綠色租稅改革的重要議題之一乃是「雙紅利假說」(double dividends hypothesis)的驗證，多數文獻認為第二重紅利的存在並非理所當然，其原因不一而足。本文延伸龐雅文、蕭代基[1]的模型，將其「預防治療活動」進一步區分為治療支出與預防支出，以區別二者對家計單位之健康效果的異質性、及其於個人所得稅之可扣除性的差別待遇，同時考慮不同性質時間（包括勞動、休閒與生病時間）的價值差異，提出正確的邊際社會損害(marginal social damage, MSD)之衡量方式，結果發現，除了過去文獻所提出的皮古效果、收入循環效果、租稅交互效果、效益面租稅交互效果、以及醫療的效益面租稅交互效果外，尚可衍生「預防支出的效益面租稅交互效果」，理論上符號不確定，使得第二重紅利的存在性更增添不確定性因素。

The double dividend hypothesis plays important role in recent agenda on green tax reform. Its existence of the second dividend, however, is doubtful for various reasons. We extend Pang and Shaw’s (2007) model by dividing the medical expenditure into treatment expenditure and precautionay expenditure to capture their difference in health effects and income tax deductability. Differentiating the value of time of different properties and redefinition of the marginal social damage (MSD) of dirty goods consumption. The result shows that the prevention-based benefit-side tax-interaction effect emerges besides the Pigouvian effect, the revenue-recycling effect, the tax-interaction effect, the benefit-side tax-interaction effect and the mitigation-based benefit-side tax-interaction effect in the previous literature. This study indicates that the second dividend is theoretically indeterminate.