政府與非政府總體經濟預測之比較

Abstract

過去許多學者都研究關於經濟預測的誤差，並且探討導致長期預測偏誤的原因。在這篇論文中，我們蒐集中國大陸、法國、德國、日本、台灣、英國以及美國政府以及Consensus Economics Inc.從1997到2008年的經濟指標預測數據，並依照預測機構的特性，將他們分成政府與非政府，並且觀察政府的預測偏誤是否比非政府機構還大。我們使用Wilcoxon 符號等級檢定以及Fair - Shiller 檢定，並發現德國和美國的政府機構表現較差，而中國的預測則較準確。我們又將預測的時間分為衰退與非衰退時期，最後我們發現在衰退時期，所有的機構皆會高估GDP，但是政府機構的預測誤差會較非政府還大。我們將政府與非政府的預測準確度在同一期間排名，而發現某些國家的政府預測準確度有較差的情形。最後我們使用Granger 因果關係檢定，而發現政府的預測有領先非政府的趨勢。

Many researchers find biased in economic forecasts, and discuss the phenomenon why institutions make continuous biased forecasts. In this thesis, we classify the forecaster into government and non-government sectors, and investigate whether government forecasts are more biased than non-government sectors. We use Wilcoxon signed-rank test and Fair-Shiller test, and then find Germany and United States are less accurate in government forecasts in GDP and CPI, but in opposite in China. In addition, we separate the forecasts region into recession and non-recession period, and observed that all forecasters are much more overestimated, and the government forecasters are more inaccurate than non-government in recession period. We rank the forecast bias of government in peers, and find some predicted most inaccuracy than other countries. Finally, we use the Granger causality test and find government forecast is a leading indicator of non-governmnets‘.