• 作者： 粱國源; 周炳宏
• 作者服務機構： 國立清華大學經濟學研究所
• 中文摘要： 本文旨在延續梁國源(1995)之研究，將重點轉移至定性方面之探討。我們首先回顧HM（Henriksson and Merton, 1981）檢定的基本原理(Merton, 1981），以及一些與它有關的統計檢定步驟的推導。其次，本研究利用行政院經濟建設委員會對某些關鍵性的總體經濟變數所發布的事前經建目標數據，應用各種變型的HM檢定從事實證分析。檢定結果顯示經建會對這些變數所發布的經建目標值，雖然以傳統非條件機率公式所求算的方向上精確度指標在某些情況下尚屬合理，唯它們卻大多無法通過各種變型的HM檢定，因此它們大多在Merton(1981)的涵義下，對使用者而言，並不具有意義也不能產生作用。
• 英文摘要： A previous study (Liang,1995) attempted a critical evaluation of macroeconomic forecasting in Taiwan.In that paper, the emphasis was on the quantitative sides of forecasting accuracy. Although we follow the samespirit of Liang (1995), this paper however, concentrates on the qualitative aspects of forecasting accuracy. Forthis purpose, we apply three variants, hypergeometric (Henriksson and Merton, 1981), contingency table (Stekler,1994）and regression（Cumby and Modest, 1987）, of the Henriksson-Merton test to the case of Taiwan. Specifi-cally, we use annual target values of several key macroeconomic variables released by the Council for EconomicPlanning and Development (CEPD). The empirical results suggest that the correlation between thesemacroeconomic targets and their corresponding actual values are generally not statistically significant. In otherwords, the hypotheses that these targets have no value for their users cannot be rejected statistically in mostcases.
• 中文關鍵字： 方向上的預測精確度; 預測誤差; 總體經濟經建目標值; Henriksson-Merton檢定﹑預測有用性
• 英文關鍵字： directional forecasting-accuracy; forecasting errors;macroeconomic targets; Henrilsson-Merton Test; usefulness of forecast