The OECD cross-country comparison has been conducted by Daniel Mügge, the principal investigator. This subproject leveraged differences across the countries studied to understand better which factors have been decisive in shaping local measurement choices. For example, why do some countries include real estate prices in their inflation figures while others do not? Why do some jurisdictions count people as unemployed when they work less than 12 hours per week, while in other countries, one hour or more is enough for the “employed"-status? Like the other qualitative subprojects, this project has leveraged a mix of data sources: official reports, secondary – often rather technical – literature, and most importantly personal interviews, prominently including national statisticians in the countries studied, but also representatives of international organizations with which these have interacted, for example from the International Monetary Fund, the World Bank or the World Trade Organization, and civil society stakeholders.
PhD candidate Daniel DeRock’s research has focused on the work that international organizations such as the World Bank and the International Monetary Fund do to promote the dissemination of statistical standards around the world. These dissemination activities are not politically neutral. Often, DeRock has found, the standards in question are ill-fitting for developing countries. Hence, he has asked why the standards are promoted there nevertheless, who does so in particular, with what motivation, and what the response from the target countries is.
At the same time, we have dug into a number of specific country case studies. PhD candidate Joan van Heijster has ventured to China to understand better how this country, still rule by the Chinese Communist Party, has embraced an essentially capitalist economic measure since the late 1980s. PhD candidate Roberto Aragao has done the same for a whole range of economic measures in Brazil, post docs Juliette Alenda-Demoutiez and Francisca Grommé have investigated unemployment in South Africa and inflation measures in the Dutch Caribbean, respectively.
The quantitative subproject in FICKLEFORMS, focusing on trade data, has taken a completely different approach. Every international transaction is recorded twice – once by the sending country, and once by the receiving country. In theory, these two values should match. In practice, we find significant and persistent differences, indicating that such international economic data is a lot less reliable than the official figures suggest.
Using such mirror statistics, we have built an alternative dataset and replicated major academic studies. They show that academic insights are indeed sensitive to measurement error, even if to different degrees. But also after thorough analysis, it is not possible to specify the sources of data defects. I have therefore devised a range of strategies researchers should employ to improve the robustness of inferences they draw from large economic datasets.
