本文利用1961-1999之樣本期間來探討台灣之失業率與產出成長率間之歐肯關係是否成立。實證結果指出歐肯關係存在,且此一結果不受不同模型設定的影響。此外,實證結果亦指出除失業率外,資本生產力與勞動力的變化對產出成長亦有影響,遺漏此二變數將高估失業對產出的影響效果。然而,不同模型所估計的歐肯係數值不盡相同,在一階差分模型中,歐肯係數之估計值為-2.1,然此一模型不具結構穩定性。在應用Beveridge-Nelson與Hodrick-Prescott分解法之缺口模型中,歐肯係數之估計值則分別為-7.0與-5.2,且模型具結構穩定性,同時未發現失業率對產出成長的影響具不對稱性。
This paper examines the validity of Okun's law in Taiwan during 1961 to 1999. Using three detrending methods (first differencing, the Beveridge-Nelson decomposition procedure and the Hodrick-Prescott filter) to decompose variables, we find that there exists a negative relationship between the cyclical output growth and the cyclical unemployment rate. In addition to changes in unemployment rate, changes in capital productivity and labor force are also important variables in explaining changes in output growth; omission of the two variables would cause the estimates of the Okun coefficient to be upwardly biased. Estimates of the Okun coefficient and the stability of Okun's relationship are sensitive to the trend/cycle decomposition. In the first-difference model, the estimate is -2.1 and Okun's relationship is unstable. In the BN and HP model, the estimates are -7.0 and -5.2, respectively, and both models exhibit stability; there is no asymmetric effect of the unemployment rate on output growth.