Improving the accuracy of outlook price forecasts
研究了爱荷华州立大学发布的生猪展望价格预测，与时间序列模型和期货市场预测比较，发现组合多种预测方法能显著降低误差。
This study investigates the predictive ability of outlook hog price forecasts released by Iowa State University relative to alternative time-series and market forecasts. Under root mean squared error (RMSE), the futures market forecast is most accurate at the first and second horizon but less accurate than Iowa outlook and the other forecast methods at the third horizon. In terms of the individual time-series models, some vector autoregressions (VARs) and Bayesian VARs flexible in specification and estimation and model averaging tend to perform better than Iowa outlook forecasts. Evidence from encompassing tests, more stringent tests of forecast performance, indicates that many price forecasts can add incremental information to the Iowa forecast. Simple combinations of these models and outlook forecasts are able to reduce forecast errors by economically significant levels. Overall, the results indicate that it is possible to provide more accurate forecasts than Iowa outlook at every horizon.