The relationship between non-interest income and bank
insolvency risk
Nick Rustenhoven Student number: 2225433 First Supervisor: Silviu Ursu
Abstract
During the last decades many countries experienced turbulent financial times. A stable banking sector can definitely reduce turbulence. Several factors impact the stability of individual banks and the banking sector as a whole. Banks no longer only earn money by charging interest on loans. A bank can also earn money with a broad range of non-interest activities. This papers studies the relationship between the share of these non-interest generating activities and bank insolvency risk. This paper find that a shift towards non-interest income increases the insolvency risk of cooperative banks. This paper does not find significant results for other categories of banks.
Keyword: non-interest income, interest income, bank insolvency risk
2
Table of content
Table of content ... 2
1. Introduction ... 3
2. Literature review ... 6
2.1.1Benefits of non-interest income ... 6
2.1.2 Costs of non-interest income ... 7
2.2Empirical literature ... 8 2.3 Hypothesis ... 9 2.4 Control variables ... 10 3. Methodology ... 12 4. Data ... 13 5. Results ... 17 5.1.1 Commercial banks ... 17 5.1.2 Cooperative banks ... 17 5.1.3 Savings banks ... 19
5.1.4 Implications for supervisors and regulators ... 21
5.1.5 Implications for managers and shareholders ... 21
5.2 Robustness tests ... 21
5.3 Endogeneity ... 22
6. Conclusion ... 23
3
1. Introduction
One of benefits of the existence of banks exists is that they facilitate economic growth. Banks are able to allocate resources efficiently. They move money from those who have a surplus (depositors) to those who have a deficit (lenders). This stimulates economic activities. Banks pay interest to depositors and receive interest from lenders. The difference between it is called net interest income and is the main source of income for banks. Nowadays, banks can generate income in other ways than providing loans This new source of income is called non-interest income. It contributed more and more to the total income of banks in the last decades. The recent financial crisis remained us of the fact that banks can also fail and harm the economy, therefore it is important to understand the factors that impact bank insolvency risks. The easing of regulations has for example impacted the business models of banks
significantly (Mercieca et al 2007). This paper addresses the question if a larger share of non-interest contributes to insolvency risk of banks in Europe.
This paper adds empirical evidence on the relationship between the share of non-interest income and bank insolvency risk. The results are mainly relevant for regulators and supervisors, but also, to a more limited extent, for shareholders and managers.
Regulators and supervisors are interested in the prevention of bankruptcy, because bank failure can have serious negative consequences for governments and even for financial stability. The explanation behind this claim is related to the fact that banks are quite different from non-financial companies. Firstly, the creditors of banks are different from the creditors of non-financial companies. The latter are often large banks, professional investors or other large institutions, where for banks a large share of their creditors exist of depositors. Problems arise in case of bankruptcy of a bank. Normally if a company goes bankrupt, the company will often be forced to liquidate its assets. The proceeds will first cover the costs of the
4 Furthermore the failure of a bank can cause problems for other banks. Banks make loans to other banks. These loans are called interbank loans. If a bank defaults, it also defaults on its interbank loans. This will lead to losses for the supplier of the loan, which is in this case another bank. If the loans are sufficient large, the supplying bank can also end up in financial distress, which might cause problems for other banks. A bank failure might thus trigger a chain of subsequent failures, which can have a negative impact on financial stability (Rochet & Tirole 1996).
From a manager and shareholder perspective insolvency risk is theoretically less important. Financial theory suggests that firms should not use valuable resources to reduce risk, because investors can diversify risk away by holding a diversified portfolio (Stiroh & Rumble 2006). However, there are also several reasons why shareholders should care about insolvency risk. Shareholders want managers to maximize the value of the company. Direct and indirect bankruptcy costs are important factors in the valuation of bank. For direct
bankruptcy costs it is obvious that higher insolvency risks means that a bank is more likely to incur direct bankruptcy costs. These costs are any legal, administrative, accounting, and advisory payments associated with bankruptcy. Higher insolvency risk also leads to higher indirect bankruptcy costs. First of all, financial distress might trigger debt covenants that limited the power of the management. Hence, managers might be unable to make investments in positive net present value (NPV) projects. Furthermore, banks will be unable to attract external capital or they can only attract external capital at higher costs. This might again lead to the inability of a bank to make investments in positive NPV projects. Finally, financial distress might lead to the loss of key employees and/or key customers. (Schroek 2002)
For managers there is an additional incentive to reduce insolvency risk. If a company goes bankrupt, managers will lose their job. Furthermore, they might have large share of their wealth invested in the bank’s stocks (Stiroh & Rumble 2006).
It is important to understand how banks can generate non-interest income before one could discuss the possible relationship with insolvency risk. Non-interest activities exist of all financial products and services a bank can offer, except for the providing of loans and taking deposits. Examples of non-interest activities are investment banking, securities brokerage, insurance activities (DeYoung & Rice 2004).
5 z-score the lower the insolvency risk of the bank. The z-score is calculated as the sum of the return on assets (ROA) and the capital to assets ratio (CAR) divided by the standard deviation of the return on assets (STDROA). A change in z-score can be caused by a change in ROA,
STDROA or CAR. I make a distinction between savings banks, commercial banks and
cooperative banks. Commercial banks are fundamentally different from cooperative and savings banks on several points. Cooperative and savings banks operate locally and focus on providing loans and collecting deposits of households and small businesses. Commercial banks are bigger and operate on a larger scale. Furthermore they are profit maximization organizations and engage more in non-interest generating activities. (Köhler 2015).
I find that an increase in the share of non-interest income increases the insolvency risk of cooperative banks. This increase in insolvency is caused by a change in STDROA. More specifically, an increase in non-interest income increase the STDROA of cooperative banks; in other words is causes profits to fluctuate more. Most of the income of cooperative banks is interest income. This makes them theoretically vulnerable to shocks that affect the interest income of a bank, which leads to large fluctuations in profits. Adding non-interest income should therefore lead to a decrease in insolvency risk. There are several arguments that explain why I find the opposite. Cooperative banks are regionally focused and therefore able to build up long-term relationships with customers. These long-term relationships reduces information asymmetry, which has several advantages. First, it allows cooperative banks to improve their judgment about the creditworthiness of borrowers. This helps them in their decision to grant (additional) credit. Second, the continuous contact between the borrower and the bank gives value information on how to price loans, if collateral is needed and if it is necessary to attach other conditions to the loan. Third this makes it easier to renegotiate loans in case the borrower faces distress (Mercieca et al 2007; Elyasiani & Goldberg 2004). The regional focus might at the same time explain why an increase of non-interest income even leads to an increase in STDROA. Cooperative banks sell their non-interest and interest generating products and services probably for a large share to the same customers. If a
6
2. Literature review
A bank becomes insolvent if its assets are worth less than its debt. Banks have often only low levels of equity and high levels of debt. This means they have only small buffers to absorb losses. To prevent financial distress it is crucial for banks to generate stable profits and/or increase profitability. Banks traditionally generate income with lending activities. Some researchers claim that adding non-interest income can help stabilize profits and/or increase profits and hereby reducing insolvency risk. I will discuss the potential benefits and costs of non-interest income.
2.1.1Benefits of non-interest income
The first explanation of how an increase in the share of non-interest income can lead to lower insolvency risk is trough income stabilization. The logic behind this claim is the following: the major part of bank income still exist for most banks of interest income (Kohler 2015). Diversifying into non-interest reduces the dependence on interest income. If activities that generate non-interest income are uncorrelated, or, at least imperfectly correlated with interest generating activities, then combining these activities can stabilize the income stream of a bank (Chiorazzo Milani, & Salvini 2008). In other words combining multiple sources of income that are not perfectly correlated can improve the stability of total income. Suppose a random shock leads to an decrease in interest income. The income of banks that mainly generate income with loans would be seriously affected in a negative way. Now suppose this same shock does not lead to a decrease of non-interest income, but would even lead to an increase of non-interest income. This would mean that the income of banks with significant levels of non-interest income generating activities will not decrease as much as for banks that solely rely on interest income.
7 This increases the number of potential customers. If they can services these potential
customers without having to make major additional costs such as investments in technology or hiring additional personal, they can achieve costs savings. This is called economies of scope. Secondly, lending activities create long-term relationship between the bank and its customers. This provides banks with a large base of easy reachable potential customers in case they start offering new non-interest generating products. Moreover, the relationships between a bank and its clients can provide valuable information on the specific needs of each customer.
2.1.2 Costs of non-interest income
There are also arguments that suggest insolvency increases as risk as a result of an increase of non-interest income. First, in certain situations increasing the share of non-interest income does not necessarily stabilize the total income of banks. This is especially likely to happen if non-interest income is much more volatile than interest income. In this case adding non-interest income exposes banks to a more volatile income source. This will certainly happen if correlation between non-interest and interest income is high. In this case a random shock thus will affect both non-interest and interest income in the same way. If these two conditions hold, then banks with only a small proportion of non-interest income will already have a less stable income stream then banks without non-interest income. Banks with larger shares of non-interest income will have an even more unstable stream of income. Given the above conditions, it can be beneficial for banks to have an income stream that is almost entirely generated with interest income. It is hard to exactly quantify at what percentage an increase of non-interest income decreases income stability. However in general if the volatility of interest income is higher than interest income and correlation between non-interest income and non-interest income is high, then adding non-non-interest income decreases income stability for most levels of non-interest income.
If correlation between non-interest and interest income is low and non-interest income is more volatile, matters become more complicated. Berger et al. (2015) show with
simulations that sometimes stabilizes profits and other times destabilize profits.
8 often long-term relationships. These relationships are characterized by high switching costs, which makes it hard to terminate the relationship. Therefore, banks are more likely to lose non-interest income generating customers than interest generating customers. Of course this means that it is also easier to attract customers from competitors. This moving around of customers leads to large fluctuations of profits. Second, non-interest income generating activities could require a high investment in technology or labour. This rise in fixed costs increases the operational leverage of a bank. As long as revenues are stable this is not a problem, however a sudden drop in revenues might lead quicker to losses. Contrary, banks profitability will sharply increase once fixed cost are covered and revenues continue to increase.
A reason why correlation between interest and non-interest income can be high is because of cross-selling different products to the same customer. This means that banks provide both loan services and non-interest products to the same customers. This limits the benefits of providing a broader range of products (Stiroh 2004). For example if a customer defaults, it leads to a reduction in both interest income and non-interest income.
Not only a decrease in the stability of profits can increase insolvency risk, but also a decrease in profitability can make banks more likely to default. Offering too many products increase the complexity of the banks. This can lead to an increase in agency costs, which will reduce profits (Baele et al 2007).
2.2Empirical literature
The empirical results on the relationship between non-interest income and insolvency risk are mixed. A part of the literature finds that more non-interest income leads to less insolvency risk, while others find the reverse. I start with the literature that finds an decrease in insolvency risk. Secondly, I mention the literature that finds an increase in insolvency risk. Finally, I will describe literature that studies the impact of non-interest income on income stability and profitability.
Kohler (2014) studies the relationship between the share of non-interest income and bank insolvency risk of German banks in the 2002 – 2012 period. He examines the
9 for both cooperative and savings banks. Kohler (2015) also investigates the relationship between non-interest income and bank insolvency risk for a sample consisting of banks located across Europe. He finds that commercial, cooperative and savings banks lower their insolvency risk if they increase their share of non-interest income.
Other researchers find a positive relationship between the share of non-interest income and insolvency risk. Mercieca et al (2007) find that an increase of non-interest income
increases the insolvency risk of banks. They use a sample of 755 small European banks for the period 1997 – 2003 and perform a panel data analysis with z-score as dependent variable. Dermiguc-Kunt and Huizinga (2010) find similar results for an international sample that consist mainly of commercial banks. However, they perform a cross-sectional regression. As inputs for the z-score they use the averages of the return on assets, standard deviation of assets and capitalization over the 1995 – 2007 period. Stiroh (2004) performs quite a similar study as Demeriguc-Kunt and Huizinga (2010). He uses the same methodology, but a
different sample. His sample period consist of the years 1984 up to and including 2001 and consist of commercial banks located in the United States. Not surprisingly, he also find that banks with a larger share of non-interest income have higher insolvency risk.
Furthermore a part of the researchers investigates the relationship between the share of non-interest income and profitability and/or income stability separately. Note that these results provide no direct link between the share of non-interest income and insolvency risk. However, profitability and income stability are components of the z-score and therefore relevant for this study. Lee et al. (2013) investigate what the impact of non-interest income is on bank profitability and income stability in Asia over the period 1995 and 2009. They find an positive relationship between the level of non-interest income and total income stability of commercial banks. For cooperative and savings banks they find a negative relationship. Furthermore, they find that non-interest income positively affect profitability for cooperative banks, but negatively for savings banks.
2.3 Hypothesis
10 Hypothesis 2: A larger share of non-interest income will decrease insolvency risk (increase the z-score).
The literature and theory provide mixed results on the relationship between non-interest income and bank insolvency risk. Therefore it is hard to predict if non-interest income reduces insolvency risk or increases insolvency risk. The same story holds for the impact of non-interest income on income stability. Therefore, I form the two different hypothesizes (1,2).
Hypothesis 3: A larger share of non-interest income will increase income stability (decrease
STDROA).
Hypothesis 4: A larger share of non-interest income will decrease income stability (increase the STDROA)..
The literature provides and theory provides mixed results on the relationship between the share of non-interest income and the STDROA. Therefore I form two different hypothesizes (3,4).
Hypothesis 5: A larger share of non-interest income will increase ROA.
I expect a positive relationship between non-interest income and profitability(5). High competition in the market for interest income reduces the margins for banks.
Furthermore combining non-interest income and income activities can yield economies of scope. Finally relationships can be used to sell non-interest generating products to existing customers.
2.4 Control variables
11 There are two theories on how size affects bank insolvency risk. Firstly, larger banks have less insolvency risk. Their size allows them to spread fixed costs over a larger number of financial services and products. Furthermore they are often more geographically diversified than small banks, which makes them less vulnerable to region specific shocks. The second theory predicts that large banks are more risky. Larger banks can become too complex, which makes them hard to manage. Furthermore larger banks can be too-big-to-fail. Especially the largest banks expect to get bailed out in case of insolvency, because of their systematical importance. More specific, the negative impact of insolvency on the economy would be much larger than the costs of saving the bank. These too-big-too-fail banks might abuse this
situation by taking excessive risk, which can increase their insolvency risk (Rahman et al. 2015).
Furthermore, I control for short-term funding. Basically, short-term funding can be divided into two main categories. These categories are customer deposits and other short term non-deposit funding. Customer deposits are funds that people put in their savings account, while other non-deposit funding exist for example out of interbank loans and commercial paper. Other non-deposit funding are characterized by fixed maturities shorter than 1 year. Banks often borrow new short-term debt to repay existing short-term debt when it is about to mature. This is called the rollover of debt. Banks that cannot borrow new debt, despite being solvent, can struggle to repay their maturing liabilities. This liquidity problem can force a bank to sell some of its assets in order to repay their liabilities. This is especially undesirable when they have to sell assets at a large discount. This will lead to large losses, which
12
3. Methodology
I use accounting data in order to investigate the relationship between the share of non-interest income and bank risk. The use of accounting data increases the number of banks that can be investigated. This increases the heterogeneity in the sample. A limitation of accounting data is the smoothing of profits, which is particularity undesirable when estimating volatility. On the other hand stock markets overreact to announcements.
I use a time-varying z-score as proxy for bank insolvency risk. This is consistent with other papers that study bank insolvency risk(e.g. Hesse & Cihák 2007 ; Berger et al 2015). The z-score is a measure of bank insolvency risk; more precisely it is the number of standard deviations the return on assets needs to drop to completely wipe out the bank’s equity.
𝑧 − 𝑠𝑐𝑜𝑟𝑒𝑖𝑡 = 𝑅𝑂𝐴𝑖𝑡+ 𝐶𝐴𝑅𝑖𝑡 𝑆𝑇𝐷𝑅𝑂𝐴𝑖𝑇 (1)
The z-score is equal to the return on assets of a bank (ROA) plus the equity to assets ratio (CAR) divided by the standard deviation of the return on assets (STDROA). Where ROA is the return on assets of bank i in year t and CAR is the equity to asset ratio of bank i in year t.
STDROA is calculated over a three year time window. This measure of the STDROA is
consistent with several other studies (Chiaramonte et al. 2015). Changes in the z-score can be caused by changes in the profitability, capitalization and income stability of a bank. In my empirical analysis I use the natural logarithm of the z-score which is consistent with most researchers (e.g. Kohler 2014; Kohler 2015 ).
Most researchers calculate ROA as net income over assets (e.g. Lown et al 2000; Stiroh 2004), while others use pretax profits(Köhler 2015; Bertay et al 2016). I use pretax profits, because banks can carry taxes forwards or backwards to smooth net income, which induces a downward bias on the standard deviation of the ROA.
I measure the share of non-interest income (NNII) as |total non-interest operating income| / (|total non-interest operating income| + |total net interest income|). Furthermore, I measure the size of a bank (SIZE) as the natural logarithm of assets. I use the ratio of
overhead costs to assets as a proxy for overhead costs (OVERHEAD) and net interest income to total assets as a proxy for the net interest margin (NIM). I use customer deposits to total short term deposits as a proxy for the funding structure of a bank (FUNDING).
13 Yit = α + β1NNIIit−1+ β2SIZEit−1+ β3CARit−1+ β4FUNDINGit−1 (2)
+ β5OVERHEADit−1 + β6NIMit−1+ αi + ϒit + εit
The dependent variable is the natural logarithm of the z-score of bank i at time t. I control for cross-sectional and time fixed effects (ϒ). Because risk is likely correlated within a bank over time, I adjust standard errors for clustering at the bank level (Berger et al 2015).
A positive coefficient for NNII means that a larger share of non-interest income will improve the z-score of banks; in other words a larger share of non-interest income will reduce the insolvency risk of a bank. This change in z-score can be caused by three things. First, a bank can become more profitable. Second, a bank can improve the stability of its profits. Finally, a bank can increase its capital ratio. Theory provides arguments for why the share of non-interest income can affect profitability and income stability, therefore I also run
regressions with ROA and STDROA as dependent variable. This allows me to interpret how non-interest income impacts the individual components of the z-score. Furthermore, it is possible that the share of non-interest income has no impact on the insolvency risk of bank. This does not mean that it has no impact on profitability or income stability. It is perfectly possible that non-interest income has a positive impact on profitability, but a negative impact on the stability of profits. If these effects offset each other there will be no impact on bank insolvency risk.
4. Data
14 consisting of 1388 banks. The total unbalanced panel consists of 12482 observations.
Furthermore I make a sample split based on the specialization of a bank. The banks in my sample are commercial banks, cooperative banks and savings banks.
Table 1: Bank characteristics
Panel A shows the characteristics of commercial, cooperative and savings banks. It shows the mean, standard deviation and median. Panel B shows the score and its components. The z-score is calculated as (return on assets + equity to asset ratio) / the standard deviation of the
return on assets.
The sample of commercial banks consist of 259 banks. Table 5 in the appendix shows
that these banks are located across 26 countries. Almost every country has commercial banks. Bankscope does not provide commercial bank observations for Norway and Liechtenstein. Banks from Germany and Switzerland dominate the sample, together they make up for 40% of the sample. Commercial banks are profit maximizing organizations owned by shareholders (Köhler 2015). Commercial banks provide a broad range of financial services and products. They are active in the traditional lending market, but they also generate income with non-interest activities. Furthermore commercial banks are larger than cooperative and savings banks (Hesse & Çihák 2007). Panel A of table 1 confirms these claims. On average
commercial banks’ assets size is 51.9 billion euro, however the distribution is highly skewed. Therefore, it is better to look at the median which is 1.7 billions of euro. This makes them bigger than cooperative and savings banks. Commercial banks also have the largest share of non-interest income. They earn approximately 39.1% of their income with non-interest generating activities. Furthermore, commercial banks have an average ROA of 0.9%, which makes them the most profitable. However, the STDROA of commercial banks is the largest,
Panel A: Bank’s characteristics
Commercial banks Cooperative banks Savings banks Mean S.D Median Mean S.D Median Mean S.D Median
Non-interest income (%) 39.1 21.9 35.2 27.2 7.6 27.1 21.3 7.6 22.6
Size in billions 51.9 211.0 1.7 4.5 72.9 0.5 3.2 16.8 0.9
Net interest margin (%) 2.6 1.9 2.0 2.5 0.6 2.6 2.1 0.6 2.1
Overhead ratio (%) 2.7 2.6 2.0 2.3 0.6 2.3 1.7 0.6 1.8
Funding ratio (%) 75.0 22.6 81.9 84.1 9.2 85.0 80.6 11.1 82.0
Panel B: Bank’s Z-score and its components
Commercial banks Cooperative banks Savings banks
Mean S.D Median Mean S.D Median Mean S.D Median
Z-score 103.1 166.3 40.6 112.2 136.0 68.1 138.0 153.3 84.9
Equity to asset ratio (%) 10.9 8.2 8.7 7.2 2.4 6.9 5.5 2.9 5.3
Return on assets (%) 0.9 - 0.6 0.6 - 0.5 0.4 - 0.3
Standard deviation of the return on assets (%)
15 which makes their income stream the least stable. This income volatility is probably the reason that commercial banks have to highest capitalization ratio. This high level of capital must prevent financial distress in case of severe losses. However, commercial banks have, despite these large buffers, the highest insolvency risk. This is reflected in a lower average (103.1) and median (40.6) of the z-score. The large difference between average and mean again indicates that the z-score is highly skewed.
The two other samples consist respectively of 568 cooperative banks and 561 savings banks. Both samples are dominated by German banks. Table 5 shows that cooperative banks are almost solely operating in Germany. Only 7 other countries report a small number of cooperative banks. In total German banks make up more than 96% of the cooperative bank sample. Most of the savings banks are located in Germany(62%), but also a large share of them are located in Switzerland(33%). Only 4 other countries have savings banks. The business models of cooperative and savings banks are very similar. They both have a traditional banking business model. This is characterized by (1) a strong focus on taking deposits and providing loans to households and small and medium businesses and (2) a strong regional focus(Bülbül et al 2013). This strong regional focus implies that cooperative and savings banks have a smaller base of potential customers than commercial banks. Table 1 shows that cooperative and savings banks are indeed smaller than commercial banks. The assets of cooperative banks have an median value of 0.5 billion euro. The assets of savings are with a value of 0.9 billion euro somewhat higher. The data confirms that cooperative and savings banks focus mostly on generating income in a traditional way. Cooperative and savings banks earn respectively on average 27.2% and 21.3% of their income with non-interest generating activities. Furthermore, cooperative and differ fundamentally in their organizational structure compared to commercial banks. Cooperative banks are owned by their members. They cannot sell their shares for a profit, therefore they do not have an incentive to put pressure on the management to increase the value of the bank. Hence, the goal of cooperative banks is twofold. First they want to support the economic undertakings of their clients. Second they want to achieve this in a profitable way. This implies that
16 indication that banks with lower income stability have on average higher buffers (7.2% for cooperative banks vs. 5.5% for savings banks). Finally, the median values of the z-scores for cooperative and savings banks are respectively 68.1 and 84.9. These results suggest that savings banks have the lowest insolvency risk compared to cooperative and commercial banks.
Figure 1 shows that the share of non-interest income was most of the years stable. Only in the pre-crisis period(2005-2006) and the crisis period (2007-2009) shows some variability in the average share of non-interest income. The share of non-interest income peaked for cooperative banks in the year 2006 with a value of 35%. The share declined again after the year 2006. Commercial banks had the highest levels of non-interest income in the 2005-2007 period. This results indicates that banks reduced or were forced to reduce their non-interest share as result of the crisis. Another explanation could be the drop in non-interest income is caused by trading losses. (Bertay et al. 2015)
Figure 1
This figure shows the share of non-interest income for commercial, cooperative and savings banks. This figure shows that the average share of non-interest for banks in the 2010-2014 period was stable.
17
5. Results
In this section I will provide and discuss the results of the panel data analysis. Furthermore I will answer the hypothesizes.
5.1.1 Commercial banks
The descriptive data shows that commercial banks are the largest banks and have the highest insolvency risk. This makes commercial banks the most interesting for regulators and supervisors. First I will discuss the impact of the share of non-interest income on commercial bank insolvency risk. Panel A of table 2 shows a negative coefficient, but it is insignificant. This suggest that increasing the share of non-interest income does not significantly impact the insolvency risk of commercial banks. I do find a significant positive relationship between the share of non-interest income and ROA. The economical meaning is limited, because a 10% increase in non-interest income increases the return on assets of a commercial bank by 0.07%, where 0.9% is the average ROA for commercial banks. This result provides evidence that increasing the share of non-interest income can enhance profitability by a small amount. This positive effective on profitability is apparently not enough to significantly reduce insolvency risk. Furthermore I find that the share of non-interest income has no significant impact on income stability. Commercial banks have already a relative large share of non-interest
income, which implies that they do not depend on one single source of income. This might be an explanation why commercial banks do not benefit from additional non-interest income. Theoretically increasing the share of non-interest income can reduce income stability if they provide non-interest generating products and services to the customers to which they also provide loans. This would make them more vulnerable to customers specific shocks. This also seems not the case. An explanation could be that commercial banks operate on a large scale, which could mean that they provide non-interest and interest income generating products and services to different segments.
5.1.2 Cooperative banks
18 z-score. A lower z-score means higher insolvency risk. Furthermore I find that non-interest income significantly increases the STDROA. This finding suggest that the income stream of cooperative becomes less stable if they increase their share of non-interest income. A 10% increase in the share of non-interest income increases the STDROA with 0.03%. The average
STDROA of cooperative banks is 0.2%. The economical impact is moderate. These findings
are in alignment with the findings of Mercieca et al (2007). Cooperative banks rely mostly on interest income. This dependence should theoretically make them vulnerable to shocks that affect interest income. The business model of cooperative banks can provide an explanation why I do find the opposite. They are relative small banks, which focus on providing loans to households and small businesses. In these segments cooperative banks have a comparative advantage over large banks. This comparative advantage can be contributed to the
development of long-term relationship with their customers (Mercieca et al 2007). This in contrast to large bank, which have difficulties in building up long-term relationships with small customers. Especially these long-term relationships can help cooperative banks to minimize the impact of interest specific shocks. Long-term relationships provide banks with lots of information about their customers. This reduce information asymmetry, which provides several benefits. First, it allows cooperative banks to improve their judgment about the creditworthiness of borrowers. This helps them in their decision to grant (additional) credit. Second, the continuous contact between the borrower and the bank gives value information on how to price loans, if collateral is needed and if it is necessary to attach other conditions to the loan. Third it makes it easier to renegotiate loans in case the borrower faces distress (Mercieca et al 2007; Elyasiani, Goldberg 2004). Additionally, the local focus of cooperative banks leads to cross-selling of non-interest products or services and loans. Selling multiple products to the same customer does not necessarily reduce exposure to shocks in interest income. If a shock causes multiple customers to default on their loans, it is likely that the bank will also lose the profits they are generating on non-interest products or services. This can even decrease the stability of total income. Interestingly, I find that the share of non-interest income positively affects ROA. This finding again confirms that increasing the share of non-interest income can enhance profitability. However, this increase in ROA is small. A 10% increase in the share of non-interest income increases ROA with 0.03%. The average
19 5.1.3 Savings banks
20
Table 2: Panel data analysis
This table shows the results of the regression analysis. The dependent variables are the z-score, STDROA and ROA. Panel A shows the results for commercial banks. Panel B shows the result for cooperative banks. Panel C shows the results for savings banks. standard errors are reported in parentheses. I cluster standard errors at the bank level * indicates statistical significance at the 10% level, ** at the 5% level, and *** at the 1% level.
Panel A: Commercial banks
Z-score STDROA ROA
NNII – share of non-interest income -0.002 (0.003) 0.001 (0.002) 0.007* (0.004) SIZE 0.000 (0.072) 0.171 (0.104) -0.181 (0.136) FUNDING - Funding ratio 0.003** (0.001) -0.002* (0.001) -0.003 (0.003) OVERHEAD – overhead ratio -0.064*** (0.021) 0.109*** (0.033) -0.126 (0.103) CAR – capitalization 0.020*** (0.003) 0.008 (0.007) 0.010 (0.011) NIM – net interest margin 0.007 (0.025) 0.073*** (0.024) 0.217*** (0.055) Constant 3.578** (1.587) -3.597 (2.381) 4.460 (3.033)
Adjusted R2 0.57 0.51 0.53
Observations 2193 2193 2193
Redundant fixed effect test 0.000 0.000 0.000
Hausman test 0.000 0.000 0.000
Panel B:Cooperative banks
Z-score STDROA ROA
NNII – share of non-interest income -0.005*** (0.002) 0.003*** (0.000) 0.003* (0.002) SIZE 0.281*** (0.100) -0.0688*** (0.023) -0.197*** (0.073) FUNDING - Funding ratio 0.011*** (0.004) -0.003*** (0.000) 0.006*** (0.001) OVERHEAD – overhead ratio -0.011 (0.064) -0.013 (0.018) -0.126*** (0.043) CAR – capitalization 0.159*** (0.043) -0.022 (0.019) -0.085** (0.031) NIM – net interest margin -0.136* (0.078) 0.055*** (0.017) 0.121** (0.052) Constant -2.903 (2.188) 1.737*** (0.528) 4.501*** (1.634)
Adjusted R2 0.39 0.29 0.39
Observations 5171 5171 5171
Redundant fixed effect test 0.000 0.000 0.000
Hausman test 0.000 0.000 0.000
Panel C: Savings banks
Z-score STDROA ROA
NNII – share of non-interest income -0.003 (0.004) 0.002** (0.001) 0.003 (0.002) SIZE 0.005 (0.078) 0.042* (0.025) -0.104** (0.052) FUNDING - Funding ratio -0.004 (0.004) 0.000 (0.000) 0.003*** (0.000) OVERHEAD – overhead ratio -0.021 (0.091) 0.047** (0.022) 0.015 (0.038) CAR – capitalization 0.127*** (0.025) -0.001 (0.010) 0.001 (0.013) NIM – net interest margin -0.249*** (0.087) 0.080*** (0.014) 0.160** (0.082) Constant 4.678** (1.839) -1.040* (0.564) 1.893 (1.212)
Adjusted R2 0.429 0.47 0.60
Observations 5118 5118 5118
Redundant fixed effect test 0.000 0.000 0.000
21 5.1.4 Implications for supervisors and regulators
‘
The results regarding insolvency risk are important for regulators and supervisors, because it is their job to prevent the bankruptcy of banks. An increase in the share of non-interest income seems to increase the probability that a cooperative bank defaults. It is therefore in the interest of supervisors and regulators that cooperative banks do not increase their non-interest share. However, cooperative banks might not be the first priority of supervisors and regulators, because of their relative small size and historically low default rates (Bülbül et al 2013). More important are commercial banks, because they are larger and more likely to fail. A larger share of non-interest income does not affect the insolvency risk of commercial banks. This implies that regulators and supervisors do not have to spend
additional resources to prevent commercial bank in varying their share of non-interest income
5.1.5 Implications for managers and shareholders
The results show that non-interest income can enhance profitability. Especially commercial banks seem to profit. This mean that managers of commercial banks should definitely consider to invest in non-interest generating products and services. However, economic significance is small. This makes it especially for managers of cooperative banks unattractive to diversify into non-interest income, because for managers of cooperative banks maximizing profits is not the main goal. Furthermore, non-interest income increases the insolvency risk of a cooperative bank. Therefore it is unwise for managers of cooperative banks to diversify into non-interest generating activities.
5.2 Robustness tests
I provide two additional robustness checks. First, the sample of cooperative banks exist almost entirely out of German banks. This means that an increase in the share of non-interest income could possibly increase the insolvency risk of all banks in Germany.
22 This German sample consist of 934 banks. Most of them are cooperative banks 55.6% and savings banks 37.5%. A small proportion of this sample exist of commercial banks (3.9%). Panel A in table 3 shows that an increase in the share of non-interest income does not significantly increase insolvency risk of all German banks.
This sample exist of cooperative, commercial and savings banks. On average banks in this sample have a share of 15%. Most of the banks in the sample are savings banks (65.2%) and cooperative banks (22.5%). The rest of the sample consist of commercial banks (12.3%). The share of non-interest income has again no significant impact on the z-score of banks.
Table 3: Panel data analysis
This table shows the results of the regression analysis. The dependent variables is the z-score. Panel A shows the results for German banks and banks with low levels of non-interest
income. The standard errors are reported in parentheses. I cluster standard errors at the bank level. * indicates statistical significance at the 10% level, ** at the 5% level, and *** at the 1% level.
Panel A: Robustness tests
German Low non-interest income
NNII – share of non-interest income -0.001 (0.005) .0.001 (0.002)
SIZE 0.262*** (0.05) 0.130** (0.054)
FUNDING - Funding ratio -0.002 (0.003) -0.001 (0.001) OVERHEAD – overhead ratio -0.111** (0.05) -0.115*** (0.058) CAR – capitalization 0.05*** (0.011) 0.003 (0.008) NIM – net interest margin -0.000 (0.025) -0.022 (0.054) Constant -0.885 (1.271) -2.556 (1.117)
Adjusted R2 0.39 0.45
Observations 8503 3009
Redundant fixed effect test 0.000 0.000
Hausman test 0.000 0.000
5.3 Endogeneity
A limitation of this study is that I cannot completely solve the possible endogeneity problems in a econometric way. I control for bank fixed effects to make sure that
23 cooperative banks decide to increase their share of non-interest income as result of an increase in insolvency risk. In practice this is highly unlikely, because one of the goals is to provide support the economic undertakings of their clients (Bülbül et al 2013). It is therefore more likely that cooperative banks increase their non-interest income in case their clients demand financial non-interest generating products or services.
6. Conclusion
I find that a large share of non-interest income does increase the insolvency risk for cooperative banks, but not for commercial and savings banks. The increase in insolvency risk for cooperative banks is caused by a decrease in the stability of total profits. This decrease in income stability can be explained by the fact cooperative banks have a more stable income stream if they focus on interest income. In this market they have a comparative advantage over commercial banks, because they are better in developing and maintaining long-term relationships. These relationships provides several advantages that can help smooth interest income. Adding non-interest income to their income stream does only destabilize the total income stream. Cooperative and banks do probably offer products and services that generate non-interest income to their existing customers. This makes them more vulnerable to the default of a customer. They do not only lose their interest income, but also their non-interest income. I do find the same result for savings banks with regards to income stability, but this decrease in income stability is not significantly reflected in a lower insolvency risk.
This study has several limitations. First, I cannot completely solve the endogeneity issues from a econometric perspective. Therefore repeating this study with more sophisticated econometric tools could improve the validity of this study. Furthermore non-interest income can be generated with a heterogeneous group of activities. Some activities might reduce insolvency risk. This might offer an opportunity for future research.
.
7. References
Baele, Lieven, Olivier De Jonghe and Rudi Vander Vennet, 2007, Does the stock market value bank diversification, Journal of Banking and Finance 31, 1999-2023.
24 Beltray, A. C., Demirgüç-Kunt, A., Huizinga, H., 2015 Is the financial safety net a barrier to cross-border banking? Working paper, World Bank, Washington, DC
Berger, Allen N., El Ghoul, Sadok, Guedhami, Omrane, Roman, Raluca A., 2013. Internationalization and bank risk. Working Paper. University of South Carolina.
Brunnermeier M. K., Dong G. N., Palia D. 2011. Banks Non-Interest Income and Systemic Risk. Working Paper, Princeton University.
Bulbul, D., R. Schmidt, and U. Sch¨uwer (2013). Savings Banks and Cooperative Banks in Europe. White Paper Series Center of Exellence SAFE No.5
Chiaromente, L., Croci E, Poli., F., 2015 Should we trust the z-score? Evidence from the european banking industry. Global Finance Journal
Chiorazzo, V., Milani, C., Salvini, F., 2008. Income diversification and bank performance: evidence from italian banks. Journal of Financial Services Research 33, 181–203.
Demirgüç-Kunt, A., Huizinga, H., 2010. Bank activity and funding strategies: the impact on risk and return. Journal of Financial Economics 98, 626–650
DeYoung and Rice (2004), ‘How do banks make money? The fallacies of fee income’, Federal Reserve Bank of Chicago Economic Perspectives, Vol. 28, pages 34-51.
DeYoung, R. and Roland, K. P. (2001), Product Mix and Earnings Volatility at Commercial Banks: Evidence from a Degree of Total Leverage Model, Journal of Financial
Intermediation. 10, 54-84.
Elyasiani, E., and L. G. Goldberg (2004): "Relationship Lending: A Survey of the Litera ture," Journal of Economics and Business, 56, 315-330.
25 Hesse, H., Cihak, M., 2007. Cooperative Banks and Financial Stability. IMF Working Paper 07/02. International Monetary Fund, Washington D.C.
Köhler, M. 2014. Does non-interest income make banks more risky? Retail-versus investment-oriented banks. Review of Financial Economics, 23, 182–193.
Köhler, M. 2015. Which banks are more risky? The impact of business model on bank stability. Journal of Financial Stability, 16, 195–212
Lee, C.C., Yang, S.J., Chang, C.H. (2014a) “Non-interest income, profitability and risk in banking industry: A crosscountry analysis”, North American Journal of Economics and Finance, vol. 27, pp. 48 - 67.
Lown, Cara S., Carol L. Osler, Philip E. Strahan, and Amir Sufi 2000. The Changing Landscape of the Financial Service Industry: What Lies Ahead? Economic Policy Review, Federal Reserve Bank of New York 6, 39-54.
Mercieca, S., Schaeck, K., Wolfe, S., 2007. Small European banks: benefits from diversification? Journal of Banking and Finance 31, 1975–1998.
Rahman, M.M., Zheng, C., Ashraf, B.N., 2015. Bank size, risk-taking and capital regulation in Bangladesh. Eurasian Journal of Business and Economics 8, 95-114
Rochet, Jean Charles and Jean Tirole (1996) “Interbank Lending and Systemic Risk.” Journal of Money, Credit, and Banking 28(4):733–762.
Schroeck, Gerhard (2002). Risk Management and Value Creation in Financial Institutions.
Stiroh, K. J., 2004.Diversification in banking: Is noninterest income the answer? Journal of Money, Credit, and Banking , 36 , 853–882
26 Rochet, Jean Charles and Jean Tirole (1996) “Interbank Lending and Systemic Risk.” Journal of Money, Credit, and Banking 28(4):733–762.
8. Appendix
Table 4
Panel A shows the variables that I use in my regression analysis. The data I use is collected with the help of Bankscope.
* Size and z-score are two highly skewed variables, therefore I use the natural logarithms of these two variables.
Panel A: Regression analysis variables
Description Measurement
NNII Share of non-interest income
ABS(operating non-interest income)
[ABS(operating non-interest income) + ABS(net interest income)]
SIZE* Bank size Assets in billions of euro
NIM Net interest margin Net interest income
Total Earning assets
OVERHEAD Overhead ratio Overhead costs
Total Assets
FUNDING Funding ratio Customer deposits
Total short-term funding
Z-SCORE* Insolvency measure CAR+ROA
STDROA
CAR Capitalization Equity
Assets
ROA Profitability Pretax profits
Total Assets
27
Table 5
Panel A shows the sample composition. Together the samples contain banks located across 28 countries in Europe. I make a distinction between commercial, cooperative and savings
banks, because they have different business models.
Panel A: Sample composition
Commercial banks Cooperative banks Savings banks
Count (%) Count (%) Count (%)
