領先指標預測能力之研究

A Study on the Predictive Ability of Leading Indicators

領先指標可作為預先衡量景氣變化的工具，鑑於該指標領先性與所選構成項目組成之間有高度關連性，且其良好特性可能隨著經濟結構變遷而減弱，因而必須定期檢討及改進。實證結果顯示：1.現行項目中，工業及服務業加班工時自2000年以後逐漸喪失領先性，並趨向同時。製造業存貨量指數（倒數）之領先性穩定，但其本質趨向落後。而核發建照面積的循環對應性較不穩定。2.建議加入服務業廠商對景氣的預期、讓製造業存貨量指數回歸其本質，將領先指標構成項目修訂為外銷訂單指數、實質貨幣總計數M1B、股價指數、SEMI半導體接單出貨比、核發建照面積、工業及服務業淨進入率、服務業營業氣候測驗點。3.新領先指標（不含趨勢）與現行領先指標（不含趨勢）的預測能力相當，但新領先指標構成項目反映經濟結構改變。

While the leading character and high level of connectivity between its components make the CEPD’s Composite Leading Index (CLI) a good tool for pre-measuring changes in the economic cycle, its predictive ability may be weakened by changes in the economic structure. Hence, it is necessary to regularly review and improve it. The empirical results of this study are as follows: 1. Among the current components of the CLI, the average monthly overtime in industry and services has been gradually losing its predictive character since 2000, tending toward concurrent rather than leading indicativity. The index of producer's inventory (inverted) has retained stability in its leading character, but is essentially of a lagging nature. Building permit issuance has been relatively unstable in its cyclical correspondence. 2. We suggest increasing the service sector element of economic forecasting, and letting the index of producer's inventory return to its essential nature, adjusting the components of the CLI to (1) the index of export orders, (2) real monetary aggregates M1B, (3) the stock price index, (4) the SEMI book-to-bill ratio, (5) building permits, (6) the net accession rate of employees in industry and services, and (7) the service sector composite indicator. 3. The new trend-adjusted CLI’s predictive ability is similar to the current trend-adjusted CLI’s, but its components can reflect the change in the economic structure.