We partnered with a micro‐lender in Mali to randomize credit offers at the village level. Then, in no‐loan control villages, we gave cash grants to randomly selected households. These grants led to higher agricultural investments and profits, thus showing that liquidity constraints bind with respect to agricultural investment. In loan‐villages, we gave grants to a random subset of farmers who (endogenously) did not borrow. These farmers have lower – in fact zero – marginal returns to the grants. Thus we find important heterogeneity in returns to investment and strong evidence that farmers with higher marginal returns to investment self‐select into lending programs.