Impact of the Great Financial Crisis of 2007-09 on European Bank business models

(1)

Impact of the Great Financial Crisis of 2007-09 on European

Bank business models.

Master Thesis

Master in International Finance

University of Amsterdam, Amsterdam Business School By: Thijs Luijf (10684522)

Supervisor: dhr. prof. dr. A.W.A. Boot Date: October 2016

(2)

Abstract

This study examines the effect of the Great Financial Crisis (GFC) of 2007-09 on the diversity of business models of banks located in the European Union. The diversity between business models is measured as the distance between a single bank and its peers for each of the components of its business model. A lower distance is interpreted as greater homogeneity between banks. A change in this distance is measured between the period before the GFC (2005-2007) and after (2012-2015). To determine the significance of the change in homogeneity since the GFC, a control period (1998-2000) is included in the study.

Results show that banks have changed their business models since the GFC, as the composition of assets, funding, and income has changed for the average bank since that period. Against expectations of this study, it was found that only the funding structure of banks’ business models has become more homogeneous since the GFC compared to the period before the GFC. This was found despite the fact that the funding structure has not become more homogeneous during the period after the GFC itself, but as concluded above only compared to the period before. Other business model structures, income, assets, and funding, have not become more homogeneous over the same period.

Although the study did not completely confirm the expected outcomes, it does provide input for future research. As not much research has been performed on the topic of this study, with the same scope, it offers researchers a starting point on the subject of homogeneity within the banking sector.

(3)

1. Rationale and aim of research

1.1. Background

Since the Great Financial Crisis (GFC) of 2007-09, the environment in which banks operate has changed drastically. As the global economy went into a recession which lasted years, governments and central banks intervened to combat the economic downturn. Failing banks were bailed-out or even nationalised by governments, while central banks provided additional liquidity in order to increase credit facilities. These interventions resulted in an increase in supply of money, shifting the yield curve down along the whole range. At the same time, the conduct of international banks and the consequences of this demeanour showed that more stringent banking regulation was required. In Europe, banking regulation profoundly changed when Basel III enter into effect under the Capital Requirements Regulation (CRR)1_{and the Capital Requirements Directive IV (CRD}

IV)2_{in July 2013. These events affected the monetary and regulatory environment of the global}

banking sector.

The lower yields on government bonds and central bank facilities are the clearest evidence of the change in monetary environment. A common banking activity is term transformation, by which a bank funds itself through short-term deposits for low interest rates and lends that money to customers for a longer term and higher interest rate. As the yield curve shifted downwards in Europe and the United States of America (US) as a response to the GFC, it also flattened, leaving the interest spread earned by banks lower. Claessens, Coleman, & Donnelly (2016) argue that persistently low interest rates may depress the profitability of banks as they are reluctant to fully price the low, or even negative, interest rate into their deposit products. The lower profitability from these activities might put pressure on banks’ management boards to find activities or investments which generate higher returns.

1_{Regulation (EU) No 575/2013 of the European Parliament and of the Council of 26 June, 2013 on prudential} requirements for credit institutions and investment firms and amending Regulation (EU) No 648/2012, OJ L 176/1 27.6.2013.

2_{Directive 2013/36/EU of the European Parliament and of the Council of 26 June 2013 on access to the activity of} credit institutions and the prudential supervision of credit institutions and investment firms, amending Directive 2002/87/EC and repealing Directives 2006/48/EC and 2006/49/EC, OJ L 176, 27.6.2013.

(5)

2 Under CRR and CRD IV, capital planning changed completely for European banks. The capital definitions for banks have become more restrictive, resulting in less available capital3. Some capital types (preferred stock) were prohibited from counting towards capital requirement ratios. Other deductions (deferred tax assets and minority interest) reduced the contribution of remaining capital towards the requirements. While Total Capital requirements remained at 8% of Total Risk Exposure Amount (TREA), or Risk-Weighted Assets, the required composition was altered. A new capital component, Common Equity Tier 1 (CET1), was introduced and the minimum required ratio was set at 4.5%, while the Tier 1 (T1) capital component (including CET1) increased from 4% to 6%. The remainder of 2% can be held in Tier 2. Additionally, banks are required to maintain buffers on top of the minimum required capital ratios in the form of a capital conservation buffer (CCB), a systemic risk buffer, a countercyclical capital buffer, and an additional buffer for systemically important banks. The CCB is applicable for all banks and is phased in between 2015 (0%) and 2019 (2.5%) in steps of 0.625% CET1. The countercyclical capital buffer can be imposed on banks by the National Competent Authorities (National prudential banking supervisors, NCAs) by own discretion and can range between 0% and 2.5% CET1. The other two buffers depend on the systemic relevance of the individual banks. As available capital decreased, required capital increased, and the allowed capital composition changed, weights attributed to certain asset classes, such as securitisations, increased. Resulting in an increased TREA for most banks, requiring those banks to hold more capital in order to meet the minimum capital ratios.

On top of the capital requirements, additional prudential supervision measures are imposed on banks. Banks are required to have a minimum leverage ratio of 3%. Where capital ratios are calculated using TREA, the leverage ratio is expressed as the ratio between Tier 1 capital, and total assets and off-balance exposures. The liquidity coverage ratio (LCR) measures the available High Quality Liquid Assets (HQLA) of a bank compared to the net cash-flows expected during a 30-day stress period. The requirement is being phased in between 2015 (60%) and 2019 (100%). NCAs are able to impose stricter requirements for an individual bank on all measures (including capital, leverage ratio, and LCR) when the risks run by the individual bank are not sufficiently covered by the minimum requirements.

(6)

3 As the environment changed, certain activities should have become less attractive for banks as such activities generate lower net interest margins and/or higher capital requirements through higher asset weightings. Having to comply with higher capital requirements decreases the profitability of banks as capital is typically more expensive than other funding sources, in particular due to the tax advantages from higher leverage, the leverage indifference theorem of Modigliani and Miller (1958) does not apply in the situation of banks. Moreover, the opaqueness of banks generally leads to additional costs for raising equity (Myers & Majluf, 1984). Also a debt overhang effect is in play. Once capital is low, raising capital gives a windfall gain to debt holders, and this further reduces banks’ incentives to raise equity. While countervailing effects might be present, higher leverage may then decrease their weighted average cost of capital (Greenbaum, Thakor, & Boot, 2015). Assuming all commercial banks are looking to maximise shareholder value, such effective ‘scarcity’ of capital may then induce banks to discontinue lower return activities. As a result, the diversity between business models of European banks should decrease.

1.2. Research question

Considering the changes in the banking environment, the following questions will be the key-research question in this study:

In order to answer the key-research question, the following research questions need to be answered first:

1) How have balance sheets of banks changed since the GFC? 2) How have income sources of banks changed since the GFC?

3) How have individual bank’s balance sheets changed compared to their peers’ since the GFC?

4) How have income sources of individual banks changed compared to their peers’ since the GFC?

5) Is the development of business models of banks compared to their peers’ since the GFC different from the development since the end of 2000?

The period for which these questions are to be answered are pre-GFC (2005 to 2007) and post-GFC (2012 to 2015). The post-post-GFC period is one year longer to accommodate for data not being

“In what way have the distances between business models of banks changed since the GFC?”

(7)

4 available for 2015 in some situations, see more in section 4. Data. In research question 5, any change found to be significant in answering research question 3 and 4 is tested against the period 1998 to 2000 (T0), leaving the same amount of time between measured periods.

Answering the key-research question can contribute to the prudential supervision on the banking sector. More homogeneous business models of banks result in increased competition between banks for business in those activities. Schaeck, Cihak & Wolfe (2009) argue that increased competition between banks decreases the chance on systemic crises. However, decreased chance on a systemic crisis does not imply a chance of zero percent. As more banks would be undertaking the same less-risky activities, the effect on the economy when the risk related to those activities would materialise could be enormous as a large number of banks would be affected. The outcome of this study can potentially inform NCAs about specific business model developments which cause a more homogeneous banking sector.

1.3. Outline

Following this introduction is firstly, a literature review covering business models of banks, how recent literature has measured those, and how distances between business models are measured. This literature review forms the theoretical framework for this study. Secondly, the methodology and hypotheses are discussed. Thirdly, the description of the data is presented. Fourthly, the results of the performed tests will be analysed and interpreted. Finally, this study ends with a conclusion and recommendations towards future research.

2. Literature review

This section forms the theoretical framework of this study. At first, an overview is provided of existing research into the effect of changes in the banking environment on the business models of banks. Secondly, an overview is provided of financial variables used in existing literature in order to define business model of banks. Finally, quantitative methods used to carry out the analyses needed to answer the research questions of this study are evaluated.

Research like this study has not been performed yet with the same scope, leaving this study with no benchmark of earlier outcomes or tested guidelines of how to perform such a study. Earlier research has however been conducted on the effects of changes in banking regulation on the business models of individual banks. These studies, amongst others, are discussed below.

(8)

5

2.1. Changes in the banking environment

In most developed countries, the banking sector has been subject to a number of regulatory changes over the past decades. Most of these changes have been deregulation, or liberalization, of the sector and not more stringent regulation, therefore most literature is focused on this aspect of a changing banking environment.

In the early 1980s, US Congress passed acts deregulating interest costs, thereby increasing funding costs for banks. Humphrey and Pulley (1997) show that banks initiated three adjustments following this cost-increasing regulatory change, namely cost offset and reduction, cost shifting, and revenue augmentation. The last two adjustments are of interest for this study as they involve changes in the focus of banks. Cost shifting was achieved through offering more floating-rate loans to borrowers and selling securitised assets to investors. The augmentation of revenue was achieved by lending to a riskier and concentrated set of borrowers, this generated a greater expected return. A concrete example of deregulation altering the business models of banks is the adoption of the Gramm-Leach-Bliley act, in 1999. This act abolished the separation between investment and commercial banks in the US, thereby allowing the existence of universal banks undertaking a range of banking activities under a holding company (Komai & Richardson, 2011).

Multiple studies (see Fischer & Chénard, 1997, and Caprio & Klingebiel, 1997) find that following liberalization of financial markets, financial crises become more frequent. Especially Fischer & Chénard (1997) find that banks show typical symptoms following deregulation of the system, these symptoms include: growth of banking assets and rapid quality-deterioration of the loan portfolio. Hellmann, Murdock, & Stiglitz (2000) argue that the liberalization of the financial market lowers profits and the franchise values of banks, which in effect lowers the incentive for making good loans.

Financial and technical innovation can drive changes in the banking environment, and business models of banks, as well. Merton (1995) argues that the increased use of derivatives has affected the activities of banks and future innovations will continue to change the structure and activities of banks. Boot & Thakor (2014) show that the banking system is becoming more and more intertwined with financial markets. As a result, banks are better able to hedge risks using the markets, but also have the option of securitization of assets in order to offload those to investors. Both options increase the non-interest income generated by banks, altering their business models.

(9)

6 Ultimately, Claessens & Kodres (2014) name both financial innovation and financial liberalization as common causes for crises. Those factors drive banks to develop and use new financial products of which the risks are not fully understood and remove barriers due to which banks were unable to engage in certain activities.

Changing minimum capital requirements for banks has different effects on risk taking of individual banks. Greatly undercapitalized banks have the incentive to maximize risk taking after minimum capital requirements are increased, as there is little capital to lose in the case of a bankruptcy (in most countries depositors are insured up to a certain value) and those banks have everything to gain from the upside potential of risky investments. This moral hazard becomes less for banks which are moderately undercapitalized, as those banks, and their shareholders, benefit more from not losing any capital available to the bank. Banks that are marginally overcapitalized show the same behavior. Banks that are greatly overcapitalized have the tendency to increase risk taking, which does not have an adverse effect on moral hazard. The probability of default of those banks is low, therefore these banks are optimizing their returns while not increasing the risk for deposit holders. This is found by Calem & Rob (1999) and they describe it as a U-shaped relationship between capital and risk taking of banks. According to their findings, increasing capital of greatly undercapitalized banks reduces risk in those banks as their management will shift from maximizing risk to minimizing risk in order to stay solvent.

Changes to business models of banks need not come from monetary or regulatory changes, they can also come from within the sector. Ghemawat (1991) identified four threats to the sustainability of the business model of a company in general: imitation, hold up, slack, and substitution.

Imitation is the risk that competitors try to copy a successful business model and make it their own, taking away customers from the source they are imitating. Components of business models which require certain reputation, experience, or privileged relationships cannot simply be replicated by competitors. For banks, this risk is higher for taking on retail deposits and lending than for activities like private banking, investment services, or mergers and acquisitions. The former requiring less expertise from a bank than the latter. Imitation would lead to greater homogeneity amongst a selected group of banks, while causing greater heterogeneity between the imitating bank(s) and the original peers. It could pose an opportunity for the imitated bank to start new activities.

(10)

7 Substitution is the risk that a new product is introduced in the market which offers customers the same benefits as the existing product in a different way. For example, traveling by planes has been a great substitution for traveling by trains. In the banking sector, the Fintech revolution is creating a number of substitution services for the traditional banks. Dapp (2014) of Deutsche Bank writes that this movement has given rise to a new competitive environment for banks, due to new players offering “digital payment solutions and information services, savings and deposit-taking right through to modern online banking, multi-channel advisory and securities trading services as well as simple financing solutions and the use of compatible financial software.” (Dapp, 2014). Substitution should cause greater homogeneity as new business activities are undertaken within the sector, whilst still servicing the same group of customers.

Hold up is the risk that a corporation becomes too dependent on existing customers or business, making it increasingly difficult to enter into new activities while exiting the existing ones. Hold up is therefore a limiting factor for banks who look to expand and/or replace activities.

Slack is corporate complacency, which when identified by management should be countered by the right incentives or a change in strategy. A change in strategy would require the bank to undertake new activities.

Imitation and substitution are risks that arise from external sources, while hold up and slack come mostly from within the company. Imitation, substitution, and slack are drivers of a change in business models of banks over time without need for external monetary or regulatory changes, although these could directly and indirectly effect imitation and substitution. Hold up is a driver of static business models over time, which is also directly and indirectly influenced by monetary and regulatory changes.

2.2. Identifying business models of banks

The aim of this study is to determine how banks’ business models have changed compared to their peers since the GFC. As discussed earlier, in order to test this, first it has to be established that banks business models have changed individually during the same period. Therefore, this section focuses on what variables should be used to measure business models of banks, rather than how to define different types of business models. A full list of types and a definition of the bank type can be found in appendix (Table A.1, p. 27). The fourth section of this study (4. Data) covers which types of banks have been included in the analyses performed in this study.

(11)

8 The business model of banks has been covered plentifully in recent literature. Multiple studies show that business models of banks can be defined by any of the four following components: capital structure, asset structure, funding structure, and income structure (see Altunbas, Manganelli, & Marques-Ibanez, 2011; Martel, van Rixtel, & González Mota, 2012; and Ayadi, et al., 2016). The first three of these are balance sheet components and are covered in this study by research questions 1 and 3, while the last one is covered by research question 2 and 4.

As the regulatory capital requirements changed for banks within the European Union due to the introduction of the CRR and CRD IV (as explained in 1. Rationale and aim of research), the capital structure has changed for banks by definition. Although this is the situation, the new regulation also adjusted the definition of some capital components (T1 capital) and introduced (CET 1), making the comparison of capital structure between pre-CRR and post-CRR situations more difficult. Ayadi, et al. (2016) take a broader approach for their study and look at regulatory & supervisory indicators. Due to the changes in the regulations, the regulatory environment has become more restrictive for banks.

The asset structure of a bank can be defined in three main categories, namely loans, securities, and cash and equivalents (Greenbaum, et al., 2015). The loans category can be split by counterparty into two types, consumer loans, and commercial and industrial loans. Consumer loans consist of residential mortgage loans, and consumption credit. A loan in this last category is typically for the purchase of a durable good. Commercial loans are granted to businesses for daily operations (working capital loan), financing of a specific purchase (transactional loan), to invest in fixed assets (term loan), or a combination of these. Business can be granted mortgages as well in the form of construction or commercial mortgages. Securities encompass government bonds, corporate bonds and shares, and derivatives. Usually, most government bonds and some high-quality corporate securities are held for liquidity purposes as they qualify as HQLA (see chapter 15 of Greenbaum, et al. (2015)). Otherwise, securities can be held as investments generating dividend or interest income for the bank. Derivatives are normally held for the purpose of hedging risks of other assets or liabilities. Cash and equivalents consists of coins and banknotes, and central bank reserves.

The funding structure consists of the remaining balance sheet part, the liabilities. For most banks, the largest source of funding is deposits. These deposits come in several forms and can be categorized into two groups: demand deposits, and saving or term deposits. Other sources of

(12)

9 funding are repurchase agreements (repos); in which the bank sells securities to a counterparty for cash under the agreement to repurchase the securities for the cash amount plus interest, debt; both debt securities and borrowings, and trading liabilities.

The income structure can be split in interest income and non-interest income. The first is generated by activities that involve loans and investments in debt securities. Non-interest income is a broader definition. It can contain dividend income from investments in equity, fees and commissions, and other operating income.

The composition of these structures is different for each bank business model. Ayadi, et al. (2016) show that the five bank business models found in their study (focused retail, diversified retail (Type 1), diversified retail (Type 2), wholesale, and investment) have very different structures. For example, in the asset structure loans to banks as a percentage of total assets ranges from 6.6% to 52.2%. Similar differences in structure composition were found for the funding structure, the income structure, and the regulatory & supervisory indicators.

Section 4. Data covers which variables have been included in the study in order to determine the different compositions of each structure. These variables are further referred to as business model components.

3. Methodology and hypothesis

The main steps of this study are the following: determine sample of banks, selection of variables for each business model structure, run test procedures for research question 1 and 2, calculate pair wise distance for each individual bank compared to all other banks, run test procedures for research question 3 and 4, run test procedures for research question 5, and finally present empirical results, interpretation of outcome, and conclusion. The first two steps are covered in detail in the next section (4. Data).

In order to answer research questions 1 and 2, the difference between business models of banks pre-GFC and post-GFC is calculated and tested for significance. This is done by means of a classical dependent, paired sample, t-test. The tested means are taken over the full sample of banks for each business model component to ensure that each small change in business model composition is measured. Let 𝑑̅̅̅ be the mean difference between pre-GFC and post-GFC periods 𝑏

(13)

10 selected), where research question 1 focuses on the asset, funding, and capital structure components and research question 2 focuses on the income structure of the banks’ business model.

The null hypothesis is that 𝑑̅̅̅ for all business model components is equal to zero, so that no 𝑏

significant change in business models has been detected between the two periods. The alternative hypothesis is that 𝑑̅̅̅ for at least one business model component is not equal to 0, so that there has _𝑏 been at least one business model component which has significantly changed between both periods.

𝐻₀1,2: 𝑑̅̅̅̅ = 𝑑_𝑏₁ ̅̅̅̅ = ⋯ = 𝑑_𝑏₂ ̅̅̅̅̅ = 0; _𝑏_𝐵 𝐻_𝑎1,2: 𝑑̅̅̅̅ ≠ 𝑑_𝑏₁ ̅̅̅̅ ≠ ⋯ ≠ 𝑑_𝑏₂ ̅̅̅̅̅ ≠ 0; _𝑏_𝐵

( 1 )

Formula (1) contains the null and alternative hypotheses for both research question 1 and 2, therefore they have a superscript 1,2.

The expectation is that for each business model structure a number of components show a significant change since pre-GFC. As the income structure is a result of the assets (income) and funding (expenses) structures of a business model, it is expected that of the four structures most components changed within the income structure. Changes in the monetary environment (lower interest) should result in a shift towards non-interest income, these changes are expected to have an effect on the asset structure of banks as they shift towards non-interest earning assets. At the same time, interest expense should have decreased since pre-GFC as interest-bearing funding has become cheaper.

Regulatory changes are expected to mostly affect the assets and capital structure. Assets due to the fact that the risk weight applied to some asset classes has been changed. Capital is affected by the requirement to hold more, and different types of, capital. However, this last part is hard to measure as the definition of capital types has changed.

In order to answer research question 3 and 4, the pair wise distance (PWD) between each individual bank and its peers is calculated for each business model structure. The mean of PWD is evaluated to have changed significantly or between the pre-GFC and post-GFC periods. Cai, Saunders, and Steffen (2011) use the PWD method in their study to determine the interconnectedness of banks in the syndication of loans in the US. The formula used in their study has been adjusted to fit the goal of this study.

(14)

11 Let 𝑤𝑖,𝑠,𝑐,𝑡 be the weight of each business model component 𝑐 within its respective business

model structure (asset, capital, funding, and income) 𝑠 for each bank 𝑝. Note that for all pairs of 𝑝, 𝑡, and 𝑠, ∑𝐶 𝑤_{𝑝,𝑠,𝑐,𝑡} = 1

𝑐=1 , where 𝐶 is the amount of business model components of the relevant

business model structure. The business model components are specified in the next section of this study.

The distance in this C-dimensional space between banks 𝑝 and 𝑞 (𝑝  𝑞) in period 𝑡 for business model structure 𝑠 is then computed using the following formula, which is an adaption of the formula used by Cai, et al (2011):

𝐷_{𝑝,𝑞,𝑠,𝑡} = √∑(𝑤_{𝑝,𝑠,𝑐,𝑡}− 𝑤_{𝑞,𝑠,𝑐,𝑡})2

𝐶

𝑐=1

( 2 )

The measure 𝐷_{𝑝,𝑞,𝑠,𝑡} has a lower limit of zero and an upper limit of √2 by construction. A lower 𝐷_{𝑝,𝑞,𝑠,𝑡} indicates that banks 𝑝 and 𝑞 are closer to each other compared to a higher value of 𝐷_{𝑝,𝑞,𝑠,𝑡}. This means that a significantly lower average 𝐷𝑠,𝑡 is an indication that business models of banks

have become more homogeneous on that business model structure. This average distance per bank 𝑝 in period 𝑡 for business model structure 𝑠 can be calculated using the following formula:

𝐴𝑣𝑒𝐷𝑝,𝑠,𝑡 = (∑ 𝐷𝑝,𝑞,𝑠,𝑡 𝑄 𝑞≠𝑝 ) ×1 𝑄 ( 3 )

Where 𝑄 is the number of banks not equal to 𝑝 and 𝐷𝑝,𝑞,𝑠,𝑡 is the distance calculated in (2) per

pair of banks 𝑝 and 𝑞, for all pairs where 𝑝  q. The result of (3), 𝐴𝑣𝑒𝐷_{𝑝,𝑠,𝑡}, is then used to calculate the mean difference between pre-GFC and post-GFC business model structures. Let this mean difference be denoted by 𝑑̅̅̅, for each business model structure 𝑠. Using a dependent, paired 𝑠

sample, t-test, the difference for each business model structure is evaluated to be significant or not. Due to the fact that the regulatory environment has become more restrictive for banks, the expectation is that banks are left with less freedom in their activities and as a result see their business models become more homogeneous with other banks. Greater homogeneity is indicated by a smaller PWD post-GFC compared to pre-GFC. Therefore, the null hypothesis for research questions 3 and 4 is that 𝑑̅̅̅ is smaller than or equal to than zero for all four business model 𝑠

(15)

12 structures. The alternative hypothesis is that 𝑑̅̅̅ is greater than zero for at least one business model 𝑠

structure.

𝐻₀3,4: 𝑎𝑙𝑙 𝑜𝑓 (𝑑̅̅̅̅, 𝑑𝑠1 ̅̅̅̅, 𝑑𝑠2 ̅̅̅̅, 𝑑𝑠3 ̅̅̅̅) ≤ 0; 𝑠4

𝐻𝑎3,4: 𝑎𝑡 𝑙𝑒𝑎𝑠𝑡 𝑜𝑛𝑒 𝑜𝑓 (𝑑̅̅̅̅, 𝑑𝑠1 ̅̅̅̅, 𝑑𝑠2 ̅̅̅̅, 𝑑𝑠3 ̅̅̅̅) > 0; 𝑠4

( 4 )

As described in 2. Literature review, business models of companies change due to different factors. In order to determine whether a potential change found in 𝐻₀3,4 is due to regular developments in the banking sector, or due to a specific event (possibly changing monetary or regulatory environments), another period should be tested as benchmark. Therefore, the same test is carried out between T0 and pre-GFC to determine whether the average PWD between banks’ business model structures have decreased during that period. The hypotheses for this test are the same as presented in (4), only the periods are different.

Subsequently, in order to answer research question 5, the average distance per bank 𝑝 is calculated for 𝑡 = 𝑇0 using the same method as described above in (2) and (3). The difference between 𝐴𝑣𝑒𝐷𝑝,𝑠,𝑇0 and, the earlier calculated, 𝐴𝑣𝑒𝐷𝑝,𝑠,𝑝𝑟𝑒−𝐺𝐹𝐶 (let this be denoted as 𝐷𝑃𝑟𝑒𝑝,𝑠),

and 𝐴𝑣𝑒𝐷_{𝑝,𝑠,𝑝𝑟𝑒−𝐺𝐹𝐶} and 𝐴𝑣𝑒𝐷_{𝑝,𝑠,𝑝𝑜𝑠𝑡−𝐺𝐹𝐶} (let this be denoted as 𝐷𝑃𝑜𝑠𝑡_𝑝,𝑠) are then calculated for each bank 𝑝 and business model structure 𝑠. The mean difference, 𝑑̌, is then calculated for each 𝑠

business model structure 𝑠. Using a dependent, paired sample, t-test, the difference between 𝐷𝑃𝑟𝑒_𝑝,𝑠 and 𝐷𝑃𝑜𝑠𝑡_𝑝,𝑠 is evaluated to be significant or not. The expectation is that the change in the banking environment between pre-GFC and post-GFC affects the PWD more than any possible change between T0 and pre-GFC, therefore the difference is expected to be greater than zero. Hence, the hypotheses for this test are as follows.

𝐻₀5_{: 𝑎𝑙𝑙 𝑜𝑓 (𝑑} 𝑠1 ̌ , 𝑑̌ , 𝑑𝑠2 ̌ , 𝑑𝑠3 ̌ ) ≤ 0; 𝑠4 𝐻𝑎5: 𝑎𝑡 𝑙𝑒𝑎𝑠𝑡 𝑜𝑛𝑒 𝑜𝑓 (𝑑̌ , 𝑑𝑠1 ̌ , 𝑑𝑠2 ̌ , 𝑑𝑠3 ̌ ) > 0; 𝑠4 ( 5 ) 3.1. Limitations

The above presented methodology has certain limitations, these are touched upon in this section so that the reader can read and interpret the results with knowledge of these limitations.

Firstly, there is the time limitation. As discussed in 2. Literature review, business models within a sector change over time due to external and internal factors. This study focuses on the external

(16)

13 factors (monetary and regulatory environment), while controlling for the development over time that is caused by internal factors. This is done by performing the analyses on the changes seen before the GFC and the differences between PWD measured between T0 and GFC, and pre-GFC and post-pre-GFC. Nonetheless, this period is possibly, and likely, also biased for external changes to the banking environment and therefore not presented a fully unbiased development of business models over time. Since 1998-2000, the world has witnessed the IT-bubble burst and has undergone dramatic change after the horrific events of 11 September 2001.

Secondly, this study aims to capture the effect of combined changes in the banking environment on the banks’ business models and how they have evolved compared to their peers’. This is captured through the proposed methodology, thereby it does not directly measure the effect of changes in the banking environment caused by monetary policies or regulation on the business models of banks. This applies as well for any correlation between or causal effect of these changes and banks’ business models. In order to capture these correlations and/or causal effects, a study could select specific regulatory or monetary changes and control for those in a regression analysis. This issue is elaborated on in 7. Follow up research.

4. Data

This section describes all data related steps of this study. Starting with a description of the source of the data. Secondly, a description of how the sample of banks has been selected. Thirdly, the selection of data points (business model components) ensuring that each business model structure is properly represented in the selection. Fourthly, what manipulations have been carried out in order to prepare the data for analyses. Lastly, the descriptive statistics of the data used in the study is presented.

4.1. Source of data

The data used in this study has been retrieved from Bankscope. This database offers financial data of banks in a variety of formats. One of these formats is the Fitch Universal Format4, which was developed by Fitch Ratings in order to assist with comparisons across different accounting standards around the world. Using this data allows for uniform use of all available data without any additional manual manipulations, filters, or mapping of data points.

(17)

14 It is important to note that it is not mandatory for banks to report financial data to Bankscope and if they report, they can stop doing so without any notification.

4.2. Sample of banks

As implied by the title of this study, the scope of this study encompasses banks located in Europe. Within Europe the focus will be on the European Union as the regulatory environmental changes affecting banks business models can be aligned with the implementation of the CRR and CRD IV. Based on this selection, Bankscope’s Universal Format database contains data for 8,388 banks, with a total balance sheet (last available year) of over € 100,000 billion. Table 4.1 contains all information on the composition of the sample and the effect each selection criterion had on the sample size.

Banks have been selected for the research sample based on the following four criteria. The first criterion is already mentioned above, banks located in the European Union. The second criterion is the type of business model, as identified by Bankscope. Certain types of banks are excluded, amongst others: central banks, micro-financing companies, bank holding companies, and group-financing companies, due to their unique non-banking activities. A description of all business models used by Bankscope can be found in the appendix (Table A.1, p. 27). The types of banks that are included in the sample can be found in Table 4.1. This filtered 859 banks from the sample. The third criterion is balance sheet size. Bank with total assets of over € 1 billion in the last available year of reporting have been included in the sample. This decreased the sample by 4,773 banks.

The last criterion is availability of data, at least two years have to be available pre-GFC and post-GFC, and at least one year in T0; this criterion is included in order to make sure the bank was active and has reported data during all periods. The post-GFC period is a year longer as some banks had not released data for 2015 during this study. In order to be able to include those banks in the study the period includes 2012 as well. As T0 is not the main period of interest in this study and is only used as benchmark, it is deemed sufficient when a bank only has data available for one year for this period.

These sampling criteria resulted in a final sample of 778 banks with combined assets of over € 6,267 billion. Table 4.1 shows the number of banks included per type and the proportion of assets account for by each type of bank.

(18)

15

Table 4.1: Frequency and asset of banks in sample used in study

Bank type Frequency Total Assets (€ million) Asset Share

Savings Banks 276 1,106,155 1.04%

Commercial Banks 264 2,524,557 2.37%

Cooperative Banks 152 447,646 0.42%

Real Estate & Mortgage Banks 45 647,479 0.63%

Finance Companies 17 114,684 0.11%

Private Banking & Asset Mgmt 13 78,959 0.07%

Investment Banks 11 1,320,237 1.21%

Sample total 778 6,266,717 5.88%

Filtered availability of data 1,978 68,546,803 64.27% Selected balance sheet 2,756 74,813,520 70.15% Filtered balance sheet 4,773 1,435,787 4.35% Selected business models 7,529 76,249,307 74.50% Filtered business models 859 30,399,001 25.50%

Bankscope EU banks total 8,388 106,648,308 100.00%

4.3. Selection of data points

The Universal bank format of Fitch and Bankscope has a total of 117 variables that can be divided into assets, liabilities, capital, and income. The full list of variables is included in the appendix (Table A.2, p. 29), this table displays the business model structure each component is part of and has an indicator of whether the component is included in the study. Of these 117 variables, 86 are included in the analyses performed in this study. The variables used to calculate ratios (total assets, total liabilities, and total capital) are the first variables that were excluded, as those ratios would always be 100%. Any subtotals or net variables are also excluded as they are a function of at least two or more other variables. The selected variables are chosen to generate a view of the business model through the balance sheet and income statement.

4.4. Data manipulation

Before the data could be used in the analyses of this study, some manipulations were performed on the data. This section describes all manipulations done on the data.

In order to compare the data of banks of different balance sheet sizes and income levels, the data was recalculated as a ratio of its respective total. Total assets, total liabilities, total capital, and the sum of gross interest income and total non-interest operating income (total operating income) were used for all data points of, respectively, the asset structure, funding structure, capital structure, income structure.

(19)

16 Next, the pre-GFC and post-GFC mean is calculated for each business model component of each bank. This is done in order to run the analyses in this study where the differences between business models pre-GFC and post-GFC are evaluated.

After the earlier described sample was established, the data of outliers more than eight standard deviations away from the mean of the respective business model structure was excluded from the sample for the respective business model structure. This resulted in data for the following banks to be excluded.

Table 4.2: Banks excluded due to being PWD outliers.

Income structure Capital structure

Bank of New York Mellon (International) Ltd (The) ABH Financial Limited

Cyprus Turkish Cooperative Central Bank Limited Caisse Centrale du Crédit Mutuel SA Daiwa Capital Markets Europe Limited

Merrill Lynch International Nova Kreditna Banka Maribor d.d.

SG Hambros Bank (Channel Islands) Limited Sparkasse Dachau

4.5. Descriptive statistics

This section covers the descriptive statistics of the data used in this study, after all the steps discussed in the sections above have been completed. As can be expected, the fewest data points available per year is for 2015 (17,378). The most data is available for 2012 (31,764) and 2013 (31,647). Table 4.3 presents a full overview of total amounts of data points and the average data points per bank available per year.

Table 4.3: Available data points per year and average per bank

Year 1998 1999 2000 2005 2006 2007 2012 2013 2014 2015

Count 24,333 25,515 25,657 26,445 26,417 26,731 31,764 31,647 28,992 17,378

Average per bank 30.9 32.4 32.6 33.6 33.5 33.9 40.3 40.2 36.8 22.1

Summary statistics for each individual business model component are available in the appendix (Table A.2, p. 29).

5. Results

Table 5.1 shows the results of the paired sample t-test performed on the individual business model components, grouped by their respective business model structure. The numbers after the sub header of each business model structure indicate for how many components 𝐻₀1,2 is rejected due to the t-statistic.

(20)

17

Table 5.1: Outcome of paired sample t-test per business model component

Pre GFC (%) (Post GFC) 95% CI for

Business model component Mean SD Mean SD Mean Difference t df

Income (14/28)

Interest Income on Loans _65.3 _63.5 _14.2 _18.2 _0.5 _2.7 _2.89* ₆₇₀ Other Interest Income _22.4 _17.5 _25.5 _20.4 _3.2 _6.2 _6.10* ₇₅₆

Dividend Income 4.4 4.0 5.1 4.1 0.0 0.6 2.01* 669

Interest Expense on Customer Deposits _33.9 _21.8 _22.2 _19.2 _4.5 _11.8 _4.48* ₁₀₈ Other Interest Expense _41.8 _27.3 _18.5 _17.1 _13.5 _15.4 _29.52* ₇₄₀ Net Gains on Trading and Derivatives _0.9 _1.1 _8.3 _6.1 _-0.8 _0.8 _-0.05 ₄₉₃ Net Gains on Other Securities _2.2 _3.5 _7.5 _22.9 _-6.0 _2.2 _-0.90 ₁₃₉ Net Gains on Assets at FV through Income Statement _0.8 _0.7 _1.9 _2.8 _-2.0 _1.1 _-0.60 ₂₆

Net Insurance Income _0.0 _0.0 _2.5 _5.5 _-4.5 _4.5 _0.01 ₈

Net Fees and Commissions _12.1 _15.8 _9.1 _13.2 _-4.1 _-3.1 _-13.51* ₇₇₀ Other Operating Income _3.8 _4.6 _6.1 _10.5 _-1.5 _-0.2 _-2.46* ₇₆₇ Personnel Expenses _20.6 _25.5 _8.1 _10.7 _-5.4 _-4.4 _-18.73* ₇₆₂ Other Operating Expenses _16.0 _20.6 _8.2 _10.5 _-5.2 _-4.0 _-15.20* ₇₇₆ Equity-accounted Profit/Loss - Operating _0.5 _0.9 _0.8 _1.7 _-0.3 _0.6 _0.64 ₁₉ Loan Impairment Charge _7.0 _2.5 _4.9 _18.7 _3.3 _6.1 _6.57* ₇₄₅ Securities and Other Credit Impairment Charges 0.2 0.9 0.9 3.2 -3.1 0.7 -1.29 47 Equity-accounted Profit/Loss - Non-operating 0.9 1.0 3.4 3.3 -3.5 4.0 0.14 12 Non-recurring Income 3.5 0.4 8.6 11.5 1.5 3.2 5.36* 270 Non-recurring Expense 1.6 0.9 5.7 3.5 -0.3 1.2 1.21 280 Change in Fair Value of Own Debt 0.1 -0.3 0.3 0.6 -1.5 2.4 1.00 2 Other Non-operating Income and Expenses -1.6 -8.6 6.5 8.0 6.2 8.0 15.70* 572

Tax expense 3.7 5.5 3.1 3.7 -2.0 -1.4 -11.40* 769

Profit/Loss from Discontinued Operations 1.8 -1.4 4.8 12.2 -6.2 13.0 0.98 4 Change in Value of AFS Investments 0.0 2.5 2.1 3.6 -3.9 -1.4 -4.24* 49 Revaluation of Fixed Assets 0.6 0.6 1.0 2.1 -6.7 7.9 1.00 1 Currency Translation Differences 0.2 -1.5 0.8 4.1 -0.5 4.7 1.71 19 Remaining OCI Gains/Losses -1.8 0.0 8.6 2.9 -5.5 1.7 -1.10 30 Fitch Comprehensive Income 7.1 8.1 10.7 29.5 -3.1 1.1 -0.98 777

Assets (15/33)

Residential Mortgage Loans 31.0 28.9 22.7 14.1 -4.1 0.6 -1.46 66

Other Mortgage Loans 0.0 20.2 0.0 25.6 0

Other Consumer/Retail Loans 17.9 16.2 16.6 15.6 -2.1 5.1 0.82 37 Corporate & Commercial Loans 40.1 11.2 29.2 18.7 0.3 7.6 2.16* 120

Other Loans 53.8 31.0 21.5 20.0 24.0 26.8 34.99* 627

Less: Reserves for Impaired Loans/NPLs 1.7 1.7 1.7 3.7 -2.7 -1.4 -6.10* 133 Memo: Impaired Loans & Advances to Customers

incl. above 2.1 3.0 1.8 4.0 -6.0 -3.3 -6.90* 104

Memo: Loans at Fair Value included above 7.1 37.0 12.3 40.7 -3.7 3.1 -1.00 1

Memo: Loans to the Public Sector _5.3 9.8 0

Memo: Total Impaired Loans & Assets 2.1 3.0 1.8 4.0 -6.0 -3.3 -6.91* 104 Memo: Total Impaired Loans 2.1 3.0 1.8 4.0 -6.0 -3.3 -6.91* 104

(21)

18

Loans and Advances to Banks 15.9 12.6 16.2 15.3 2.4 3.9 8.35* 770 Memo: Impaired Loans & Advances to Banks 0.1 0.2 0 Reverse Repos and Cash Collateral 16.6 10.1 2.0 17.6 -18.1 28.7 0.98 2 Trading Securities and at FV through Income 4.1 1.8 7.1 7.3 -1.6 0.7 -0.77 86

Derivatives 1.5 3.0 3.3 9.2 -1.5 0.0 -2.07* 57

Available for Sale Securities 9.4 2.2 12.2 6.0 -0.4 3.2 1.56 108 Held to Maturity Securities 6.6 8.2 10.3 9.9 -4.4 0.4 -1.65 53 At-equity Investments in Associates 1.6 1.3 2.9 1.3 0.1 0.5 3.18* 707 Other Securities 21.1 21.0 13.3 13.3 -1.7 -0.2 -2.46* 593 Memo: Government Securities Included Above 3.4 4.8 5.8 6.3 -1.8 -1.0 -6.61* 598

Memo: Total Securities Pledged 16.9 6.6 13.1 11.1 0

Investments in Property 0.6 0.4 2.7 0.6 -0.3 0.0 -1.88 20

Insurance Assets 2.2 1.5 3.0 3.3 -3.9 5.6 0.57 3

Cash and Due From Banks 2.4 2.9 4.0 5.4 -0.8 0.0 -1.97* 754 Foreclosed Real Estate 0.0 0.7 0.0 1.5 -1.2 0.8 -1.00 2

Fixed Assets 1.2 1.0 1.5 2.1 0.1 0.3 4.65* 760

Goodwill 0.3 0.3 0.6 0.5 -0.2 0.3 0.60 37

Other Intangibles 0.1 0.1 0.3 0.2 0.0 0.0 1.37 709

Current Tax Assets 0.1 0.1 0.1 0.2 0.0 0.0 1.52 35

Deferred Tax Assets 0.1 0.3 0.3 0.5 -0.3 -0.1 -3.12* 57 Discontinued Operations 0.0 0.6 0.0 0.7 -1.3 0.8 -1.00 2

Other Assets 1.5 1.2 4.4 4.9 0.1 0.5 2.57* 778

Funding (9/18)

Customer Deposits - Current 24.5 39.0 15.5 20.0 -15.5 -13.7 -32.24* 737 Customer Deposits - Savings 32.2 29.9 16.2 16.5 1.6 3.0 6.71* 612 Customer Deposits - Term 19.0 13.3 16.5 15.0 4.4 6.4 10.54* 699 Deposits from Banks 22.5 18.4 19.3 17.8 3.2 4.9 9.24* 763

Repos and Cash Collateral 13.5 8.9 16.1 0

Other Deposits and Short-Term Borrowings 2.0 2.1 6.9 8.3 -0.2 1.2 1.30 290 Senior Debt Maturing after 1 Year 14.2 7.8 22.8 15.2 3.8 9.0 4.90* 95 Subordinated Borrowing 1.6 1.0 3.6 3.1 0.4 0.9 5.66* 535

Other Funding 5.8 2.0 10.4 6.5 3.8 5.1 13.93* 472

Trading Liabilities 3.0 2.1 6.1 8.7 -4.1 5.6 0.35 12

Fair Value Portion of Debt 0.0 0.0 0.1 0.0 0.0 0.0 1.00 4 Credit Impairment Reserves 0.2 0.7 0.5 1.4 -1.5 0.8 -0.65 9 Reserves for Pensions and Other 1.1 1.1 1.1 0.9 -0.1 0.0 -1.59 722 Current Tax Liabilities 0.1 0.1 0.2 0.1 0.0 0.0 1.82 601 Deferred Tax Liabilities 0.2 0.1 0.3 0.2 0.0 0.1 0.61 48 Other Deferred Liabilities 0.4 0.3 2.4 2.1 0.0 0.2 3.22* 607

Insurance Liabilities 2.1 3.0 3.0 5.4 -5.7 3.3 -0.63 6

Other Liabilities 1.7 1.3 7.1 6.0 0.0 0.8 2.03* 778

Capital (2/7)

(22)

19

Pref. Shares & Hyb. Capital accounted for as Equity 28.0 12.2 70.2 103.2 -14.5 92.7 1.53 19

Common Equity 99.5 99.7 2.6 2.4 -0.4 0.0 -1.78 778

Non-controlling Interest 1.9 0.8 3.1 3.7 0.0 3.2 2.06* 34 Securities Revaluation Reserves 2.4 2.1 4.3 3.5 -0.9 1.3 0.41 51 Foreign Exchange Revaluation Reserves -3.2 -1.0 8.6 7.8 -3.2 3.1 -0.03 3 Fixed Asset Revaluations & Other Accumulated OCI 3.9 1.6 13.1 7.4 -1.5 4.7 1.03 35

* P < 0.05

From Table 5.1. can be read that for almost half of all components (40 out of 86) the null-hypothesis can be rejected. This is an indication that for an average bank those components have changed since the GFC. The only structure that has not changed much in the period after the GFC is the capital structure, with only two relatively non-essential components (Preferred shares & hybrid capital accounted for as debt and non-controlling interests) changing significantly.

As stated before, only if 𝐻₀1,2 can be rejected is it interesting to continue with research questions 3 and 4. The results of the analyses needed to answer those questions is covered below.

Table 5.2: Outcome of paired sample t-test on mean PWD by business model structures (preGFC – postGFC)

Pre GFC (%) Post GFC (%) 95% CI for

Business model structure Mean SD Mean SD Mean Difference t df

Income 42.1 21.8 46.2 24.8 -4.9 infinity -9.21 765 Assets 31.2 14.5 38.5 15.3 -8.0 infinity -17.58 771 Funding 35.8 16.2 36.5 16.3 -1.1 infinity -2.26 769 Capital 5.6 6.6 2.7 4.2 2.6 infinity 16.16* 767

* P < 0.05

The results presented in Table 5.2 indicate that 𝐻₀3,4can be rejected only for the capital business model structure, while for the rest the null hypothesis cannot be rejected.

As discussed in 3. Methodology and hypothesis, the same test is performed on the data of T0 and pre-GFC, the results of that test are presented in Table 5.3.

Table 5.3: Outcome of paired sample t-test on mean PWD by business model structures (T0 – preGFC)

T0 (%) Pre GFC (%) 95% CI for

Income 42.9 23.2 42.1 21.8 0.2 infinity 2.20* 765 Assets 29.5 14.9 31.2 14.5 -2.2 infinity -6.28 771 Funding 34.5 17.9 35.8 16.2 -1.8 infinity -4.39 769 Capital 12.1 12.7 5.6 6.6 5.7 infinity 15.72* 767

* P < 0.05

For the period T0 until pre-GFC, 𝐻₀3,4 can be rejected for the income and capital business model structure of banks. It can be seen from Table 5.2 and Table 5.3 that only the income structure of

(23)

20 banks’ business models is developing differently between T0 and pre-GFC, and pre-GFC and post-GFC. This is seen from the t-value in both tables.

In order to test the significance of the difference between these two developments, Table 5.4 presents the outcome of the test on difference between 𝐷𝑃𝑟𝑒𝑝,𝑠 and 𝐷𝑃𝑜𝑠𝑡𝑝,𝑠.

Table 5.4: Outcome of paired sample t-test on difference between PWD by business model structures

DPrep,s (%) DPostp,s (%) 95% CI for

Income -0.8 10.0 4.1 12.4 -6.0 infinity -7.56 765 Assets 1.7 7.6 7.3 11.5 -6.5 infinity -10.11 771 Funding 1.3 8.2 0.7 8.0 -0.1 infinity 1.42* 769 Capital -6.4 11.3 -2.9 5.0 -4.3 infinity -7.43 767

* P < 0.05

Table 5.4 present the results of the final test, indicating that 𝐻₀5_{can only be rejected for the}

funding structure of banks’ business models.

The next section presents the conclusions based on these results.

6. Conclusion

In order to answer the main research question of this study, the research questions have to be answered first. This section covers each question and then presents a conclusion on main research question.

Research questions 1 and 2 focuses on the balance sheet and income statement components of banks and especially whether those had undergone change since the GFC. In this study an analysis was done to test whether a selected group of business model components had changed significantly since the GFC. The expectation was that at least the asset, funding, and income structure of banks would have changed since the GFC. From Table 5.1. it can be seen that half of the components tested as part of the asset, funding, and income structure have changed significantly since the GFC. The income structure shows changes in the interest income and expenses, the most likely cause for this is the decrease in interest rates. At the same time personnel expenses have changed as well. This can be a result of the downsizing banks went through during the GFC. The asset structure has changed most in the loan book. Regulatory changes might have caused this by imposing higher capital requirements for loans to certain counterparties. The funding structure shows changes in funding from deposits and wholesale funding. Based on the result, it can be concluded that the balance sheet and income sources of banks has indeed changed since the GFC.

(24)

21 The results found in research questions 1 and 2 are positive for this study, as research questions 3 and 4 assumed that the balance sheet and income source of banks have changed since the GFC. Research question 3 focuses on how the balance sheet of individual banks have changed compared to their peers’ since the GFC. This has been tested by calculating the pair wise distance (PWD) of each bank compared to its peers’ on three business model structures of banks, namely assets, funding, and capital. After calculating each banks’ PWD for each of the three structures, this study tested whether the mean difference between the PWD before and after the GFC was equal to or smaller than zero, or alternatively greater than zero. If the mean difference is greater than zero, this would indicate that the balance sheet of banks on average has become more homogeneous since the GFC, as expected for the assets and funding structures in this study. Table 5.2 presents the outcomes of the analysis for the business model structures. It can be concluded that the capital structure of banks has become more homogeneous since the GFC. This indicates that, on average, banks have increasingly started using more of the same capital. This was found against expectations of this study. As regulation introduced new definitions of capital and altered others after the financial crisis, it was expected to result in an insignificant result for capital.

Research question 4 focuses on the change of income sources of banks compared to their peers’. In order to answer this question, the methodology used to answer question 3 was used on the data of the income structure. The expectation was that the difference between PWD of pre-GFC and post-pre-GFC would be found to be significantly greater than zero. Table 5.2 presents the outcome of this analysis. As the null hypothesis cannot be rejected for income, it cannot be concluded that, on average, banks’ income sources have become more homogeneous since the GFC.

The analyses carried out for research question 3 and 4 have been performed in order to partially answer research question 5 as well, the only alteration being that differences between T0 and pre-GFC have been evaluated to be significantly greater than zero or not. This was the first step in controlling for general developments in business models that take place over time. Table 5.3 presents the outcomes of this analysis. It can be concluded that both income and capital structures of banks’ business models have become more homogeneous between T0 and pre-GFC.

The second step in answering research question 5 is determining whether the mean difference in PWD between pre-GFC and post-GFC is significantly greater than the mean difference in PWD between T0 and pre-GFC. The expectation was that the mean difference for the asset, funding, and

(25)

22 income structures is greater for the period after the GFC than for the period before GFC. The outcome of the analysis is presented in Table 5.4. It can be concluded from the results that only the funding structure has become more homogeneous during the period following the GFC than before the GFC. Although partially confirming the expectation, this is an interesting outcome as for each period on its own there is no indication that the funding structure has become more homogeneous. Therefore, it can be concluded that the development over time for the funding structure is not increased homogeneity, but compared to the trend (albeit it a two-period trend) the funding structure has become more homogeneous since the GFC.

Having answered research question 1 through 5, the main research question can also be answered. It can be concluded that banks’ business models have changed compared to their own business models since the GFC. Subsequently, it can be concluded that, on average, banks have become more homogeneous in their funding sources compared to their peers since the GFC.

This means that banks have used more of the same funding sources since the GFC than the development seen in the period before the GFC. As a result, more demand for the same funding sources could result in higher prices for these sources. However, it is expected that the market will adjust to this. Instead of specializing in the funding with higher demand, there will be a substitute effect where other types of funding will be supplied by investors. After a transition period, the supply and demand of funding sources should have come to an equilibrium again. These expectations are not tested in this thesis as they are outside of the scope. The development of demand and supply of funding sources is included in the next section, as research into it could provide new insights.

Although not covered in this study, deposits and wholesale funding are most likely the causes for the greater homogeneity within the funding sources of banks, as it was earlier presented that changes in the funding structure were mainly driven by deposits and wholesale funding. However, this should be researched in further studies.

The fact that only the funding structure of banks has become more homogeneous since the GFC, compared to the period before the GFC, indicates that regulatory and monetary changes do not have the effects which were expected at the start of this study. One explanation for this could be that banks were already fairly homogeneous in their asset and income structures. Resulting in small insignificant changes due to any changes since the GFC.

(26)

23

7. Follow up research

This section provides some interesting follow-up research areas, which were not in scope of this study due to various reasons.

As discussed in 3.1. Limitations, this study does research the effects of regulatory and/or monetary environmental changes on banks’ business models. Further research could single-out selected environmental changes and control for them in a regression over two, or more, periods and study any correlations or causal effects.

The expectations of this study were broader than the results which were found. The asset and income structure of banks’ business models have not become more homogeneous. It might be that these two structures were already more homogeneous before the GFC than the funding structure was. Possibly leaving little room to become even more homogeneous over time. This would be interesting to study in a follow-up paper.

As it was found in this study that the funding structure of banks’ business models has become more homogeneous since the GFC, a study to the underlying causes of this finding could have intriguing results.

A more homogeneous funding structure means greater demand for the same funding sources, the market is expected to react to this by supplying different or new funding source causing a transitional period. After this period, a new supply and demand equilibrium should consist. The developments in the funding market could provide new insights in the working of the market after changes in the banking environment.

Additionally, the risks of having greater homogeneity, and thus less diversity, in funding of banks can be subject of future research. It could be studied what would happen to the banking sector, given a scenario in which the more commonly used funding sources dry up.

This study controls for the effects developments over time by comparing the differences in pair wise distance between T0 and pre-GFC (the control period), and pre-GFC and post-GFC (the period of interest in this study). This provides a two-period trend for the differences. Further research could identify a more general development over a longer period, making it possible to perform analyses on the development of business models controlled for this long-term development.

(27)

24

8. References

Altunbas, Y., Manganelli, S., & Marques-Ibanez, D. (2011). Bank risk during the financial crisis: do business models matter? Working paper no. 1394. European Central Bank.

Ayadi, R., de Groen, W. P., Sassi, I., Mathlouthi, W., Rey, H., & Aubry, O. (2016). Banking

Business Models Monitor 2015 Europe. HEC Montréal. International Research

Center on Coopertaive Finance.

Boot, A. W., & Thakor, A. V. (2014). Commercial Banking and Shadow Banking: The Accelerating Integration of Banks and Markets and its Implications for Regulation. In A. Berger, P. Molyneux, & J. Wilson, Oxford handbook of banking (2nd ed.). Oxford: Oxford University Press.

Cai, J., Saunders, A., & Steffen, S. (2011). Syndication, interconnectedness, and systemic risk.

Calem, P., & Rob, R. (1999). The Impact of Capital-Based Regulation on Bank Risk-Taking. Journal of Financial Intermediation, 8, 317-352.

Caprio, G. J., & Klingebiel, D. (1997). Bank insolvencys: Bad luck, bad policy or bad banking? Proceedings of the 1996 World Bank Conference on Development

Economics (pp. 79-104). Washington, DC: World Bank.

Claessens, S., & Kodres, L. (2014). The Regulatory Responses to the Global Financial Crisis: Some Uncomfortable Questions. Working Paper No. 14/46. International Monetary Fund.

Claessens, S., Coleman, N., & Donnelly, M. (2016). "Low-for-long" interest rates and net

interest margins of banks in Advanced Foreign Economies. Retrieved from Federal

Reserve System: 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

Dapp, T. F. (2014). Fintech - The digital (r)evolution in the financial sector. Deutsche Bank. Frankfurt am Main: Deutsche Bank Research.

(28)

25 Fischer, K. P., & Chénard, M. (1997). Financal Liberalization Causes Banking System Fragility. Working Paper No 97-12. Université Laval, Centre de recherche en économie et finance apliquées.

Ghemawat, P. (1991). Commitment: the Dynamic of Strategy. New York: The Free Press. Greenbaum, S. I., Thakor, A. V., & Boot, A. W. (2015). Contemporary Financial

Intermediation (3rd ed.). Academic Print, Elsevier.

Hellmann, T. F., Murdock, K. C., & Stiglitz, J. E. (2000, March). Liberalization, Moral Hazard in Banking, and Prudential Regulation: Are Capital Requirements Enough?

The American Economic Review, 90(1), 147-165.

Humphrey, D., & Pulley, L. (1997). Banks' Responses to Deregulation: Profits, Technology, and Efficiency. Journal of Money, Credit and Banking, 29(1), 73-93. Komai, A., & Richardson, G. (2011). A Brief History of Regulations Regarding Financial Markets in the Unites States: 1789 to 2009. Working Paper No. 17443. Cambridge, MA, USA: National Bureau of Economic Research.

Martel, M., van Rixtel, A., & González Mota, E. (2012). Business models of international banks in the wake of 2007-2009 global financial crisis. Bank of Spain, Revista de

Estabilidad Financiera, 22, pp. 99-121.

Merton, R. C. (1995). Financial innovation and the management and regulation of financial institutions. Journal of Banking & Finance, 19, 461-481.

Modigliani, F., & Miller, M. H. (1958). The Cost of Capital, Corporation Finance and the Theory of Investment. The American Economic Review, 261-297.

Myers, S. C., & Majluf, N. S. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics, 187-221.

Schaeck, K., Cihak, M., & Wolfe, S. (2009). Are Competitive Banking Systems More Stable? Journal of Money, Credit and Banking, 41(4), 711-724.

(29)

26

9. Abbreviations

Abbreviation Definition

CCB Capital Conservation Buffer CET1 Common Equity Tier 1

CRD IV Capital Requirement Directive IV CRR Capital Requirement Regulation GFC Great Financial Crisis

HQLA High Quality Liquid Assets LCR Liquidity Coverage Ratio NCA National Competent Authorities Post-GFC Period 2012-2015

Pre-GFC Period 2005-2007

PWD Pair Wise Distance

T0 Period 1998-2000

T1 Tier 1

(30)

27

Appendix: Tables and Figures

Table A.1: Type of banks and definition

Type of bank Definition In study

Bank holdings & Holding companies Holding companies of bank groups, which usually have very limited business activities Central banks Supervising national banking systems

Clearing & Custody institutions Institutions providing clearing and custody services

Commercial banks Mainly active in a combination of Retail Banking (Individuals, SMEs), Wholesale Banking (large corporates) and Private banking (not belonging to groups of saving banks, co-operative banks)

x

Cooperative banks Cooperative banks have a cooperative ownership structure and are mainly active in Retail Banking (Individuals, SMEs) x Finance companies Consumer Finance Companies, Credit Card Companies, Factoring Companies, Leasing Companies, Trade Finance

Companies

x

Group finance companies Companies mainly active in attracting funding for and lending on behalf of the group

Investment & Trust corporations Investment Corporations/Investment Trust Companies and Private Equity Companies/Property Developers and Covered Bond Issuers investing in various asset

Investment banks Mainly active in Corporate Finance, Debt/Equity Issues, Mergers & Acquisitions, Securities Trading and usually in Private Banking

x

Islamic banks An "Islamic bank is an institution that mobilises financial resources and invests them in an attempt to achieve

predetermined Islamically-acceptable social and financial objectives. Both mobilisation and investment of funds should be conducted in accordance with the principles of Islamic Shari'a".

Micro-financing institutions Providing micro finance to individuals and very small companies Multi-lateral governmental banks Active in multi-lateral development finance

Other non-banking credit institutions Institutions providing guarantees, money transfer companies, companies providing banking and non-banking financial services to groups of financial institutions

Private banking/Asset management companies Banks mainly active in private banking and asset management x Real Estate & Mortgage banks Mainly active in Mortgage Financing and Project Development x Savings banks: Mainly active in Retail Banking ( Individuals, SMEs) and usually belonging to a group of savings banks x

(31)

28 Securities firms Mainly active in Securities Trading/Arbitrage activities/Securities Brokerage/Derivatives

Specialized governmental credit institutions Institutions providing National Development Finance, Sectoral Finance or Export / Import Finance. This specialisation category includes Public Institutions acting on privileged or protected segments or benefiting from Governmental guarantee or sponsoring.

Impact of the Great Financial Crisis of 2007-09 on European Bank business models