本文探討體型對於失業的影響，以相對BMI與BMI定義體型，採用半無母數模型(Seminonparametric model)估計，並以此和Probit模型估計結果進行比較，半無母數分配假設較為寬鬆，不需如同Probit模型需將分配假設為常態。 研究資料使用2010年最新釋出的美國國民健康調查(National Health Interview Survey,NHIS)資料，並將樣本分為男女進行討論。此外，考慮到相對BMI或BMI與失業的內生性，使用Control Function approach處理此問題，使用的工具變數為區域肥胖盛行率。根據本研究結果，半無母數模型結果有半數皆拒絕Probit模型。半無母數模型的實證結果顯示：不分年齡下，男性相對BMI對失業沒有顯著影響，BMI對失業有負向顯著影響，即BMI上升使失業率下降；女性的相對BMI對失業沒有顯著影響，BMI則對失業有正向顯著影響，即BMI上升使失業率上升。年輕男性的相對BMI以及BMI皆對失業沒有顯著影響，年輕女性的相對BMI以及BMI對失業皆有正向顯著影響；中年男女性的相對BMI以及BMI皆對失業沒有顯著影響；老年男性的相對BMI以及BMI皆對失業有負向顯著影響，老年女性的相對BMI以及BMI對失業皆沒有顯著影響。本研究並繪出邊際效果圖形，可從圖形看出男女體型對於失業的變動情形。總而言之，體型只對老年男性失業有負向顯著影響；體型只對於年輕女性失業有正向顯著影響。 This study examines the impact of body size on the unemploymentby defining the body size as relative BMI and BMI. It estimates the effects by using Seminonparametric model, compared with the results of Probit model. The distribution assumption is relaxed in the Seminonparametric model, which does not assume normal distribution such as Probit model. According to the results of this research, half of Seminonparametric models reject Probit model. Data using the latest release of the 2010 National Health Interview Survey (NHIS) of the U.S., and the sample isdivided into male and female samples and samples of young, middle-aged and old cohorts. In order to take the endogeneity of the relative BMI or BMI into account, we tackle this problem by using Control Function approach the instrumental variable is regional prevalence rate of obesity. The results indicate that effect of relative BMI is insignificant, BMI is significantly nagtive for the whole sample of males.While the effect of relative BMI is insignificant, BMI is significantly positivefor the whole sample of females. Neither the effect of relative BMI nor the effect of BMI is significant for young males but a positive and significant effect is found for young females. Both relative BMI and BMI are not significant for middle-aged male and middle-aged female. Both relative BMI and BMI are negatively significant effect on unemployment for elderly male but insignificant effect for elderly female. This study also presents the patterns of marginal effects by using graphs. All in all, body size has a negatively significant impact on unemployment only for elderly male. Increasing in body size will decrease the unemployment rate for elderly male. Body size has a positively significant impact on unemployment only for young female. Increasing body size will increase the unemployment rate for young female.
