Abstract:
When it comes to the inequality-growth relationship, the empirical literature offers contradictory assessments: Estimators based on time-series variation only (i.e., differences-based estimators) indicate a strong positive link while estimators also exploiting the cross-sectional variation (i.e., level-based estimators) suggest a negative relationship. Taking advantage of a new dataset, the present paper confirms this conflicting pattern - but also offers an explanation on the basis of a simple model: The seemingly inconsistent empirical results just mirror different aspects of reality. More specifically, we argue that the differences-based methods are prone to reflect the (mostly) positive short- or medium-run implications of inequality while the level-based estimators also incorporate the negative consequences - which require more time to materialize. Thus, the latter estimates come close to reflect the adverse overall impact of inequality in the long run.