Asymptotic Distribution of JIVE in a Heteroskedastic IV Regression with Many Instruments

This paper derives the limiting distributions of alternative jackknife IV (JIV) estimators and gives formulae for accompanying consistent standard errors in the presence of heteroskedasticity and many instruments. The asymptotic framework includes the many instrument sequence of Bekker (1994) and the many weak instrument sequence of Chao and Swanson (2005). We show that JIV estimators are asymptotically normal and that standard errors are consistent provided that \frac{\sqrt{K_{n}}}{n} \to \infty as n \to \infty, where K_n and r_n denote, respectively, the number of instruments and the concentration parameter. This is in contrast to the asymptotic behavior of such classical IV estimators as LIML, B2SLS, and 2SLS, all of which are inconsistent in the presence of heteroskedasticity, unless \frac{K_n}{r_n}\to 0. We also show that the rate of convergence and the form of the asymptotic covariance matrix of the JIV estimators will in general depend on the strength of the instruments as measured by the relative orders of magnitude of r_n and K_n.