本文利用1978-002年之《人力運用調查》,觀察台灣近二十多年來之薪資差異變化,發現包含勞動者未觀察到能力報酬的大專教育溢酬自1980年以後男性約下降了30.1%,而女性亦減少了9.2%。此外,男性與女性每年薪資方程式之殘差項變異數亦呈下降趨勢。因此本文進一步檢測大專教育溢酬之減少,以及性別間變化之差異是否因勞動者未觀察到能力(unobserved ability)報酬之變化。本文以Chay and Lee(2000)之誤差分解模型(error component model),假設勞動者本身天賦並不隨時間而改變,在誤差項裡形成一永久性組成份子(permanent error component);因此藉由相同天賦族群之薪資殘差變化,可估計勞動者未觀察到能力報酬之變化。實證發現1990-2000年大專男性之勞動者未觀察到能力報酬約下降了11-30 %,女性則為5-11%。而此男性勞動者未觀察到能力報酬的減少約解釋了10.0-31.3%之大專教育溢酬下降,女性則為5.0-14.3%。而且因男性之勞動者未觀察到能力報酬降幅大於女性,使得大專生的性別薪資差異減少。
Results from the 1978-2002 Taiwan's Manpower Utilization Surveys show that the observed college premiums experienced a 30.1% decline among males and 9.2% decline among female, and the variances of ”within group” log wages de creased over time. The paper adopts the Chay and Lee (2000) error component model approach, and estimates the unobserved ability premium by allowing un observed productivity to be the permanent component, in which estimation of changes in the return to unobserved ability is based on variation in within-group log-wage variances across groups and over time. The model estimates suggest that 11-30% decrease in the unobserved ability premium for male and 5-11% for female during the 1990s. The evidences also imply that the decrease in the return to ability can account for at most 10.0-31.3% of the observed decrease in the college premium for males, and 5.0-14.3% for females. In addition, the finding that the well-educated male experienced more decline in returns to ability relative to the female reduces the sex-related wage differentials.