NATIONAL INSTITUTE OF BANK MANAGEMENT

FINAL PROJECT REPORT

TOPIC: Macroeconomic, bank-specific, regulatory and

institutional factor that influence the formation of credit risk in
Indian Banking Industry and degree of persistence in credit risk
in Indian banking industry

By
Komal Chaudhary
Roll No.- R1801034
PGDM (B&FS) Batch 2018-20

Guide:
Dr. Richa Verma Bajaj, NIBM
 INTRODUCTION

The global financial crisis of 2007/08 has not only revived the interest of economists in the
consequences that a banking crisis can have on the economy, but also stimulated them to look
again at the factors that may trigger a crisis. It is an established view that increasing credit risk
(as widely measured in the literature by ex post loan losses)1 due to the accumulation of a huge
pile of non-performing loans (NPLs) or impaired loans generally tends to increase the likelihood
of a banking crisis.
In the present study, I intend to seek the answer of a key research question: What drives credit
risk in the Indian banking industry? The motivation of the present study stems from the fact that
the credit quality of Indian banks has deteriorated substantially in the post-crisis era. This is
evident from the official statistics released by the Reserve Bank of India (India’s central bank)
that the ratios of gross and net NPLs to total advances rose to 4.4 percent and 2.4 percent
respectively in the year 2014/15, which is almost double relative to 2.3 percent and 1.1 percent
as observed in the year 2007/08. This compelled me to look into the factors that cause loan losses
in the Indian banking industry. In particular, the objective of this research is to identify the key
determinants that drive credit risk in the Indian banking industry over the period from 2005/06 to
2017/18. The extant literature highlights that, along with the regulatory and institutional factors,
there are a wide range of bank-specific, industry-specific and macroeconomic factors that
influence the formation of bank credit risk.
The findings of this study are of immense significance for regulatory and supervisory authorities
concerned with bank stability in India, since the study not only identifies the key systematic
(external) and unsystematic (internal) factors that might be responsible for the formation of
credit risk in the banking industry, but also quantifies the degree of persistence in the occurrence
of loan losses. Furthermore, Indian banks’ NPLs as a percentage of gross loans have consistently
been found to be far above the levels seen in other Asian economies. Overall, the present study
aims to help regulators and bank managers in identifying the possible factors that may lead to a
deterioration in credit quality and increase the burden of default risk.

METHODOLOGICAL FRAMEWORK

In order to explore the key determinants of credit risk along with testing the degree of persistence
in credit risk in the Indian banking industry, the present study employs a dynamic panel
estimation method based on the generalized method of moments (GMM).

Two-step system GMM estimation approach

The literature suggests that panel models with a lagged regressor suffer from problems of
autocorrelation as individual fixed effects correlate with the error term, which may lead to the
problem of endogeneity. Under such circumstances, the traditional panel estimators (pooled OLS
and fixed or random effects estimators) are seriously biased. Thus, the system GMM retains
more information and reduces the potential biases and inaccuracies associated with the
traditional panel and first-difference GMM estimators, especially for short and unbalanced panel
data, and when the explanatory variables have strong persistence. Further, system GMM
estimator(s) are asymptotically more efficient and robust to heteroscedasticity and second-order
autocorrelation.
DATA, VARIABLES AND TESTABLE HYPOTHESES

1. DATA

The dataset used in the present study consists of an unbalanced panel of commercial banks that
were actively operating in India during the period 2005/06 to 2018/19. In addition to an
aggregate analysis of the Indian banking industry, I conduct the analysis in a disaggregated
manner by classifying the banks in accordance to their ownership. The required bank-level data
have been culled from the annual editions of the ‘Statistical Table Relating to Banks in India’,
which are available on the official website of the Reserve Bank of India (RBI). The real GDP
growth rate at 2010 constant prices (in percent) and the inflation rate (in percent) for each sample
year have been obtained from the World Bank Database. The data on real effective exchange
rates (REER) are sourced from the Bank of International Settlements, calculated as geometric
weighted averages of bilateral exchange rates adjusted by relative consumer prices from 2010
(with 61 economies included in the basket). An increase in REER indicates an appreciation in the
nation’s currency.

2. VARIABLES AND TESTABLE HYPOTHESES

Macroeconomic variables

Real GDP growth rate: In the literature, the real GDP growth rate is used to control the effect of
the macroeconomic environment. I hypothesize a negative relationship between the real GDP
growth rate and credit risk. The justification for the negative relationship is that growth in GDP
entails a higher income level, which improves debt servicing in the economy. An improvement
in debt servicing leads to a reduction in NPLs.

Inflation rate : The empirical literature provides no clear impact of the inflation rate on credit
risk. Therefore, I am not sure about the sign of the coefficient of this variable.

Real effective exchange rate: The REER has been used to reflect the transmission of external
currency shocks to credit quality in the domestic economy. There are two strands of literature
establishing the link between REER and credit risk. One argues that an appreciation of the
domestic currency (as indicated by the value of REER index beyond 100) brings a dip in exports,
which reduces the debt repayment capacity of firms in export-oriented sectors. Thus, an increase
in international competitiveness contributes to an increase in bank defaults. The other strand tries
to model the impact of currency depreciation on asset quality. In situation of a weakening of the
exchange rate vis-a-vis the foreign currency, the banks are exposed to severe credit risk when
firms borrow in a foreign currency and do not hedge currency risk. Similar to inflation, I expect
either a negative or a positive effect of REER on credit risk.

Bank-Specific Variables

Bank profitability: Bank profitability is measured by return on assets (ROA), which measures the
profit a bank can generate given total assets. A higher ROA indicates better profit prospects for
growth and resilience to shocks, and should thus be associated with lower credit risk. Highly
profitable banks have fewer incentives to engage in high-risk activities. Thus, bank profitability
is correlated negatively with default risk. I expect a negative impact of bank profitability on
credit risk.

Income diversification: The bank’s total income is the sum of interest and non-interest income.
In recent years, the significance of non-interest income as a source of bank income has increased
tremendously. This is evident from the fact that many banks have diversified and moved into
non-traditional activities (like investment banking, asset management and insurance
underwriting, feepaying and commission-paying services, trading and derivatives) that earn fees
rather than interest. It is claimed that more diversification in the bank’s business model improves
loan quality and reduces credit risk. In the present study, the ratio of non-interest income to total
income is used as a measure of income diversification. I hypothesize a negative impact of
income diversification on credit risk.

Credit growth: Credit growth has been considered an important determinant of credit risk.
In order to obtain new business, banks reduce loan rates and ease credit standards, which
eventually lead to a rise in loan losses. A priori I hypothesize a positive relationship between
credit growth and NPLs. I use the loans-to-assets ratio as a proxy for credit growth. It is worth
noting here that this ratio also reflects liquidity risk since loans are less liquid and riskier but
havea greater expected return than other assets, like government securities in banks portfolio.

Bank Size: Bank size (SIZE) is proxied as the natural logarithm of a bank’s total assets. The
empirical literature does not provide clear-cut evidence about the relationship between bank size
and credit risk. Bigger banks might be better able to control NPLs due to the adoption of better
risk management systems and procedures, which lowers the default rate by ensuring a proper
screening of loan applications. In addition, a bigger size of the bank allows for more
diversification opportunities, which lower credit risk. Following this view, one can expect a
positive effect for leverage on NPLs conditional on size.

Cost (in)efficiency: In the literature, the link between credit risk and cost (in)efficiency is
obscure. Cost efficiency is generally captured by the ratio of operating expenses (i.e., non-
interest expenses) to total assets. After the loans become past due or non-accruing, the bank
begins to expend additional managerial effort and incurs higher costs in order to monitor these
problem loans. Thus, soaring problem loans lead to higher levels of cost inefficiency. Managers
with poor skills in credit scoring, appraisal of pledged collaterals and monitoring and controlling
borrowers after loans are issued do not sufficiently control their operating expenses and give out
poor quality loans. This hypothesis suggests that a bank that seeks to enhance cost efficiency
as a way of maximising long-run profits may decide to cut down the cost of operations in order
to be cost-efficient in the short run by skimping on the resources allocated to loan screening and
monitoring (underwriting cost). In other words, the skimping strategy may result in the bank
experiencing greater bad loan problems in the long run.

Dummy Variables

Financial crisis: In order to see whether the recent global crisis of 2007-09 has an adverse impact
on the credit risk of Indian banks, we use a dummy variable FINCRISIS with a value of 1 for the
year 2007/08 and 2008/09 and 0 otherwise.
EMPIRICAL RESULTS

Persistence effect
In order to see the persistence effect of credit risk, we should focus on the estimated coefficient
of the lagged dependent variable in a model specification. A value of ˆ between 0 and 1 implies
persistence of credit risk. First, the coefficient is positive and statistically significant in all the
model specifications. This clearly provides evidence of time persistence in the accumulation of
bad loans in the Indian banking industry. The magnitude of coefficient in different model
specifications suggests that if there is a shock to NPL levels in the current year then about 40 to
60 percent of its effect will persist in the following year. Overall, the results indicate that the
recovery of NPLs in the Indian banking industry is not instantaneous, and Indian banks retain a
significant portion of their bad loans from year to year. Second, the degree of persistence is
higher for the ratio of gross NPLs relative to net NPLs across different specifications. The
relatively low magnitude of persistence in net NPLs could perhaps be due to an improvement in
the credit appraisal process, greater transparency, new legal initiatives aimed at faster NPLs
resolution, and greater provisions and write-offs, which were enabled by greater profitability
levels in the Indian banking industry during the study period.

Bank-specific effects
Following are the interesting observations regarding the bank-specific determinants of credit
risk. First, as expected, we note a consistently negative and significant relationship between
ROA and credit risk. This suggests that, if the profitability of Indian banks increases, they
engage in more prudent lending with more careful screening and monitoring of borrowers, which
may lead to a reduction in the risk of defaults. Second, the sign of the coefficient of the variable
SIZE in a sense runs contrary to our a priori expectations. The positive effect (though not very
strong in terms of statistical significance of the coefficients) of bank size on credit risk suggests
that large banks take excessive risks by increasing their leverage under the “too-big-to-fail”
presumption and therefore have more gross NPLs. This reflects that large banks take excessive
risks and extend their credit without proper screening and monitoring of the borrower’s
creditworthiness. Third, there is surprising evidence indicating that greater diversification (as
proxied by the NONIT) is insignificant and so influencing bank credit risk negatively does not
take into consideration. Fourth, as expected, the relationship between credit growth and GNPLs
is found to be negative and significant in model specification. The key reason behind this finding
is that the deterioration in credit quality has followed a lagged cyclical pattern with regard to
credit growth in the Indian banking industry (Reserve Bank of India, 2010). It is noteworthy here
that during the period 2005–2007, commercial banks adopted a liberal credit policy and made
aggressive lending to stressed sectors (like infrastructure, coal mining and aviation, etc.).

Macro-economic effects
The findings of this study suggest a low probability of risk of default during periods of inflation
in India. This may be due to adjustments in policy rates by the Reserve Bank of India as a step to
contain inflation, which reduces the real value of outstanding loans and makes debt servicing
easier for borrowers. In this study the coefficient estimates of GDP have not shown any
significant impact of economic activity in the formation of credit risk during the period. The
changes in the real effective exchange rate show a significant negative impact on gross and net
NPLs ratios, implying that the depreciation of the Indian rupee goes along with an increase in the
risk of defaults. This supports the view that a worsening of the local currency makes it difficult
to serve debt denominated in a foreign currency, and thus worsen the fragility of the banking
system to the next level.

Dummies effect
The dummy variable for financial risk has a significant positive impact on the credit risk in the
Indian banking system. Thus shows that the recent global crisis of 2007-09 has an adverse
impact on the credit risk of Indian banks.

Variables Definition Expected GMM

Sign
Real GDP Growth Real GDP growth rate (at (-) (-)
Rate 2011 constant prices)
Inflation Annual inflation rate (+/-) (-)
Real effective Annual real effective (+/-) (-)
exchange rate exchange rate
Return on Assets Ratio of return to average (-) (-)
assets
Non-interest income Ratio of non-interest (-) (-)
income to total assets
Credit growth Ratio of loans to total (+) (-)
assets
Size Log of total assets (-) (+)
Cost (in)efficiency Ratio of operating (-) (+)
expenses to total assets
Financial crisis 1 for the period 2007- (+) (+)
2009 and 0 otherwise

CONCLUSION

In all, the empirical findings suggest that both systematic (macroeconomic) and unsystematic
(bank-specific) factors explain the formation of credit risk in the Indian banking system. In
addition, greater market concentration, more diversification, higher credit growth, and a large
size of banks increase the probability of defaults in the Indian banking industry. To deal with the
problem of enormous and rising bad loans in banks, attention on bank-specific factors is a must.
To begin with, the regulators may focus on the resolution of NPLs in two or three big banks by
providing sufficient additional capital to write down NPLs to appropriate levels that ensure their
financial stability. This is because a big bank failure has more serious repercussions and is likely
to generate a disastrous effect on the real economy.