Modelling NPLs and identifying convergence phenomenon of banks: a case of pre-during and immediate after the crisis years in India
| DOI | https://doi.org/10.1108/JFRC-09-2020-0083 |
| Published date | 24 August 2021 |
| Date | 24 August 2021 |
| Pages | 1-125 |
| Subject Matter | Accounting & finance,Financial risk/company failure,Financial compliance/regulation |
| Author | Anju Goswami |
Modelling NPLs and identifying
convergence phenomenon of banks:
acaseofpre-duringandimmediate
after the crisis years in India
Anju Goswami
Department of Economics and International Business,
University of Petroleum and Energy Studies, Dehradun, India
Abstract
Purpose –Comparing conventional data envelopment analysis (DEA) model with contemporary Seiford
and Zhu model,this study aims to evaluate the technical efficiency(TE) of Indian banks from 1998/99 to 2016/
17 in the presenceof non-performing loans (NPLs).
Design/methodology/approach –To examine TE, this study has considered a novel approach, i.e. linear
monotone decreasing transformation as suggested by Seiford and Zhu (2002), which treats undesirable output as a
desirable output in the framework of Charnes, Cooper and Rhodes (CCR)-based output-oriented DEA approach. In
particular, to remove the biasness from the estimated efficiency scores, Simar Wilson (1998, Algorithm #1) has been
applied, which is perhaps the first attempt in this kind of literature till now. This study further tries to investigate the
notion of sigma and unconditional
b
-convergence in TE using two-step system generalized method of moments
model in dynamic panel framework.
Findings –Treatment of NPLs using conventional DEAmodel misinterprets the TE scores, while a true
picture emerges when theNPLs are correctly accounted as an undesirable output in banks’loansproduction
process. Efficiency has declinedduring the crisis years, but it recovered immediately after the crisisyears in
India. However, a sudden and steep deterioration in efficiencyscores has been seen from 2013 till the most
recent studyperiod. Public sector banks and old private bankshave reported higher average efficiency scores
than new private banks (NPBs) and foreign banks (FBs) in India. However, FBs are the only commercial
banks that maintained their efficiencylevels during crisis years in India. This study alsosaw the persistence
and presenceof
s
-convergencephenomena in TE for Indian banks, reflecting the abilityto reach up to “Catch-
up”phenomenonowing to the lower dispersion and persistenceof convergence in TE by the Indian banks.
Practical implications –The actual efficiency score can only be estimated when the NPL will be
considered as an undesirable output rather than a desirable output when designing the loan production
process of banks.Although the ownership clusters of all commercial banks in Indianeed to formulate stricter
policies to increasethe level of assets quality and efficiency, but, NPBs need to pay somemore efforts in this
direction. This study’s outcome has the potential to provide useful information for regulators and
policymakers, which suggests that in whichdirection or in which clusters improvement are needed to raise
the level of asset qualityand technical efficiency in the coming years.
Originality/value –For a long time, there has been the existence of trade-offs between researchers, like
which is the best model for accountingfor NPLs? Traditional or contemporary? Thus, our study aims to add
knowledge to the limited stock of NPL modelling in the efficiency literature. Dynamic convergence in TE
scores in Indianbanks has yet not to be tested, which is anothernovelty of the study.
Keywords Credit risk, Banking, Banking crisis, Econometrics, Technical efficiency, Bootstrapped
DEA model, Nonperforming loans, Indian banks, Dynamic convergence analysis
Paper type Research paper
JEL classification –G21, G32, G01, C50
The author would like to thank anonymous referee(s), and Prof. John Ashton, for their useful
suggestion(s) and sharp observations, which helped to improve the quality of the paper markedly.
Identifying
convergence
phenomenon
of banks
1
Received4 September 2020
Revised14 February 2021
12May 2021
Accepted15 June 2021
Journalof Financial Regulation
andCompliance
Vol.30 No. 1, 2022
pp. 1-24
© Emerald Publishing Limited
1358-1988
DOI 10.1108/JFRC-09-2020-0083
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/1358-1988.htm
1. Introduction
An efficient and robust banking sector is necessary for the development of any country as it
provides vital inputs for designing andformulating the macroeconomic policies in general and for
financial system, in particular. In the banks’production process, loans and advances are
considered as major outputs, because they generate revenue. But these outputs sometimes appear
in the form of non-performing loans (NPLs) [1], when borrowers default to make payments as
obligated contractually (Freixas and Rochet, 2008). The bank efficiency literature suggests that
rising NPLs may contribute to the inefficiency in the bank’s performance causing instability
(Zago and Dongili, 2011;Podpiera and Podpiera, 2005;Hamid et al.,2017). Therefore, NPLs
should be incorporated as an undesirable output in the vector of bank’s output for achieving a
meaningful evaluation of its performance. With this, the present study aims to evaluate the
technical efficiency of Indian bank(s) in the presence of increasing levels of NPLs as undesirable
outputs along with other desirable output(s) and input(s).
In recent years, the most prominent development in the Indian banking sector is the
accumulation of hugeamounts of non-performing assets in the accounts of banks. As per the
Central bank’s (i.e. Reserve Bank of India [RBI]) statistics report of end-March 2018, gross
non-performing assets and net non-performing assets for the Indian banking system as a
whole stood at 11.6% and 6.1% oftotal advances, respectively, which is almost sixfold from
the level of 2.3% and 1.1%, respectively, in 2007/2008. The excessive lending to troubled
borrowers, weak creditappraisal system and mismanaged information regarding borrowers
coupled with inadequate banking practices have contributed to delinquencies in loans
repayment (Reserve Bank of India, 2015b). This enormous increase in NPLs raised the
concerns for the policymakers, regulators and other stakeholders of the banking system.
Therefore, policymakershave adopted a multilateral strategy aimedat maintaining stability
and reducing potential negative shocks on Indian banks in the context of improving bad
debts at the bank level, hence immediately after the crisis period they shifted their focus
from de-regulation to re-regulation policies. Nevertheless, more rigorouspolicies are needed
to improve asset quality and efficiencylevels in Indian banks.
Because NPL is a critical factor to determine the bank’sefficiency, still no effort has been
made to incorporate NPLs in the output vector (Sarkar and Bhaumik, 1998;Bhaumik and
Dimova, 2004;Sahoo and Tone, 2009;Tabak and Tecles, 2010). More recently, Das and
Kumbhakar (2012) stressed on the quality of inputs and outputs in productivity
measurement but have not explicitly accounted for the influence of NPLs as undesirable
output in the production process of Indian banks. In particular, for a long time, there has
been the existence of trade-offs between researchers, like which is the best model for
accounting for NPLs? “Traditional or contemporary”? Additionally, existing studies still
lack to cover two important issues, namely,linearity and convexity while formulating linear
programming model for estimation of robusttechnical efficiency scores (Simar and Wilson,
1998;Scheel, 2001). Thus, our study aims to add knowledge to the limited stock of NPLs
modelling in the efficiency literature. Furthermore, researchers have paid very little
attention to investigatingdynamic efficiency and its convergence properties,particularly, in
the presence of NPLs for Indian banks as a whole and separate ownership group, which is
another innovationin the study.
With the aforementioned backdrop, we have formulated the following research
questions: Is the Seiford and Zhu model a more powerfulapproach than the traditional data
envelopment analysis (DEA) model for estimating robust scores of technical efficiency?
Does the accountability of NPLs as bad output affect the technicalefficiency level of Indian
banks? If yes, then to what extent? To what extent does efficiency converge within and
across distinct ownership groups in the Indian banking industry? To answer these
JFRC
30,1
2
To continue reading
Request your trial