Developing New Business Cycle Composite Indicators Using Wavelet Analysis

作为本研究项目的一部分，我们对日本官方领先指标综合指数(CLI)及其组成序列的特征进行了评估。更具体地说，我们将Baxter-King过滤器分别应用于CLI及其成分系列，并提取了每个序列的商业周期部分，其频率在32个月到128个月之间。利用过滤后的序列确定了它们的业务转折点，并将它们与日本政府官方确定的业务转折点日期进行了比较。实证研究的部分结果是：(A)商业周期指标中的景气周期部分平均领先官方业务转折点9个月。(B)指标的整体表现与最终需求商品的库存指数相似。(C)大多数指标对于1991.2的转折点的领先月份太长。我们还指出，如果使用成分序列的评估属性，我们可以构建一个具有所要求的某些理想属性的综合领先指标。作为说明，我们提供了这样一个综合领先指标，它将官方业务转折点提前了大约4个月，而且样本方差比日本官方CLI的样本方差小。

As a part of this research project, we evaluated the characteristics of the Japanese official composite index of the leading indicators (CLI) and its component series. More concretely, we applied the Baxter-King filter to the CLI and its component series individually, and extracted the business cycle part of each series of which frequency lies between 32 months and 128 months. Making use of the filtered series, we determined business turning points of them, and compared them with the business turning point dates officially determined by the Japanese government. Parts of our findings from the empirical investigations are:(a) The business cycle part in CLI leads the official business turning points by 9 months on average.(b) The overall performance of CLI is similar to that of Inventory Indexes of Final Demand Goods.(c) The leading months of most indicators for the turning point at 1991.2 are too long.We also pointed out that if we use the evaluated properties of component series, we may construct a composite leading indicator which has some desirable properties as requested. As an illustration we provided such a composite leading indicator, which leads the official business turning points roughly by 4 months and has the smaller sample variance than that of the Japanese official CLI.