The relation between banking sector competition and the degree of pass-through of the term premium decreasing effect of quantitative easing to long-term
commercial interest rates in the Euro area.
BSc Thesis Economics Melvin Vooren
Student number: 10190481
Table of contents
1. Introduction ... 3
2. Literature review ... 5
2.1 Unconventional monetary policy ... 6
2.2 Transmission mechanism ... 8
2.3 A new measure of competition ... 10
3. Methodology ... 12
4. Results and analysis ... 16
4.1 Relation with the Boone indicator ... 18
4.2 Case studies ... 19 4.2.1 Ireland ... 19 4.2.2 Greece ... 20 4.2.3 Spain ... 21 4.2.4 France ... 22 4.2.5 Italy ... 23 4.2.6 Austria ... 23 4.2.7 Portugal ... 24 5. Conclusion ... 24 Reference list ... 26 Appendix ... 28
1. Introduction
With the short-term nominal interest rate at its zero lower bound, policymakers have to consider different policy measures to stimulate the economy during a recession. Quantitative easing is one example of an unconventional tool to stimulate demand when the interest rates are at their zero lower bound. It has first been applied by the Bank of Japan to combat the country’s deflation following the collapse of the real estate bubble in the 1990’s. More recently it has been employed by various central banks to deal with the recession accompanied by the global financial crisis following the bankruptcy of Lehman Brothers in 2008. The European central bank has also utilised quantitative easing, particularly to deal with the macroeconomic effects of the European sovereign debt crisis (Joyce et al. 2012).
Quantitative easing operates mainly by reducing the term premium on long-dated assets. An important long-term interest rate is the variable mortgage rate charged by commercial banks. A decrease of this interest rate could eventually transmit into consumption, should households consume part of their savings on their interest payments. But this is only the case if commercial banks adequately transmit the term premium decreasing effect of quantitative easing into a reduction of their variable mortgage rates. Banking sector competition is an important driver of this kind of pass-through. Literature by for example Van Leuvensteijn et al. (2013) have analysed the impact of bank competition on the interest rate pass-through in the Euro area.
The literature regarding interest-rate pass and banking sector competition is quite diverse. Yet, no literature has linked the pass-through of quantitative easing with banking sector competition. This thesis will attempt to estimate the relation between the pass-through of the term premium decreasing effect of quantitative easing and the
degree of banking sector competition. The following research question will be
answered: “What is the relation between the degree of banking sector competition and the amount of pass-through of the term premium decreasing effect of quantitative easing to long-term commercial interest rates in the euro area?”
To answer this question, the amount of pass-through will be estimated for several European countries, regarding three long-term commercial interest rates. The countries that will be analysed are Austria, France, Greece, Ireland, Italy, Portugal and Spain. The long term commercial interest rates that will be examined are the variable mortgage rate, the interest rate charged by commercial banks on corporate debt smaller than one million euros and the interest rate charged by commercial banks on corporate debt larger than one million euros. After estimating the degree of pass-through to each of these interest rates for these seven countries, the results will be compared with a measure of competition, to see whether there is a relation between the two variables.
The remainder of this thesis is structured as follows. The following part is a literate review in which both the existing literature, as well as the theoretical background on the subject will be discussed. In section three, the empirical models that have been estimated will be described. Section four will discuss the results as well as the analysis, in order to arrive at the conclusion in section five.
2. Literature review
Before the bankruptcy of Lehman Brothers in 2008 initiated a global financial crisis, traditional monetary policy seemed secure and straightforward. The short-term interest rate was used as an instrument to achieve low and stable inflation (Joyce et al 2012). The interest rate was determined by the Taylor rule (Taylor 1993). The Taylor rule can be described as an interest rate feedback rule in which the optimal rate is determined by the output gap and the inflation gap, that is the actual rate of inflation minus the target inflation. Usually, a weight of three-thirds is given to the inflation gap, and a weight of a half is given to the output gap (Woodford 2001).
Before the crisis, policymakers had mainly two conventional tools at their disposal to influence the interest rate. Standing facilities give banks the possibility to either lend or deposit liquidity at a pre-announced rate. The European Central Bank (ECB) offers two overnight standing facilities: the Marginal Lending Facility and the Deposit Facility. These facilities allow a broad amount of banks to infinitely deposit or lend liquidity, as long as they post sufficient eligible collateral for the Marginal Lending Facility. These standing facilities create a corridor for the overnight interest rate. The Deposit Facility functions as the lower bound for the overnight rate, while the Marginal Lending Facility confines the upper bound (Lenza et al. 2010).
Conducting open market operations is another method to influence the interest rate. To lower the interest rate, the central bank can inject liquidity into the economy by purchasing financial assets, or through repurchase agreements. Repurchase agreements involve the central bank purchasing a financial asset, while agreeing to sell in back in the future. The central bank can withdraw liquidity from the market to raise the interest rate. To do so, the central bank can either sell financial assets, or make reverse repurchase agreements. In a reverse repurchase agreement, the central
bank sells a financial asset, while guaranteeing to buy it back in the future (Lenza et al. 2010).
According to Joyce et al. (2012) there are several issues with the ability of conventional monetary policy to stimulate the economy after financial crises. The first concern is the zero lower bound on nominal interest rates. In many European
countries, the Taylor rule would suggest negative nominal interest rates to deal with the output gaps brought on by the preceding financial crises. Negative nominal interest rates are impossible due to the fact that economic agents can always hold cash, which does not bear interest. Therefore, other forms of monetary policy need to be considered.
2.1 Unconventional monetary policy
There are many forms of unconventional monetary policy, but the most important one is Quantitative Easing (QE). This form of unconventional monetary policy was first introduced by the Bank of Japan to deal with the country’s deflation following the collapse of the real estate bubble in the 1990’s. While the Japanese interest rates were already at their zero lower bound, the Bank of Japan endeavoured at purchasing government bonds from the Japanese banking sector. The intention was to expand the amount of cash reserves the banking sector held. The ambition was that by targeting a high enough level of reserves, this would transmit into lending into the nation’s economy. This would ultimately drive asset prices up and heave Japan’s economy out of the deflationary spiral1 (Joyce et al. 2012, p. F274).
Other central banks have followed the Bank of Japan by implementing QE as a response to the crisis, although each in a slightly different manner. The Federal Reserve (Fed) has bought US Treasury Bills, but large quantities of mortgage backed
securities as well. The Bank of England (BoE) has intensely bought UK government bonds (gilts). In reality, the differences between the assets bought by the Fed and the BoE are not that big, because the mortgage backed securities bought by the Fed are backed by US government agencies (Joyce et al. 2012, p. F274).
The ECB has implemented QE via Target balances. The term TARGET is an acronym standing for Trans-European Automated Real-Time Gross Settlement Express Transfer. Target balances measure the total surplus or deficit of each Eurozone national central bank (NCB) with the ECB. This can basically be
understood as the current account balance with all other Eurozone countries (Sinn & Wollmershäuser 2012, 473-473).
The euro crisis caused strong monetary imbalances within the euro-area banking sector. Many banks, particularly those in Greece, Ireland, Italy, Portugal and Spain (GIIPS), have suffered serious and stable capital outflows, mainly towards Dutch and German banks. These capital outflows resulted in Target deficits in the GIIPS
countries, and Target surpluses in the other countries (Sinn & Wollmershäuser 2012).
1 Japan has suffered from prolonged deflation since the start of the 2000’s. The situation did
not improve much after embracing QE in March 2001. After committing to expand the asset purchase programme by increasing the monetary base at an annual pace of 60 to 70 trillion yen in April 2013, inflation started to increase. A year after this statement, inflation started to increase sharply, peaking at twice the inflation target of two per cent per year.
Source: Bank of Japan, statement on monetary policy, dated 4 April 2013. Inflation based on the Japanese Corporate Goods Price Index (CGPI).
Figure 1: Target balances in the Eurozone (end of 2011) Source: Sinn & Wollmershäuser (2012)
The ECB supported extensive money creation by the NCB’s of the GIIPS countries to finance these deficits. This went at the expense of money creation in the countries in the core of the Eurozone. Instead of printing and lending it, the ECB encouraged NCB’s of the core countries to borrow and remove euro currency from circulation. Bypassing the European Parliament, this ‘reprinting’ of money can be seen as a rescue program that aided the crisis countries in the same way the official euro rescue facilities the European Financial Stability Facility (EFSF) and the European Financial Stability Mechanism (EFSM) did, but much earlier (Sinn & Wollmershäuser 2012, 469).
2.2 Transmission mechanism
Joyce (2012) describes two main channels through which asset purchases by a central bank can affect the economy. The portfolio substitution channel is a channel that always operates. When investors sell government bonds to the central bank, they exchange these long-term assets for bank deposits, which are short-term assets. Many holders of government bonds, like pension funds and insurance companies, have long-dated liabilities. They prefer to match these liabilities with long-long-dated assets. Hence,
they are very likely to invest the proceedings of their bond sales in other long-term assets such as corporate bonds, to rebalance their portfolio. This initiates a reduction of the term premium on other long-term assets, making it very likely that the prices of these long-term assets will increase, including corporate bonds and equities. As a result, government and corporate bond yields will decline. This will make it easier to raise funds for many firms. These firms will subsequently generate more capital for households, which are their ultimate owners. When these households consume part of this, this will eventually increase demand, and thereby boost GDP.
The other channel through which QE affects the economy is the bank funding channel. This channel might improve the availability of bank credit, but will only operate if the availability of funds to individual banks is under threat. If banks are concerned about their ability to refinance themselves, they are less likely to grant loans. As described in the previous section, bank’s deposits as well as reserve
balances by the central bank increase as a result of government bond purchases by the central bank. This will make the bank more likely to expand lending (Joyce et al. 2012, F281).
An important factor that determines the effectiveness of QE is the degree of interest rate pass-through by commercial banks. As described earlier, QE reduces the return on government and corporate bonds through a reduction of the term premium. Myron & et al. (1983) concluded that commercial loan rates are primarily determined by market interest rates. Therefore, it is likely that commercial loan rates will
decrease as a result of QE.
However, this might not be as indisputable as is seems. Bank market structure is an important factor that influences a bank’s price-setting behaviour. An uncompetitive banking market might lead to high commercial interest rates. One may hypothesise
that an uncompetitive banking market might obstruct the effectiveness of QE, arguing that with a lack of competition, banks are not incentivised to decrease their
commercial lending rates, neglecting to transmit the low market rates into low commercial lending rates (Kopecky et al. 2012).
Kopecky et al. (2012) have designed a dynamic model of bank adjustment costs to examine pass-through of market rates into bank retail deposit and loan rates. They confirmed that an imperfectly competitive banking market affects the degree of market rate pass-through by commercial banks. When the degree of competition in the banking sector decreases, the impact of lagged and expected rates on current retail rates increases, while the effect of current security rates decreases. This impedes pass-through.
2.3 A new measure of competition
To incorporate the degree of competition in the analysis, an accurate measure of competition is essential. One commonly used measure of competition is the price cost margin (PCM), which is defined as the output price minus the marginal costs divided by the output price. The PCM is commonly used as an empirical approximation of the theoretical Lerner index. However, there are issues with the theoretical foundations of this measure. In papers by Bulow and Klemperer (1999), Rosentahl (1980) and Stiglitz (1979), models are presented in which higher levels of competition may lead to an increase in the PCM. In spite of these contradictory results, the PCM still is a popular measure of competition. Boone (2008) believes that there are two reasons for this. First, it is still unknown how often these contradictory situations occur in
practice. A second, more technical reason for the popularity of PCM is the wide availability of the data required to estimate the PCM.
A novel, more robust measure of competition has been introduced by Boone (2008). In his paper, he introduces the Relative Profit Differences (RPD) as a measure of competition. This measure of competition is further referred to as the Boone
indicator. The central intuition behind the Boone indicator is that when the level of competition in an industry rises, efficient firms gain in terms of relative profits or market shares. This effect increases in the amount of competition. A technical advantage of using the Boone indicator is that there are no additional data
requirements in order to make a sound estimate, relative to the price cost margin. The first application of the Boone indicator to the banking industry has been carried out by Van Leuvensteijn et al. (2011). The author states that direct
measurement of competition in the banking industry is often problematic, because prices and costs of single banking products are not widely available. Therefore, indirect measurement of competition is required. One way to do this is through the Herfindahl-Hirschmann Index (HHI). The HHI is a measure of market concentration, defined by the sum of squared market shares. A disadvantage of this method is that the HHI does not distinguish between small and large countries. Also, concentration may increase after forced consolidation due to heavy competition. Therefore, the HHI is an inappropriate measure.
Given these practical limitations, Van Leuvensteijn et al. (2011) utilise the Boone indicator as a measure of competition the banking sector. In their paper, Van Leuvensteijn et al. estimate the following model:
In which ln !! represents the natural logarithm of the market share of bank i. The market share of bank i is defined as its total revenue divided by the total revenue of the entire market:
!! = !!!! !!!!
!
ln !"! represents the natural logarithm of the bank’s marginal costs. Boone et al. (2004) take the ratio of a firm’s average costs and revenues as an approximation of its marginal costs. The estimated coefficient for ! will be referred to as the Boone indicator. The model has been specified in logarithmic terms to deal with
heteroscedasticity. Furthermore, the specification implies that the Boone indicator is an elasticity, which simplifies the interpretation. An increase in the Boone indicator is interpreted as a reduction of the competitive nature of the market.
3. Methodology
The aim of this thesis is to analyse the pass-through of quantitative easing to long-term commercial interest rates, and to relate this with the degree of competition in the banking sector. As described in the literature review, quantitative easing operates through a reduction of the term premium on long-dated assets. This would suggest that quantitative easing (QE) initiates a reduction of long-term interest rates.
Christensen & Rudebusch (2012) have confirmed this. They have used dynamic term structure models to decompose US and UK zero-coupon government bond yields into expectations about the future short-term interest rates and term premiums. They have found that declines in the US long-term bond yields can mainly be attributed to the fact that the expectations about future short-term interest rates scaled down. In the UK however, declines in the long-term bond yields mainly reflect reduced term
premiums. According to the authors, these differences are due to the differences in policy communication. In contrast to the Bank of England, the Federal Reserve was more willing to provide monetary policy forward guidance at the zero lower bound.
An important long-term interest rate is the mortgage rate charged by
commercial banks. If the term premium decreasing effect of QE is passed through to this commercial interest rate, the reduction of the interest payments by households can transmit into consumption. This effect can also stimulate credit-constrained
households to take out a mortgage to buy a new-built home or to renovate their existing homes. Altogether these processes boost the economy and stimulate recovery.
The following model has been specified to estimate the degree of pass-through of quantitative to the mortgage rates of seven European countries:
!"#$%&%!_!"#$!" = !!+ !!!"#$%_!"#$! + !!!"#_!"#$_!"#$%!"+ !! (1)
The countries that have been analysed are Austria, France, Greece, Ireland, Italy, Spain and Portugal. As described in the literature review, these countries received aid in the form of extensive money creation during the crisis period. This is the reason why these countries have been selected. The time period that has been researched is January 2003 until March 2014. Using monthly data, this time period yields 137 observations. !"#$%&%'_!"#$!" is the average mortgage rate in country x in
period i. These mortgage rates have been obtained from the European Central Bank’s website (Euro area and national MFI interest rates, 1.2.1.5). !"#$%_!"#$! is the European Central Bank’s main overnight refinancing rate in period i. These rates are
freely available on the ECB’s website (Key ECB interest rates, main refinancing operations).
To incorporate quantitative easing in the analysis, an appropriate proxy is required. As described in the literature review, quantitative easing comes conjointly with a reduction of government bond yields. Christensen and Rudebusch (2012) have also confirmed this using a dynamic term structure model. Therefore, 10-year
government bond yields have been added to the regression as a proxy for quantitative easing. !"#_!"#$_!"#$%!" represents the yield in period i on a 10-year
government bond issued by the government of country x. Finally, !! represents an
econometric error.
Another important interest rate is the rate charged by commercial banks to non-financial corporations. As described in the literature review, a decreasing interest rate on corporate debt makes it easier for firms to raise funds. This increases their profits, generating capital for their owners. The ultimate owners of these firms are
households. If these households invest or consume part of this capital, this will increase demand, which will ultimately boost GDP.
Because of the importance of these interest rates, two additional regressions have been run, adopting the same explanatory variables as the first regression. The first model has been specified to analyse the pass-through of quantitative easing to relatively small corporate loans:
!"#$_!"#$_!!" = !!+ !!!"#$%_!"#$! + !!!"#_!"#$_!"#$%!" + !! (2)
Where !"#$_!"#$_!!" represents the average interest rate charged by commercial
country x in period i. These rates have been gathered from the same source as the average mortgage rates (Euro area and national MFI interest rates, 1.2.2.8). !! represents another econometric error. To analyse the pass-through of quantitative easing to relatively large corporate loans, the following model has been specified:
!"#$_!"#$_!!" = !! + !!!"#$%_!"#$! + !!!"#_!"#$_!"#$%!" + !! (3)
Where !"#$_!"#$_!!" portrays the average interest rate charged by commercial banks to non-financial corporations for loans larger than one million euros in country x in period i. These rates have also been collected from the European Central Bank’s website (Euro area and national MFI interest rates, 1.2.2.9). !! represents another econometric error.
After running these three regressions, the estimated coefficients !!, !! and !! will be compared with the average Boone indicator for each country. Each country will have another estimate for !!, !! and !!, which will be interpreted as the degree of pass-through of quantitative easing to variable mortgage rates, corporate loans to financial institutions smaller than one million euros and corporate loans to non-financial institutions larger than one million euros in this country, respectively.
The average Boone indicator has been calculated using yearly data from the Federal Reserve Bank of St. Louis from 2003 to 2011. A monthly Boone indicator is not available. The average Boone indicator has been taken because it would be troublesome to make a yearly estimate for !!, !! and !! with monthly data. This would yield only twelve observations, making it likely that the estimated coefficients will be insignificant. Still, it may be possible with weekly data. Yet, weekly data is unavailable at this time.
4. Results and Analysis
Before estimating the models, White’s test for heteroskedasticity has been run for each model and each individual time series. The results of this set of tests show that, in most of the cases, the null hypothesis was rejected. Therefore, heteroskedasticity-consistent standard errors have been used in the remainder of this thesis.
The estimation of the first model, regarding the pass-through of quantitative easing to mortgage rates, has yielded the following results. For Austria, !! has been estimated at 0.37. For Italy, France and Spain the coefficient has been estimated at 0.25, 0.22 and 0.21, respectively. For Portugal, the estimated coefficient for !! equals
0.16. All of these estimates are significant at a one per cent significance level. The estimated coefficient for Greece is 0.01, at a five per cent significance level. No significant estimate has been found for Ireland. R-squared ranges from 0.72 to 0.91.
The second model has yielded the following estimates for !!. For Spain, Italy, Portugal and Greece and Ireland, !! has been estimated at respectively 0.51, 0.40,
0.23, 0.12 and 0.06, at a significance level of one per cent. For Austria, the coefficient has been estimated at 0.08, although at a five per cent significance level. A
remarkable estimate has showed up for France: -0.16 at a significance level of one per cent. R-squared ranges from 0.41 to 0.95.
The final model, regarding the pass-through of quantitative easing to corporate bank loans larger than one million euros, has generated the following estimates for !!.
For Italy, Portugal, Spain and Greece, !! had been estimated at respectively 0.32, 0.27, 0.23 and 0.13, at a significance level of one per cent. No significant estimates were found for Austria and Ireland. Furthermore, a negative estimate for !! has turned up for France: it has been estimated at -0.20 at a significance level of one per cent.
R-squared ranges from 0.46 to 0.96. A detailed table containing all estimates,
standard errors and values for R-squared can be found in the appendix (Table 2, A-C). A time series of the variable mortgage rate has been plotted in graph 1. The interest rate on corporate loans smaller than one million euros has been plotted in graph 2. The same has been done in graph 3 for the interest rate on corporate loans larger than one million euros. These graphs can be found in the appendix.
It is interesting to see that the spread on corporate loans widened significantly after the crisis when compared with the spread on variable mortgage rates, especially in the GIIPS countries. An increase in the risk might be a possible explanation for this. Many businesses got into trouble after the crisis, increasing risk involved in corporate loans.
Small corporate loans have a bigger spread, because these loans are mainly provided to small and medium enterprises (SME’s). Therefore, these loans are usually more risky than large corporate loans, which are usually provided to large
corporations. Mortgages, on the other hand, have a collateral. This dampens the effect of the crisis on the variable mortgage rate spread, because banks can always get (part of) their money back once a customer goes into default, lowering the risk involved.
4.1 Relation with the Boone indicator
After running these regressions, the average Boone indicator has been calculated for each country using data from the Federal Reserve Bank of St. Louis from 2003 to 2011.
Table 1
Country( Indicator((2003(–(2011)(Average(Boone( !!( !!( !!(
Ireland( !0.009% 0a% 0.06**% 0a% Greece( !0.065% 0.01*% 0.12**% 0.13**% Spain( 0.101% 0.21**% 0.51**% 0.23**% France( !0.045% 0.22**% !0.16**% !0.20**% Italy( !0.040% 0.25**% 0.40**% 0.32**% Austria( !0.058% 0.37**% 0.08*% 0a% Portugal( !0.085% 0.16**% 0.23**% 0.27**%
a : not significantly different from zero * : p ≤ 0.05
** : p ≤ 0.01
To display the relation between the estimates of !!,!!!, !! and the average Boone indicator, the data from the preceding table has been plotted in a scatterplot. In scatterplot 1, the estimates for !! have been plotted against the average Boone
indicator for each country. The estimates !! have been plotted against the average Boone indicator in scatterplot 2. The same thing regarding !! has been done in scatterplot 3. These plots can be found in the appendix.
Scatterplot 1 reveals one outlier (Spain). With exception of this outlier, a slightly negative correlation between the estimates for !! and the average Boone indicator can be seen. Disregarding the outlier, the correlation coefficient is -0.25. When performing Ordinary Least Squares to estimate the effect of the average Boone indicator on the estimated !!, a strongly negative (-1.4) coefficient is found, although not significant. This may be because of small sample size, because only six
Scatterplot 2 reveals one outlier as well. It is the same country as in the first scatterplot, and the only country with a positive average Boone indicator.
Disregarding this outlier, a slightly negative correlation can be seen again. The
correlation coefficient is -0.19, leaving out the outlier. Ordinary Least Squares reveals a strongly negative (-1.38) but insignificant effect of the average Boone indicator on !!.
Due to the positive average Boone indicator of Spain, scatterplot 3 shows more or less the same outlier as the preceding scatterplots. Ignoring this outlier, the
correlation coefficient measures -0.36. Ordinary Least Squares reveals a strongly negative (-2.7) estimate for the effect of the average Boone indicator on !!, yet again insignificant.
4.2 Case studies
Given the average values of the Boone indicator and the estimates for !!,!!! and !!, an evaluation of these results can be made. In the following section, a brief case study of each country will be discussed, in which the results will be elaborated.
4.2.1 Ireland
No significant effect has been found for the effect of government bond yields on variable mortgage rates in Ireland. The same applies to the effect of government bond yields on corporate loans larger than 1 million euros. A significant effect of
government bond yields on small corporate loans has been found. Yet, the estimated coefficient is only 0.06.
Ireland was struck hard by the financial crisis. In January 2009, the Irish government nationalised Anglo Irish Bank after it was concluded that a government
recapitalisation scheme would not be enough to prevent the bank from failing. In addition to that, the two largest banks in Ireland both received 3.5 billion euros of government aid (Dieckmann & Plank 2012, p. 920). In November 2010, the government had to acquire an 85 billion euro bailout from the European Union, International Monetary Fund and the European Financial Stability Facility (Martin & Milas 2012). By the end of 2011, Ireland had a negative Target balance of 120 billion euros (see figure 1).
This confirms that Ireland has received a lot of aid during the crisis. Therefore it is surprising that the estimates for !!and !! are not significantly different from zero, and that!!! is only estimated at 0.06. The average Boone indicator of -0.009 partially explains these findings, as the Irish banking sector is among the least competitive in the sample of countries studied in this thesis. An uncompetitive banking sector hinders pass-through of the term premium lowering effect of quantitative easing, as already described in the literature review.
4.2.2 Greece
The estimated pass-through of the term premium lowering effect of quantitative easing to Greek mortgage rates is fairly minimal, only 0.01. In contrast, the estimated pass-through of the term premium lowering effect of quantitative easing to Greek corporate loans is much higher. For small corporate loans it has been estimated at 0.12, and correspondently at 0.13 for large corporate loans.
With the introduction of the euro, government bond spreads in the Eurozone started to converge. Yet, following the beginning of the financial crisis in August 2007, European government bond spreads started to diverge again to their levels in the period 1999-2001. Until the second half of 2008, when the global financial crisis
escalated, yield spreads started to increase again, reaching records heights of over 300 basis points (Bernoth et al. 2012).
Greece was severely affected by the increase of its government bond spread, mainly caused by financial distress due to high debt-to-GDP levels and large budget deficits. The recession following the increase in these spreads lead to several
government bailouts, on the condition of austerity measures. The first bailout
occurred in May 2010. Still, the situation did not improve. In July 2011, the European Union agreed upon a new support package. The total scale of these support packages is close 240 billion euros (Martin & Milas 2012). According to figure 1, Greece had a Target deficit of 104 billion euros by the end of 2011.
The degree of banking sector competition is fairly decent according to the average Boone indicator, relative to the other countries studies in this thesis.
Therefore, it is remarkable that the pass-through of the term-premium lowering effect of quantitative easing to Greek variable mortgage rates has been estimated at only 0.01. A reason for this might be the relative amount of non-performing loans (NPL’s) in the Greek banking sector. The percentage of NPL’s increased sharply after the Greek economic crisis. This might also be the reason why the estimates for corporate loans are so small, relative to the average Boone indicator (Louzis et al. 2012).
4.2.3 Spain
Spain is among the countries with the highest degree of pass-through. The degree of pass-through to corporate loans smaller than 1 million euros has been estimated at 0.51. This is interesting, because the average Boone indicator in the period 2003 – 2011 is 0.101, the highest among the sample of countries investigated in this thesis. This is remarkable, because Van Leuvensteijn et al., who have estimated the Boone
indicator for the time period 1993 – 2004, have concluded that the Spanish banking sector is among the most competitive of the European Union.
Although, according to the results of Van Leuvensteijn et al. (2013), the degree of competition deteriorated slightly in the second half of the time period they have investigated. They also mention that the Spanish banking sector has made a lot of progress since extensive liberalisation reforms in the late 1980’s and early 1990’s. This may still be driving the pass-through nowadays. However, there might possibly also be an error in the dataset containing the Boone indicators for Spain.
4.2.4 France
The degree of pass-through of the term premium lowering effect of quantitative easing to mortgage rates has been estimated at 0.22 for the French banking sector. Interestingly, the estimates for the amount of pass-through to corporate loan rates are both negative. This remarkable result may be due to omitted variable bias, since corporate loan rates are usually not only determined by policy rates and central government bond yields. Another important determinant of the interest charged on corporate debt, is the amount of risk involved in the debt. The omission of this variable might explain the remarkable results.
According to the results of Van Leuvensteijn et al. (2011), the French banking sector is amongst the least competitive of the euro countries investigated in their paper. Although many French banks have been privatised by now, the role of the government in the French banking sector is still fairly substantial. Many important entities are still controlled by the government (Van Leuvensteijn et al 2011, p. 3164).
4.2.5 Italy
Fairly high pass-through has been estimated for the Italian banking sector. This can mostly be contributed to the deregulation and liberalisation of the Italian banking sector during the early 1990’s (Van Leuvensteijn et al. 2011, p. 3162). Still, the average Boone indicator is not very low when compared to the other countries. Yet, the effect of the deregulation and liberalisation measures of the early 1990’s may still be affecting the pass-through nowadays, even though the Boone indicator increased gradually afterwards. Among all Eurozone countries, Italy had the largest Target deficit by the end of 2011, approximately 191 billion euros according to figure 1.
4.2.6 Austria
The estimate of the degree of pass-through to mortgage rate is 0.37 for this country. This is the highest among the all countries investigated in this thesis. Yet, the degree of through to small corporate loans is fairly small, and the degree of pass-through to large corporate loans is not significantly different from zero.
This may be due to the same reason why the estimates for France are negative: omitted variable bias. Another explanation might come from the fact that Austria has a relatively low Target deficit, and thus has received little aid compared to the other countries. Austria’s target deficit amounts 33 billion euros, which is the lowest among the investigated countries.
4.2.7 Portugal
Portugal has the most competitive banking sector among the examined countries. The average Boone indicator in the period 2003 – 2011 amounts -0.085. In addition, Van Leuvensteijn et al. (2011) have confirmed that that the degree of competition in the Portuguese banking sector has increased over the period 1992 – 2004. The estimated coefficients suggest a moderate degree of pass-through compared to the other
countries. This is in line with the hypothesis that the pass-through of quantitative easing increases when the banking sector gets more competitive.
5. Conclusion
In this thesis, the effect of quantitative easing in the Eurozone on long-term
commercial interest rates has been analysed by estimating three linear models using ordinary least squares. The countries that have been analysed are Austria, France, Greece, Ireland, Italy, Portugal and Spain. The long-term commercial interest rates that have been incorporated in the analysis are the average variable mortgage rate, the average commercial interest rate on loans with a face value smaller than one million euros and the average commercial interest rate on loans larger than one million euros. The yield on central government bonds with a maturity of ten years was used as a proxy for quantitative easing.
The estimated effect of this proxy on each of the aforementioned interest rates has been interpreted as the degree of pass-through of the term premium decreasing effect of quantitative easing to that particular interest rate. The estimated coefficients were subsequently matched and correlated with the average Boone indicator, a measure of the degree of competition in a country’s banking sector.
After doing this analysis the research question can be answered. Given the negative correlation between the estimated coefficients representing the degree of pass-through, and the average Boone indicators in the examined countries, the
research question can be answered as follows. The degree competition in the banking sector is positively correlated with the degree of pass-through of the term premium decreasing effect of quantitative easing to long-term commercial interest rates in the Euro area. Regarding the external validity, it is not a matter of fact that these findings apply to other countries, because quantitative easing has been implemented in
different ways in various countries.
To substantiate the findings of this thesis, a larger sample is required. This allows for a more robust estimation of the relation between the Boone indicator and the estimated coefficient for pass-through. A recommendation for further research is to do the analysis again with a larger sample. One method to do this is to run the analysis again for the same countries, yet with yearly estimates for pass-through and with yearly values of the Boone indicator. This would require a more extensive dataset than analysed in this thesis.
Reference list
• Kerstin Bernoth, Jürgen von Hagen & Ludger Schuknecht, “Sovereign risk premiums in the European government bond market” Journal of International Money and Finance 31, no. 5 (2012): 975-995.
• Jan Boone, “A new way to measure competition” The Economic Journal 118, no. 531 (2008): 1245-1261.
• Jeremy Bulow, Paul Klemperer, “Prices and the Winner’s Curse” The RAND Journal of Economics 33, no. 1 (2002): 1-21.
• Jens H.E. Christensen & Glenn D. Rudebusch, “The Response of Interest Rates to US and UK Quantitative Easing” The Economic Journal 122, no. 564 (2012): F385-F414.
• Stephan Dieckmann & Thomas Plank, “Default Risk of Advanced Economies: An Empirical Analysis of Credit Default Swaps during the Financial Crisis” Review of Finance 16, no. 4 (2012): 903-934
• Stefan Gerlach, Gert Schnabel, “The Taylor rule and interest rates in the EMU area” Economics Letters 67, no. 2 (2000): 165-171.
• Michael Joyce, David Miles, Andrew Scott & Dimitri Vayanos, “Quantitative easing unconventional monetary policy, an introduction” The Economic Journal 122, no. 564 (2012): F271-F288.
• Kenneth J. Kopecky & David D. Van Hoose, “Imperfect Competition in Bank Retail Markets, Deposits and Loan Rate Dynamics” Journal of Money, Credit and Banking 44, no. 6 (2012): 1185-1205.
• Michele Lenza, Huw Pill & Lucrezia Reichlin, “Monetary policy in exceptional times” Economic Policy 26, no. 62 (2010): 295-339.
• Michiel van Leuvensteijn, Jacob A. Bikker, Adrian A.R.J.M. van Rixtel &
Christoffer Kok Sørensen, “A new approach to measuring competition in the loan markets of the eura area” Applied Economics 43, no. 23 (2011): 3155-3167. • Michiel van Leuvensteijn, Jacob A. Bikker, Adrian A.R.J.M. van Rixtel &
Christoffer Kok Sørensen, “Impact of bank competition on the interest rate pass-through in the euro area” Applied Economics 45, no. 11 (2013): 1359-1380. • Dimitrios P. Louzis, Angelos T. Vouldis & Vasilios L. Metaxas,
“Macroeconomic and bank-specific determinants of non-performing loans in Greece: A comparative study of mortgage, business and consumer loan portfolios” Journal of Banking and Finance 36, no. 4 (2012): 1012-1027. • Christopher Martin & Costas Milas, “Quantitative easing, a sceptical survey”
Oxford Review of Economic Policy 28, no. 4 (2012): 750-764.
• Robert W. Rosenthal, “A Model in which an Increase in the Number of Sellers Leads to a Higher Price” Econometrica 48, no. 6 (1980): 1575-1579.
• Hans-Werner Sinn & Timo Wollmershhäuser, “Target loans, current account balances and capital flows: the ECB’s rescue facility” International Tax and Public Finance 19, no. 4 (2012): 468-508.
• Myron B. Slovin & Marie Elizabeth Sushka, “A Model of the Commercial Loan Rate” The Journal of Finance 38, no. 5 (1983): 1583-1596.
• Joseph E. Stiglitz, “Equilibrium in Product Markets with Imperfect Information” The American Economic Review 69, no. 2 (1979): 339-345.
• John B. Taylor, “Discretion versus policy rules in practice” Carnegie-Rochester Conference Series on Public Policy 39 (1993): 195-214.
• Michael Woodford, “The Taylor Rule and Optimal Monetary Policy” The American Economic Review 91, no. 2 (2001): 232-237.
Appendix
Graph 1: Time series variable mortgage rate
Graph 2: Time series corporate loans < € 1 million !0!%! !1!%! !2!%! !3!%! !4!%! !5!%! !6!%! !7!%! 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Ireland Greece Spain France Italy Austria Portugal !0!%! !1!%! !2!%! !3!%! !4!%! !5!%! !6!%! !7!%! !8!%! !9!%! 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Ireland Greece Spain France Italy Austria Portugal
Graph 3: Time series corporate loans > € 1 million
Scatterplot 1: !!, average Boone indicator (2003 – 2011) !0!%! !1!%! !2!%! !3!%! !4!%! !5!%! !6!%! !7!%! !8!%! 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Ireland Greece Spain France Italy Austria Portugal Ireland! Greece! Spain! France! Italy! Austria! Portugal! 0.0! 0.1! 0.2! 0.3! 0.4! C0.10! C0.05! 0.00! 0.05! 0.10! 0.15! Average'Boone'Indicator'(2003'4'2011)'
Scatterplot 2: !!, average Boone indicator (2003 – 2011)
Scatterplot 3: !!, average Boone indicator (2003 – 2011) Ireland! Greece! Spain! France! Italy! Austria! Portugal! C0.2! 0.0! 0.2! 0.4! 0.6! C0.10! C0.05! 0.00! 0.05! 0.10! 0.15! Average'Boone'Indicator'(2003'4'2011)' Ireland! Greece! Spain! Italy! Austria! Portugal! C0.2! 0.0! 0.2! 0.4! C0.10! C0.05! 0.00! 0.05! 0.10! 0.15! Average'Boone'Indicator'(2003'4'2011)'
Table 2A: Results model 1
! Ireland! Greece! Spain! France! Italy! Austria! Portugal!
Average!Boone!Indicator! (2003=2011)! !0.00899853( !0.06547641( 0.101175233( !0.04535128( !0.03997907( !0.05780819( !0.08492069( const! 2.504209**( (0.1206737)( (0.0312024)(2.835564**( (0.2452283)(1.179625**( (0.1194496)(2.19438**( (0.3011401)(1.274601**( (0.0995746)(1.163156**( (0.1655825)(1.449304**( REFIN_RATE! 0.593482**( (0.0288181)( 0.5475068**((0.0312024)( 0.7738803**((0.0341896)( 0.4117396**((0.0415611)( 0.6989918**((0.0336593)( 0.7065436**((0.0365193)( 0.7176462**((0.0435181)( GVT_BOND_YIELD! !0.0080356( (0.0118261)( (0.0062833)(0.0126395*( 0.2103597**((0.0425998)( 0.2156986**((0.0530242)( 0.2524818**((0.0603715)( 0.3738566**((0.0453917)( 0.1584939**((0.0132997)( n! 137( 137( 137( 137( 137( 137( 137( R=squared! 0.842( 0.7182( 0.8261( 0.7433( 0.7759( 0.9129( 0.7787(
Numbers between parentheses are robust standard errors * : p ≤ 0.05
Table 2B: Results model 2
! Ireland! Greece! Spain! France! Italy! Austria! Portugal!
Average!Boone!Indicator! (2003=2011)! !0.00899853( !0.06547641( 0.101175233( !0.04535128( !0.03997907( !0.05780819( !0.08492069( const! 3.198503**( (0.2007787)( 4.098763**((0.198347)( (0.4250379)(1.653809**( (0.1397671)(2.020786**( 1.581483**((0.349975)( 1.505262**((0.095792)( (0.1358993)(4.116607**( REFIN_RATE! 0.6835357**( (0.0458112)( 0.4325919**((0.0569572)( 0.3519973**((0.0597679)( (0.0350308)(1.056921**( 0.5333993**((0.0436384)( 0.8864853**((0.0258729)( (0.0376925)(0.506097**( GVT_BOND_YIELD! 0.0583269**( (0.0222519)( 0.1225294**((0.0113553)( 0.5137149**((0.0736809)( !0.155703**((0.0544366)( (0.0639221)(0.396684**( (0.0349595)(0.0793738*( 0.2297257**((0.011761)( n! 137( 137( 137( 137( 137( 137( 137( R=squared! 0.735( 0.593( 0.4051( 0.9351( 0.6297( 0.9466( 0.7185(
Numbers between parentheses are robust standard errors * : p ≤ 0.05
Table 2C: Results model 3
! Ireland! Greece! Spain! France! Italy! Austria! Portugal!
Average!Boone!Indicator! (2003=2011)! !0.00899853( !0.06547641( 0.101175233( !0.04535128( !0.03997907( !0.05780819( !0.08492069( const! 2.214864**( (0.1273595)( (0.2863419)(3.034338**( (0.3204953)(0.7585712*( (0.1189551)(1.673514**( (0.285088)(0.508563( (0.0861038)(1.211183**( (0.2242178)(2.265499**( REFIN_RATE! 1.0012**( (0.0277957)( 0.3452458**((0.0845115)( 0.7735338**((0.0438261)( (0.0371676)(1.024844**( 0.6980422**((0.0365238)( 0.9466874**((0.0243415)( 0.4317451**((0.0602115)( GVT_BOND_YIELD! !0.023300( (0.0155072)( 0.1336339**((0.0147261)( 0.2299969**((0.054231)( !0.202447**((0.0461117)( 0.3177491**((0.049627)( (0.0302362)(!0.0257647( 0.2652862**((0.0197841)( n! 137( 137( 137( 137( 137( 137( 137( R=squared! 0.9422( 0.4559( 0.8106( 0.9227( 0.7942( 0.9571( 0.5458(
Numbers between parentheses are robust standard errors * : p ≤ 0.05
