• Credible robust inference;
  • Partial identification;
  • (Un)conditional (limiting) distributions;
  • (Un)restrained repeated sampling;
  • Weak instruments

Summary  In simple static linear simultaneous equation models, the empirical distributions of IV and OLS are examined under alternative sampling schemes and compared with their first-order asymptotic approximations. We demonstrate that the limiting distribution of consistent IV is not affected by conditioning on exogenous regressors, whereas that of inconsistent OLS is. The OLS asymptotic and simulated actual variances are shown to diminish by extending the set of exogenous variables kept fixed in sampling, whereas such an extension disrupts the distribution of IV and deteriorates the accuracy of its standard asymptotic approximation, not only when instruments are weak. Against this background, the consequences for the identification of parameters of interest are examined for a setting in which (in practice often incredible) assumptions regarding the zero correlation between instruments and disturbances are replaced by (generally more credible) interval assumptions on the correlation between endogenous regressor and disturbance. This yields OLS-based modified confidence intervals, which are usually conservative, as is established by simulation. Often they compare favourably with IV-based intervals and accentuate their frailty. The latter is demonstrated in an empirical illustration.