Robust Small Sample Accurate Inference in Moment Condition Models

Procedures based on the Generalized Method of Moments (GMM) (Hansen, 1982) are basic tools in modern econometrics to estimate the parameters and make inference in moment condition models. In general, the inferential tools (p-values and confidence intervals) are based on first order asymptotic theory which is not very accurate in moderate to small samples. Moreover, in the presence of small deviations from the assumed model, p-values and confidence intervals based on classical GMM procedures can be drastically affected (non-robustness). Several alternative techniques have been proposed in the literature to either improve the accuracy of GMM approximation (information and entropy econometrics (IEE)) or address the lack of robustness (Robust GMM estimators and tests).
In this work, we propose a new alternative procedure which combines both robustness properties and accuracy in small samples. Specifically, we combine the IEE techniques of Imbens, Spady, Johnson (1998) with the robust approach of Ronchetti and Trojani (2001). This leads to a new robust estimator and overidentification test in moment condition models with excellent finite sample accuracy. Finally, we illustrate the performance of the new procedure by simulations for Chi-squared moments and Stochastic lognormal volatility models.

Where: Red Centre 4082
When: 4pm Friday 31st August, 2007.
Please join us for wine and cheese after the seminar!