Kaplow (1992, AER)在投保者為單一類型的假設下發現，所得稅損失扣除政策會降低投保人購買保險的意願，因此會扭曲投保人的保險決策，導致社會福利水準的降低。本文應用Rothschild and Stiglitz (1976)的保險模型建構一個包含政府、保險公司以及投保人的三階段性賽局(three-stage game)，分析所得稅損失扣除政策的福利效果。文中有兩點重要發現：首先，所得稅損失扣除政策會改變Rothschild and Stiglitz (1976)一文分離均衡的特質：所得稅率高到超過高風險者的最大忍受程度時，私人的保險市場將不復存在；其次，該項政策具有三種福利效果，分別為所得重分配效果、高風險者之風險承擔效果以及低風險者之風險承擔效果。當上述三項的總效果為正，則引進所得稅損失扣除政策將使得社會福利水準獲得改善。另外，數值分析的結果顯示：惟有當所得稅率高時，引進損失扣除政策才具有福利增進的效果，不過此時私人的保險市場並不存在；當所得稅率較低，使得私人保險市場得以存在時，引進損失扣除政策的福利效果幾乎都為負，僅有少數近乎於零。

Under the assumption of identical insurees, Kaplow (1992, AER) shows that income tax deduction will induce insurees to buy partial insurance, which in turn distorts their insurance decision. Thus welfare with deduction will be lower than without deduction. This paper extends the setting of Rothschild and Stiglitz (1976) and constructs a three-stage game which consists of the government, insurers and insurees to study the welfare effects on income tax deduction policy for losses. It is shown in this extended model that: (1) if there exists a critical value for the marginal tax rate such that high-risk insurees do not insure, then the insurance market cannot exist in equilibrium, and (2) the choice of loss deduction policy involves three kinds of effects: the income redistribution effect; and the risk-taking effects for high-risk and low-risk insurees. If the three kinds of effects are positive in aggregate, then welfare with deduction will be higher than without deduction. However, when an insurance market exists, we show that welfare effects with deduction are negative mostly from numerical examples, but welfare effects will be improved when no insurance market exists.