勞動市場若存在樣本選擇（sample selection)問題，本研究嘗試建構 聯立迴歸模型，同時包含隨機工資與工時兩條邊界方程式，可探討 男女兩性勞工薪資低付程度課題。因存在組合誤差項，必須使用關 聯結構法推導出聯合機率密度函數，進而建構隨機邊界關聯結構模 型。整理民國94、96、98、100與102等五年的台灣「人力運用調 查」資料，進行迴歸分析。發現考慮樣本選擇的平均薪資效率遠低 於未考慮樣本選擇者，與以往文獻有相當差異，可能原因爲以往文 獻探討薪資效率時，未同時考慮樣本選擇問題，將無工作者樣本完 全排除，導致迴歸分析結果僅適用於有工作者。

This paper compiles the “Manpower Utilization Survey” data, to study the issue of wage underpayment. Assuming sample selection, we apply copula methods to derive the joint probability density function for composed errors in the equations of wage and hours of work. The likelihood function can take both workers, who have observed wages, and non-workers, who have no observed wages, into account. Non-workers are usually excluded by previous researchers, studying similar issues, since they overlook the role of sample selection. The empirical results show that the trend of wage efficiency in the categories of working identity and working area are almost the same in each gender, whether correcting for the sample selection problem or not. However, in the remaining 4 categories, the sample selection bias appears to play an important role on the determination of wage efficiency. With the correction of the samples election bias, most of the findings differ from the past literature that considers only workers with wages and salary.