LAUSR

Abstract We revisit the estimation algorithm of Pettenuzzo and Timmermann (2011) and show how to apply the posterior simulation test of Geweke (2004) to locate and correct an error in the original posterior sampling algorithm. The main modification for the new algorithm is the introduction of a Metropolis–Hasting step to draw the precision parameters of the return and predictor equations, taking into account the correlation between the error terms of the two equations. We find that the modification of the original algorithm has very minor effects on the empirical results reported in Pettenuzzo and Timmermann (2011). In particular, for both the dividend yield and Treasury bill predictors the updated algorithm finds that the models best supported by the data, as measured by the SIC, match those reported in the original paper.