地方財政政策對臺灣縣市長選舉之影響（2001~2018）

Abstract

許多學者的研究及媒體的報導都顯示臺灣縣市政府存在明顯的政治預算循環現象，也就是在選舉年時，縣市長會明顯地增加財政支出及調整財政支出結構，但究竟這樣的操弄是否真能對縣市長選舉有正面助益？本研究以2001年、2005年、2009年、2014年以及2018年臺灣縣市長選舉為研究對象，來探討地方財政總支出、財政政事別支出及財政赤字對與現任縣市長同黨籍候選人得票率之影響。 統計結果顯示，縣市長在任期內增加財政總支出，對於下一次選舉中同黨籍候選人的得票率，效果不具統計上顯著性，表示財政支出具有政治預算循環的現象，卻無法提高下屆同黨候選人當選的機會，而縣市長在非選舉年及選舉年時，增加三個主要財政政事別科目支出時，亦無法高下屆同黨候選人當選的機會。 此外，縣市長在任期內若產生較前任多的財政赤字，對同黨籍的縣市長候選人沒有明顯的影響，不會影響縣市長選舉結果，民眾對於政府赤字增加，可能抱持較中性的看法。補助款收入的資源規模龐大，但統計上亦未達顯著，補助收入的多寡，對於現任縣市長同黨籍候選人之得票率並無影響。 控制變數的部分，經濟社會變數對縣市長選舉的影響並不明顯，地方經濟數據如各縣市實質家戶所得變化額、地方犯罪件數之變化額，對縣市長選舉結果無顯著影響。 政治變數都有達到統計上的顯著性，若縣市長為現任者，對得票率有非常明顯的助益，顯示縣市長的現任優勢很大。此外，縣市長候選人與總統同黨籍有顯著負面影響，與總統同政黨會使縣市長候選人得票率降低，可能是選民會根據中央政府和地方政府的施政表現，進行回溯性投票的結果。

Many scholars have shown that local governments in Taiwan significantly expand their budgets and manipulate the composition of budgets during the election year, which corresponds to the theory of political budget cycle. However, whether such manipulation has a positive impact on local election results remains to be seen. This paper aims to answer the above question by examining the election results in 2001, 2005, 2009, 2014, and 2018. The results suggest that higher public expenditures executed during the whole political term have positive but not statistically significant effect on local elections. Higher spending on economic affairs, social welfare, and higher expenditures for education, science, and culture over the years preceding the elections have not statistically significant effect on local elections. In addition, higher fiscal deficit of local governments have not statistically significant effect on local elections, and we fail to find the evidence supporting the political business cycle in local deficit. On the other hand, the size of grants during the election years has not statistically significant effect on local elections. As for control variables, economic and social variable, including the rate of real family income change, and the change of crime rate have no statistically significant effect on local elections. Political variables have significant impacts on local elections. The incumbent advantage is beneficial to the results of county magistrates and city mayors elections. Additionally, county magistrates and city mayors being from the same party as the president has significant negative impacts on local elections. It might be the consequence of the models of retrospective voting—voters’ decision made after taking into consideration factors like the performance of central government and local government.