From tables to microdata: A multivariate synthetic approach for studying economic inequality, with an application to The Netherlands 1950–1976

Long-run studies of income and wealth distributions based on tabular tax data focus almost exclusively on the top 10% or 1% and do not consider underlying dimensions of social stratification. This paper proposes a methodology which overcomes these limitations by creating multivariate synthetic data through a combination of three techniques: ungrouping, imputations and reweighting. The reliability of the methodology is shown through an evaluation based on modern EU-SILC data. The methodology is then applied to grouped income tax data from the Netherlands between 1950 and 1976. First results show that a narrow focus on the top of the distribution neglects a lot of patterns and trends that shape total income inequality. The multivariate synthetic microdata open up a large scala of research potential.