Governments manipulate official Statistics: Institutions matter

Abstract

Many governments have been reported to systematically manipulate official statistics. However, scholarly research has not extensively dealt with the determinants of data manipulation, beyond the effect of autocracy. We extend the literature by including institutional factors hypothetically affecting data manipulation. Regressing the gap between GDP – predicted by night-time lighting data – and „official“ GDP on these institutional factors suggests that economic openness and democracy decrease manipulation, while decentralization increases manipulation. Political openness decreases manipulation for countries under-reporting GDP and increases manipulation for countries over-reporting GDP. Surprisingly, no effects are found for press freedom and the independence of the statistical office.