The Reliability of U.S. Gross National Product
分析了美国GNP估计的修订和误差两个可靠性方面，指出修订幅度虽大但初步估计仍有信息量，误差可能大于或小于修订，并探讨了减少误差的方法，对关注经济数据质量的学者和政策制定者有用。
Two aspects of reliability—revisions and errors—need separate analysis. Revisions in gross national product (GNP) estimates are large in relation to the size of changes that matter to businesses and policy makers; nevertheless, preliminary estimates convey a lot of information about what the revised estimates will show. Errors in GNP estimates can plausibly be either larger or smaller than revisions. The distinction between gathering errors and adjustment errors helps us understand (a) the absence of any measure of the size of GNP errors and (b) the ways in which national-income statisticians go about reducing errors. These ideas suggest several guidelines for improving GNP estimates.