Buffer-Stock Money: Interpreting Short-Run Dynamics Using Long-Run Restrictions: A Comment
评论了Lastrapes和Selgin（1994）用双变量VAR检验缓冲库存货币假说的论文，指出该假说要求货币存量外生且个人以货币作为收支缓冲，但LS的识别方法可能存在问题。
In a recent paper, Lastrapes and Selgin (1994) [LS] test the buiTer-stock hypothesis using a bivariate vector auto-regression (VAR) system. The buiTer-stock hypothesis as originally exposited in Laidler (1982) is that increases in the money supply engineered by the monetary authorities (that is, independent of money demand factors) remain in individuals' bank accounts instead of triggering portfolio substitutions. Since the price level is assumed to be slow to adjust, this implies an increase in real money balances. This real balance eSect is supposed to increase consumption which slowly increases the price level until real money balances are restored to their original position. During the transition, real money balances are above their equilibrium level; this is oiTered as an explanation for the instability of money demand equations. There are two necessary components of the traditional buiTer-stock hypothesis. First, individuals must use money as a buiTer (hence the name) between receipts and expenditures; that is, they do not monitor their cash balances at each moment but allow cash balances to wander within certain bounds. This assumption is necessary but not sufficient. In addition the money stock must be exogenous. This is discussed more fully in Laidler (1982) and Milboume (1988) and the reason is as follows. Suppose that an increase in the money supply comes through an increase in the money balance of one individual. If this individual spends this, thus starting oiT the transmission process, it goes to another individual. If this triggers a portfolio adjustment by the second individual, which results in a removal of the excess money from the system, then the aggregate money stock retums to its original level, and no further transmission takes place. Thus, for the buiTer-stock transmission mechanism to work, money cannot be expunged from the system as a result of a portfolio substitution. The aggregate money stock must be able to be controlled by the authorities and unaiTected by the actions of individuals. The LS paper is innovative in that it tests the hypothesis using a bivariate VAR system with a just-identifying restriction. The authors argue that this avoids the