Canadian Banking Industry Profitability: Exploring the Relevance of Two Competing Hypotheses

This study investigates the Canadian banking industry profitability and seeks to determine if there is evidence of market power hypothesis (MPH) and/or efficiency structure hypothesis (ESH) in the industry. Using GMM and data from 2006 to 2018, it finds no support for the structure-conduct-performance (SCP) component of MPH. However, evidence of relative market power (RMP) in the industry reflects partial support for MPH, implying that banks that offer differentiated products are able to exercise market power, increase market share, and achieve better profitability. The lack of support for X-efficiency (ESX) and scale efficiency (ESS), which are components of ESH, indicates no support for the ESH. The finding that QLH holds in the industry suggests the level of competition is inadequate and that managers in the industry exhibit suboptimal behaviour. Therefore, increasing the level of competition in the industry will stimulate managerial effectiveness. Findings relating to the control variables show that spread impedes profitability, whereas capitalization and the joint influence of spread and liquidity risk have facilitating effects. Credit risk is immaterial to profitability however, the effect of economic growth can be positive. This study provides a better understanding of the industry, which is important to managers, regulators, and policy makers. The robustness checks affirm the consistency of the findings and policy implications.

2017; Tregenna, 2009), the evidence reflects full support for the MPH or the ESH. Nonetheless, studies such as Simatele, Mishi, and Ngonyama (2018), Sharma, Gounder, and Xiang (2013), Kapopoulos and Siokis (2005), and Berger (1995) show concurrent presence of MPH and ESH components. These inconsistencies in findings suggest the need for an empirical study with more insight. A good understanding of MPH and ESH is important to managers in the industry seeking to improve profitability. It is also important to policy makers trying to avoid wrong policy decisions. Studies on the Canadian banking industry are inconsistent and not contemporary. For example, analyzing data from 1972 to 1974, Short (1979) seems to indicate that profitability is due to market power, but the period investigated encompasses early changes in the industry's evolution. On the other hand, Shaffer (1993) suggests that the industry is perfectly competitive, which the study considers similar to findings in Nathan and Neave (1989). The view that the industry is perfectly competitive could be due to investigating periods in the 1980s when the 1980 Bank Act amendment that changed the industry's competitive landscape took effect. Based on data from 1994 to 2001, Claessens and Laeven (2004) examination of banking industry around the world show that the Canadian banking industry is monopolistically competitive. Also, using data from 2000 to 2006, Allen and Engert (2007) insinuate no evidence of collusive behavior in the industry, but the study notes that the industry reflects monopolistic competition. In general, studies on Canada are insufficient, not current, and differ in findings. In addition, they give no clear understanding of how the industry's structure affects profitability. Furthermore, existing studies provide no clear information to help discern if MPH and/or ESH holds in the industry, which are inquiries addressed in this study. Since banking industry structure changes with time (Mendes & Rebelo, 2003), wrongly proclaiming MPH and/or erroneously declaring that ESH holds in the industry could lead to incorrect policies, strengthening the rationalization for a new empirical study on Canada. By focusing on MPH and ESH, this study sheds light on the industry's structure, allowing regulators and policy makers to discern the impact of each of the two hypotheses on profitability prior to instituting new measures or initiatives. This study has six sections. Following the first section (i.e., Introduction), the next section (Section 2) provides a review of related literature on banking industry structure, efficiency and profitability. Section 3 describes the methodology and data. Section 4 discusses the findings. Section 5 presents the robustness checks that affirm the findings. The study concludes with policy implications and suggestions for future research.

Literature Review
Studies on banking industry structure apply non-structural and structural approaches. The nonstructural studies (e.g., Rodriguez, Bolivar, & Reyes, 2018;Demirel & Hatirli, 2017;Bikker, Shaffer, & Spierdijk, 2012;Aktan & Masood, 2010;Gischer & Stiele, 2008;Casu & Giradone, 2006) rely on Bresnahan-Lau model, Lener index, and the Panzar-Rosse model. Typically, studies of this nature identify the industry as monopolistically competitive, however, some characterize it as perfectly competitive (e.g., Shaffer, 1993;Nathan & Neave, 1989), oligopoly (e.g., Babic, Zildzovic, & Loncar, 2015;Mensi, 2010), or monopoly (e.g., Mustafa & Toci, 2017). The structural approach relies on proxies such as Herfindahl-Hirschman Index (HHI) and concentration ratio of n firms (CR n ) to elucidate the level of competition in the industry (Nguyen, 2019;Bikker, Shaffer, & Spierdijk, 2012;Berger, Demirguc-Kunt, Levine, & Haubrich, 2004). Conceptually, studies (e.g., Bucevska & Misheva, 2017;Alhassan, Tetteh, & Brobbey, 2016;Al-Muharrami & Matthews, 2009) that apply the structural approach highlight two competing hypotheses (i.e., market power hypothesis [MPH]; efficiency structure hypothesis [ESH]) in discuss relating to banking industry profitability. The MPH assumes that industry structure determines profitability, whereas the ESH presumes that efficiency results in low costs and better profitability (Alhassan, Tetteh, & Brobbey, 2016;Mensi & Zouari, 2011). Within the MPH is the structure-conduct-performance (SCP) construct and the relative market power (RMP) conjecture. The SCP construct presupposes that industry concentration affects conduct and shapes performance (Hamid, 2017), suggesting that the absence of competition in highly concentrated industry creates the environment where firms collude to achieve higher level of profitability (Lelissa & Kuhil, 2018;Berger et al., 2004). On the other hand, the RMP conjecture asserts that firms with differentiated products are able to increase market share and power, resulting in better profitability (Hamid, 2017). The ESH discerned by Demsetz (1973) posits that efficient firms can enlarge in market share and attain higher profitability than less efficient firms (Bucevska & Misheva, 2017). Similar to the MPH, the ESH has two suppositions. As espoused in Lelissa and Kuhil (2018, p. 35), Seelanatha (2010, p. 23) and Fu and Heffernan (2009, p. 42), the first is "X-efficiency" (ESX), which supports the notion that firms with efficient operations or better management achieve cost advantage, lager market share, and better profitability. In essence, the higher the efficiency, the larger will be the market share and market power, leading to an increase in the industry's concentration and high profitability for the efficient firm. The second is the "scale efficiency" (ESS) proposition, which maintains that if banks have similar management and comparable technology, the bank operating closer to the most productive scale size will generate better profitability, suggesting that profitability is the result of operating at the appropriate operational scale. Studies on banking industry structure, efficiency, and profitability ijef.ccsenet.org International Journal of Economics and Finance Vol. 12, No. 4;2020 have generated conflicting findings. Some (e.g., Hamid, 2017;Qayyum, Mehmood, Kasheer, & Haider, 2018;Nabieu, 2013) focus on MPH by investigating SCP and RMP in the industry. Although Nabieu (2013) shows the presence of SCP, findings in Hamid (2017) and Qayyum et al. (2018) reveal support for RMP. Noting no support for MPH (i.e., SCP & RMP), Dabbous & Nassereddine (2017) and Bucevska and Misheva (2017) provide evidence in favour of the ESH, signifying that profitability is due to efficient operations, not collusive behaviors. The inference is that banks with better management and operational scale are able to curtail costs, enlarge market share, and increase profitability (Casu & Giradone, 2006). On the contrary, some studies indicate partial support for MPH. A classic example is Ye, Xu, and Fang (2012) findings that show no evidence of SCP or ESH. However, its strong support for RMP signals that banks with differentiated products and services are able to increase profits by exercising market power. In addition, there are studies such as Bhatti and Hussain (2010) and Al-Muharrami and Matthews (2009) that show support for SCP, demonstrating that concentration in the industry creates the environment where banks collude to increase profitability. Furthermore, single country studies (e.g., Simatele, Mishi, & Ngonyama, 2018;Alhassan, Tetteh, & Brobbey, 2016;Doyran, 2012;Ye, Xu, & Fang, 2012;Tregenna, 2009;Lu, Fung, & Jiang, 2007) differ in findings, which could be due to differences in methodology, country characteristics and economic conditions, regulation, and the period of study. Notably, studies that examined the industry in the same country have produced conflicting findings. For example, Doyran (2012) and Tregenna (2009) differ in findings relating to the banking industry in US. Also, Ye, Xu, and Fang (2012) and Lu, Fung, and Jiang (2007) deliver contrasting results concerning the Chinese banking industry. In the case of Canada, while Short (1979) finds that market power leads to higher profitability, Nathan and Neave (1989) and Shaffer (1993) suggest that the industry is perfectly competitive. On the other hand, Allen and Engert (2007) conclude that the industry is monopolistically competitive. Understandably, some of the studies on Canada investigated periods in the early stages of change in the industry, however, others examined periods when changes to the Bank Act took effect. The justification for a Canadian study stems from comprehending that findings relating to other countries are contradictory and may not be relevant to Canada. In addition, outcome of studies from European Union and US may not reflect the prudential nature of the Canadian banking industry. Furthermore, the small number of studies on the Canadian banking industry are dated and inconsistent in findings, which could expose the industry to wrong policy decisions. Investigating the MPH and ESH reveals information that managers in the industry would find pertinent when attempting to augment profitability. Additionally, this study contributes to discussions that provide policy makers better understanding of the dominant construct in the industry. If MPH prevails, policies should guard against consolidation, preventing the concentration of the industry and/or collusive behavior (Ajlouni, 2010). On the other hand, if ESH prevails, policies should ensure that efficient banks continue to minimize costs and increase market share, even though it could lead to an increase in the concentration of the industry (Fare, Grosskopf, Maudos, & Tortosa-Ausina, 2015; Seelanatha, 2010)

Methodology and Data
Studies (e.g., Bhatti & Hussain, 2010;Molyneux & Forbes, 1995) test the underlying assumption relating to MPH and ESH using the structural approach model similar to the one shown below: The notation Π it is the profitability of bank i in period t. The profitability variable is often proxied by either return on assets (ROA) or return on equity (ROE) (Antoun, Coskun, & Georgievski, 2018;Sharma, Gounder, & Xiang, 2013). HHI t is the industry's concentration in period t. Herfindahl-Hirschman Index (HHI) is the proxy for the industry's concentration (Bikker, Shaffer, & Spierdijk, 2012;Berger et al., 2004). MS it is the market share of bank i in period t. Z it is the vector of additional control variables that include bank specific, industry specific and macroeconomic variables i in period t. β 0 is the constant term, whereas e it is the random error. As noted in (Bhatti & Hussain, 2010), traditional SCP is confirmed if β 1 > 0 and β 2 = 0, suggesting that concentration is associated with the industry's profitability. The efficiency construct is ascertained if β 1 = 0 and β 2 > 0, indicating that efficient firms are able to increase market share and profitability. While the pioneer model (Model 1) provides the foundation for understanding MPH and ESH, Sathye and Sathye (2004) and Berger (1995) remarks show that the market share (MS) variable may reflect effects that are not associated with efficiency, implying that MS is an incorrect proxy for efficiency. Similar to Berger (1995) and Fu and Heffernan (2009), the underlying model that captures the two efficiency measures in this study is: signify that banks could minimize costs by using less inputs to generate outputs, or by managerial effectiveness. As such, it is common to use technical efficiency (TE) or pure technical efficiency (PTE) as the X-efficiency (ESX) measure in conjunction with SE, which is the scale efficiency (ESS) measure. X-efficiency (ESX) indicates that banks with superior management or technology would produce outputs at minimum costs (Lelissa & Kuhil, 2017). Rather than infer cost efficiency (CE) from TE or PTE, this study adopts the approach in Bucevska & Misheva (2017), directly incorporating CE in the model to reflect X-efficiency (ESX) along with SE as the scale efficiency (ESS) measure. The modified model with ROA as profitability variable is: Banks become cost efficient by being technically and allocatively efficient, minimizing costs for a given level of outputs to increase profitability (Isik & Hassan, 2002). In addition, becoming scale efficient by operating at optimum scale size minimize costs, contributing to higher profitability (Behname, 2012;Tregenna, 2009). Similar to Oredegbe (2020), the industry's CE and SE were determined using input oriented Data Envelopment Analysis (DEA), which is a nonparametric approach requiring no price information. Additionally, the intermediation approach, which is the common approach for input and output selection is relied on in the selection of the three inputs (i.e., deposits, employment, non-interest expense) and two outputs (i.e., loans and net-interest income). Studies that have used DEA efficiency scores include Guillen, Rengifo, and Ozsoz (2014) and Seelanatha (2010). Although studies vary in methodology and estimation approach, the generalized method of moments (GMM) estimation technique is prevalent. It deals with potential endogeneity bias, and mitigates erroneous estimation (Ullah, Akhtar, & Zaefarian, 2018;Alhassan, Tetteh, & Brobbey, 2016). Furthermore, it is a good estimator in situations where the explanatory variables are not decisively exogenous (Garza-Garcia, 2012). This study uses the single equation instrumental variable regression with GMM as the estimator, robust weighting matrix, and reports findings at 95% confidence level. Some studies (e.g., Bhatti & Hussain, 2010;Tregenna, 2009) incorporate bank and industry specific variables in the model, however there are studies (e.g., Dabbous & Nassereddine, 2017;Mensi & Zouari, 2010) that include a combination of bank specific, and industry and macroeconomic variables in the model. The additional control variables in the model in this study consist of bank and industry specific variables such as credit risk (LLPTL), spread (SPREAD), liquidity risk (LOTD), and capitalization (EQTA). In addition, it includes economic growth (i.e., change in GDP per capita [CGDPPC]) as the macroeconomic variable. Also, an interaction term (SPREAD*LOTD) is included in the model to assess the joint impact of spread and liquidity risk. The model with bank and industry specific and the macroeconomic variable is: The data is from the Canadian Bankers Association database, annual financial statements of banks, and Conference Board of Canada. To remove the impact of inflation, variables in Canadian dollars have been adjusted using consumer price index (CPI) with 2015 chosen as the base year. As noted in Berger (1995) and Fu & Heffernan (2009), for MPH to hold, coefficients of variables reflecting SCP and RMP have to be greater than zero (i.e., β 1 > 0, β 2 > 0, β 3 = 0, and β 4 = 0). Similarly, for ESH to prevail, coefficients of variables reflecting X-efficiency (ESX) and scale efficiency (ESS) have to be greater than zero (i.e., β 1 = 0, β 2 = 0, β 3 > 0 and β 4 > 0). In addition, support for a unique instance of market power, which is known as Hicks' (1935) If concentration and market share variables have statistically significant negative influence on CE (i.e., the proxy for X-efficiency [ESX]) and SE (i.e., the proxy for scale efficiency [ESS]), support for QLH is evident (Fu & Heffernan, 2009;Berger & Hannan, 1998). It would suggest that market power allows managers in the industry to exhibit slack behavior, and shirk responsibilities to increase revenue and/or minimize costs (Fare et al., 2015). Similarly, to carry out confirmation tests for the ESH, the two tests below (Model 5 & Model 6) are necessary.
The CE and SE variables denoting X-efficiency (ESX) and scale efficiency (ESS) should have positive and statistically significant effects on concentration (HHI) and market share (MS). ijef.ccsenet.org International Journal of Economics and Finance Vol. 12, No. 4; 2020

Results and Discussion
For insight on the structure of the industry, HHI, which is the proxy for industry concentration was determined by finding the sum of the square of the market share of firms in the industry (Ferreira, 2013;OECD, 2010). The mean HHI is 1894 when the HHI information is based on total assets, but it is 1879 when the data is based on deposits (Table 1). The industry's concentration based on assets (i.e., HHI a ) increased by 4.1 percent in the study period, signifying that larger banks are investing more in assets than smaller banks, which could allow them to increase access to customers and compete in the global financial markets. On the other hand, the increase in the industry's concentration when the concentration level is based on deposit (i.e., HHI d ) is 2.9 percent, signalling that large banks have not increased market share by much, or that smaller banks are gaining market share in the industry. The latter seems plausible as HHI d declined in the last two years (Table 1). Overall, the HHI a and HHI d trends in Figure 1 reveal an increase in the industry's concentration in the study period. Using the HHI classification in US Department of Justice and the Federal Trade Commission (2010, p. 19), it is reasonable to conclude that the Canadian banking industry is "moderately concentrated." This buttresses the outlook in Allen and Engert (2007), which shows that a handful of banks control more than 90 percent of the assets in the industry. Given that a concentrated industry could still be competitive (Owen & Pereira, 2018;Claessens & Laeven, 2004), finding that the Canadian banking industry is somewhat concentrated may not indicate that competition is lacking in the industry. The industry faces competition from non-banks (e.g., insurance), virtual banks, and emerging financial tech industry, suggesting that it is competitive despite the indication that it is somewhat concentrated (Department of Finance Canada, 2016). The efficiency trends in Figure 2 show decline in all areas of efficiency measure. This could be due to the concentration of the industry as studies indicate that high concentration hinders efficiency (OECD, 2010;Berger & Hannan, 1998).   Table 2 shows the regression outputs when HHI and MS information are based on assets and deposits respectively. Regression 1 reveals that HHI a as proxy for industry concentration is negative and statistically insignificant, indicating the rejection of SCP construct. On the other hand, the MS a variable denoting market share is positive and statistically significant, suggesting support for RMP proposition. Because SCP and RMP are components of MPH, these findings signal partial support for MPH. Since SCP presumes that banks achieve high profitability by colluding (Hamid, 2017;Berger & Hannan, 1998), the lack of support for SCP signifies absence of collusive behaviour in the Canadian banking industry, which is consistent with Allen and Engert (2007). Regression 2 with HHI and MS data based on deposits reveals similar findings. The evidence in support of RMP is similar to Ye, Xu, and Fang (2012) finding regarding the Chinese banking sector. It is also consistent with Garza-Garcia (2012) findings relating to the Mexican banking industry. However, it contradicts Alhassan, Tetteh, Brobbey (2016) outcomes concerning the Ghanaian banking industry. Nonetheless, the presence of RMP in the Canadian banking industry indicates that banks that differentiate on products and services attain better profitability than banks that do not. Such banks could accumulate market power and enlarge market share irrespective of the industry's concentration (Hamid, 2017;Fu & Heffernan, 2009). Product and service differentiation that allow banks to charge higher price could be achieved through a combination of unique products and services, good brand image and reputation, and network of branches for customers' convenience (Kavale, Mugambi, & Namusonge, 2016;Northcott, 2004).  Table 2), but positive and insignificant link with profitability (Regression 2; Table 2) suggests plausible evidence of X-efficiency (ESX), and partial support for ESH. The X-efficiency (ESX) finding is inconsistent with Doyran (2012) views relating to US financial institutions. Nonetheless, it supports Kapopoulos & Siokis (2005) findings regarding Euro area banking industry. The plausible evidence of X-efficiency (ESX) could indicate that banks with better management and technology are able to control costs, leading to better profitability, enlarged market share, and a possible increase in the industry's concentration (Bucevska & Misheva, 2017;Kapopoulos & Siokis (2005). While these findings offer no support for the scale efficiency (ESS) component of ESH, the negative relationship between SE and profitability signifies that banks in the industry have not been able to increase profitability by adjusting operational scale. In the interim, scale adjustment may have resulted in high costs and decline in profitability. However, in the long run, it could improve profitability as banks are better able to respond to customers' needs. The observed partial support for ESH is tested using Model 5 and Model 6. To confirm the plausible ESH, the coefficients of CE, which is the X-efficiency (ESX) variable and SE, which is the scale efficiency (ESS) variable in Model 5 and Model 6 have to be positive and statistically significant. The results of the tests in Table 3 (Table 3).
Because the results show that the requisite tests to proof ESH are not satisfied, the projected partial support for ESH in the industry is tenuous, leading to its rejection.   Hicks (1935, p. 8) "quiet life" hypothesis (QLH), which posits that as banks amass market power when industry concentration increases, managers can raise price and would exhibit no motivations to curtail costs (Lelissa & Kuhil, 2017;Berger & Hannan, 1998). The finding that QLH holds in the industry is consistent with Gavurova, Kocisova, and Kotaskova (2017) view of EU banking, Repkova and Stavarek (2013) outlook of the Czech banking industry, and Berger and Hannan (1998) assessment of the US banking industry. However, it contradicts Mensi & Zouari (2010) and Casu and Giradone (2009) findings that show no support for QLH. Nonetheless, the finding that QLH holds in the case of Canada signifies that banks in the industry are able to exercise market power, charging higher rates on loans to customers and paying lower rates on deposits.
Given the similarities in signs of the control variables in Regression 1 and Regression 2 (Table 2), the discussion of the control variables focuses on Regression 1. Among the control variables, LLPTL as a measure of credit risk (Pervan, Pelivan, & Arneric, 2016;Sharma, Gounder, & Xiang, 2013) is negative and statistically insignificant, suggesting that it has no relevant effect on profitability of banks. While this finding aligns with Hamid (2017) observation, it contrasts Bucevska and Misheva (2017) report of a negative and statistically significant effect on profitability of banks. Spread (SPREAD) denotes the difference between the rate banks charge on loans to customers and the rate paid on deposits (Aboagye, 2012). It has a negative effect on profitability in the two regressions ( Table 2). The finding that the effect is statistically significant in Regression 1 shows that spread could diminish profitability. To increase spread, banks may charge a higher lending rate, which increases the cost of borrowing (Saona, 2016). This could reduce incentive for new bank loans and cause profitability to decline. In addition, the negative impact of spread on profitability may reflect exposure of banks to losses if customers default due to the high rates on loan (Addae-Korankye, 2014). It is also possible for banks to increase spread by reducing rates on deposits, however, this makes deposits in banks unattractive to depositors (Soledad, Peria, & Mody, 2004). The reduction in deposits could constrain banks' ability to make new loans, leading to profitability decline. Change in GDP per capita (CGDPPC), which denotes economic growth (Majumder & Li 2018) has a positive link with profitability in Regression 1 and Regression 2 (Table 2). However, the impact on profitability in Regression 1 is statistically insignificant, implying that economic growth could be immaterial to the profitability of banks. While this finding refutes Ayaydin and Karakaya (2014) and Staikouras and Wood (2004) remarks that it undermines profitability, it is consistent with findings in Majumder and Li (2018) and Sharma, Gounder, and Xiang (2013) that show no effect on profitability. Although economic growth may result in more deposits with banks (Owen & Pereira, 2018), it could be unimportant to profitability unless banks utilize the funds in new loans and profitable investments. While the focus is on Regression 1, the statistically significant positive influence in Regression 2 (Table 2) aligns with Goddard, Molyneux, and Wilson (2004). In addition, it corroborates Yao, Haris, and Tariq (2018) finding that improved economic conditions create demand for loans, leading to higher profitability. LOTD reflects liquidity risk and the ability of banks to convert deposits to loans (Kocisova, 2016;Yudaruddin, 2012). The two regressions reveal a negative association between LOTD and profitability. The finding that the negative effect is statistically significant in Regression 1 (Table 2) shows that increasing loan to deposit ratio could inhibit profitability. This observation substantiates Garza-Garcia (2012), but it contradicts Simatele, Mishi, and Ngonyama (2018) finding that indicate statistically significant positive influence on profitability. High LOTD reduces liquidity and creates the need for banks to raise funds at higher costs (Gropp, Sorensen, & Lichtenberger, 2007), leading to profitability decline. EQTA, which is the proxy for capitalization (Kocisova, 2016) has a statistically significant positive effect on profitability. While this finding departs from Almumani (2013) remark that the effect is negative and statistically insignificant, it is consistent with findings in Bucevska and Misheva (2017) and Garza-Garcia (2012) that show that capitalization drives profitability. The positive effect on profitability indicates that banks with high capitalization will achieve better profitability. High capitalization reduces the risk of failure, conferring to banks benefits such as good reputation, and the ability to secure funds at low costs (Menicucci & Paolucci, 2016). The interaction term (i.e., SPREAD*LOTD) has a positive and statistically significant impact on profitability, implying that the joint effect of the two variables is important to profitability.

Robustness Check
While this study adopts CE as X-efficiency (ESX) measure, other studies use either TE or PTE to reflect X-efficiency (ESX). For this reason, the robustness checks when either TE or PTE replaces CE is necessary to determine if findings in this study are sensitive to parameter change. Similar to the original findings, the results of the robustness checks in Table 5 and Table 6 show that the MS variable is positive and statistically significant ijef.ccsenet.org International Journal of Economics and Finance Vol. 12, No. 4;2020 in each of the regressions, pointing to evidence of RMP in the industry. Additionally, the results in Table 5 and  Table 6 provide no support for the ESH, making it unnecessary to perform further tests associated with ESH (Mensi & Zouari, 2010). Nonetheless, in order to avoid wrong declaration, confirmation tests were performed. The test results in Table 7 and Table 8 show that the required conditions for ESH confirmation are not satisfied, signifying no evidence of ESH in the industry. In addition, the QLH test results with either TE or PTE as X-efficiency (ESX) variable (Table 9A and Table 9B) show that QLH holds in the industry. The R 2 in the robustness checks are similar to the initial findings. These R 2 are comparable to prior studies (e.g., Yuanita, 2019;Lelissa & Kuhil, 2017, Guillen, Rengifo, Ozsoz, 2014Sharma, Gounder, & Xiang, 2013;Berger, 1995). The R 2 should not be a concern. This is because the motive is not to predict but to identify variables with influence on profitability (Guillen, Rengifo, & Ozsoz, 2014). In general, the robustness checks reveal outcomes that are identical to the original results, thereby confirming the consistency of the findings in the study.

Conclusion
This study examines profitability in the Canadian banking industry and tests the two competing hypotheses (i.e., MPH & ESH). The HHI level indicates that the industry is somewhat concentrated, suggesting that banks have market power, however, concentration does not preclude competition (Claessens & Laeven, 2004). The tendency for competition in the industry is buttressed by finding no support for SCP. This signifies no evidence that banks in the industry attain better profitability by colluding on rates and fees. The support for RMP, which shows that banks achieve profitability by offering differentiated products and services harmonize with prior empirical studies including Pawłowska (2016), Sharma, Gounder, and Xiang (2013), and Garza-Garcia (2012). Nonetheless, the moderate level of concentration signals that banks in the industry possess some market power.
To prevent implicit collusion, regulatory policy should ensure that banks are transparent when setting rates and fees. The finding that CE (i.e., X-efficiency [ESX]) variable enhances profitability seems to provide partial support for ESH. However, because the requisite tests failed to confirm ESH, the purported partial support for ESH is unfounded, necessitating its rejection. The finding that QLH holds in the industry suggests that managers exercise market power, and display suboptimal behaviours by foregoing revenue increase and/or cost minimization opportunities (Subramaniam, Ab-Rahim, & Selvarajan, 2019;Berger & Hannan, 1998;Hicks, 1935). In view of this, banks need to establish proper monitoring mechanism to dissuade managers from engaging in behaviours that undermine efficient operations. Furthermore, the finding that QLH holds in the industry and the potential efficiency loss should concern regulators. Regulators need to aim at increasing the level of competition in the industry. The increase in competition could motivate managers to improve resource allocation, spur product and service innovations, and ensure access to better services and rates (Ab-Rahim, 2016).
Credit risk as denoted by LLPTL appears to have unimportant negative effect on profitability. Although unimportant, the negative relationship suggests that banks should not ignore credit risk management (Saona, 2016). As part of credit risk management, banks should properly screen loan applicants to reduce the likelihood of defaults and profitability decline (Muhamet & Arbana, 2016). The finding that spread impedes profitability signifies the need for banks not to pursue profitability objectives by arbitrarily increasing rates on loans, or decreasing rates on deposits. In addition, regulators need to ensure that rules that banks set are fair to customers. Because macroeconomic factors affect spread (Soledad, Peria, & Mody, 2004), government policy makers need to ensure that policy objectives are not contributing to spread increase. The findings relating to economic growth reveal that unless banks make new loans and profitable investments during economic growth, profitability may not increase. Liquidity risk as reflected by high loan to deposit (LOTD) ratio is detrimental to profitability. It could leave banks with insufficient funds, which undermines the ability to meet new loan requests or unplanned withdrawals by customers (Gropp, Sorensen, & Lichtenberger, 2007). Also, it could create the need for banks to close funding gaps with high cost funds, leading to profitability decline (Gropp, Sorensen, & Lichtenberger, 2007). To reduce exposure to liquidity risk, regulators need to ensure that banks are able to meet funding requirements. In addition, banks need to be amenable to investing in high liquid assets, which are convertible to cash to meet anticipated funding needs (Hlebik & Ghillani, 2017). The finding that capitalization improves