運用隨機共同邊界模型比較我國金控與非金控銀行之生產效率--考慮要素內生性問題

Abstract

本研究運用 Huang et al.(2014) 提出之隨機共同邊界模型，比較我國金控與非金控銀行生產效率；採用 Amsler et al.(2016) 的方法解決隨機生產邊界模型有內生生產要素問題，容許無效率項受到外生環境變數之影響，必須另行推導機率密度函數。群組邊界估計結果，發現增設分行不利樣本銀行提升技術效率；民營銀行的效率優於泛公營銀行；金控銀行不支持寧靜生活假說；存放款利差越大的非金控銀行，技術效率越佳；產品多樣化僅對非金控銀行有利。共同邊界模型發現金控銀行無效率主因是欠缺管理能力，非金控銀行主因是採用較差生產技術；樣本期間內生產效率有改善趨勢。

This paper applies the stochastic meta-frontier model, developed by Huang, Huang, and Liu (2014), to compare the production efficiency of Taiwan’s financial holding banks (FHBs) with non-financial holding banks (NFHBs). Furthermore, we solve the problem of endogenous inputs, using the method proposed by Amsler, Prokhorov, and Schmidt (2016), link the inefficiency term with a set of environmental variables, and derive the required probability density function. On the basis of group frontiers, the increase in the number of branches fails to enhance bank’s efficiency, private banks outperform pan-public banks, FHBs reject the quiet life hypothesis, NFHBs with higher net interest margin have higher efficiency, and product diversification improves efficiency of NFHBs only. According to the results from the meta-frontier model, the main source of inefficiency for FHBs comes from the lack of managerial inabilities, while the adoption of inferior technology is the primary source of inefficiency for NFHBs. It is found that production inefficiency lessens during the sample period.