GMM and misspecification correction for misspecified models with diverging number of parameters (Q2300520)

This paper considers statistical estimation by the generalized method of moments (GMM) when the model is misspecified. Focusing on the situation where the numbers of parameters and predictors tend to infinity as the sample size grows, the authors prove local and global consistency to and asymptotic normality around the solution to the estimation equation, what they call the pseudoparameter vector. Then the authors propose a semiparametric method to correct the misspecification and show its asymptotic properties. Although the numerical experiments with a simple regression model demonstrate the effectiveness of the correction method, they are a little too limited to verify the asymptotic properties.