本研究採分量迴歸與重複抽樣 (bootstrapping)，搭配1990 年至2014 年長達25 年期間的個人年度勞動所得資料，以新方法和新資料型態探討勞動所得不均度變化，歸因於薪資所得者的教育年數、工作經驗、性別、婚姻狀態、公私部門與產業別等，各種特性所佔比例，並進行分配變化趨勢成因的擬真分解。結果發現，教育年數及教育報酬率的提高，有助於勞動所得的全面提升，尤以中間和基層勞工的受益較大；產業別的報酬率變化，反映金融海嘯之後的產業結構演進，對低所得族群有較大的負面衝擊。至於其他特性和報酬率的分配，影響效果都屬其次。本研究結果隱含，透過合理的補貼和在職訓練，提升弱勢族群的教育報酬率，有助於勞動所得不均度的改善。

Using twenty-five years of data from 1990 to 2014 on individual annual labor income, this study analyzes the determinants of Taiwan’s labor income inequality based on a quantile regression technique with the bootstrap method. We utilize new econometric methods and data sources to examine whether changes in labor income inequality are attributable to workers’ education years, working experience, gender, marital status, public or private sector employment, and industry sectors. The results suggest that more schooling years and larger marginal rates of return to education lead to increases in an individual’s labor income, especially for low- and middle-income workers. Changes in returns to different industry sectors show that industrial structure adjustments arising from the 2008-2009 global financial tsunami may have an adverse effect on low-income labors. Other factors that affect wage inequality are secondary. The results therefore imply that governments can reduce labor income inequality through education subsidies and on-the-job trainings for the disadvantageous workers.