This paper studies the robustness of estimated policy effects to changes in the distribution of covariates, a key exercise to evaluate the external validity of (quasi)-experimental results. I propose a novel scalar robustness metric. It measures the magnitude of the smallest covariate shift needed to invalidate a claim on the policy effect (say, ATE greater than 0) supported by the (quasi)-experimental evidence. I estimate my robustness metric using de-biased GMM, which guarantees a parametric convergence rate while allowing for machine learning-based estimators of policy effect heterogeneity (including LASSO, random forest, boosting, neural nets). I apply my procedure to study the robustness of policy effects’ estimates for health-care utilization and financial strain outcomes in the Oregon Health Insurance experiment. I find that, among all outcomes, the effect of the insurance policy on outpatient visits is most robust to shifts in the distribution of context-specific covariates.

This paper studies the implication of a fraction of the population not responding to the instrument when selecting into treatment. We show that, in general, the presence of non-responders biases the Marginal Treatment Effect (MTE) curve and many of its functionals. We obtain partial identification results for the MTE, the LATE and the MPRTE. Moreover, in a linear-in-covariates model we obtain a testable implication for the presence of non-responders. Simulation results show that our proposed estimators, using the range of the propensity score, work well.

This project focuses on the exercise of comparing coefficients across regression specifications, broadly known as robustness checks, a very popular tool in applied economic research. The econometric literature has proposed to formalize this coefficient stability exercise in linear models by turning robustness checks into a robustness test. Unfortunately, for available procedures, a rejection in their proposed test may be due to either a failure in the notion of robustness or to a failure of the linearity of the model, making the robustness test non-specific. I bypass this difficulty by proposing a semi-parametric test based on linear sieve estimators that allows for non-linearity and is specific to testing the coefficient stability.

The classic Arrow-Debreu (1954) general equilibrium model cannot sustain or account for the existence of money. This lacuna arises because each household and firm faces a single budget constraint summarizing revenue and expense in all commodities. Money, a carrier of value between transactions, has no function when all credits and debits are rolled into a single expression. A trading post model of $N \geq 3$ commodities and transaction costs generates \(\frac{1}{2} N (N-1) \) separate budget constraints with distinct bid and ask prices. General equilibrium, market-clearing prices and transactions at each trading post, exists under conventional continuity and convexity conditions. Commodities acquired by an agent at one trading post and disbursed at another constitute commodity money.

A thick market external effect is applied to a trading post model of \( N\geq 3 \) commodities with transaction costs and distinct bid and ask prices. We state and prove an existence theorem for general equilibrium with external effects in the trading post model. Media of exchange occur endogenously as liquid commodities, characterized by a narrow bid/ask price spread. The thick market externality can lead to concentration of the endogenously determined media of exchange towards an equilibrium with a single medium. In a class of examples, we show that if the households have sufficiently heterogeneous tastes relative to the size of the economy, the monetary equilibrium leads to higher consumption than the barter equilibrium.