A0985
Title: On sandwich variance estimation: Bayesian versus frequentist
Authors: Cy Sin - National Tsing Hua University (Taiwan) [presenting]
Abstract: It is well known the Eicker-Huber-White variances are not only heteroskedasticity-robust and nonlinearity-robust but also nonnormality-robust. Recently, some of the Eicker-Huber-White variances have been reviewed. Among other things, it is concluded: (a) Simulation studies suggest HC(4), a variant of robust variance estimator proposed in another study, does not over-reject or mildly under-rejects even in cases of non-normal distributions; (b) The original robust variance (denoted by HC(0)) and its variants considered by the prevalent statistical software (such as R and STATA), are all asymptotically equivalent. The focus is on a Bayesian approach, considering the balanced loss function (BLF) proposed in a recent study. Unlike the conventional inference loss function (ILF), this function balances estimation error and lack of fit. This function is, in turn, generalized upon the first proposed in another study, where the attention to normality-type likelihoods is confined. The Bayesian estimator of the variance-covariance matrix is asymptotically equivalent to the frequentist estimator. Non-normal likelihoods are covered. Simulation studies that compare the Bayesian estimator with the conventional estimators are performed.
