Bayesian and other procedures are developed and applied to forecast annual output growth rates for eighteen countries year by year for the period 1974-84. This work extends earlier work relating to nine countries' data, 1974-81. The new calculations indicate that previous methods work well in forecasting using the new, extended data base. Comparisons of the performance of various Bayesian shrinkage procedures, other procedures and OECD forecasts are provided. It is found that the autoregressive-leading indicator models used in past work continued to perform well in the present study based on an enlarged data base.