Abstract
The empirical literature on the relationship between inequality and growth offers a contradictory assessment: Estimators based on time-series variation indicate a positive link while estimators (also) exploiting the cross-sectional variation suggest a negative relationship. The present paper (i) confirms this conflicting pattern in an expanded dataset; (ii) proposes a simple theoretical framework to highlight the biases associated with the different techniques. We argue that mechanisms generating a positive inequality-growth relationship work mainly in the short-run and are reflected in difference-based estimators. In contrast, mechanisms generating a negative relationship work over the longer term and are reflected in level-based estimators.