性別薪資差異之選擇性偏誤與薪資拆解—臺灣高等教育擴張下之實證

Gender Wage Gap, Selection Bias, and Wage Decomposition: Evidence from Higher Education Expansion in Taiwan

李明純(Ming-Chun Lee)

指導教授 ： 黃芳玫

國立臺灣大學/生物資源暨農學院/農業經濟學研究所/碩士(2019年)

https://doi.org/10.6342/NTU201803425

摘要

臺灣自1980年代以來，經歷了一連串的高等教育擴張歷程，而在這進程中，高等教育淨在學率由原本男性高於女性，自1988年後，轉變為女性超越男性的情形，同步地，性別間的薪資差距也逐年在縮小。又性別間的薪資差異一直是各國勞動學者關注的議題之一，臺灣方面亦不缺乏對於此議題的探討，針對高等教育與性別薪資差距關聯的討論更不在少數，但由於女性勞動參與行為具有選擇性偏誤的問題，故修正選擇性偏誤對於探討性別薪資差異相當重要。臺灣的文獻雖然有部分考慮到此問題，但對於選擇性偏誤矯正前後的結果比較，還未有系統性的討論。故本研究之主要目的為探討過去35年，臺灣高等教育快速擴張對性別薪資差異之影響，再細分為以下兩點：(1)採用Gronau-Heckman-Roy模型，考量女性全職工作之選擇性偏誤對性別薪資差異之影響，(2)再利用Neuman and Oaxaca薪資拆解模型探討價格、要素因子以及選擇性偏誤在性別薪資差異中之角色。
本研究利用行政院主計處「人力資源暨人力運用調查」資料，選取1983~1987、1993~1997、2003~2007、2013~2017年的四群時間點資料作為高等教育擴張前、高等教育擴張後初期、中期及近期的樣本，並限制樣本年齡在壯年的25~54歲。於採用Gronau-Heckman-Roy模型的實證結果顯示，女性的選擇行為在高等教育擴張前為顯著的負向選擇行為，意即女性全職勞動市場中有較高的比例為低人力資本勞工，而高等教育擴張後初期也為負向但較不顯著，直到高等教育擴張後中期轉變為正向選擇行為，但到擴張後近期才為顯著，表示女性全職勞動市場已經轉變為有利於高人力資本的勞工，另外，其他條件不變下，四段期間的比較中，教育程度為專科或大學以上學歷之選擇性偏誤變化最小，婚姻狀況則又以未婚的選擇性偏誤變化最小。
而性別薪資差距在四個時期裡逐漸縮小的過程中，對當期做薪資拆解之結果為：價格因子皆為造成當期性別薪資差距的主要因素，選擇性偏誤在高等教育擴張前以及擴張後初期是造成差距的因素之一，到高等教育擴張後中期及近期，則轉變為抵銷薪資差距的效果，而要素因子除高等教育擴張前的階段外，都有降低薪資差距的效果，並隨著時間的推移，要素因子的貢獻逐漸增大，然而，分解各解釋變數對於要素因子的貢獻後，發現專科及大學以上的貢獻逐年增大，故高等教育擴張確實可以透過提升女性的平均特性，來達到縮小性別薪資差距的效果。最後，分解跨期的薪資差距，由高等教育擴張前至近期，經過約35年，三大項因素的改變都是縮小薪資差距的功臣之一：要素因子透過高等教育擴張帶動女性的人力資本提升而縮小差距，性別歧視的減少也為性別薪資差距縮小做出貢獻，而女性全職工作選擇行為的改變，更是促成了大部分縮小的性別間薪資差距。

關鍵字

薪資差異 ； 選擇性偏誤 ； 薪資拆解 ； 高等教育擴張

並列摘要

Since the 1980s, Taiwan has experienced a series of higher education expansions. In this process, the net enrollment rate of higher education has changed from male to female. Since 1988, it has changed to female surpassing male. The wage gap between genders is also shrinking year by year. Gender differences in wages have always been one of the topics of concern to labor scholars in various countries. There is no shortage of discussion on this topic in Taiwan. However, choices for female labor participation behavior causes selection bias problem. This problem might make estimates bias, so correcting selection bias is very important for exploring gender pay differences. Although Taiwan's literature has partly considered this problem, there has been no systematic discussion on the results before and after the correction of selective errors. Therefore, the main purpose of this study is to explore the impact of the rapid expansion of higher education in Taiwan on gender pay differences over the past 35 years. That further subdivides into the following two methods: (1) using the Gronau-Heckman-Roy model to consider the selection bias of women's full-time work on the impact of gender pay differences, (2) using the Neuman and Oaxaca model to decompose wage gap into the role of price, factor factors, and selection bias.
This study used the "Manpower Utilization Survey" of Taiwan to select four groups of periods from 1983 to 1987, 1993 to 1997, 2003 to 2007, and 2013 to 2017 as the pre-expansion of higher education, and initial, intermediate and recent period of post-expansion of higher education. Also, limit the age of the sample to 25 to 54 years old. The empirical results of the Gronau-Heckman-Roy model show that women's choice behavior is a significant negative selection behavior in pre-expansion of higher education, which means that a high proportion of female full-time labor market is low human capital labor. After the expansion of education, it was also negative but less significant in the initial stage. It was not until the intermediate expansion of higher education it changed to positive selection behavior. It indicates that the female full-time labor market has transformed into a labor force that is conducive to high human capital. In addition, ceteris paribus assumption, in the comparison of the four periods, the degree of education has the least change in the selective bias of the some college degree or above, and the marital status has the smallest change in the selective bias of the unmarried.
In the process of narrowing the gender pay gap in the four periods, the result of the current wage decomposition is that the price factor is the main factor causing the current gender pay gap. The selection bias is one of the factors that caused the gap before and after the expansion of higher education. In the intermediate and recent expansion of higher education, it turned into an offsetting effect on the wage gap. Except to the pre- expansion period of higher education, the characteristic factor has the effect of reducing the wage gap in other periods and the contribution has gradually increased. Additionally, after decomposing the contribution of each explanatory variable to characteristic factor, it founds that the contribution of some colleges and universities is increasing year by year. As a result, the expansion of higher education can indeed narrow the gender pay gap through improving the human capital of women. Last, decompose the inter-temporal wage gap from the expansion of higher education to the recent. After about 35 years, the three major factors change is one of the roles to narrow the wage gap. Characteristic factors narrows the gender wage gap through the expansion of higher education to enhance women's human capital. The reduction of gender discrimination have also contributed to the narrowing of the gender pay gap, and the change in women’s full-time job choices has contributed to it the most.

並列關鍵字

Wage Differential ； Selection Bias ； Wage Decomposition ； Higher Education Expansion

參考文獻

行政院主計總處(1984-2018)，72-106年人力運用調查(AA020006-AA020040)【原始數據】取自中央研究院人文社會科學研究中心調查研究專題中心學術調查研究資料庫。

徐美 (2004)，「有偶婦女在臺灣勞動市場轉型中參與行為的變動」，發表於2004年臺灣人口學會年會暨「人口、家庭與國民健康政策回顧與展望」研討會。台北，2004年12月。

徐美、陳明郎、方俊德 (2006)，「臺灣產業結構變遷和性別歧視對男女薪資溢酬變動趨勢之影響」，《經濟論文》，34(4)，505－539。

國家發展委員會 (2017)，《都市及區域發展統計彙編》，台北：國家發展委員會。

教育部 (2016)，《教育程度標準分類第5次修正版 (105年1月)》，台北：教育部。