Anonymized Firm Data under Test: Evidence from a Replication Study

This paper contributes to the literature on the use of anonymized firm level data by reporting results from a replication study. To test for the practical usefulness of anonymized data I selected two of my published papers based on different cross sections of firm data. The data used there were anonymized by micro aggregation. I replicated the analyses reported in the papers with the anonymized data, and then compared the results to those produced with the original data. Frequently, the reported levels of statistical significance differ. Furthermore, statistically significant coefficients sometimes differ by order of mangitude. Therefore, at least for the moderate sample sizes used here micor-aggregated firm data should not be considered as a tool for empirical research.