Net interest margin, bank risks and business cycle: evidence from Vietnam

By investigating the behavior of bank net interest margins over business cycle in the period 2007-2016, the results show that net interest margins are countercyclical, i.e.. In addition,

NET INTEREST MARGIN, BANK RISKS AND

BUSINESS CYCLE:

EVIDENCE FROM VIETNAM

TRAN PHUONG THAO University of Economics HCMC – tranthao@ueh.edu.vn

THAN THI THU THUY University of Economics HCMC – thuynh@ueh.edu.vn

LE VAN LAM

University of Economics HCMC – levanlamtcnh@ueh.edu.vn

Abstract

The purpose of the study is to explore the relationships between net interest margin, bank risks, and business cycle as well as examine the role of business cycle on the relationship between bank risks and net interest margin in Vietnam By investigating the behavior of bank net interest margins over business cycle in the period 2007-2016, the results show that net interest margins are countercyclical, i.e they significantly increase in recessions In addition, business cycle also significantly affects the association between credit risk and net interest margins Banks with higher credit risk have higher net interest margins and the positive effects of credit risk on net interest margins are greater in booms than in recessions The empirical findings of the study also show that bank-specific and industry indicators are closely contributed to the bank net interest margin

Keywords: net interst margins; business cycle; bank risk; Vietnamese banking system

1 Introduction

The relation between price-cost margins and aggregate shocks has been attracted voluminous theoretical and empirical studies In financial markets, the seminal papers contributed by Bernanke and Gertler (1989) and Bernanke, Gertler, and Gilchrist (1999) document that external finance premium, measured by the difference between the cost of external funds and internal funds, can act as a “financial accelerator” which exacerbates the aggregates shocks over business cycle

Bank net interest margins, usually calculated by the difference between interest income and interest expense divided by total asset, is one of the important metrics of bank profitability (Angbazo, 1997) Furthermore, it is also considered as an indicator to evaluate the efficiency of banks’ intermediation role in financial markets High interest margins not only signal a hardship

in investment due to high borrowing costs but they also discourage depositers to save when the deposit rates are low In the recessions, the phenomenon of high net interest margins (i.e countercyclical net interest margins) can amplify the adverse effects of economic shocks since they diminish opportunity to access sources of cheap funds which in turn reduces investment and production at marco level, resulting in a deeper recession period

Bank operation faces inevitable sources of risks in which credit risk and liquidity risk are shown to link to the probability of default In the period of the recent global financial recession, the occourence of these risks partially leads to the failure of the majority of banks (Imbierowicz & Rauch, 2014) Credit and liquidity risks are also documented to be bank-specific determinants of bank net interest margins (Demirgüç-Kunt & Huizinga, 1999; Saunders & Schumacher, 2000) However, very few studies examine the effects of these risks on net interest margins over business cycle Thus, the cyclical behavior of net interest margins as well as the effects of bank risks on bank net interest margin over business cycle should be investigated to understand the efficiency

of banks’ intermediation role

The Vietnamese banking system has experienced a rapid development since 1990 in terms of growth and profitability Despite not seriously affected by the recent global financial crisis, the credit growth rate in Vietnam significantly dropped in 2008 and 2009, revealing the effects of business fluctuation on the operation of Vietnamese banking industry When it comes to the issue,

an analysis of the relation between net interest margins and business cycle in Vietnam is critical

to provide implications for policy makers These observations motivate us to explore the impacts

of cyclical behavior on the relationship between bank net interest margins and bank risks in Vietnam

The results show that bank net interest margins in Vietnam are countercyclical Another interesting finding is that business cycle also significantly affects the relation between credit risk and net interest margins Credit risk is found to have a positive impact on net interest margins and its effect is greater in good time than in bad time These results provide an insight into Vietnamese banks’ intermediation role, suggesting important implications for interest rates policy

in Vietnam to mitigate the adverse influence of net interest margins during crisis

The remainder of the paper is structured as follows A brief review of literature is mentioned

in section 2 Section 3 describes data and methodology Section 4 provides empirical results while the conclusion is presented in section 5

2 Literature review

The theoretical model suggested by Ho and Saunders (1981) is considered as the starting point for empirical studies of net interest margins in banking sector In this seminal study, banks are viewed as risk-averse intermediaries between the supply side and the demand side of funds The existence of interest margins results from uncertainty faced by banks in transaction process Specifically, in a single period model, banks set the price of deposits and loans at the beginning of the period while facing asymmetric arrival time between deposits and loans demand In the purpose of maximising wealth at the end of the period, deposit and loan rates are set to minimise the asymmetry of arrival time between the two side of funds Ho and Saunders (1981) demonstrate four determinants of interest margins including (1) degree of managerial risk aversion, (2) size of transactions, (3) market structure of banking industry, (4) interest rates volatility

Known as a measure of bank profitability, net interest margins significantly depend on bank risks Based on Ho and Saunder’s work, Angbazo (1997) incorporates not only interest rate risk but default risk into the model Employing the data for the US banks from 1989 to 1993, the study demonstrates that riskier loans result in higher net interest margins Another type of bank risks, credit risk, is found to have positive effects on net interest margins in various studies, with the data for European Union member countries in Maudos and De-Guevara (2004) and Kasman, Vardar, and Tunç (2011) or for many countries around the world (Demirgüç-Kunt & Huizinga, 1999) Nassar, Martinez, and Pineda (2014) suggest a similar evidence in Honduras when examining quarterly data for the period 1998-2003 Exceptionally, Williams (2007) illustrates an opposite result to the conventional hypothesis when concluding that credit risk negatively affects net interest margins in the Australian banking system Net interest margins are also posited to have a positive relation with liquidity risk, i.e a negative relation with liquidity ratio This hypothesis is examined and confirmed in Angbazo (1997), Drakos (2003) and King (2013)

Liquidity ratio is also found to inversely affect bank spreads in Turkey (Aydemir & Guloglu, 2017) However, the negative sign between liquidity ratio and net interest margins is not confirmed in other studies such as Afanasieff, Lhacer, and Nakane (2002) or Islam and Nishiyama (2016) The relation between liquidity ratio and net interest margins is also found insignificant for the case of

Australia in Williams (2007) or Philippines and Indonesia in Doliente (2005)

In terms of cyclical behavior, Bernanke and Gertler (1989) and Bernanke et al (1999) provide evidence that the adverse influence of aggregate shocks can be amplified by external finance premium Using a “pricipal – agent” model, they conclude that the agency cost and external finance premium negatively affect borrowers’ net worth Based on the argument that the net worth of borrowers is procyclical, external finance premium is posited to be countercyclical In the banking sector, the impact of business cycle on net interest margins is firstly explored by Aliaga-Díaz and Olivero (2011) Using quarterly data in the US banking industry for the period 1979-2005, the study concludes that net interest margins are countercyclical It offers some potential explanations that net interest margins are significantly affected by several factors over the business cycle including monetary policy, interest rates volatility, the degree of financial deepening, liquidity and capital holdings and the proportion of total assets owned by major banks

in the system The countercyclical behavior of interest margins is also found in Turkish banking system by Turgutlu (2010) Aydemir and Guloglu (2017) conduct an investigation for the period 2002-2013 in Turkey and point out that business cycle affects not only bank spreads but also the relation between credit risk, liquidity risk and bank spreads

Based on the aforementioned literature, not much research capturing the fluctuation of net interest margin with regard to business cycle and the moderating effect of business cycle on the relationships the bank risks and net interest margins in emerging countries Furthermore, the fluctuation of the banking sectors in emerging markets after the 2007 – 2008 global financial crisis is of interest of many researchers Thus, the study may fill this gap by investigating the cyclical behavior of bank net interest margin and the effects of bank risks on bank net interest margin during business cycle in the Vietnamese banking sector

3 Methodology

To examine the relationship between net interest margin, bank risks and business cycle, three empirical specifications are developed in this study

The effect of business cycle and bank risks on bank net interest margin

Model 1: NIMit = 𝛽0 + 𝛽1GDPit + 𝛽2CRit + 𝛽3LIQit + 𝛽4INFit + 𝛽5LARit + 𝛽6CIit + 𝛽7CAPit +

The moderating effect of business cycle on the relationship between bank risks and bank net interest margin

Model 2.1: NIMit = 𝛽0 + 𝛽1GDPit + 𝛽2(LIQ*GDP)it + 𝛽3CRit + 𝛽4INFit + 𝛽5LARit + 𝛽6CIit +

Model 2.2: NIMit = 𝛽0 + 𝛽1GDPit + 𝛽2(CR*GDP)it + 𝛽3LIQit + 𝛽4INFit + 𝛽5LARit + 𝛽6CIit +

where 𝜀it denotes for error terms of bank i at time t

In the first empirical specification, the dependent variable is net interest margins denoted by NIM, measured by banks’ net interest income to total assets ratio Our main concern is the impact

of business cycle and bank risks on net interest margins Hence, following the study of Aydemir and Guloglu (2017), the first explanatory variable is the growth rates of real Gross Domestic Product (GDP) The sign of 𝛽1 is expected to be negative to reflect the countercyclical behavior of net interest margins as the argument in our literature Controlling for bank risks, we respectively include credit risk (CR) and liquidity risk (LIQ) into the model as mentioned in the studies of Williams (2007) and Aydemir and Guloglu (2017) We employ the ratio of liquid assets to total assets as a proxy for liquidity risk Since the effects liquidity risk on net interest margins is found with mixed results in literature, we do not expect the sign of this relation and consider it as an empirical issue To measure credit risk, we employ the ratio of preserves of risky loans to total loans Facing higher credit risk, banks are more likely to require higher risk premium on the loan rates, leading to higher net interest margins Therefore, the coefficient of CR in the model is expected to be positive Additionally, the interaction terms of LIQ with GDP in the second specification and CR with CDP in the third specification to respectively capture the relation between liquidity risk and credit risk with net interest margins over business cycle

The study incorporates the inflation rates (INF) into the model since higher inflation rates motivate banks to increase their loan rates to offset an increase in costs In Demirgüç-Kunt and Huizinga (1999) and Mendes and Abreu (2003), the inflation rates is found to positively impacts upon net interest margins The loan size (LAR), that is measured by the ratio of total loans to total asset, is documented to be positively related to net interest margins since a higher loan size signals

a riskier degree of bank operation that requires higher net interest margins (Maudos & Solís, 2009; Maudos & De-Guevara, 2004) The operating cost to total income ratio (CI) is found to inversely affect net interest margins (Maudos & Solís, 2009) The higher this ratio is, the less efficiently banks operate, leading to lower net interest margins Captial structure (CAP) is shown

to be one of the determinants of net interest margins (Saunders & Schumacher, 2000) Using the equity to total assets ratio as a proxy for capital structure, we expect a negative relation between CAP and NIM since a lower banks’ financial leverage ratio is likely to reduce banks’ risk-taking degree in lending activities and hence, reduce interest income

In terms of industry indicators, this study employs the proportion of banks’ assets in total assets of Vietnamese banking system (MKS) to measure banks’ market power McShane and Sharpe (1985) and Maudos and De-Guevara (2004) suggest a positive relation between market power and net interest margins Suspecting that the restructuring process in the Vietnamese banking system can significantly affect banks’ operation, we include a dummy variable (DRES) that is equal to one for the period 2012-2016 and equal to zero for the remaining years

Regarding empirical methodology, we firstly employ pooled OLS (Ordinary Least Square) estimators to estimate the coefficients of explanatory variables However, in panel data analysis, unobserved heterogenity can cause pooled OLS estimators become inefficient and inconsistent Therefore, we respectively use fixed effects model (FEM) and random effects model (REM) to control for this empirical issue We conduct F-test (pooled OLS versus FEM), Hausman test (FEM versus REM) and Breusch and Pagan Lagrangian test (pooled OLS versus REM) to select the most appropriate estimation method among them Furthermore, to achieve more robust results, we use Feasible General Least Square (FGLS) estimators to control for heteroskedasticity and autocorrelation

The study uses data on bank specific variables from the annual financial reports of 24 commercial banks for the period 2007-2016 These 24 banks account for 65% of the total assets

of Vietnamese commercial bank system in 2016 Due to some missing observations, our unbalanced panel dataset consists of 227 observations The data on macroeconomic variables including the inflation rates and the real Gross Domestic Product are collected from Asian Development Bank dataset Table 1 below presents the descriptive statistics for our sample while the correlation matrix of variables is provided in Table 2

Table 1

Sample descriptive statistics

Variables Number of

observations

Mean Standard

deviation

Minimum value

Maximum value

Table 2

Correlation matrix of variables

NIM GDP CR LIQ CR*GDP LIQ*GDP INF LAR CI CAP MKS DRES

NIM 1.000

GDP -0.175 1.000

CR 0.0845 -0.321 1.000

LIQ -0.245 0.288 -0.269 1.000

CR*GDP 0.0513 -0.086 0.958 -0.237 1.000

LIQ*GDP -0.274 0.537 -0.342 0.951 -0.259 1.000

INF 0.0872 0.018 -0.034 0.364 -0.023 0.303 1.000

LAR 0.3930 0.005 -0.076 -0.539 -0.058 -0.469 -0.132 1.000

CI -0.185 -0.396 0.087 -0.314 -0.012 -0.382 -0.123 -0.050 1.000

CAP 0.3220 0.080 -0.267 0.134 -0.274 0.157 0.234 0.125 -0.086 1.000

NIM GDP CR LIQ CR*GDP LIQ*GDP INF LAR CI CAP MKS DRES

MKS -0.057 0.016 0.366 -0.200 0.419 -0.190 -0.037 0.243 -0.227 -0.470 1.000

DRES 0.0652 -0.429 0.234 -0.569 0.120 -0.588 -0.615 0.088 0.498 -0.218 0.042 1.000

4 Empirical results

Figure 1 NIM over business cycle

To capture the time series of net interest margins in Vietnam, the study computes the average

of banks’ net interest margins for each year Figure 1 provides the information of the average net interest margins in Vietnam over business cycle It could be seen that the average net interest margins fluctuates over the examined time It is noted that the average net interest margin gradually increases from 2007 to 2009 whereas this period witnesses a decrease in the growth rates of real GDP On the contrary, during the post crisis period (2010-2012) with a drop in the real GDP growth rates, there is an upward trend of the average net interest margins in Vietnam The findings presented in Model 1 are consistent with our hypothesis Specifically, GDP shows its significant and negative effects on NIM, revealing that net interest margins in Vietnam are countercyclical, i.e they increase in bad economic condition This leads to the difficulty to invest and produce at macro level while discouraging savers to deposit their funds because of the bigger gap between loan rates and deposit rates The contraction period, hence, is likely to be lengthened

In other words, net interest margins in Vietnam can be considered as an “amplifier” of aggregate

-2.0000

4.0000

6.0000

8.0000

10.0000

12.0000

2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Average NIM and real GDP growth rates

Average NIM (%)

shocks, which is similar to external finance premium in the seminal studies of Bernanke and Gertler (1989) and Bernanke et al (1999) Therefore, it could be argued that banks’ intermediation role is less efficient in the bad time

There is no significant change in our results in terms of the signs of relations and p-values for the Model 2.1 and Model 2.2 when we respectively include the interaction terms of LIQ and GDP,

CR and GDP into the model Interestingly, the business cycle affects not only net interest margins but also the association between credit risk and net interest margins In table 5, CR*GDP shows positive and significant effects on NIM, suggesting that higher credit risk raises more net interest margins in booms than in recessions This finding may be due to that banks with higher credit risk find it easier to charge more borrowing cost to offset the risk exposure in a good economic condition than in a bad economic condition

Table 4

Estimation results on the moderating effect of business cycle and liquidity risk on net interest margin

**, * indicate significant levels of 5% and 10% respectively

Table 5

Estimation results on the moderating effect of business cycle and credit risk on net

interest margin

Note: **, * indicate significant levels of 5% and 10% respectively

The signs of coefficients of control variables are the same with our expectation and with the results in most previous studies, except MKS, the proxy for banks’ market power and CAP, the ratio of equity to total assets In the case of Vietnamese banks, market power shows a negative effects on net interest margins A potential explanation is that the Vietnamese bank market is dominated by a few and highly efficient banks with very high proportions of market share whereas the majority of banks share insignificant market power resulting in an inefficiency in cost management Additionally, it is possible that the Vietnamese banks compete to achieve more market shares and hence, they are willing to reduce the loan rates while increasing the deposit rates to attract more customers, resulting in lower net interest margins Similarly, the negative relation between net interest margins and equity ratio is not supported It could be argued that equity capital in the Vietnamese banks shows its advantages over its costs In other words, higher proportions of equity capital to total capital motivate the interest of banks’ shareholders to control and monitor banks’ operation, leading to higher net interest incomes This finding is similar to the case of Egyptian banks (Naceur & Kandil, 2009)