Abstract:
Is there misallocation in firm production, as we might expect to result from market power, contracting constraints, taxes, regulations, corruption, or other potential distortions? And if there is misallocation, how severe are its resulting welfare consequences? In this paper we propose a new test for misallocation that is nonparametric in the sense that it does not restrict any firm’s production technology, demand, market structure, or optimizing behavior. We also develop a new procedure to quantify losses from misallocation via a nonparametric instrumental variable random coefficient model. Our procedures exploit exogenous shocks that induce some firms to alter their input use and then measure the average level of, and cross-firm dispersion in, the rate at which firms’ output value increases, on the margin, for a given increase in inputs. We apply these results to a setting in which thousands of firms experience exogenous demand shocks due to a lottery-based assignment of public procurement contracts for construction services in Ecuador. Using monthly data on firm-to-firm transactions and employer-employee payments, a randomization inference version of our test rejects (at standard levels) the null of overall allocative efficiency (AE) but the costs of this misallocation appears to reduce aggregate output among this set of firms by only 1% relative to the first-best. This derives roughly half from an insufficient aggregate use of inputs and half from cross-firm dispersion in the marginal products of the inputs that are used. Standard parametric assumptions applied to the same setting would suggest losses that are many times larger.