Influence of central bank policy and competition on interest margins

(1)

Influence of central bank policy and competition on interest margins

Bachelor thesis by Pepijn Schreurs, student number 6152112

University of Amsterdam, programme: Economics and business, specialization Economics

Supervisor: C.G.F van der Kwaak MSc Final version: August 1st₂₀₁₃

(2)

This paper investigates the effect of monetary policy on interest margins. The hypotheses are that interest margins react to changes in a central bank’s policy rate, and that this effect is stronger under high levels of competition. Specifically, this is tested in the European banking market in the 1999-2009 period, using annual country-level data. Combining the literature on interest rate pass-through and on interest margin determinants, it arrives at a model of interest margins of banks in an

oligopolistic environment. Doing so, it may provide some insights into the dynamics of the European banking sector.

Interest income is one of the main sources of income for any bank. It is a result of its income from loans and its cost on deposits, which derive from its deposit rate id, its loan rate il and the amounts of deposits and loans. Although due to diversification the relative importance of interest income has decreased in recent decades (Lepetit et al., 2008), it is still an interesting topic for academics and policymakers. There is a large amount of literature on what exactly determines interest income, starting with a 1981 paper by Ho and Saunders. More recent contributions, for example Carbó and Rodriguez (2007) are extensions of their model. The literature consistently mentions market structure or the level of competition as an important factor.

One of the main instruments of monetary policy is a central bank’s policy rate. The pass-through of policy rates to market loan and deposit rates is essential to the efficiency of this instrument and therefore the mechanisms underlying pass-through are often studied. A high level of competition in the market for loans and deposits is commonly associated with quick interest rate pass-through. (van Leuvensteijn et al., 2013)

This paper aims to extend the literature on interest income determinants by searching for a link between policy rate changes and changes in interest margins. The second question is whether the level of competition is an influential factor in this process. Theory is tested using a panel data model of eurozone banking sectors in the 2000-2009 period.

Following this introduction, section 2 gives an overview of the existing literature on interest income and interest rate pass-through. The third section elaborates on the theory and hypotheses

underlying this research, and the fourth section presents the model and data. Section five gives the results and section six concludes.

(4)

2. Literature review

The following section first gives an overview of the evolution of the literature on interest income models. After that it looks at the literature on competition in interest rate pass-through and summarizes the implications for interest income models.

Interest income determinants

Most of recent research on interest margins values the 1981 model by Ho and Saunders. The main motivation behind their paper is a growing concern about the effects of interest rate fluctuations on banks. They start their analysis by first looking at the previous two strands of literature that tried to explain the behaviour of banks, namely the hedging hypothesis and the microeconomics strand. The hedging hypothesis sees banks as balancing a portfolio of assets (loans) and liabilities (deposits) and calibrating the maturities of these assets and liabilities so that it minimizes the risk of changes in interest rates. The microeconomics strand tries to explain banks’ behaviour by looking at it as a profit maximizing entity, setting its loan and deposit rates depending on the level of risk. Ho and Saunders are not satisfied with the hedging hypothesis, because according to them it does not explain clearly why this hedging behaviour occurs. The microeconomics strand did not give sufficient explanation of the difference between loan and deposit rates, and did not include any other

explanatory variable than risk.

The model by Ho and Saunders views the bank as a dealer of deposits and loans. It has a certain cost, resulting from the uncertainty that it faces on the in- and outflow of deposits and loans whilst it has to immediately provide those to any customer that asks for it. The uncertainty of in- and outflows makes that the bank has to hold some kind of buffer, which is costly. The size of this buffer depends on a number of variables which together determine the banks’ possible net income from interest. These are:

• level of risk aversion • market structure • size of transactions • variance of interest rates

It is shown that even at low levels of risk there will always remain some margin between loan and deposit rates, the so-called pure margin. In an empirical analysis it is shown that this pure margin exists and that margins are indeed influenced by variance of interest rates. There is also a number of other variables, not present in the model, which seem significant, namely implicit interest payments through service fees, opportunity costs of holding required reserves with the central bank, and finally a mark-up for default risk on loans. The results of Ho and Saunders (1981) are confirmed using a different dataset in Saunders and Schumacher (2000).

(5)

In a discussion of the model, Lerner (1981) argues that Ho and Saunders miss a very important factor of the banking business, namely that banks have to cover for a cost function of wages, buildings, and having to provide a return to their stockholders. The model where banks are only trading

intermediaries is too simple. Lerner advocates a model of banks being a loan- and deposit producing firm with a cost function, that in addition also has the pure margin resulting from risk and variance in interest rates.

The model has been extended by various authors (for example Angbazo, 1997), mainly to include different types of risk. Maudos and de Guevara (2004) are the first to include operating cost as a factor. Their empirical analysis leads to the following list of determinants of interest income:

• market structure • interest rate risk • credit risk • operating costs • risk aversion of banks • quality of management • implicit interest

• opportunity cost of required reserves

According to a recent strand of literature (Carbó and Rodriguez, 2007)(Lepetit et al., 2008) diversification of the banking sector has changed the determinants of interest income. They note that banks often use some products as a loss leader, meaning that they sell them for a loss just to get the opportunity to sell customers their other products, for example insurance. There is evidence that especially customer loans are used as loss leaders. Also, the relative importance of interest income as a source of revenue is decreasing. In Europe, interest income as a share of total income decreased from 74% in 1989 to 59% in 1998 (Lepetit et al., 2008). Keeping this in mind, the model of banks being just dealers of loans and deposits may be too simple for the current European banking sector. The effects of diversification are important to keep in mind when investigating changes in interest margins.

Meanwhile, the microeconomics strand mentioned earlier arrives at the same determinants of interest margins, although using a different model. As Wong states (2007, p.253): “our model should be viewed as confirmation of rather than confrontation with the existing literature.” For the

remainder of this research the list from Maudos and de Guevara (2004) is used as a guide.

Effect of competition on interest rate pass-through

(6)

The policy rate is the most important instrument of the ECB (and other central banks). Monetary policy is most effective when a change in the policy rate is quickly passed through to retail loan and deposit rates. Therefore the speed of interest pass-through is closely monitored by the ECB and analysts alike. In addition to the speed of pass-through, the level of symmetry is also important. For example, loan rates may adjust quicker than deposit rates. These so-called rigidities can have a number of causes, and the theory presented later in this study makes assumptions on the effect of competition on these rigidities.

A 2010 study by Karagiannis et al. tests pass-through from the policy rate and from the money market rate to retail rates. They find that the central bank rate is passed symmetrically to the retail loan rate. The money market rate is however only passed to depositors when it decreases. Also, changes in the money market rate are passed to loan rates mostly when it increases. Pass-through from central bank rates to deposit rates did not yield significant results. The results from this study confirms that, for the eurozone, the so-called bank-collusive hypothesis on pass-through holds. This hypothesis, from Berger and Hannan (1989), states that banks set their rates exactly like firms with a small amount of market power.

There is a large number of studies confirming that the level of competition affects the speed of pass-through. The ECB itself mentions it as one of the two factors influencing pass-through (the other is credit risk)(ECB, 2009). Van Leuvensteijn et al. (2013) specifically investigates this bank-collusive

hypothesis for the euro area in 1994-2004 period. They find that the spread between money market

rates and retail loan rates is significantly lower in countries with a higher level of competition. Also, the response of long-term loan rates to market rates (thus, the pass-through to loan rates) is quicker in countries with a higher level of competition. This confirms the bank-collusive hypothesis at least for the loan market. However, an interesting result is that in countries with a higher level of loan competition, deposit rates are lower. It seems that banks compensate for their smaller profits in the loan market by decreasing their deposit rates. Competition is not so fierce in the deposit market. Sander and Kleimeier (2004) find, also for the European market, that pass-through speed generally increases and rigidities decrease with the level of competition. In slight contrast with the lack of competition in deposit markets that van Leuvensteijn et al. find, they note that with less competition deposit rates adjust faster downward than upward.

It is important to notice that in all this literature there is not a single deposit rate or loan rate. As Corvoisier and Gropp (2002) note there are marked differences in the responses of for example demand deposits and time deposits. The level of competition is also different for each type of

(7)

product. For the remainder of this research it is assumed that there is one eurozone deposit rate and for each country only one loan rate, expecting that – all rigidities combined – on average they move as one.

(8)

3. Theory and hypotheses

This section proposes a theory of the determinants of changes in interest margins, using the existing literature on the subject. It ends with hypotheses following from the theory.

Having investigated the literature on interest margins there appears to be no research on what determines the changes in these margins. What conditions would lead to quick increases or

decreases? How fast do changes occur? Is there an equilibrium point or are margins forever unstable? Interest income is a result of the level of retail loan rates il , retail deposit rates id and the total

amount of loans and deposits. For any bank, annual interest expense is the amount of deposits times the (average) deposit rate. Annual income is the amount of loans times the (average) loan rate. Changes in net interest income come from these four variables. To correct for changes in the

amount of loans and deposits, this study investigates determinants of the interest margin, defined as

net interest income divided by total earning assets. It is assumed that banks set their retail interest rates to maximize their interest margin, restricted by the factors such as competition and default risk. A change in the ECB’s policy rates affects retail rates, and therefore a change in its policy rates should have an effect on interest margins. The size and direction of this effect depends on the rigidities that are present in interest rate through. The existing evidence on interest rate pass-through gives some clues to the expected direction of the changes in interest margins after the central bank changes its rates. Also, the bank-collusive hypothesis gives an idea of what should happen under different levels of competition. The research question springing from this theory is as follows: What effect does the ECB have on interest margins, and is competition an important factor? As mentioned in the previous section, existing literature on interest income views banks as dealers of deposits and loans. The following analysis does just that, assuming banks interact with

competitors and customers in an oligopolistic environment:

• When the ECB increases its rates, competitive pressure in the market for deposits increases the retail deposit rate id. Any bank that would fail to increase its deposit rate risks losing customers.

• The increase in the ECB rate increases prices in the money market, which lowers profit margins on loans. Banks will want to increase loan rates il, but if they do this faster than their competitors they risk losing customers.

• With a decrease in the ECB rate, banks will feel competitive pressure to decrease their loan rates il. At the same time they may want to decrease deposit rates, but doing so too quickly might entail losing customers.

(9)

Using this framework of competitive rigidities one can make predictions about what should happen after a central banks changes its deposit rate. It brings us to two hypotheses:

• Changes in the ECB deposit rate are negatively linked to changes in interest margins.

• There is a joint effect of rate changes and competition.

The second hypothesis comes from the assumption that all rigidities are weaker under a high level of competition.

Of course there are many other factors that may cause a change in interest margins. The existing literature does not look at changes in interest income, but the determinants of interest income found in other studies are likely to also be of importance in affecting changes. For this reason, they are used as control variables in the model that follows.

(10)

4. Methodology

The following section describes the model that is used to test this paper’s hypothesis. After that it gives a description of the dataset and anticipates a number of problems with this methodology.

The hypotheses are tested using the following model on panel data.

ΔIMit = β0 + β1*LAGIM + β2*(CBt-CBt-1) + β3*ΔHERF + β4*((CBt-CBt-1) *ΔHERF) + β5*ΔGDP + β6*DEFSPREAD + β7*STDEVEURIBOR + β8*GDPLAG + αi + uit

where IM = the interest margin, calculated as interest income/total earning assets;

ΔIM = percentage change in net interest margin;

LAGIM = the first lag of change in the interest margin; CB = ECB deposit rate;

HERF = the Herfindahl competition measure;

ΔHERF = percentage change from previous year in the Herfindahl competition measure; ΔGDP = percentage change from previous year in a country’s gross domestic product;

DEFSPREAD = the credit spread, the difference between yields on government and corporate bonds; STDEVEURIBOR = the volatility of the euribor interest rate;

Uit is the error term and αi are the unobserved effects for each panel.

The CB and HERF variables are of importance to the hypotheses tested in this paper. More specifically, CB and the joint effect should, according to the theory, have a significant value. Other variables are added as control variables, because they are determinants of interest margins according to the existing literature.

Change in GDP is added as a proxy for the business cycle. The credit spread is added as a proxy for default risk, and the volatility of euribor is used to control for the uncertainty of funding rates that banks face (interest rate risk).

With these variables accounted for, there are a number of unobserved effects left. These are

mentioned in the relevant literature as determinants of interest income (for example in Maudos and de Guevara, 2004), therefore they are assumed to be relevant for changes in interest income as well.

• Changes in the level of risk aversion

• Changes in the level of diversification: there is evidence that banks are moving away from interest income as their main source of income

(11)

• Changes in quality of management, implicit interest payments and opportunity cost of required reserves

The unobserved effects can be captured by using a fixed effects panel data model. Due to the use of a lag of the dependent variable, the estimates of this model are biased. To counter this bias an Arellano-Bond estimator is often used, but this is not possible here due to the small size of the sample.

Data description

Countries

The countries investigated are the twelve countries who introduced the euro in 2001 with the exemption of Greece, Portugal and Luxemburg. Data on interest income and default spreads is not available for those three. The remaining nine countries are Germany, Ireland, the Netherlands, Finland, Austria, France, Belgium, Italy and Spain.

Time period

The time variable is measured in years. The lower bound is the introduction of the euro in 1999 and the upper bound is 2009, since more recent data is unavailable. It seems likely that better results could be obtained by using quarterly or even monthly data, but this was not available.

Financial markets have seen an unusual amount of turmoil in the last two years of this period. For external validity and consistency of this research it is useful to question whether the mechanisms behind interest rate pass-through have functioned normally in this period. In a 2009 report, the ECB stated that “the co-movement between retail bank interest rates and market rates has not seemed to

differ markedly from past patterns”. However “bank credit standards have been tightened

signiﬁcantly in recent quarters, countering, to some extent, the smooth pass-through to retail bank lending rates”.

As Karagiannis and et al. 2010 note, the link between the ECB’s policy rate, money market rates and retail interest rates was disturbed from 2007 onwards. Credit spreads widened because banks tightened their rules for providing loans. The disturbance is visible in the figure in appendix A. When looking at the changes in interest income it seems that there was more volatility during the crisis period, but it is undecided whether margins either increased or decreased more than normally. In conclusion, it is hard to say whether changes in interest margins had significantly different causes during the crisis. The use of credit spreads as a control variable should take away at least some of

(12)

the disturbance from this external shock. Due to the size of the sample used in this research the 2000-2009 period is maintained, since dropping the last two years would entail a significant loss of data.

Interest margins

Interest margins are calculated as a percentage, (interest income/total earning assets)*100.1_The data are annual and aggregates for each country. The change is just the difference in percentage points between this year and the previous year. Interest margins in this period were on average 1,49%. Changes are approximately normally distributed with a mean of -0,47% and standard deviation 0,144.

ECB rate

ECB’s deposit rate is used.2_{The variable is the change in value between the start and end of each} calendar year.

Herfindahl competition measure

The Herfindahl index is the sum of the squares of all the credit institutions’ market shares. The ECB publishes these data annually.2_{The use of this index as a proxy for competition is not unquestioned,} for example van Leuvensteijn et al. (2013) make a case for the use of Boone-indicators, and Maudos and de Guevara (2004) use a Lerner index in addition to the Herfindahl index. Using these indicators would however be too complicated for the scope of this research.

There are large differences in the value of this measure between countries. The mean is 944 with a standard deviation of 818. The year-to-year percentage changes however are approximately normally distributed with a mean of +3,1% and a standard deviation of 9,9%.

GDP

Annual changes in GDP are obtained from the IMF.3_{Changes are approximately normally distributed} with a mean of +2,1% and a standard deviation of 2,9%.

Credit spread

The credit spread for each individual country is measured in percentage points4_{. It is the difference} between yields on government bonds and corporate bonds, and is thus a measure for the

probability of default on corporate bonds. This is then used a a proxy for default risk in the entire

1_{Data from OECD (2013)}

2_{From the ECB website (2013)}

3_{On their website (2013)}

4_{From Datastream (2013)}

12

(13)

financial system. It seems likely that with higher default risk banks tighten their criteria for loan provision, thereby changing interest income. The figure in appendix B plots the credit spreads for each country.

Euribor volatility

The volatility of the euribor 3-month interest rate is calculated as the standard deviation of all end-of-day values in a year.4_{The figure in appendix C shows the standard deviation. Most notably,} volatility spiked in 2008.

Possible problems

Is ECB’s deposit rate the right rate?

One might argue that banks don’t necessarily respond to the ECB’s deposit rate. They might also respond to the rate on the marginal lending facility, or money market rates such as EONIA or euribor. For simplicity this study only uses the ECB deposit rate.

Correlation of deposit rate and GDP

As the ECB sets its policy rate in response to macro-economic conditions, a high correlation between the two explanatory variables is to be expected. The table below shows the correlation between a change in the ECB’s deposit rate and (lagged variables of) changes in GDP.

Change deposit rate Change GDP 1st GDP lag 2nd GDP lag 3rd GDP lag

Change deposit rate 1,0000

Change GDP 0,7924 1,0000

1st GDP lag 0,2910 0,5420 1,0000

2nd GDP lag -0,2096 -0,0518 0,4868 1,0000

3rd GDP lag -0,3942 -0,0724 0,3348 0,5757 1,0000

Table: correlation of policy rate and changes in GDP

This high correlation suggests that changes in the ECB’s deposit rate are already a fairly good proxy for the business cycle. Including both might be redundant and give a biased estimate of the

coefficients.

Diversification

One unobserved variable, diversification, may be correlated with the competition measure. It can be argued that due to increasing levels of competition interest margins have deteriorated, which has led banks to try and find other sources of income. Diversification is however hard to observe. In the dataset, the Herfindahl index and interest income have a correlation of -0,58. This suggests that there is indeed a negative link between competition and interest income. In addition, Carbo and Valverde (2007) find evidence for a link between diversification/specialization and bank margins. The

(14)

unobserved effect of diversification may be of importance and disturb the results of this paper’s analysis. In particular, the estimated coefficient of competition may be biased.

Other omitted variables

There are a number of other unobserved variables that might cloud the results. Some of them were dropped from the model because no significant effect could be found in the dataset, this is further discussed in the results section.

Cross-country effects and country-aggregate problems

The dataset consists only of data on country-level. This makes the use of, for example, the interest margin, difficult. If one institution in a country has a drastic change in interest income then it affects the aggregate. This disturbs all the coefficients that the model estimates. A solution to this issue would be to use bank-level data.

Another problem arises from the fact that banks are often multinational companies. They get income, risk exposure and costs not just from their home country, but probably also from other European and non-European countries. Once again this makes it difficult to estimate the effects of single country-level data on single country-level income. A model that uses only a pan-European dataset might yield better results, although there would be a much smaller number of data points available.

Size of the sample

The model is tested with only a small amount of data, therefore the results must be interpreted with caution. Some variables that ideally should have been in the model could not be fit in the sample. With nine countries, ten years, and a lagged dependent variable there are only 80 data points.

(15)

5. Results

This section describes the results from the model that was explained in the previous section. The following table lists the regression results:

Fixed effects panel regression on change in interest margins R-squared: 0,3148

Observations: 80

Variable Coefficient Standard deviation P-value

Lag interest margin 0,0927 0,1262 0,466

Change GDP 0,0285 0,0119 0,019*

1st_{lag GDP} _-0,0221 _0,0155 _0,160

Change deposit rate -0,0581 0,0264 0,031*

Change Herfindahl

index 0,0028 0,0017 0,118

Change Herfindahl x change deposit rate (crossterm)

-0,0002 0,0018 0,988

Credit spread 0,0234 0,0137 0,092

Euribor volatility 0,2747 0,0852 0,002**

Intercept -0,1631 0,0317 0,000**

Table: Regression results

* = significant at 5% level ** = significant at 1% level

Interpretation of coefficients

The lag of the interest margin is far from significant. Omission of this variable, however, deteriorates the fit of the model, and it makes more economic sense to keep it.

GDP growth is positively correlated with interest margins. This indicates that in times of high economic growth, margins increase. It was noted in the previous section that GDP is highly

correlated with changes in the ECB’s rate. So it seems like banks can somehow exploit deposit rate increases to improve their margins. This is consistent with the finding mentioned in section 2, that deposit rates are stickier than loan rates. In a world where deposit rates are perfectly sticky and loan rates are perfectly flexible, an increase in the deposit rate means just more interest revenue while cost stays the same.

(16)

The coefficient of change in the deposit rate is significant, thus the model does not reject the first hypothesis.

Change of the Herfindahl index is not significant. It is expected that this result is biased by diversification of the banking sector (see discussion in section 4).

The joint effect of the deposit rate and the Herfindahl index is far from significant, therefore the second hypothesis is rejected. Part of this may be explained by multicollinearity with the change in the deposit rate.

The credit spread is positively correlated with margins (although not significantly). This is to be expected: increased default risk will lead banks to tighten lending conditions, asking a higher price for the same service. It is safe to assume that retail deposit rates are not affected by this risk factor. As a result, interest income increases.

Volatility of euribor, the proxy for interest rate risk, has the expected positive correlation with interest margins. This follows Ho and Saunders’s 1989 model, where the pure margin depends mainly on the volatility of market rates.

Further interpretation

The figure in Appendix D plots the residual values of the model. The sudden decline in the final period seems to indicate that there is a shock from a variable which is not in the model. The residuals have a correlation of 0,56 with the dependent variable.

Different specifications

The model does not include all variables that could theoretically affect changes in interest margins. Some variables were dropped from the final model because they did not significantly improve it. Further lagged variables of changes in GDP and deposit rates should theoretically have some effect, but this could not be proven. This is possibly due to the size of the sample. Also, changes in expense ratios in the industry should, according to the improved Ho and Saunders model with operating cost, have some effect. This could not be verified either.

(17)

6. Conclusion and discussion

The following section evaluates the hypotheses and the obtained results and gives some suggestions for further research.

This research has tried to answer questions about what effect central banks have on interest margins, and about the effects of competition. The contribution lies in the analysis of interest income, resulting from existing literature on interest pass-through and competition. A panel data regression on European bank data in the 2000-2009 period was used to estimate the coefficients of variables that may affect changes in interest margins. The resulting model fits the data reasonably well. The first hypothesis, stating that ECB policy has an effect on interest margins, is not rejected. The second hypothesis – predicting a joint effect of ECB policy and competition is rejected in this dataset. From this model and dataset, the main determinants of changes in interest margins appear to be: the business cycle; default spreads; and volatility of money-market interest rates. These variables were used as control variables following earlier research on interest margins.

Results from this regression should be treated cautiously because of the size of the sample. Also, the sample included the start of a major global financial crisis, which may have clouded the results. A number of other problems have already been discussed in the methodology section.

Although this paper rejects the hypothesis on the joint effect of central bank policy and competition, they might very well be confirmed with a larger dataset. Further research should be conducted with bank-level rather than country-level data and should have a time variable that is smaller than one year. Furthermore, it would be useful to look at multiple types of interest income (short-term, long-term) to see in which areas banks feel the most competitive pressure. Preferably further research would also include data from non-European countries, and take diversification of the banking sector into account. Finally, the literature suggests that the Herfindahl index is not an ideal measure of competition, another option would be to estimate Lerner or Boone indices.

(18)

7. References

Angbazo, L. (1997). Commercial bank net interest margins, default risk, interest-rate risk, and off-balance sheet banking. Journal of Banking & Finance, 21(1), 55-87. doi:10.1016/S0378-4266(96)00025-8

Berger, A. N., & Hannan, T. H. (1989). The price-concentration relationship in banking. The Review

of Economics and Statistics, 71(2), 291-299.

Carbó Valverde, S., & Rodríguez Fernández, F. (2007). The determinants of bank margins in european banking. Journal of Banking & Finance, 31(7), 2043-2063.

doi:10.1016/j.jbankfin.2006.06.017

Corvoisier, S., & Gropp, R. (2002). Bank concentration and retail interest rates. Journal of Banking

& Finance, 26(11), 2155-2189. doi:10.1016/S0378-4266(02)00204-2

Datastream. (2013). Credit spreads, euribor

Recent developments in the retail bank bank interest rate pass-through in the euro area, (2009). ECB. (2010). ECB banking structures. Retrieved 7/6, 2013, from

http://www.ecb.int/pub/pdf/other/eubankingstructures201009en.pdf ECB. (2013). Key ECB interest rates. Retrieved 7/6, 2013, from

https://www.ecb.int/stats/monetary/rates/html/index.en.html

Haughton, A. Y., & Iglesias, E. M. (2012). Interest rate volatility, asymmetric interest rate pass through and the monetary transmission mechanism in the caribbean compared to US and asia. Economic Modelling, 29(6), 2071-2089. doi:10.1016/j.econmod.2012.06.034 Ho, T. S. Y., & Saunders, A. (1981). The determinants of bank interest margins: Theory and

empirical evidence. The Journal of Financial and Quantitative Analysis, 16(4, Proceedings of 16th Annual Conference of the Western Finance Association, June 18-20, 1981, Jackson Hole, Wyoming), 581-600.

IMF. (2013). World economic outlook database 2013. Retrieved 7/6, 2013, from

http://www.imf.org/external/pubs/ft/weo/2013/01/weodata/index.aspx

Karagiannis, S., Panagopoulos, Y., & Vlamis, P. (2010). Interest rate pass-through in europe and the US: Monetary policy after the financial crisis. Journal of Policy Modeling, 32(3), 323-338. doi:10.1016/j.jpolmod.2010.02.006

Lepetit, L., Nys, E., Rous, P., & Tarazi, A. (2008). The expansion of services in european banking: Implications for loan pricing and interest margins. Journal of Banking & Finance, 32(11), 2325-2335. doi:10.1016/j.jbankfin.2007.09.025

Maudos, J., & Fernández de Guevara, J. (2004). Factors explaining the interest margin in the banking sectors of the european union. Journal of Banking & Finance, 28(9), 2259-2281. doi:10.1016/j.jbankfin.2003.09.004

Maudos, J., & Solís, L. (2009). The determinants of net interest income in the mexican banking system: An integrated model. Journal of Banking & Finance, 33(10), 1920-1931.

doi:10.1016/j.jbankfin.2009.04.012

Memmel, C. (2011). Banks' interest rate risk: The net interest income perspective versus the market value perspective. Quantitative Finance, , 1-10. doi:10.1080/14697688.2011.630326

OECD. (2013). Statextracts. Retrieved 7/6, 2013, from http://stats.oecd.org/

(19)

Sander, H., & Kleimeier, S. (2004). Convergence in euro-zone retail banking? what interest rate pass-through tells us about monetary policy transmission, competition and integration.

Journal of International Money and Finance, 23(3), 461-492.

doi:10.1016/j.jimonfin.2004.02.001

Saunders, A., & Schumacher, L. (2000). The determinants of bank interest rate margins: An international study. Journal of International Money and Finance, 19(6), 813-832. doi:10.1016/S0261-5606(00)00033-4

van Leuvensteijn, M., Sørensen, C. K., Bikker, J. A., & van Rixtel, A. A. R. J. M. (2013). Impact of bank competition on the interest rate pass-through in the euro area. Applied Economics,

45(11), 1359-1380. doi:10.1080/00036846.2011.617697

Wong, K. P. (1997). On the determinants of bank interest margins under credit and interest rate risks. Journal of Banking & Finance, 21(2), 251-271. doi:10.1016/S0378-4266(96)00037-4

(20)

Appendix A

Figure: developments in eurozone interest rates (Karagiannis et al., 2010)

Appendix B

Figure: credit spreads

-6 -4 -2 0 2 d ef s pr e ad 2000 2002 2004 2006 2008 2010 year aus bel fin fra ger ire ita lux net spa 20

(21)

Appendix C

Figure: volatility of 3-month euribor

Appendix D

Figure: residuals of regression

-.2 -.1 0 .1 .2 Li ne ar pr e di c ti on 2000 2002 2004 2006 2008 2010 year aus bel fin fra ger ire ita net spa 21

Influence of central bank policy and competition on interest margins