The Impact of Competition on the Profitability and Risk-Taking of Commercial Banks in India

Abstract

The purpose of this article is to investigate the impact of competition on the performance of Indian banks. The survey includes banks from both the public and private sectors. The study will collect data for four years, from 2015 to 2019. Dynamic and static panel data are applied to estimate the association between competition and the bank's performance. Profitability and risk-taking are the performance measures used in the study. The study's main findings are that competition does not impact the banks' profitability in India. However, the findings concerning risk-taking are mixed. Therefore, it can be inferred that overall competition does not impact the banks' performance in India. Other measures of performance of the banks could have been used in the study. It is a limitation to use data of four years. Data for a much more extended period could have also been used. This is one of the few papers on the subject. Therefore, its contribution is very significant. The gap in studies on the topic of competition versus performance of the banks is veritably filled by the current study's findings.

1. Introduction

Competition in banks is one of the areas, which attracts scrutiny from academia and practitioners alike. Discussion on competition in banks is partly started by Keeley (1990). However, he raises the issue in the context of deregulation in the US during the 1970–80s. Since then, competition has been in vogue, and it is part of the ongoing discussion of its significance in the banks. Furthermore, Allen and Gale (2004) try to find its association with efficiency and financial stability, but the results are inconclusive. Agoraki et al. (2011) discussed competition and regulation and linked them with risk-taking in banks. Jiménez et al. (2013) also explored competition and risk-taking (risk-taking is linked with non-performing assets) and found that low competition or high concentration of market power reduces the risk. The common thread in all such studies is competition, which is also the opposite of banks’ market power or concentration.

The performance of banks is not as transparent as bank rivalry. A large body of research exists that examines the bank’s performance in several areas, such as accounting, efficiency, risk-taking, and stock-market performance (Athaley et al., 2020). Some combinations of regulation, profitability, and risk-taking are also routinely taken into account in a larger definition of bank performance (Rastogi et al., 2021; Agoraki et al., 2011; Triki et al., 2017). The two primary indices of bank performance used in this paper are profitability and risk-taking. We found that these two measures are linked to the fundamentals of bank performance, while other measures such as efficiency, valuation, and stock-market performance are second-stage performance indicators.

Literature is replete with evidence of utter confusion on the role of competition for banks. There is no consensus on this issue, and it is still part of the ongoing debate (Allen & Gale, 2004). Jiménez et al. (2013) present in their study that competition and financial stability are adversely associated with one another. In contrast, Allen and Gale (2004), in their seminal work on the topic, find the issue inconclusive. Fungáčová and Weill (2013), in their study on Russian banks, found that competition is the root cause of bank failures. Boyd and De Nicolo (2005) presented that banks’ risk-taking grows as a result of competitiveness.

Au contraire, another set of studies find the competition is good for the bank, however with some riders. Barra and Zotti (2017) and Schaeck et al. (2009) emphasized that competition injects financial stability into banks if the institutions have market power (or concentration banking prevails). The concern of Schaeck et al. (2009) for concentration banking is also supported in the literature (Tabak et al., 2015; Beck et al., 2006; Ariss, 2010; Kiran & Jones, 2016). However, the positive role of concentration banking to fetch the sound effects of competition for financial stability is not supported by Kasman and Kasman (2015). Instead, the findings of Kasman and Kasman (2015) confound the role of concentration banking because, unlike the other studies, they argue that competition and concentration in banking have a negative impact on bank financial stability. The issue is further beclouded when many studies find evidence that efficiency in the banks is reasonably increased by competition (Arrawatia et al., 2015).

The prevailing confusion regarding competition and its effect on the banks (Allen & Gale, 2004) is the primary motivation for the study. This study aims to clarify the problem of competition and its significance for Indian banks. Profitability and risk-taking, the authors feel, are the best indicators of a bank’s performance. The competition is empirically assessed against both the bank’s performance measures to help provide a concrete view of the ongoing debate on competition and its relevance for the banks. Disclosures and the efficiency of the banks are used as the controlling variables. Therefore, both have no effect on the impact of competition on the bank performance (measured by profitability and risk-taking).

The current work makes two important contributions to the existing body of knowledge on the subject. Firstly, the profitability of Indian banks is unaffected by competition. Secondly, competition does not adversely affect risk-taking contrary to popular belief in the literature for the same. Both the contributions have significant implications for the Indian banking sector as unplugging the two (profitability and risk taking from competition) would open vistas of opportunities and challenges simultaneously.

The remainder of the paper is organized as follows. The first section is an introduction, which is followed by a discussion of literature and the formulation of hypotheses in the second section. The third portion gives the theoretical formulation of the subject. The fourth section of the paper goes over the data and methods used to achieve the paper’s goal. The data analysis results are provided in the fifth section, followed by a discussion of the findings in relation to previous studies in the sixth section. The seventh and final sections of the document bring the paper to a close.

2. Literature Review and Hypotheses

Literature is replete with examples that competition, in totality, does not help at all. In bits and pieces, it may help some banks but hurt others, and in sum, it is an economic loss in general. Boot and Thakor (2000) found evidence to endorse this point of view. Allen and Gale (2004), having gone through more than one model, also concluded that it is essentially difficult to fix whether competition adds value or depletes because such generalization may not be practically possible. In reality, it is like that only. Cetorelli (2001) found evidence while going through the literature survey opines that neither extreme no competition nor complete competition is good for the banking sector. Furthermore, he endorses, contrary to popular belief, that competition in the banking sector does more harm than good and subsequently adversely impacts economic activities. Alam et al. (2019) manifested that excessive competition may lead to issues of solvency. Moreover, there are many instances where it is seen that competition leads to bank failures as well (Martinez- Miera & Repullo, 2010).

Keeley (1990) was the first to suggest that competition could be harmful to banks. He claims that competition erodes the banks’ charter value (Acharya, 1996). As a result, the risk of default increases, and bank capital decreases. Following this, Agoraki et al. (2011) extended Keeley’s (1990) notion, link it to regulation, and argued that competition (which he refers to as market power) may raise credit risk and the possibility of default. Hellmann et al. (2000), in their seminal work on the topic, reinforced the point that competition induces poor bank behavior and decision making. On the contrary, the other determinants and factors of bank performance (mainly regulation for the capital) support prudent bank decision-making.

2.1. Competition and Profitability in Banks

As we switch to talking about the effect of competition on profitability from a general discussion on competition in the bank, we realize that the literature on the topic is entirely inadequate. It is a less researched area. Competition is always considered suitable for any firm, especially from the efficiency and customer-centric approach, albeit in theory. Research on competition and its benefits to the customers is ambiguous. In an industry-agnostic discussion, Banker et al. (1998) could not present the unequivocal idea of the role of competition on the quality of the services. Such a prevaricate view about quality in the services sector due to competition is given by Allon and Federgruen (2007). This ambiguity on the role of competition for profitability also prevails in the banking sector.

Tan (2016) presented a study on Chinese banks during 2003–2011 using the GMM (Generalized Method of Moments) method that competition does not impact profitability. However, Tan et al. (2017) found evidence that competition impacts profitability but negatively (on the same Chinese set of banks), which further sparked the controversy surrounding the role of competition on bank profitability. In addition to this, Moudud-Ul-Huq et al. (2020), in their study on MENA (the Middle East North Africa) countries from 2011-to 17, found supporting outcomes to the findings of Tan et al. (2017) that competition greatly decreases bank profitability. However, Llewellyn (2005) and Pham et al. (2020) explicitly contradicted the findings of Tan et al. (2017) and Moudud-Ul-Huq et al. (2020).

Llewellyn (2005), in his work on the UK and European banks, found that competition encourages European banks to boost their profitability to that of UK banks. Moreover, the change of strategic decision from the SHV (Share Holder Value) proposition to the STV (Stake Holder Value) proposition in European banks (on a similar line to the UK banks) may have kicked in due to competition only. In their study on Ukraine-based banks, Pham et al. (2020) also posited similar positive findings on the role of competition for profitability. Similar findings are reported by other studies (Kuknor & Rastogi, 2021; Singh & Rastogi, 2020). Thus, the following hypothesis is framed in the alternate form to test the role of competition on profitability empirically.

H1: Competition influences the profitability of the banks.

2.2. Competition and Risk-Taking (NPA) in Banks

In the banking industry, the association of competition with risk-taking, particularly NPA, is oriented toward minimizing the negative relationship between the two. Except for a few exceptions, most of the paper finds evidence that the competition is detrimental to the banks concerning NPA. Some studies are done from the market-power point of view, which is another way to look at the competition (Agoraki et al., 2011). However, the same results are shown. Jiménez et al. (2013), in a study on the Spanish banking sector, evidence that through the reduction in competition or an increase in the market power through concentration of banking services, risk in the bank is reduced. Alam et al. (2019) showed that competition increases the risk-taking in the banks, and stringent capital regulations further aggravate the association. Their study is of 208 banks in 10 highly developed Islamic banking countries. Agoraki et al. (2011) also advocated the same results and highlight that increased market power coupled with capital requirements aptly reduces all types of risks in banks (credit, liquidity, and default risks).

Boyd and De Nicolo (2005), on the other hand, found inconsistent results when it comes to the relationship between competition (market power) and bank risk-taking. He adds that taking either the extreme position that competition is beneficial or the opposite position that competition is harmful is not the best approach. They advise that the nature of the relationship between competition and risk-taking at a bank is dependent on numerous unforeseen elements and cannot be generalized. The stock market also adds lots of volatility and risk in the estimation of bank performance. Competition adds stock market-based volatility (Sharma et al., 2020; Patil & Rastogi, 2020b; Patil & Rastogi, 2020a; Patil & Rastogi, 2019). Dividend-related concerns also add volatility to the bank stocks (Pinto et al., 2019; Pinto & Rastogi, 2019). Therefore, to empirically explore the role of competition for risk-taking in banks, the following hypothesis is framed in the alternate form:

H2: Competition impacts risk-taking (NPA) in the banks.

3. Theoretical Framework

The study proposes the following two models to empirically test the theoretical framework developed in the current paper. It is believed that the rivalry would have some influence on bank profitability and risk-taking. Literature could not provide an acceptable answer to the issue; therefore, the following two models are built for empirical testing. To analyze the potential difference in the association between public and private sector banks in terms of competition, a dummy variable is created. Moreover, it is also believed that disclosure (Bhimavarapu & Rastogi, 2020; Solikhah et al., 2020; Qizam, 2021) and efficiency (Kanoujiya et al., 2021; Rastogi et al., 2021) may impact the profitability and risk taking in the banks (Mazumder & Hossain, 2018; Nguyen & Nguyen 2020; Nguyen et al., 2020). Therefore, controlling them is essential to study the actual impact of competition on both variables.

Profitability = f (Competition, DV for Bank type, Control Variables)       (1)

Risk taking = f (Competition, DV for Bank type, Control Variables)      (2)

Where

Profitability: NIM (net interest margin)

Risk-taking: NPA (non-performing assets)

DV: Dummy Variable for commercial and public sector banks

Control variables: Technical Efficiency and Disclosure Index for banks

Profitability is measured by NIM (Net Interest Margin) (Hamadi & Awdeh, 2012). Risk-taking is measured by NPA (Sen & Sen, 2015). DV for the bank type: dummy variable (1 for public sector bank and 0 for private sector). Two variables are used as control variables: Technical efficiency measured by DEA (Data Envelopment Analysis) method (Jemric & Vujcic, 2002) and the bank disclosure index. The construction of the disclosure index is explained in the coming sections. Similarly, the variable used for competition (Learner’s Index) is described in detail in the coming section (Sec 3.2).

3.1. Construction of T&D Index

According to Hossain (2008), in the Indian economy, banks play a vital role. The significance of the T&D concerning banks is discussed by authors (Wang et al., 2008; Silva et al., 2008; Al-Mashat et al., 2018) research studies. An in-depth review of the literature found that studies considered the T&D index for non-financial firms, but none of the studies focused on banks. Authors (Patel & Dallas, 2002; Aksu & Kosedog, 2006; Arsov & Bucevska, 2017) considered the S&P database for their study. Adopting the same, we considered the compiled database provided by S&P as a base, and added a few newer sets of variables unaddressed in the literature but would be of greater importance to stakeholders’ well-being for constructing the Customised T&D index.

While Arsov and Bucevska (2017) and Patel and Dallas (2002) considered 98 attributes, Aksu and Kosedag (2006) customized and made it to a total of 106 attributes. To build an effective T&D model, we finalized with 102 desirable attributes, considering the three broad categories identified by authors (Patel & Dallas, 2002; Aksu & Kosedog, 2006; Arsov & Bucevska, 2017) and introducing a new broad category of “Strategic, Technology and Basel Disclosures.” To brief in detail on the newly constructed T&D index, below are the broad categories;

1. Ownership Structure & Investor Relations (10 attributes),

2. Board & Management Structures & Processes (29 attributes),

3. Financial Transparency and Information Disclosure (30 attributes), and

4. Strategic, Technology, and Basel Disclosures (33 attributes).

Greater importance has been given to the newly added category as it was ignored in the earlier studies. As the current research focuses on the Indian banking sector, we consider our sample’s mix of government and commercial banks. A total of 34 (15 government and 19 commercial) banks were included. Initially, a thorough examination of the annual reports and websites of the individual institutions is carried out. When the information is missing in the primary source, we opted for the alternative source, i.e., BSE India, NDTVprofit.com, MoneyControl.com, etc. The time span is confined to four years from 2015–16 till 2018–19 because the Indian banking industry is experiencing considerable transformation for various reasons; statistics prior to 2016 may not represent recent changes in the Indian banking business. As a result, the traditional five-year period has skirted, and just four years of data have been used in the study.

The T&D index score is calculated either using a weighted or unweighted method. Adopting the methodology (Hossain, 2008; Turrent & Ariza, 2012; Kumar & Kidwai, 2018) current study considers the unweighted method. Binary values are allowed in this method, i.e. one for the availability of the data and zero if data is not present or unavailable for each attribute present in the index.

3.2. Learner’s Index for Market Power (Competition)

From the multiple options available to calculate market power for the banks, the learner index is the most convenient way (Coccorese, 2014). The learner index is widely used to determine a bank’s current market dominance (Haque & Brown, 2017; Ariss, 2010; Elzinga & Mills, 2011). For the current study, we customized the methodology used by Agoraki et al. (2011).

\(\mathrm{Lit}=\frac{\left(p_{i t}^{q}-\mathrm{MC}_{t}\right)}{p_{i t}^{q}}\)       (1)

p^q_it is the final price for an i_th bank for the time t. The price is the change in the revenue earned through the bank’s interest on the total assets. The following formula is used to calculate the marginal cost (mc).

\(\begin{aligned} \operatorname{In} C_{i t}=& b_{0}+b_{1} \overline{\operatorname{In} q_{i t}}+\frac{1}{2} b_{2}\left(\overline{\operatorname{In} q_{i t}}\right)^{2}+b_{3} \overline{\operatorname{In} d_{i t}}+\frac{1}{2} b_{4}\left(\overline{\operatorname{In} d_{i t}}\right)^{2} \\ &+b_{5} \overline{\operatorname{In} w_{i t}}+\frac{1}{2} b_{6}\left(\overline{\operatorname{In} w_{i t}}\right)^{2}+b_{7}\left(\overline{\operatorname{In} q_{i t}}\right)\left(\overline{\operatorname{In} w_{i t}}\right) \\ &+b_{8}\left(\overline{\operatorname{In} q_{i t}}\right)\left(\overline{\operatorname{In} d_{i t}}\right)+b_{9}\left(\overline{\operatorname{In} d_{i t}}\right)\left(\overline{\operatorname{In} w_{i t}}\right)+e_{i t} \end{aligned}\)       (2)

C_it denotes the cost of the i_th bank at the year t. variables q, d, w, and e represent the banks’ total assets, bank deposits, input costs, and stochastic disturbance terms. Deviations of the respective means help resolve the multicollinearity issue (Wang et al., 2008; Uchida & Tsutsui, 2005) are symbolized by the bars on the variables. As discussed above, it represents i_th bank at the year t of Lit, where L = (0, 1) signifies perfect competition and monopoly. If the value of the L is inverse, it represents the bank’s non-optimizing actions. Where C symbolizes total expenses, q denotes real incomes, and d represents short-term capital.

In addition, the sum of interest, employees, and other running expenses yields total expenditure (C ). The value of total earning assets such as equity shares, loans, and other assets is represented by banks’ output (q) (Altunbaş et al., 2001; Hughes & Mester, 1993). The input cost(W) is the sum of w1, w2, and w3 (Agoraki et al., 2011). Where w1 is the deposit price, computed as the interest expense to total deposits ratio, w2 represents the labor price, determined as the ratio of staff expenses to total assets, and w3 represents the capital price, computed as the ratio of other operating expenses to total fixed assets. (Berger & Humphrey, 1997; Berger et al., 2009).

4. Data and Methodology

4.1. Data

Our analysis includes 136 observations from a balanced panel dataset of 34 banks comprising public (owned by the Government Bodies) and private sector (owned by the private players) banks observed from 2015 to 2019. There are 19 nationalized banks, with the State Bank of India and its affiliates accounting for 9. There were 15 banks in the private sector. Scheduled and cooperative banks were excluded from our study as they are not controlled by the Reserve Bank of India (RBI). The Annual Reports and corporate Websites for each of the banks in question are the primary sources of information. We used the PROWESS database from the CMIE (Centre for Monitoring Indian Economy), RBI (Reserve bank of India), and BSE India websites in the absence of any data.

Since the Indian banking sector underwent significant restructuring for several reasons, we collected data for the study from the financial year 2015 till 2019 only. Data before 2015 might not be able to capture recent developments in Indian banks. Our data is a balanced panel dataset. It means that each attribute is replicated every financial year. We had to exclude the 2019–2020 financial year from our data collection because the Indian banking industry experienced significant turmoil in 2019 due to government-induced mergers of public sector banks. Some studies (Lai et al., 2013; Wanderi, 2016; Amidjaya & Widagdo, 2019; Srairi, 2019) support using a short panel of 4–5 years validating the utilization of 4 years of data for the current study.

Profitability, which is measured using NIM (Hamadi & Awdeh, 2012), and risk-taking, measured using NPA (Sen & Sen, 2015), are the variables used in the current study to measure the performance of the performance the banks. Net Interest Income (NIM) is defined by Tan (2016), Nguyen (2012), and Hamadi and Awdeh (2012) as the difference between advances and deposits in terms of interest earned and paid to depositors. According to Sarita et al. (2012), NIM can monitor conventional loans and the banks’ credit granting activities. Further net interest margin is a measure of the difference between interest paid and interest received, adjusted for the total amount of interest-generating assets held by the bank. NPA (Non-Performing Assets) is defined by Midthanpally (2018) and Sen and Sen (2015) as a percentage of Net NPA (Gross NPA-Provisions) of total advances. The variables disclosure index and technical efficiency are used as control variables in the study. It is believed that they may affect the variables chosen for the study, namely bank profitability and risk-taking. The control variable, the T&D index, is estimated based on the prevailing literature (Patel & Dallas, 2002; Thinh, 2018; Turrent & Ariza, 2012). According to Jemric and Vujcic (2002) and Saha and Ravisankar (2000), Technical Efficiency is measured using DEA using the constant return to scale approach as the input orientation.

4.2. Methodology

The current study deploys panel data regression to test the hypothesis. The following equation is specified for the static panel data model using equation one theoretical model.

NIM_it = β₀ + β₁ LI_it + β₂ TDI_it + β₃ EFF_it + u_it       (3)

NPA_it = β₀ + β₁ LI_it + β₂ TDI_it + β₃ EFF_it + u_it       (4)

Where u_it = μ_i + ν_it; μ_i is the individual effect term (entity term for banks), and ν_it is the remainder error term. The variable notations are reported in Table 1.

Table 1: Descriptive Statistics

Note: NIM is net interest margin, measured in percentage. NPA is non-performing assets measured as a fraction of total advances. LI is Learner’s index. TDI is the transparency and disclosure index. Eff is technical efficiency measured through the DEA method. Values in the correlation matrix are correlation coefficients. Values in parenthesis are p-values.*significant at 5%.

As opined by Baltagi and Raj (1992), Baltagi and Song (2006), and Hsiao (2005), panel data analysis has perceptible advantages as compared to other types of data sets. Lower risk of collinearity problems, adaptability for complex conditions, availability of additional information, and individual heterogeneity are essential features of panel data. According to Rastogi et al. (2021), Panel data are observations of various phenomena collected over time for the same or different firms. Static panel data model and dynamic panel data model are considered for the study as using both static and dynamic panel data specifications will aid in the confirmation of the robustness of the results (Rastogi et al., 2021). The static panel data model considers both time series and cross-sectional data simultaneously (Arellano, 1987). The dynamic panel data model includes lagged values in the list of exogenous variables, as highlighted in this model (Arellano & Bond, 1991). The panel used in this analysis analyzed data from 34 banks over four years, indicating that the data may be skewed by Nickell’s bias (Nickell, 1981). As our chosen data set represents a short panel, i.e., the number of cross-sections which is denoted by “N”, is greater than the period considered in the panel “T”, exacerbates the issue of endogeneity by the addition of a lagged dependent variable as an exogenous variable.

Leitão (2012) explained that static panel data faces difficulties such as serial correlation, heteroskedasticity, and endogeneity of some explanatory variables. The GMM- system (GMM-SYS) estimator helps solve those issues the scope for using the dynamic panel data model. Further reasons for using dynamic panel data in bank-related studies concerning profitability, as opined by Bolarinwa et al. (2019), Dietrich and Wanzenried (2011), and Goddard et al. (2011). According to Wooldridge (2006), lagged T&D values may have endogeneity problems, and thus dynamic panel data specification may be better suited to address the problem. Due to the persistence issue of weak instruments, Anderson and Hsiao’s (1981) Two-Stage Least Square method might not solve the problem. As a result, to estimate complex panel data specification, this analysis employs the system GMM (Generalized Method of Moments) proposed by Arellano and Bond (1991) and Arellano and Bover (1995). As Bolarinwa et al. (2019) applied, system GMM is favored over differenced GMM because differenced GMM suffers from possible biasness and inconsistencies. As a result, the following dynamic panel data model is estimated using the theoretical model developed in equation 2.

NIM_it = β₀ + NIM_it–1 + β₁ LI_it + β₂ TDI_it + β₂ EFF_it + u_it       (5)

NPA_it = β₀ + NPA_it–1 + β₁ LI_it+ β₂ TDI_it + β₃ EFF_it + u_it       (6)

Where error terms signify as in equations 4 and 5, the variable notations are reported in Table 1.

5. Results

Table 1 contains descriptive data for all variables utilized in the study. The net interest margin (NIM) is calculated as the difference between net interest earned and interest paid as a percentage of advances. 2.6118 % is the mean value of NIM in the present dataset, with 1.02% as a minimum and 4.63% as the maximum percentage for the same. Another measure of performance of banks under consideration is NPA, which is defined as non-performing advances adjusted for provisions as a percentage of total advances. NPA has a mean value of 4.3236%, with a minimum value of .28% and a maximum value of 16% (standard deviation is 3.1792%). The measure for the competition is Linters Index (LI). The mean value of LI is 0.2474, with a standard deviation of 4.5430. The T&D index means the value is 0.5146, with a standard deviation is 0.0932. The technical efficiency mean value is 0.8744 with a standard deviation of 0.1855. Moreover, no two exogenous variables have a significant correlation coefficient equal to 0.80 or more. Therefore, the results of correlation rule out any issue of multicollinearity.

Static panel data results of both the performance measures (NIM and NPA) are reported in Table 2. The panel data regression of NIM with LI is not statistically significant. T&D, which is used as a control variable in the regression, is substantial, but the other variable used as the control variable, technical efficiency (Eff), is not significant. Moreover, the dummy variable (DV) for the public sector versus public sector bank is significant (as the associated p-value is 0.0000). DV takes zero (“0”) for private banks and “1” for public sector banks. The coefficient of the dummy variable is negative (–0.8890) implies that NIM is statistically better for private sector banks by (1.6566– (–)0.8890 =) 2.5456 %. The data is found to have a heteroscedasticity problem (Table II) (but no autocorrelation issue). Therefore, robust estimates are reported in Table 2.

Table 2: Result of Regression Analysis (Static Panel Data)

Note: 1 Wald test of heteroscedasticity has the null of no heteroscedasticity. 2Wooldridge test of autocorrelation in the panel has the null of no autocorrelation (with 1 lag). SER is the standard error of the regression. Theta estimates the fitness of the random effect model (higher is better). { } describe standard error and ( ) p-value at 5%. ^@ Regression out are robust estimates.

The panel data regression of NPA with LI gives more lucid results in static models (reported in Table 2). LI is significantly associated with NPA. The association between them is positive as the coefficient is .04371. This condition implies that more competitiveness results in higher levels of NPA. Both the control variables are also significant in the static panel data model. Moreover, public sector banks have a higher level of NPA as the dummy variable coefficient is 4.8470. This situation also implies that public sector banks have a 1.7467 % higher NPA level than private sector banks. A significant heteroscedasticity problem is found in the data. Therefore, robust standard errors are also reported (here again, the autocorrelation problem is not observed).

Table 3 shows the dynamic panel data estimates for both performance measures. LI is not significantly related in any situation. The dynamic panel data with NIM does not have a problem with overidentification and autocorrelation. Out of two control variables (T&D and Eff) in the estimate, for NIM, T&D is significant. In dynamic panel data regression of NPA with LI, both the control variables (T&D and Eff) are not significantly associated with NPA. However, overidentification is in the dynamic panel data estimation; therefore, robust estimates are reported.

Table 3: Regression Results (Dynamic Panel Data)

Note: Saran test is the test of over-identification issues under the GMM framework. The null hypothesis of the Sargan test is that there is no over-identification problem in the dynamic panel data model. Arnello-Bond test used in the analysis is for serial autocorrelation in the first differenced error terms of the order 1. Values in brackets [ ] are standard errors and parenthesis ( ) are p-values at 5%. ^@ Robust Estimates.

6. Discussion

Hypothesis one is rejected that competition influences profitability (measured by NIM). In both the models, static and dynamic panel data models, the coefficients are not statistically significant. Therefore, this can be inferred that competition does not impact profitability in banks in India. The empirical testing of the second hypothesis that competition influences the NPA gives mixed evidence. The relevant coefficient in the static panel data model is significant, but it is statistically insignificant in the dynamic panel data model. As a result, it is unclear whether competition has an effect on NPA levels in Indian banks. The current study’s findings support Tan’s (2016) contention that competition has no effect on bank profitability.

The current paper contradicts both the other school of thought, which supports that competition either negatively impacts profitability (Moudud-Ul-Huq et al., 2020; Tan et al., 2017) or positively impacts profitability (Llewellyn, 2005; Pham et al., 2020). The exact opposite evidence by two schools of thought endorses the findings of the current paper that competition, in general, may not impact the bank’s profitability. Moreover, the study of Tan et al. (2017) and Pham et al. (2020) use the banks from China and Russia for their study, which have non-democratic dispensations. Moudud-Ul-Huq et al. (2020) use relatively less developed banks of African nations (MENA countries). Llewellyn (2005)’s findings may be biased because of the internal conflict of intra-European nations, mainly between the UK and other European nations. As a result, the current paper’s findings are supported.

Furthermore, the current paper’s conclusions concerning competitiveness and risk-taking (NPA) in Indian banks are supported by Boyd and De Nicolo (2005). Our findings endorse the mixed view (Boyd & De Nicolo, 2005) or contingent view of Niinimäki (2004) that the role of competition in risk-taking cannot be generalized. Rather it depends upon other contingent factors, for example, regulation. Such findings of the current paper contradict the widespread view that competition increases banks’ risk taking (NPA) (Agoraki et al., 2011; Alam et al., 2019; Bolt & Tieman, 2004).

The new paper greatly contributes to the existing body of knowledge on the subject. The findings of the study support the assumption that competition does not improve the performance of India’s banking system. Considering the paucity of empirical evidence on the subject, the current paper’s contribution is remarkable. This paper is against the popular belief that competition makes firms more profitable and safer is not supported in the article. The outcome of the current paper is more pertinent because the performance considers both profitability and risk-taking. Therefore, this paper hugely contributes and provides guidelines to be followed by the decision-makers.

This paper has considerable implications for both: managers and policymakers. From the managers’ perspective, they should be concerned that competition may not be able to increase profitability while decreasing bank risk-taking. Therefore, they may concentrate on creating value for the stakeholders even by circumventing the competition. From the policymakers’ point of view, they may devise long-term banking plans that are bereft of competition. Undue competition may be avoided, which appears to be of no good to both profitability and risk-taking in the banks. The findings of the current paper can help both the concerned parties, managers, and policymakers, boost the banking fraternity’s morale, which seems to be lost in the labyrinth of competition. In the banking sector, a one-size-fits-all approach to the competition may not yield a reasonable profit. As a result, a policy reconsideration is required. For banks, a new competition policy is needed. This situation may provide the needed momentum to the global banking sector.

7. Conclusion and Limitations

According to the study’s findings, competition has little effect on the performance of Indian banks in general. Profitability and risk-taking in banks are used to assess bank performance. Furthermore, the current study investigates the impact of competition on bank profitability. Following empirical research, it was discovered that competition has no significant impact on bank profitability. However, the impact of competition on the other performance metric, risk-taking, is equivocal. It has been discovered that either competition dramatically raises NPA levels or it has no effect. As a result, it is concluded that the impact of competition on risk-taking is ambiguous.

Other, though subtler, popular performance measures for banks also exist, for example, stock market return or valuation. These other performance measures may have also been examined when measuring the influence of competition on bank performance. Furthermore, data for a longer period of time may have added value to the product. These are some of the paper’s shortcomings. Such difficulties may be addressed in future research on the subject.