Reviving the Federal Statistical System: The View From Academia
指出官方统计指标看似明确，实则依赖假设和估算方法，其合理性取决于研究问题；例如零售数据估算消费对生产规划足够，但检验经济理论可能有害。
There is a tendency to think of official government statistics as unambiguous measures of economic activity. In truth, however, nearly all of the available series are based heavily on assumptions, and are sensitive to the estimation techniques used. Furthermore, many of these assumptions and estimation techniques have been refined and improved over time. Whether these underlying assumptions are reasonable and whether the refinement of assumptions over time has been useful depends crucially on the questions one is trying to answer. For example, an estimate of consumption derived from data on retail sales may be perfectly adequate for planning future production, or setting government budgets, but may be disastrous for testing a subtle economic theory. Similarly, gathering more genuine consumption data might improve our current estimates of consumption, but a series that reflects retail sales for one era and genuine consumption for another could wreak havoc when used in estimating a time-series rela