Reassessing the Role of Buffer Stock Money Under Oil Price Shocks.(Brief Article)

NICHOLAS APERGIS [*]

This paper uses the structural vector autoregressive approach to assess the significance of buffer stock money under alternative real shocks in the U.S. economy over the 1960-96 period. Buffer stock effects are shown to play a minor role when oil price shocks are explicitly considered. (JEL E41)

Introduction

The buffer stock approach to money demand, as developed by Carr and Darby [1981], argues that agents are willing to hold a proportion of unanticipated increases in money supply in a short-run buffer [Cuthbertson and Taylor, 1986]. According to this approach, anticipated changes in nominal money supply are quickly reflected in the price level with no effect on real money balances. By contrast, unanticipated changes in nominal money supply cause short-run changes in real money holdings. Mackinnon and Milbourne [1984] provide evidence against the buffer stock hypothesis, whereas Browne [1989] and Boughton and Tavlas [1990] provide evidence in favor of the hypothesis.

Lastrapes and Selgin [1994] investigated the buffer stock hypothesis using a bivariate vector autoregression (VAR) system with a just-identifying restriction, which renders a structural VAR system. The restriction prevents nominal money supply shocks from affecting real money balances in the long run. In particular, the long-run interaction is modeled between real and nominal money as:

[delta]m = [v.sup.[delta]M]

[delta]M = [a.sub.1] [delta]m +[v.sup.[delta]M], (1)

where M is nominal money and m is real money, whereas [v.sup.[delta]m] and [v.sup.[delta]M] are exogenous shocks, which are assumed to be serially and contemporaneously uncorrelated. They find that nominal shocks, [v.sup.[delta]M], significantly affect real money balances, mainly in the short run, which supports the buffer stock hypothesis. Lastrapes and Selgin [1994] argue that more proxies for real shocks should be used in the empirical analysis. In fact, Lastrapes and Selgin [1995] extend their original paper by including real income and a rate of interest as additional endogenous variables. The resulting empirical evidence still supports the buffer stock hypothesis.

This paper uses the methodology of Lastrapes and Selgin [1994] to consider oil price shocks as an explicit proxy for real shocks. Vaez-Zadeh [1989] argues that oil price shocks directly affect money demand due to the presence of a wealth effect, while Miller [1990] maintains that money demand schemes become highly volatile because of oil price shocks. Although these oil price shocks disturb equilibrium in the money market, the demand for money adjusts relatively quickly as these shocks allow interest rates, real income, and the price level to adjust rapidly. Thus, under oil price shocks, the buffer stock hypothesis should not receive strong support. Since 1973, the U.S. economy has experienced several sizeable (both negative and positive) oil price shocks. Hamilton [1983] makes a convincing case for their importance. In this paper, oil price shocks are evaluated relative to other supply shocks. The introduction of oil price shocks explicitly allows a disaggregation of supply shocks into oil price shocks and other supply shocks. It appears that this distinction has not been considered in literature so far.

The Model

The model of Lastrapes and Selgin [1994], which allows oil price effects to play an explicit role, …