SEASONALITY, FORECAST EXTENSIONS AND BUSINESS CYCLE UNCERTAINTY
研究了季节性调整对商业周期测量的影响，发现其贡献在序列转折点和样本端点较大，尤其对高通滤波器影响显著，并讨论了预测扩展的作用。
Abstract Seasonality is one of the most important features of economic time series. The possibility to abstract from seasonality for the assessment of economic conditions is a widely debated issue. In this paper we propose a strategy for assessing the role of seasonal adjustment (SA) on business cycle measurement. In particular, we provide a method for quantifying the contribution to the unreliability of the estimated cycles extracted by popular filters, such as Baxter and King and Hodrick–Prescott (HP). The main conclusion is that the contribution is larger around the turning points of the series and at the extremes of the sample period; moreover, it much more sizeable for highpass filters, like the HP filter, which retain to a great extent the high‐frequency fluctuations in a time series, the latter being the ones that are more affected by SA. If a bandpass component is considered, the effect has reduced size. Finally, we discuss the role of forecast extensions and the prediction of the cycle. For the time series of industrial production considered in the illustration, it is not possible to provide a reliable estimate of the cycle at the end of the sample.