Interest rates and bank
risk-taking: empirical evidence from
the U.S.
Mark de Boer 2047829
Abstract
There is a lot of growing consensus that a long period of low interest rates can increase the level of bank risk-taking. Using a panel dataset from the U.S. banking sector during 2009-2015, this paper finds a significant and positive relationship and finds therefore no evidence of increased risk-taking by U.S. banks. In exploring this result the fixed effects model and the general method of moments are used. The result is a robust relationship due to the use of different measures of risk-taking and interest rates.
Field keywords: Bank risk-taking, interest rates, monetary policy & U.S. banks.
2 1. Introduction
There has been considerable academic and regulatory interest in how to mitigate bank risk-taking behaviour in the recent years. Undue risk-taking by banks jeopardizes the safety and soundness of individual institutions as well as the stability of the entire financial sector when contagion causes risks to spill over to other financial institutions as discussed in other studies (see, e.g., Srivastav and Hagendorff, 2015; Maddeloni and Peydró, 2011).
Claessens, Coleman, Donnelly, (2016) reported that after the crash of the banking market between 2007 and 2009, the Fed felt it needed to pull out all of the stops to prevent the economy from collapsing into a new Great Depression and therefore the Fed stimulated the economy by cutting the costs of borrowing. The interest rate levels1 are still at historical low levels today and according to Fed policymakers,
interest rates are not going to return to pre-crisis levels for a long time period (Claessens, Coleman, Donnelly, 2016).
Traditionally the role of banks is to take deposits and make loans (Hull, 2015). The interest charged on these loans is greater than the interest being paid on deposits and this covers up for administrative costs and loan losses. The low interest rates of the beginning of the 21th century drove net interest margins down and caused banks to take more risk. Delis and Kouretas (2011) explain that a low interest-rate environment drives, ceteris paribus, bank margins and informational asymmetries down. As a consequence, Banks react softening their lending standards, thus raising the level of risk assets in their portfolios and worsening the equilibrium risk of failure. Especially an environment of low interest rates following a period of high rates is particularly problematic, because the bank can have the incentive to increase their risk-taking behaviour in search for return (Rajan, 2005). A banks fixed liabilities can be relatively high and a change in the interest environment from high to low can persuade a bank to increase their risk-taking behaviour.
3
According to Claessens, Coleman, Donnelly, (2016) low interest rates help economies recover and can enhance a bank’s balance sheets and performance by supporting asset prices and reducing non-performing loans, but persistently low interest rates may also erode the profitability of banks as low rates are typically associated with lower net interest margins. So how do U.S. banks respond to a low interest environment after the financial crisis of 2007 till 2009?
This papers aim is to understand the impact of interest rates on bank risk-taking in the United States of Amerika during a post crisis time period from 2009 till the end of 2015. Interest rates are low for a long period of time and a lot of research conclude that low interest rates are increasing the risk-taking behaviour of banks, but most recent literature focus on time periods before and during the crisis and use a European based dataset (see, e.g., Delis and Kouretas, 2011; Jiménez, Ongena, Peydró and Saurina, 2008).
To investigate the relationship between interest rates and bank risk-taking, this study constructs a panel dataset consisting of commercial, savings and cooperative banks from the United States. The three bank risk-taking measures which are used are: risk assets to total assets, non-performing loans and loan loss provisions to total assets. This paper employs the short-term interest rate, the long-term interest rate and the bank-level lending interest rate as measures for the explanatory variable interest rate. The methodology used in this paper builds on the methodology used by Delis and Kouretas (2011).
4
This paper contributes to previous literature by further exploring the relationship between interest rates and bank risk-taking, but taking an up to date database with the focus on the U.S.. In addition, the period after the crisis is explored in contrast to other studies (see, e.g., Maddaloni and Peydró, 2011; Cebenoyan and Strahan, 2004). The combination of U.S. banks, a post crisis focus and an up to date panel dataset are reasons why this study adds meaningful contribution to existing literature.
The remainder of this article will be as follows: section two discusses related theory and testable hypothesis. Section three reports, in detail, data and methodology and research methods which will be used. Section four will discuss the empirical result in combination with a discussion. Section five is the last section of this paper and ends with a conclusion.
2. Literature review
2.1 Monetary policy transmission and interest rates
The early finding of Friedman and Schwartz (1963) shows that monetary policy actions are followed by movements in real output that may last for two years or more, was on the basis of today’s research on interest and bank risk-taking according to Bernanke and Gertler (1995).
5
a focus on the impact of changes in monetary policy on the supply of loans by depository institutions. When interest rates are low, banks should search for new financing sources at a higher cost due to the treat of deposit withdrawals as discussed in Andries, Cocriş and Pleşcău (2015).
2.2 Interest rates and bank risk-taking
The focus of this paper is the impact of interest rates on bank risk-taking. Borio and Zhu (2008) identified the risk-taking channel and they state that the transmission channel of monetary policy should take into account the changes in the risk-taking activity of banks (Andries, Cocriş and Pleşcău, 2015). They describe the idea behind the risk-taking channel as the link between monetary policy and the perception and pricing of risk by economic agents. The risk-taking channel is caused by three factors following Borio and Zhu (2008). First they describe the impact of interest rates on valuation, income and cash flows. Lower interest rates increases asset values, collateral values, incomes and profits which can reduce risk perceptions. Secondly they identify the risk-taking channel through the term of Rajan (2005): ‘in search for yield’. Rajan (2005) explains that in a persistently low interest rate environment, the investment manager and the bank will stretch for yield by taking on risk. The investment manager has a salary based on returns and high returns can attract new investors. The bank takes more risk, because high long-term fixed liabilities have to be paid off. The last factor which is identified by Borio and Zhu (2008) is communication policies and reaction function of the central bank. High transparency by the central bank can reduce uncertainty about future expectation and compresses the risk premia.
2.3 Empirical evidence of the risk-taking channel
6
Jiménez, Ongena, Peydró and Saurina, (2008) use data of Spanish banks consisting over a time period of 23 years. They suggest that following a monetary expansion will increase the borrowers net worth an the appetite for liquidity risk, but that these factors are not the only reasons for the banks ‘new engagements’. Banks also want to take more credit risk. However they state that the impact is not equal for every bank. The negative relationship is stronger for small and commercial banks. The state that there are three important factors which determine the impact of monetary policy and bank risk-taking: the Bank, the borrower and the market characteristics.
Ioannidou, Ongena, Peydró (2009) build on the research of Jiménez, Ongena, Peydró and Saurina, (2008) and explored Bolivian banks from 1991 till 2003 and find that relaxing monetary conditions enhances risk-appetite of banks. They used the Fed funds rate as an interest variable measure which is particularly interesting for this paper. An interesting finding is that when they observed loans with a subprime credit rating or loans to riskier borrowers with current or past non-performance also becomes more likely when the federal funds rate is low, but banks do not seem to price this additional risk.
According to Brissimis and Delis (2009) the possible heterogeneity of the response of banks following a change in monetary policy is not addressed by Jiménez, Ongena, Peydró and Saurina, (2008) and Ioannidou, Ongena, Peydró (2009). This heterogeneity arises from differential bank balance sheet characteristics, such as liquidity, capitalization and size and this may also have implications for their risk-taking. The analysis of Bressimis and Delis (2009) showed that banks with healthier balance sheets and market power behave differently than banks with weaker balance sheets and market power. However in times of adverse economic shocks, even the balance sheets of the healthiest banks quickly decline and banks which are exposed to abnormal high risk may become insolvent.
7
bank risk-taking. Low interest environment increases risk-related bank assets and alters the composition of euro area bank portfolios to a more risk position. This relationship is stronger for banks with non-traditional banking activities and weaker for banks with high levels of capitalization.
Delis and Kouretas (2011) only explored the relationship between interest rates and bank risk-taking with a European dataset. Maddaloni and Peydro (2011), find evidence in Europe, but also in the U.S. that low short-term interest rates lower the lending standards for household and corporates loans. They do not find this relationship for long-term interest rates. Countries with lower lending standards before the crisis experienced a lower performance after the crisis. They provide also evidence that too low for too long monetary policy rates increases the risk of bank assets and were a key factor leading to the financial crisis. Finally they conclude that monetary policy affects financial stability.
This study will build heavenly on the study of Delis and Kouretas (2011). However in this paper a dataset consisting of U.S. banks is used. Empirical evidence of the existence of an risk-taking channel in the USA is developing. Studies such as Buch, Eickmeier and Prieto (2011), Delis, Hasan and Mylonidis (2012) and Dell’Ariccia, Laeven and Suarez (2016) are one of the most recent papers which perform direct empirical analysis of the existence of a risk-taking channel.
However monetary policy can affect short-term interest by expansionary monetary policy, Buch, Eickmeier and Prieto (2011) do not find evidence for a risk-taking channel for the entire banking system after expansionary monetary policy shocks. They find that the risk-taking channel works mainly through small banks and that large banks and foreign banks do not change their exposure to new risk.
8
Dell’Ariccia, Laeven and Suarez (2016) find a negative relationship between short-term interest rate environment and bank risk-taking which suggest that lower interest rates increases bank risk-taking. However this effect is less pronounced for poorly capitalized banks.
If we look at previous research about the risk-taking channel of monetary policy, mainly all of the research conclude that there is a negative relationship between interest rates and bank risk-taking which suggest that low interest rates increase the level of bank risk-taking. Therefore the following hypothesis is developed based on previous literature:
Hypothesis 1: Interest rates have a negative impact on bank risk-taking during the time period starting from 2009 till the end of 2015.
3. Data and Methodology
This paper builds upon the methodology used by Delis and Kouretas (2011). The general empirical model in Eq.(1) is estimated by using a panel regression model to investigate the relationship between interest rates and bank risk-taking. The general empirical model used in this paper is in the following form:
𝑟
𝑖𝑡= ∝ + 𝛽1𝑖𝑟𝑖𝑡+𝛽2𝑏𝑖𝑡+ 𝛽3𝑦𝑐𝑖𝑡+ 𝜇𝑖𝑡 ,(1)
Where 𝑟 is the bank risk-taking variable, of bank 𝑖, written at time 𝑡 as a function of the interest rate variable 𝑖𝑟, in country 𝑖 at time 𝑡 and controlled for, bank-level variable 𝑏, for bank 𝑖 at time 𝑡 and country macroeconomic variable 𝑐, for bank 𝑖 at time 𝑡.
The panel dataset consists of time series data and cross-sectional data. Annual U.S. Bank specific data, interest rate data and macroeconomic data for the measures of the dependent, explanatory and control variables are collected from the database Bankscope, the database of the Worldbank and the database of the Federal Reserve. Data of the dependent variable is collected from Bankscope and consists of annual data about active2 U.S. commercial, savings and cooperative
9
banks between the time period 2009-2015. Following Delis and Kouretas (2011), investment banks are not explored, because they do not take deposits. The starting date 2009 is particularly used as starting point due to lack of data availability3.
Data of the explanatory variables short-term interest rate and long-term interest rate are collected from the website of the Federal Reserve and data of the bank-level lending rate is obtained from Bankscope. At last the macroeconomic data is collected from the database of the Worldbank. A type of criticism could arise due to the use of annual data. According to Delis and Kouretas (2011) the analysis of the interest rates-bank risk nexus is possible with using annual data (see, Ashcraft, 2006, p.760).
Following Duprey and Lé, (2015), I use consolidation codes in Bankscope which capture the balance sheet sensitivity of banks and not the actual size of the banking market4. To capture the influence of possible outliers, some variables are
winsorized, because trimming would have led to a loss of valuable information. The outlier labelling rule from Hoaglin and Iglewicz (1987) is used to control for possible outliers and also the non-normal distribution. Therefore some variables are winsorized at the 1% and 99% levels for each year. The variables which are winsorized due to extreme outlier are: risk assets to total assets, non-performing loans, loan loss provisions to total assets, bank-level lending rate, capitalization, size, lagged profitability, off-balance sheet items and efficiency. After adapting all these changes, the final sample consists of 6309 banks and 44.163 bank year observations.
3.1 Dependent variables
To what extent affect interest rates the dependent variable bank risk-taking is the main issue addressed in this paper. To measure bank risk-taking I will use three different measures which are taken from existing literature. I use three measures to test for robustness and to capture different characteristics of bank risk-taking. In line with Delis and Kouretas (2011) I use the ratio risk assets to total assets and non-performing loans to total assets. Risk assets are all bank assets minus cash,
10
government securities and balances due from other banks. The ratio risk assets to total assets reflects a direct measure of risk-taking behaviour of a particular bank. An increase of this ratio indicates that the bank takes a more risky position due to less risk-free assets. The ratio non-performing loans to total loans reflects credit risk. Credit risk is the primary driver of risk for most banks, although other risks obviously exist according to Jiménez, Lopez and Saurina (2013). In other words, it reflects the quality of bank assets. A higher ratio indicates that the bank is taking more risk, because the quality of the bank assets will go down and this causes more credit risk on the loan portfolio.
Following Andries, Cocriş and Pleşcău (2015) the third ratio in this study which will be used is the loan loss provisions ratio. They explain when loans become non-performing, banks are faced with the possibility of not recovering the principle and the interest, so they have to use a mechanism through which they create a kind of cushion that protects them against unexpected losses on loans. A negative ratio between interest rates and loan loss provisions will include that banks take higher risk when interest rates are low.
3.2 Explanatory variables
11
rate consists of the market yield on U.S. Treasury securities at 10-year constant maturity.
The bank-level lending rate is displayed in figure 1, 2 and 3. Figure 1 shows a simple regression between the relationship between risk assets to total assets and the bank-level lending rate. The bank-level lending rate is the ratio of interest income to total customer loans and is calculated from the data of Bankscope. The regression line in figure 1 indicates a first impression that a negative relationship exist between interest rates and bank-risk taking. However figure 2 and 3 indicate a positive relationship between interest rates and bank risk-taking. Later in this chapter these results will be further explored.
Fig. 1: The relationship between risk assets to total assets and the bank-level lending rate is displayed. The relationship points out towards a negative relationship between bank risk-taking and the bank-level lending rate.
.2 .4 .6 .8 1 R isk a sse ts/ t o ta l a sse ts .02 .04 .06 .08 .1
12
Fig. 2: The relationship between non-performing loans and the bank-level lending rate is displayed. The relationship points out towards a positive relationship between bank risk-taking and the bank-level lending rate.
Fig. 3: The relationship between loan loss provisions to total assets and the bank-level lending rate is displayed. The relationship points out towards a positive relationship between bank risk-taking and the bank-level lending rate.
0 .0 5 .1 .1 5 N o n -p e rf o rmi n g l o a n s .02 .04 .06 .08 .1
Interest income/ total loans
0 .0 1 .0 2 .0 3 .0 4 L o a n l o ss p ro vi si o n s to t o ta l a sse ts .02 .04 .06 .08 .1
13 3.3 Control variables
Interest rates are not the only variables which can influence bank risk-taking behaviour and therefore there are some control variables added which will be used in the research of this paper to ensure that results are not biased and to test for robustness.
In this study I control for some bank-specific characteristics. The first bank-specific control variable added in this study is capitalization which is measured as the ratio of equity to total assets. Buch, Eickmeier and Prieto (2011) conclude that the riskiness of banks declines with the degree of capitalization. Jokipii and Milne (2011) also note that the management of short-term adjustments in capital and risk are dependent on the amount of capital the bank holds in excess of the required minimum. Another measure used in this paper as a control variable is size which is measured as logarithm of the total assets. Bhagat, Bolton and Lu (2015) conclude that size has a positive relationship with risk-taking of a bank, but they suggest that financial firms should more focus on capitalization than on bank size alone. Martynova, Ratnovski and Vlahu (2015) show that more profitable banks not always have lower risk-taking incentives. They conclude that higher bank profitability (or, similarly, higher bank franchise value or bank capital) is not panacea against risk-taking. High profitability banks can borrow more and engage in side investments on a larger scale. Therefore I add a profitability measure which is measured by profits before tax divided by total assets. In addition I add a ratio off-balance sheet items to total assets to control for engagement in side effects which can increase bank risk-taking according to Mikati (2012). Due to technology changes I include an efficiency proxy which is measured by total revenue to total expenses. Technology efficient banks may be more capable in managing risks; however, higher risks may also explain technical efficiency levels if they are responsible for the level of bank income (Delis & Kouretas (2011).
14
countries. The second macro-economic variable which I added is domestic credit by banks over GDP. Larrain (2006) explains that some banks are incapable in attracting capital which is not attracted from domestic credit offered by banks. The last macroeconomic variable which is included is a banking concentration variable. Boyd and de Nicolo (2005) show that there exists a positive relationship between the number of bank competitors and risk-seeking, but it is fragile. Therefore this variable is added as a control variable which is measured using the three-bank concentration ratio in terms of total assets (Hoque et al., 2015).
I do not control in this study for regulation which is used in previous research. Delis and Kouretas (2011) and Drakos, Kouretas and Tsoumas (2014) use regulatory variables to control for their empirical research and are constructed by Barth et al. (2008). These regulatory variables consist of capital stringency, official supervisory power and market discipline, but due to a lack of data these variables could not be constructed.
3.4 Descriptive statistics
15
Table 1 Variable Obs Mean Std. Dev. Min Max
Risk assets 40,798 0.756 0.147 0.290 0.973
Non-performing loans 40,607 0.046 0.028 0.000 0.150 Loan loss provisions to total assets 42,463 0.004 0.006 -0.003 0.038 Capitalization 42,468 0.111 0.035 0.042 0.264
Size 42,468 12.190 1.322 9.658 16.823
Profitability 42,468 0.008 0.011 -0.045 0.035
Efficiency 42,440 0.733 0.206 0.326 1.711
16
Table 2: reports correlation coefficients. Risk assets is the ratio of risk assets to total assets. Non-performing loans is the ratio of non-performing loans to total loans. loan loss provisions to total assets is the ratio loan loss provisions to total assets. Capitalization is the ratio of equity capital to total assets. Size is the natural logarithm of total assets. Lagged profitability is the ratio of profits before tax to total assets in year t-1. Efficiency is the ratio of total revenue to total expenses. Off-balance sheet items is the ratio of off-balance sheet items to total assets. Economic growth is the GDP growth. Importance of banks is the domestic credit provided by the banking as a share of the GDP. Concentration is the 3-bank concentration ratio. Short-term interest rate is the Federal funds effective rate. Long-term interest rate is the Market yield on U.S. Treasury securities at 10-year constant maturity and the bank-level lending rate is the ratio of interest income to total customer loans.
Table 2 Risk Assets Non-performing loans provisions to total
assets Capita-lization Size
Lagged
profit-ability Efficiency
balance sheet
items Economic growth Importance of banks Concen-tration
17 4. Results and discussion
4.1 Panel data regression model
As mentioned in section 3, a panel dataset is constructed to investigate the relationship between interest rates and bank risk-taking. To test this relationship, I will use a panel regression model. Panel data can exhibit unobserved heterogeneity. In other words, the panel data of this paper contains many different banks which have different characteristics which can influence the risk-taking behaviour of one particular bank in the sample of this paper which is not included in the models used to perform the research.
Brooks (2004) indicates two models which are frequently used when analysing financial panel datasets. He describes that the panel dataset can be analysed by a panel data regression model with fixed effects or a panel data regression model with random effects. To explore if unobserved heterogeneity is a major concern and which model should be applied, a Hausman test is performed. The null hypothesis of the Hausman test assumes that the preferred model is random effects and the alternative hypothesis states that the preferred model is fixed effects. A rejection of the null hypothesis will indicate that there are fixed effects present and that a fixed effects model should be used. After performing the Hausman test for all independent variables, the results indicate that there are fixed effects present and the null hypothesis will be rejected. Therefore the preferred panel data regression model which should be use is the panel data regression model with fixed effects.
According to Torris-Reyna (2007) when using fixed effects, we assume that something within the entities may impact or bias the explanatory or dependent variables and therefore we need to control for this. This is the rationale behind the assumption of the correlation between entity’s error term and explanatory variables. Fixed effects remove the effect of those time-invariant characteristics so we can assess the net effect of the explanatory variable on the dependent variable.
18
that a lower interest rate will result in lower risk-taking by the bank. First I regressed the proxy risk assets to total assets with the short and long-term interest rate which results in a positive and significant relationship. However when I used the bank-level lending rate the relationship tends to be significant and negative. Secondly, when non-performing loans are used for the dependent variable bank risk-taking and the short-term, long-term and the bank-level lending rates are used for the explanatory variable interest rates, the relationship turns out to be significant and positive. At last, when loan loss provisions to total assets is used for the dependent variable bank risk-taking and the short-term, long-term and bank-level lending interest rates are used for the explanatory variable interest rates, the relationship turns also to be significant and positive. The relationship between bank risk-taking and interest rates has the most influence when the bank level lending rate is used for the explanatory variable interest rates. These results indicate that U.S. banks are behaving differently in contrast to European banks. European banks are taking more risk when interest rates drop and U.S. banks are taking less risk when interest rates drop. One interesting thing resulting from the results is that the relationship between lagged profitability and the dependent variables risk assets to total assets, non-performing loans and loan loss provisions is almost in every regression significant and negative. This indicates that if the bank’s profitability decreased in the previous period, bank’s will increase risk-taking in the subsequent period.
However the fixed effect model does not take into account whether the dynamic and persistence of bank risk-taking and the possible endogeneity of some control variables affect the results. Therefore these results should be carefully examined.
4.2 Generalized method of moments
19
Delis and Kouratas (2011) state that an essential concern in estimating Eq. (1) using an fixed effects model is that bank risk persists and therefore will deviate from its equilibrium in the short-run. They state that there are at least four arguments to explain the dynamic nature of bank risk and therefore will deviate from its equilibrium in the short-run. In line with Keeley, 1990, Cordella and Yeyati, (2002) persistence can reflect the existence of intense competition, which tends to enhance the risk-taking of bank. The second argument which Delis and Kouretas (2011) explain is that relationship-banking with risky borrowers will have a long term influence on the levels of bank risk-taking (Claudio Borio, Craig Furfine and Philip Lowe, 2001). Lastly risk can be influenced by regulation. Allen and Gale (2000) show that regulation can have a negative influence on risk-taking when using static models and a positive influence within a dynamic model. So to control for the persistence and dynamic nature of bank risk, I use a dynamic model instead of a static model which is biased. Therefore the following model (2) will be used which is a variant of model (1):
𝑟
𝑖𝑡= ∝ +𝛿(𝑟𝑖,𝑡−1)+ 𝛽1𝑖𝑟𝑖𝑡+𝛽2𝑏𝑖𝑡+ 𝛽3𝑦𝑐𝑖𝑡+ 𝜇𝑖𝑡 ,
(2)
In order to perform the generalized method of moments equation (2) is used. The generalized method of moments is a dynamic model, including a lagged dependent variable which captures the persistent character of bank risk and provide unbiased results according to Andries, Cocriş and Pleşcău (2015). This is not possible with the fixed effects model, because the lagged dependent variable is correlated with the error term (Baum, 2006). A value of 𝛿 statistically equal to 0 implies that bank risk adjust quickly, a value equal to 1 implies slow adjustment and a value between 0 and 1 implies that risk persist, but will eventually return to average level (Delis & Kouretas, 2011). If values for 𝛿 become negative, there probably is a problem with the dataset.
20
instruments. Variables which will be treated as endogenous variables are: interest rate variables, capitalization, lagged profitability, efficiency and off-balance sheets items. Size is treated as a predetermined variable. The macroeconomic control variables, economic growth, importance of banks and concentration are treated as exogenous variables.
The results of using the generalized methods of moments estimated by equation (2) are displayed in table 4. The results are in line with the results from the fixed regressions in table 3, if we look at the relationship between interest rates and bank risk-taking. The results still indicate a positive relationship between interest rates and bank risk-taking and are not in line with Delis and Kouretas 2011. However Delis and Kouretas (2011) investigate the relationship between interest rates during 2001-2008. This paper investigates this relationship during a post-crisis period. In contrast to Delis and Kouretas (2011), Dell’Ariccia, Laeven and Suarez (2016) find evidence that reductions in interest rates have a disproportionately positive effect on bank risk-taking during periods when there are relatively few bank failures, which is consistent with our results.
The first dependent variable which is used is risk assets. The results for the short-term interest are not significant and therefore not valuable. The results from the long-term interest rate and the bank level lending rate are significant and indicate a positive and negative value. These results add less economic significance to this paper, because they are contradictory. However it should be noted that the generalized method of moments use a dynamic setting, all risk coefficients are closer to zero.
21
subsequent. The impact of efficiency and off-balance sheet items is zero and therefore adds no economic significance to this paper. This result also applies to the macroeconomic variables.
The third dependent variable, loan loss provisions to total assets, shows also a significant and positive value between the short-term, long-term and bank-level lending rate which indicates that lower interest variables decrease bank risk-taking. The impact of capitalization on loan loss provisions is significant and negative which indicates that high capitalized banks take less risk. The impact of size is zero and adds therefore no economic significance to this paper. The impact of lagged profitability is significant and positive which indicates that banks which were less profitable in the previous period will take less risk in the subsequent period. The impact of efficiency is significant and positive which indicates that banks whch are more efficient will take more risk. The impact of off-balance sheet items is significant and negative which indicates that banks which have more off-balance sheet items take less risk. The impact of the macroeconomic variables is almost zero and adds therefore no economic significance to this paper.
22
Table 3: Fixed effects I II III IV V VI VII VIII VIIII
Capitalization 0.275* 0.268* 0.175* -0.009* -0.109* -0.100* -0.031* -0.038* -0.041* (0.026) (0.025) (0.025) (0.001) (0.003) (0.003) (0.002) (0.002) (0.002) Size 0.022* 0.021* 0.007* -0.002* -0.012* -0.010* 0.003* 0.002* 0.003* (0.002) (0.002) (0.002) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Lagged Profitability -0.025 0.015 -0.012 0.012* -0.026* -0.032* -0.072* -0.074* -0.078* (0.044) (0.043) (0.042) (0.002) (0.005) (0.005) (0.003) (0.003) (0.003) Efficiency -0.048* -0.051* -0.060* 0.000 0.000 0.002* 0.001* 0.001* 0.002* (0.003) (0.003) (0.003) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Off-balance sheet items 0.035* 0.034* 0.036* 0.001* -0.000 -0.001 -0.000** -0.000** -0.000**
(0.004) (0.004) (0.004) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Economic growth 0.005* 0.004* 0.006* -0.013* -0.004* -0.003* -0.001* 0.000* 0.000* (0.001) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Importance of banks 0.000* -0.000* -0.001* -0.000* -0.001* -0.001* -0.000* -0.000* -0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Concentration 0.004* -0.008* 0.011* 0.016* 0.015* 0.014* 0.002* 0.001* 0.001* (0.000) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Short-term interest rate 0.089* 0.280* 0.026*
(0.014) (0.001) (0.001)
Long-term interest rate 0.024* 0.000* 0.001*
(0.001) (0.000) (0.000)
Bank-level lending rate -2.336* 0.408* 0.104*
(0.063) (0.008) (0.005)
Obs 35042 35042 34987 34754 34754 34745 36131 36131 36067 Number of banks 6187 6187 6180 5801 5801 5801 6189 6189 6184 R-squared 0.099 0.096 0.097 0.391 0.180 0.206 0.106 0.123 0.125 Hausman 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Table 3: reports coefficients and standard errors (in parentheses). In regressions I-III the dependent variable is risk assets to total assets. In regressions IV-VI the
dependent variable is non-performing loans. In regressions VII-VIIII the dependent variable is loan loss provisions to total assets. The explanatory variables are:
23
Table 4 GMM I II III IV V VI VII VIII VIIII
Lagged Risk assets 0.555* 0.214* 0.379* (0.029) (0.027) (0.021) Lagged NPL 0.999* 1.004* 1.002* (0.001) (0.000) (0.001) Lagged LLPTA 0.463* 0.479* 0.397* (0.015) (0.014) (0.014) Capitalization 0.547* 0.455* 0.093*** 0.002* 0.002* 0.001* -0.093* -0.095* -0.100* (0.075) (0.053) (0.055) (0.000) (0.000) (0.000) (0.006) (0.005) (0.005) Size -0.006 0.024* -0.022* 0.000* 0.000* 0.000* 0.000 0.001*** 0.001*** (0.010) (0.005) (0.005) (0.000) (0.000) (0.000) (0.001) (0.001) (0.001) Lagged Profitability -0.269* -0.034 -0.180* -0.003* -0.003* -0.003* 0.059* 0.068* 0.028* (0.047) (0.041) (0.041) (0.000) (0.000) (0.000) (0.008) (0.008) (0.008) Efficiency -0.017** -0.049* -0.015* 0.000* 0.000* 0.000* 0.010* 0.010* 0.010* (0.007) (0.005) (0.005) (0.000) (0.000) (0.000) (0.001) (0.000) (0.000) Off-balance sheet items 0.046* 0.066* 0.077* 0.000*** 0.000*** -0.000 -0.008* -0.007* -0.004* (0.018) (0.016) (0.016) (0.000) (0.000) (0.000) (0.001) (0.001) (0.001) Economic growth 0.001 0.001*** 0.002* -0.000* -0.000 0.000 0.001* 0.001* 0.001* (0.001) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Importance of banks 0.001* -0.000** 0.000 0.000 -0.000 0.000*** -0.000* -0.000* -0.000* (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Concentration -0.003* -0.007* 0.003* 0.000* 0.000*** -0.000 0.001* 0.000** 0.000* (0.001) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Short-term interest rate 0.007 0.002* 0.004**
(0.023) (0.000) (0.002)
Long-term interest rate 0.021* 0.000* 0.001*
(0.001) (0.000) (0.000)
Bank-level lending rate -1.374* 0.005* 0.124*
(0.140) (0.001) (0.011) Obs 28287 28287 28245 28949 28949 28939 29934 29934 29874 AR1 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 AR2 0.004 0.482 0.043 0.086 0.247 0.327 0.020 0.019 0.358 AR3 0.196 0.661 0.903 0.367 0.639 0.651 0.358 0.457 0.385 Sargan 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
24 5. Conclusion
The aim of this paper is to study the relationship between interest rates and bank risk-taking during the time period starting from 2009 till the end of 2015. Using a panel dataset consisting of U.S. banks, a fixed effects model and the generalized method of moments is performed.
Previous research on the debate of interest rates and bank risk-taking suggest that there is a negative relationship between interest rates and bank risk-taking. Low levels of interest rates tend to increase bank risk-taking. This relationship is not captured in the dynamic model approach used in this paper. The relationship between interest rates and bank-risk taking is significant and positive which indicates that lower interest rates decrease bank risk-taking. The results are robust due to the use of different risk-taking measures and different levels of interest rates. However there is still some endogeneity present in the models which could not be eliminated. Therefore the results should be examined carefully.
The difference in results with previous research from for instance Delis and Kouretas (2011) can appear from the use of a different time period and the use of a different country. However recent research from Dell’Ariccia, Laeven and Suarez (2016) find evidence that reductions in interest rates have a disproportionately positive effect on bank risk-taking during periods when there are relatively few bank failures, which is consistent with our results.
Implications for future research are the use of a larger and richer dataset of U.S. banks to capture the amount of endogeneity which is still present in the models of this paper. In addition, future research should focus on studying the effect of low interest rates after adverse economic shocks on bank risk-taking in the U.S., because there is less research about bank taking in the U.S. than bank risk-taking in Europe.
25
Allen, F., Gale, D., 2004. Competition and Financial Stability. Journal of Money,
Credit, and Banking, (36):3.
Andries, A. m., Cocriş, V., Pleşcău, I., 2015. Low interest rates and bank risk-taking: has the crisis changed anything? Evidence from the Eurozone. Review of
Economic business studies, 8(1): 125-148.
Ashcraft, A. B., 2006. New evidence on the lending channel. Journal of Money,
Credit, and Banking, 38:751-776.
Baum, C. F., 2006. An Introduction to Modern Econometrics using Stata. Stata Press, Texas, U.S..
Bernanke, B. S., Gertler, M., 1995. Inside the black box: the monetary credit channel of monetary policy transmission. Working paper No. 5146. National bureau of economic research.
Bhagat, s., Bolton, B., Lu, L., 2015. Size, leverage, and risk-taking of financial institutions. Journal of Banking & Finance, 59:520-537.
Binder, M., Hsiou, C., Pesaran, M. H., 2003. Estimation and inference in short panel vector auto-regressions with unit roots and cointegration. Economic Theory, 21,795-837.
Borio, C., Furfine, C., Lowe, P., 2001. Procyclicality of the financial system and financial stability: issues and policy options. Marrying the macro- and
micro-prudential dimensions of financial stability, (1):1-57
Borio, C., Zhu, H., 2008. Capital regulation, risk-taking and monetary policy: a missing link in the transmission mechanism? Bank for international Settlements Working Paper, No. 268.
Boyd, J. H., De Nicolo, G. 2005. The theory of bank risk taking and competition revisited. The journal of Finance, 60(3):1329-1343.
26
Brooks, C., 2014. Introduction Econometrics for Finance. Cambridge University Press, Cambridge, United Kingdom.
Buch, C.M., Eickmeier, S., Prieto, E., 2011. In search for yield? Survey-based evidence on bank risk taking. Working paper No. 10/2011.
Cebenoyan, A. S., Strahan, P. E., 2004. Risk management, capital structure and lending at banks. Journal of Banking & Finance. 28:19-43.
Claessens, S., Coleman, N., Donnelly, M., 2016. "Low-for-long" interest rates and net interest margins of banks in Advanced Foreign Economies. Retrieved from: https://www.federalreserve.gov/econresdata/notes/ifdp-notes/2016/low-for-long- interest-rates-and-net-interest-margins-of-banks-in-advanced-foreign-economies-20160411.html.
Delis, M. D., Hasan, I., Mylonidis, N., 2012. The risk-taking of monetary policy in the USA: Evidence from micro-level data, Mimeo, Rensselaer Polytechnic Institute.
Delis, M. D., Kouretas, G. P., 2011. Interest rates and bank risk-taking. Journal of
Banking & Finance, 35: 840-855.
Dell’Ariccia, G., Leaven, L., Suarez, G. A., 2016. Bank leverage and monetary policy’s risk-taking channel: evidence from the United States. Working paper No. 1903.
Drakos, A. A., Kouretas, G. P., Tsoumas, C., 2014. Ownership, interest rates and bank risk-taking in Central and Eastern European countries. International Review
of Financial Analysis.
Duprey, T., Lé, M., 2015. Bankscope dataset: getting started. Retrieved from: http://ssrn.com/abstract=2191449.
Cordella, T., Yejati, E. L., 2002. Financial opening, deposit insurance, and risk in a model of banking competition. European Economic Review, 46:471-485.
Friedman, M., Schwartz, A. J., 1963. A monetary history of the United States.
27
Hoaglin, D., Iglewicz, B., 1987. Fine-tuning some resistant rules foroutlier labelling, Journal of the American Statistical Association, 81, No.396.
Hoque, H., Andriosopoulos, D., Andriosopoulos, K., Douady, R., 2015. Bank regulation, risk and return: Evidence from the credit and sovereign debt crises.
Journal of Bnking & Finance, 50:455-474.
Hull, J. C., 2015. Risk Management and Financial Institutions. John Wiley & Sons, Hoboken, New Jersey.
Ioannidou, V. P., Ongena, S., Peydro, J. L., 2009. Monetary policy, risk-taking and pricing: Evidence from a quasi-neutral experiment. CentER Discussion paper. No. 2009-31S.
Jimenez, G., Lopez, J.A., Saurina, J., 2013. How does competition affect bank risk taking? Journal of financial stability, 9:185-195.
Jimenez, G. S., Ongena, J. L., Peydro, J., Saurina, J., 2008. Hazardous times for monetary policy: what do twenty-three million bank loans say about the effect of monetary policy on credit risk taking? Working Paper, Banco de Espana.
Jokipii, T., Milne, a., 2011. Bank capital buffers and risk adjustments decisions. Journal of financial stability, 7:165-178.
Keeley, M. C., 1990. Deposit insurance, risk, and market power in banking.
American Economic Review, 80:1183-1200.
Landier, A., Sraer, D., Thesmar, D., 2013. Banks’ exposure to interest rate risk and the transmission of monetary policy. Working paper No. 18857.
Larrain, B., 2006. Do Banks Affect the Level and Composition of Industrial Volatility? Journal of Finance, 61(4):1897-1925.
Maddaloni, A., Peydró, J. L., 2011. Bank Risk-Taking, Securitization, Supervision, and low Interest Rates: Evidende from the Euro-area and the U.S. Lending Standards. Review of Financial Studies, 24:2121-2165.
28
Martynova, N., Ratnovski, L., Vlahu, R., 2015. Bank Profitability and Risk-Taking. IMF Working paper WP/15/249.
Mikati, N. Z., 2012. Bank Risk Exposure, Bank Failure and Off Balance Sheet Activities: an Empirical Analysis for U.S. Commercial Banks. Finance Meeting
EUROFIDAI-AFFI Paper. Retrieved from:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2080857.
O’brein, R. M., 2007. A Caution Regarding Rules of Thumb for Variance Inflation Factors. Quality & Quantity, 41:673–690.
Rajan, R.G., 2005. Has financial development mad the world riskier? Working paper No. 11728.
Srivastav, A., Hagendorff, J., 2016. Corporate Governance and Bank Risk-taking.
Corporate Governance: An International Review, 24(3): 334–345.
The Financial Times, 2016. In depth: US interest rates. Retrieved from: https://ig.ft.com/sites/when-rates-rise/.
29 Appendix A
