A Bayesian inference of autoregressive panel data model was applied to real data of Nelore sires Expected Progenie Difference (EPD) during a five-year period (2000-2005). The exact likelihood function and predictive distributions of future observations were considered. The results indicated the importance of sires grouping in homogeneous groups according to accuracy and showed forecast efficiency for EPD values in a future year around 80%.
Nelore cattle; time serie; MCMC method; EPD