Bayesian Generalized Least Squares with Autocorrelated Error

Autocorrelation plays significant role in both time series and cross sectional data. More often than none it rendered the inference of parameter estimates invalid those other statistics that use parameters. This study investigates asymptotic behaviour generalised least squares with Autocorrelated errors cum Marcov Chain Monte Carlo simulation. Bias Mean Squares Error criteria were used to evaluate finite properties estimator. The following sample sizes: 25, 50,100, 250 constructed used. Thus 11,000 simulations varying level error carried out. is subjected convergence. Minimum revealed improving performance asymptotically regardless error. observed consistency efficiency obtained at large which obey law number point fact variance terms tend towards zero distribution tends normal when applied. therefore recommended samples should be obtain make inferences stable