利用領先指標預測美國景氣循環之衰退－以二元邏輯斯模型分析

Using Leading Indicators to Forecast Recession in the US Business Cycle—A Binary Logit Analysis

吳思緩(Sih-Huan Wu)

指導教授 ： 柯大衛

淡江大學/國際研究學院/美洲研究所碩士班/碩士(2014年)

https://doi.org/10.6846/TKU.2014.00800

摘要

為了解美國總體經濟指標跟各州領先指數之關聯性，本文使用二元邏輯斯模型檢測美國國家經濟研究局所公布之景氣循環指數，再針對美國50州的領先指標比較，選擇研究數據時間為西元1982年至2013年。根據Stock and Watson的論點確定變數組合，使用動態因子模型和主要元成份分析建立模型，產生同期和領先指標代表目前經濟所呈現的狀態。
實證結果發現肯塔基州的領先指數無論是使用兩個月或三個月的落後期，對於預測美國總體經濟指標皆有顯著的效果。在比較肯塔基州與美國總體經濟指標時，顯示肯塔基州的預測能力較高。由於美國國家經濟研究局的數據以及國家和州的領先指標對外公布的時間較久，因此無法肯定落後期為一個月時何者的預測能力較優，此外，本文也探討了使用時間頻率為每季或每月兩者的不同之處。最後歸納出影響美國全國經濟的因素在於某些州的經濟活動與全國經濟活動關聯性較大時，較能夠準確預測出未來景氣循環的走向。

關鍵字

美國景氣循環 ； 領先指標 ； 衰退 ； 二元邏輯斯模型

並列摘要

In this paper, we seek to study the relation between data on leading indicators and U.S. business cycle by using binary logit model to examine business cycle index of the National Bureau of Economic Research and compare to the leading indicators of the 50 states of the United States. from 1982-2013. According to Stock and Watson’s theory to determine the variable combination. The use of dynamic factor models and principal components to model the current state of the economy has led to a series of coincident and leading indicators. These indicators are specifically designed to help predict the future movements in the state of the economy.

The empirical result shows Kentucky’s leading index using both of the two or three month lags, the predict ability is significant effect. The conclusion that can be drawn is that two and three months prior, predictions made using the Kentucky leading index will generally be more reliable a predictor for NBER recessions than those using the national leading indicator. One month ahead forecasts are not likely to be very reliable since the NBER number is not available within one month and the state and national leading indicator will not also generally be available so early.

Finally, this paper also discusses the use of time for quarterly or monthly frequency difference between the two. When the state’s economic activity is similar to the country, it can forecast the future direction of the business cycle more accurately.

並列關鍵字

U.S business cycle ； leading indicator ； recession ； binary logit model

參考文獻

陳仕偉、沈中華(2003)金融領先指標與實質領先指標訊息一致嗎？台灣領先指標的實證分析，《人文及社會科學集刊》vol.15:4，pp627-660

黃聿銘。「台灣領先指標對景氣循環預測能力的探討」。中央大學經濟研究所，碩士論文，民國97年。

林大超。「台灣與美國兩地景氣循環指標關聯性之研究」。成功大學企管研究所，碩士論文，民國92年。

Burns, A.F. and Mitchell, W.C. (1946), “Measuring Business Cycles.” New York: NationalBureau of Economic Research.

Barhoumi, K., Darné, O. and Ferrara, L. (2013), “Testing the Number of Factors: An Empirical Assessment for a Forecasting Purpose.” Oxford Bulletin of Economics and Statistics, 75: 64–79.