摘要：The two-stage least-squares (2SLS) instrumental variables estimator is commonly used to address endogeneity. However, the estimator suffers from bias that is exacerbated when the instruments are only weakly correlated with the endogenous variables and when many instruments are used. In this article, I discuss jackknife instrumental variables estimation as an alternative to 2SLS. Monte Carlo simulations comparing the jackknife instrument variables estimators to 2SLS and limited information maximum likelihood (LIML) show that two of the four variants perform remarkably well even when 2SLS does not. In a weak-instrument experiment, the two best performing jackknife estimators also outperform LIML.
