本文運用台灣1998-2005年的縣市別動態追蹤資料，估計所得分配不均對於各種犯罪率的影響。在研究方法方面，除了OLS與固定效果模型外，並採用系統GMM方法進行估計。研究結果顯示，如果在模型中不考慮時間虛擬變項，則OLS、固定效果模型、動態追蹤資料模型GMM估計方法都相當一致地顯示出縣市以吉尼係數代表的所得分配不均程度，對於普通竊盜、故意殺人、強制性交犯罪率有顯著正向影響。同時在動態追蹤資料模型中，所得分配不均顯現出對於刑案犯罪率與暴力犯罪率都有顯著正向影響。如果在模型中考慮時間虛擬變項，則OLS、固定效果模型、動態追蹤資料模型GMM估計方法相當一致地顯示出所得分配不均程度對於普通竊盜犯罪率與故意殺人犯罪率有顯著正向影響。我們的研究結論與1990年代以後國外文獻中普遍發現所得分配不均與謀殺犯罪率之間有顯著正向關連的結論一致。

This paper applies the system GMM approach along with a traditional OLS and fixed-effect model to investigate the impact of income inequality on various crime rates based on Taiwan's regional data from 1998 to 2005. In the models without time dummy variables, we find that all models consistently show that the Gini coefficient, the income inequality measure we use in this paper, has significantly positive impact on common theft, murder and forcible rape. The estimation from the system GMM approach also implies that the income inequality leads to higher crime rates and especially higher violent crime rates. Once we introduce time dummy into our models, all approaches show that the common theft and murder rate are higher with greater income inequality, which is also consistent with the connection between income inequality and the murder rate found in the literature from other countries.