Based on large-sample stochastic approximations, we obtain asymptotic bias and mean squared error for the k-class estimators of an identified but misspecified equation with an arbitrary number of endogenous variables in linear simultaneous-equations model
