Tolerance and growth: modeling the empirical relationship

摘要

We showed, in Berggren and Elinder (2012), that tolerance toward homosexuals is negatively and quite robustly related to economic growth. In a comment, Bornhoff and Lee (this issue) question this finding on model-specification grounds. By undertaking three changes, they purport to show that our main result does not hold. In this article, we demonstrate that one of these changes is inconsequential (replacing GDP per capita by its logarithm in controlling for conditional convergence) and argue that two of them are questionable. First, the removal of certain central control variable risks introducing omitted variable bias and inconsistent estimates. Second, regional dummy variables are added on arbitrary grounds. For example, by using regional dummy variables that are just as reasonable as the Baltic dummy used by Bornhoff and Lee, we find that significance for tolerance toward homosexuals reappears in our empirical model. In all, this implies that there are good grounds for considering the negative relationship between tolerance towards homosexuals and growth valid, Bornhoff and Lee's claims notwithstanding.