台灣經濟預測引用國際數據之檢視－以WEFA、IMF與OECD為例

On the Use of Inputting WEFA, IMF and OECD Forecasts into Taiwan's Forecasting Models

梁國源(Kuo-Yuan Liang) ； 周大森(Ta-Sheng Chou)

《臺灣經濟預測與政策》 33卷1期 (2003/03) Pp. 41-74

https://doi.org/10.29629/TEFP.200303.0004

摘要

多年來WEFA、IMF與OECD發布的經濟預測數據，不但為媒體所關注，也常為台灣各預測機構所引用。不過在關注與引用的過程中，有一奇特的現象是，人們在使用該等數據時，往往照單全收，鮮少人會去深思其中的預測品質，與值不值得被引用等重要的基本課題。
基於此，本文旨在以WEFA為主，另輔以IMF與OECD等機構，對美、日與中國大陸，發布的實質GDP成長率預測值為對象，分別由不同觀點，研析各自的預測精確度，與預測誤差的性質；並透過Ashley(1983,1985,1988)提出的預測有用性指標，檢視使用者是否認為該等機構提供的預測資訊為有意義，且能對其產生有益的作用。作者期望透過本文的實證結果，能對台灣各預測機構及媒體在引用該等數據時，增進若干評判的能力。
本文的實證結果大致上支持以下五點結論。第一，不論WEFA的季預測或三個機構的年預測，RMSE、MAE與Theil U值，隨預測時距加長而增加；對美、日的當年預測，尚且呈現該三種誤差值按各機構的預測發布先後順序遞減。第二，WEFA的季預測在方向上的精確度，比三個機構的年預測，表現較佳。第三，檢視最適預測、不偏性、及效率性等預測誤差性質時，顯示三個機構的年預測明顯地比WEFA季預測的情況較佳。第四，Ashley的預測有用性指標分析，顯示年預測明顯的優於季預測，其中尤以當年預測的表現最優；另，該指標值還呈現隨預測時距拉長而越大於1的現象，即越遠離預測有用性的值域。第五，綜合研判，最能讓使用者放心引用的預測，為各機構發布的當年預測。

並列摘要

Over the last decade, GDP forecasts for the U.S.A., Japan, and Mainland China produced by Wharton Econometric Forecasting Associates (WEFA), International Monetary Fund (IMF), and Organization for Economic Cooperation and Development (OECD) have been used as inputs to Taiwan's forecasting models. Strangely, although directly substituting such projections into a Taiwanese forecasting model as exogenous inputs is considered a natural practice, the viability of such a procedure has seldom been deeply explored. Thus, this paper strives to analyze the accuracy and properties, as well as the relative worthiness (in the Ashley sense) of these GDP forecasts. Basically, samples for the quarterly forecasts produced by WEFA range from 1993 Q2 to 2000 Q4. Meanwhile, the samples for the annual projections released by WEFA, IMF and OECD are between 1990 and 2000. The empirical results support the following conclusions: (1) the RMSE, MAE and Theil U statistics reveal that the accuracy of the quarterly and annual forecasts generally increases with a decline in the forecast horizon; (2) the directional accuracy statistics show that quarterly forecasts are better than the annual projections; (3) the optimality, unbiasedness and efficiency tests support that the annual projections released are superior to quarterly forecasts; (4) Ashley's usefulness statistics indicate that annual forecasts are better than quarterly forecasts; and (5) overall, the current-year forecasts prepared in the middle of the same year perform the best.

參考文獻

Ash, J. C. K., Smith, D. J., Heravi, S. M.(1991).The Accuracy of OECD Forecasts of the International Economy: Demand, Output and Price.International Journal of Forecasting.6(3)

Ashley, R.(1983).On the Usefulness of Macroeconomic Forecasts as Inputs to Forecasting Models.Journal of Forecasting.2(3)

Ashley, R.(1985).On the Optimal Use of Suboptimal Forecasts of Explantory Variables.Journal of Business and Economic Statistics.3(2)

Ashley, R.(1988).On the Relative Worth of Recent Macroeconomic Forecasts.International Journal of Forecasting.4(3)

Ball, L., Croushore, D.(2002).How Do Forecasts Respond to Changes in Monetary Policy?.Business Review.Q4