| Abstract: | The impact of international trade, or “openness,” on economic growth is difficult to quantify because of reverse causality. In this article, I use recent advances in gravity equation estimation to generate a geography‐based instrument for openness à la Frankel and Romer (1999). In contrast with the benchmark, the new instrument is constructed using consistent and unbiased estimates of the impact of geography on bilateral trade. As a result, the instrument provides stronger identification of the impact of trade on income and increases the efficiency of the two‐stage least square estimation. An important advantage of the corrected procedure over the benchmark is that the estimated effect of trade on income remains large, positive, and statistically significant even after controlling for regional indicators and endogenous institutional quality. |
