Demand for M1 in Nigeria

Abstract

This study addresses many of the pitfalls involved in the empirical modeling of M1 demand with data from an economic setting where there is no consensus on the background information for the choice of proxies for the variables suggested by theory. We explore alternative approaches for dealing with these choices and demonstrate the sensitivity of modeling outcomes to the different empirical specifications and the various econometric estimators. Our specification searches with quarterly data (1960I-1995IV) show that some of the approaches in extant literature lead to dead ends and reveal fruitful alternatives. In particular, we establish the strong relevance of interest rates in Nigerian M1 demand models over that period. On the contrary, the rate of inflation, which has been favored over interest-rates by some authors, had no place in the long run demand for M1 models recovered in this study. The exchange rate or its rate of depreciation also plays no role whatsoever in models. No role was found for either foreign interest rates or foreign interest rate differential in the empirical models of the demand for M1 in Nigeria. We recover economically sensible, econometrically sound, and policy relevant long run and short run empirical models of the demand for M1 in Nigeria. We find that the assumed long run relations hold between M1, total domestic expenditure, its deflator and two domestic interest rates. We establish that domestic absorption, not gross domestic product, is the relevant measure of transactions in the model for M1 in Nigeria. We also confirm that the deflator for domestic absorption, not consumer price index, is the relevant proxy for the general price level in the M1 model. We document significant 'own' and 'cross' price effects in the long run relations. The two domestic interest rates that entered the relations were the three-month time deposit rate and the Treasury bill rate. The time deposit rate mimicked the 'own-rate' of money in the equation, while the treasury bills rate captured the domestic 'cross-price effect', representing the domestic opportunity cost of money holding. Deletion of any of the two domestic interest rates in the long run models led to loss of the cointegrating relationships. Not only is a meaningful cross rate effect recovered only when the own rate is allowed for, both must be included for cointegration to occur. To be valid, empirical models for M1 in Nigeria must include at least two domestic interest rates, as proxies for the own-rate and the opportunity costs.