利用動態Probit模型預測台灣景氣衰退

Forecasting Taiwan's Recessions with Dynamic Probit Model

黃建銘(Huang, Jianming)

指導教授 ： 李偉銘

國立中正大學/管理學院/經濟系國際經濟研究所/碩士(2012年)

摘要

台灣有關景氣循環的研究主要是利用馬可夫轉換模型來定義台灣的景氣循環轉折點，而較少著重在預測景氣衰退的部分。因此，我們採用動態Probit模型來預測台灣之景氣衰退。由於過去許多實證研究發現期間利差為較佳的預測變數，因此本論文亦以期間利差為預測變數。另外，由於能源價格一直以來受大家許多的關注，所以本篇論文也探討能源價格對景氣退的預測能力。我們發現在實證結果中，相對於靜態模型而言，動態模型不論在樣本內或樣本外預測皆表現較佳，但在樣本外的結果中，動態模型會隨著預測期數的增加而逐漸失去預測能力。另外在預測變數的表現上，
能源價格之波動表現較佳。

關鍵字

動態Probit模型 ； 景氣衰退 ； 期間利差 ； 能源價格

並列摘要

Many researches of Taiwan's business cycles primarily focus on identifying the turning points of business cycles with Markov-switching models. The purpose of this thesis is to investigate the predictability of Taiwan's recession. We adopt the dynamic Probit model and use the term spread as a predictive indicator to forecast Taiwan's recessions because many papers have found the superior predictive power of the term spread. On the other hand, we also consider the energy price as a predictor. Our findings suggest that compared to static models, dynamic models provide more reliable predictions in in-sample and out-sample results. However. dynamic models become inaccurate gradually with the increase of the forecast horizon in out-of-sample results. From the performances of predictive variables, the volatility of energy prices is a better predictor.

並列關鍵字

dynamic Probit model ； recession ； term spread ； energy prices

參考文獻

Chen, H.-C., B.-S. Kuo, and S.-J. Lin (2012). Energy price shocks and Taiwan's

Chen, S-W. and J.-L. Lin (2000). Identifying turning points and business cycles in Taiwan: a multivariate Markov-Switching factor model approach, Academia Economic Papers, 28, 289-320.

Burbidge, J. and A. Harrison (1984). Testing for the eects of oil-price rises using vector autoregressions, International Economic Review, 25, 459-484.

varying Markov-Switching models, Academia Economic Papers, 28, 17-42.

Diebold, F. X. and G. D. Rudebusch (1989). Scoring the leading indicators, Journal of Business, 62, 369-391.