The statistical trilemma: built-in limitations of international economic statistics

Abstract

Economic statistics are central to global economic governance. They are the informational background to the Sustainable Development Goals, conditional lending by international organizations, and other dimensions of development policy. But there is a growing chasm between aspirations for economic statistics and what they can deliver on the ground. We argue that these shortcomings are rooted in what we call a trilemma of official statistics, a general limitation that goes beyond methodological deficiencies of individual indicators. Data users demand that economic statistics should (a) use harmonized standards to be comparable, (b) be guided by standards prescriptive enough to guarantee reliability and prevent manipulation, and (c) be suited to local socioeconomic contexts. Yet as we show, complex statistics can only satisfy two of these conditions at once. Importantly, we can only increase the suitability of such statistics to local contexts if we make concessions on either prescriptiveness or harmonization. We examine three cases in detail: national accounts statistics, poverty lines, and unemployment statistics. To strengthen external validity, we also briefly consider inflation, trade, and debt statistics as additional cases. The statistical trilemma clarifies the inevitable trade-offs statisticians face when designing measurement standards and thus the role they can – and cannot – play in global governance.