A Critique of Variance Bounds Tests for Monetary Exchange Rate Models: Note
指出Huang（1981）和Vander Kraats与Booth（1983）的方差界检验存在计算错误，他们使用的货币需求利率半弹性适用于年度数据而非月度数据，修正后汇率不再违反方差界。
The monetary approach to exchange rate determination attributes the observed volatility of floating rates to revisions of expectations about current and future disturbances that affect the supply and demand conditions in the money market see, for example, Frenkel and Mussa (1980). A number of recent studies have proposed and implemented variance bounds tests of monetary exchange rate models along the lines of bond-price and stock-price volatility tests pioneered by LeRoy and Porter (1981), Shiller (1979, 1981), and Singleton (1980). Huang (1981) presents evidence apparently demonstrating that some exchange rates are more volatile than a simple monetary model based on the assumptions of purchasing power parity and uncovered interest parity can rationalize. Meese and Singleton (1983) derive (but do not test) the variance bounds implied by a number of monetary models. Vander Kraats and Booth (1983) (henceforth referred to as V&B) implement volatility tests for three monetary models. They conclude that only a variant of the sticky-price model developed by Dornbusch (1976) can potentially explain (under certain rather implausible assumptions about parameter values) the observed volatility of the exchange rates included in their sample. In the literature, this evidence of excess volatility of floating rates is considered damaging to the monetary approach and/or to the rational expectations hypothesis. For example, in his recent survey of the literature, Richard Levich remarks that, on the basis of Huang's results, it follows that either the market is inefficient . . . or that the rational expectations monetary model is incorrect, or both (Levich 1985, p. 994). The primary purpose of the present paper is to point out that the excess volatility evidence presented by Huang and by AI&B arises from a calculation error. Namely, the value they assign to the interest semielasticity of money demand may be appropriate for a study using annual time series but is inappropriate for their monthly data. Once this error is corrected, their exchange rate series no longer violate the variance bounds they test.