Forecasting with an econometric model: Some recent results using the national institute model
通过复制国家研究所的近期预测，证明大型计量经济模型无需人为干预即可产生与官方发布相近的预测结果，且机械模型预测在事后比较中不劣于发布版本。
Abstract It is widely believed that the large econometric models cannot be used for forecasting without considerable intervention on the part of the forecaster. In this paper we challenge this view by reproducing a number of recent forecasts published by the National Institute but without the ad hoc interventions used at the time. We show that in no case would the forecast, produced by the model used mechanically, have been radically different from that actually published. Further, in an ex‐post comparison against actual out‐turns, the mechanical model forecast is not obviously dominated by the published version.