酒駕違規與事故之因果關係分析

警察取締量增加是否能有效減少交通事故發生，一直是學理及實務尚未證實的議題。儘管如此，不論在政府的施政，或是民眾的期待，警察執法還是普遍被視為改善違法行為最有效之手段。基於此，本研究試圖探討酒駕違規取締與酒駕事故發生之間是否具有相互因果關係，並分析造成酒駕違規取締及酒駕事故發生之關鍵因素。本研究蒐集民國96年1月至105年12月，台灣22縣市酒駕取締及事故之縱橫資料，使用Granger causality test，檢驗酒駕違規取締件數與酒駕事故件數間潛在的相互因果關係。並依Granger causality test之分析結果及其落後期程，透過聯立迴歸方程式，進一步確認導致酒駕違規和酒駕事故之關鍵因素。 Granger causality test分析結果顯示，酒駕違規取締與酒駕事故發生之Granger因果關係，存在顯著的城鄉差異。酒駕違規取締與酒駕事故在都市地區(如：台北市、新北市、桃園市、台中市、台南市)產生明顯之Granger因果互動關係，且其影響皆於滯後1個月產生；而其餘縣市酒駕違規與事故間並無明顯之Granger因果關係。聯立迴歸估計結果顯示，民國102年6月13日開始實施更嚴格的酒精濃度取締標準，此法規修訂有效地減少酒駕事故之發生。另一方面，酒駕違規取締件數、公共運輸市佔率上升，人口密度、擁車率下降，也減少了酒後駕車事故發生的數量。至於酒駕違規取締件數方面，因人口密度及平均每人警政支出的增加，有助於酒駕違規行為減少。因此，本研究提出了相應的對策，以減少酒駕違規和酒駕事故的發生。

關鍵字

酒駕違規；酒駕事故； Granger因果關係；聯立迴歸

並列摘要

Whether the increase in police enforcement can effectively curtail the occurrence of traffic accidents has always been an issue that has not yet been confirmed both academically and practically. Nevertheless, from the perspectives of government administration or people expectation, enforcement is generally regarded as the most effective way to mitigate illegal behaviors. Based on this, this study attemtps to investigate the mutually causal relationship between drunk-driving violations and drunk-driving accidents, along with the key factors contributing to the violations and the accidents. To do so, this study first collects a panel data of drunk-driving violations and accidents of 22 cities/counties in Taiwan from January 2007 to December 2016. Granger causality test is used to examine the potential mutually causal relationship between the number of drunk-driving violations and the number of drunk-driving accidents. Afterward, a simultaneous equation regression model is used to further identify the key factors contributing to the violations and accidents based on the causal relationship and the number of lagged months signficnatly tested by Granger causality. The results of Granger causality test show that the Granger causal relationship between the vioaltions and the accidents has significant difference between urban and rural areas. In urban cities (such as Taipei, New Taipei, Taoyuan, Taichung and Tainan), the violations and accidents have a causal interaction with a one month lagged effect. However, the Granger causality between the violations and accidents does not exist in rural cities/counties. The estimated simultaneous equation model shows that the stricter new standards for alcohol levels, effective on June 13 2017, have effectively curtailed the number of drunk-driving accidents. Meanwhile, the number of violation enforcement, increase in public transportation market share, population density, reduce in motorcycle ownership also reduce the number of drunk-driving accidents. As to the number of drunk-driving violations, population density and expenditures in police administration would help to mitigate the violations. Corresponding countermeasures to curtail the number of drunk-driving violations and accidents are then proposed accordingly.