The effect of interest rates on bank performance and performance
based CEO compensation in the long run and short run: Evidence from
European Banks.
Master Thesis
Marco Helder
University of Groningen Faculty of Economics and BusinessMSc Finance
Supervisor Boris van Oostveen January 2018
ABSTRACT Words: 10.614
By using a unique dataset, we study the effects of long and short term interest rates on bank performance and performance based CEO compensation for 56 banks in Europe. We show that interest rates have a negative effect on return on assets for both the short run and long run, which is caused by banks taking significant long positions in interest rate sensitive derivatives and securities in order to gamble on an interest rate decrease. We also show that interest rates have a negative effect on CEO stock compensation. However, this effect is not present when CEO’s receive cash compensation, which leads us to believe there is some aspect of agency theory present in current CEO’s compensations structures.
2
I. Introduction
In 2008 the financial world got shaken up by a global financial crisis. During this period it became clear that financial stability is crucial to the economy as we know and use it. Especially banks play an important role in our economy, since they hold a major portion of the capital in circulation. In order to better withstand such shocks as witnessed in 2008, the banking sector should be financially stable and sustainable. This is also where the monetary policy has a direct impact on stability and sustainability. Through mainly interest rates and several other channels monetary policy is able to influence the stability and sustainability of banks to a great extent (Aydemir and Ovenc, 2016). As a result of current monetary policy, we currently live in an all-time low and negative interest rate environment, which is unique in economy and finance as we know it. At the same time, headlines of many worldwide newspapers claim that bank CEO’s receive substantial compensation packages. Especially during recent times when
unemployment was high and wage increase was low, these compensation packages lead to incomprehension and heavy discussions (Oxelheim, 2008).
This study will cover both these topics and will try to see if there is a relation between these all-time low interest rates on bank performance and thereafter on the substantial CEO bonuses. We will do so by studying the long run and short run1 effect of interest rates during 2011 to 2016 on bank performance, and thereafter the effect of interest rates on CEO compensation, all while using a unique dataset. This will also be the structure of this study going forward, where the first part of a section will be dedicated to the effect of interest rates on bank performance, while the second part focuses on the effect of interest rates on CEO compensation. Different studies have tried to link bank performance to performance based CEO compensation in the banking sector, but the results are not consistent. Positive relationships are found by Barro and Barro (1990), Duffhues and Kabir (2008), Faleye et al., (2013), while negative relationships have been found by Zhou et al., (2011) and Aduda (2011). Yusuf and Abubakar (2014) find no
significant results. The one thing these studies have in common, besides testing the relation between bank performance and performance based CEO compensation, is that they do not account for interest rates to any extend. Charumathi (2008) concluded that banks rely primarily on income through interest rate. Therefore, we argue that neglecting the interest rate will result in mixed results. Because of this we will research to what extent interest rates influence
3 bank performance, through net interest margin and return on assets. After that, we research the effect of interest rates on performance based CEO compensation for both cash and stock compensation. To the best of our knowledge we are the first to study the long term and short term effects of interest rates on both bank performance and performance based CEO
compensation in Europe, which leads to some interesting insights and results.
The contribution of this study exists of two parts. Firstly, the effect of interest rates on bank performance is in general nothing new. However, current studies only focus on one particular country (e.g. Alessandri and Nelson, 2015; United Kingdom, Aydemir and Ovenc, 2016; Turkey). These studies are interesting and resourceful, but they overlook the fact that the countries are more interconnected than one might generally think. Since the introduction of the European Union banks have physically easier access in the way that they can easily set up subsidiaries, but also financial markets are more open, connected and accessible. Therefore, for our study we look at Europe as a whole and include 16 different European countries in our final sample. Secondly, this study focuses on a less researched topic, namely the long term and short term effects of interest rates on performance based CEO compensation. Previous research on the effect of interest rates on performance based CEO compensation consist of Oxelheim (2008) and a replicated study on US data from Chiu et al. (2010). However, these two studies focus on the effects of anticipated and unanticipated macroeconomic factors. We slightly adjust the random effects model proposed by Oxelheim (2008) to test for the long run and short run effects of interest rates and slope of the yield curve on performance based CEO compensation. We adjust the model to capture long run and short run effects of interest rate and slope of the yield curve instead of the anticipated and unanticipated macroeconomic factors that Oxelheim (2008) uses. In order to do this, we hand collected the data for 2011 to 2016 on the CEO compensation for 56 banks across 16 European countries.
The first part of this study shows that the effect of interest rates on net interest margin are as expected. Where the short run effect is negative and significant, as a result of repricing
4 interest sensitive derivatives and securities through their trading operations, indicating a
gamble on interest rates decreases.
The second part of this study shows that, because of performance based CEO compensation, both in the long run and short run interest rates show significant and negative effects on stock compensation. The witnessed effects of interest rates on return on assets spread through to the CEO stock compensation. However, when CEO’s receive cash compensation we find no significant effect, which leads us to believe that there is indeed a form of agency theory present in the current stock compensation structures.
The next section reviews the literature regarding bank performance and performance based CEO compensation. Section III describes the methodology, which includes a system GMM and random effects approach. Section IV describes the unique data set, which includes hand collected data from 56 annual reports from 2011 to 2016. Section V shows the results of this study. Section VI provides a conclusion and discusses limitations and provides suggestions on further research. As mentioned before, for each section the first part is dedicated to bank performance, while the second part will be dedicated to performance based CEO
compensation.
II. Literature Review
As mentioned in the introduction, the first part of the literature review is dedicated to the effect of interest rates on banks performance, while the second part will be dedicated to CEO compensation.
Bank Performance
As Aydemir and Ovenc (2016) concluded, there are two channels through which bank performance is affected. The first one being bank-specific factors, the second being
5 Similar research is conducted by Goddard et al (2004), where they conclude that a positive relation between size and bank performance could be explained as economies of scale, market power and monetary protection (too-big-to-fail). Goddard et al. (2004) use ordinary least squares for their cross-sectional models and two step generalized method of moments for their dynamic panel models to identify the size-profitability relation, where they use return on equity as their profitability measure. Their results indicate that there is some evidence to support the size-profitability relation, but it is not systematic2. Demirguc-Kunt and Huizinga (2000) run regressions for Profit/Total Assets and Net Interest Margin/Total Assets and conclude that both are interrelated. They show that capitalization and the size of the loan portfolio have a positive effect on bank performance. They also conclude that the Net Interest Margin/Total Assets is a better estimator of how the financial structure of the bank influences bank performance. In order to determine bank profitability, Short (1979) uses country specific and bank-specific variables, which include leverage rate and asset growth. He argues that if two banks have the same return on assets, the one with the higher leverage rate will have a higher return on equity. The growth rate of assets is included because banks might sacrifice on their
performance now in order to gain market share, to earn more profits later on. Bourke (1989) researches determinants of bank profitability and uses a variant of net interest margin and return on assets as a dependent variable. For his bank-specific independent variables Bourke (1989) applies capital ratios and liquidity ratios. His results show that banks with high capital ratios, are better capitalized, have easier access to outside capital, resulting in higher profits. In a similar study Molyneux and Thronton (1992) replicate the methodology of Bourke (1989) and extend the sample from 1972 till 1981 for 12 countries located in Europe, North America and Australia to 18 countries in Europe with a time span over 1986 till 1989. Molyneux and Thronton (1992) find the same results as Bourke (1989) in regards to bank-specific variables. In more recent studies Alessandri and Nelson (2015) and Aydemir and Ovenc (2016) use similar bank-specific variables as Short (1979), in order to research the effect of interest rates and the yield curve on bank profitability. They incorporate two bank-specific variables, namely leverage rate and growth of assets. Aydemir and Ovenc (2016) suggests that the effect of leverage on profitability is negative, because higher leverage results in either more debt, or less assets, and therefore limits the banks available capital, resulting in lower profits. However, his findings show that in the Turkish market the leverage rate has a positive effect on the return on equity of banks, because as banks gain more debt, a higher leverage results in more profits. As
2
6 opposed to Short (1979), Aydemir and Ovenc (2016) expects a positive relation between asset growth and bank performance, because total assets includes a portion of loans, which would also increase as assets increase, resulting in higher profits. Alessandri and Nelson (2015) and Aydemir and Ovenc (2016) also incorporate a variable for GDP growth. They conclude that GDP growth is an estimator of a growing economy and as a result, there will be an increase in capital demand. This has a positive impact on banks, since they are able to offer more loans and thus increase bank performance.
Many of these studies that research bank profitability, use macroeconomic variables as well in order to explain macroeconomic effects, for example Short (1979), Bourke (1989) and
Molyneux and Thronton (1992) use interest rates and market growth3 to research the effect on bank profitability. All three studies find evidence for a positive relation between interest rates and profitability. Demirguc-Kunt and Huizinga (1999) run a regression analyses to show the effect of bank-specific factors and interest rates on bank performance through net interest margin and profit before taxes over total assets. They conclude that here is a significant size effect, measured through loans/total assets, but the sign changes as the dependent variable changes from net interest margin to profit before tax. They also show a positive and significant effect of interest rates on both net interest margin and profit before tax. Beckmann (2007) uses a regression analyses to determine the effects of macroeconomic factors and balance sheet structures on return on assets. The macroeconomic factors GDP growth and interest rate exhibit highly significant properties. However, opposed to earlier mentioned studies, Beckmann (2007) finds a negative relation between interest rates and return on assets, which he
concludes is caused by procyclical profitability pattern. Beckmann (2007) also concludes that the reason why the yield curve is not significant, is because of interest rate risk hedging. It is important to note that these studies mentioned do not make a distinction between long-term and short-long-term interest rates. However, Alessandri and Nelson (2015) argue that this distinction, between short run and long run effects, is important. This is because in the short run there will be repricing frictions as a result of maturity mismatch, resulting in a short term loss. However, banks might be able to benefit from higher interest rates in the long run.
Therefore, in order to make a clear distinction between long run and short run effects, we need to incorporate a variable for both these effects. Alessandri and Nelson (2015) and Aydemir and Ovenc (2016) use such variables in their system generalized method of moments regression,
3
7 where they test the effects of interest rate, yield curve slope and GDP on net interest margin and return on assets. In order to distinguish between short run and long run effects they incorporate a first difference variable for both the interest rates and yield curve. Their results show that for the net interest margin the long run the effect of interest rates is positive and significant, while in the short run they are negative and significant. Their conclusion is that banks suffer from repricing frictions caused by a mismatch in the duration of their assets and liabilities, therefore leading to negative effect of interest rate changes in the short run. However, when banks overcome these repricing frictions they are able to benefit from higher interest rates in the long run. Hence, in the long run interest rates have a positive effect on the net interest margin and return on assets of banks, while in the short run banks suffer from repricing frictions, leading to a negative effect of interest rate changes.
CEO Compensation
The effect of interest rates on performance based CEO compensation is not as widely
researched as the effect of interest rates on bank performance. However, Oxelheim (2008) and Chiu et al. (2010) research the effect of macroeconomic fluctuations on performance based CEO compensation in the long run, but they define these fluctuations as luck, and therefore research the effect of luck on performance based CEO compensation. Oxelheim (2008)
researches the effect of macroeconomic fluctuations of Swedish corporations during the period 2001-2006. They make a distinction between anticipated an unanticipated changes (i.e. luck). Oxelheim (2008) uses a random effect model to estimate the effect of macroeconomic fluctuations on performance based CEO compensation. The independent variables used are Sales, Tobin’s Q and anticipated and unanticipated interest rate and exchange rate changes. All variables are defined as natural logarithms, and should therefore be interpreted as elasticities rather than sensitivities. Oxelheim (2008) proposes two advantages to the use of natural logarithms, the first one being that ‘’it produces a better fit in terms of marginal effects’’ Oxelheim (2008), the second being that elasticity is not as sensitive to bank size as sensitivities. Alternative measures for Tobin’s Q were return on assets and return on equity, but Oxelheim (2008) concluded that Tobin’s Q has the least correlation with other variables and has more explanatory power. Oxelheim (2008) shows that the anticipated interest rate change is negatively and significantly related to performance based CEO compensation.
8 concludes that both the anticipated and unanticipated interest rates changes are negatively related to performance based CEO compensation, but the unanticipated interest rate changes are not significant. However, the magnitude of effects seems way larger in the US than in Sweden, which Chiu et al (2010) explain by the difference in variable compensation (i.e. stock compensation or cash compensation), where stock compensation is much higher in the US. It is important to note that variable compensation in both studies do not differentiate between cash or stock compensation in their models. It is important to distinguish between both
compensation methods, because CEO’s are able to influence the value of their compensation, if it is paid out in stocks. This is because of the fact that in general variable remuneration is not paid out on the day it was set. During this period CEO’s can influence stock prices, and
therefore the value of their compensation4, by investing in riskier projects. This phenomena is also known as agency theory (Tosun, 2016). Houston and James (1995) research if performance based CEO compensation schemes promote risk taking in the banking sector by using a sample from 1980 to 1990, which contains 134 commercial US banks. In order to determine this, they use equity-based incentives, measured as the percentage of stock compensation, and the level of risk taking, measured by the variance of stock returns. They conclude that there is no
evidence between CEO compensation and excessive risk taking. Different results are found by Cheng et al. (2015), who use a US based sample over a time period from 1992 to 2008 to research this effect of agency theory, and show that there is a strong relation between risk-taking and performance based CEO compensation. However, they conclude that it is uncertain whether CEO compensation structure leads to excessive risk taking, or that employees at a riskier bank require more compensation.
Hakenes and Schnabel (2014) use a theoretical model to research the effect of performance based CEO compensation with or without bail-outs in play, and conclude that compensations schemes induce CEO’s to take excessive risks even when there are no anticipated bail-outs. Taking agency theory in consideration we expect that there is a significant difference between cash and stock compensation. We expect that there is no significant effect of interest rates on cash compensation. However, because of agency theory, we expect that the effect of interest rates on stock compensation is negative and significant. Since there is sufficient evidence of
4
9 performance based CEO compensation5, we expect that interest rates have the same effect on CEO compensation as bank performance. This is because the value of the performance based CEO compensation relies for 78.97% on bank performance. Therefore, we expect that interest rates have a positive effect on stock CEO compensation in the long run, but a negative effect in the short run.
III. Methodology Bank Performance
For the first part of our research regarding the effect of interest rate on bank performance we use a two-step system generalized methods of moments model (GMM), as proposed by Alessandri and Nelson (2015). The use of two-step GMM has several advantages, especially when dealing with highly persistent data, such as interest rates (Alessandri and Nelson, 2015). Roodman (2006) lists several other advantages of GMM, including: “1) “small T, large N” panels, meaning few time periods and many individuals; 2) a linear functional relationship; 3) a single left-hand-side variable that is dynamic, depending on its own past realizations; 4) independent variables that are not strictly exogenous, meaning correlated with past and possibly current realizations of the error; 5) fixed individual effects; and 6) heteroskedasticity and
autocorrelation within individuals, but not across them”. Windmeijer (2005) concludes that two-step GMM performs better and is more accurate than one-step GMM. We report the Hansen J test statistic for over-identification restrictions, where H0 = no over-identification and
the Arrelano-Bond test (AR(2)) for autocorrelation to the second order, where H0 = no
autocorrelation.
The proposed model is as follows:
Yit = αYit-1 + β’Xit + γ’Mt + εit
εit = ηi + vit
Equation 1
10 Where Yit represents the income component (i.e. net interest margin and return on assets). Xit is
a vector for of bank-specific controls, Mt is a vector of macroeconomic variables, ηi is a bank
effect and vit is an idiosyncratic disturbance (Alessandri and Nelson, 2015)
As discussed earlier we employ two different performance measures as dependent variables, namely net interest margin and return on assets. Non-interest income through trading income might cancel out the effects of net interest margin on bank performance, because banks use their trading income to hedge interest rate risks. Therefore we use return on assets as an alternative measure to capture the effects of interest rates on bank performance. In equation 1 the bank-specific variables are Leverage t-1 (the lag of debt/total assets ratio) and Asset growth t-1 (the lag of growth of total assets). The macroeconomic variables in equation 1 are
GDPGrowth (growth of real GDP), GDP Growth t-1 (the lag of growth of GDP), Interest Rates (the
yearly average short-term interest rates), D Interest Rates (the first difference of interest rates), D Interest Rates t-1 (the lag of the first difference of interest rates), Slope Yield Curve6 (the slope
of the yield curve), D Slope Yield Curve (the first difference the slope) and D Slope Yield Curve t-1
(the lag of the first difference of the slope).
Under system generalized methods of moments we treat bank-specific variables (i.e. leverage and asset growth) as strictly endogenous and assume that macroeconomic effects are strictly exogenous. When doing this, we can use all macroeconomic variables as instruments
(Alessandri and Nelson, 2015).
As discussed in the literature review, we expect leverage to have a negative impact on bank performance. Higher leverage results in either more debt, or less assets, resulting in less available capital and therefore reduce the ability to gain profits. We expect asset growth to have a positive effect on bank performance, because if assets grow, the proportion of loans grow as well, increasing the income of a bank. We expect GDP growth to have a positive effect on bank performance, since banks are able to offer more loans in a growing economy. For long term interest rates and slope of the yield curve we expect a positive relation, while for short term we expect a negative relation.
CEO Compensation
For the second part of this study we estimate a random effect model to estimate the long run and short run effect of interest rates on performance based CEO compensation, for both stock and cash compensation. Oxelheim (2008) and Chiu et al. (2010) propose the initial model,
6
11 which consists of interest rates, exchange rate and consumer price index. We slightly alter the model of Oxelheim (2008), because it is the closest model that covers our subject, to capture the effect of both short-term and long-term interest rates. We drop the variables exchange rate and consumer price index, since they do not hold additional value in our regression and results. The proposed model is as follows:
Yit = α + β’Xit + γ’Mt + uit + εit
Equation 2
Where Yit is the vector of compensation method, Xit is a vector for of bank-specific controls, Mt
is a vector of macroeconomic variables, uit is an idiosyncratic disturbance and εit is the error
term. All variables are in natural logarithms.
As discussed earlier we employ two different compensation measures as dependent variables, namely cash compensation and stock compensation. Cash compensation might behave
different to interest rates than stock compensation, as a result of agency theory, which is discussed in the literature review. CEO’s are able to influence the value of their stock
compensation ex post. Which leads to ability to take risky investments, hoping the value of the shares will rise, resulting in a higher value of stock compensation. We use natural logarithms for all the variables, because it produces a better fit and that elasticity is not as sensitive to bank size as sensitivities (Oxelheim, 2008). In equation 2 the bank-specific factors are Sales (natural logarithm of total sales) and Tobin’s Q (natural logarithm of market value/book value). The macroeconomic factors are Interest Rate (natural logarithm of the yearly average short-term interest rates), D Interest Rates (natural logarithm of the first difference of interest rates), Slope Yield Curve (natural logarithm of the slope of the yield curve) and D Slope Yield Curve (natural logarithm of the first difference the slope).
As discussed in the literature review, we expect interest rates to have a positive effect in the long run, while we expect a negative effect in the short run.
IV. Data
12 we require detailed bank-specific information like leverage rate and asset growth. The
information on CEO compensation is hand-collected from analyzing Annual Reports and Remuneration Reports for all banks, over a time period from 2011 to 2016. Since not all banks list the remuneration for CEO’s, or do not even mention variable compensation in their reports, we had to eliminate 51 banks, resulting in a final sample of 56 over 16 European countries. It is important to note that not all banks in European report in euro’s. For example, Swiss banks still report in Swiss Francs, therefore the yearly average exchange rate is used to convert these domestic currencies to euro’s. Macroeconomic variables are gathered from OECD.stat, which include GDP growth and interest rates.
The final sample provides detailed information on the location of the bank, bank-specific factors such as asset growth, leverage rate, revenue, net interest margin and return on assets. Macroeconomic factors such as, short-term interest rate, long-term interest rates, slope of the yield curve and GDP growth. Compensation characteristics such as short-term compensation of CEO’s in either cash payments or share payments and details on how linked the variable
compensation is to financial performance. A detailed list of the banks and their locations can be found in Appendix 2. It might be beneficial to point out that during the collection of variable compensation most Eastern and Southern European countries were eliminated from the sample, which might be caused by Euro crisis, where banks from Eastern and Southern Europe suffered huge losses. As a results banks in those areas did not report on variable remunerations due to poor financial performance.
13
Table 1: Definitions of bank, country and compensation characteristics for 16 European countries in the period 2011 to 2016.
Definitions Source
Bank characteristics
Return on Assets Net profit before tax/Total Assets Orbis Bank Focus
Leverage Rate Debt/Total Assets Orbis Bank Focus
Asset Growth Yearly growth of Total Assets Orbis Bank Focus
Net interest margin Net interest income/Total Assets Orbis Bank Focus
Sales (ln) Natural logarithm of Revenue Orbis Bank Focus
Tobin's Q (ln) Natural logarithm of Market capitalization/Total Assets Orbis Bank Focus
Country characteristics
GDP Growth Annual GDP Growth Worldbank.org
Interest Rate Three months treasury rate OECD.Stat
D Interest Rate First difference of Interest Rate OECD.Stat
Interest Rate (ln) Natural logarithm of Interest Rate OECD.Stat
D Interest Rate (ln) Natural logarithm of D Interest Rate OECD.Stat
Slope Yield Curve 10 Year government bond yield - Three months treasury rate OECD.Stat
D Slope Yield Curve First difference of Slope Yield Curve OECD.Stat
Slope Yield Curve (ln) Natural logarithm of Slope Yield Curve OECD.Stat
D Slope Yield Curve (ln) Natural logarithm of D Slope Yield Curve OECD.Stat
Compensation characteristics
Cash compensation (ln) Natural Logarithm of yearly variable remuneration in cash Annual Reports/Remuneration Reports
Stock compensation (ln) Natural Logarithm of yearly variable remuneration in stock Annual Reports/Remuneration Reports
Financial weight Percentage of variable remuneration related to financial
14
Dependent variables
The choice of two different performance measures as dependent variable for bank
performance is important since banks earn income through more than one channel. Since net interest margin only captures one stream of income in a bank, it is important to include the return on assets as a dependent variable, since this results in a more complete overview of the effects of interest rates on bank performance. For the performance based CEO compensation we use the natural logarithms of both cash compensation and stock compensation. This results in the fact that variables should be interpreted as elasticity rather than sensitivity. Using natural logarithms provides two advantages, the first one being that elasticity is not as sensitive to bank size as sensitivities, and the second one being that natural logarithms provide a better marginal fit.
Control Variables
In order to test the effect of interest rates on bank performance we will control for leverage rate, asset growth and GDP growth. However, from the literature review we concluded that more control variables are possible, we preclude to use them in order to prevent over-identification of the model7. Leverage rate is measured as the Total Debt/Total Assets and is collected from Orbis Bank Focus. According to Aydemir and Ovenc (2016) higher leverage leads to lower profits, as banks have either more debt or less assets. This compresses the ability for banks to issue more loans. Therefore, a negative relation between leverage and bank
performance is expected. Asset growth is measured as yearly growth of total assets. Short (1979) argues that there is a negative relation between growth of assets and bank
performance. This is because banks might sacrifice profits now, in order to gain more market share in the future. However, Aydemir and Ovenc (2016) expect that asset growth will have a positive relation on bank performance. As assets grow, banks have more capital to issue loans, generating more profits. Therefore, we expect asset growth to have a positive effect on bank performance. GDP Growth is measured as the annual GDP growth per country and is used to control for the real activity of the European market. We expect that GDP growth has a positive effect on bank performance. This is because as GDP increases, banks have more possibilities to issue loans.
7
15 To test the effects of interest rates on CEO compensation we will also control for bank size, measured through sales (Oxelheim, 2008), where we measure Sales as total revenue. We use the natural logarithm of sales in order to fulfill the normality constrains.
Descriptive statistics
Table 2 shows the descriptive statistics of the variables used during this research. For bank characteristics we have a minimum of 254 observations for all variables. We used Orbis Bank Focus to get data on Tobin’s Q, but in 26 of the total 280 observations Orbis Bank Focus could not provide us with this data, resulting in 254 observations. The variable Sales was unavailable in Orbis Bank Focus for 3 observations, resulting in a total of 277 observations. The amount of observations is sufficient for system GMM. As Roodman (2006) discussed, system GMM is useful for ‘’small T, large N’’ panels. We see that the average return on assets is 0.4%, the average leverage rate is 93.6%, average asset growth is 0.6% and the average net interest margin is 1.2%.
For country characteristics we witness a low interest rate environment, with an average
interest rate of 0.25%. This is caused by current monetary policy, where the ECB pumps money in the capital market by selling bonds. This puts pressure on the interest rates, resulting in an all-time low interest rate. At some points during the sample short-term interest rates drop below the 0% mark, resulting in negative interest rates. We also see a negative D Interest Rate, namely -0.28%, which means that over our sample the average interest rate decreased. As shown in table 2, the slope of the yield curve has an average value of 1.61% over our sample, with a D Slope Yield Curve of -0.29%, which means that over our sample the slope of the yield curve became flatter.
16 2000). We see in table 2 that the average weight of financial performance is 78.97%, which means that banks still reward CEO’s mainly for achieving financial goals or targets.
Table 2: Descriptive statistics for bank, country and compensation characteristics for 16 European countries in the period 2011 to 2016.
Mean Stdev Min Max Obs
Bank characteristics
Return on Assets 0.004 0.008 -0.079 0.019 280
Leverage Rate 0.936 0.022 0.863 1.021 280
Asset Growth 0.006 0.119 -0.376 1.077 280
Net interest margin 0.012 0.006 -0.002 0.032 280
Sales (ln) 14.875 1.548 12.146 17.810 277 Tobin's Q (ln) -3.086 0.691 -5.809 -0.703 254 Country characteristics GDP Growth 1.12% 2.10% -7.30% 26.27% 280 Interest Rate 0.25% 0.57% -0.78% 2.22% 280 D Interest Rate -0.28% 0.27% -0.82% 0.05% 280 Interest Rate (ln) -5.033 1.72 -9.21 -1.71 280 D Interest Rate (ln) -3.74 2.01 -9.21 -1.66 280
Slope Yield Curve 1.61% 2.02% -0.14% 21.93% 270
D Slope Yield Curve -0.29% 1.14% -12.10% 7.57% 270
Compensation characteristics
Cash compensation (ln) 2.552 2.888 0 7.932 280
Stock compensation (ln) 1.919 2.795 0 7.719 280
Financial weight 78.97% 23.92% 25% 100% 39
Percentage Stock compensation 38.23% 32.91% 0.00% 100% 39
17
Table 3: Pair wise Correlation Matrix
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (1) Return on Assets - (2) Leverage -0.38 - (3) Asset Growth 0.15 0.01 - (4) Net Interest Margin -0.02 -0.35 -0.12 - (5) Sales (ln) -0.16 0.44 -0.2 -0.01 - (6) Tobin's Q (ln) 0.41 -0.55 0.06 0.13 -0.17 - (7) GDP Growth 0.37 -0.3 0.03 -0.1 0.02 0.33 - (8) Interest Rate 0.04 -0.1 -0.08 0.02 0.05 -0.02 -0.09 - (9) Slope Yield Curve -0.47 0.18 -0.15 0.39 0.06 -0.18 -0.47 0.03 - (10) Cash compensation (ln) 0.17 0.08 0.09 -0.18 0.04 0.04 0.03 -0.08 -0.26 - (11) Stock compensation (ln) 0.1 0.16 0.02 -0.11 0.34 -0.02 0.06 0.11 -0.15 0.43 - V. Results Bank Performance
We estimate equation 1 to examine the effects of interest rate on the net interest margin for banks in 16 European countries. Alessandri and Nelson (2015) research the effect of interest rate and the yield curve in a strong and developed market, namely, the United Kingdom. Aydemir and Ovenc (2016) research the same effects in an emerging market, in their case Turkey. This provides us with excellent background material to research the European market as a whole. Tables 4 and 5 contain our primary results.
18 and the interest rates would increase by 100 basis points, the net interest margin would
increase by 17.6% annually. Assuming the same for the yield curve, and 100 increase in basis points leads to an increase of the net interest margin of 12.6% annually, both are significant, respectively, at a 5% and 1% significance level.
These results can be explained by the fact that banks usually borrow on the short-term (i.e. deposits) and lend on long-term (i.e. mortgages). The yield curve steepens if there is an
increase in the long-term interest rates, or a decrease in the short term interest rates. A steeper yield curve often indicates a strong economic environment and bull markets. This opens up the market for banks because the supply and demand for capital will increase during these times. Therefore it is safe to say that the banks’ net interest margin, and hence profitability, is positively influenced by a steeper yield curve. Since the swords cuts two ways, the opposite is true as well. A flatter yield curve (i.e. increase in short-term interest rates and/or decrease in long-term interest rates) indicates a weak economic environment, resulting in a drop for supply and demand for capital. This affects the net interest margin of banks, and hence profitability, in a negative way (Estrella and Hardouvelis, 1991; Adrian et al., 2010). The results for the variable Interest Rates and Slope Yield Curve are in line with the findings of Alessandri and Nelson (2015) and Aydemir and Ovenc (2016).
When we look at the effect on interest rate changes or the short-term effect of interest rates, we see that both D Interest Rates and D Slope Yield Curve enter the regression negatively and significantly. The first difference of the slope of the yield curve is significant at the 1%
significance level and is, to some extent, of similar magnitude as the long-term slope. The first difference of interest rates is significant at the 5% significance level and is of similar magnitude as the long-term interest rates. This negative effect can be interpreted as, when there are unexpected interest rate changes, the bank loses money through their net interest margin in the short run. Combining this with the positive effects of long-term interest rates and the slope, we can conclude that there are short-term repricing frictions when banks encounter
unexpected interest rate changes. When repricing becomes available in the long run, banks are able to overcome these repricing frictions and benefit from higher interest rates (Alessandri and Nelson, 2015; Aydemir and Ovenc 2016).
19 significance level. Asset growth has a positive effect and is significant at the 10% significance level. However, GDP growth shows no significant effects on the net interest margin.
Table 4: Estimation results System GMM Net Interest Margin for 16 European countries, period 2011-2016.
Net Interest Margin
Coefficient Standard Error P-value
NIM t-1 -0.247 (0.331) 0.46 Leverage t-1 -0.096** (0.038) 0.014 AssetGrowth t-1 0.003* (0.001) 0.073 GDPGrowth 0.004 (0.006) 0.517 GDPGrowth t-1 0.062+ (0.041) 0.138 Interest Rates 0.176*** (0.065) 0.009 D Interest Rates -0.152** (0.067) 0.026 D Interest Rates t-1 -0.180* (0.097) 0.071
Slope Yield Curve 0.126** (0.052) 0.018
D Slope Yield Curve -0.191*** (0.057) 0.002
D Slope Yield Curve t-1 -0.056 (0.034) 0.102
Constant 0.101*** (0.037) 0.008 AR (2) 0.12 AR (2) p-value 0.901 Hansen 3.36 Hansen p-value 0.972 Number of observations 216 Number of banks 54
Notes: ***, **, *, + denote significance level at respectively 1%, 5%, 10% and 20%. First column refers to variable, second column refers to the coefficient of the variable, third column refers to the standard error, which is displayed in brackets, forth column refers to p-value. Variables: NIM (net interest margin), NIM t-1 (lag of net interest margin), Leverage t-1 (the lag of debt/total
assets ratio), AssetGrowth t-1 (the lag of growth of total assets), GDPGrowth (growth of real GDP), GDPGrowth t-1 (the lag of
growth of GDP), Interest Rates (the yearly average short-term interest rates), D Interest Rates (the first difference of interest rates), D Interest Rates t-1 (the lag of the first difference of interest rates), Slope Yield Curve (the slope of the yield curve), D Slope
Yield Curve (the first difference the slope), D Slope Yield Curve t-1 (the lag of the first difference of the slope), Constant (constant).
AR (2) statistic test for autocorrelation, H0 = No autocorrelation, Hansen J Test for over-identification, H0 = no over-identification. Since the income of banks is not only through their net interest income, but also through other activities, such as trading, we also need to take these into account. Unfortunately Orbis Bank Focus can’t provide us with information regarding trading income of banks. Therefore, we will apply the variable return on assets to capture the effect of interest rates on the total
20 including interest risks, we would observe no impact of interest rates on the return on assets if this hedging was perfect. However, if we look at the results in table 5, we see that both the Interest Rates and Slope Yield Curve are weakly significant8, and therefore we can, although in a weakly manner, conclude that banks do not perfectly hedge their interest rate risks.
The variable Interest Rates is of higher magnitude for return on assets than for net interest margin, but is not significant. However, the variable D Interest Rates is significant at the 20% significance level. The negative effect is expected if we take into account that banks do not perfectly hedge their interest rate risks, because banks suffer from repricing frictions in the short run.
Looking at the Slope Yield Curve we see that the sign of the variable changed when looking at the return on assets as opposed to net interest margin. This means that the banks lose money through their trading income when the slope becomes steeper. During our sample period, we witnessed a decrease in the slope of the yield curve (i.e. flattening of the yield curve), because D Slope Yield Curve is -0.29%, as can be seen in table 2, descriptive statistics. In general we would see a positive relationship between Slope Yield Curve and return on assets, because this would signal a bull market, which is discussed before. However, we witness a negative
relationship between Slope Yield Curve and return on assets, while this relationship was positive for the net interest margin. This would suggest that in the long run banks lose money through their trading income when the slope of the yield curve would increase (i.e. become steeper). Since we already concluded that banks do not perfectly hedge their interest rate risks, we can assume that banks take a long position in interest sensitive derivatives (i.e. bonds) through their trading income in order to profit from the gamble on interest rate decrease. Therefore, we can conclude that banks have a significant large position in long-term derivatives as opposed to short-term derivatives. Because of this, when the yield curve becomes steeper, long-term derivatives drop in value, resulting in a loss on the return on assets (Gros et al., 2016). This is also in line with findings of Demiralp et al. (2016), who conclude that if banks have excess liquidity, which is the case in the current low interest rate environment generated by the ECB, banks tend to increase their position in sovereign bonds.
The eventual difference between the positive effect of interest rates and negative effect of the slope of the yield curve suggests that for the return on assets the general effect of interest rate
8
21 changes is negative. This is contradicting to results by Alessandri and Nelson (2015) and
Aydemir and Ovenc (2016) who both find that interest rate changes have a positive effect on both the net interest margin and return on assets in the long run.
Table 5: Estimation results System GMM Return on Assets for 16 European countries, period 2011-2016.
Return on Assets
Coefficient Standard Error P-value
ROA t-1 -0.602 (0.735) 0.417 Leverage t-1 -0.160+ (0.120) 0.187 AssetGrowth t-1 0.001 (0.004) 0.748 GDPGrowth 0.093 (0.084) 0.274 GDPGrowth t-1 -0.095 (0.190) 0.617 Interest Rates 0.215 (0.192) 0.267 D Interest Rates -0.507+ (0.376) 0.183 D Interest Rates t-1 -0.26 (0.488) 0.596
Slope Yield Curve -0.446+ (0.269) 0.103
D Slope Yield Curve -0.038 (0.196) 0.846
D Slope Yield Curve t-1 0.013 (0.087) 0.879
Constant 0.160+ (0.118) 0.178 AR (2) 0.47 AR (2) p-value 0.638 Hansen 8.16 Hansen p-value 0.613 Number of observations 216 Number of banks 54
Notes: ***, **, *, + denote significance level at respectively 1%, 5%, 10% and 20%. First column refers to variable, second column refers to the coefficient of the variable, third column refers to the standard error, which is displayed in brackets, forth column refers to p-value. Variables: NIM (net interest margin), NIM t-1 (lag of net interest margin), Leverage t-1 (the lag of debt/total
assets ratio), AssetGrowth t-1 (the lag of growth of total assets), GDPGrowth (growth of real GDP), GDPGrowth t-1 (the lag of
growth of GDP), Interest Rates (the yearly average short-term interest rates), D Interest Rates (the first difference of interest rates), D Interest Rates t-1 (the lag of the first difference of interest rates), Slope Yield Curve (the slope of the yield curve), D Slope
Yield Curve (the first difference the slope), D Slope Yield Curve t-1 (the lag of the first difference of the slope), Constant (constant).
22 strong economic environment. These findings are in line with Alessandri and Nelson (2015) and Aydemir and Ovenc (2016).
The results for return on assets are less defined, as expected and discussed earlier. The interest rate shows a weak and negative relationship in the short run, which means hedging takes place to some extent, but is not perfect. In the long run we witness a weak and negative relationship for the slope of the yield curve, which is opposite from results of Alessandri and Nelson (2015) and Aydemir and Ovenc (2016). This is caused by banks gambling on interest rate decreases by holding large amounts of bonds in their trading portfolio.
CEO Compensation
We estimate equation 2 to examine the effects of interest rate on the compensation for CEO’s for both stock and cash compensation. As mentioned before, previous studies focused on macroeconomic effects. However, because we solely focus on the interest rate effects, and therefore the low interest rate environment, we only use variables for interest rates and slope of the yield curve.
Table 6 shows the random effects model for stock compensation for CEO’s. It shows that CEO compensation increases with 0.65% for every 1% increase in Sales. The regression also shows that interest rates have a negative economic and significant impact on the stock compensation in both the long run and short run. An increase in interest rates by 100 basis points, results in a drop of 17.6% of the value of stock compensation in the long run. As argued before, this effect should be insignificant for cash compensation, which it is. An increase in the slope of the yield curve by 100 basis points, results in a drop of 33.7% of the value of stock compensation in the long run. Both are significant at the 1% and 20% significance level respectively.
23 CEO can’t influence the value of the cash compensation after it has been determined. Since there are no significant variables for the cash compensation regression, we included the table in the appendix, and can be found in Appendix 1. The absence of significant variables in this regression, combined with results from the stock compensation regression, leads us to believe that there is indeed a problem of agency theory occurring when CEO’s get compensated through stock.
The negative and weakly significant short run effects can also be related to findings earlier in this paper. For return on assets we witness a negative and weakly significant effect for short run interest rates, which spill over to CEO compensation in the short run. As mentioned before, CEO compensation is dependent on financial goals. It is important to note that both these effects (return on assets and CEO compensation) are only weakly significant at the 20% significance level, and therefore should be interpreted with caution.
Table 6: Estimation Results Random Effects model Natural Logarithm of Stock CEO Compensation
Natural Logarithm Stock CEO compensation
Coefficient Standard Error P-value
Revenue (ln) 0.652*** (0.196) 0.001
Tobin's Q (ln) 0.271 (0.280) 0.332
Interest Rates (ln) -0.176*** (0.065) 0.007
D Interest Rates (ln) -0.082+ (0.064) 0.194
Slope Yield Curve (ln) -0.337+ (0.228) 0.139
D Slope Yield Curve (ln) -0.018 (0.036) 0.61
Constant -9.712*** (3.408) 0.004
Hausman p-value 0.6075
Number of
observations 253
Number of banks 52
Notes: ***, **, * and + denote significance level at respectively 1%, 5%, 10% and 20%. First column refers to variable, second column refers to the coefficient of the variable, third column refers to the standard error, which is displayed in brackets, forth column refers to p-value. Variables: Revenue (ln) (natural logarithm of revenue), Tobin’s Q (ln) (natural logarithm Tobin’s Q), Interest Rates (ln) (natural logarithm of yearly average short-term interest rates, D Interest rates (ln) (natural logarithm of the first difference of interest rates), Slope Yield Curve (ln) (natural logarithm of the slope of the yield curve), D Slope Yield Curve(ln) (natural logarithm of the first difference the slope), Constant (constant), Hausman test for random effects, H0 = random effects.
VI. Conclusion
24 by using a random effect models, on performance based CEO compensation in 16 European countries. Current literature show mixed results when studying bank performance and performance based CEO compensation of banks. However, they do not account for interest rates during their research. Charumathi (2008) argues that interest rates are the most
important influencer of bank performance for banks, therefore, the absence of interest rates in previous studies seem like a shortcoming, which might cause the mixed results.
The first part of our research focuses on the effect of interest rates on bank performance, for net interest margin and return on assets, for both the short run and long run. Earlier studies by Alessandri and Nelson (2015) and Aydemir and Ovenc (2016) predict a negative relation in the short run and a positive relation in the long run. By running the regression for net interest margin, we found similar results as Alessandri and Nelson (2015) and Aydemir and Ovenc (2016) and can conclude that interest rates and slope of the yield curve have a negative and significant effect, at the 5% significance level for interest rates and 1% significance level for the slope of the yield curve, on the net interest margin in the short run, which is caused by repricing frictions in the short run. However, once these repricing frictions have been overcome, banks should be able to profit from higher interest rates. We have shown that this is indeed the case and that there is a positive and significant relationship between net interest margin and long-term interest rates and slope of the yield curve at the 1% significance level and 5% significance level, respectively.
When we run the regression for return on assets our results weaken a bit, this is in line with findings of Alessandri and Nelson (2015). Banks hedge their interest rates through trading income, and therefore, the impact of interest rates on return on assets is weaker. If banks would perfectly hedge their interest rates we would witness no significant impact of interest rates on return on assets. However, this is not the case and for the short run we witness a negative and weakly significant impact of interest rates, at the 20% significance level. The slope of the yield curve also exhibits negative proportions in the short run, but is to no extent
25 on the return on assets (Gross et al., 2016). This is contradicting with findings of Alessandri and Nelson (2015) and Aydemir and Ovenc (2016). However, Demiralp et al. (2016) offers an alternative explanation, where if banks have excess liquidity, as a result of monetary policy, banks tend to invest more in foreign bonds.
The second part of our research focuses on the effect of interest rates on variable CEO
compensation. We distinguish between both stock and cash compensation and short and long run effects. We expect that interest rates have a negative and significant effect on stock compensation but no effect on cash compensation as a result of agency theory. In short this means that CEO’s can influence the value of their stock compensation by taking risky positions in derivatives. However, doing this has no effect on the value of cash compensation, since the value is fixed in the amount of euros. Our results show that there is indeed a difference between cash and stock compensation, where interest rates show negative and significant effects in both the long run and short run on stock compensations, whereas interest rates exhibit no significant effects on cash compensation. Combining this with findings earlier in the paper we can conclude that CEO’s indeed exhibit problems of agency theory, where they make risky investments in order to privately benefit through variable remuneration.
This study examines the effect of interest rates in short run and long run on both bank performance and CEO performance. However, this study has several limitations. Firstly, the effects between strong economic markets and less developed market might differ (Alessandri and Nelson, 2015; Aydemir and Ovenc, 2016). Due to a relative small sample size, we are unable to differentiate between differences in maturity levels between economies. Secondly, we were not able to collect data on other activities through which a bank can create income, such as trading income. For a more complete study this would be necessary. Thirdly, we ignore any ‘’too-big-to-fail’’ assumptions, which might lead to bail-out policy, and therefore excessive risk taking by CEO’s (Hakenes and Schnabel, 2012). And lastly, we assume that the weight of financial indicators is strictly related to bank performance. However, when looking closely at CEO contracts, differences might be examined to the extent that only a small percentage is related to actual performance and more to key performance indicators like capital ratio’s etc. Further research could focus more on the specifics of remuneration policies for CEO’s in order to determine the effects of interest rates and risk-taking more clearly.
26
Reference list
Adrian, T., Estrella, A., Shin, H.S., 2010. Monetary cycles, financial cycles and the business cycle. FRB N. Y. Staff Rep. 421.
Aduda, J., 2011. The relationship between executive compensation and bank performance in the Kenyan banking sector. Journal of Accounting and Taxation 3(6), 130–139.
Agrawal, A., Mandelker, G.N., 1987. Managerial incentives and corporate investment and financing decisions, Journal of Finance 42, 823–837.
Alessandri, P., and Nelson, B.D., 2015. Simple banking: profitability and the yield curve. Journal of Money Credit and Banking 47(1), 143–175.
Aydemir, R., and Ovenc, G., 2016. Interest rates, the yield curve and bank profitability in an emerging market economy. Economic Systems 40(4), 670-682.
Barro, J.R., Barro, R.J., 1990. Pay, performance, and turnover of bank CEOs. Journal of Labour Economics 8(4), 448–481.
Beckmann, R., 2007. Profitability of western European banking systems: panel evidence on structural and cyclical determinants. Banking and Financial Studies 17.
Blot. C., Hubert, P., 2016. Negative interest rates: Incentive or hindrance for the banking system? In: Gross, D., Blot, C., Hubert, P., Demertzis, M., Wolff, B., 2016. How do low and negative interest rates affect banks’ activity and profitability in the euro area? Monetary Dialogue.
Bourke, P., 1989. Concentration and other determinants of bank profitability in Europe, North America and Australia. Journal of Banking and Finance 13, 65–79.
Charumathi, B., 2008. Asset liability management in Indian banking industry - with special reference to interest rate risk management in ICICI bank. International Conference Proceedings of the World Congress on Engineering 21, 1149-1154.
Cheng, I., Hong, H., Scheinkman, J.A., 2015. Yesterday’s heroes: compensation and risk at financial banks. The Journal of Finance 70(2), 839-879.
27 Demiralp, S., Eisenschmidt, J., Vlassopoulos, T., 2016, The impact of negative interest rates on bank balance sheets: Evidence from the euro area. Mimeo ECB.
Demirguc-Kunt, A., Huizinga, H., 1999. Determinants of commercial bank interest margins and profitability: some international evidence. World Bank Economic Review 13, 379–408.
Demirguc-Kunt, A., Huizinga, H., 2000. Financial structure and bank profitability. Policy Research Working Paper 2430.
Duffhues, P., Kabir, R., 2008. Is the pay-performance relationship always positive? Evidence from the Netherland. Journal of Multinational Financial Management 18(1), 45–60.
Estrella, A., Hardouvelis, G.A., 1991. The term structure as a predictor of real economic activity. Journal of Finance 46(2), 555–576.
Faleye, O., Reis, E., Venkateswaran, A., 2013 The determinants and effects of CEO –employee pay ratios. Journal of Banking and Finance 37(8), 3258–3272.
Goddard, J., Molyneux, P., Wilson, J.O., 2004. The profitability of European banks: a cross-sectional and dynamic panel analysis. The Manchester School 72(3), 363–381.
Gross, D., 2016. Low rates = low banks’ profit? In: Gross, D., Blot, C., Hubert, P., Demertzis, M., Wolff, B., 2016. How do low and negative interest rates affect banks’ activity and profitability in the euro area? Monetary Dialogue.
Hakenes, H,. Schnabel, I., 2014. Bank bonuses and bail-outs. Journal of Money, Credit and Banking 46(1), 259-288.
Houston, J.F., James, C., 1995. CEO compensation and bank risk is compensation in banking structured to promote risk taking? Journal of Monetary Economics 36, 405-431.
Jensen, M.C., Murphy, K.J., 1990. Performance pay and top-management incentives. The Journal of Political Economy. 98(2), 225–264.
Matolcsy, Z.P., 2000. Executive cash compensation and corporate performance during different economic cycles. Contemporary Accounting Research Vol. 17(4), 671-692.
28 Oxelheim, L., Wihlborg, C., Zhang, J. 2008. Executive compensation and macroeconomic
fluctuations, in: Oxelheim, L., Wihlborg, C. (Eds.), Markets and compensation for executives in Europe. Oxford, pp. 233-261.
Roodman, D., 2006. How to do xtabond2: an introduction to ‘’difference’’ and system’’ GMM in Stata. Working Paper 2013.
Short, B.K., 1979. The relation between commercial bank profit rates and banking
concentration in Canada, Western Europe and Japan. Journal of Banking and Finance 3, 209– 219.
Smirlock, M., 1985. Evidence on the (non) relationship between concentration and profitability in banking. Journal of Money Credit and Banking 17(1), 69–83.
Tosun, K., 2016. The effect of CEO option compensation on the capital structure: a natural experiment. Financial Management, 953-979.
Windmeijer, F., 2005. A finite sample correction for the variance of linear efficient two-step GMM estimators. Journal of Econometrics 126, 25–51.
Yusuf, I. and Abubakar, S., 2014. Chief executive officer’s pay in Nigerian banks: are corporate governance, bank size, and performance predictors?. 4TH International Conference on a Century of Public Sector and Corporate Governance in Nigeria, 1914–2014, Nasarawa State University Keffi Nigeria, 7–9 October.
29
Appendix I.
Table 7: Estimation Results Random Effects model Natural Logarithm of Cash CEO Compensation
Natural Logarithm Cash CEO compensation
Coefficient Standard Error P-value
Revenue (ln) 0.153 (0.198) 0.440
Tobin's Q (ln) 0.264 (0.292) 0.366
Interest Rates (ln) -0.054 (0.070) 0.440
D Interest Rates (ln) -0.062 (0.068) 0.357
Slope Yield Curve (ln) -0.303 (0.240) 0.206
D Slope Yield Curve (ln) -0.012 (0.038) 0.752
Constant -0.745 (3.466) 0.830
Hausman p-value 0.1698
Number of
observations 253
Number of banks 52
30
Appendix II.
HSBC Holdings Plc UNITED KINGDOM
BNP Paribas FRANCE
Deutsche Bank AG GERMANY
Crédit Agricole S.A. FRANCE
Barclays Plc UNITED KINGDOM
Société Générale SA FRANCE
Lloyds Banking Group Plc UNITED KINGDOM
Royal Bank of Scotland Group Plc (The) UNITED KINGDOM
UniCredit SpA ITALY
ING Groep NV NETHERLANDS
Credit Suisse Group AG SWITZERLAND
Banco Bilbao Vizcaya Argentaria SA-BBVA SPAIN
Nordea Bank AB (publ) SWEDEN
Standard Chartered Plc UNITED KINGDOM
Natixis SA FRANCE
Commerzbank AG GERMANY
Danske Bank A/S DENMARK
ABN AMRO Group N.V. NETHERLANDS
DnB ASA NORWAY
KBC Groep NV/ KBC Groupe SA-KBC Group BELGIUM
Svenska Handelsbanken AB SWEDEN
Skandinaviska Enskilda Banken AB SWEDEN
Nationwide Building Society UNITED KINGDOM
Swedbank AB SWEDEN
Dexia SA BELGIUM
Erste Group Bank AG AUSTRIA
Bankia, SA SPAIN
Deutsche Postbank AG GERMANY
Banque Nationale de Belgique SA BELGIUM
Unione di Banche Italiane Scpa-UBI Banca ITALY
Allied Irish Banks plc IRELAND
Julius Baer Group Ltd SWITZERLAND
Jyske Bank A/S (Group) DENMARK
National Bank of Greece SA GREECE
Banco Comercial Português, SA-Millennium bcp PORTUGAL
Deutsche Pfandbriefbank AG GERMANY
Storebrand Group-Storebrand ASA NORWAY
Aareal Bank AG GERMANY
Banque Cantonale Vaudoise SWITZERLAND
EFG International SWITZERLAND
Banco BPI SA PORTUGAL
31 Azioni
Banca Piccolo Credito Valtellinese-Credito Valtellinese Soc Coop
ITALY
Valiant Holding SWITZERLAND
Investec Plc UNITED KINGDOM
Basellandschaftliche Kantonalbank-Banque Cantonale de Bale-Campagne
SWITZERLAND
SpareBank 1 SR-Bank ASA NORWAY
Banque Cantonale de Genève SWITZERLAND
Sydbank A/S DENMARK
Liechtensteinische Landesbank AG-National Bank of Liechtenstein
LIECHTENSTEIN
Vontobel Holding AG-Vontobel Group SWITZERLAND
Edmond de Rothschild (Suisse) S.A SWITZERLAND
SpareBank 1 SMN NORWAY
Van Lanschot Kempen NV NETHERLANDS
VP Bank AG LIECHTENSTEIN
