超越身體質量指數：體型對就業的影響

Beyond BMI: The impact of body size on employment

吳婉鈺(Wan-Yu Wu)

指導教授 ： 胡登淵

淡江大學/商管學院/產業經濟學系碩士班/碩士(2014年)

https://doi.org/10.6846/TKU.2014.00684

摘要

近年來，肥胖在許多已開發國家中，成為一個普遍盛行的健康問題。然而，肥胖除了會影響健康外，還可能進而影響生產力與勞動市場的表現。現有許多肥胖和勞動市場的研究大多是以BMI作為肥胖的衡量指標，但在醫學文獻中，BMI並非總是被認為是一個有效的肥胖衡量，因為它沒有區分脂肪與無脂肪質量(肌肉、骨骼等)。
有鑑於此，本文使用三種肥胖衡量指標，分別為BMI、體脂肪百分比與腰圍來研究體型對就業的影響，資料取自1999~2004年共三波的美國國民健康與營養調查(National Health and Nutrition Examination Surveys, NHANES)，因為此三波調查資料涵蓋這三種肥胖衡量指標。本文採用假設分佈較具彈性的半無母數(Seminonparametric)與二元選擇的半無母數(Seminonparametric estimation of bivariate binary-choice)模型進行估計，並採用肥胖盛行率作為工具變數，以Control function approach來控制肥胖與就業間的內生性問題。
研究結果顯示：在半無母數模型中， BMI、PBF與腰圍對於男性就業機率之影響呈現倒U型，其中又以腰圍最為顯著，而女性BMI、PBF與腰圍對就業機率的影響呈負向線型；在二元選擇的半無母數模型中，不論男性和女性，BMI≧30的人，其就業機率都會顯著較低。然而，體脂肪肥胖和腰圍肥胖僅對女性就業有顯著的負向影響，即女性體脂肪百分比＞32%者、腰圍＞88公分者，會有相對較低的就業機率，在三種指標中，腰圍肥胖對女性就業的負向影響是最大的。
本文進而嘗試控制肥胖相關疾病－高血壓與糖尿病─作為控制變數，並採用高血壓盛行率與糖尿病盛行率作為工具變數，以Control function approach處理兩種疾病的內生性問題。所有模型在控制此二項健康條件後，大多會稍微弱化肥胖對就業的影響。
倘若僅以BMI作為肥胖的衡量指標，那麼男、女性肥胖對其勞動市場結果皆會有負向的影響，但若進一步考量PBF和腰圍指標，不僅能捕捉到腰圍對男性就業之倒U型影響，亦能看出肥胖對女性就業的負向影響較明顯。

關鍵字

就業 ； BMI ； 體脂肪百分比 ； 腰圍 ； 半無母數模型 ； 二元選擇的半無母數模型

並列摘要

In recent years, obesity has become a widespread prevalence of health problems in many developed countries. However, obesity will not only affect health but also labor market outcomes. Many existing studies of obesity almost universally use BMI as a measure of obesity. In the medical literature, the belief that BMI is an accurate measure of obesity is not always well-founded, because it does not distinguish fat from muscle, bone, and other lean body mass.
　　This study, hence, uses three measures of obesity--BMI, percent body fat and waist circumference--to examine the impact of body size on employment. The data come from three waves of National Health and Nutrition Examination Survey (NHANES), including the 1999-2000, 2001-2002 and 2003-2004 surveys, since these three waves of data have those three measures of obesity. By making flexible distribution assumption, it uses the seminonparametric model and seminonparametric estimation of bivariate binary-choice model. To take the potential endogeneity of BMI, percent body fat and waist circumference into account, it uses regional prevalence rate of obesity as our instrumental variable and uses Control function approach to tackle this problem.
　　The results of seminonparametric model indicate an inverted U-shaped curve for the effects of BMI, percent body fat or waist circumference on employment for men, especially significant in the waist circumference model. On the other hand, the results show a negative linear effect for the impact of BMI, percent body fat or waist circumference on employment for women, and they are all significant. The results of seminonparametric estimation of bivariate binary-choice model indicate that both men and women whose BMI is greater than or equal to 30 have significantly lower probability of employment. However, we only find significantly negative impact of percent body fat obesity and central obesity on employment among women. In the three measures of obesity, central obesity has the greatest impact on employment for woman.
　　This study, furthermore, attempts to use obesity-related diseases--hypertension and diabetes--as control variables. To take the potential endogeneity of hypertension and diabetes into account, it uses regional prevalence rate of hypertension and diabetes as instrumental variables, and it, again, uses Control function approach to tackle this problem. After controlling for those two health conditions, it finds that the impact of obesity on employment is slightly weakened, in general.
　　If we only use BMI as a measure of obesity, obesity will have the negative impact on labor market outcomes for both men and women. However, if we further consider the impact of percent body fat or waist circumference on employment, we not only capture an inverted U-shaped curve for the effects of waist circumference on employment for men, but also find that obesity will negatively affect women relatively more than men.

並列關鍵字

Employment ； BMI ； Percent body fat ； Waist circumference ； Seminonparametric model ； Seminonparametric estimation of bivariate binary-choice model

參考文獻

侯季志. (2012). 體型對失業的影響-以半無母數分析. 淡江大學產業經濟學系碩士班學位論文, , 1-125.

李冠威. (2013). 體型對職業的影響─ 以半無母數模型分析. 淡江大學產業經濟學系碩士班學位論文.

許雅棠、胡登淵、林東瓏 (2011). 相對體型對薪資的影響: 台灣老少群體的半母數分析. 台灣經濟學年會.

Asgeirsdottir, T. L. (2011). Do body weight and gender shape the work force? The case of Iceland. Economics & Human Biology, 9(2), 148-156.

Baum, C. L., & Ford, W. F. (2004). The wage effects of obesity: A longitudinal study. Health Economics, 13(9), 885-899.