有許多文獻討論綠色租稅改革的租稅交互效果(Bovenberg and de Mooij, 1994)，其中有三篇文章(schwartz and Repetto, 2000; Williams, 2002, 2003)討論健康效果於綠色租稅改革的福利效果以及對最適環境稅的影響。首先，Schwartz and Repetto (2000)的模型將健康效果隱含於效用函數之中，接著，Williams (2002)明確地將代表健康的疾病函數納入模型之中。然而，Schwartz and Repetto (2000)與Williams (2003)的結論並不一致。Schwartz and Repetto (2000)認為最適環境稅可能大於邊際污染傷害； Williams (2003)則認為最適的污染稅率小於邊際污染傷害。本文認為Williams (2003)用兩個函數的形式代表健康效果的表示方式並不適當，本文認為合理的健康生產函數的投入有預防治療支出與環境品質兩者，而其產出應該同時影響效用函數與時間限制式，本文建立較合理的健康生產函數，得到一個新的租稅交互效果－預防治療的效益面租稅交互效果，透過模擬分析發現此效果為正。因此考量預防治療的效益面租稅交互效果後，我們認為最適的污染稅應該大於邊際污染傷害，此結論異於Williams (2003)的結論。

There are three articles (Schwartz and Repetto, 2000; Williams, 2002, 2003) that have introduced health effects into the tax-interaction effect in the green tax reform literature and examined their impacts on the optimal environmental taxes. While the health effect is implicitly represented in the utility function in Schwartz and Repetto (2000), it is explicitly modeled using a sickness function in Williams (2002, 2003) and a health production function in Williams (2003). In Williams' (2003) model, sickness enters time constraint and health affects utility function. However, Schwartz and Repetto (2000) and Williams (2003) reach different conclusions. Schwartz and Repetto (2000) suggest the positive impacts of better environmental quality on labor supply will increase welfare and the optimal environmental tax may be larger than the marginal pollution damage. Williams (2003) find that the impact of health effects on welfare is negative and the optimal environmental tax should be less than the marginal pollution damage. In this paper we argue that the two-function representation of the health effect in Williams (2003) is inappropriate. By introducing a health production function which depends on mitigating activities and environmental quality and affects both the utility and time constraint at the same time in a general equilibrium framework, our model shows that the mitigation-based benefit-side tax-interaction effect emerges. We find that its effect on welfare is positive using a numerical analysis of the model. Thus, contrary to Williams (2003), we conclude that the optimal pollution tax should be larger than the marginal pollution damage after taking the mitigation-based benefit-side tax-interaction effect into consideration.