- 學位論文
中國大陸證券業效率與生產力分析
Efficiency and Productivity Analysis of China’s Securities Industry
摘要
本研究目的為分析2006年至2012年中國證券業經營效率以及各年生產力變動。第一階段以「資料包絡分析法」( Data Envelopment Analysis, DEA)衡量證券公司的經營效率，投入變數為固定資產及職工薪酬、產出變數為經紀業務收入、承銷及保薦業務收入、投資收入、以及其他業務。為處理投資收入及其他業務收入的負值問題，選用的DEA模型為SORM-DEA( Semi-Oriented Radial Measure─Data Envelopment Analysis)。生產力的部分則應用「麥氏生產力指數」(Malmquist Productivity Index)分析跨年生產力以及各指數拆解項變動。第二階段使用Simar and Wilson (2007) 所提出的雙重拔靴法截斷常態迴歸(Double Bootstrapping─Truncated Normal Regression)，運用拔靴法(Bootstrap)校正效率值偏誤後建立「斷尾迴歸模型」(Truncated Normal Regression)，試圖找出影響經營效率的因素，迴歸係數的估計亦使用拔靴法建立信賴區間以提升估計的有效性。 第一階段分析結果顯示麥氏生產力指數變動除2007年及2009年為成長，其餘皆呈現衰退的狀態，全期間年平均衰退為11.1%，整體累積衰退幅度達56.1%。股市低迷的2008年及2011年的生產力衰退情形最為嚴重，分別為56.6%及37.7%。指數拆解後發現技術退步為證券業生產力衰退的主要原因，全期間年平均衰退為12.6%，整體累積衰退幅度達62.6%。2008年及2011年則分別衰退61.8%、44.7%。技術效率、規模效率、純技術效率大致平穩不變，僅在2008年及2011年出乎預料的技術效率及純技術效率皆有超過10%的成長。 第二階段迴歸分析結果顯示總資產報酬率、金融資產佔總資產比重、位於京滬深地區，這些解釋變數對於經營效率有正向影響。總資產規模大小以及是否為上市公司與經營效率則不具顯著關係。
並列摘要
This study aims at investigating the efficiency and productivity of the China’s securities industry between 2006 and 2012. First, we use Data Envelopment Analysis (DEA) to assess technical, pure technical and scale efficiencies of each firm and then apply Malmquist Productivity Index (MPI) to estimate the components of productivity change. In order to deal with negative data, a semi-oriented radial measure DEA (SORM-DEA) model that permits the presence of variables which can take both negative and positive values is implemented. Four outputs and two inputs are specified in the efficiency analyses. The outputs of securities firms are divided into four categories: brokerage income, underwriting income, Investment income, and other income. On the input side, Fixed asset and employee salaries are two inputs to be considered. Second, Double Bootstrapping─Truncated Normal Regression (Simar and Wilson, 2007) is applied to examine the determinants of pure technical efficiency. The empirical results of SORM-DEA and MPI analyses indicate that the China’s securities industry is in a recession which is largely due to technical regress. On average, MI decrease 11.1% and technical change decline 12.6% each year between 2006 and 2012 despite CRS efficiency and VRS efficiency are in upward trend. The results of Double Bootstrapping-Truncated Normal Regression demonstrate that the return on assets ratio , financial assets to total assets ratio and firms which are located in Beijing, Shanghai or Shenzhen have positive effects on pure technical efficiency.