本篇文章使用「台灣地區家庭收支調查」資料,研究受教年數對薪酬的影響,並以二階段迴歸模型(工具變數法)去除其中可能會有的內生性問題。本篇使用出生季作為工具變數,受教年數作為第一階段的被解釋變數,薪酬(取對數值)作為第二階段的被解釋變數,年齡、年齡的平方、家戶人數、婚姻狀態、城鄉以及出生年作為第一階段與第二階段的其他解釋變數;觀察樣本為民國31~40年、41~50年以及51~60年的樣本,男女樣本分開評估。我們發現二階段迴歸模型中受教年數對薪酬的影響,比起一般最小平方迴歸模型中受教年數對薪酬的影響,有較高的趨勢,符合過往國外文獻的實證狀況。
We use "Taiwan Area Family Income/Expenditure Survey" data, and look into how number of schooling affects wage with two-stage regression model (also known instrumental variable method) which can remove endogeneity. In our model, quarter of birth is instrument; number of schooling is response variable in the first-stage regression; wage is response variable in the second-stage regression; age, square of age, size of household, marriage, city and country, and year of birth are the other explanatory variables in the first-stage regression and in the second-stage regression. The observed sample is in 1942~1951, 1952~1961, and 1962~1971; the sample is assessed separately for men and women. We found that the effect of number of schooling on wage in two-stage regression model is larger than the effect of number of schooling on wage in ordinary-least-square regression model. It is in line with the foreign empirical literature.