https://scholars.lib.ntu.edu.tw/handle/123456789/54339
標題: | 失業真的會導致犯罪嗎?以臺灣1978年至2003年縣市資料為例 Does Unemployment Increases Crime? Evidence from County Data of Taiwan 1978-2003 |
作者: | 林明仁 劉仲偉 Lin, M.J. Liu, C.W. |
關鍵字: | 失業;犯罪率;工具變數;unemployment;crime rates;instrumental variables | 公開日期: | 十二月-2006 | 卷: | 34 | 期: | 4 | 起(迄)頁: | 445-482 | 來源出版物: | 經濟論文叢刊 | 摘要: | 國外文獻一般認為失業率上升只會使財產犯罪增加(Levitt,2004);然而幾乎所有回歸結果都未考慮內生性的問題(Piehl,1998)。本文首先簡單說明經濟理論如何預測失業與犯罪的關係,並討論同時性(simultaneity)如何對最小平方法(OLS)的估計造成偏誤。我們接著提出美元匯率、日圓匯率以及能源價格三者分別與製造業就業人口比例乘積作為失業率的工具變數,並從理論與弱工具變數測試(weak Ⅳ test)兩方面同時探討該組工具變數之有效性。在實證結果方面,雖然在最小平方法下失業率對各類犯罪影響幾乎都為正且顯著,但在兩階段最小平方法(2SLS)下,失業率只對財產犯罪(主要在其中的竊盜一項)有正的顯著影響,對暴力犯罪則無。且2SLS估計值皆大於OLS的結果。最後,過度認定測試(over identification test)的結果也從統計上支援這些工具變數的外生性。 Unemployment has a small positive effect only on property crime (Levitt, 2004). However, most research has not considered the problem of simultaneity (Piehl, 1998). In this paper, we first describe how economic theory predicts the effect of unemployment on crime. Exchange rates for the US dollar and Japanese Yen against the New Taiwanese dollar, along with oil price are then used as novel instrumental variables to perform a 2SLS. Under OLS, the effect of unemployment is positive and significant on both violent and property crime. Under 2SLS, unemployment has positive significant effect only on property (theft) but not violent crime. Furthermore, the estimates are larger under 2SLS than OLS. Our results also show robustness in the Weak Ⅳ test at the first stage and an over-identification test at the second stage. |
URI: | http://ntur.lib.ntu.edu.tw//handle/246246/282135 |
顯示於: | 經濟學系 |
在 IR 系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。