Drivers of Bank Solvency,Risk Provisioning and Profitability in the Armenian Banking System

Abstract

The aim of this study is to examine the effects of bank-specific, regulatory and macroeconomic determinants on solvency, risk provisioning, and profitability in the Armenian banking sector. We show that abnormal loan growth is associated with a decrease in regulatory capital ratios, an increase in loan loss provisions, and a reduction in loan portfolio profitability. In addition, we observe an inverse relationship between GDP growth and bank solvency as well as profitability. Regarding regulation, we identify a decrease in regulatory capital ratios as well as a drop in profitability after the implementation of the Basel II Accord.

JEL Classification: G32, G21, G28

Keywords

Risk management profitability solvency credit risk Armenian banks loan growth

1. Introduction

The aim of this article is to investigate potential drivers of Armenian banks’ solvency, risk provisioning and profitability. Empirical studies show that both bank-specific and macro-environmental factors can have a significant impact on a bank’s credit risk exposure, profitability and regulatory capital (Foos, Norden, & Weber, 2010; Guidara, Lai, Soumaré, & Tchana, 2013; Hess, Grimes, & Holmes, 2009; Laeven & Majnoni, 2003; Pasiouras & Kosmidou, 2007). Moreover, recent research, both at the international level and at country level, shows that the implementation of the Basel II framework has had positive effects on bank efficiency (Pasiouras, 2008; Pasiouras, Tanna, & Zopounidis, 2009), as well as on banks’ risk taking behaviour (Hakenes & Schnabel, 2011). However, as Barth, Dopico, Nolle, and Wilcox (2002) show empirical results obtained for developed countries may not hold for transition countries. In line with this notion, based on a sample of 582 banks in 22 transition countries Delis, Molyneux, and Pasiouras (2011) show that regulatory capital requirements and super visory power tend to have no significant effect on banks’ productivity under normal market conditions.

Following a period of restructuring and privatisation, in several transition countries in Central and Eastern Europe rapid growth in bank lending to the private sector (even exceeding GDP growth) could be witnessed (Cottarelli, Dell’Ariccia, & Vladkova-Hollar, 2005). Factors driving this development are studied, for instance, by Backé and Zumer (2005), Hilbers, Otker-Robe, Pazarbasioglu, and Johnsen (2005), and Coricelli, Mucci, and Revoltella (2006). Another line of research that is closely related to our study investigates the impact of abnormal loan growth (ALG) on banks’ core characteristics, and on the extent to which banks with growing loan portfolios are fragile (Foos et al., 2010; Laeven & Majnoni, 2003; Salas & Saurina, 2002; Shehzad, de Haan, & Scholtens, 2010). Moreover, considering ALG as a main determinant of banks’ fundamental dimensions (viz., riskiness, solvency and profitability), Foos et al. (2010) found a significantly positive relation with loan losses lagged by two to four years, as well as a negative association with a bank’s relative interest income (RII) and solvency. In addition, prior research indicates the existence of pro-cyclical elements in credit risk, such as easing credit standards and institutional memory (Berger & Udell, 2004), as well as a negative co-movement between a bank’s excess capital, and business cycles (Jokipii & Milne, 2008; Shim, 2013; Stolz & Wedow, 2011). However, Guidara et al. (2013) report a positive association between Canadian banks’ capital buffer and business cycles, indicating a countercyclical effect.

Building on the aforementioned literature, our study focuses on firm-specific, regulatory and macroeconomic factors driving bank characteristics in Armenia. The analyses are based on a unique database provided by the Central Bank of the Republic of Armenia. Our final sample consists of 800 quarterly observations across 22 banks from 2003 to 2014.

In a first step, we analyse the impact of ALG, loans outstanding, bank liquidity and dollarisation level on regulatory capital, controlling for several bank-specific variables as well as for the impact of the Basel II Accord. Our results show that loans outstanding, dollarisation level of liabilities and ALG significantly and negatively influence total regulatory capital. These findings indicate that banks pursuing an aggressive growth strategy, granting more credit and holding more foreign-exchange (FX) liabilities are more exposed to credit risk and maintain lower regulatory capital ratios. Moreover, we find that the implementation of the Basel II Accord in Armenia leads to a decrease in regulatory capital ratios. Regarding GDP growth, we observe a negative and significant association with regulatory capital ratios, which is consistent with banks easing credit standards and aggressively expanding their loan portfolios during periods of high economic growth.

We then analyse whether excessive loan growth, bank liquidity, credit risk exposure and the macroeconomic environment influence banks’ credit risk provisioning while controlling for banks’ ownership type, size and age, as well as for bank’s dollarisation ratio, and for the size of their loan portfolio relative to total assets. Our findings support the hypothesis that banks with a more aggressive lending strategy experience higher loan losses. Moreover, the results confirm that loan loss provisioning is positively related to both increases in banks’ lending activity in general, and increases in foreign currency denominated refinancing.

Finally, we examine how ALG, changes in the loans outstanding ratio and changes in the dollarisation level of liabilities affect banks’ profitability. We again control for the impact of the adoption of Basel II as well as for macroeconomic conditions in Armenia. Our results show that ALG and GDP growth are negatively related to profitability. Interestingly, we find that the implementation of Basel II reduces RII in the Armenian banking system. Theoretically, foreign ownership is likely to reduce borrowing costs for three reasons: first, agency costs may be lower for foreign-owned banks due to better monitoring. Second, subsidiaries of foreign banks may benefit from adapting their parents’ more sophisticated internal control systems, and risk management techniques, and may receive better training. Third, they may have access to intragroup funding at privileged interest rates. As a consequence, lower borrowing costs may lead to lower interest rates on loans granted to customers. However, the results of the regression analyses show no support for this hypothesis. Finally, other bank-specific and regulatory factors included in our analysis show no significant association with banks’ profitability, a finding that runs to some extent contrary to prior empirical evidence for other countries.

Overall, this study yields new insights into the Armenian banking system relevant both from an academic and a regulatory perspective. In particular, we identify several factors at bank and at macro level that may influence risk taking behaviour, profitability and solvency of banks. Thus, our findings may be instructive with respect to the appropriate design of financial institutions’ risk management systems in a transition country.

The rest of the article is organised as follows. Section 2 summarises the current state of research on the determinants of bank solvency, risk and profitability. Section 3 provides a description of the Armenian banking system. In Section 4 we describe the data underlying our study, motivate our choice of variables and present the empirical models adopted. Section 5 discusses our empirical results. Section 6 contains an additional robustness check corroborating our findings. Section 7 contains concluding remarks with respect to our main results as well as their implications.

2. Related Literature

Related literature can be divided into studies exploring the determinants of banks’ solvency, studies focusing on banks’ risks and studies on banks’ profitability. Based on 16,000 individual banks across 16 countries, Foos et al. (2010) study the impact of ALG on banks’ solvency controlling for bank size and several other bank-specific effects. Taking banks’ capitalisation ratio (equity to total assets) as a measure of solvency, they observe a significantly negative association with ALG and a positive relationship with bank size. Based on a sample of 800 banks from 50 countries, Shehzad et al. (2010) analyse the determinants of banks’ capital adequacy ratio and show that banks with higher loan growth and lower cost-to-income ratios have higher capital adequacy ratios. In addition, Francis and Osborne (2010) examining 202 financial institutions in the UK observe that firm size, return on equity (ROE), risk (proxied by regulatory risk weights) and GDP are negatively associated with capital adequacy ratios.

Another stream of studies provides evidence on the impact of bank-specific and macro-environmental factors on bank risk. Laeven and Majnoni (2003) study the impact of credit growth on loan loss provisions based on 1,419 banks among 45 countries for the period 1988–1999. They find a negative and significant coefficient for credit growth, suggesting that banks behave less prudently when they experience a rapid credit expansion. Foos et al. (2010) analysing 16,000 banks in 16 countries show that ALG lagged two to four periods positively affects loan losses of individual banks. They argue that banks pursuing an aggressive credit growth strategy grant credit to previously rejected, unknown or less credible borrowers and hence suffer more credit losses.

On a country-specific level, Hess et al. (2009) analyse the determinants of loan losses based on 33 Australasian banks from 1980 to 2005. They detect that GDP growth and change in the unemployment rate affect significantly banks’ annual loan loss provisions. Furthermore, their findings suggest that strong loan growth transmits to higher credit losses with a lag of two to four years. Studying UK banks Pain (2003) finds a negative relationship between GDP growth and loan loss provisions. Moreover, he shows that well diversified commercial banks bear less loan losses. However, the results show no significant association between rapid loan growth and commercial banks’ loan losses. Salas and Saurina (2002) examine credit risk determinants using Spanish savings and commercial banks data for the period 1985–1997. The results indicate that credit expansion and new market penetration influence future loan losses three and four years later, respectively. Moreover, they observe a negative coefficient for GDP growth implying that macroeconomic fluctuations affect loan loss provision of commercial and savings banks in Spain. Similarly, Jiménez and Saurina (2006) show that Spanish banks’ nonperforming loans are positively related to rapid credit growth (lagged four years), as well as negatively related to GDP growth.

Another line of banking research has focused on the extent to which bank-specific, macroeconomic and regulatory variables are related to banks’ profitability. In this context, several studies provide evidence that bank-specific variables such as firm size (see Foos et al., 2010; García-Herrero & Vázquez, 2013; Goddard, Molyneux, & Wilson, 2004; Pasiouras & Kosmidou, 2007; Stiroh & Rumble, 2006), ALG (see Foos et al., 2010), share of deposits (see García-Herrero, Gavilá, & Santabárbara, 2009), cost-to-income ratio (Pasiouras & Kosmidou, 2007) and capitalisation (see Demirgüç-Kunt & Huizinga, 1999; Foos et al., 2010; García-Herrero et al., 2009; Pasiouras & Kosmidou, 2007) affect bank profitability. Moreover, prior research shows that foreign banks tend to be more profitable than the domestic ones (Berger, Hasan, & Zhou, 2009; Brown, Maurer, Pak, & Tynaev, 2009; Demirgüç-Kunt & Huizinga, 1999; IMF, 2000). With respect to macro-environmental variables the literature provides evidence that inflation, real interest rates and GDP growth positively impact profitability (see Demirgüç-Kunt & Huizinga, 1999; García-Herrero et al., 2009; Pasiouras & Kosmidou, 2007).

Finally, regarding the regulatory framework, based on 87 Chinese banks García-Herrero et al. (2009) show that government intervention impacts bank profitability negatively. However, contradictory results are observed by Guidara et al. (2013) who analyse the six largest Canadian banks. They show that the implementation of Basel I and Basel II improved bank performance in terms of return on assets (ROAs).

3. The Armenian Banking System

Tying in with the aforementioned literature, we assess empirically which are the main factors driving Armenian banks’ solvency, risk provisioning and profitability. Following the collapse of the USSR, the Armenian economy underwent major structural changes during its first decades of independence. Reforms aimed to build a banking sector which would contribute to the stabilisation and sustainable economic growth of a market-based economy. The structural changes took place in three stages. The first stage includes the period from 1991 to 2002. In the early 1990s, private banks started to operate in Armenia, though most of them were small and unhealthy ‘pocket’ banks created by firms and individuals. Due to stricter banking regulation and supervision, consolidation and elimination of ‘pocket’ banks started in 1996 and led to a certain degree of stabilisation in the banking sector in the early 2000s. The second stage includes the period from 2003 to 2006. The main reforms in this period were aimed at creating the financial infrastructure needed for future development and proper risk management. The third stage began in 2006 and continues to the present day. During this stage, reforms were mainly focused on creating a favourable business environment for foreign investors, and on improving banking regulation and supervision in line with international best practices.

The Armenian banking sector consists of 21 commercial banks and one state-owned development bank. Banks still dominate the financial system, holding approximately 90 per cent of financial assets. Although the banking sector exhibits relatively low credit penetration, in recent years a rapid increase in financial intermediation was observed. Compared to 2008, the banking sector credit-to-GDP ratio almost doubled to about 40 per cent in 2014. The banking sector is dominated by foreign-owned banks. More than half of the banking sector (in terms of total assets) consists of subsidiaries of foreign banks from the UK, France, Germany, Cyprus, Russia, Lebanon, Iran and Kazakhstan. Banks typically offer simple traditional credit products focusing mainly on trade, manufacturing, consumer lending and mortgages (IMF, 2013). Hence, credit risk is considered the most significant risk in the Armenian banking system. Credit risk accounts for almost 90 per cent of total risk-weighted assets, and, in this sense, its proper management is vital from a financial stability point of view (see CBA, 2013).¹

The supervisory authority has implemented a tight and robust regulatory as well as supervisory framework. While implementing the Basel II Accord, tier 1 and total capital adequacy requirements were set at the minimum level of 8 per cent and 12 per cent, respectively. Within the scope of Basel II, in 2008 the standardised approach to credit risk measurement, the standardised and the basic indicator approaches to operational risk measurement and the standardised approach to market risk measurement have been adopted to determine risk-weighted assets. Moreover, banks are not allowed to use advanced approaches to measure capital adequacy ratios (CBA, 2007). According to the IMF’s Financial Sector Assessment Program (FSAP), the banking sector of Armenia meets ‘Core Principles for Effective Supervision’ of the Basel Committee on banking supervision by nearly 93 per cent at the level described as compliant or largely compliant (CBA, 2013). The supervisory authority has initiated a harmonisation process of its regulatory framework with European Union (EU) directives. Following the Basel III Accords, the supervisory authority has already released a regulatory framework for gradually implementing respective capital requirements for 2015, whilst for liquidity standards, the supervisory authority is still in the process of collecting data and assessing the necessity and the impact of an implementation.

4. Data, Model and Variables

4.1. Data

Our analysis is based on balance sheet and income statement data, as well as on regulatory data for the period 2003–2014. Our sample includes all commercial banks operating in the Republic of Armenia. We exclude observations for one bank operating under a special supervision regime during the period 2003–2004. Sample banks’ total assets cover about 98 per cent of total assets in the entire Armenian banking sector. Since during the period 1993–2003 the Armenian banking sector underwent major structural changes, we choose the period 2003–2014 which is characterised by a relatively stable development of the country’s banking industry. Overall, the sample comprises an unbalanced panel of 800 quarterly observations from 22 commercial banks.

Financial data are obtained from the supervisory database of the Central Bank of the Republic of Armenia. For the period 2003–2011, financial reporting is based on Armenian Accounting Standards, whereas 2011 marks the transition to International Financial Reporting Standards (IFRS). However, since Armenian Accounting Standards were very closely linked to International Accounting Standards (IAS), the change in accounting standards does not lead to a structural break in financial statement data (The World Bank, 2008). Data on economic indicators have been provided by the National Statistical Service of the Republic of Armenia.

4.2. Model and Variables

To assess the factors driving Armenian banks’ solvency, risk provisioning and profitability, we conduct the following fixed effects regression analyses²:

Solvency:

\begin{matrix} R E G C A P_{i, t} = α + β_{1} A L G_{i, t - 1} + β_{2} C A S H_T A_{i, t - 1} \\ + β_{3} C R E D I T_{i, t - 1} + β_{4} R O E_{i, t - 1} + \\ β_{5} F X F N D_{i, t - 1} + β_{6} B_A G E_{i, t} + β_{7} B N K G R_{D u m_{i, t}} \\ + β_{8} G D P g r_{t - 1} + β_{9} B A S E L_{D u m_{t}} + β_{10} R W_{D u m_{t}} \\ + δ y e a r d u m i e s + ε \end{matrix}

(1)

Risk Provisioning:

\begin{matrix} Δ L L_{i, t} = α + β_{1} A L G_{i, t} + β_{2} C A S H_T A_{i, t - 1} + β_{3} B_S I Z E_{i, t} \\ + β_{4} Δ C R E D I T_{i, t} + β_{5} C R E D R I S K_{i, t - 1} \\ + β_{6} Δ F X F N D_{i, t} + β_{7} B_A G E_{i, t} \\ + β_{8} m e a n G D P g r o w t h_{t (o v e r o n e y e a r)} \\ + β_{9} B N K G R_{D u m_{i, t}} + δ y e a r d u m i e s + ε \end{matrix}

(2)

Profitability:

\begin{matrix} Δ R I I_{i, t} = α + β_{1} Δ C R E D I T_{i, t - 1} + β_{2} A L G_{i, t} \\ + β_{3} Δ F X F N D_{i, t} + β_{4} G D P g r_{t} + β_{5} B N K G R_{D u m_{i, t}} \\ + β_{6} B A S E L I I_{D u m_{t}} + δ y e a r d u m i e s + ε \end{matrix}

(3)

Table 1 summarises the variables used in our empirical analysis as well as the explanatory variables’ expected signs. Our first dependent variable is required total regulatory capital (REGCAP) which is defined as the ratio of total regulatory capital to total risk weighted assets as a proxy for bank solvency. The second dependent variable, loan loss provisioning (LL) is measured as the amount of net loan loss provisions made between t – 4 and t relative to the size of the average customer loan portfolio between t – 4 and t. Finally, profitability is proxied by RII which is measured as the ratio of interest income from customer loans over the average size of the customer loan portfolio.

Descriptive statistics in Table 2 show that Armenian banks have an average total regulatory capital ratio of 38 per cent, with 25th and 75th percentile values of 19 per cent and 47 per cent, respectively. Overall, Armenian banks seem to be well-capitalised. Loan losses (LL) amount to about 1 per cent over the one-year horizon, with 25th and 75th percentile values close to the median (0% and 2%, respectively), suggesting that given the transition stage of the Armenian economy, banks seem to succeed in ensuring relatively high loan portfolio quality. Average RII is 15 per cent, with 25th and 75th percentile values of 14 per cent and 17 per cent, respectively. The tight interquartile range suggests that in general RII does not differ too much among most banks.

Our selection of explanatory variables is motivated as follows. We choose the difference between a bank-specific loan growth rate and the overall banking sector loan growth rate as a proxy for ALG. On the one hand, ALG should lead to a decrease in liquidity implying an increase in liquidity risk. On the other hand, rapid expansion of the outstanding loan portfolio may be associated with easing credit standards, and an increase in loans to less credible borrowers, which may translate into higher future loan losses. In this sense, we expect to observe a negative association of ALG with the regulatory capital ratio, and a positive association with loan losses which would be in line with prior research (Foos et al., 2010; Hess et al., 2009; Salas & Saurina, 2002; Sinkey & Greenawalt, 1991). We expect also to see a negative relationship between ALG and RII, as in order to rapidly expand loan portfolios banks are likely to lower interest rates on loans granted while expecting lower loan losses due to portfolio diversification and economies of scale from non-interest expenses.

Table 1.

Definition of Variables

Variables	Expected Sign			Definition
Variables	H1	H2	H3	Definition
Dependent Variables
REGCAP_i,t	x			Total regulatory capital to risk– weighted assets
LL_i,t		x		Net loan loss provisions over total customer loan portfolio
RII_i,t			x	Interest income from loans over customer loan portfolio
Explanatory Variables
ALG_i,t	–	+	–	Bank loan growth rate minus sector loan growth rate
CASH_TA_i,t	+/–	+/–		Cash and cash equivalents over total assets
CREDIT_i,t	–	+	–	Customer loan portfolio over total assets
ROE_i,t	+			Profit after taxes over average equity
FXFND_i,t	–	+	–	FX liabilities over total liabilities
CREDRISK_i,t		+		Credit risk-based capital over customer loans
B_AGE_i,t	–	–	–	Bank age in months
B_SIZE_i,t		–	–	Natural logarithm of total assets
BNKGRDum_i,t	–	–	–	Dummy variable controlling for type of ownership
GDPgr_t	–	–	–	Real GDP growth
BASELDum_t	–		+	Dummy variable controlling for Basel II effects
RWDum_t	–		+	Dummy variable controlling for FX-related regulatory changes

Source: The authors.

We choose the ratio of cash and cash equivalents to total assets (CASH_TA) as a measure for liquidity and expect a negative association with regulatory capital ratios, and loan losses. On average, Armenian banks hold 5 per cent of total assets in cash and cash equivalents, and half of them keep cash and cash equivalents between 3 and 7 per cent of total assets. We take a high share of cash in the balance sheet as a potential indicator of cash management problems due to low or unstable loan portfolio quality. In this context, as variability in loan portfolio quality negatively impacts the total regulatory capital ratio, we expect the share of cash and cash equivalents in total assets to be negatively associated with the regulatory capital ratio, and positively related to loan loss provisions (Acharya, Davydenko, & Strebulaev, 2012). However, one could also argue that banks holding more cash lower their overall credit risk by restricting lending activities, and hence experience lower loan losses and report a higher regulatory capital ratio.

The ratio of customer loans³ to total assets (CREDIT) is taken as a measure of concentration in banking activities, and we expect to observe a negative relationship with both the solvency ratio and profitability as well as positive relation to bank risk. Since banks with a higher share of customer loans to total assets are holding smaller portfolios of securitised assets and are more active in lending, they are supposed to be more exposed to credit risk (Maudos & Fernández de Guevara, 2004). Moreover, banks expanding their lending activity tend to lower lending rates to encourage customers to borrow.

We choose ROE as a common measure for overall bank profitability and expect to see a positive association with the regulatory capital ratio. As retained earnings are one of the main sources of capital funding for Armenian banks, highly profitable banks are assumed to be able to increase their regulatory capital easily using internal sources, hence they are supposed to maintain higher total regulatory capital ratios.

The ratio of credit risk-based capital to customer loans (CREDRISK) serves as a proxy for loan riskiness. We expect to see a positive association with loan losses, as higher risk taking implies a higher probability of loan losses.

The share of foreign exchange denominated liabilities in total liabilities (FXFND) is used as a proxy for the dollarisation⁴ level of banks’ liabilities. As Table 2 shows, the Armenian banking system seems to be highly dollarised, a common feature for emerging economies. On average, FX denominated liabilities of banks account for 63 per cent of total liabilities. We expect to see a negative association of FXFND with the regulatory capital ratio and with RII. We assume that banks tend to match assets’ and liabilities’ foreign exchange exposure in order not to be exposed to foreign currency risk. Since the Armenian economy is import-oriented (the country’s balance of trade is consistently negative) lending in foreign currencies is assumed to be riskier, due to potential differences between borrowers’ cash-flows and debt servicing flows, consequently resulting in higher relative loan losses. Due to potential higher risk in FX loans and relative loan losses, banks with a higher share of FX funding are supposed to have lower total regulatory capital ratios. Our arguments are consistent with the current regulatory framework requiring higher risk weights for FX denominated assets, as well as requiring gross FX exposure not to exceed 10 per cent of total regulatory capital. We expect also to see a negative association of dollarisation with RII, because banks are supposed to offer a low interest rate both for FX deposits and loans due to clients’ exposure to foreign currency risk.

Table 2.

Descriptive Statistics

				Quantiles
Variables	n	Mean	S.D.	Min	0.25	Mdn	0.75	Max
REGCAP_i,t	800	0.3792	0.2900	0.1210	0.1852	0.2532	0.4669	1.1981
LL_i,t	800	0.0126	0.0454	–0.2861	0.0041	0.0113	0.0230	0.2672
RII_i,t	800	0.1509	0.0262	0.0434	0.1338	0.1501	0.1661	0.2329
ALG_i,t	800	0.0360	0.3649	–0.9001	–0.1734	–0.0042	0.2034	0.9128
CASH_TA_i,t	800	0.0510	0.0292	0.0000	0.0300	0.0400	0.0700	0.2500
CREDIT_i,t–1	800	0.5211	0.1542	0.0100	0.4500	0.5500	0.6300	0.8200
ROE_i,t	800	0.1004	0.1165	–0.4291	0.0322	0.0881	0.1667	0.4509
FXFND_i,t	800	0.6267	0.1585	0.2100	0.5300	0.6150	0.7400	0.9800
CREDRISK_i,t	800	17.0572	1.2547	13.1593	16.1289	17.1965	18.0071	19.5756
B_AGE_i,t	800	168.6496	60.8181	13.0000	126.1500	176.0850	213.0800	278.1300
B_SIZE_i,t	800	17.5048	1.0613	14.0900	16.6700	17.5800	18.3200	19.6400
BNKGRDum_i,t	800	0.3113	0.4633	0.0000	0.0000	0.0000	1.0000	1.0000
GDPgr_t	800	0.0583	0.0888	–0.1970	0.0310	0.0725	0.1216	0.1911
BASELDum_t	800	0.6913	0.4623	0.0000	0.0000	1.0000	1.0000	1.0000
RWDum_t	800	0.3937	0.4889	0.0000	0.0000	0.0000	1.0000	1.0000
Z_SCORE_i^3y	800	31.5198	21.2909	1.0430	14.6768	27.5431	41.6735	88.9722

Source: The authors.

GDP growth (GDPgr) is the real GDP growth rate in the t-th quarter of the year relative to the previous year’s respective quarter. We use this variable in order to capture potential procyclicality in banking activities.

Next, in line with Foos et al. (2010) and Shim (2013), we use the natural logarithm of total assets (B_SIZE) to control for bank size. As Table 2 shows, bank size does not seem to deviate substantially among Armenian banks. This finding suggests a rather low level of concentration in the banking sector which is in line with earlier evidence reported by the Central Bank (CBA, 2013). Regarding the variable’s coefficient, we expect to see a negative relationship between bank size on the one hand and loan losses on the other hand. Large banks are supposed to have more diversified loan portfolios, and can afford spending more on developing better risk management practices, internal control processes and internal capital adequacy assessment programs (ICAAP). Therefore, severe loan losses are assumed to be less likely and large banks are expected to suffer less loan losses. At the same time, large banks also benefit from economies of scale with respect to non-interest expenses, and they tend to refinance based on relatively low interest rates. Therefore, we expect to observe a negative association with RII.

We expect bank age (B_AGE), measured in months since inception, or, if applicable, since its most recent major restructuring due to a merger, to be negatively associated with the regulatory capital ratio, with loan losses and with RII. More mature banks are supposed to have a deeper understanding of local markets, to have established long-term relationships with clients, and to have advanced risks assessment systems. In this sense, older banks are assumed to be able to maintain lower total regulatory capital ratios and have lower relative loan losses, as they are supposed to suffer less severe losses resulting from an underestimation of risk exposure. At the same time, an increase in older banks’ solvency ratios due to well-implemented risk management practices is unlikely, as Armenian banks typically use the standardised approach to credit risk. Regarding profitability, we expect a negative effect of bank age since in a competitive market banks with more mature customer relationships may tend to offer better loan conditions and relatively lower interest rates to key customers. As of December 2014, average age of our sample banks is about 14 years, with the oldest bank being about 23 years old.

Tying in with prior research on the effects of ownership structure, or ownership concentration on bank performance (e.g., Booth, Cornett, & Tehranian, 2002; García-Herrero et al., 2009; Liang, Xu, & Jiraporn, 2013), we control for the affiliation of banks to foreign financial groups using a respective dummy variable (BNKGRDum). We expect subsidiaries of foreign-owned financial groups⁵ to have better risk management practices and internal control processes than their domestic competitors, leading to lower relative loan losses.⁶ Moreover, these subsidiaries are likely to receive emergency capital support in case of financial distress, which combined with the joint impact of relative lower loan losses, allows them to maintain lower total regulatory capital ratios. With respect to RII, we expect foreign subsidiaries to offer relatively low lending rates, due to the availability of cheap funding.⁷

Finally, we include two more dummy variables to capture the impact of regulatory changes relating to the regulatory capital ratio, and RII. First, we expect that the shift from Basel I to Basel II in 2008 (BASELDum) negatively influences the regulatory capital ratio. On the one hand, in addition to credit risk the Basel II framework forces banks also to back market and operational risks with regulatory capital. On the other hand, it imposes stricter risk weights for both on and off balance sheets items than before. Moreover, the shift from Basel I to Basel II is assumed to be positively associated with RII, since ceteris paribus regulatory tightening limits potential lending, hence leading banks to increase interest rates charged on loans to compensate for the reduction in lending volume. Second, we use a dummy variable to capture the impact of changes in the local regulatory framework effective late 2010/early 2011 imposing higher risk weights for FX denominated assets, and higher minimum loan loss provisioning rates for non-performing FX denominated loans (RWDum). Overall, the latter regulatory changes further aggravate capital requirements (which are by nature a regulatory tightening), and therefore are assumed to influence the regulatory capital ratio and RII in a similar way as the shift from Basel I to Basel II.

5. Results

5.1. Factors Driving Bank Solvency

Regression model (1) investigates how the regulatory capital ratio varies with changes in bank-specific variables, and regulatory and macroeconomic conditions. Our empirical results support the hypothesis that loans outstanding and ALG affect the regulatory capital ratio. In particular, we find that the prior quarter’s ALG is inversely related to banks’ total regulatory capital ratio (columns 1 and 2 of Table 3), which is consistent with banks pursuing aggressive growth strategies increasing risk-weighted assets and thereby putting pressure on their regulatory capital ratios.

Foos et al. (2010) argue that loan losses typically occur several periods after a loan has been granted. Moreover, ALG more than one quarter ago might even be positively associated with regulatory capital. If loan portfolio expansion turns out profitable due to increases in commissions and interest rates charged, and banks deliberately replenish statutory capital, this could also result in a considerable increase in the regulatory capital ratio. For this reason, we repeat the aforementioned analysis replacing the ALG variable lagged one period with the average ALG over one year (four quarter average). Our findings shown in columns 3 and 4 of Table 3 indicate that average ALG leads to a decrease in the regulatory capital ratio. The negative coefficient of 0.11 for average ALG implies that ALG of 10 per cent over one year leads to a decrease in the regulatory capital ratio by almost 1.1 per cent.

Regarding the credit ratio we observe a negative and significant relationship with the regulatory capital ratio across all model specifications, which is consistent with the Armenian banking sector’s current stage of development. Since Armenian banks’ securities portfolios are strongly biased towards government bonds, and since the latter have a zero risk-weight in the calculation of the regulatory capital ratio under the standardised approach, the loan-to-total assets ratio is a key determinant of Armenian banks risk-weighted capital.

Regarding the dollarisation ratio, we find a significantly negative coefficient, which supports our assumption that lending in a foreign currency is riskier for import-oriented countries. Moreover, in line with our initial expectation we find evidence for a negative influence of the implementation of Basel II on the regulatory capital ratio.

With respect to GDP growth which enters our model either with a one-quarter lag (column 1), or as an annual average (column 3) we find a negative and significant association with the regulatory capital ratio. This result supports the hypothesis that under favourable macroeconomic conditions banks are prone to ease their credit standards and grant loans to borrowers of lower creditworthiness, which leads to higher risk exposure and hence to lower regulatory capital.⁸ Moreover, banks may raise capital or retain earnings (to increase the numerator) during favourable economic conditions, and grant fewer loans during economic downturns (to decrease the denominator).

Table 3.
Factors Driving Bank Solvency

Variables 1
2
3
4

REGCAP t–value REGCAP t–value REGCAP t–value REGCAP t–value

ALG_i,t–1 –0.0728 (–2.585) –0.0726 (–2.418)

meanALG_i,t^1y –0.110* (–3.816) –0.105* (–3.265)

CASH_TA_i,t–1 0.322 (0.532) 0.435 (0.683) 0.327 (0.556) 0.476 (0.762)

CREDIT_i,t–1 –0.486 (–2.706) –0.491 (–2.588) –0.422 (–2.578) –0.446 (–2.520)

ROE_i,t–1 –0.283* (–1.907) –0.263* (–1.785) –0.214 (–1.453) –0.209 (–1.452)

FXFND_i,t–1 –0.230 (–2.518) –0.301 (–2.282) –0.238** (–2.274) –0.266* (–1.951)

B_AGE_i,t –0.000807 (–1.539) –0.00275* (–3.978) –0.00108 (–1.647) –0.00286* (–4.064)

BNKGRDum_i,t 0.0488 (0.458) 0.0545 (0.537) 0.0536 (0.518) 0.0569 (0.555)

GDPgr_t–1 –0.130* (–1.735)

meanGDPgr_t^1y –0.285* (–1.763)

BASELDum_t –0.104* (–1.908) –0.117 (–2.271)

RWDum_t –0.0448 (–1.685) –0.0293 (–0.890)

Constant 1.006* (6.147) 1.294* (5.557) 1.027* (5.408) 1.262*** (5.432)

Time dummies NO YES NO YES

Observations 800 800 800 800

No. of B_Code 22 22 22 22

R-squared 0.488 0.508 0.500 0.515

Variables	1	2	3	4

ALG_i,t–1	–0.0728**	(–2.585)	–0.0726**	(–2.418)
meanALG_i,t^1y					–0.110***	(–3.816)	–0.105***	(–3.265)
CASH_TA_i,t–1	0.322	(0.532)	0.435	(0.683)	0.327	(0.556)	0.476	(0.762)
CREDIT_i,t–1	–0.486**	(–2.706)	–0.491**	(–2.588)	–0.422**	(–2.578)	–0.446**	(–2.520)
ROE_i,t–1	–0.283*	(–1.907)	–0.263*	(–1.785)	–0.214	(–1.453)	–0.209	(–1.452)
FXFND_i,t–1	–0.230**	(–2.518)	–0.301**	(–2.282)	–0.238**	(–2.274)	–0.266*	(–1.951)
B_AGE_i,t	–0.000807	(–1.539)	–0.00275***	(–3.978)	–0.00108	(–1.647)	–0.00286***	(–4.064)
BNKGRDum_i,t	0.0488	(0.458)	0.0545	(0.537)	0.0536	(0.518)	0.0569	(0.555)
GDPgr_t–1	–0.130*	(–1.735)
meanGDPgr_t^1y					–0.285*	(–1.763)
BASELDum_t	–0.104*	(–1.908)			–0.117**	(–2.271)
RWDum_t	–0.0448	(–1.685)			–0.0293	(–0.890)
Constant	1.006***	(6.147)	1.294***	(5.557)	1.027***	(5.408)	1.262***	(5.432)

Time dummies	NO		YES		NO		YES
Observations	800		800		800		800
No. of B_Code	22		22		22		22
R-squared	0.488		0.508		0.500		0.515

Source: The authors.

Notes: Models (1)–(4) are estimated using a fixed effects approach. Models (2)–(4) are estimated using time dummies while omitting time-varying but bank-independent variables. Regarding time dummies, we run a Wald test to verify the hypothesis that the coefficients of these dummies are jointly 0, which is strongly rejected for all models. Adequacy of the fixed effects approach is scrutinised based on a Hausman test. In model (3) we use average ALG (meanALG_i,t^1y) and average GDP growth over one year (meanGDPgr_t^1y). We omit RWDum and BASELDum in models (2) and (4) due to a relatively high Pearson correlation coefficient of 0.5386. Robust t-statistics are stated in parentheses. *, ** and ** indicate significance at the 10%, 5% and 1% levels, respectively.

Table 4.

Factors Driving Bank Risk Provisioning

VARIABLES	1		2
VARIABLES	ΔLL	t-value	ΔLL	t-value
ALG_i,t	0.00792*	(1.859)	0.00713*	(1.759)
CASH_TA_i,t–1	0.0203	(0.548)	0.0234	(0.682)
ΔCREDIT_i,t	0.0795*	(1.917)	0.0737*	(1.853)
ΔFXFND_i,t	0.0459**	(2.215)	0.0429**	(2.022)
CREDRISK_i,t–1	0.00474	(0.730)	0.00584	(0.945)
B_AGE_i,t	2.79e-06	(0.282)	5.86e-06	(0.567)
B_SIZE_i,t	–0.00552	(–0.672)	–0.00634	(–0.803)
BNKGRDum_i,t	–0.000158	(–0.149)	0.000145	(0.138)
meanGDPgr_t^1y	–0.00593	(–0.375)
Constant	0.0145	(0.396)	0.00767	(0.210)
Time dummies	NO		YES
Observations	800		800
No. of B_Code	22		22
R-squared	0.0342		0.0804

Source: The authors.

Notes: All models are estimated using a random effects approach. Model (2) is estimated using time dummies while omitting time-varying but bank-independent variables. Regarding time dummies, we run a Wald test to verify the hypothesis that the coefficients of these dummies are jointly 0, which is strongly rejected for all models. Adequacy of the random effects approach is scrutinised based on a Hausman test. meanGDPgr_t^1y is average GDP growth over one year. Results do not change qualitatively if we conduct a fixed effects regression analysis based on models (1) and (2), or if we omit B_Size (due to a relatively high Pearson correlation coefficient of 0.9687 with CREDRISK_i,t–1). Robust t-statistics are stated in parentheses. * and ** indicate significance at the 10% and 5% levels, respectively.

5.2. Factors Driving Bank Risk Provisioning

As Table 4 shows, we test two specifications of Equation (2) to asses which factors explain changes in Armenian banks’ loan loss provisions. Regarding ALG, we observe a positive and significant regression coefficient. This result is again consistent with earlier findings by Foos et al. (2010) who argue that in order to expand their loan portfolios, banks following an aggressive growth strategy may be tempted to approach borrowers with lower creditworthiness, thus sowing the seeds for higher future losses.

Table 5.
Factors Driving Bank Profitability

VARIABLES 1
2

ΔRII t-value ΔRII t-value

ALG_i,t –0.00222 (–2.282) –0.00229 (–2.307)

ΔCREDIT_i,t–1 0.0115 (1.021) 0.0124 (1.067)

ΔFXFND_i,t –0.00755 (–0.891) –0.00911 (–1.006)

BNKGRDum_i,t –0.00133 (–0.933) –0.000948 (–0.580)

GDPgr_t –0.0146 (–2.814)

BASELDum_t –0.00205 (–2.166)

Constant 0.00174* (2.004) –0.00120 (–1.498)

Time dummies NO YES

Observations 735 735

No. of B_Code 22 22

R-squared 0.038 0.046

VARIABLES	1	2
ALG_i,t	–0.00222**	(–2.282)	–0.00229**	(–2.307)
ΔCREDIT_i,t–1	0.0115	(1.021)	0.0124	(1.067)
ΔFXFND_i,t	–0.00755	(–0.891)	–0.00911	(–1.006)
BNKGRDum_i,t	–0.00133	(–0.933)	–0.000948	(–0.580)
GDPgr_t	–0.0146**	(–2.814)
BASELDum_t	–0.00205**	(–2.166)
Constant	0.00174*	(2.004)	–0.00120	(–1.498)
Time dummies	NO		YES
Observations	735		735
No. of B_Code	22		22
R-squared	0.038		0.046

Source: The authors.

Notes: Both models are estimated using a fixed effects approach. Adequacy of the fixed effects approach is scrutinised based on a Hausman test. Model (2) is estimated using time dummies while omitting time-varying but bank-independent variables. Results do not change even when we include several other variables that may serve as additional factors explaining change in RII (B_SIZE_i,t, B_AGE_i,t, RWDum, meanGDPgr_t^1y and ΔFXFND_i,t–1). Moreover, since the local regulatory framework implies higher risk-weights and higher loan loss provisioning rates for FX loans, we also interact FXFND with RWA (FXFND * RWDum_i,t) in both models, but the coefficient on this variable is insignificant. Regarding time dummies, we run a Wald test to verify the hypothesis that the coefficients of these dummies are jointly 0, which is strongly rejected for both models. Robust t-statistics are stated in parentheses. * and ** indicate significance at the 10% and 5% levels, respectively.

We further observe a positive and statistically significant relationship between changes in bank’s lending activity (∆CREDIT) and changes in loan loss provisions. As banks with an increasing share of customer loans to total assets should be subject to higher credit risk, this result is again in line with our expectations. Moreover, with respect to the dollarisation ratio the regression analysis confirms that loan losses are higher for banks which are more active in FX funding. For all other factors, regression coefficients are statistically insignificant.

5.3. Factors Driving Bank Profitability

Regarding profitability, the empirical findings support our expectation that ALG is associated with a reduction in RII⁹ (see Table 5), which may be a result of banks trying to underbid their competitors to boost their lending business. However, in contrast to our expectations we find a negative and statistically significant coefficient for the Basel II dummy implying that the implementation of Basel II is associated with a decrease in Armenian banks’ profitability.

Moreover, we find a negative and statistically significant relationship between GDP growth and changes in RII, suggesting that while in a booming economy banks may well extend their lending activities, profitability decreases due to substantial competitive downward pressure on interest rates. During economic downturns, however, banks tend to tighten credit standards, and they tend to be more sensitive to credit risk implying higher mark-ups on customer loans.

6. Robustness Test

To scrutinise the robustness of our choice of LL as a measure of bank (credit) risk, we restate Equation (2) using banks’ Z-scores as an alternative dependent variable. Being a measure of distance-to-default, Z-score is calculated as the average capitalisation (equity-to-asset ratio, Eq/A) adjusted by the RoA during the three preceding years, divided by the standard deviation of RoAs over the same time horizon (Fang, Hasan, & Marton, 2014; Foos et al., 2010):

Z_S C O R E = \frac{M e a n (E q / A + R o A)}{S t d . D e v . (R O A)}

Replacing the loan loss provision variable (LL) with Z-score we analyse whether average excessive loan growth (meanALG_i,t^3y) and dollarisation (meanFXFND_i,t^3y) over three years, bank liquidity, banks activity concentration, risk-based capital, bank size and ownership structure are related to bank risk by controlling for time effects.¹⁰ We observe a negative and significant association with average ALG (see Table 6), indicating that banks pursuing an aggressive growth strategy tend to be exposed to higher credit risk in the future. Moreover, in line with our initial expectation, we find a negative and significant coefficient for the dollarisation ratio, providing evidence that lending in foreign currencies bears higher default risk. This result is consistent with findings by Luca and Petrova (2008), who show that banks in transition countries tend to lower their direct exposure to FX risk (by rolling over this risk to borrowers), but also to increase their exposure to currency-induced default risk, hence increasing the country’s overall sensitivity to foreign exchange-related financial crises.

Table 6.
Empirical Results for Z-score

VARIABLES 1
2

Z_SCORE t-value Z_SCORE t-value

meanALG_i,t^3y –5.020* (–1.854) –4.418* (–1.828)

CASH_TA_i,t 150.0* (1.835)

meanCASH_TA_i,t^3y 81.49 (0.594)

CREDIT_i,t –0.759 (–0.0448)

meanCREDIT_i,t^3y 21.42 (0.772)

meanFXFND_i,t^3y –31.97 (–2.325) –39.64 (–2.100)

CREDRisk_i,t 14.05 (1.718)

meanCREDRisk_i,t^3y 3.061 (0.541)

B_SIZE_i,t 1.792 (0.225) 11.07 (1.707)

BNKGRDum_i,t 2.547 (0.417) 1.604 (0.186)

Constant –234.3 (–2.384) –210.1 (–2.084)

Time dummies YES YES

Observations 624 624

N. of B_Code 22 22

R-squared 0.190 0.156

VARIABLES	1	2
meanALG_i,t^3y	–5.020*	(–1.854)	–4.418*	(–1.828)
CASH_TA_i,t	150.0*	(1.835)
meanCASH_TA_i,t^3y			81.49	(0.594)
CREDIT_i,t	–0.759	(–0.0448)
meanCREDIT_i,t^3y			21.42	(0.772)
meanFXFND_i,t^3y	–31.97**	(–2.325)	–39.64**	(–2.100)
CREDRisk_i,t	14.05	(1.718)
meanCREDRisk_i,t^3y			3.061	(0.541)
B_SIZE_i,t	1.792	(0.225)	11.07	(1.707)
BNKGRDum_i,t	2.547	(0.417)	1.604	(0.186)
Constant	–234.3**	(–2.384)	–210.1**	(–2.084)
Time dummies	YES		YES
Observations	624		624
N. of B_Code	22		22
R-squared	0.190		0.156

Source: The authors.

Notes: Both models are estimated using a fixed effects approach. Adequacy of the fixed effects approach is scrutinised based on a Hausman test. In model 1, the variables CASH_TA_i,t, CREDIT_i,t and CREDRISK_i,t are included as control variables whereas in model (2) they are used as predictor variables. Consistent with the estimation period of the dependent variable we determine meanALG_i,t^3y, meanCASH_TA_i,t^3y, meanCREDIT_i,t^3y, meanCREDRISK_i,t^3y and meanFXFND_i,t^3y as the respective explanatory variables’ three-year averages. Regarding time dummies, we run a Wald test to verify the hypothesis that the coefficients of these dummies are jointly 0, which is strongly rejected for both models. Robust t-statistics are stated in parentheses. * and ** indicate significance at the 10% and 5% levels, respectively.

7. Conclusion

Based on a comprehensive supervisory database provided by the Central bank of the Republic of Armenia, this article examines how firm-specific, regulatory and macroeconomic factors affect bank stability and bank performance in the Armenian banking sector. To the best of our knowledge this study is the first to assess the impact of factors like ALG, dollarisation, GDP growth or the implementation of Basel II on solvency, risk provisioning and profitability in a post-soviet transition country. We find that both ALG and GDP growth have a significant and negative impact on banks’ solvency, and profitability. In addition, we find that both ALG and dollarisation positively affect subsequent loan losses. These findings suggest that in order to attract new business, banks pursuing an aggressive growth strategy tend to ease their credit standards, and to grant loans at lower interest rates than justified by the default risks being taken.

From a supervisory perspective, our findings may serve as a starting point for determining on which indicators an effective early warning system might be built to be able to identify threats to financial market stability in the Republic of Armenia. In addition, they emphasise the importance of credit risk management and risk-adjusted pricing for banks operating in a transition country. Although Armenian banks have significantly improved their risk management systems during the past two decades, Western risk management practices have not yet been fully implemented, as domestic banks’ reliance on the standardised approach to credit risk measurement illustrates. In this context, the drivers of solvency, risk provisioning and profitability identified in our study might be taken into account by Armenian banks in future attempts to develop more advanced risk management approaches.

Declaration of Conflicting Interests

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

Funding

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

Footnotes

Acknowledgements

The first draft of this article was developed while Suren Pakhchanyan was a guest researcher at the Central Bank of the Republic of Armenia. He is grateful for all the support provided there. Moreover, we wish to thank staff members at the Central Bank, seminar participants at University of Oldenburg, as well as Thomas Kaspereit for their helpful comments. The views expressed in this article are those of the authors and do not necessarily reflect those of the Central Bank of the Republic of Armenia or Deloitte.

Notes

References

Acharya

Davydenko

S. A.

& Strebulaev

I. A.

(2012). Cash holdings and credit risk. Review of Financial Studies, 25(12), 3572–3609.

Backé

P. F.

& Zumer

(2005). Developments in credit to the private sector in central and Eastern European EU member states: Emerging from financial repression—A comparative overview. In Oesterreichische Nationalbank (Ed.), Focus on European economic integration (p. 83). Vienna: Oesterreichische Nationalbank.

Barth

J. R.

Dopico

L. G.

Nolle

D. E.

& Wilcox

J. A.

(2002). Bank safety and soundness and the structure of bank supervision: A cross-country analysis. International Review of Finance, 3(3–4), 163–188.

Berger

A. N.

Hasan

& Zhou

(2009). Bank ownership and efficiency in China: What will happen in the world’s largest nation? Journal of Banking & Finance, 33(1), 113–130.

Berger

A. N.

& Udell

G. F.

(2004). The institutional memory hypothesis and the procyclicality of bank lending behavior. Journal of Financial Intermediation, 13(4), 458–495.

Booth

J. R.

Cornett

M. M.

& Tehranian

(2002). Boards of directors, ownership, and regulation. Journal of Banking & Finance, 26(10), 1973–1996.

Brown

Maurer

M. R.

Pak

& Tynaev

(2009). The impact of banking sector reform in a transition economy: Evidence from Kyrgyzstan. Journal of Banking & Finance, 33(9), 1677–1687.

CBA (2007). Regulation of banking, prudential standards for banking. Yerevan, Armenia: Central Bank of the Republic of Armenia.

CBA (2013). Financial stability report 2012. Armenia: Central Bank of the Republic of Armenia.

10.

Clarke

G. R. G.

Cull

& Martinez Peria

M. S.

(2001). Does foreign bank penetration reduce access to credit in developing countries evidence from asking borrowers (Policy Research Working Paper Series). Washington, DC: The World Bank.

11.

Coricelli

Mucci

& Revoltella

(2006). Household credit in the new Europe: Lending boom or sustainable growth? (CEPR Discussion Paper). London: Centre for Economic Policy Research.

12.

Cottarelli

Dell’Ariccia

& Vladkova-Hollar

(2005). Early birds, late risers, and sleeping beauties: Bank credit growth to the private sector in Central and Eastern Europe and in the Balkans. Journal of Banking & Finance, 29(1), 83–104.

13.

Crystal

J. S.

Dages

B. G.

& Goldberg

L. S.

(2002). Has foreign bank entry led to sounder banks in Latin America? Current Issues in Economics and Finance, 8(1), 1–6.

14.

Dabla-Norris

& Floerkemeier

(2007). Bank efficiency and market structure: What determines banking spreads in Armenia? (IMF Working Paper No. 07/134). Washington, DC: International Monetary Fund.

15.

Delis

M. D.

Molyneux

& Pasiouras

(2011). Regulations and productivity growth in banking: Evidence from transition economies. Journal of Money, Credit and Banking, 43(4), 735–764.

16.

Demirgüç-Kunt

& Huizinga

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

17.

Detragiache

Tressel

& Gupta

(2008). Foreign banks in poor countries: Theory and evidence. The Journal of Finance, 63(5), 2123–2160.

18.

Fang

Hasan

& Marton

(2014). Institutional development and bank stability: Evidence from transition countries. Journal of Banking & Finance, 39(1), 160–176.

19.

Foos

Norden

& Weber

(2010). Loan growth and riskiness of banks. Journal of Banking & Finance, 34(12), 2929–2940.

20.

Francis

W. B.

& Osborne

(2010). On the behavior and determinants of risk-based capital ratios: Revisiting the evidence from UK banking institutions. International Review of Finance, 10(4), 485–518.

21.

García-Herrero

Gavilá, S., & Santabárbara

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

22.

García-Herrero

& Vázquez

(2013). International diversification gains and home bias in banking. Journal of Banking & Finance, 37(7), 2560–2571.

23.

Goddard

Molyneux

& Wilson

J. O. S.

(2004). Dynamics of growth and profitability in banking. Journal of Money, Credit and Banking, 36(6), 1069–1090.

24.

Greene

W. H.

(2008). Econometric analysis. Upper Saddle River, NJ: Prentice Hall.

25.

Guidara

Lai

V. S.

Soumaré

& Tchana

F. T.

(2013). Banks’ capital buffer, risk and performance in the Canadian banking system: Impact of business cycles and regulatory changes. Journal of Banking & Finance, 37(9), 3373–3387.

26.

Hakenes

& Schnabel

(2011). Bank size and risk-taking under Basel II. Journal of Banking & Finance, 35(6), 1436–1449.

27.

Hess

Grimes

& Holmes

(2009). Credit losses in Australasian banking. Economic Record, 85(270), 331–343.

28.

Hilbers

Otker-Robe

Pazarbasioglu

& Johnsen

(2005). Assessing and managing rapid credit growth and the role of supervisory and prudential policies (SSRN Scholarly Paper). Rochester, NY: Social Science Research Network.

29.

IMF (2000). International capital markets: Developments, prospects, and key issues. Washington, DC: Author.

30.

IMF (2013). Republic of Armenia: Financial system stability assessment. Washington, DC: Author.

31.

Jiménez

& Saurina

(2006). Credit cycles, credit risk, and prudential regulation. International Journal of Central Banking, 2(2), 65–98.

32.

Jokipii

& Milne

(2008). The cyclical behaviour of European bank capital buffers. Journal of Banking & Finance, 32(8), 1440–1451.

33.

Laeven

& Majnoni

(2003). Loan loss provisioning and economic slowdowns: Too much, too late? Journal of Financial Intermediation, 12(2), 178.

34.

Liang

& Jiraporn

(2013). Board characteristics and Chinese bank performance. Journal of Banking & Finance, 37(8), 2953–2968.

35.

Luca

& Petrova

(2008). What drives credit dollarization in transition economies? Journal of Banking & Finance, 32(5), 858–869.

36.

Maudos

& Fernández de Guevara

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

37.

Ongena

Popov

& Udell

G. F.

(2013). ‘When the cat’s away the mice will play’: Does regulation at home affect bank risk-taking abroad? Journal of Financial Economics, 108(3), 727–750.

38.

Pain

(2003). The provisioning experience of the major UK banks: A small panel investigation (Bank of England Working Paper). London, UK: Bank of England.

39.

Pasiouras

(2008). International evidence on the impact of regulations and supervision on banks’ technical efficiency: An application of two-stage data envelopment analysis. Review of Quantitative Finance & Accounting, 30(2), 187–223.

40.

Pasiouras

& Kosmidou

(2007). Factors influencing the profitability of domestic and foreign commercial banks in the European Union. Research in International Business and Finance, 21(2), 222–237.

41.

Pasiouras

Tanna

& Zopounidis

(2009). The impact of banking regulations on banks’ cost and profit efficiency: Cross-country evidence. International Review of Financial Analysis, 18(5), 294–302.

42.

Rajan

R. G.

(1994). Why bank credit policies fluctuate: A theory and some evidence. The Quarterly Journal of Economics, 109(2), 399–441.

43.

Salas

& Saurina

(2002). Credit risk in two institutional regimes: Spanish commercial and savings banks. Journal of Financial Services Research, 22(3), 203–224.

44.

Shehzad

C. T.

de Haan

& Scholtens

(2010). The impact of bank ownership concentration on impaired loans and capital adequacy. Journal of Banking & Finance, 34(2), 399–408.

45.

Shim

(2013). Bank capital buffer and portfolio risk: The influence of business cycle and revenue diversification. Journal of Banking & Finance, 37(3), 761–772.

46.

Sinkey

J. F.

& Greenawalt

M. B.

(1991). Loan-loss experience and risk-taking behavior at large commercial banks. Journal of Financial Services Research, 5(1), 43–59.

47.

Stiroh

K. J.

& Rumble

(2006). The dark side of diversification: The case of US financial holding companies. Journal of Banking & Finance, 30(8), 2131–2161.

48.

Stolz

& Wedow

(2011). Banks’ regulatory capital buffer and the business cycle: Evidence for Germany. Journal of Financial Stability, 7(2), 98–110.

49.

The World Bank (2008). Report on the Observance of Standards and Codes (ROSC)—Accounting and auditing: Armenia. Washington, DC: Author.

Variables	1		2		3		4
Variables	REGCAP	t–value	REGCAP	t–value	REGCAP	t–value	REGCAP	t–value

ALG_i,t–1	–0.0728**	(–2.585)	–0.0726**	(–2.418)
meanALG_i,t^1y					–0.110***	(–3.816)	–0.105***	(–3.265)
CASH_TA_i,t–1	0.322	(0.532)	0.435	(0.683)	0.327	(0.556)	0.476	(0.762)
CREDIT_i,t–1	–0.486**	(–2.706)	–0.491**	(–2.588)	–0.422**	(–2.578)	–0.446**	(–2.520)
ROE_i,t–1	–0.283*	(–1.907)	–0.263*	(–1.785)	–0.214	(–1.453)	–0.209	(–1.452)
FXFND_i,t–1	–0.230**	(–2.518)	–0.301**	(–2.282)	–0.238**	(–2.274)	–0.266*	(–1.951)
B_AGE_i,t	–0.000807	(–1.539)	–0.00275***	(–3.978)	–0.00108	(–1.647)	–0.00286***	(–4.064)
BNKGRDum_i,t	0.0488	(0.458)	0.0545	(0.537)	0.0536	(0.518)	0.0569	(0.555)
GDPgr_t–1	–0.130*	(–1.735)
meanGDPgr_t^1y					–0.285*	(–1.763)
BASELDum_t	–0.104*	(–1.908)			–0.117**	(–2.271)
RWDum_t	–0.0448	(–1.685)			–0.0293	(–0.890)
Constant	1.006***	(6.147)	1.294***	(5.557)	1.027***	(5.408)	1.262***	(5.432)

Time dummies	NO		YES		NO		YES
Observations	800		800		800		800
No. of B_Code	22		22		22		22
R-squared	0.488		0.508		0.500		0.515

VARIABLES	1		2
VARIABLES	ΔRII	t-value	ΔRII	t-value
ALG_i,t	–0.00222**	(–2.282)	–0.00229**	(–2.307)
ΔCREDIT_i,t–1	0.0115	(1.021)	0.0124	(1.067)
ΔFXFND_i,t	–0.00755	(–0.891)	–0.00911	(–1.006)
BNKGRDum_i,t	–0.00133	(–0.933)	–0.000948	(–0.580)
GDPgr_t	–0.0146**	(–2.814)
BASELDum_t	–0.00205**	(–2.166)
Constant	0.00174*	(2.004)	–0.00120	(–1.498)
Time dummies	NO		YES
Observations	735		735
No. of B_Code	22		22
R-squared	0.038		0.046

VARIABLES	1		2
VARIABLES	Z_SCORE	t-value	Z_SCORE	t-value
meanALG_i,t^3y	–5.020*	(–1.854)	–4.418*	(–1.828)
CASH_TA_i,t	150.0*	(1.835)
meanCASH_TA_i,t^3y			81.49	(0.594)
CREDIT_i,t	–0.759	(–0.0448)
meanCREDIT_i,t^3y			21.42	(0.772)
meanFXFND_i,t^3y	–31.97**	(–2.325)	–39.64**	(–2.100)
CREDRisk_i,t	14.05	(1.718)
meanCREDRisk_i,t^3y			3.061	(0.541)
B_SIZE_i,t	1.792	(0.225)	11.07	(1.707)
BNKGRDum_i,t	2.547	(0.417)	1.604	(0.186)
Constant	–234.3**	(–2.384)	–210.1**	(–2.084)
Time dummies	YES		YES
Observations	624		624
N. of B_Code	22		22
R-squared	0.190		0.156