B0790
Title: DiD with as few as two cross-sectional units: An application to the impact of democracy on growth
Authors: Gilles Koumou - Universite Laval (Canada)
Emmanuel Tsyawo - Universite Mohammed VI Polytechnique (Morocco) [presenting]
Abstract: The effect of democracy on economic growth remains a largely unresolved issue, as empirical findings tend to be varied because of substantial heterogeneity. The objective is to take a different approach that targets the impact on as few as one country (Benin) using as few as one control (Togo). Under weak identification and sampling assumptions, a consistent and asymptotically normal difference-in-differences (DiD) estimator is developed that exploits time variation in order to estimate an average treatment effect on the treated under fixed $N$, large $T$ asymptotics. Benin's experience of democracy since 1991 accounts for a 4.4\% yearly increase in GDP on average.
