We examine empirically the generalizability of internally valid micro estimates of causal effects in a fixed population over time when that population is subject to aggregate shocks. Using panel data we show that the returns to investments in agriculture in India and Ghana, small and medium...