The effect of internal variables on the profitability of the banks in the United States

1

The Effect of Internal Variables on the

Profitability of the Banks in the United States

From 2003 until 2014

Bachelor of Science

Fawad Tajqurishi 10358854

Bsc. Finance & Organization Specialization: Finance

Supervised by I. Sakalauskaite

June 29, 2016

2 Statement of originality

This thesis is written by Fawad Tajqurishi, who takes the full responsibility for the contents of this paper. Hereby, I declare that the work and the text that are presented in this thesis are original. Except the sources that are mentioned, no other sources have been used to create this thesis. The University of Amsterdam, Faculty of Economics and Business, is the only responsible party for the supervision and completion of this thesis. However, the contents are excluded from the scope of the responsibilities of the aforementioned faculty.

3 Abstract

This thesis examines the determinants of profitability of the banks in the United States, from 2003 until 2014. During the given period, there was a financial crisis. Therefore, this thesis examines also the effects of the recent financial crisis on the profitability of the banks. To be more specific, this thesis examines the impact of internal variables on the profitability of the banks during the financial crisis period, which is from 2008 until 2010, and compares it with the non-crisis period, from 2003 until 2007 and 2011 until 2014. Internal variables have a strong or weak effect on the profitability of the banks. The main purpose of this thesis is to examine if the internal variables have the same effect on the profitability of the banks during the crisis as before the crisis. Moreover, this thesis controls for the external variables, which also have an influence on the profitability of the banks. The banks in the United States faced great changes over the last decade, caused by the crisis. The financial crisis, which started in September 2007, had a major impact on the performance and structure of the banks in the United States. Using a sample of 362 banks, I will show that internal variables, namely capital structure, bank size and credit risk, indeed affect the profitability of the banks. All three variables are positive and significant. The statistical results show that during the recent financial crisis, only two of the three internal variables, namely capital structure and bank size, have a significant and weaker effect on the profitability of the banks in the United States. The third internal variable, namely credit risk, has a positive coefficient. However, it is not significant at any significance level. To conclude, the internal variables have a weaker effect on the profitability of the banks in the United States during the recent financial crisis.

4

Contents

1. Introduction

2. Literature review

2.1. The determinants of profitability 6

2.2. Profit of banks 7

2.3. Internal determinants 7

2.3.1. Credit risk 7

2.3.2. Capital structure 8

2.3.3. Size 9

2.4. The external determinants 9

2.4.1. Gross domestic product 10

2.4.2. Inflation rate 10

2.4.3. Interest rate 11

2.5 The complete model 11

3. Econometric Analysis

3.1 Data 12

3.2 The variables 13

3.2.1 The dependent variable 13

3.2.2 The variable of interest 14

3.2.3 The control variables 15

4. Methodology

5. Results

6. Conclusion

7. Bibliography

5

1. Introduction

Banks are financial intermediaries. Banks have a major role in the operations of most economies (Demirguc-Kunt and Huizinga, 1999). According to Demirguc-Kunt and Huizinga (1999), the activities of the banks affect the economic growth and vice versa. The banks in the United States faced great changes over the last decade. The financial crisis, which started in September 2007, had a major influence on the performance and structure of the banks in the United States. The banks in the United States were under significant pressure during the financial crisis. The recent financial crisis was the largest financial crisis in 80 years. The financial crisis began in the sub-prime sector of the securitized United States mortgage market (Carmassi et al, 2009). According to Carmassi et al (2009), the securitization process and the rating agencies failed to spot excessive risk-taking for loans. As a result, the financial sector in the United States collapsed. This also affected other financial markets overseas. Such crises are costly for the banks and the economy. For example, during the sovereign debt crises banks in America, Greece, Ireland, Italy and Spain received a huge amount of funds from their governments to continue with their operations.

Financial crises are not a new phenomenon to the world. For example, countries in Europe had to deal with a financial crisis between the period of 1875 until 1930. Of course this was not the only one. It also happened in South-America (oil-crisis, 1975 until 1985) and Asia (early 2000). Countries like Brazil, Argentina, Venezuela, Hong Kong, Indonesia and Singapore were affected as well. The United States also experienced, for example, financial crisis between the period of 1875 until 1915. Another example is the great depression. It can be concluded that almost every country faced a crisis. Financial institutes are not immune to the effects of a financial crisis. There are different kinds of financial crises. First of all, there is a currency crisis. This is the case when a certain currency is pegged to another currency, most of the time to the American dollar. A currency crisis occurs when there is a speculative attack. Secondly, there is a stock market crash. In this scenario, the prices of stocks fall dramatically and this affects the economy. Finally, there is the banking crisis. The recent financial crisis is an example of a banking crisis. During this period, banks had to perform all kinds of operations to survive from crisis. Hence, analyzing the factors that can help banks better to survive such period is important.

6

banks. They have divided the profitability of the banks into two main categories, namely internal and external variables. External variables are macroeconomic variables, whereas internal variables are specific to each bank. Banks face changes in macroeconomic variables, which affect their profitability. These variables are beyond their own self-control. However, banks can affect the internal variables. As mentioned before, during the crisis, the

operations of the banks became more relevant. These operations can be seen as strategic managerial decisions. Therefore, these decisions have an influence on the internal variables of the banks, which in the long-term affect the profitability of banks.

The method and data of this thesis are slightly different than previous researches. First, the time period is different. The time frame of this thesis will be from 2003 until 2014. Secondly, the number of banks varies. As last, none of the researchers have examined the specific effect of the internal variables on the profitability of the banks, taking the crisis effect into account. The pre-crisis period will be considered from 2003 until end of 2007. The crisis period will be from 2008 until end of 2010. From 2011 until end of 2014 is the post effect of the crisis. Hence, the purpose of this thesis is to examine the contribution of the internal determinants on the profitability of the banks during the recent financial crisis and to control for the external variables. The research question of this thesis is therefore:

‘’Have the internal variables a weaker effect on the profitability of the banks during the financial crisis’’?

In order to be able to answer the research question, first of all, a detailed review will be provided about the existing literature in section 2. In section 3, the econometric analyses used in this thesis will be elaborated. Section 4 will present the results that are obtained from regression analysis and the final section gives a conclusion.

2. Literature review

In this section a detailed review will be given of the existing literature on the determinants.

2.1. Literature on determinants of bank profitability

7

the researchers selected different methodologies to reach the same conclusion. Some authors have done a cross-section analysis and others have focused on the banking system of an individual country. No matter which method was applied, cross-section or panel data, the determinants, which affect the profitability of banks, can be categorized into two main groups (Trujillo-Ponce, 2013 & Demirguc-kunt and Huizinga, 1999), namely internal and external variables. The first group consists of variables that are specific to each bank and are controlled by the management through their decisions. The second group consists of

determinants, which relate profitability to the macroeconomic environment. These variables are beyond the control of the banks and are therefore external variables.

2.2 Measuring profit of bank

According to Demirgiuc-Kunt and Huizinga (1999), the best way to measure the banks’ profitability is to look at the return on assets. Return on assets is calculated by taking the net income divided by total assets. Return on assets is the net profit that is generated by the banks’ total assets. Return on assets also shows management efficiency.

The profitability of the banks could also be measured by looking at the return on equity. Return on equity is calculated by taking the net income divided by equity. According to Nicolae et al. (2015), return on equity measures the net return of capital invested by shareholders. Goddard et al. (2004) concludes that return on assets is a better ratio in comparison to return on equity to determine the banks’ profitability. The reason is that return on assets gives a better overview about banks’ profitability and shows management efficiency. The profitability of the banks can be expressed as a function of internal and external determinants (Dietrich and Wanzenried, 2011).

2.3 Internal variables

2.3.1 Credit risk

Credit risk is one of the important variables that affect banks’ profitability. Credit risk will give an insight into defaults of loans. Credit risk is computed by total loans divided by total assets. Providing loans is one of the major tasks of the banks. Providing more loans will increase the profit of the banks. Providing more loans has also its consequences. According to Rasiah (2010), providing more loans comes alongside risk. The consequence and risk is

8

that banks have to deal with more defaults, which has a negative effect on the profitability of the banks. According to Garcia-Herrero et al. (2009), banks’ profitability will increase by increasing portfolio of loans in comparison to other assets. They have also stated that high operating costs of holding a large portfolio of loans is not an issue for banks’ profitability. The credit risk also increases as the proportion of loans in the portfolio increases. Goddart et al. (2004) also stated that there is direct relationship between the percentage of loans in bank assets and the profitability of the banks. The results of Vong et al. (2009) have shown that there is a negative relation between credit risk and banks’ profitability. During the recent financial crisis, the asymmetric information increased and, as result, credit risk

increased. Banks were more careful in providing loans. So, the effect of loan issuance should be weaker during the crisis. Another reasoning is that banks are in the risk taking business, for which they are rewarded, but during a crisis, these risks become less profitable because of higher default rates. Credit risk might have no effect on effect on the profitability of the banks. Hence, the first hypothesis to be tested is:

Hypothesis 1: Credit risk has a weaker or no effect on the profitability of the banks during the crisis.

2.3.2 Capital structure

Capital structure can be calculated by dividing equity by total assets. The share of equity in bank’s liabilities determines how their assets are financed and how it can cover the losses. Although equity is the most expensive bank liability in terms of expected return (Garcia et al., 2009), according to Demirguc-Kunt and Huizinga (1999), Kosmidou and Pasiouras (2007), Goddard et al. (2004) and Garcia-Hererro et al. (2009), banks who perform the best are the ones who have a higher ratio of equity compared to assets. There are multiple explanations for the positive relation between capital structure and bank profitability: banks with a high level of equity face lower bankruptcy risk as banks with enough capital are able to repay their debt, despite the profit loss. Financing with debt is more risky since a bank would expose itself to credit and liquidity risk (Vong et al., 2009). According to Dietrich and Wanzenriend (2011) and Molyneux and Thornton (2011), this reduces their cost of borrowing because of lower expected bankruptcy costs. It also reduces risk-taking

9

leads to credit and liquidity risk. According to Molyneux and Thornton (1992), banks that have higher equity are able to reduce cost of capital. This can imply a positive effect on the profitability of banks. The results of Molyneux and Thornton (1992) are similar with the results of Demirguc-Kunt and Huizinga (1999). In the crisis period, the relationship between equity and bank profitability could be more important because banks were seeking for equity to increase their profitability but there was a lot of asymmetric information and banks experienced a fall in their assets.

Hypothesis 2: Capital structure has a weaker effect on the profitability of the banks during the crisis.

2.3.3 Size

In the research of Nicolae Petri et al. (2015), banks’ size was computed by taking the natural logarithm of total assets. Bank size is the size of its assets. Natural logarithm can be used to deal with large numbers and to ease analysis, where the effect can be interpreted in terms of percentages rather than units. According to Nicolae Petri et al. (2015), economies of scale will increase, as the size of the banks gets larger. According to Kosmidou and Pasioureas (2007), there is a positive and significant relationship between the size and the profitability of the banks. This is because larger banks have more loan diversification and a higher degree of products than smaller banks. According to Barros et al. (2007), a bank’s size can also have a negative effect on a bank’s profitability. According to them, larger and more diversified banks will perform poorly because of the asymmetric information. Smaller banks can lend more efficiently, which reduces the asymmetric information. Micco et al. (2007) have found no correlation between the size and profitability of the banks. The coefficient of size was positive but not significant. During the crisis, all banks realized a fall in their assets.

Hypothesis 3: Bank size has a weaker effect on the profitability of the banks during the crisis.

2.4 External determinants

As mentioned before, external variables are indirect factors. The management of the banks cannot control these external factors but these factors can have great impact on the

10

factors are the gross domestic product, the interest rate and the inflation rate.

2.4.1 Gross domestic product

According to Shehzad et al. (2013), gross domestic product gives an overview of the economic growth. Their findings have shown that gross domestic product has a positive effect on the profitability of the banks. They have also written that the loan defaults are low when the growth of economy is stable and vice versa. Demand for loans will be higher in case of high economic growth. As result of higher demand, the interest income and non-interest income of banks will increase and this results in a higher profit for the banks. On the other hand, Yong and Christos (2012) have found a significant and negative effect of gross domestic product growth related to the profitability of the banks in China. In their research, they found that high economic growth would improve the business environment and lower the bank entry barriers. As a result, competition will increase and the profitability of the banks will decrease. During the recent financial crisis, the economic growth was not stable and there was no incentive to enter the banking industry.

2.4.2 Inflation rate

According to Demarguc-Kunt and Huizinga (1999), inflation rate can cause variation in the profitability of the banks. They have found that increase in inflation rate causes an increase in the profitability of the banks. Their results have shown that there is a positive relationship between inflation rate and the profitability of the banks, because the income of the banks will increase more with inflation than their costs. High inflation comes together with high loan interest, and as a result, high incomes for the banks. Naceur and Goaied (2001) have not found any significant relationship between the inflation rate and the profitability of the banks in Tunisia. They stated that a further research is needed to emphasize the relationship between inflation rate and the profitability of the banks. Rasiah (2010) has stated that the cost of borrowing will increase as the inflation rate increases. According to Rasiah (2010), a rise in inflation rate will have a negative effect on the profitability of the banks. The reason is that the value of real assets of the banks will decrease comparing to the liabilities of the banks. During the recent financial crisis, the inflation rate was almost zero.

11

2.4.3 Net interest income

According to Aburime (2008), interest rate is an important macroeconomic variable to determine the profitability of the banks. In his research, he founds a significant effect of interest rate on the profitability of the banks. According to Rasiah (2010), in many

researches the interest rate is captured as a profitability determinant of banks because of the net interest income. Net interest income is the difference between the interest income and the interest expense. This difference has an impact on the profitability of the banks. In his research, he described that the interest rate should be taken as an external determinant because the government has a huge roll in interest rate changes. The interest rate is also affected by the market conditions. Moreover, he stated that interest rate changes affect the short term and long term portfolios of the bank. According to him, banks have to improve their revenue sources and cost of funds corresponding to the change.

Demarguc-kunt and Huizinga (1999), Shehzad et al. (2013) and Kosmidou and Pasiouras (2007) have found in their research a positive effect of interest rate on the profitability of the banks.

2.5 Complete model

So far, the theory was given about the determinant for the profitability of the banks. The complete model is provided in the formula (1). The variables that will be used in the regression are listed in figure 1.

YROA = β0 + β1 CSi,t + β2 (CS*Crisis)i,t + β3 Sizei,t + β4(Size*Crisis)i,t + β5 CRi,t + β6 (CR*Crisis)i,t +

β7(NII)i,t + β8 IRt + β9 GDPt + BFEt + εi,t (1)

Where:

εi is the error term

Crisis: is a dummy variable Crisis = 1 for the years 2008 until end of 2010 Crisis = 0 for other periods

12

3. Econometric analysis

In the previous sections the theoretical background and the general model were provided. First, in this section the data that is used in this thesis will be explained in detail. Specifically, the source of the data will be indicated, what kind of data it is and as last, the descriptive statistics will be discussed. Secondly, the dependent and three main explanatory variables will be discussed, namely the internal variables, and as last, the three control variables, which are the external variables, will be discussed.

3.1 Data

In this thesis only data from the United States is used. In order to get a large sample, this thesis uses data from all banks in the United States in the form of panel data. Panel data is used because all information of the banks in the United States are chosen across time. Panel data analysis has an advantage in contrast to time series analysis. Panel data gives a better understanding of the dynamics of adjustments. The main source of these empirical data is Wharton Researches Data Services. The variables that are used in the model are observed yearly. The time period is from 2003 until 2014. This provides twelve observations for each bank. As described before, in this thesis, all banks in the United States are observed as long as the data was available for the given time period. In this thesis, 362 banks are observed in the United States for the given time frame. Hence, the total observation is 4344.

Figure 1. All variables used in this thesis

Year Year [2003-2014]

Country United States

# Banks 362

Return on Assets (ROA) Net profit over return in percentages Capital Structure (CS) Equity over assets

CS * crisis Interaction term CS & crisis Credit Risk (CR) Total loans over total assets CR * crisis Interaction term CR & crisis Ln(Size) Natural logarithm of assets Size * crisis Interaction term ln(Size) & crisis

Net interest income (NII) Interest income minus interest expense

GGDP Growth gross domestic product

Inflation Inflation rate

13

3.2 The variables

Prior to explaining the variables, a list of the used variables in this thesis is already provided in figure 1.

3.2.1 The dependent variable

In this thesis, the return on assets is the dependent variable, which is a measurement for the profitability of the banks. Return on assets is calculated as followed:

Return on assets = (Net income)/(Total assets) (2)

Net income and total assets of all banks in the United States are derived from Wharton Research Data Services for the period 2003 until 2014. The data contains yearly return on assets. From Table 1, one can see that the mean is 0.00579 and the volatility is 0.0148. The research question of this thesis is: ‘’Have the internal variables a weaker effect on the

profitability of the banks during the financial crisis?’’. To get an answer to this question,

graph 1 is presented for return on assets of the banks in the United States for the period of 2003 until 2014. From graph 1, it can be derived that till end of 2007 the return on assets were quite stable. Beginning 2008, when the financial crisis started, there are some extreme peaks, which indicate the effect of the crisis on return on assets of the banks. There are two peaks in 2012. This may be the after-shock of the crisis. It looks like that the return on assets are stable again early 2014. To conclude, the crisis affects the profitability of the banks in the United States.

14

Graph 1.

3.2.2 Internal variables

The formulas for the three internal variables, namely capital structure, credit risk and size, are:

Capital structure = Equity / Assets (3)

Credit risk = (Total loans) / (Total Assets) (4)

Size = Ln(Size) (5)

The internal variables are the main explanatory variables. All internal variables are derived from Wharton Research Data Services. Each variable contains annually data of all banks in

-0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3 0,4 0,5 0,6 2003 2003 2003 2004 2004 2004 2005 2005 2006 2006 2006 2007 2007 2008 2008 2008 2009 2009 2010 2010 2010 2011 2011 2011 2012 2012 2013 2013 2013 2014 2014 ROA (% ) Years

Return on Assets

Variable #Observations Mean Std. Dev. Min Max

Return on Assets 4344 0.0057981 0.0147885 -0.2915129 0.500956 Inflation Rate 4344 0.0232129 0.0109158 -0.0035555 0.038391 Net interest income 4344 0.0007581 0.0043832 -0.0013283 0.0546520 Capital Structure 4344 0.1021234 0.0459049 -0.036 0.7923

Credit Risk 4344 0.6623952 0.1478778 0.0003005 0.9719463

Size 4344 14.01366 1.902549 9.370672 21.90809

15

the United States. Capital structure and credit risk are ratios. Size of banks has been converted into percentages by taking the natural logarithm of size.

3.2.3 Control variables

Gross domestic product is a measurement of the value of all the goods and services that are produced in the United States. Gross domestic product is derived from the database of Federal Reserve Bank of St. Louis, specifically, the source of gross domestic product is the U.S Bureau of Economic Analysis. The gross domestic product contains annually data from the United States. The unit of gross domestic product are in billions of dollars. In this thesis, the growth of gross domestic product has been taken. The growth of gross domestic product is calculated as followed:

Gross domestic product growth = (GDPt+1 - GDPt)/GDPt (6)

When the growth of gross domestic product is positive, it means that economy is stable and this will affect the profitability of the banks in a positive way. The gross domestic product growth was positive until 2008. In 2009 was the growth negative, which is the effect of the crisis. In the correlation matrix, the effect of gross domestic product growth and banks’ profitability is positive, the coefficient is 0.2127. So, gross domestic product and return on assets are linearly correlated, but not perfectly.

The inflation rate is also derived from the database of Federal Reserve Bank of St. Louis. The source of inflation rate is the World Bank. The inflation rate contains annual data from the United States. It contains consumer price index. Inflation rate can be seen as the change in price that consumers pay for goods and services between two periods. The inflation rate for the United States is every year around 2% except for year 2009. The

inflation rate in 2009 was -0.356%. This is because of the crisis. After 2009, the inflation rate is around 0%. When the inflation rate is high, people are careful with purchasing. The purchasing activities will be suspended for a period. As a result, the inflation rate will affect the profitability of the banks. According to table 2, the correlation between inflation rate and return on assets is positive, the coefficient is 0.1119. inflation rate and return on assets are also linearly correlated, but not perfectly.

16

described before, net interest income is the difference between interest income and interest expense. This method is taken on purpose because the interest rate is not only determined by the banks. Demand and supply of consumers, government and the economy affect the interest rate as well. Hence, the net interest income is an in external variable by taking the difference of interest income and interest expense. If a ratio would have been taken, then interest-income and non-interest-income would have become relevant and those variables are internally determined. According to the correlation matrix, the correlation between net interest income and return on assets is 0.007. This correlation coefficient is very low. It means that there is almost no correlation between the net interest income and return on assets.

The multicollinearity problem arises when there is a high correlation between two independent variables. According to Grewal et al.(2004), many researchers ignore the multicollinearity problem because of practical considerations. However, high correlation should be avoided between the independent variables in order to have unbiased

estimations. According to them, two independent variables are highly correlated when the correlation between the two variables are at least 0.80. In such case type II errors, which is a failure to detect a significant effect, are extremely high. Variables that are highly correlated should not be used in the analysis. According to the correlation matrix, none of the

independent variables are highly correlated. Therefore, the multicollinearity problem is not an issue for this analysis.

Table 2

Correlation matrix

ROA IR GDP NII CS CR Size

Return on Assets 1

Inflation Rate 0.1119 1

Gross Domestic Product 0.2127 0.6363 1

Net Interest Income 0.007 -0.0192 -0.0217 1

Capital Structure 0.0616 -0.017 0.0311 -0.0444 1

Cedit Risk 0.0237 0.0482 -0.0249 -0.1634 -0.2338 1

17 4.

Methodology

The research question of this thesis is: ‘’Have the internal variables a weaker effect on the

profitability of the banks during the financial crisis?’’. To enable to answer this question, the

effect of the variables should be estimated with a proper model. Also, the estimation of the model should be free from selection bias. To be able to analyze the effect of internal

variables during the crisis on the profitability of the banks in the United States, in other words the effect of variables that are different over time, the fixed effects model is used. The advantage of the fixed effects model is that it controls for characteristics, which are not observed, hence, it corrects the selections bias. All these characteristics do not change over the time. So, the fixed effects control for bank-specific characteristics, such as business models and culture. Furthermore, the regression analysis that is used in this thesis is a multivariate regression with one dependent variable, three explanatory variables and three control variables. As described before, the regression (formula 1) is as followed:

YROA = β0 + β1 CSi,t + β2 (CS*Crisis)i,t + β3 Sizei,t + β4(Size*Crisis)i,t + β5 CRi,t + β6 (CR*Crisis)i,t + β7

(NII)i,t + β8 IRt + β9 GDPt + BFEt + εi,t

The intercept of the regression is β0. Remaining variables are described in figure 1. A

multivariate regression is used rather than the single regression model. Omitted variable bias will arise when only one independent variable is used. That variable is probably correlated with another variable that is not included in the model. Hence, multivariate regression is used. The F-test can be used to test the significance of the model. The significance of each variable can be tested by the t-test.

5. Results

5.1 Regression output

Table 3 provides the empirical results for the main profitability measure return on assets. The first column of table reports only the effect of the internal variables on return on assets. The R-squared, which explains the sample variance of return on assets, is 4% and only capital structure and credit risk have a significant effect on return on assets. Column two reports the effect of the internal variables and the interaction term between the internal variables

18

and the crisis dummy variable on return on assets. The R-squared is 11.3% and the variables capital structure, credit risk, interaction term capital structure*dummy crisis and

size*dummy crisis have a significant effect on return on assets. Column three reports the effect of the whole regression model on return on assets. The empirical results will be discussed on the basis of this column. The R-squared is 12.2%. It is logical that the R-squared will increase as more variables are added to the model.

The results suggest that the share of equity in banks total liabilities has significantly positive effect on its assets at 0.01% significance level. A 1-percentage point increase in the ratio of equity to assets will increase return on assets of the banks by 7.01%. The coefficient of the interaction term between capital structure and crisis is -0.0332. It means that the positive effect of equity on the banks’ profitability is by 3.32% lower during the crisis period. The interaction term is significant at 1% significance level. The hypothesis was that capital structure has a weaker effect on the profitability of the banks during the crisis. This

hypothesis is consistent with the result. The coefficient of credit risk 0.0349. A 1-percentage point increase in credit risk coefficient will increase the banks’ profitability by 3.49%. Credit risk has a significant effect on the return in assets at 0.01% significance level. The coefficient of the interaction term between credit risk and crisis is positive but not significant, 0.00335. The result show that the effect of credit risk on the profitability of the banks did not change during the crisis period. The hypothesis was that credit risk has a weaker or no effect on the profitability of the banks during the crisis. The hypothesis is consistent with the result. Banks faced defaults in loans during the crisis period. They were more careful in providing loans during the crisis. Also, banks faced a fall in their assets and the risks they took became during the financial crisis less profitable for the banks. Hence, no effect of credit risk on return on assets during the crisis period. The coefficient of Ln(Size) is 0.00196. A 1- percentage point increase in the coefficient of Ln(Size) will increase the bank’s profitability by 0.196%. Besides, the coefficient of Ln(Size) has a significant and positive effect on the profitability of the banks at 1% significance level. The coefficient of the interaction term between Ln(Size) and crisis is -0.0003 and it is significant at 5% significance level. It means that the positive effect of size on bank’s profitability is by 0.03% lower during the crisis period. The hypothesis was that bank size has a weaker effect on the profitability of the banks during the crisis. The hypothesis is consistent with the results. A fall in the assets may be the cause of this effect. Also, it corresponds to the literature. The asymmetric information increased during the

19

crisis. As last, the control variable will be briefly discussed. The coefficient of gross domestic product growth is 0.0934. It means that if the coefficient of gross domestic product increases by 1-percentage point, the return on assets will increase by 9.34%. Gross domestic product is significant at 0.01% significance level. If the economy is in a healthy state, it is obvious that the profitability of the banks will increase. The coefficient of inflation rate is negative and is -0.0449. The coefficient of the inflation rate is not significant. This result is consistent with the result of Naceur and Goaied (2001). They have also found no significant relationship between the inflation rate and the profitability of the banks. The coefficient of net interest income is also negative and is -0.0629. This coefficient has also no significant effect on the profitability of the banks in the United States. This result is contrary to expectations. One would expect that if the net interest income increases it would affect the return on assets in a positive way and not in a negative way. A possible explanation for this unexpected

coefficient can be the error between the coefficient of net interest income and the constant. This reasoning can explain why the constant is significant. Finally, the complete regression model is significant at 0.01% significance level (F=61.30), which means that at least one of the coefficient has an effect on return on assets of the banks in the United States.

20

Table 3 Regression analysis

(1) (2) (3)

ROA ROA ROA

Capital structure 0.0621*** 0.0664*** 0.0701*** (0.00691) (0.00685) (0.00688) Credit risk 0.0318*** 0.0358*** 0.0349*** (0.00291) (0.00286) (0.00288) Ln(Size) -0.000477 0.000885 0.00196** (0.000603) (0.000585) (0.000621)

Capital structure*dummy crisis -0.0331** -0.0332**

(0.0105) (0.0104)

Credit risk*dummy crisis 0.00130 0.00335

(0.00245) (0.00250)

Ln(Size)*dummy crisis -0.000414** -0.000300*

(0.000133) (0.000136)

Gross domestic product 0.0934***

(0.0160)

Inflation -0.0449

(0.0243)

Net interest income -0.0629

(0.162)

Constant -0.0149 -0.0351*** -0.0532***

(0.00874) (0.00850) (0.00916)

FE Yes Yes Yes

N 4344 4344 4344

R-squared 0.040 0.113 0.122

Adj. R-squared -0,048 0.032 0.040

F-statistic 55.45*** 84.82*** 61.30***

21

5.2 Robustness check

To check whether the results, which are obtained, are robust, the data has been trimmed. Trimming is the process of transformation of the data. This process has excluded 1% of the extreme values of the statistical data. It reduces the effect of possible outliers. The results are presented in table 4. Only column 3 of table 4, which present the whole regression model, will be discussed in order to check if the results obtained are robust.

The coefficient of capital structure was 0.0701 in table 3 and has become 0.0588. Both coefficients are significant at a 0.01% significance level. The effect of capital structure is less strong but still significant. The effect of credit risk is also less strong but still significant at 0.01% significance level. The coefficient of Ln(Size) was 0.00196 and has become 0.00144. Moreover, Ln(Size) was significant at 1% level and is now significant at 0.1% level. The interaction term between the coefficient of capital structure and crisis was -0.0332 and has become 0.124. This means that the positive effect of the interaction term was by 12.4% higher during the crisis period. This result might be the cause of getting rid of extreme values, which are caused by the crisis. Moreover, the interaction term has become significant at 0.01% level. The interaction term between the coefficient of credit risk and crisis has become -0.0126, which means that the positive effect of credit risk was lower by 1.26% during the crisis period. This result is consistent with the hypothesis. The interaction term has also become significant at 0.01% level. The interaction term was not significant according to table 3. The coefficient of Ln(Size) was -0.0003 and was significant at 5% level. The coefficient has become now -0.00056 and is significant at 0.1% level. The coefficient of gross domestic product has become less strong but it is still significant at 0.01% level. The coefficient of inflation rate was negative and insignificant. It has become now less negative and significant at 0.01% level. The coefficient of net interest was insignificant and it still is. Moreover, the sample variance of return on assets is now explained by 22.2% and the complete regression model is significant at 0.01% level. Given the results, it can be concluded that the results obtained in table 3 are not robust.

22

Table 4

Robust regression analysis

(1) (2) (3)

ROA ROA ROA

Capital structure 0.0915*** 0.0599*** 0.0588***

(0.00825) (0.00788) (0.00775)

Credit risk 0.0171*** 0.0201*** 0.0203***

(0.00226) (0.00212) (0.00209)

Size (logarithmic scale) -0.000987* 0.0000864 0.00144***

(0.000436) (0.000408) (0.000430)

Capital structure*dummy crisis 0.128*** 0.124***

(0.0144) (0.0141)

Credit risk*dummy crisis -0.0146*** -0.0126***

(0.00267) (0.00263)

Size (log)*dummy crisis -0.000644*** -0.000560***

(0.000152) (0.000150)

Gross domestic product 0.136***

(0.0124)

Inflation -0.155***

(0.0271)

Net interst income -1.198

(0.766)

Constant -0.00107 -0.0135* -0.0345***

(0.00631) (0.00591) (0.00628)

FE Yes Yes Yes

N 3363 3363 3363

R-sq 0.060 0.189 0.222

Adj. R-squared -0.052 0.091 0.128

F-statistic 63.76 116.24*** 95.16***

23

6. Conclusion

The purpose of this thesis was to examine the relationship between banks’ profitability and the internal variables. In order to get an answer, a panel data regression was applied for 362 banks in the United States during the period from 2003 until 2014. This thesis adds value to the existing literature by taking the recent financial crisis into account. Also, the relationship between profitability of banks and internal variables in the United States was not

investigated before.

To answer the research question, first of all, a detailed review of existing literature was given. The results have shown that the internal variables indeed have a weaker effect on the profitability of the banks in the United States during the crisis period. The positive effect of capital structure on the profitability of the banks was during the crisis lower by 3.32%. The positive effect of size on the profitability of the banks was during the crisis lower by 0.03%. According to the results, the effect of credit risk on the profitability of the banks did not change during the crisis period. During the non-crisis period, all internal variables had a positive and significant effect on the profitability of the banks in the United States. Gross domestic product was the only external variable that had an effect on the profitability of the banks. The robustness check showed that the obtained results are not robust.

A considerable remark is that this thesis has limitations. To provide more evidence, more internal and external variables could be included. Industry specific variable could be included as well, using Herfindahl-Hirschman Index. Further research is required to analyze the determinants, which affect the profitability of the banks.

24

7. Bibliography

Aburime, U. T. (2008, September 02). Determinants of bank profitability: Macroeconomic evidence from Nigeria. Retrieved from

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1231064

Acaravci, S. K., & Calim, A. E. (2013). Turkish Banking Sector's Profitability Factors.

International Journal of Economics and Financial Issues, 3(1), 27-41.

Barros, C. P., Ferreira, C., & Williams, J. (2007). Analysing the determinants of performance of best and worst European Banks: A mixed logit approach. Journal of Banking &

Finance, 31, 2189-2203.

Borio, C., Gambacorta, L., & Hofmann, B. (2015, October). The influence of monetary policy on bank profitability. Retrieved from

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2668188

Carmassi, J., Gros, D., & Micossi, S. (2009). The Global Financial Crisis: Causes and Cures.

Journal of Common Markets Studies, 47, 977-996.

Demirguc-Kunt, A., & Huizinga, H. (1999). Determinants of Commercial Banks Interest Margin and Profitability: Some International Evidence. The World Bank Economic

Review, 13(2), 379-408.

Dietrich, A., & Wanzenried, G. (2011). Determinants of bank profitability before and during the crisis: Evidence from Switzerland. Journal of International Financial Markets,

Institutions & Money, 21, 307-327.

Flannery, M. J. (1981). Market Interest Rates and Commercial Bank Profitability: An empirical Investigation. The Journal of Finance, 36(5), 1085-1101.

Garcia-Herrero, A., Gaila, S., & Santabarbara, D. (2009). What explains the low profitability of Chinese banks? Journal of Banking & Finance, 33, 2080-2092.

Goddard, J. A., Molyneux, P., & Wilson, J. (2004). Dynamics of Growth and Profitability in Banking. Journal Money, Credit, and Banking, 36, 1069-1090.

Grewal, R., Cote, J. A., & Baumgartner, H. (2004). Multicollinearity and Measurement Error in Structural Equation Models: Implications for Theory Testing. Marketing Science,

23(4), 519-529.

Jaffee, D., Lynch, A., Richardson, M., & Van Nieuwerburgh, S. (2009). Mortage Origination and Securitization in the Financial Crisis. In B. Salomon, & S. Wiley (Eds.), Mortage

Origination and Securitization in the Financial Crisis (pp. 141-143). New York,

Amerika: Periodicals.

Kosmidou, K., & Pasiouras, F. (2007). Factors influencing the profitability of domestic and foreign commercial banks in the European Union. Research in International Business

25

Molyneux, P., & Thornton, J. (1992). Determinants of European bank profitability: A note.

Journal of Banking & Finance, 16, 1173-1178.

Naceur, S. B., & Goaied, M. (2001). The determinants of the Tunisian deposits banks' performance. Applied Financial Economics, 11(3), 317-319.

Rasiah, D. (2010). Theoretical Framework of Profitability as Applied to Commercial Banks in Malaysia. European Journal of Economics, Finance and Administrative Sciences, 1(19), 74-93.

Saha, A., Subramanian, V., Basu, S., & Mishra, A. K. (2009). Networth exposure to interest rate risk: An empirical analysis of Indian commercial banks. European Journal of

Operational Research, 193, 581-590.

Shehzad, C. T., De Haan, J., & Scholtens, B. (2013). The relationship between size, growth and profitability of commercial banks. Applied Economics, 45(13), 1751-1765. Trujillo-Ponce, A. (2013). What determines the profitability of banks? Accounting and

Finance, 53, 561-586.

Vong, A. P. I., & Chan, H. S. (2006). Determinants of Bank Profitability in Macao.