DP12980 Instrumental Variables in the Long Run
We study the interpretation of instrumental variable (IV) regressions that use historical or geographical instruments for contemporary endogenous regressors. We find that conventional IV regressions generally cannot estimate the long-run causal effect of an endogenous explanatory variable when there is a time gap between the instrument and the endogenous variable. We develop a model that can overcome this problem and apply our results to important topics in the field of economic growth, including the effect of institutions on economic growth. We find effects that are smaller than those estimated in the existing literature, demonstrating the quantitative importance of our study.