本研究運用我國90 年與94 年國民健康訪問調查資料，分男女按體 重分組估計體重控制對糖尿病衝擊中老年就業的緩和效果。為能考 量糖尿病與體重控制為內生，使用帶工具變數三元Probit 模型， 工 具變數包含糖尿病家族病史和糖尿病與體重控制三區域盛行率。比 較二年度分析可知： 90 年因缺乏糖尿病家族病史且二區域盛行率變 異不夠大， 緩和效果較不明確﹔惟94 年之分析則無此二項限制，過重 以上（ BMI 三24）男性之罐車日效果呈顯著為1.98% ，其它各組之效 果多偏小或不顯著。政策制定者估算體重控制成求效益時， 宜將中 老年過重以上男性之緩和效果納入考量。

This study examines whether weight control ameliorates the effect of diabetes on employment among middle-aged and older adults (aged 40 to 64) by gender and weight group. It analyzes two waves of the National Health Interview Survey of Taiwan (200 1, 200）. Allowing for endogeneity of diabetes and weight control, it uses a trivariate probit model with instrumental variables, including the family history of diabetes, area prevalence rates of diabetes d weight control. Due to lack of famihisto ry of diabetes d insufficient variations in the two area prevalence rates, the findings from the 2001 wave are not totally according to expectation. However, the analysis of the 200data does not suffer similar constraints. The moderating effect r overweight or obese (defined as BMI 24) males is statistically significa， on the order of 1. 98%. Most effects are either small or insignificant for other groups . τ'a.king into account the moderating effect r overweight or obese, middle-aged and older males will enhance completeness of cost-benefit analysis in evaluating weight control policy