Net Interest Margins of Banks in India

Abstract

Net interest margin (NIM) is the raison d’être of banking. It is an important measure of efficacy of the banking sector. At the system level, it is indicative of the cost of financial intermediation, health of the banking sector and its pricing power. In recent years, the Indian banking sector has experienced a major metamorphosis, with increasing competition and changing norms of liquidity, income recognition and capital. The end of regulatory forbearance and asset quality review (AQR) unearthed significant non-performing assets. This article traces the influence of various factors on the NIM, using bank-level data for 42 Indian banks over 25 quarters from March 2011 to March 2017. The study employs a dynamic panel generalised method of moments (GMM) framework to trace the impact of three distinct set of factors in setting banks’ NIMs: bank-level factors (like the share of low interest-bearing deposits held, the extent of gross non-performing assets [GNPAs], the capital-to-risk weighted assets ratio [CRAR], size of the loan book, operating costs and lending rates); system-level factors such as the monetary policy rate, credit growth and yields on government securities and macro-variables such as GDP growth and inflation. The results indicate that the main determinants of NIMs are banks’ CRAR levels, the proportion of current account and savings account deposits (CASA) to total deposits, operating costs and size of the loan book. Macro-factors like the growth of the economy and repo rate have a positive influence on the NIM.

JEL Classification: C23, E43, G21

Keywords

Net interest margin (NIM)Margins Determinants Commercial banks Panel

1. Introduction

The net interest margin (NIM) of banks is the ratio of net interest income to average interest-earning assets. More specifically, NIM is the difference between a bank’s interest income and interest expense divided by its average interest-earning assets. Globalisation has altered the dynamics of the banking sector in India. The sector has evolved and has adopted international norms and has penetrated various layers of the economy. Bank lending rates assumed greater importance with the strengthening of channels of monetary policy transmission to the real economy. The NIM is an important measure of the efficacy of the banking system and is also indicative of the costs of financial intermediation, a bank’s health and its pricing power.

The central bank influences interest rates in various ways: by altering the cost of funds by changing the policy rate and modulating liquidity in the market. Commercial banks transmit changes in policy rates through their lending and deposit rates. While there is a definite causal relationship between the repo rate and lending/deposit rates, the extent of pass-through of the rate change to various lending and deposit rates is determined by banks and their policies on their NIMs. Banks’ intermediation costs or NIMs determine the cost of credit to the real sector and thus the efficacy of transmission of monetary policy to the rest of the economy, especially in a bank-dominant financial system. In deciding their lending rates, banks take decisions based on factors such as the proportion of current account and savings account deposits (CASA) in total deposits, cost of funds, operating costs and the interest spread that has to be maintained. Decisions regarding the NIM thus determine the extent of transmission of monetary policy to its depositors and borrowers. These bank-level decisions are a crucial ingredient in the banking sector’s willingness and ability to transmit monetary policy signals to the rest of the economy.

The determinants of NIM have varied widely in the recent period following the major metamorphosis in the banking scenario, with increasing competition, changing liquidity, income recognition and capital norms. Also, the end of regulatory forbearance and unearthing of ever-greened loan books have significantly raised non-performing assets (NPAs). How have banks’ NIMs reacted to the rise in NPAs? Do operating costs and capitalisation of banks matter for NIMs? Have these bank-level, banking system-level and macroeconomic factors affected the NIMs of banks? The quest for answers to these questions underlines the theme of the article.

The article contributes to exploration of the determinants of the NIM after the unearthing of large gross non-performing assets (GNPAs) as part of the asset quality reviews (AQRs) conducted by the Reserve Bank of India (RBI) and stringent NPA regulation thereafter. How does this deterioration in credit risk affect the margin behaviour of banks? This article uses the dynamic panel model to examine the impact of the three set of factors on the NIMs of banks and therefore the efficacy of the banking sector. It uses a panel of 42 banks over 25 quarters from March 2011 to March 2017. It applies the Arellano–Bond (AB) models in the dynamic panel generalised method of moments (GMM) framework. The GMM model tries to gauge the extent of influence of bank-level factors such as the size of the loan book, risk weighted assets (CRAR), extent of NPAs and CASA. System-level factors such as the central bank policy rate, credit growth and yields on government securities also influence the NIM. Besides this, the model traces the quantum of impact of macroeconomic factors like GDP growth and inflation.

The rest of the article is organised as follows. The second section gives the theoretical underpinnings of the NIM in the literature. The third section traces the movement of bank NIMs in India. The fourth section focuses on the data and methodology used in the study while the fifth section discusses the empirical results, and the sixth section concludes the article.

2. Literature Review

Literature on the determinants of NIMs is relatively thin. In the area, the Ho and Saunders (1981) model is a pioneering model. This dealership model on US data integrates the hedging and expected utility approaches to analyse bank margins. It is a two-stage model that segregates the influence of bank-level factors and system-level determinants of bank spreads. It postulates that a bank is a risk-averse dealer or intermediary between depositors and borrowers, and its interest margin is a function of factors such as the market structure in which it operates, degree of competition, the extent of its risk-aversion abilities and the interest rate risk it is exposed to. In effect, a bank’s optimal NIM is a mark-up that compensates it for transactional uncertainties it faces due to the stochastic nature of deposits and loan demands. Ho and Saunders’ results indicate that margins are a function of the banking structure, the liability and asset structure of banks, their size and attitude towards risk.

The literature contains different variants of this model applied to different countries and sample periods. Some studies have found that implicit and explicit taxation, level of inflation and market structure affect NIM (Hanson & Rocha, 1986). Demirguc-Kunt and Huizinga (1999) developed the Ho and Saunders model, using cross-country data from 80 countries for the years 1988–1995. Their model goes beyond a mere application of the original dealership model as it takes into account various sets of factors that affect the NIM at the bank level and at the macroeconomic level. The latter includes factors like the GDP of the country, its tax structure (both implicit and explicit), its financial structure, the existence of deposit insurance and legal structure. The authors found a high correlation between interest rates prevailing in an economy and the interest margins of banks. They found that in low-income developing countries, foreign banks enjoyed higher margins than local banks, while the reverse was true in developed countries. They also found that while corporate tax was totally passed on to borrowers, higher reserve requirements were not passed on, especially in developing countries.

Due to wide differences in bank-level factors and the financial environment within which the banks operate, a comparison across countries is a difficult proposition. A variety of factors influence the NIM including differences in institutional structures and the regulatory environment at the bank level. All these need to be accounted for, before proceeding to model the ‘pure spread’. Hence, studies have used two-step models. Saunders revisited this area of research (Saunders & Schumacher, 2000) and used a two-stage model applied to a multi-country framework using data for six European countries and the USA for 1988–1995. Their results show that the NIM depended on a bank’s capital-to-asset ratio, market power and risk aversion. The authors point out that there was a trade-off between assuring bank solvency in the form of a high capital/asset ratio and reducing the cost of financial intermediation. Claeys and Vennet (2008) using a 36-country model covering Central and Eastern Europe examined the determinants of NIMs across these countries. Their results show that capital adequacy and risk attitude were important determinants of bank margins but the most crucial factor was operational efficacy. They also found that NIMs reduced across this region. Their results are in sync with Maudos and de Guevara (2004) who found that NIMs fell during the period due to the lowering of credit risks, interest rate risks and operating costs which are crucial parameters that affect NIMs. The Maudos and de Guevara study used a large panel of banks from European countries for 1993–2000. They identified crucial factors that affected bank NIMs as operating expenses and market power.

Some authors consider lower NIMs to be an indicator of an efficient banking sector (García-Herrero, Gavilá, & Santabárbara, 2009). Higher NIMs adversely affect real savings and real investments in the economy as they are a wedge between what borrowers pay for their loans and what depositors receive. The banking sector’s NIM is also the cost of financial intermediation, so the existence of high-interest margins is considered undesirable as it works against the banking sector leading to disintermediation (Brock & Suarez, 2000). However, not all researchers are in agreement on this as some show that the NIM also depends on a bank’s operations and its health and bargaining/market power. The NIM is defined as a function of a bank’s market power, operational costs, risk aversion and its volume of loans (Hawtrey & Liang, 2008). There is a significant relationship between the health of a bank and its NIM (Brock & Suarez, 2000).

Brock and Suarez (2000) studied the impact of NPLs and operating costs on banks’ NIMs and found that they exerted significant upward pressure on NIMs. The authors also found that global and macro-conditions in the economy and financial cycles affected NIM levels, as they softened NIMs. Studies that used NIMs as a proxy for financial intermediation costs found that banks’ NIMs rose with an increase in the riskiness of the credit portfolios held by banks, lack of competition and an increase in the size of their loan books (Poghosyan, 2012). Poghosyan’s study examined the causes of higher NIMs in lower-income countries and found that NIMs were higher and stayed elevated, as market structures were concentrated and competition was often lacking. Moreover, a large part of the variation in banks’ NIMs was due to variations in bank-level factors. The literature also shows that banks’ quality of assets affects their NIMs. As asset quality deteriorates, banks require greater provisioning and this has to be recouped from borrowers, so banks pass on the credit risk premium to borrowers (Brock & Suarez, 2000; Maudos & de Guevara, 2004).

The NIM affects the profitability of the banking sector and its ability to support growth in the real sector. After the global financial crisis, there was a phase when interest rates in advanced economies remained low, and these low rates were able to support higher asset values and stronger aggregate demand. However, an IMF study cautions that prolonged phases of low interest rates may impair bank margins and thereby bank profitability (Jobst & Lin, 2016). Global events like the occurrence of a financial crisis also affect NIMs (Gunter Krenn, & Sigmund, 2013). Their study used a supervisory dataset of Austrian banks and found that while factors such as the extent of non-interest incomes, competition and macro-factors were responsible for movements in the NIM, global factors like the occurrence of a financial crisis also mattered.

According to Saksonova (2014), the NIM can also be used to gauge the stability of the banking system. She analysed the banking sectors in the Baltic region, Euro area and the USA. Her results show that the NIM mirrored growing tensions and vulnerabilities in the banking sector in a country and that it was the most appropriate parameter for gauging the efficacy of a bank and stability of its operations. She found that it was superior to other parameters like the return on assets and was an indicator of how well the bank managed its interest-bearing assets and liabilities. Considering the NIM as one of the most important criteria for asset structure optimisation, the study used data for Greece and other countries that had volatile interest margins to illustrate how the NIM reflected the state of the banking sector in the country.

Islam and Nishiyama (2016) studied neighbouring countries, India, Nepal, Bangladesh and Pakistan for 1997–2012, based on data from 230 banks in a modified Ho and Saunders’ dealer model. They found that variables like size ownership, asset structure, attitude towards risk and the structure of banking affected banks’ NIMs. They also found that operating expenses, equity and reserve requirements were crucial for determining NIMs in this region and that GDP growth had a negative effect on margins.

Studies in the Indian context (Das, 2013; Sarkar, Sarkar, & Bhaumik, 1998) have considered the NIM to be an indicator of banks’ efficacy. They find that ownership is crucial in determining NIMs. Most studies have also found that bank size was not a significant determinant of NIMs in the Indian context (Kannan, Narain, & Ghosh, 2001). These studies also observe that the existence of higher interest incomes and the extent of NPAs affected NIM. In addition, some studies like Sensarma and Ghosh (2004) found that size was also a crucial determinant of NIM and they also found that banks with higher capital adequacy could command better NIMs.

3. A Backdrop On Movements Of NIMs In India

This section looks at the evolution of bank NIMs alongside the development of India’s banking system. In the era of administered interest rates during the late 1970s and 1980s, banks had very little control on their NIMs because interest rates were fixed both for deposits and for lending. Hence, banks’ NIMs during this period were policy determined. It may be mentioned here that this phase was also marked by a large pre-emption of resources by way of the statutory liquidity ratio and cash reserve ratio which accounted for as much as 63.5 per cent of the net demand and time liabilities at one point of time. These constraints on banks were gradually removed with the liberalisation of interest rates in 1994–1997.

Successive years saw Indian banks gaining a better handle on their NIMs, with more and more autonomy. However, there were also greater demands on their resources with the requirements for more stringent income recognition, asset classification norms and larger provisioning as per Basel standards. NPAs were fairly low till the global financial crisis, rose gradually after the crisis but steadily rose after regulatory forbearance was done away with, in April 2015. In 2015–2016, a wide-scale asset quality review (AQR) was initiated that helped unearth a large number of NPAs. RBI’s multipronged approach to addressing these issues is showing some impact not only in recognising the NPAs but also in taking concrete steps to resolve them. Through these years, the NIMs of banks have seen various changes: from around 3 to 4 per cent in the reform period in 1991, the average rose through the higher margins of foreign banks in India. With financial reforms and deregulation of interest rates, the NIM started decreasing and came to around 3 per cent by 1997, and by 2010, the NIMs of Indian banks were around 2.5 per cent. The literature in other countries shows that banks with a higher market share have better NIMs; however, while Indian public sector banks (PSBs) have a large market share, they have continued to operate with a reasonable NIM, while the new private sector banks have higher NIMs (Figure 1).

Figure 1

Source: Statistical Tables Relating to Banks in India, RBI, various issues, https://dbie.rbi.org.in

The global slowdown and its aftermath adversely affected banks’ asset quality; it also affected their NIMs. Foreign banks, however, continued to have larger NIMs in India than in their home countries. This is in line with the findings of other researchers (Demirguc-Kunt & Huizinga, 1999). Both foreign banks and private banks in India have NIMs which are above those of the PSU banks. In keeping with global trends, the NIMs of PSBs and private banks fell during the global financial crisis while those of foreign banks expanded and remained elevated. Figure 2 compares the NIMs of Indian banks with those in the USA and EU. PSBs’ NIMs are comparable with the average of US banks; in fact, they are lower than US banks almost through the entire period after the global financial crisis.

Figure 2

Source: Federal Reserve Bank of St Louis, https://fred.stlouisfed.org

4. Data And Methodology

This study uses bank-level quarterly data for 42 banks for 25 quarters from March 2011 to March 2017 from various sources. It uses quarterly data on bank-wise advances, NIMs, operating costs and total assets, the policy (repo) rate and lending rates (base rate), CRAR, GNPAs, operating profits and return on assets; quarterly data reported by banks; and macro-data from the Database on the Indian Economy (DBIE), the CMIE database, Banking Statistics and Statistical Tables Relating to Banks in India.

We now formulate our model using three distinct sets of factors that influence the banks’ NIMs, see Table1 for expected signs of the variables in the model.

\begin{matrix} N I M = F [B a n k ​ - ​ l e v e l F a c t o r s, B a n k i n g S y s t e m ​ - ​ l e v e l ​ \\ F a c t o r s a n d M a c r o e c o n o m y ​ - ​ l e v e l F a c t o r s] \end{matrix}

(1)

The appendix table gives results of different empirical models estimated here using the dynamic panel GMM. Alternative equations were useful to be able to include different combinations of determinants.

We use the dynamic panel GMM model using the Arellano-Bond (AB) technique for this estimation (Arellano & Bond, 1991). The GMM estimation has an advantage over fixed and random effect models, as OLS (Ordinary Least Squares) models of this genre suffer from endogeneity and heterogeneity problems. The GMM equation is written as:

Y_{i t} = δ Y_{i, t - 1} + X_{i t} β + u_{i} + ε_{i t}

(2)

where Y_it is the dependent variable, Y_it₋₁ is the lagged dependent variable, X_it are the explanatory variables in the equation, u_i is the individual effect of ith company and ε_it is the error term. Our model is dynamic due to the inclusion of the lagged dependent variable in effort to model the partial adjustment of the dependent variable. As pointed out by Nickell (1981), the existence of a lagged term causes unobserved heterogeneity. To overcome this, we use the ‘AB method’ or the ‘difference GMM approach’. The AB model uses first-difference transformations of equations to get rid of the heterogeneity. The AB model tackles one more issue that arises due to the occurrence of correlation between the error term and the regressor in the process of demeaning of the variables, known as the Nickell bias, which may be sizable when the number of cross-sections (N) is more than the time (T) dimension.

Table 1

Expected Signs of Variables

Dependent Variable Net Interest Margin
Bank-level Factors
NIM (−1)	Net interest margin (previous period)	+
GNPA/GA (%)	GNPA to gross assets (%)	−
CRAR	Capital-to-risk assets ratio (%)	+
Loans and advances	Log (loans and advances, SIZE)	+/−
Lending rate	Base rate/marginal cost of funds based lending rate (MCLR)	+
CASA deposits	Log (current and capital account deposits)	+
Operating expense	Log (operating expenses)	+
Fee income	Log (fee income of the banks)	−
Banking System-level Factors
Repo	Policy rate	+
NFC growth	Growth in credit to non-agricultural sector (growth in non-food credit)	+
G-Sec 10 yields	Yields on govt. securities of 10-year maturity	+/−
Macroeconomic Factors
GDP growth	% Year on year growth in gross domestic product	+
Inflation	% Year on year inflation	+

Source: Authors’ a priori expectations.

The choice of instruments is very crucial for GMM models to obtain consistent estimates. The general rule is that there have to be at least as many instruments as the number of explanatory variables. The instruments are required to be related to the explanatory variables but should not be correlated with the disturbance terms. The AB approach works out the instruments for each of the lags and for each time period separately and uses the GMM model to weigh them.

In models with few time periods and many individual units, a linear functional relationship of the dependent variable that is dynamic depends, inter alia, on its own past realisations (right-hand variables) that are not strictly exogenous. So the AB method is the right model to deal with a number of issues. In this technique, the GMM estimator optimally uses all linear moment restrictions, provided there is no serial correlation of errors in the estimated equation. This technique uses present and past values of strictly exogenous variables for constructing instruments after differencing out the permanent effects (Arellano & Bond, 1991).

5. Empirical Results

The results given in Table A1 are examined here. We performed the robustness checks suggested by Arellano and Bond (1991) for our model. The first robustness check was Sargan’s test statistic for overidentifying restrictions, computed to see if overidentifying assumptions hold in instrumental estimations. This test should be rejected for the model to be good. The second test—serial correlation testing of models estimated by GMM—is crucial for the consistency of the estimator. It depends on whether the lagged values of the variables can be used as instruments in the model and for this there should be no second-order serial correlation as proposed by Arellano and Bond (1991). Both AR(1) and AR(2) are two different test statistics and the requisite outcome is that the first-order statistic is required to be ‘negative and significant’ (with a negative autocorrelation coefficient) while the next test of AR(2) is required to be insignificant. Our model equations satisfy all the requisite conditions (refer Table A1 for the results of the empirical models).

5.1. Bank-level Factors

Past values of the NIM: The previous-period NIM has a positive effect on the current-period NIM due to the legacy factor, and NIMs adjust gradually as indicated by a significant coefficient of lagged NIMs. This is acceptable due to the fact that the banks can adjust their NIMs only gradually given the contractual nature of saving and fixed-term deposits as well as some fixed rate loans. Moreover, even floating rates have a reset period of at least a quarter.

Quality of loan books: Non-performing loans is the main reason why ex ante NIM does not translate into an actual NIM. As expected, a priori, an increase in the level of default risks requires banks to make more provisions for these. Our results show that as the GNPA/GA percentage rises, the NIM rises to compensate for the rise in provisioning requirements. Banks try to raise their NIMs and pass on the higher cost of provisioning to their customers as indicated by a positive and significant coefficient of GNPA/GA. Higher levels of NPAs inflict banks with additional costs, but banks try to protect their NIMs through other means, like lower transmission of rate cuts and greater transmission of rate hikes. NIMs increase with the riskiness of banks’ credit portfolios (Poghosyan, 2012). The literature corroborates that banks do in fact pass the credit risk premium to borrowers (Brock & Suarez, 2000; Maudos & Guevara, 2004).

CRAR: Our study finds that well-capitalised banks are considered safer and pay lower interest on their deposit liabilities which in effect supports their NIMs. We get a positive relationship between the CRAR and NIMs, as customers regard a higher CRAR as an indicator of the strength and stability of a bank; being well capitalised helps banks instil confidence and make depositors/bond holders accept a lower interest rate, thus helping boost the NIM. Hence, there is a positive relationship between the CRAR and NIM. Sensarma and Ghosh (2004) also found that banks with higher CRARs could command a higher NIM.

Operating costs have a significant role in determining banks’ NIMs. Our model gives us significant evidence of the fact that banks with higher operating costs need to beef up their NIMs. Operating expenses in turn are a function of a number of factors including the productivity of the bank staff, extent of overheads and systems at the bank level and their efficacy. Our results indicate that operating costs push up NIMs. Maudos and de Guevara (2004) also found that operating expenses were crucial in the process of determining the NIM.

Current account and savings accounts deposits: It is a well-accepted tenet in banking that retail deposits make the funding of banks more stable and offer the bank a low-cost source of funds that boosts its NIM. A wider deposit base garners large saving bank deposits that bear low interest rates, and the current accounts pay nothing on the float. Both are low-cost deposits, and banks use these funds for lending at the base rate or MCLR that is much higher than the deposit rate. Hence, higher the share of CASA in total deposits, the higher is the NIM. Our results confirm this a priori expectation and show that the NIM is supported by a higher CASA ratio or a higher proportion of low-cost CASA. This is indicated by a significant coefficient of the log (CASA) variable in the equation.

Size of the bank’s loan book: The size of the loan book is taken as a variable reflecting the size or market power in many international studies, while some studies observe that it impacts the NIM significantly and in the reverse direction. In India, large banks have the ability to raise deposits at lower rates as they have the reach, power, credibility and ability to mobilise deposits. However, many of the larger banks are in the public sector and they do not generally use this market power to extract a higher NIM. They use their size to spread their overheads over their larger base, so size is observed to affect their NIMs inversely. Our results are in line with Kannan et al. who find that in India size is not an important determinant of the NIM.

Fee-based income: Ideally, banks with a larger fee-based income can depend less on their NIMs. In recent years, banks’ fee-based income has increased and our study tried to see whether this has had any impact on their NIMs; the variable was observed to be insignificant.

5.2. Banking System-level Factors

Policy rate: The NIM moves in tandem with the repo rate and our results support this. We observe that an increase in the policy rate raises NIMs, because when there is a rise in the repo rate, banks can easily transmit the higher interest rate to their asset side (lending portfolio, for fresh lending) well before they adjust it on their liability side (i.e., pay higher interest on the deposits they hold). When the repo rate rises, banks are able to raise their margins by raising lending rates before they raise their deposit rates. So, an increase in the repo rate affects NIM positively.

G-Sec yields: We tested the sensitivity of NIMs to yields on government securities. The variable was found to be significant but the coefficient was very small, probably because of the fact that to the extent that the government securities are in HTM category they are shielded from the impact of the yield changes. The impact of the yield increases is observed only for the securities that are not in the held to maturity (HTM) category in which case, the bank had to mark to the market and suffer losses. So, the net impact depends on the proportion of government securities held in different categories-HTM, held for trading (HFT) or available for sale (AFS).

Credit cycle: Overall, credit growth in the economy is a crucial factor for banks; when there is high credit demand, we expected the NIMs to go up. We tried to gauge the impact of system-level credit on the NIM of individual banks. The variable we have used here is the growth in bank credit to non-agricultural activities (this variable is called non-food credit [NFC]). The impact of the NFC on the NIM was observed to be insignificant. So the NIM is relatively unaffected by system-level credit growth.

5.3. Macroeconomic Factors

GDP growth: The NIM was found to improve with higher GDP growth. The sign was as per a priori expectations and it was found to be significant. During a boom or an upcycle phase of the GDP, the banks are in a position to expand their net margins, as the borrowers expecting better profits continue to borrow even if lending rates rise. This is explicable as when economic growth is picking up, banks are able to lend at relatively higher rates and hence are able to expand their interest margins due to increased business optimism. Hence, our results were expected and show that NIMs are positively related to the GDP growth rate.

Inflation: As expected, the inflation variable exerted an upward pressure on the NIM, as is evident from the positive and significant coefficient of the inflation variable in Table A1. This is in line with results in other developing countries where the real rate of interest received by depositors sometimes does not keep pace with inflation, while borrowers face higher interest rates. During inflation, NIMs tend to rise gradually (note the small dimension of the coefficient of inflation).

6. Concluding Observations

Using bank-level data, we found that NIMs were influenced by macroeconomic factors such as growth and inflation, and system-level factors such as the policy rate and yields on government securities. They were also influenced by a number of bank-level factors, such as the proportion of CASA to total deposits, operating costs, GNPAs, banks’ CRAR and the size of banks’ loan books. In studies on other countries, higher market share invariably translates into a higher NIM. In India, larger banks do not necessarily have larger NIMs because many of them are in the public sector and do not use their market power to extract higher NIMs; they use their size to spread their overheads over their larger base as indicated by a negative relation between the size of the bank and its NIM.

It is often said that in developing countries, NIMs are inflated and affect savings and investments in the real economy, but our study found that Indian banks’ NIMs were fairly comparable with NIMs in advanced economies, especially with banks in the USA. In the Indian context, we observed that it was the quantity and the quality of bank capital which influenced its NIM level. Our empirical results indicate that the most important determinants of NIMs are the extent of a bank’s CASA, CRAR and operating costs, while macroeconomic factors such as GDP growth and repo rate also have a positive influence on the NIM. An increase in NPAs puts pressure on banks to increase their NIMs to cover rising provisioning costs. With the resolution efforts that are currently being undertaken, overtime the extent of NPAs and the provisions required are likely to decrease and this is likely to bring down the cost of intermediation.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

Footnotes

Acknowledgements

The authors acknowledge support for data reorganisation from Onkar Swami, AGM, DBR and Venkat Sreedhara, Manager, DBR RBI. The usual disclaimer applies.

Appendix

Table A1

Dependent Variable: NIM, Method: Panel Generalised Method of Moments Method: Arellano–Bond Transformation: First Differences, Sample (adjusted): 2011–2017 Cross-sections included: 42 Periods included: 25 Quarters, Total panel (unbalanced) Observations 920
	EQ1	EQ2	EQ3	EQ4	EQ5	EQ6	EQ7	EQ8	EQ9
Bank Level Factors
NIM(−1)	0.6563	0.6561	0.6561	0.6691	0.6707	0.6837	0.6547	0.679	0.6773
	(21.71)	(21.59)	(21.69)	(21.44)	(21.59)	(22.04)	(21.59)	(28.50)	(22.68)
Gnpaga per	0.0172		0.0182	0.0143	0.0143	0.0094	0.0179	0.0087
	(4.58)		(4.81)	(3.41)	(3.49)	(2.28)	(4.78)	(2.81)
CRAR	0.0111	0.0116	0.012	0.0114	0.113	0.0157	0.0121		0.0136
	(3.34)	(3.50)	(3.58)	(3.41)	(3.10)	(4.51)	(3.61)		(3.93)
log loansadv								−0.276
								(−14.82)
l (CASA)		0.0287
		(2.05)
log Opexp				0.0336	0.336			0.7138	0.0657
				(3.00)	(3.00)			(7.04)	(5.53)
log feeinc					−0.0051
					(−0.01)
Banking System-level Factors
Repo	0.0702	0.0729	0.072	0.0722	0.0722	0.1045	0.07
	(5.63)	(5.79)	(5.75)	(5.20)	(5.20)	(7.19)	(5.61)
NFC growth				0.0035
				(1.58)
G-sec yield			0.0003						0.0004
			(3.17)						(2.53)
Macroeconomic Factors
GDP growth							0.0087	0.0177	0.0114
							(1.95)	(4.25)	(2.39)
log GDP						0.4826
						(4.61)
Inflation									0.0081
									(2.07)
Robustness Checks
AR(1)	−9.72	−9.56	−9.82	−9.73	−9.55	−9.99	−9.73	−10.62	−9.83
p -Value	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
AR (2)	−0.16	−0.31	−0.31	−0.41	−0.41	−0.06	−0.25	−0.79	−0.69
p -Value	0.87	0.74	0.75	0.63	0.68	0.95	0.81	0.42	0.48
Sargan p	0.34	0.26	0.26	0.26	0.26	0.29	0.19	0.26	0.25

Data Definitions
NIM	Net Interest Margin
Nim(–1)	Lagged Net Interest Margin
Gnpagaper	Gross NPA to Gross Assets percentage
CRAR	Capital to Risk Asset Ratio
log(Total Assets)	Log of Total Assets
log(CASA)	Log of Current Account and Saving Accounts Deposits
log Opexp	Log of Operating Expenses
Logfeeinc	Log of Fee Income
log loansadv	Log of Loans and Advances
Repo Rate	Policy rate announced by the RBI
NFCgr	Non-Food Credit Growth
Tenyr(yield)	Yield on 10 year Government Securities

GDPgr	GDP Growth rate
lgdp	Log of GDP
Inflation	Inflation, CPI combine
lGDP	Log of GDP

Source: The authors.

Note: Figures in brackets are ‘t’ values.

References

Arellano

, & Bond

(1991). Some tests of specification for panel data: Monte-Carlo evidence and an application to employment equations. The Review of Economic Studies, 58(2), 277–297.

Brock

P. L.

, & Suarez

R. L.

(2000). Understanding the behaviour of bank spreads in Latin America. Journal of Development Economics, 63(1), 113–134.

Claeys

, & Vennet

V. R.

(2008). Determinants of bank interest margins in Central and Eastern Europe: A comparison with the West. Economic Systems, 32(2), 197–216.

Das

T. B.

(2013). Net interest margins, financial crises and bank behaviour: Experience of Indian banks (RBI Working Paper Series No. 10). New Delhi: RBI.

Demirguc-Kunt

, & Huizinga

(1999). Determinants of commercial bank interest margins and profitability: Some international evidence. World Bank Economic Review, 13(2), 379–408.

García-Herrero

, Gavilá, S., & Santabárbara

(2009). What explains the low profitability of Chinese banks? Journal of Banking and Finance, 33(11), 2080–2092.

Gunter

, Krenn

, & Sigmund

(2013). Macro-economics, market and bank-specific determinants of net interest margins in Austria (Financial Stability Report). Vienna: Oesterreichische Nationalbank.

Hanson

J. A.

, & Rocha

R. de R.

(1986). High interest rates spreads the costs of intermediation: Two studies (Industry and Finance Series, 18). Washington, DC: The World Bank.

Hawtrey

, & Liang

(2008). Bank interest margins in OECD countries. North American Journal of Economics and Finance, 19(3), 249–260.

10.

T. S. Y.

, & Saunders

(1981). The determinants of bank interest margins: Theory and empirical evidence. Journal of Financial and Quantitative Analysis, 16(4), 581–600.

11.

Islam

M. S.

, & Nishiyama

(2016). The determinants of bank net interest margins: A panel evidence from South Asian countries. Research in International Business and Finance, 37(C), 501–514.

12.

Jobst

, & Lin

(2016). Negative Interest Rate Policy (NIRP): Implications for monetary transmission and bank profitability in the euro area (IMF Working Paper, 16/172). Washington, DC: IMF.

13.

Kannan

, Narain

, & Ghosh

(2001). Determinants of net interest margins under regulatory requirements: An econometric study. Economic and Political Weekly, 36(4), 337–344.

14.

Maudos

, & de Guevara

J. F.

(2004). Factors explaining the interest margin in the banking sectors of the European Union. Journal of Banking and Finance, 28(9), 2259–2281.

15.

Nickell

(1981). Biases in dynamic models with fixed effects. Econometrica, 49(6), 1417–1426.

16.

Poghosyan

(2012). Financial intermediation costs in low-income countries: The role of regulatory, institutional, and macroeconomic factors (IMF Working Paper, WP/12/140). Washington, DC: IMF.

17.

Saksonova

(2014). The role of net interest margins in improving banks’ asset structure and assessing the stability and efficiency of their operations. Procedia—Social and Behavioral Sciences, 150, 132–141.

18.

Sarkar

, Sarkar

, & Bhaumik

S. K.

(1998). Does ownership always matter? Evidence from the Indian banking industry. Journal of Comparative Economics, 6(2), 262–281.

19.

Saunders

, & Schumacher

(2000). The determinants of bank interest rate margins: An international study. Journal of International Money and Finance, 19(6), 813–832.

20.

Sensarma

, & Ghosh

(2004). Net interest margins: Does ownership matter? Vikalpa, 29(1), 41–47.