由於傳統探討逃漏稅行為之文獻皆假設商品之消費可無限細分，然而假設商品之消費皆可無限細分並不符合現實世界的情況。因此，本文設定一模型包含了可無限細分之財貨及不可無限細分之財貨，目的在於探討納稅人在存在不可無限細分之財貨的環境下是否會改變其逃漏稅決策，進而得出不同於傳統逃漏稅文獻之結論。而為了與傳統逃漏稅文獻相對應，除了商品具不可分割性的假設之外，其餘參數仍沿用傳統逃漏稅文獻之設定：所得、稽徵機率和處罰倍率皆為外生。 透過模型設定，本文得到了三個和傳統逃漏稅文獻十分不同的結論。第一，在商品不可無限分割之下，納稅人願意進行逃漏稅的條件相對於藉由傳統逃漏稅模型所得之條件寬鬆，甚至是願意在對其不利的條件之下進行租稅逃漏。第二，逃漏稅額與稅率之間不全然為單調遞減的關係，其方向並不明確。因而可以改善傳統逃漏稅模型預測最適逃漏稅額與稅率之間為單調遞減關係之結果，使其更加符合現實世界的情況。 第三，在商品具有不可分割性的假設之下，納稅人的風險偏好會因而改變使其不為一純粹的風險趨避者。因此，本文發現若是政府藉由調降政策變數以允許納稅人進行租稅逃漏，則政府之預期稅收會因為允許納稅人逃漏稅而提高，進而得以提升整體之社會福利。最後，在本文的設定之下，最適的處罰倍率不必然為無窮大。 / Traditional tax evasion researches all assumed that consumptions of goods are divisible, but those assumptions do not conform reality. Thus, our research set a model include one divisible good and one indivisible good in order to inquire that if taxpayers change their tax evasion decision under the assumption. In order to correspond to traditional tax evasion researches, our model still practices the settings of traditional tax evasion researches, which include exogenous income, fine rate and probability rate. Through the setting, our research obtained three important conclusions that are different from traditional tax evasion researches. First, under our assumption, taxpayers are more willing to evolve in tax evasions even the conditions are unfavorable to them. Second, the relationship between tax evasion and tax rate is no longer monotone decreasing. Therefore, the conclusion is able to reform the impractical conclusion obtain by traditional tax evasion researches. Third, under our setting, taxpayers are no longer pure Risk-Averter. Hence, we discover that if government slash the policy parameter to encourage taxpayers evolve in tax evasions, expect revenue of government will rise. Since the expect revenue of government rises, social welfare rises as well.