Purpose

This paper examines the technical efficiency (TE) of Indian commercial banks during 1998–2015.

Design/methodology/approach

This study uses mathematical programming-based data envelopment analysis (DEA) methodology to measure technical efficiency of Indian banks. Further, Simar and Wilson (2007) double bootstrap procedure is applied to examine the determinants of efficiency of the Indian banks, by examining the effects of various bank specific and other contextual variables.

Findings

The results indicate substantial upward bias in the conventional efficiency estimates of the Indian commercial banks. Needless to note, such upward bias is consistent with the theoretical postulates. The bootstrapped regression results show that increasing capital adequacy ratio is positively associated with bank efficiency. The popular belief that non-performing assets have a dampening effect on performance of banks is validated. Among others, ownership category is observed to be an important determining factor of bank efficiency. Specifically, state-owned banks (SOBs) are relatively lagging behind the foreign banks. Moreover, larger banks are observed to have a significantly higher level of efficiency, therefore, recent official policy initiatives toward consolidation of SOBs are validated.

Originality/value

As this study uses Simar and Wilson (2007) bootstrap approach, it enables the authors to have an estimate of the extent of bias in the traditional DEA TE scores. It also helps us drawing consistent inferences by rectifying the problem of serial correlation in the conventional second stage regression in this regard.

中文翻译：

---

印度银行业的压力资产、表外业务活动和业绩：DEA 双引导方法

目的

本文考察了 1998-2015 年印度商业银行的技术效率 (TE)。

设计/方法/方法

本研究使用基于数学规划的数据包络分析 (DEA) 方法来衡量印度银行的技术效率。此外，Simar 和 Wilson（2007）通过检查各种银行特定变量和其他上下文变量的影响，应用双引导程序来检查印度银行效率的决定因素。

发现

结果表明印度商业银行的传统效率估计存在很大的向上偏差。不用说，这种向上的偏差与理论假设是一致的。自举回归结果表明，提高资本充足率与银行效率呈正相关。不良资产对银行业绩有抑制作用的普遍观点得到了证实。其中，所有权类别被认为是银行效率的重要决定因素。具体而言，国有银行（SOB）相对落后于外资银行。此外，观察到较大的银行具有显着更高的效率水平，因此，最近对国有银行合并的官方政策举措得到了验证。

原创性/价值

由于本研究使用 Simar 和 Wilson (2007) 引导方法，因此作者能够估计传统 DEA TE 分数的偏差程度。它还通过在这方面纠正常规第二阶段回归中的序列相关问题来帮助我们得出一致的推论。