This paper gives a new jackknife estimator for instrumental variable inference with unknown
heteroskedasticity. The estimator is derived by using a method of moments approach
similar to the one that produces LIML in case of homoskedasticity. The estimator
is symmetric in the endogenous variables including the dependent variable. Many instruments
and many weak instruments asymptotic distributions are derived using high-level
assumptions that allow for the simultaneous presence of weak and strong instruments for
different explanatory variables. Standard errors are formulated compactly. We review
briefly known estimators and show in particular that the symmetric jackknife estimator
performs well when compared to the HLIM and HFUL estimators of Hausman et al.
(2011) in Monte Carlo experiments.