United States Trade Balance Announcements: The Nature of its Data Revisions
研究美国商务部月度贸易平衡公告中数据修正的性质，发现初步公告更像是真实值加测量误差，而非有效预测，对使用这些数据的分析师有参考价值。
This paper examines the nature of the data revisions in the monthly United States trade balance announcements made by the Commerce Department. Previous work in this area have considered agent reactions to only the preliminary trade balance announcements on various asset prices and have not taken into account the revised figures of the previous month's trade balance announcement. Since Ghosh and Lien (1995) found that the revisions in the data announced by the government are taken into account by the agents, it leads us to further investigate the nature of these revisions. This paper uses the econometric approach described in Mankiw, Runkle and Shapiro (1984) and Mork (1990) to analyze whether the revisions in the US trade balance announcements represent measurement errors, efficient forecast errors, or inefficient forecast errors. Using the generalized method of moments estimation procedure the researcher finds that the preliminary announcements of the trade balance data are better characterized as observations of the true series measured with error rather than as efficient forecasts. The results are of interest to the users of these preliminary US trade balance announcements since it suggests to them that they could use statistical procedures to estimate the underlying value and thus obtain more efficient estimates of these government announcements.