總體經濟預測的準確度與預測者的獨立性

Abstract

本篇研究欲比較獨立預測者與非獨立預測者在預測實質國內生產毛額、消費者物價指數及失業率的準確度。實證採用1995年至2008年Consensus Economics所發佈的七個國家的預測資料，包含台灣、中國、美國、英國、法國、德國和日本。非獨立預測者是挑選自2009年被Bloomberg評為全球前二十大的投資銀行。根據實證結果顯示，獨立預測者在預測表現上相對樂觀。在非衰退期，非獨立預測者在預測國內生產毛額與失業率時均較為準確；而在衰退期，雖然預測誤差均較非衰退期高，但檢定結果發現不論是獨立預測者或非獨立預測者，都擁有可減少預測誤差的有用資訊。

This paper compares the accuracy of GDP growth, inflation and unemployment rate forecasts produced by independent and non-independent forecasters. The data come from the monthly Consensus Economics forecasting service, for seven countries over the period from 1995 to 2008. Investment banks (which are on the list of the Bloomberg 20 in 2009) are the target non-independent forecasters. The empirical results show that independent forecasters are more optimistic in most cases. And with few exceptions, the independent forecasts are more biased in terms of root mean square error. Formal tests show these differences are statistically significant for forecasts of GDP growth and unemployment rate in the non-recession periods. And it is striking that although the forecasts in the recession period are more biased, both forecasters contain useful information that could be used to reduce the error.