This repository contains Matlab code for implementing the tests proposed in Andrews, Roth, and Pakes (ARP), "Inference for Linear Conditional Moment Inequalities". The paper proposes tests for moment inequalities of the form E[Y_i(beta) - X_i delta | Z_i] <=0, where beta is the target parameter, delta is a nuisance parameter, and X_i is non-random conditional on Z_i.
The package contains the following main functions:
- hybrid_test_fn implements the hybrid test, our preferred method
- conditional_test_fn implements the conditional test
- lf_critical_value_fn calculates the least-favorable critical value
- etahat_fn calculates the test statistic for the LF test (the same test stat is used for the conditional/hybrid tests, but done internally)
- lf_ci_for_linear_params calculates an LF CI for the case where the target parameter also enters the moments linearly.
- conditional_variance_fn estimates the conditional variance using the method of Abadie, Imbens, and Zheng (2014)
The file LinearMomentInequalities_Examples.m contains examples for how to use all of these functions. It also illustrates some computational shortcuts that can be taken when the target parameter also enters the moments linearly.