Commodity Prices, Money Surprises, and Fed Credibility: Comment
纠正了Frankel和Hardouvelis（1985）论文中因商品价格数据收集日期错误导致的回归结果误差，重新估计后发现货币供应公告意外对商品价格的影响在1980-82年货币目标期更为显著，支持预期假说。
A recent paper by Frankel and Hardouvelis (1985) examined the impact of money supply announcement surprises on the commodity markets. The purpose of this comment is to correct some errors we discovered in the regression results reported in their paper. We received a copy of Frankel and Hardouvelis' data set from the Journal of Money, Credit, and Banking. While examining this data we discovered that approximately one-half of the commodity price data had erroneously been collected a day earlier than intended. Therefore we decided to reestimate their regression equations with commodity data collected from the Dow Jones News Retrieval Service. In each equation the dependent variable is the difference between the closing price of the next maturing contract one hour before the weekly money announcement and the opening price of the same contract the following morning. The independent variable is the difference between the announced change in the money supply (M 1) and the change forecasted (based on a survey by Money Market Services). As with Frankel and Hardouvelis, the equations were estimated for both the 1978-79 (pretargeting), and 1980-82 periods. The results are qualitatively similar to those reported by Frankel and Hardouvelis, although there are substantial differences for individual commodities such as sugar, cocoa, silver, and gold. Perhaps the most significant difference is that during the money targeting period (1980-82), the nine-commodity average is now significant at the .001 significance level (as compared to a .05 significance level found by Frankel and Hardouvelis). The new estimates provide even stronger evidence in favor of the anticipations hypothesis which views the market response during 1980-82 as an anticipation of future policy actions by the Fed. Neither the inflationary anticipations hypothesis nor the real activity signaling hypothesis would seem consistent with such a large qualitative change in market responses between 1978-79 and 1980-82.