The Political Economy of Data Production

Abstract: Cross-national macroeconomic statistics are nearly ubiquitous in international relations and comparative politics research. While we know that these data can only be
measured with error, our reliance on them implies a belief that those errors are random, or, at a minimum, unrelated to the political phenomena we use them to understand. But
that is implausible. Measuring the economy is largely a state function, and the political-economic backdrop against which it occurs inherently shapes it. The implicit belief that
the politics of data production are inconsequential to political science research should be scrutinized. We examine this belief using a newly available dataset of ex post revisions to World Development Indicators data, with a particular focus on GDP growth statistics. We find that revisions to these data reveal a form of measurement error that is both consequential—simple political economy relationships vary substantially depending on which version of the data is used—and systematic. We focus particularly on the IMF’s role in the political economy of data production, but our analysis reveals other political factors that inform the scale and direction of ex post revisions.

摘要：跨国宏观经济统计数据在国际关系与比较政治学研究中几乎无所不在。尽管我们知道这些数据只能以存在误差的方式进行测量，但我们对其的依赖暗示了一种信念，即这些误差是随机的，或者至少与我们所用以理解的政治现象无关。然而，这一观点并不可信。衡量经济活动在很大程度上是一种国家职能，而其发生的政治经济背景本质上对其产生了塑造作用。那种认为数据生产中的政治因素对政治科学研究无足轻重的隐含信念，应当受到严格的审视。我们利用新近可获得的《世界发展指标》数据事后修订数据集，尤其关注国内生产总值增长率统计数据，来检验这一信念。我们发现，对这些数据的修订揭示了某种测量误差，这种误差既具有影响性——简单的政治经济学关系会因使用的数据版本不同而出现显著差异——又具有系统性。我们特别关注国际货币基金组织在数据生产政治经济学中的作用，但我们的分析揭示了其他影响事后修订规模和方向的潜在政治因素。