Making Data Measurement Errors Transparent: The Case of the IMF

This article takes Oskar Morgenstern’s seminal On the accuracy of economic observations seriously. Morgenstern (1950) pointed out that – just as in the natural sciences – absolute precision and certainty are impossible to obtain in economic observations. Indeed our observations are often hampered by substantial measurement error. Unlike the natural sciences, economics in general does not report measurement errors for the key concepts such as prices, value or production that it seeks to measure and explain. Economics is not a natural science and its methodology must be different but the non-reporting of measurement error is unscientific and carries an important cost for society. Indeed, using data without consideration of the extent of measurement error has potentially significant implications for evidence-based policy-making (Reiss, 2016). [...]