Long run money demand in the EU: Evidence for area-wide aggregates

Abstract

Long-run properties of EU-wide money aggregates are analysed.For each of the three aggregates considered -Currency, M1 and M3H -it is possible to obtain cointegrating relationships with GDP and interest rates (long or short term market interest rates). Results are not improved when traditional aggregates, obtained by aggregating existing national aggregates, are extended by the inclusion of various measures of Cross-Border Holdings. Specific attention is also paid to aggregation issues and the relative performance of area-wide and national equations. The results show that aggregation bias is not a major problem and that the relatively good area-wide performance is largely a consequence of a statistical averaging effect.

Currency Substitution and Money Demand in Euroland

“…This suggests that the omission of the foreign interest rate from the VAR system involves a misspecification error. If that is so, then this problem will be affecting all recent estimates addressing the euro-area money demand [Brand and Cassola, 2004;Funke, 2001;Coenen and Vega, 2001;Hayo, 1999;Fagan and Henry, 1999;Spencer, 1997]. Third, when the U.S. dollar interest rate is included among the regressors, the magnitude of its coefficient becomes similar to that obtained for the domestic interest rate.…”

“…As noted, for example, by Fagan and Henry [1999], the superiority of EU-wide estimates can be attributed either to a statistical averaging effect or to asymmetric shocks that offset each other in the aggregate, of which intra-European CS is a only special case. For direct tests suggesting the existence of intra-European currency substitution, see Melvin [1985].…”

“…The CS hypothesis is assessed by the sign and significance of b 2 : If the two currencies are substitutes, a decline in the user cost of foreign money will induce a cash switching from domestic money to foreign money. Recent estimates for the money demand in the euro-area may be interpreted as imposing b 2 = 0 in equation (1) [Brand and Cassola, 2004;Coenen and Vega, 2001;Funke, 2001;Hayo, 1999;Fagan and Henry, 1999;Spencer, 1997]. In the exercise below, this restriction is subject to a statistical scrutiny: If it holds, then the CS hypothesis is rejected.…”

“…Regardless the quality of the proxies used, it is remarkable that in all these studies the respective coefficient was found to be significant. It is, thus, surprising that the most recent money demand estimates for the euro-area [Brand and Cassola, 2004;Coenen and Vega, 2001;Funke, 2001;Fagan and Henry, 1999;Fase and Winder, 1998;Spencer, 1997] have been entirely devoted to the closed economy portfolio balance model.…”

“…A controversy exists, however, as to whether such stability is applicable to forecasting purposes. Indeed, since empirical studies have been based on constructed pre-EMU monetary aggregates, their validity has been questioned both in light of the Lucas critique and aggregation bias [Beyer et al, 2001;Arnold and de Vries, 2000;Fagan and Henry, 1999;Hayo, 1999;Spencer, 1997;Arnold, 1994].…”