Ingrid VAN KEILEGOM (KU LEUVEN) – “Tests of exogeneity in duration models with censored data”
December 8 @ 4:00 pm - 5:15 pm
PSE Seminar :
Time: 16:00 pm – 17:15 pm
Date: 8th of december
Room : 3001
Ingrid VAN KEILEGOM (KU LEUVEN) – “Tests of exogeneity in duration models with censored data”
Abstract :
| “Consider the setting in which a researcher is interested in the causal effect of a treatment $Z$ on a duration time $T$, which is subject to right censoring. Given a vector of baseline covariates $X$, we would like to test whether the treatment is endogenous, meaning that $Z$ is not independent of the error term in the structural model for $T$, given $X$. In this paper, we propose nonparametric tests for this problem. The test statistics rely on the presence of an instrumental variable $W$ that is independent of the error term in the structural model for $T$ conditional on $X$. We assume that $X,W$ and $Z$ are all categorical and that $T=varphi(X,Z,U)$, where $varphi(X,Z,U)$ is strictly increasing in the error term $U$ for each $(X,Z)$ and $Usim mathcal{U}[0,1]$. Therefore, the model is nonparametric and nonseparable. We construct test statistics for the hypothesis that the conditional rank $V_T= F_{T mid X,Z}(T mid X,Z)$ is independent of $(X,W)$ jointly. Under an identifiability condition on $varphi$, this hypothesis is equivalent to $Z$ being independent of $U$ given $X$, meaning that $Z$ is exogenous. However, note that $T$ being censored by $C$ implies that $V_T$ is censored by $V_C =F_{T mid X,Z}(C mid X,Z)$, which complicates the construction of the test statistics significantly. We derive the asymptotic properties of the proposed tests, which is challenging due to the presence of right censoring both in the data and in the estimated conditional ranks. Moreover, we prove the interesting result that our estimator of the distribution of $V_T$ converges to the uniform distribution at a rate faster than the usual parametric $n^{-1/2}$-rate. We also demonstrate that the test statistics and bootstrap approximations for the critical values have a good finite sample performance in various Monte Carlo settings. Finally, we illustrate the tests with an empirical application to the National Job Training Partnership Act (JTPA) Study”.
Joint work : Gilles Crommen and Jean-Pierre Florens |
Organizer :
Laurent DAVEZIES (Pôle économie du CREST)
Sponsors:
CREST