本文旨在探討學校教育對個人薪資所得的影響。因為許多學者均認為:對薪資與學歷進行迴歸分析時,忽略個人先天能力及家庭背景可能造成教育報酬率在估計上的偏誤;所以,我們嘗試納入代表家庭別的虛擬變數,以控制個人先天能力及家庭背景,進而凸顯學校教育對個人薪資所得的直接效果。本文使用1989年、1998年及2007年橫跨十八年之三個不同時期的「人力運用調查」資料。實證結果顯示:第一、忽略個人先天能力及家庭背景將高估教育報酬率,而在解釋變數中加入家庭虛擬變數及加入父母親教育年數,均可改善教育報酬率高估的情形;比較兩種修正方式,加入家庭虛擬變數能使被高估的教育報酬率較大幅度地向下修正。第二、教育報酬率隨著時間有些微上升的趨勢;如果以教育年數代表教育程度,大致而言,台灣的教育報酬率約為4%。 This thesis aims to analyze the schooling''s influence on the individual salary. Many scholars think that omitting the innate ability variable or the family background variable will cause bias in estimation of returns to schooling. In order to highlight the schooling''s direct effect on the individual salary, we attempt to add the family dummy variable to control the innate ability and the family background. The data for this thesis are taken from the “Manpower Utilization Survey” in 1989, in 1998 and in 2007. The empirical results can be summarized as follows: First, both adding the family dummy variable and adding the number of the parent''s education years can improve the overestimated situation of the returns to schooling. However, adding the family dummy variable is a better method because the magnitude of its correction is bigger. Second, Taiwan''s rate of return to schooling has slightly risen during the observed period and it is approximately 4%.