警察人數增加是否可以降低犯罪率，一直是學理和實務上的爭論焦點。警力與犯罪率之間是否有雙向因果關係，也是值得釐清的公共行政問題。本研究依據我國縣市1998 至2007 年之動態追蹤資料，透過Granger因果測試方法，測試竊盜犯罪率與警民比之間的雙向因果關係。研究結果發現，提高警民比（警察局警察人數對居民人口比率）可導致竊盜犯罪率顯著降低，且此一Granger 影響具恆久性；至於竊盜犯罪率升高，在短期內雖然能導致警民比增加，但是在長期並無顯著的Granger 影響。本實證研究結果支持政府可藉由提升警力及見警率來有效減少竊盜犯罪之發生。

Whether increasing the number of police can reduce the crime rate is a topic of much debate in researches and policies. The possible bi-directional causality between the number of police and the crime rate is also an issue worth clarifying in public administration. Using the dynamic panel data of regions in Taiwan between 1998 and 2007, this study employs the method of Granger causality test to investigate the bi-directional causality between the number of police and the theft rate. It is found that an increase in the police-population ratio (the ratio of police to population) can permanently Granger-cause a decrease in the theft rate. An increase in the theft rate can temporary Granger-cause an increase in the police-population ratio, but this causality is not permanent. Our results support the policy thinking that increasing the frequency of seeing police can effectively reduce the occurrence of theft.