Article Sectoral Loan Journal of Emerging Market Finance19(1) 66–99, 2020 Portfolio © 2019 Institute for Financial Management and Research Concentration Reprints and permissions:in.sagepub.com/journals-permissions-india and Bank Stability: DOI: 10.1177/0972652719878597journals.sagepub.com/home/emf Evidence from an Emerging Economy Baah Aye Kusi1 Lydia Adzobu1 Alex Kwame Abasi2 Kwadjo Ansah-Adu3 Abstract In this study, the effect of sectoral loan portfolio concentration on bank stability is investigated in the Ghanaian banking sector between 2007 and 2014. Specifically, we investigate the linearity and non- linearity effects of sectoral loan concentration on bank stability given the limited exploration of this nexus. Employing a two-step generalized method of moments (GMM) robust random and fixed effects panel models of 30 banks, the study provides evidence showing that sectoral loan concentration weakens the stability of banks. This confirms the concentration-fragility hypothesis and the diversification theory of traditional banking but may promote bank stability beyond a certain 1 Department of Finance, University of Ghana Business School, Accra, Ghana. 2 Department of Accounting, Banking & Finance, Wisconsin International University College Ghana, Accra, Ghana. 3 Department of Banking and Finance, Valley View University, Accra, Ghana. Corresponding author: Baah Aye Kusi, Department of Finance, University of Ghana Business School, Accra, Ghana. E-mail: baahkusi@gmail.com Kusi et al. 67 threshold point. This implies that bank sectoral loan concentrate has a direct non-linear U-shape effect on bank stability in Ghana. We argue that although sectoral loan concentration may weaken stability of banks in the short run, it may however enhance the stability of banks in the long run through prolonged expert knowledge, experience and understanding of sectors. From these findings, policymakers, regulators and bank managers must not only develop and design policies and regulations that prohibit sectoral loan concentration but should also incorporate plans and policies that encourage banks to develop core competence and competitive advantage to take advantage of advancing bank stability through sectoral loan concentration. Keywords Stability, loan concentration, sectoral loans, banks, Ghana JEL Codes: G10; G18; G41 1 Introduction The important functions undertaken by banks cannot be overemphasized in both developing and developed economies. Banks generally perform the core functions of mobilising savings and allocating these funds mobilized through the administration of loans for effective and efficient use (Mishkin, 1999; Bain & Howells, 2009). Given the dynamism and ever increasing and fierce competition in the operations of banks, banks are pressured to develop new and innovative products and services and also take on more risky operations, which increase the risk appetite and risk-taking behaviour of banks leading to reduced bank stability. The issue of bank stability has received much attention especially after the 2007/2008 global financial crises, which was mainly caused by excessive risk-taking behaviour by banks and other financial institutions, leading to huge credit losses (Crotty, 2009) that destabilized most economies. Hence, the world’s financial sector has come under serious scrutiny due to the ability of the financial sector to make or unmake an economy, if the sector is not adequately regulated as learned from the 2007/2008 global financial crises. From a pure finance and economics perspective, banks are economic agents that have profit maximisation and survival objectives. Hence, banks employ all possible means (include crude means) to maximize their perfor- mance if they are not well monitored, accessed, evaluated and regulated. 68 Journal of Emerging Market Finance 19(1) Since banks amass majority of their revenue through their lending function, banks are attempted to concentrate their lending resources in pursuit of their survival and profit maximisation objectives. To prevent banks from lending all or majority of their funds to individuals or corporate entities, most countries have strict regulations on ‘single obligor limits’, which spells out the maximum amount a bank is allowed to lend to a single corpo- rate or individual borrower in relation to the shareholders fund of that bank (Morris, 2001; Basel Committee on Bank Supervision, 1991). This single obligor limit argues for diversification of loans. However, in sharp contrast, banks practise sectoral loan concentration, where they focus their lending functions in few sectors where they have developed competitive advantage over time (Ali, Intissar, & Zeitun, 2015). The effect of loan concentration on bank stability is explained by the concentration-fragility hypothesis (destabilisation effect) and concentration-stability hypothesis (stabilising effect) (see Ali et al., 2015). Specifically, while the concentration-stability hypothesis argues in support that concentration reinforces stability in banks due to increased core competences and experience, specialisation, information accessibility and profitability (see Berger, Klapper, & Turk- Ariss, 2009; Uhde & Heimeshoff, 2009; Boyd, De Nicolò, & Jalal, 2006), the concentration-fragility hypothesis argues that concentration detracts bank stability due to increased risk exposure (see Beck, Demirgüç-Kunt, & Levine, 2006; Allen & Gale, 2000). In recent times, studies (including Chen, Polemis, & Stengos, 2018; Polemis & Stengos, 2017; Dai, Liu, & Serfes, 2014) have confirmed non-linear (direct or inverted U-shape) relationships between concentration and financial performance variables giving rise to the exploration of non-linear investigations in the financial sector studies across the globe. In the light of the arguments above, it is surprising to find that very few studies (see Adzobu, Agbloyor, & Aboagye, 2017; Kamp, Porath, & Pfingsten, 2005; Ali et al., 2015; Morris, 2001) investigate sectoral loan concentration, hence informing why very few countries (including Bulgaria, Sweden) have policies on sector loan concentration (see Morris, 2001). Again, one may argue that this lack of policies or regulation on sectoral loan concentration may be attributed to the inconsistency in the arguments of concentration in the banking sector and hence it is difficult to inform policy direction. Hence, this study takes advantage of the limited empirical studies on sectoral loan concentration in Africa and uses bank data from Ghana to investigate the effects (linear or non-linear effects) of sectoral loan concentration on bank stability. One key feature distinguish- ing this study from other concentration and stability studies is that, this study employs four measures for bank sectoral loan concentration from Kusi et al. 69 a bank level following Kamp, Porath, and Pfingsten (2005). Ghana was selected for this study due to availability and accessibility of data. Also, Ghana is among the most promising countries in Africa that have the potential to develop and fairly has a greater number of banks from different countries, making it a suitable country to undertake such a study. Hence, this study investigates the linearity and non-linearity nexus of sectoral loan concentration on bank stability in Ghana, covering periods between 2007 and 2014, following Lind and Mehlum (2010) method of testing non-linearity. To achieve this purpose, the study employs four different measures for sectoral loan concentration, namely Hirshmann–Herfindahl Index (HHI), Shannon Entropy (SE), Absolute Distance Measure (ADM) and Relative Distance Measure (RDM) following Kamp, Porath, and Pfingsten (2005), while bank stability is measured with Z-score following Boyd, de Nicoló and Jalal (2006), Beck (2007) and Laeven and Levine (2009). The rest of the article is organized as follows: literature review, methodology, empirical results, robustness checks and diagnostics and conclusions and recommendations. 2 Literature Review and Hypothesis Development 2.1 Theoretical Review The literature on bank concentration and stability is explained by two theories: the traditional banking theory and the theory of corporate finance. While the traditional banking theory argues in favour of concentration- fragility hypothesis (supports diversification), the theory of corporate finance argues in favour of concentration-stability hypothesis. Specifically, the traditional banking theory states that banks must diversify their loan portfolio in order to expand their credit lines to reduce the probability of default (Diamond, 1984). That is, once banks expand their intermedia- tion function, they are able to gather more information which helps banks information asymmetry problem leading to reduction in adverse selection and moral hazard problems. Diversification, the creation of multiple activi- ties and internationalized banks can promote financial stability, as banks are less sensitive to national economic conditions. Basel Committee on Banking Supervision (1991) showed that loan concentration is associated with banking crises, which reduces the stability of banks. Likewise, Rossi, Schwaiger, and Winkler (2009) and Bebczuk and Galindo (2008) also provided evidence in Austria and Argentina in support of concentration- fragility hypothesis. Put simply, the traditional banking theory argues 70 Journal of Emerging Market Finance 19(1) that banks should not put or lay all their eggs in one basket. On the other hand, the theory of corporate finance supports the concentration-stability hypothesis, which argues that when banks concentrate their activities on specific sectors or groups of sectors, they are able to learn and benefit from expertise of the sector or sectors, which reinforce the stability of banks. That is, banks are able to learn over time and master the sector and hence accumulate competitive advantage, which helps them special- ize and better understand the sector, hence promoting bank stability due to improved knowledge base (Acharya, Hasan, & Saunders, 2006; Denis, Denis, & Sarin, 1997; Jensen, 1986). That is, diversification risk reduces risk from the modern portfolio theory. Contrary to the concentration- stability hypothesis and concentration-fragility hypothesis, some recent studies (Chen et al., 2018; Polemis & Stengos, 2017; Dai et al., 2014) show that concentration may not have a linear or monotone effect on per- formance as suggested by the two hypotheses discussed above. They argue that the concentration may have a non-linear effect on performance which takes the form of a direct or inverted U-shape. However, very few studies investigate the non-linear nexus between concentration and performance. 2.2 Empirical Review: Sector Loan Concentration and Performance Studies that investigate sectoral loan concentration are limited and scanty in number. We identify and review a few studies on sectoral loan concentration across the globe. For instance, Adzobu et al. (2017) investigated sectoral loan diversification (concentration) and its effect on bank profits and credit risk in Ghana between 2007 and 2014. Employing Prais–Winsten, fixed and random effect regressions, they found that loan portfolio diversification (or concentration) does not improve (or worsens) bank profitability nor does it reduce bank credit risk in Ghana. Similarly, Chen, Wei, Zhang and Shi (2013) also examined the effects of sectoral loan diversification (concentration) on bank returns and risk in the Chinese banking sector using 16 banks between 2007 and 2011. They find that sectoral loan diversification (concentration) is associated with reduced returns and risk, and therefore contradicts existing findings in developing and emerging economies. Furthermore, Tabak, Fazio and Cajueiro (2011) investigated loan portfolio concentration on bank return and risk using monthly banks on Brazilian banks between January 2003 and February 2009. They reported that loan portfolio concentration propels bank returns depending on the size Kusi et al. 71 of banks while foreign and public owned banks seem to have less effect on the degree of concentration. Also, Kamp et al. (2005) investigated whether or not bank loan concentration and diversification contradict the theory of financial intermediation using banks in German banking market between 1993 and 2002. Assume different measures of concentration including HHI and other less known distance measures of concentration, they found that the different measures of loan concentration yield conflicting results and conclude that reliance on only one measure of concentration may be misleading. They further showed that bank type including credit cooperatives, savings, regional and foreign banks significantly influence sectoral loan concentration. More so, Hayden, Porath, and Westernhagen (2007) employed indi- vidual bank loan portfolios of 983 German banks for the period from 1996 to 2002 to investigate the links between portfolio diversification and bank performance. Their findings suggest that there is little evidence that large performance benefits are associated with diversification, implying that loan portfolio concentration improves bank performance and subse- quently induce stability. Finally, Acharya et al. (2006) examined the effect of loan portfolio focus and diversification on the return and risk of 105 banks in Italy from 1993 to 1999. They found that diversification does not guarantee superior performance and bank safety. That is, diversification in less stable banks reduces bank return, while diversification in more stable banks leads to marginal improvement or inefficient risk-return trade-off. While none of the studies above focused on sectoral loan concentra- tion and bank stability, they rather focus on sectoral loan concentration and bank return and risk. Also, other studies examine bank stability and concentration in general without focusing on sectoral loan concentration. For instance, Boyd et al. (2006) investigated bank concentration and performance of banks in a sample of 134 countries over the period from 1993 to 2004. Their study shows that bank concentration improves earn- ing abilities of banks and reinforces stability. That is, banks’ probability of failure is positively related to concentration. Uhde and Heimeshoff (2009) employed banks across 25 European Union (EU) countries cov- ering the periods from 1997 to 2005. They provided empirical evidence that national banking market concentration has a negative effect on the financial soundness of banks in the European Union. They showed that Eastern Europe is characterized with lower levels of competition, diver- sification and has higher fraction of government-owned banks which are more prone to financial fragility. Winton (1999) investigated whether banks should concentrate on diversification of the activities. They found that diversification across loan 72 Journal of Emerging Market Finance 19(1) sectors helps most when loans have moderate downside risk and bank’s monitoring incentives are in doubt. Also, diversification may increase bank failure when loans have sufficiently high downside. Berger et al. (2009) showed that a positive relationship exists between bank concen- tration and stability from a risk channel point of view. They showed that the overall reduction in bankruptcy risks improves the market power of banks. They argued that banks will hold a larger capital base in order to be able to absorb or soak losses. Beck et al. (2006) investigated the impact of national bank concentration and regulation on systemic banking crises. Using banking data from 69 countries covering the period from 1980 to 1997, they found that banking crises is less likely in countries that have concentrated banking systems. This finding implies that concentration reinforces stability in banks. They report that banking regulations that hinder competition are associated with greater banking system fragility. Vives (2010) analysed the impact of market power and financial stabil- ity. Their findings suggest that market power (thus concentration) induces greater profits, which increases the capital of banks and subsequently improves their ability to absorb shocks in an instable financial framework. Further, the study highlighted that banks that are more concentrated are less prone to liquidity or macroeconomic shocks. Ali et al. (2015) analysed the relationship between banking concentration and financial stability in 156 countries during 1980 and 2011. Their findings suggest that concentra- tion does not have a direct effect on financial stability but goes through profitability and interest rate to reinforce and detract financial stability, respectively. They further show the existence of a stabilising effect of concentration on financial stability in developing countries. The empirical review shows clearly the limited empirical investigation on sectoral loan concentration and bank stability although exiting stud- ies on sector loan concentration focus on its effects on bank returns and credit risk. Thus, whether or not sectoral loan concentration propels bank stability remains unsearched to the best of our research capacities. Again, majority of these studies on sectoral loan concentration, which do not focus on banks in Africa and have examined concentration, assume it to have a linear or monotone effect on performance or risk, while evidence exists to support non-linearity nexus between concentration and performance. One may argue that the conflicting effect of loan concentration as per the empirical review can be attributed to the non-linearity nature of concen- tration on performance, which is less examined and also due to context differences within which the studies were done. From these discussions, we hypothesize that concentration should have a non-linear relationship with stability in the context of Ghana. Kusi et al. 73 3 Methodology This study examines the effect of sectoral loan concentration on bank stability in Ghana from 2007 to 2014. This study employs a panel of 30 banks in Ghana in a random effect, fixed effect and two-step generalised method of moments (GMM) estimations. Arellano and Bond (1991) and Arellano and Bover (1995) argued that the generalized least squares and ordinary least squares (OLS) may not adequately capture the speed of adjustment. That is, the random effect and ordinary least squares overesti- mate the speed of adjustment while the fixed effect model underestimates the speed of adjustment. Again, given the high probability that some of the independent variables may be correlated with current and past values of the idiosyncratic component of the error term, the GMM is employed and offers consistent estimates by making use of instruments that are obtained from the orthogonality conditions that exist between the error term and the lagged variable. Following Arellano and Bond (1991) and Arellano and Bover (1995), the use of the forward orthogonal GMM technique minimizes the gaps that missing observations present in the data set. Specifically, the study employs the two-step GMM instead of the one-step because the one-step GMM assumes no autocorrelation and het- eroscedasticity, which affect the consistency of the coefficients. However, the two step-GMM controls for autocorrelation and heteroscedasticity but its standard errors are not accurate and reliable (see Blundell and Bond, 1998); hence the study employs Windmeijer (2005) method to correct the standard errors of the two-step variant GMM. While the robust random and fixed effects offer the static view of the models estimated, the GMM offers a dynamic view which is more robust to omitted variable biases, autocorrelation and heteroscedasticity. The bank data were collected from the audited annual financial statements of bank in Ghana while the mac- roeconomic data were taken from World Development Indicators (WDI). The bank stability variable is selected from prior studies and included in the models which are estimated as follows: Bank Stability Models Z- SCOREit = b0+ b1Z- SCOREit-1+ b2ADMit+ b3CAPit + b4BSIZEit+ b5NPLRATIOit+ b6 INCDIVit IHHI LIQ MANAQUA (1)+ b7 it + b8 it+ b9 it + b10ROEit+ b11GDPGRTHit+ fit 74 Journal of Emerging Market Finance 19(1) Z- SCOREit = b0+ b1Z- SCOREit-1+ b2HHIit+ b3CAPit + b4BSIZEit+ b5NPLRATIOit+ b6 INCDIVit + b7 IHHIit + b8LIQit+ b9MANAQUA (2) it + b10ROEit+ b11GDPGRTHit+ fit Z- SCOREit = b0+ b1Z- SCOREit-1+ b2SEMit+ b3CAPit + b4BSIZEit+ b5NPLRATIOit+ b6 INCDIVit + b7 IHHIit + b8LIQit+ b9MANAQUA (3) it + b10ROEit+ b11GDPGRTHit+ fit Z- SCOREit = b0+ b1Z- SCOREit-1+ b2RDMit+ b3CAPit + b4BSIZEit+ b5NPLRATIOit+ b6 INCDIVit + b7 IHHIit + b8LIQit+ b9MANAQUA (4) it + b10ROEit+ b11GDPGRTHit+ fit 3.1 Variable Definition and Selection 3.1.1 Bank Stability (Z-Score) Z-Score is the dependent variables measuring bank stability and is computed as capital adequacy ratio plus return on assets, all divided by standard deviation of return on assets (Boyd et al., 2006; Beck, 2007). The resulting ratio gives an indication of how far a bank is away from insolvency or indicates the likelihood of bank failure, thus measuring how stable a bank is. Following prior studies (see Boyd et al, 2006; Laeven & Levine, 2009), the z-score is justified as a bank stability measure and interpreted to mean as the number of standard deviations by which returns would have to fail from the mean to wipe out all the equities of a bank (Boyd & Runkle, 1993). This measure provides an indication of the safety and a direct measure of bank soundness communicates stability of banks. Higher values of the z-score indicate improved solvency and less likeli- hood of bank failure, hence financial stability. Due to high skewness and scale biases, the study takes the natural log of the z-score. Z-SCOREit-1 is the past year value of Z-Score which is used in the GMM estimation to control for endogeneity. 3.1.2 Sectoral Loan Concentration (ADM, HHI, SE, RDM) We measured sectoral concentration by employing two traditional meas- ures of concentration, namely the HHI and the SE. In addition to the two Kusi et al. 75 traditional measures, two distance measures, namely the ADM (Da) and RDM (Dr), were used to measure sectoral concentration in bank loan portfolios. To compute the HHI, the relative exposures (ri) of the bank b at time t to each economic sector i is first computed and is given as: Nominal Exposure r btibti = Total Exposure bt The HHI of bank b at time t is defined as: n HHI 2bt =| rbti i= 1 The possible values of the index are given by 1n # H # 1 where n is the number of sectors. The lower limit of the HHI is 1n and represents a perfectly diversified portfolio. HHI assumes that perfect diversification means an equal exposure of the bank to each economic sector (i). If the HHI is 1, the bank is perfectly specialized (concentrated) in one sector. The higher the HHI value, the higher the sectoral concentration in the credit portfolio. SE is an effective instrument used to indicate the variety of distributions at a given point in time and can be applied to measure industrial concentration or corporate diversification (Kamp et al., 2005). This measure is calculated by: n SEMbt =-| r 1bti $ lnc r m i= 1 bti If SEM is equal to 0, the loan portfolio is highly concentrated and the bank lends to only one industry. If SEM is equal to – ln(n), the bank’s portfolio is perfectly diversified. Distance measures can be used to measure concentration in loan portfolios by quantifying the distance between a bank’s loan portfolio (r) and the benchmark’s loan portfolio(x), thus larger values mean less diversification (more concentration) (Rudolph & Pfingsten, 2005). The benchmark loan portfolio is the industry composition of the economy’s loan market portfolio. The distance measures Da and Dr are calculated by: n D 1a(r, x) bt = 2| | rbti- xbti | i= 1 n | r - x | D (r, x) 1 bti btir bt = n| i r + x = 1 bti bti 76 Journal of Emerging Market Finance 19(1) ADM (Da) is the normalized maximum absolute difference between a bank’s portfolio (r) and the benchmark portfolio(x). RDM (Dr), on the other hand, is the normalized relative difference between a bank’s portfolio (r) and the benchmark portfolio(x). A key property of the RDM is that it takes into consideration the size of the economic sectors (Kamp et al., 2005). The values of the distance measures range between 0 and 1, where 0 means perfect diversification (less concentration) and 1 means perfect concentration (less diversification). Kamp et al. (2005) showed that these statistical measures of concentration may produce contradictory results and therefore in this research we employed both measures to compare them. These distance measures have also been employed recently by Tabak et al. (2011), and all measures pointed to a similar conclusion that the loan portfolios of banks in Brazil are moderately concentrated. 3.1.3 Capital Adequacy Capital adequacy is measured as equity to assets and is an indication of how many of a bank’s assets are funded with owner’s equity. It also shows the ability of a bank to absorb credit losses, therefore enhancing the bank stability. From a risk-return hypothesis, increase in owner’s equity improves the resilience of banks and hence reinforces bank stability (Beck, Jonghe, & Schepens, 2013; Berger, 1995). Put differently, increase in owner’s equity induces shareholders to be more vigilant and scrutinizes the management of the bank to avoid unnecessary risk-taking behaviour, which improves stability in banks. Hence, we anticipate a positive relation between stability and capital adequacy. 3.1.4 Bank Size Bank size (BSIZE) is measured using natural log of total assets. From the economies of scale and diseconomies of scale theories, the expected relationship between bank size and stability could either be positive or negative. While economies of scale argue that larger banks benefit from the size to increase the operation cost efficient which reinforces stability, the diseconomies of scale argue that large banks are associated with dupli- cation of functions, bureaucracies in operations, laxities in monitoring, and supervision increases the chance of bank failure. Hence the effect of bank size is not straightforward. 3.1.5 Nonperforming Loans Nonperforming loans (NPLRATIO) is used to measure bank credit risk. It is measured as nonperforming loan to gross loans and advances. It meas- ures past years’ credit losses and is expected to reduce the soundness and Kusi et al. 77 stability of banks (Castro, 2012; Chaibi & Ftiti, 2015). Following Adrian and Shin (2008), the traditional view of banking is that credit risk, which is the bedrock of nonperforming loans, is a major stability in the banking sector; hence credit risk has the potential to influence stability of banks. That is, increase in nonperforming loans erodes the capital base of the bank, which weakens the financial strength of a bank and the ability of the bank to absolve shocks leading to instability. Hence nonperforming loans weakens the stability of banks, implying a negative relationship between credit risk and stability. 3.1.6 Bank Management Quality Bank Management Quality (MANAQUA) is employed and measured as operating expenses divided by total assets (see Athanasoglou, Brissimis, & Delis, 2008; Naceur & Orman, 2011) and is used to measure operational cost. Following Athanasoglou et al. (2008), lower operating expenses is an indication of higher bank management efficiency, which leads to an increase in efficiency and consequently improves the stability which is also an indication of lower risk-taking behaviour. However, following Naceur and Orman (2011), bank management spend to ensure banks survive and are stable; hence increase in operational cost fosters bank stability. Hence, the expected sign between management quality and bank stability could be either positive or negative. 3.1.7 Income Diversification Non-interest income is used for proxy diversification, which aims at reducing risk and hence improved bank stability following the modern portfolio theory and the arbitrage pricing theory. Corporate finance literature (see Ross, Westerfield, Jaffe, & Jordan, 2008) suggests that diversification reduces risk and at the same time improves the return of investors, which leads to stability in banks. However, following Stiroh Stiroh and Rumble (2006), highlights the dark side of diversification, arguing that diversification may lead to increased instability when manage- ment diversify into areas where they lack core competence and competi- tive advantage. Nevertheless, we expect a positive relationship between diversification and stability. 3.1.8 Concentration To use industry-level concentration (IHHI) in this study, the HHI on loan is employed, and it is measured as the sum of squares of loans and advances. This study consequently monitors how the changes in the banking market structure (using loan concentration) have impacted on 78 Journal of Emerging Market Finance 19(1) bank stability. The relationship between bank market structure and per- formance could be positive or negative (see Beck et al., 2006; Schaeck & Cihak, 2010; Keeley, 1990). This is explained as a concentrated banking sector is regarded as less efficient and hence increases credit risk due to inefficiency. However, a low concentrated banking sector is shown to increase efficiency and hence boost stability of banks (see Turk-Ariss, 2010; Schaeck & Cihak, 2010). 3.1.9 Liquidity (LIQ) Bank liquidity is used to measure the solvency of banks and measured as the ratio of loans to total assets. Thus, liquidity measures the ability or capacity for banks to honour their short-term financial obligations like honouring deposits on demand, implying that liquid banks are less likely to be unstable. For instance, Adrian and Shin (2008) stated that the global financial crises during 2007–2008 showed that bank instability can be fuelled by the level of illiquidity of banks, hence attributing that liquidity can influence stability in the banking sector. The expectation is that liquid- ity firms are more able to pay off their debts as and when they fall due, hence reinforcing the stability (Ross et al., 2008). Put differently, liquid firms are able to settle their debts, which reduces the debt obligations and enhances the stability of that liquid firm. 3.1.10 Profitability (ROE) Profitability is computed as the ratio of net profit to total equity. The literature (see Brealey & Myers, 2003; Ross et al., 2008) argues that profit- ability banks attract goodwill and improve public corporate image, hence inducing sustainability and continuity in the operations of the bank. This therefore enhances the stability of banks and lowers risk-taking behaviour in banks. Put differently, the study argues that profitable banks are more likely to be stable and less likely to assume risky banking activities. That is, profitable banks have the ability to absorb credit losses. 3.1.11 Gross Domestic Product Growth Rate Gross domestic product growth rate is a macroeconomic indicator for improvement in economic conditions and standard of living of citizens. The variable is measured as the year on year changes in gross domestic product. Following studies such as Fofack (2005), Jiménez, Lopez, and Saurina, 2009; Rajan and Dhal (2003) which argue that an improvement in economic conditions improves the loan repayment lead- ing to lower credit risk exposure. Hence, this leads to banks’ stability in the credit market. A summary of all the variable measurements, source, indicator and expected signs are presented in Table 1. Table 1. Summary of Variable Definition and Measurement Variable Measurement Indicator Source Expected Sign Dependent variable Z-score [capital Adequacy+ ROA] Bank stability Computed by author based on data from vOA Bank of Ghana Independent variables: Corporate governance structures ADM 1 n| Bank concentration Computed by author based on data from +/–| r - x | 2 bti btii 1 auditor bank statements= HHI n| Bank concentration Computed by author based on data from +/–r2 i 1 auditor bank statements= SE n| 1 Bank concentration Computed by author based on data from +/–- rbti * ln[ r ] i 1 bti auditor bank statements= RDM 1 n | r - x | Bank concentration Computed by author based on data from +/–| bti btin | r + x | auditor bank statementsi=1 bti bti Independent variables: Control variables CAP total equity Capitalization Computed by author based on data from + total Assets auditor bank statements BSIZE ln( total assets) Bank size Computed by author based on data from + auditor bank statements NPLRATIO nonperfor min g loans Credit risk Computed by author based on data from + gross laons auditor bank statements (Table 1 continued) (Table 1 continued) Variable Measurement Indicator Source Expected Sign Dependent variable INCDIV nonoperating income Diversification Computed by author based on data from [+/–] total operating income auditor bank statements IHHI 1 n| Industry concentration Computed by author based on data from [–]2n l i auditor bank statements=1 LIQ loans Liquidity Computed by author based on data from [+] total assets auditor bank statements MANAQUA operating income management quality Computed by author based on data from [–/+] total operating income auditor bank statements ROE net income Profitability Computed by author based on data from + total equity auditor bank statements GDPGRTWH Current GDP - Previous GDP Economic condition World Development Indicators database +/– Previous GDP Source: Computed by authors based on prior studies. Kusi et al. 81 4 Empirical Results Table 2 presents the summary statistics, variance inflation factor (VIF) and normality of the variables employed in this study. From the summary statistics, outliers which have the possibility to influence the consistency, efficiency and biasedness of coefficients were not observed in the data set. Shapiro–Wilk’s normality test is used to test for the normality of the data while the VIF is used to test for the acceptability of each variable in the model. Thus, Shapiro–Wilk’s normality test has a null hypothesis of as normal distribution was rejected for all the variables, indicating that the variables were all normally distributed around their means. With the maximum acceptable VIF being 10, all the variables are below the threshold of 10, indicating that all the variable are fit to be in the model. Z-score measures how far a bank is away from insolvency or indicates the likelihood bank failure. Given its average value of 17.50, it indicates that on an average, banks in Ghana during 2007–2014 were 17.50 points away from bank failure. This is an indication that the banks are not too far away from bank failure. HHI on the average is 0.29. Since HHI value close to 1 implies that bank loans are concentrated in few sectors, the average Table 2. Summary of Statistics Variable Obs. Mean Std. Dev. Min. Max. SWILK VIF Z-Score 199 2.433 1.009 –1.647 4.658 2.935*** – ADM 187 0.360 0.163 0.033 1.000 5.489*** 2.25 SQADM 187 0.156 0.155 0.001 1.000 8.351*** 1.31 HHI 187 0.291 0.148 0.000 0.995 8.038*** 4.2 SQHHI 187 0.106 0.140 0.000 0.989 9.537*** 1.3 RDM 187 0.466 0.169 0.014 0.956 3.656*** 3.6 SQRDM 187 0.245 0.181 0.000 0.913 7.009*** 1.66 SEM 187 –1.506 0.386 –2.085 –0.017 6.73*** 6.72 SQSEM 187 2.416 0.956 0.000 4.349 3.657*** 1.48 CAP 201 0.166 0.124 0.009 0.842 8.62*** 2.29 BSIZE 201 20.126 1.114 16.204 22.458 2.952*** 2.56 NPLRATIO 179 0.110 0.093 0.000 0.450 6.371*** 1.3 INCDIV 201 0.389 0.257 0.020 3.520 10.192*** 1.28 INDHHI 210 0.074 0.019 0.059 0.113 7.846*** 3.01 LIQ 201 0.439 0.148 0.030 0.768 1.491* 1.58 MANQUA 201 0.651 0.521 0.092 6.907 10.49*** 1.86 ROE 201 0.127 0.397 –4.525 0.511 10.265*** 2.02 GDPGRTH 240 0.076 0.031 0.040 0.140 6.401*** 1.76 Source: Computed by authors using Stata 13. Notes: ***, ** and * shows significance at 1%, 5% and 10% level, respectively. 82 Journal of Emerging Market Finance 19(1) value of HHI of 0.29 indicates that bank loans are more diversified across sectors. The SE is on the average –1.51, implying that banks’ sectoral loans are well diversified across sectors since the SE value of zero indicates concentration of sectoral loans. Also, the absolute (ADM) and the relative (RDM) distance measures are on an average 0.36 and 0.47, respectively. Given that higher values of these variables are an indication of sectoral loan concentration, it is evident that sectoral loans are not concentrated in few sectors. Capital adequacy is on average 16.59 per cent, indicating that owner equity or contribution constitutes about 17 per cent of the total financing of banks. Npl ratio is a measure for credit risk which is about 11 per cent of total loans and advances, indicating that credit risk is quite high in the Ghanaian banking sector. Income diversification, which is used to measure diversification, is on average 39 per cent of total income. That is, banks on the average generate 39 per cent of the income from other alternatives sources other than their core income source (interest income). Industry HHI is on average 7.44 per cent, implying that concentration in the Ghanaian banking sector is very low. From the summary statistic table, it is observed that banks are very liquid on the average given the liquidity value of 44 per cent of total assets. Management quality, which represents operational cost management, is 65 per cent of total income, indicating that operating cost constitutes 65 per cent of total income. This means that operating cost is very high in the Ghanaian banking sector. Return to shareholders (ROE) is averagely about 13 per cent of total equity while gross domestic growth rate is about 8 per cent. These are indications that ROE and economic growth are moderately okay. Table 3 shows the correlation matrix between the variables that are employed in this study. Pearson’s correlation matrix that serves as a mechanism for checking and controlling multicollinearity is shown in Table 3. Following Kennedy (2008), the study sets the multicollinearity threshold between two independent variables to 0.7. Hence, the results presented in Table 3 shows no evidence of multicollinearity. Table 4 presents two step GMM regression results on the effect of sector loan concentration on bank stability in Ghana during 2007–2014. Thus, the table presents results on linearity and non-linearity nexus between sectoral loan concentration and bank stability in Ghana. The table reports eight different models to estimate the linearity and non-linearity effects of sector loan concentration on bank stability. In models 1, 3, 5 and 7 the direct linear relationship of sectoral loan concentration including ADM, HHI, SE and RDM are examined on bank stability while in models 2, 4, 6 and 8 the non-linear relationship (U-shape or inverted U-shape) of sectoral Table 3. Pearson’s Correlation Matrix NPL- INC- IND- MAN- GDP- Z-score ADM SQADM HHI SQHHI RDM SQRMD SEM SQSEM CAP BSZIE RATIO DIV HHI LIQ QUA ROE GRTH Z-score 1 ADM 0.0447 1 SQADM 0.0766 0.9624 1 HHI 0.0788 0.5782 0.5564 1 SQHHI 0.0915 0.5478 0.5433 0.9643 1 RDM 0.1162 0.7151 0.6153 0.6433 0.605 1 SQRDM 0.1526 0.7089 0.6416 0.6654 0.653 0.9746 1 SEM 0.1178 0.6194 0.5777 0.8414 0.816 0.768 0.796 1 SQSEM –0.1146 –0.6111 –0.5528 –0.8288 –0.7441 –0.773 –0.7712 –0.9708 1 CAP 0.3006 0.3428 0.3487 0.3005 0.3475 0.3806 0.4415 0.3633 –0.3123 1 BSIZE –0.0146 –0.0776 –0.1184 –0.1496 –0.2264 –0.1926 –0.2634 –0.2135 0.1371 –0.436 1 NPLRATIO –0.2411 –0.1393 –0.1222 –0.1746 –0.1818 –0.2412 –0.233 –0.1982 0.1898 0.1113 0.0285 1 INCDIV –0.3308 –0.1337 –0.1234 –0.2601 –0.2511 –0.263 –0.2887 –0.3073 0.3012 –0.2672 0.017 –0.0693 1 INDHHI –0.1494 –0.0818 –0.098 0.0957 0.1103 0.0133 0.0029 0.0735 –0.0568 –0.2183 –0.4109 –0.2246 0.0723 1 LIQ 0.0515 –0.4135 –0.4055 –0.3603 –0.3285 –0.4514 –0.4617 –0.4502 0.4537 –0.452 0.215 –0.0549 –0.0188 0.174 1 MANQUA –0.3372 –0.147 –0.1153 –0.1237 –0.1139 –0.1618 –0.1411 –0.1249 0.1207 –0.13 –0.2286 –0.0342 0.7759 0.1175 –0.0926 1 ROE 0.4255 0.0354 0.0268 0.0696 0.0387 0.0842 0.0553 0.0896 –0.12 0.0099 0.2492 –0.1562 –0.1682 0.0323 0.0811 –0.374 1 GDPGRTH 0.0014 0.1135 0.1343 0.0231 0.0071 0.0579 0.0744 0.0366 –0.0437 0.0951 0.0033 0.0871 0.0565 –0.6337 –0.126 0.0008 –0.0348 1 Source: Computed by authors using Stata 13. Notes: ***, ** and * shows significance at 1%, 5% and 10% level, respectively. Table 4. GMM Regression—Effect of Sector Loan Concentration on Bank Stability Variables Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 L.Z-score 0.499*** 0.448*** 0.458*** 0.407*** 0.482*** 0.536*** 0.481*** 0.516*** (0.127) (0.0978) (0.115) (0.133) (0.115) (0.112) (0.107) (0.122) ADM –0.610** –1.817 (0.262) (1.277) SQADM 1.299 (1.122) HHI –0.408* 0.668 (0.216) (1.630) SQHHI –1.154 (2.134) RDM –0.119 0.239 (0.578) (1.395) SQRDM –0.556 (1.150) SEM –0.107 –1.353** (0.216) (0.649) SQSEM –0.471* (0.235) CAP 3.841* 3.831** 3.873*** 4.485*** 3.691** 3.565** 3.743** 3.321** (1.909) (1.493) (1.343) (1.529) (1.560) (1.575) (1.400) (1.527) BSIZE –0.0293 –0.0335 –0.00905 –0.0159 –0.0422 –0.0690 –0.0330 –0.105 (0.130) (0.111) (0.0752) (0.0702) (0.108) (0.0981) (0.0924) (0.0862) (Table 4 continued) (Table 4 continued) Variables Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 NPLRATIO –1.684* –2.007 –1.827* –1.606 –1.647* –1.973* –1.648* –1.734** (0.862) (1.239) (1.066) (1.016) (0.883) (1.104) (0.876) (0.761) INCDIV 0.244 0.0605 0.322 0.401 0.128 0.185 0.162 0.286 (0.405) (0.403) (0.288) (0.305) (0.366) (0.323) (0.342) (0.340) INHHI –13.38 –12.78 –11.65** –10.48** –13.03* –12.97* –12.46* –12.83** (9.305) (8.166) (5.424) (4.777) (7.636) (6.554) (6.399) (4.968) LIQ 1.024* 0.860 0.827** 0.748 1.023* 0.916 0.931* 0.895* (0.515) (0.574) (0.368) (0.557) (0.596) (0.549) (0.508) (0.464) MANQUA 1.233* 1.370** 1.596*** 1.722** 1.278* 1.317** 1.342** 1.274** (0.609) (0.617) (0.552) (0.690) (0.653) (0.591) (0.605) (0.618) ROE 1.534*** 1.630*** 1.766*** 1.911*** 1.626*** 1.545*** 1.652*** 1.670*** (0.407) (0.403) (0.405) (0.587) (0.438) (0.347) (0.408) (0.456) GDPGRTH –2.075** –1.265 –2.005** –2.386** –2.138** –1.737* –2.143** –2.239** (0.891) (1.109) (0.857) (0.973) (0.952) (0.848) (0.854) (0.910) Constant 1.289 1.693 0.587 0.401 1.448 1.886 0.991 1.745 (3.534) (3.069) (2.105) (1.852) (3.007) (2.616) (2.407) (2.231) Observations 128 128 128 128 128 128 128 128 Number of bank code 27 27 27 27 27 27 27 27 Instrument 31 32 31 32 31 32 31 32 Sargan 39 41.08 48 51.08 45 43.66 43 43.88 Prob > |2 0.0050*** 0.002 0.0000*** 0.000*** 0.0010*** 0.001 0.0010*** 0.001 HANSEN 14.3700 13.04 10.6400 11.20 11.3000 10.57 12.1500 11.30 (Table 4 continued) (Table 4 continued) Variables Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Prob > |2 0.7620 0.837 0.9350 0.917 0.9130 0.937 0.8790 0.913 AR1 –1.6300 –1.71 –1.5300 –154 –1.5400 -1.57 –1.6000 –1.63 Prob > z 0.1040 0.088 0.1250 0.124 0.1240 0.116 0.1090 0.102 Adj. R2 0.5600 0.78 0.7100 0.16 0.7300 0.58 0.7400 0.67 Prob > z 0.5760 0.438 0.4760 0.488 0.4680 0.562 0.4580 0.503 F(11, 26) 24.6600 36.99 61.9700 38.81 60.8800 55.26 62.6400 51.61 Prob > F 0.000*** 0.000*** 0.0000*** 0.000*** 0.0000*** 0.000*** 0.0000*** 0.000*** Inflection (T-value) 0.75 0.41 0.16 1.38* Source: Computed by authors using Stata 13. Notes: Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Kusi et al. 87 loan concentration including ADM, HHI, SE and RDM are examined on bank stability. In examining non-linearity, we follow Lind and Mehlum’s (2010) approach to test the existence of the non-linearity relationship between bank stability and sectoral loan concentration. First, we test the linear or monotone relationship between sectoral loan concentration on bank stability in models 1, 3, 5 and 7. However, we find that two out of the four sectoral loan concentration variables (ADM and HHI) were negatively and significantly related to bank stability, imply- ing that increase in sectoral loan concentration (measured with HHI and ADM) reduces bank stability. This finding is consistent with the traditional banking theory which argues loan concentration reduces stability of banks. This implies that banks should not put or lay all their eggs in one basket (Winton, 1999) because as banks expand their inter- mediation function, they are able to gather more information which helps banks’ information asymmetry problem, leading to reduction in adverse selection and moral hazard problems. Put differently, diversification of loan portfolio of banks create multiple activities and internationalized banks which can promote financial stability, as banks are less sensitive to sectoral conditions. However, testing the non-linearity relationship (in models 2, 4 and 6), we find that there is no significant relationship between bank stabil- ity and sectoral loan concentration and its square except for the case of SE. However, the non-linearity test (using Lind and Mehlum, 2010) as given by the inflection point with null hypothesis of monotone or direct U-shaped relationship and alternate hypothesis of inverted U-shape; we fail to reject the null hypothesis and conclude that the relationship between sectoral loan concentration and bank stability is either monotone or direct U-shape. This finding appears to support the earlier finding that the effect of sectoral loan concentration on bank stability is monotone. Surprisingly in model 8 both SE and its squared significantly weaken stability of banks, while its inflection point considering the non-linearity rejects the null hypothesis of monotone or direct U-Shape relationship and concludes that the nexus between bank stability and concentration is an inverted U-shape. The findings at this point seem to be inconsistent but tilt more towards monotone relationship giving the results of the GMM. However, one may argue that the relationship may depend on the measurement of the sectoral loan concentration. In an attempt to resolve the inconsistency, we further estimate the relationship using robust random and fixed effect regressions (Table 5 and 6). Following the random and fixed effect results, the linear relation- ships of sectoral loan concentration using four different results are not Table 5. Fixed Effect—Effect of Sectoral Loan Concentration on Bank Stability Variables Model 9 Model 10 Model 11 Model 12 Model 13 Model 14 Model 15 Model 16 ADM –0.154 –0.280 (0.173) (0.493) SQADM 0.134 (0.436) HHI –0.214 –1.495*** (0.181) (0.476) SQHHI 1.432** (0.537) RDM –0.237 –1.797** (0.281) (0.755) SQRDM 1.621** (0.714) SEM –0.0766 0.837*** (0.0926) (0.280) SQSEM 0.363*** (09988) CAP 4.385*** 4.387*** 4.375*** 4.384*** 4.373*** 4.234*** 4.354*** 4.400*** (1.040) (1.050) (1.064) (1.058) (1.044) (1.000) (1.049) (1.040) BSIZE –0.0402 –0.0407 –0.0474 –0.0551 –0.0465 –0.0347 –0.0460 –0.0612 (0.0480) (0.0482) (0.0460) (0.0444) (0.0520) (0.0482) (0.0465) (0.0468) NPLRATIO 0.289 0.282 0.246 0.197 0.237 0.201 0.267 0.169 (0.590) (0.593) (0.568) (0.519) (0.573) (0.542) (0.581) (0.510) (Table 5 continued) (Table 5 continued) Variables Model 9 Model 10 Model 11 Model 12 Model 13 Model 14 Model 15 Model 16 INCDIV 0.221 0.219 0.191 0.160 0.189 0.187 0.190 0.103 (0.163) (0.163) (0.144) (0.140) (0.149) (0.155) (0.147) (0.137) INHHI –8.025*** –8.037*** –8.026*** –8.543*** –7.984*** –8.071*** –8.055*** –9.140*** (2.441) (2.451) (2.487) (2.355) (2.511) (2.304) (2.460) (2.292) LIQ 0.672*** 0.666*** 0.695*** 0.666*** 0.627** 0.575** 0.665*** 0.655*** (0.235) (0.239) (0.232) (0.209) (0.236) (0.232) (0.232) (0.188) MANQUA –0.0286 –0.0275 –0.0518 –0.0552 –0.0465 –0.0852 –0.0355 –0.0289 (0.313) (0.316) (0.315) (0.337) (0.306) (0.301) (0.314) (0.333) ROE 1.245** 1.247** 1.229** 1.226** 1.233** 1.252** 1.234** 1.234** (0.525) (0.528) (0.543) (0.559) (0.521) (0.500) (0.540) (0.545) GDPGRTH –1.429** –1.420** –1.480** –1.484** –1.391** –1.543** –1.475** –1.531*** (0.567) (0.556) (0.582) (0.567) (0.538) (0.612) (0.578) (0.546) Constant 2.748** 2.786** 2.929** 3.374*** 2.977** 3.156** 2.727** 3.653*** (1.276) (1.266) (1.215) (1.161) (1.400) (1.323) (1.206) (1.192) Observations 154 154 154 154 154 154 154 154 R2 0.778 0.778 0.779 0.787 0.779 0.792 0.777 0.791 Number of bank code 30 30 30 30 30 30 30 30 F(11,29) 26.01*** 38.05*** 33.55*** 49.70*** Inflection 1.05 2.09** 1.91* 2.98*** Source: Computed by authors using Stata 13. Notes: Robus standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. Table 6. Random Effect—Effect of Sectoral Loan Concentration on Bank Stability Variables Model 17 Model 18 Model 19 Model 20 Model 21 Model 22 Model 23 Model 24 ADM –0.154 –0.344 (0.175) (0.485) SQADM 0.203 (0.417) HHI –0.199 –1.414*** (0.182) (0.473) SQHHI 1.357** (0.543) RDM –0.222 –1.826** (0.277) (0.767) SQRDM 1.660** (0.715) SEM –0.0689 0.773*** (0.0883) (0.279) SQSEM 0.336*** (0.101) CAP 4.343*** 4.344*** 4.331*** 4.340*** 4.330*** 4.189*** 4.312*** 4.354*** (1.011) (1.020) (1.032) (1.024) (1.013) (0.966) (1.019) (1.007) BSIZE –0.0369 –0.0375 –0.0430 –0.0486 –0.0425 –0.0297 –0.0416 –0.0532 (0.0492) (0.0495) (0.0462) (0.0467) (0.0523) (0.0489) (0.0469) (0.0490) NPLRATIO 0.245 0.234 0.206 0.168 0.197 0.168 0.225 0.147 (0.595) (0.598) (0.575) (0.533) (0.579) (0.547) (0.588) (0.526) (Table 6 continued) (Table 6 continued) Variables Model 17 Model 18 Model 19 Model 20 Model 21 Model 22 Model 23 Model 24 INCDIV 0.214 0.210 0.184 0.156 0.183 0.181 0.183 0.105 (0.170) (0.171) (0.152) (0.146) (0.156) (0.160) (0.154) (0.144) INHHI –8.037*** –8.047*** –8.027*** –8.453*** –7.999*** –8.046*** –8.065*** –8.959*** (2.440) (2.449) (2.491) (2.426) (2.500) (2.291) (2.466) (2.377) LIQ 0.686*** 0.677*** 0.707*** 0.679*** 0.643*** 0.593*** 0.680*** 0.670*** (0.230) (0.234) (0.230) (0.207) (0.231) (0.225) (0.228) (0.186) MANQUA 0.00266 0.00536 –0.0173 –0.0193 –0.0137 –0.0523 –0.00226 0.00680 (0.307) (0.311) (0.307) (0.327) (0.300) (0.295) (0.307) (0.324) ROE 1.276** 1.279** 1.261** 1.258** 1.264** 1.282*** 1.266** 1.266** (0.517) (0.519) (0.534) (0.548) (0.513) (0.492) (0.530) (0.534) GDPGRTH –1.432** –1.418*** –1.483*** –1.477*** –1.402*** –1.550** –1.481*** –1.516*** (0.556) (0.544) (0.572) (0.566) (0.528) (0.604) (0.568) (0.549) Constant 2.656** 2.708** 2.804** 3.179** 2.861** 3.024** 2.622** 3.397** (1.348) (1.337) (1.256) (1.263) (1.442) (1.380) (1.273) (1.322) Observations 154 154 154 154 154 154 154 154 R2 0.7776 0.7777 0.7785 0.7864 0.7783 0.7917 0.7769 0.7912 Number of bank code 30 30 30 30 30 30 30 30 F(11,29) 252.70*** 260.47*** 303.18*** 429.64 328.28*** 361.13*** 280.51*** 571.54*** Inflection 0.15 1.95** 2.02** 2.77*** Source: Computed by authors using Stata 13. Notes: Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. 92 Journal of Emerging Market Finance 19(1) significant in models 9, 11, 13 and 15. However, the non-linearity test (Lind & Mehlum, 2010) as given by the inflection point with null hypoth- esis of monotone or inverted U-Shape relationship and alternate hypothesis of direct U-shape; we reject the null hypothesis and conclude that the relationship between sectoral loan concentration and bank stability is direct U-shape. This finding appears to support the earlier findings (see Chen, Polemis, & Stengos, 2018; Polemis & Stengos, 2017; Dai, Liu, & Serfes, 2014) that argue in favour of non-linearity among financial and economic variables. Thus, the relationship between bank stability and sectoral loan concentration goes beyond monotone relationship and reflects a direct U-shape and inverted U-shape depending on the sectoral loan concentration variable being used. Therefore, the findings of this study to a large extent support the existence of direct U-shape relationship between sectoral loan concentration and bank stability. We argue that sectoral loan concentration may reduce stability at the initial stage but further increases or promotes bank stability. Thus, as banks concentrate their loans in fewer sectors, they learn and gain experience through the learning curve which boosts their understanding of the patterns, cycles, conditions and players of the sector, hence giving the competitive advantage as to how to best handle loans advanced to players in that sector, which helps avoid credit losses and boosts stability of banks. Hence, this finding contributes to the existing empirical literature on stability and sectoral loan concentration by showing that the relationship between bank stability and sectoral loan concentration goes beyond a linear relationship where sectoral loan con- centration weakens stability and also presents a non-linear direct U-shape relationship between stability and sectoral loan concentration using data from Africa, especially Ghana. Additionally, on the control variables, capital adequacy from the results improves bank stability as the coefficient of capital adequacy is positive and significantly related to bank stability across all the four models. From a risk-return hypothesis, capital adequacy enables banks to absorb or soak up credit losses leading to improved bank resilience and reinforces bank stability. Furthermore, bank credit risk measured with nonperform- ing loans ratio is generally negatively and significantly related to bank stability. This implies that bank credit risk reduces the stability of banks by weakening the ability of the bank to absolve shocks, leading to insta- bility. This finding is consistent with the findings of Castro (2012) and Chaibi and Ftiti (2015). Industry loan concentration measured with IHHI is employed as a proxy for banking market structure. The results suggest that the industry market structure or industry loan concentration of the banking sector Kusi et al. 93 derails bank stability. That is, as bank loans concentrate in the industry in few banks, this leads to reduction in the stability of banks. This finding supports the diversification concept preached by the risk-return theories that concentration leads to poor monitoring and supervision of loans, which propels bad loans, leading to weakened stability. Again, bank liquidity from the results indicates a positive and significant effect on bank stability across all the four models estimated. This implies that solvency of banks leads to improved bank stability. This finding is consistent with corporate finance literature in literature on bank run and liquidity risk (Cai & Zhang, 2017), which argues that solvent or liquid firms have the ability to retire their debt obligations as and when they fall due, hence boosting public confidence in the bank and reinforcing stability. Also, management quality, which measures the operational expenses of banks, reports a positive and significant effect on bank stability. This implies that banks spend on their operations to ensure the survival and stability of the bank. Again, this finding supports Naceur and Orman (2011) who find that increase in operational cost is linked to ensuring bank survival and stability of banks. Moreover, as expected, bank profit- ability is positively and significantly related to bank stability across all the models. This is an indication that profitable banks have strong financial muscles and are less prone to financial distress, hence profitability rein- forces stability. Following Margaritis & Psillika (2010), profitable firms create goodwill and good corporate image, which leads to sustainability and continuity in the operations of banks, resulting in stable operations of banks. Surprisingly, economic growth employed to provide an indication of the effect of state of the economy is negatively and significantly related to the bank stability across all the models. This means that, contrary to our expectation that improved economic growth leads to bank stability, economic growth rather leads to the reduced bank stability. However, following the pecking order theory and loan growth theory (Keeton, 1999), banks assume more risk during periods of improved economic growth because banks relax the terms and conditions regarding lend- ing in order to attract bank clients, since bank clients rely on their own financial resources and borrow in a booming economy, which is more costly compared to using their own financial resources. By relaxing the terms and conditions, even bank clients who will not be able to pay back are attracted leading to high credit risk and hence reduced bank stability. This finding is similar to the finding of Kusi, Agbloyor, Fiador and Osei (2016) who studied bank credit risk in Ghana and found that economic growth worsens credit risk in Ghana. 94 Journal of Emerging Market Finance 19(1) 5 Robustness Checks To enhance reliability, efficiency and accuracy of the result, this study employs a number of techniques. First, using the statistic table, the study screens for outliers in order to reduce the biases caused by outliers. Hence, no evidence of outliers was identified. Second, the acceptability and nor- mality of each variable are assessed by the use of variance inflation factor and Shapiro–Wilk’s normality test, respectively. All the variables were acceptable in the model and are their means normality distributed. Third, the study employs Pearson’s correlation to screen for multicollinearity and no evidence of multicollinearity was identified among the independ- ent variables. Fourth, GMM, robust fixed and random effect regression estimations are employed to control for autocorrelation, heteroscedasticity, omitted variable biases and endogeneities in the estimated models. Fifth, 24 different models are estimated to ensure consistency and reliability in results. Hence, from the results obtained, there is evidence of consist- ency and reliability, as the signs of the variables are consistent across the 24 models estimated and three estimation strategies employed. From the GMM estimates, the Hansen test, F-statistics and Arellano–Bond tests (1 and 2) are all evidence of robustness of the models while the R-squared and F-probability tests of the robust fixed and random effect regressions are evidence of robust and consistent results. Given all the results of the robustness tests employed, the results are reliable, consistent and efficient and fit for generalisation for banks in the Ghanaian banking sector. 6 Conclusion and Policy Recommendation This study sets out the objective of investigating how sectoral loan con- centration affects bank stability in Ghana, covering period the period 2007 and 2014. To do this, we investigate both the linearity and non-linearity of sectoral loan concentration on bank stability. This objective is motivated by the lack of empirical studies on the effect of sectoral loan concentra- tion on bank stability in Ghana and Africa at large. Employing a two-step Generalised method of moments (GMM), robust random and fixed effect regressions, this study provides evidence that sectoral loan concentration measured with ADM and Herfindahl-Hirschman index (HHI) does not only have a reducing monotone (linear) effect on stability of banks in Ghana but also a direct U-shaped relationship on bank stability in Ghana. This suggests that sectoral loan concentration may weaken stability of banks in Ghana at the initial stage but at a point promotes stability of banks in Kusi et al. 95 Ghana. This implies that there is a threshold point beyond which sectoral loan concentration promotes stability in Ghana. This implies that sectoral loan concentration may detract stability till a point where banks learn and gain experience through the learning curve, which boosts their understand- ing of the patterns, conditions and players of the sector, hence giving the competitive advantage as to how to best handle loans advanced to players in that sector, which helps boost stability of banks. From these findings, the following recommendations are suggested. First, management of banks may diversify their sectoral loan portfolio in order to enhance stability but may acquire and develop expert knowledge and experience in the long run as a means of increasing sectoral loan con- centration to boost stability. Second, policymakers and regulators must not only develop and design policies and regulations that prohibit sectoral loan concentration but incorporate plans that help banks to develop core competence and competitive advantage to take advantage of sectoral loan concentration. Third, given that research on sectoral loan concentration is limited and scant, especially in Ghana and Africa at large, researchers are encouraged to further this study to compliment the findings of this study in Ghana and Africa as a whole. 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. References Acharya, V. V., Hasan, I., & Saunders, A. (2006). Should banks be diversified? Evidence from individual bank loan portfolios. The Journal of Business, 79(3), 1355–1412. Adrian, T., & Shin, H. S. (2008). Liquidity and financial contagion. Banque de France Financial Stability Review: Special Issue on Liquidity, 11, 1–7. Adzobu, L. D., Agbloyor, E. K., & Aboagye, A. (2017). The effect of loan portfolio diversification on banks’ risks and return: Evidence from an emerging market. Managerial Finance, 43(11), 1274–1291. Ali, M. S. B., Intissar, T., & Zeitun, R. (2015). Banking concentration and financial stability: Evidence from developed and developing countries (No. 2015–22). Economics Discussion Papers. Retrieved from https://www. econstor.eu/handle/10419/109217 Allen, F., & Gale, D. (2000). Comparing financial systems. MIT press. 96 Journal of Emerging Market Finance 19(1) Arellano, M., & Bond, S. (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. Arellano, M., & Bover, O. (1995). Another look at the instrumental variable estimation of error-components models. Journal of Econometrics, 68(1), 29–51. Ariss, R. T. (2010). On the implications of market power in banking: Evidence from developing countries. Journal of banking & Finance, 34(4), 765–775. Athanasoglou, P. P., Brissimis, S. N., & Delis, M. D. (2008). Bank-specific, industry-specific and macroeconomic determinants of bank profitability. Journal of International Financial Markets, Institutions and Money, 18(2), 121–136. Bain, M. K., & Howells, P. (2009). Monetary economics: Policy and its theoretical basis. Basingstoke, UK: Palgrave Macmillan. Basel Committee on Bank Supervision (1991). Retrieved from https://www.bis. org/publ/bcbsc121.htm Bebczuk, R., & Galindo, A. (2008). Financial crisis and sectoral diversification of Argentine banks, 1999–2004. Applied Financial Economics, 18(3), 199–211. Beck, T., De Jonghe, O., & Schepens, G. (2013). Bank competition and stability: Cross-country heterogeneity. Journal of financial Intermediation, 22(2), 218–244. Beck, T. (2007). Bank concentration and fragility. Impact and mechanics. In The risks of financial institutions (pp. 193–234). Chicago, IL: University of Chicago Press. Beck, T., Demirgüç-Kunt, A., & Levine, R. (2006). Bank concentration, competi- tion, and crises: First results. Journal of Banking & Finance, 30(5), 1581–1603. Berger, A. N. (1995). The profit-structure relationship in banking–tests of market-power and efficient-structure hypotheses. Journal of Money, Credit and Banking, 27(2), 404–431. Berger, A. N., Klapper, L. F., & Turk-Ariss, R. (2009). Bank competition and financial stability. Journal of Financial Services Research, 35(2), 99–118. Boyd, J. H., & Runkle, D. E. (1993). Size and performance of banking firms: Testing the predictions of theory. Journal of monetary economics, 31(1), 47–67. Boyd, J. H., De Nicolò, G., & Jalal, A. M. (2006). Bank risk-taking and competition revisited: New theory and new evidence (No. 6–297). International Monetary Fund. Retrieved from https://books.google.com.gh/books?hl=en&lr=&id=- 6lfw-6APJUC&oi=fnd&pg=PP6&dq=Boyd,+Nicol%C3%B2+%26+ Jalal+(2006)&ots=FQhAMu4sn0&sig=-GEQlMKw3-evL_QcqkcU9_nmn- c&redir_esc=y#v=onepage&q=Boyd%2C%20Nicol%C3%B2%20%26%20 Jalal%20(2006)&f=false Brealey, R. A., & Myers, S. C. (2003). Capital investment and valuation. New York, NY: McGraw Hill Professional. Blundell, R., & Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), 115–143. Kusi et al. 97 Cai, R., & Zhang, M. (2017). How does credit risk influence liquidity risk? Evidence from Ukrainian banks. Visnyk of the National Bank of Ukraine, (241), 21–32. Castro, C. (2012). Confidence sets for asset correlations in portfolio credit risk. Revista de Economia del Rosario, 15(1), 19. Chaibi, H., & Ftiti, Z. (2015). Credit risk determinants: Evidence from a cross- country study. Research in International Business and Finance, 33, 1–16. Chen, Y., Wei, X., Zhang, L., & Shi, Y. (2013). Sectoral diversification and the banks' return and risk: Evidence from Chinese listed commercial banks. Procedia Computer Science, 18, 1737–1746. Chen, C., Polemis, M., & Stengos, T. (2018). On the examination of non- linear relationship between market structure and performance in the US manufacturing industry. Economics Letters, 164, 1–4. Crotty, J. (2009). Structural causes of the global financial crisis: A critical assessment of the ‘new financial architecture’. Cambridge Journal of Economics, 33(4), 563–580. Dai, M., Liu, Q., & Serfes, K. (2014). Is the effect of competition on price dispersion nonmonotonic? Evidence from the US airline industry. Review of Economics and Statistics, 96(1), 161–170. Denis, D. J., Denis, D. K., & Sarin, A. (1997). Agency problems, equity ownership, and corporate diversification. The Journal of Finance, 52(1), 135–160. Diamond, D. W. (1984). Financial intermediation and delegated monitoring. The Review of Economic Studies, 51(3), 393–414. Fofack, H. (2005). Nonperforming loans in Sub-Saharan Africa: Causal analysis and macroeconomic implications. The World Bank. Retrieved from https:// elibrary.worldbank.org/doi/abs/10.1596/1813-9450-3769 Hayden, E., Porath, D., & Westernhagen, N. V. (2007). Does diversification improve the performance of German banks? Evidence from individual bank loan portfolios. Journal of Financial Services Research, 32(3), 123–140. Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The American Economic Review, 76(2), 323–329. Jiménez, G., Lopez, J. A., & Saurina, J. (2009). Empirical analysis of corporate credit lines. The Review of Financial Studies, 22(12), 5069–5098. Kamp, A., Porath, D., & Pfingsten, A. (2005). Do banks diversify loan portfolios? A Tentative Answer Based on Individual Bank Loan Portfolios (February 28, 2005). Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_ id=675772 Keeley, M. C. (1990). Deposit insurance, risk, and market power in banking. The American Economic Review, 80(5), 1183–1200. Keeton, W. R. (1999). Does faster loan growth lead to higher loan losses? Economic Review-Federal Reserve Bank of Kansas City, 84(2), 57. Kennedy, P. (2008). A guide to econometrics (6th ed.). Oxford, UK: Blackwell Publishing. 98 Journal of Emerging Market Finance 19(1) Kusi, B. A., Agbloyor, E. K., Fiador, V. O., & Osei, K. A. (2016). Does information sharing promote or detract from bank returns: Evidence from Ghana. African Development Review, 28(3), 332–343. Laeven, L., & Levine, R. (2009). Bank governance, regulation and risk taking. Journal of Financial Economics, 93(2), 259–275. Lind, J. T., & Mehlum, H. (2010). With or without U? The appropriate test for a U-shaped relationship. Oxford bulletin of economics and statistics, 72(1), 109–118. Margaritis, D., & Psillaki, M. (2010). Capital structure, equity ownership and firm performance. Journal of Banking & Finance, 34(3), 621–632. Mishkin, F. S. (1999). Global financial instability: Framework, events, issues. The Journal of Economic Perspectives, 13(4), 3–20. Morris, J. A. (2001). Risk diversification in the credit portfolio: An overview of country practices (No. 1–200). International Monetary Fund. Retrieved from https://books.google.com.gh/books?hl=en&lr=&id=Vdj52lE5DPoC&oi= fnd&pg=PA5&dq=Morris,+(2001)+sectoral+loan&ots=IVAoGPZo Vp&sig=VVC2Gyt8fK693KWomCZN-1zdo4w&redir_esc=y#v= onepage&q=Morris%2C%20(2001)%20sectoral%20loan&f=false Naceur, S. B., & Omran, M. (2011). The effects of bank regulations, competition, and financial reforms on banks’ performance. Emerging Markets Review, 12(1), 1–20. Polemis, M., & Stengos, T. (2017). Does competition prevent industrial pollution? Evidence from a panel threshold model. Rajan, R., & Dhal, S. C. (2003). Non-performing loans and terms of credit of public sector banks in India: An empirical assessment. Occasional Papers, 24(3), 81–121. Ross, S., Westerfield, R., Jaffe, J., & Jordan, B. (2008). Modern financial management (8th ed., p. 53). New York, NY: McGraw Hill. Rossi, S. P., Schwaiger, M. S., & Winkler, G. (2009). How loan portfolio diversification affects risk, efficiency and capitalization: A managerial behavior model for Austrian banks. Journal of Banking & Finance, 33(12), 2218–2226. Rudolph, K., & Pfi ngsten, A. (2005). Competition in the German banking sector: An empirical analysis of the concentration of commercial loan origination. In 10th Symposium on Finance, Banking, and Insurance at the University of Karlsruhe. Retrieved from https://papers.ssrn.com/sol3/ papers.cfm?abstract_id=673225 Schaeck, K., & Čihák, M. (2010). Competition, efficiency, and soundness in banking: An industrial organization perspective. European banking center discussion Paper, (2010-20S). Retrieved from https://papers.ssrn.com/sol3/ papers.cfm?abstract_id=1635245 Stiroh, K. J., & Rumble, A. (2006). The dark side of diversification: The case of US financial holding companies. Journal of Banking & Finance, 30(8), 2131–2161. Kusi et al. 99 Tabak, B. M., Fazio, D. M., & Cajueiro, D. O. (2011). The effects of loan portfolio concentration on Brazilian banks’ return and risk. Journal of Banking & Finance, 35(11), 3065–3076. Uhde, A., & Heimeshoff, U. (2009). Consolidation in banking and financial stability in Europe: Empirical evidence. Journal of Banking & Finance, 33(7), 1299–1311. Vives, X. (2010). Information and learning in markets: The impact of market microstructure. Princeton University Press. Retrieved from https://papers. ssrn.com/sol3/papers.cfm?abstract_id=1610086 Winton, A. (1999). Don't put all your eggs in one basket? Diversification and specialization in lending. Diversification and Specialization in Lending (September 27, 1999). Windmeijer, F. (2005). A finite sample correction for the variance of linear efficient two-step GMM estimators. Journal of econometrics, 126(1), 25–51.