups: Barr’ s comments on our results, and his attempt to replicate these results both in and out of sample.
I. Our Results
Barr makes two main points about our results. The first is with regard to our discussion of the estimated cointegrating equation. Here, Barr points out that the tests we report are not statistically valid in the context of a cointegrating equation. This is correct: the variables included in a cointegrating equation must all be nonstationary, and this contradicts the assumption on which the usual asymptotic tests are based. This is already well known. Nevertheless, many authors, like us, report these test results, because even if they are not strictly statistically valid they may still provide some (more informal) guidance in assessing how closely the data conform to the hypotheses. We had assumed that readers would understand that in this context such test results were only meant to be suggestive—indeed, as Barr puts it, “a useful guide but not a true statistical test” (p. 720).
Barr’s other main point concerns the interpretation of our dynamic equation—notably, the rapid adjustment speed combined with the presence of the thrice-lagged short-term interest rate in the equation. He is correct in his literal interpretation of our equation—and we agree that, taken at face value, this implies somewhat peculiar behavior. However, we thought it was generally agreed that this kind of equation was meant to be a parsimonious representation of a dynamic process that may be more complex, rather than a full description of that process.
Moreover, in line with Barr we would not expect our existing equation to be the perfect tool for a European central bank to use in formulating policy—of course, we never said it was. Our intention was much more limited: to point in the direction of cross-border monetary aggregation, suggesting the possibility that an exchange rate mechanism (ERM)-wide money demand equation might actually outperform equations estimated for the individual countries, and raise the associated policy issues.1
It seems obvious that to develop a money demand equation suitable for policymaking will require a substantial program of research. This will include a more careful interpretation of what underlies the EMS-wide money demand estimates, further empirical work with alternative specifications of money demand, alternative estimation techniques, and alternative definitions of money and other variables, as well as attempts to reconcile EMS-wide results with those for individual countries; such research on money demand would have to be supplemented by research on related topics, such as the predictive power of EMS-wide monetary aggregates. Such a body of research has already begun to develop; for example, Kremers and Lane (1992b) have examined, at a theoretical level, the implications of currency substitution and international portfolio substitution for money demand estimation, specifying what is gained and what is lost through cross-border aggregation. Artis (1991) and Bomhoff (1991) have re-examined the demand for narrow money in the ERM, providing support for our finding that this relationship may be stable. Monticelli and Strauss-Kahn (1991) have investigated the EMS-wide demand for broad money, finding a stable demand for M3. Angeloni, Cottarelli, and Levy (1991) have constructed monetary aggregates incorporating cross-border monetary holdings, showing that these outperform conventional national monetary aggregates. Lane and Poloz (1992) have estimated a common specification of money demand for the group of seven major industrial countries, looking for evidence of currency substitution and testing the implications of aggregating across the European countries, Bayoumi and Kenen (1992) test the predictive power of a European Community (EC)-wide monetary aggregate, showing that the ERM money supply performs at least as well as individual national aggregates in predicting nominal variables such as inflation and the price level. Barr’ s comment adds to this burgeoning literature, and undoubtedly, the other national central banks in Europe are currently studying these issues as well.
In short, we are mainly in agreement with the points that Barr makes about the interpretation of our results.
II. Barr’s Replication
Barr’s attempt to replicate our results in and out of sample attests to the difficulty of supporting an unchanged empirical specification on the shifting sands of substantial data revisions. A close examination of the within-sample results in Barr’s equations (3) and (4) suggests that they are actually quite different from equations (1) and (2) in our paper (Kremers and Lane (1990)). The coefficients in Barr’s static equation estimated for the same sample period2 are noticeably different from ours; although for the reasons mentioned tests of significance are not statistically valid, it is interesting that the differences between his equation and ours are more than two “standard errors” for every coefficient except that on the short-term interest rate. There are also differences between his estimated dynamic equation and ours; although these differences are only statistically significant in the case of the error correction coefficient, they are quantitatively large in other cases too (for example, his point estimate of the income elasticity of money demand is 0.4, while ours is 0.7). It is also noteworthy, as Barr reports, that his estimated equation for the same sample fails a test for structural break in 1984, which ours had passed.
Further investigation has revealed that, since our 1990 paper appeared, there have been considerable revisions in the International Financial Statistics (IFS) data for some countries included in the ERM aggregate. In particular, there have been substantial changes in the definition of money for France, Italy, and Denmark. Monetary data for France were revised in March 1990, as shown in Figure 1; this entailed an expansion of the money stock to include deposits held at rural savings institutions (Crèdit Mutuel Agricol et Rural). The Italian data were revised downward, owing to other reclassifications (in November 1989 and October 1990), as shown in Figure 2. In the case of Denmark, data were revised retrospectively in 1991, to conform to standards set in the EC’s directive of December 8, 1986 on the accounts of credit institutions. Deposits at less than one month’s notice, previously grouped with demand deposits, were classified separately as notice deposits; moreover, bank deposits held by mortgage-credit institutions, which had previously been included in money, were treated like interbank deposits and netted out in calculating narrow money (Danmarks Nationalbank (1991)). The reduction in measured money between old and new definitions is shown in Figure 3. Combining the revisions for these three countries, we have the substantial differences between old and new series for aggregate money for the ERM, as shown in Figure 4; these are associated with revisions in the quarterly changes of ERM-wide money, as shown in Figure 5.
France: Narrow Money
(Billions of francs, end of period)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.France: Narrow Money
(Billions of francs, end of period)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.France: Narrow Money
(Billions of francs, end of period)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.
Italy: Narrow Money
(Billions of lire, end of period)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.Italy: Narrow Money
(Billions of lire, end of period)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.Italy: Narrow Money
(Billions of lire, end of period)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.
Denmark: Narrow Money
(Billions of Kroner, end of period)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.Denmark: Narrow Money
(Billions of Kroner, end of period)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.Denmark: Narrow Money
(Billions of Kroner, end of period)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.
ERM Aggregate: Narrow Money
(Billions of deutsche mark)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.Note: Converted into deutsche mark at 1985 purchasing-power-parity rates. See Kremers and Lane (1990).ERM Aggregate: Narrow Money
(Billions of deutsche mark)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.Note: Converted into deutsche mark at 1985 purchasing-power-parity rates. See Kremers and Lane (1990).ERM Aggregate: Narrow Money
(Billions of deutsche mark)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.Note: Converted into deutsche mark at 1985 purchasing-power-parity rates. See Kremers and Lane (1990).
ERM Aggregate: Change in Narrow Money
(Logarithmic first differences)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.ERM Aggregate: Change in Narrow Money
(Logarithmic first differences)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.ERM Aggregate: Change in Narrow Money
(Logarithmic first differences)
Citation: IMF Staff Papers 1992, 003; 10.5089/9781451973174.024.A011
Source: IFS tape, August 1989 and November 1991.These changes in data leave us with the suspicion that we are dealing with a different monetary aggregate altogether. The fact that a specification of a money demand equation estimated for one aggregate does not provide a stable demand equation for a different aggregate should come as no surprise. Moreover, since our equation does not appear to fit the new data very well even within the existing sample period, the fact that it is unstable out of sample should raise few eyebrows.
Barr’ s results draw attention to the importance of re-examining any empirical specification, particularly when the data change. Further work of this sort will be needed in order to refine our understanding of monetary data and monetary relationships within the European currency area, particularly during the transition to economic and monetary union. Meanwhile, we would consider Barr’ s comment as no more than a useful cautionary note: clearly our results on the possibility of a stable demand for money in the EMS require further scrutiny, but for now the emerging empirical evidence quoted above would appear mainly to confirm our earlier econometric results and policy conclusions.
REFERENCES
Angeloni, Ignazio, Carlo Cottarelli, and Aviram Levy, “Cross-Border Deposits and Monetary Aggregates in the Transition to EMU,” IMF Working Paper 91/114 (Washington: International Monetary Fund, November 1991).
Artis, Michael J., “Monetary Policy in Stage Two of EMU: What Can We Learn from the 1990s?” (unpublished; Manchester: University of Manchester, 1991).
Barr, David, “The Demand for Money in Europe—Economic and Monetary Integration and the Aggregate Demand for Money in the EMS: Comment on Kremers and Lane,” Staff Papers, Vol. 39 (September 1992), pp. 718–29.
Bayoumi, Tamim A., and Peter B. Kenen, “Using an EC-Wide Monetary Aggregate in Stage Two of EMU,” IMF Working Paper 92/56 (Washington: International Monetary Fund, July 1992).
Bomhoff, Eduard J., “Monetary Policy and Inflation in the EC” (unpublished; Rotterdam: Erasmus University, 1991).
Danmarks Nationalbank, “The Money Stock and Domestic Money Creation: Amendments to the Tables and Graphs Section,” Danmarks Nationalbank Monetary Review (May 1991), p. 6.
Kremers, Jeroen J. M., and Timothy D. Lane, “Economic and Monetary Integration and the Aggregate Demand for Money in the EMS,” Staff Papers, Vol. 37 (December 1990), pp. 777–805.
Kremers, Jeroen J. M., and Timothy D. Lane, (1992a), “European Monetary Union Raises Questions About Policy Targets and Indicators,” IMF Survey, Vol. 21 (April 27), pp. 136–39.
Kremers, Jeroen J. M., and Timothy D. Lane, (1992b), “The Implications of Cross-Border Monetary Aggregation” (unpublished; Washington: International Monetary Fund, 1992).
Lane, Timothy D., and Stephen S. Poloz, “Currency Substitution and Cross-Border Monetary Aggregation: Evidence from the Group of Seven” (Washington: International Monetary Fund, forthcoming as a Working Paper, 1992).
Monticelli, Carlo, and Marc-Olivier Strauss-Kahn, “European Integration and the Demand for Broad Money” (unpublished; Economic Unit of the Committee of Governors of EC Central Banks, January 1991).
Jeroen J. M. Kremers is with the Finance Ministry of the Netherlands and Erasmus University in Rotterdam (OCFEB).
Timothy D. Lane is with the Research Department of the IMF.
The policy implications of these results are discussed in Kremers and Lane (1992a).
The sample periods in our 1990 paper were 1978:4–1987:4 for the static equation and 1979:1–1987:4 for the dynamic equation. An unfortunate typographical error in our published paper reported the sample period for the dynamic equation incorrectly as ending in 1984:4; this shorter subsample was in fact used for stability tests, but the estimates reported are for the longer sample. Barr’ s results are actually for a period with one fewer observation than ours: his samples start in 1979:1 and 1979:2 for static and dynamic equations, respectively.
