THE IMPACT OF SUBSTITUTION AMONG LIQUID ASSETS ON THE DEFINITION OF MONETARY AGGREGATES

In January, 1979, the Federal Reserve System published a set of new proposals for redefining existing monetary aggregates. The actual redefined aggregates were announced in February, 1980. These revisions were prompted by the heightened pace of regulatory changes and financial developments in recent years and by the numerous suggestions that the correlation between the conventional measures of money and total expenditure has weakened since the early 1970s. The purpose of the dissertation was to evaluate the merit of these suggestions and to determine the desirability of broader definitions of money. This required the derivation of broader monetary aggregates, which reflect the impact of these financial innovations, and a comparison of their stability and predictive performance with those of the conventional measures of the money stock. The dissertation emphasized M(,1) and M(,2) (because they are much closer to aggregate expenditure than other measures of the money stock). Four near-money assets were considered. These include bank time deposits, deposits at savings and loan associations and mutual savings banks and credit union share drafts. Because adequate data on the composition of the new forms of substitute financial assets are not available for the sample period (1964:1 - 1977:4), it was decided to derive, empirically, two broader monetary aggregates for M(,1) and M(,2) (i.e., M(,1)* and M(,2)*). The constant elasticity of substitution utility model was utilized for this purpose. First, the elasticity of substitution and the liquidity coefficient were estimated for each near-money asset. These statistics were, then, employed in aggregating all the near-money assets and M(,1) to form M(,1)* and M(,2)*. M(,1)* was considered an approximation of either M(,1)+ (in the proposal) or M(,1)B in the newly redefined monetary aggregates. M(,2)* also approximates M(,2) in the proposal but deviates slightly from the new M(,2) because M(,2)* excludes non-deposit financial assets such as RPs, money market mutual funds and the Eurodollar account. Finally the stability and predictive performance of M(,1)* and M(,2)* were compared with those of M(,1) and M(,2). Stability was evaluated by a dummy variable technique while the traditional money-income model and money supply equation (M = mB) were utilized in forecasting each of the four monetary aggregates. Specifically, for the money supply process, it was first assumed, that the monetary authorities can fully control the variations in B. Thus, the predictability of M would depend on the stability of the money multiplier. This was, later, relaxed and the money supply equation was simulated for each monetary aggregate. The statistical results obtained from the study reveal the following: (a) All the four near-money assets, considered in the study, are good substitutes for money. (b) Estimates of the liquidity coeffiicient indicate that credit union shares are the most liquid of the near-money assets included in the study. The low coefficient for bank time deposit does not support the suggestion that all bank time deposits should be included in a transactions definition of money. (c) All the four monetary aggregates were stable during the sample period. However, M(,1) was the most stable, followed by M(,2), M(,1)* and M(,2)*. This contradicts the proposition that recent financial innovations have produced a shift in the conventional monetary aggregates, particularly M(,1). (d) Overall, M(,1) was the most predictable of the four monetary aggregates. Also, M(,1)* has lower prediction error than either M(,2) or M(,2)*. This suggests that lower order definitions, whether narrowly or broadly defined, would be more useful than higher order definitions, in controlling aggregate expenditures. It also supports the recent decision by the Federal Reserve System to retain M(,1) without significant modification. (M(,1)A is not significantly different from the old M(,1))