Rational expectations estimation of money demand for south asian countries

Abstract

In recent years, the demand for money function for some of the South Asian countries, namely India and Pakistan, has been estimated by many economists and econometricians alike [e.g., Fry, JDE, 1978; Gupta, JDS, 1970; Hu, JPE, 1971; Khan, PDR, 1980, 1982; and Wong, JME, 1977]. It is theoretically postulated that the amount of real money holdings by the economic agents depends on the level of real income and on the expected opportunity costs of the holding of real balances relative to other financial assets and goods. The opportunity costs of real balances and goods, in turn, depend upon their yields which are measured as the expected interest rate and the expected rate of inflation, respectively. The expected interest and inflation variables are generally unobservable and, therefore, in order to empirically estimate a money demand function that includes these variables, an assumption about the expectations formation is required. Most of the demand for money functions estimated for these South Asian countries assumed an adaptive form of expectations for the unobserved variables. The adaptive form of expectations is ad hoc and arbitrary and based only on the past information of the variable itself. To circumvent these problems, an alternative form of expectations has been suggested, known as the "Rational Expectations (RE) Hypothesis" [Muth, Econometrica, 1961]. The essential concept of rational expectations is fairly simple: economic agents form expectations efficiently using all available information. Although the demand for money functions under RE have been extensively estimated for developed eountries [Frenkel, AER, 1977; Khan, JPE, 1977; Salemi and Sargent, IER, 1979; Sargent, IER, 1977] and also for some Latin American countries [Liederman, IER, 1981], it appears that not much work has been done in estimating demand for money under RE for some of the economically and politically important countries in the South Asian region, namely India and Pakistan. This paper estimates the demand for money function for these two South Asian countries under RE using quarterly time series data covering the period from 1972-1 to 1985-4. The estimation technique used is a Full Information Maximum Likelihood Generalized Errors-in-Variables method proposed by Wickens [J. Eco. Studies, 1982] for RE models. The significance of this study is not limited to the methodology of rational expectations estimation of the money demand function for these countries but, more importantly, it is the empirical results which seem interesting and relevant for the conduct of monetary policies in developing countries. Some of the earlier empirical work by Gupta and Khan using the adaptive expectations approach found expected interest rates significant in the money demand function for these two developing countries. Contrary to these results, the FIML findings highlighted the importance of inflation rather than interest rate expectations in the money demand function. These results seem to be consistent for developing countries where the financial sector is relatively thin and heavily controlled; consequently, the interest rates are not competitively determined in the free markets.