Title: Testing Instrument Validity with Covariates

Abstract: We develop a novel specification test of the instrumental variable identifying assumptions (instrument validity) for heterogeneous treatment effect models with conditioning covariates. Building on the common empirical settings of local average treatment effect and marginal treatment effect analysis, we assume semiparametric dependence between the potential outcomes and conditioning covariates, and show that this allows us to express the testable implications of instrument validity in terms of equality and inequality restrictions among the subdensities of estimable partial residuals. We propose jointly testing these restrictions. To improve power of the test, we propose distillation, a process designed to reduce the sample down to the information useful for detecting violations of the instrument validity inequalities. We perform Monte Carlo exercises to demonstrate the gain in power from testing restrictions jointly and distillation. We apply our test procedure to the college proximity instrument of Card1 (1993), the same-sex instrument of Angrist and Evans (1998), the school leaving age instrument of Oreopoulos (2006), and the mean land gradient instrument of Dinkelman (2011). We find that the null of instrument validity conditional on covariates cannot be rejected for Card (1993) and Dinkelman (2011), but it can be rejected at the 10% level of significance for Angrist and Evans (1998) for some levels of a tuning parameter, and it is rejected at all conventional levels of significance in the case of Oreopoulos (2006).