Turn, turn, turn: Predicting turning points in economic activity
评估了亚特兰大联储贝叶斯向量自回归模型预测经济转折点的能力，并与先行经济指标指数及专门设计的转折点模型比较，发现后者预警信号质量更优。
Policy and investment decisions are made with an eye toward future economic conditions, and an econometric model that can correctly forecast directional changes in the business cycle would be a boon to policymakers, the business community, and the general public. This article provides some evidence on econometric models' ability to predict these directional changes, also known as turning points, in an effort to answer the question, How good is the state of the art in turning point forecasting? ; The author first discusses the definition of turning points and describes different approaches to turning point forecasting, along with their relative advantages and disadvantages. Next, the article assesses the performance of the Atlanta Fed Bayesian vector autoregression (BVAR) model in terms of forecasting turning points relative to a well-known alternative, the Leading Economic Indicators (LEI) Index. The author concludes that the BVAR model forecasts contain information on future recessions that appears superior to that embodied in the LEI Index, at least when simple rules of thumb are used to extract information from the index. ; Relative to a turning point model proposed by Arturo Estrella and Frederic Mishkin, however, the Atlanta Fed BVAR model is far less precise in indicating the exact timing of a recession. In general, the warning signals from models that are specifically designed to forecast turning points appear to be of better quality than those from econometric models like the BVAR model, suggesting that it is worthwhile to supplement the BVAR with a turning point model.