The Shadow Banking System : an estimation of the determinants driving the growth of shadow banking in the euro area

(1)

The Shadow Banking System

An estimation of the determinants driving the growth of

shadow banking in the euro area

Roos S. van Gaans

Student number: 10354980

BSc Economics and Business, Specialization Economics & Finance

University of Amsterdam

Supervisor: Lucyna A. Górnicka

This version: June 28, 2015

(2)

Statement of Originality

This document is written by Roos van Gaans who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

Abstract

The purpose of this study is to identify the determinants of the size of the shadow banking system. The empirical part of this thesis uses macro financial data for twelve countries in the euro area to examine the effects of eight key variables on shadow banking growth. The results suggest that when there are systemic effects of a banking crisis present, the size of shadow banks is higher. The size of institutional investors has a positive effect, whereas a lower market concentration seems to have a negative effect on the size of shadow banks. The positive influence of the growth in size of ICPF supports the regulatory arbitrage argument. The thesis represents an attempt of filling the gap of quantitative research on shadow banking in Europe while contributing to a better understanding of the causes of shadow banking growth.

(4)

Glossary of terms

ABS Asset-Backed Securities, 10-13

EC European Commission, 13

ECB European Central Bank, 6, 13-14, 16-19, 23, 25

EU European Union, 22, 25

GDP Gross Domestic Product, 16, 19-23, 26

HHI Herfindahl-Hirschman-Index, 18-22, 24, 26

ICPF Insurance Corporations and Pension Funds, 13-15, 17, 19-24, 26

IMF International Monetary Fund, 6, 13, 15-18, 22, 24, 26

MFI Monetary Financial Institutions, 17, 19-23

MMMF Money-Market-Mutual Funds, 11, 12, 15, 17, 26

NAV Net Asset Value, 11

OECD Organisation for Economic Co-Operation and Development, 17-19

OFI Other Financial Intermediaries, 14, 16-17, 24-25

SBS Shadow Banking System, 19-23

(6)

1 Introduction

In June 2008 Timothy Geithner, then President and Chief Executive Officer of the Federal Reserve Bank of New York, described the emerge of a parallel or ‘shadow’ financial system as an important cause of the recent financial crisis (Geithner, 2008). During the crisis the vulnerabilities of a shadow banking system, outside of the regulatory constraints that are placed on banks, were exposed. What is this shadow banking system Geithner referred to?

The term “shadow banking” was coined by Paul McCulley (2007), economist and money manager, who referred to it as "the whole alphabet soup of levered up non-bank investment conduits, vehicles, and structures”. From the financial crisis onwards the term shadow banking is widely used, referring to a system operating in the shadows of the traditional banking system. A concise definition of the shadow banking system is not straightforward, even though the importance of its entities and activities is recognized (FSB, 2011). The definition agreed upon by the Financial Stability Board (2011) is used in most of the research on shadow banking: “The shadow banking system can be broadly defined as the system of credit intermediation that involves entities and activities outside the regular banking system”.

Why is the shadow banking system a relevant subject for this research? That banking has changed is an understatement (Boot and Marinc, 2008). Despite the fact that the term shadow banking remains vague, its role in the financial crisis of 2007 cannot be ignored. The shadow banking system is seen as one of the main sources of financial stability concerns and therefore the topic of much international debate. Although the size of shadow banking activities is smaller in Europe than in the United States, it is significant (ECB, 2012). Whereas there is growing analytical literature estimating the determinants of shadow banking growth in the US, only a few analytic studies are yet available for Europe. This study represents an attempt to fill part of this gap, building on earlier research done by the International Monetary Fund (IMF, 2014b) and the European Central Bank (ECB, 2012).

The objective of this thesis is to perform a study on the size of shadow banking in the euro area by performing a data analysis based on the available macro financial statistics. This is not without difficulties due to the available data on the subject. A comprehensive assessment of all the determinants of the shadow banking system is nearly impossible and beyond the scope of this thesis. The main question of the research is: What are the key factors explaining shadow banking growth in the euro area?

This thesis is structured as follows: in section two shadow banking is described, based on existing literature. Here, the properties of the shadow banking system, and how

(7)

these contributed to the financial crisis and the systemic risk inherent in shadow banking are presented. In section three, the literature analysis outlined in chapter two is used to form the hypotheses. In addition, the methodology and data sources used in the quantitative research are described. Regression results and discussions are presented in section four. In the final section conclusions are drawn.

(8)

2 The shadow banking system

“If it looks like a duck, quacks like a duck, and acts like a duck, then it is a duck—or so the saying goes. But what about an institution that looks like a bank and acts like a bank? Often it is not a bank—it is a shadow bank.”1

A lot has been written and said about the shadow banking system and its role in the financial crisis of 2007, but it is important to look further than this somewhat shady term to understand what shadow banks actually are and do. Where did they come from, what are they, how do they operate, what part did they play in the financial crisis and how are they regulated going forward?

2.1 Traditional banking

Shadow banking typically describes banks that are operating outside the regulated banking sector (Gennaioli et al., 2013). To understand the shadow banking system, it can be helpful to first look at the traditional banks that do operate inside this regulated sector. Traditional banks function as intermediaries to collect funds from depositors and issue loans to borrowers. These transactions are on-balance-sheet activities (Gorton and Metrick, 2010). Edwards and Mishkin (1995) describe these transactions in more detail: long-term loans are recognized as assets, and are funded by short-term deposits, which are recognized as liabilities on the balance sheet. This process is called maturity transformation: fund less-liquid long-term assets with more-liquid short-term liabilities.

The important types of risks in this maturity transformation process are the solvency of the bank and its ability to access short-term investment. How are traditional banks regulated to minimize this risk? The main factor in the regulatory system is the safety net that is available to banks (Barth et al., 2006). The two most important components of the safety net are the Central Bank acting as a lender of last resort and the deposit insurance system. First, the solvency risk is managed by the deposit insurance system as well as by the capital requirements adopted by the Basel guidelines2_{. Second, the liquidity risk is managed by the} Central Bank, willing to offer loans as a last resort and by traditional banks, lending loans to

1

_{Quote by Laura E. Kodres, IMF staff (d}

_{ivision Chief for the Global Financial Stability Division)}

2

_{The Basel Accords are recommendations on banking laws and regulations issued by the Basel}

Committee on Banking Supervision. The Basel Accords relie on three pillars: (1) minimum capital requirements, (2) supervisory review, and (3) disclosures) (Decamps et al., 2004).

(9)

each other for a specified time. The effect of this safety net can be summarized in twofold: ‘protecting the banking system from bank runs, as well as protecting the depositor’ (Barth et al., 2006).

Edwards and Mishkin (1995) argue that deposits have become less important as a source of funding, making it harder for financial intermediaries to maintain a stable position. Gorton and Metrick (2010) agree with them on the change in stability, pointing out that the deposit insurance works well for small investors, but raises a challenge for large institutions. These large institutions, for example pension funds, need more access to safe short-term investments. An important question Gorton and Metrick (2010) raise is whether the shadow banking system can provide this access via off-balance-sheet activities. In the next sub-section the specific channels and driving forces of the rise of shadow banking are described.

2.2 Rise of shadow banking

The nature of banking has been changing due to numerous financial innovations. Pozsar (2008) summarises these financial innovations as the increasing competition in providing loans; changes in regulation rules; and, innovation in credit securitization. With these innovations, the traditional model of banking has been reshaped and gradually a new system emerged: the so-called originate-to-distribute model of banking (Pozsar, 2008). Banks started to distribute the loans they originated, by selling (a part of) the existing loans in the secondary loan market (Bord and Santos, 2012). The use of this originate-to-distribute model has been critically changing credit intermediation as banks now make less on-balance-sheet transactions and more off-balance-sheet transactions in capital markets (including the secondary loan market). Both Pozsar (2008) and Bord and Santos (2012) explain in more detail how non-bank financial institutions started to behave like banks. Three important properties of shadow banking that changed the nature of banking are described in the next subsections.

2.2.1 Securitization

Securitization is the process in which different types of assets are pooled and transformed into more liquid securities. One of the important aspects of this financial securitization process is that by repackaging an illiquid asset, the credit risk can be sold to separate parties (Mehrling, 2009). Mehrling refers to the words of Fischer Black3 spoken in 1970; Black suggested that three different parties would be involved in this separation. The first party would supply the capital, the second would bear the interest rate risk and the last party would bear the default

3_{Fischer Black was an American economist, best known for his development (with Myron Scholes) of} the widely applied Black-Scholes model on option pricing.

(10)

risk. The instruments he was imagining became quite accurate for the capital market twenty years later. This process is now accomplished by creating Special Purpose Vehicles (SPVs). In Figure 2.1 the securitization process is described by Gorton and Metrick (2010), using a schematic diagram. They argue that the originating firm, the bank, issues loans4_to borrowers. These loans are pooled together and sold to an SPV, which is the master trust in this process. The SPV then tranches this pool of assets and sells them in the capital market to finance the purchase of the portfolio of loans. The tranches are classified with a rank, because each tranche has a different risk associated with it, and sold as asset-backed securities (ABS). The securitization process thus takes illiquid loans that are sold from the balance sheet of the original bank to an SPV, which creates more liquid tranches that are sold and traded off-balance-sheet to investors. Pozsar (2008) argues that the bank insulates itself from the risk by transferring it to outside investors, who invest in the SPV by buying the ABS. These outside investors are often large institutions, for example insurance companies, whose systemic insurance risk propagates over a longer time horizon than systemic risk in banking. They need more access to safe financial assets, so they invest in ABS and take-over the liquidity risk from banks (Pozsar, 2008).

Figure 2.1 The Securitization Process

Source: Gorton and Metrick, (2010)

2.2.2 Sale and repurchase agreements (The repo market)

A repurchase agreement, a repo, is the sale of a security and the agreement to repurchase it at

(11)

a later date for a specified price (Sanches, 2014). The repurchase price is normally higher than the initial selling price: the difference is called the repo-rate and represents interest. In the repurchase market collateral is used to make the loan safe, usually via Treasury securities. Sanches (2014) argues that until 2011 large commercial depositors could not earn interest on their short-term deposits, so they needed an alternative place to store their funds. The institution that owns the financial assets, for example an investment bank, could act as the repo lender and earn the repo-rate. Here, in turn, the ABS can serve as collateral for the repo transactions. Gorton and Metrick (2010) point out that as later described by Sanches (2014), that the collateral can be sold when the borrower does not have the proceeds to buy back the security. This works well, when the underlying collateral is safe and secure. For this reason repos became popular and its volume grew substantially. In the five years prior to the crisis, its size doubled “with gross amounts outstanding at year-end 2007 of roughly $10 trillion in each of the US and euro repo markets, and another $1 trillion in the UK repo market” (Hördal and King, 2008). The authors have estimated that the size of the repo market before the financial crisis of 2007 was approximately as big as the size of the traditional banking system (as measured by total assets).

2.2.3 Money-Market Mutual Funds (MMMFs)

Money-Market Mutual Funds are an important development in financial intermediation. MMMFs pool funds from investors to construct an own portfolio, with the objective to earn interest for the investors while maintaining a NAV5_{of 1 dollar per share (Domian, 1992).} According to the Committee of European Securities Regulators (2010), MMMFs are required by law to invest in short-term assets to minimize their risk. They can invest in cash, repos with Treasury bills as collateral, or in government securities themselves. Gorton and Metrick (2010) point out that it is this feature that makes MMMFs competition to traditional banks. Investors expect their invested money not to decline in nominal terms, as an equivalent of keeping their money in secured depository institutions. MMMFs serve as cash pools and provide the funding source at the end of the chain by buying for example the ABS (Krishnamurthy, Nagel and Orlov, 2014).

MMMFs are one of the most significant financial product innovations of the last years, with an increase of over 2000 percent in the last three decades (Gorton and Metrick, 2010). The growth of the repo market prior to the financial crisis was extraordinary (Sanches, 2014) and the securitization of loans also experienced a spectacular growth in this period (Bord and Santos, 2012). These three properties of shadow banking provide credit enhancement for the financial system, which works well when the used collateral is safe.

(12)

2.3 The financial crisis of 2007

“Those financial institutions that mostly looked like banks. These institutions borrowed short term in rollover debt markets, leveraged significantly, and lent and invested in longer term and illiquid assets. In 2007–2008, we effectively observed a run on the shadow banking system that led to the demise of a significant part of the (then) unguaranteed financial system.” (Acharya et al., 2009).

In the panic of 2007, the ‘bank run’ did not occur in the traditional banking system but in the securitized system: “a run on repo” (Gorton, 2009). The author expresses how the bank run in the recent crisis was not observed by anyone other than those involved in capital markets because the repo market does not involve regular people; it was a run by banks and institutional investors on other banks. In this study three causes of the financial crisis are identified. First, the Subprime Mortgage crisis, which seems to be a small part of what is really a bigger problem. Second, the run on the repo market, and last the run on short-term debt. These last two are directly related to the collapse of the shadow banking system as confirmed by Krishnamurthy et al. (2014).

As described in the previous section, the market for derivative securities has grown spectacularly in the last 25 years, which in turn has created an increased demand for collateral. More tranches were created by SPVs and used as collateral but these ABS were not all as safe (Gorton, 2009). Gorton explains how the increasing demand together with the concerns about the value and liquidity of the collateral caused the depositors to demand a higher haircut6. This created a problem for the shadow banks, which needed to borrow the extra cash. In addition, financial funds like MMMFs, who are required to invest in short-term assets, began to invest part of their collected funds into the short-term repurchase markets (Chang, 2011). When it became clear that the collateral used in the repo market turned risky and illiquid, the run on the shadow banks started: the depositors in the repo market backed away (Krishnamurthy et al. 2014). The financial intermediaries that have been rolling over repurchase loans backed by illiquid assets now suddenly had no access to capital and the shadow banking system collapsed.

2.4 Systemic risk and banks

The Financial Stability Board (2011) affirms that the financial crisis has shown that the shadow banking system can become a source of systemic risk, either directly or through the

6_{A ‘haircut’ is the difference between the value of the deposit and the colleteral. Previously reffered to} as the repo-rate.

(13)

interconnectedness with the traditional banking system. The authors point out that there are important linkages between both systems, as banks often provide support to shadow banks or are part of the credit intermediation chain in the shadow banking system (FSB, 2011). Furthermore banks use the shadow banking system to avoid financial regulation, which can lead to an increase of non-transparent risks and increased leverage in the financial system. The European Commission (2012) agrees with the statements of the FSB that failures due to shadow banking activities can lead to contagion and spillover effects.

Summarizing the reports by the FSB (2011) and the EC (2012), risk taken by shadow banks can be transmitted to traditional banks through different key systemic risk channels: (a) interconnectedness between the systems to provide and take credit in the capital market, (b) use of the credit intermediation chain by maturity transformation, liquidity transformation, credit-risk transfer and the increased use of leverage; and, (c) massive sale of assets with backlash on prices of financial and real assets.

Since the financial crisis, global regulatory reforms coordinated by the FSB (2013) have called for “greater disclosure of asset valuations, improved governance, ownership reforms, and stricter oversight and regulation of shadow banks” (FSB, 2013). This tightening of bank regulations may be encouraging a shift of traditional banks into the shadow banking system (IMF, 2014b).

This section has outlined that the shadow banking system refers to the entities and activities that take place outside of the traditional banking system. These entities perform banking functions without being regulated as traditional banks. The activities exercised by shadow entities are not as transparent and involve financial intermediation through a wide range of securities that are bound along the credit intermediation chain (such as repos and ABS) (ECB, 2012). As the growth of the shadow banking system is seen as one of the main sources of financial stability concerns it is important to study the dynamics of shadow banking and to know its determinants, such as the regulatory framework; the expanding sector of ICPF and the costs of financial crises.

(14)

3 Methodology

3.1 Conceptual framework

The purpose of this study is to identify the key determinants of the growth of the shadow banking system. Whereas there is growing analytical literature on the shadow banking system in the United States, only a few analytical studies are yet available for Europe or the euro area. This study represents an attempt to add an econometric analysis to fill part of this gap. The research that will be conducted is in line with recent research done by the IMF (2014b) and the ECB (2012). The analysis conducted by the IMF is chosen as a guideline because it is one of very few that is estimating the shadow banking size by performing an econometric analysis. In addition, the paper written by the authors at the ECB is chosen because it is one of the first investigations of the size and structure of shadow banking entities in the euro area. In this paper the shadow banking system is defined as: “a system of credit intermediation consisting of financial institutions outside of the regular banking system engaged in activities related to securitization and derivatives”.

3.2 Assessing the size of shadow banking in the euro area

An ideal working definition of how to measure shadow banking is not straightforward. A quantitative assessment of its size is dependent of the available data. The analysis in this study uses primarily data from the European Central Bank database because the research is focused on the size of shadow banking activities in the euro area. Therefore the measure of the size of the shadow banking system used for this study is in line with the investigation recently presented by the ECB (2012). The activities of shadow banking in the euro are measured by examining the balance sheets of Other Financial Intermediaries (OFIs) (ECB, 2012). The ECB defines this sector as “a sector which covers most of the agents engaging in shadow banking, including securitization vehicles but excludes insurance corporations and pension funds”. This definition is supported by the fact that in the path leading to the financial crisis the size of OFIs expanded quickly and its size declined sharply as the crisis unfolded (ECB, 2012). The definition excludes insurance corporations and pension funds (ICPF) but as stated in recent research conducted at the IMF, ICPFs are by definition neither traditional nor shadow banking because of the specific nature of the financial intermediation services they provide (IMF, 2014b). The IMF defines the OFI sector as “(1) all non-bank financial

(15)

corporations and quasi corporations engaged mainly in financial intermediation and (2) entities providing primarily long-term financing” (IMF, 2014b).

3.3 Hypotheses

There are three main explanations for the growth of the shadow banking system as suggested by the literature. These three hypotheses are tested in this study and can be described as: (1) Regulatory arbitrage: The off-balance sheet vehicles were motivated by regulatory arbitrage, as these vehicles were not required to hold a minimum level of capital. Pozsar (2008) motivates this hypothesis by looking at the effects of the Basel Accords, which required more capital restrictions to protect banks from liquidity risk. These accords motivated traditional banks to move assets off the balance sheet and “into the shadow” to avoid the regulation rules (Pozsar, 2008).

(2) Increased demand for safe short-term assets by large institutions: A rapidly expanding sector of large institutional investors caused the growth of the demand side for interest earning financial assets. The shadow banking system provided a solution to this problem of increased demand with the supply of institutional tranches of securitized loans (Yago and McCarthy, 2004). As explained in the second section, securitized loans are structured by an SPV and the institutional investors receive collateral to make the investment safe (Gorton and Metrick, 2010). This second hypothesis is also known as the search-for-yield effect: institutional investors were under pressure from clients to ‘search for yield’ and turned to shadow banks to supply interest-earning assets (Lysandrou, 2011).

(3) Complementarities with the rest of the financial system: Earlier research shows evidence that the growth of the shadow banking system is positive related to the growth of the rest of the banking sector (IMF, 2014a). It shows that in emerging markets the growth of ICPF is accompanied by the growth of non-bank intermediaries as hedge funds and MMMFs.

In the next subsection the variables used in the panel regression are linked with these hypotheses as the drivers of shadow banking.

3.4 Data and methods

3.4.1 Sample and data set

The data set contains country-level variables for twelve countries that are in the euro area, for which the data range from 1999 to 2012. The countries that are included in this analysis are: Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg,

(16)

To estimate the effect of the drivers, a sufficient time horizon with available data is needed and therefore only countries that have adopted the euro before the recent financial crisis are included in this analysis. The timeframe should include data prior to the financial crisis of 2007 because the crisis had an important role in the rise and decline of shadow banks (as explained in section two).

A quarterly data frequency is used to measure shadow banking. This frequency is chosen to capture all the available differences in size and to collect a big enough sample. Quarterly data is available for the measure of the growth in GDP; the size of traditional banks, shadow banks, and insurance companies and pension funds; and, the term spread. For the other variables only data with an annual frequency is available. Therefore an assumption is made that the value of the indices will not change substantially during the year.

A panel regression will be used to predict the value and the relative contribution of each of the explanatory variables. A specification of the regression equation used in this study is:

𝛥𝑆𝐵𝑆!"= ∝ + 𝛽𝑗!𝐺𝐷𝑃 + 𝛽!!𝐶𝑅𝐼𝑆𝐼𝑆 + 𝛽!!𝑀𝐹𝐼 + + 𝛽!!𝐼𝐶𝑃𝐹 + 𝛽!!𝑇𝐸𝑅𝑀 + 𝛽!!𝑅𝐸𝐺 + 𝛽!!𝑇𝐴𝑋 + 𝛽!!𝐻𝐻𝐼 + 𝜀!"

The regression analysis builds on the research done by the IMF (2014b) but some variables are measured differently and two variables are added to the equation. The left-hand of the equation represents the shadow banking growth, measured by the flow of funds in the OFI sector reported by the ECB as explained in section 3.2. The right-hand of the equation consists of possible determinants of the size of shadow banking in which: ∝ is a constant to be estimated; 𝛽_!" (k= 1, .., 8) are coefficients to be estimated; and, 𝜀_! is an error term for the size of the shadow banking system in country j at time t. These explanatory variables are based on the literature on the subject and are linked to the hypotheses of section 3.3 in the next subsection. The variables used are summarized in Table 3.1, which is presented at the end of the section.

3.4.2 Variables used in the regression analysis

GDP represents the quarterly growth rate of the Gross Domestic Product (GDP) at market

prices based on a constant local currency. It controls for changes in the macroeconomic environment and can be linked to the third hypothesis: the change in the size of shadow banks is complementary with the change in the rest of the financial system. Expected thus is a positive relation between the growth in the GDP and the growth in the shadow banking system. The data source for the quarterly growth of GDP per country is the database of the

(17)

Organisation for Economic Co-Operation and Development (OECD).

CRISIS is a dummy-variable, which has a value of 1 when there are systemic effects

of the banking crisis in the country and a value of zero otherwise. Laeven and Valencia (latest update in 2012) prepared a “Systemic Banking Crises Database”, which proposes a methodology to date banking crises based on indices. The researchers stated that a banking crisis is defined as systemic if two conditions are met: “(1) significant signs of distress in the banking system (as indicated by significant bank runs, losses in the banking system, and bank liquidations); and, (2) significant banking policy interventions in response to significant losses in the banking system.” (Laeven and Valencia, 2012). They give a start and (in some cases) an ending date for the effects of the recent financial crisis per country. This dummy variable is included to capture the specific timing of events and largely to improve the overall fitting of the model (Stock and Watson, 2012). By holding extraneous variables constant, the coefficients come closer to estimating the true effect of the independent variable on the dependent variable

.

The expected sign of this determinant is negative, because as explained by Krishnamurthy, et al. (2014) a systemic banking crisis has a negative effect on the size of shadow banking activities.

MFI, or the abbreviation for “Monetary Financial Institutions”, is included as an

explanatory variable and covers the regulated banking system as defined by the ECB (2012). The ECB database is used as data source to measure the size of the MFI sector per country. A limitation to this measure is that MMMFs are included in the MFI sector but a lack of a sufficiently long time horizon prevents this study from rearranging MMMFs as institutions that belong in the OFI sector. MFI is related to the third hypothesis because of the complementary argument: growth of shadow banks is positively related to growth of traditional banks (Pozsar et al., 2010). Shadow banks are involved in activities of traditional banks, where Pozsar et al. (2010) highlight the maturity transformation process. Mandel, Morgan and Wei (2012) confirm this by pointing out that banks are important in providing credit enhancement in the securitization process.

ICPF refers to “Insurance Corporations and Pension Funds” as defined and measured

by the ECB. This variable is directly related to the third hypothesis where is explained that the growth of ICPF often is accompanied by the growth of non-bank intermediaries (IMF, 2014a). Second, ICPF are related to the second hypothesis because as stated in section 3.3, the growth of institutional investors caused the demand for safe financial assets to rise and these assets were provided by the shadow banking system. This determinant is thus expected to have a positive sign.

TERM represents the term spread and is included to measure financial conditions.

(18)

variable can be linked to the search-for-yield-effect as explained in the second hypothesis. Lysandrou (2011) analysed the evidence for the growth in securities prior to the crisis as a result from dropping yields in government markets. Investors turned to shadow banking activities, which often supply the assets earning a higher yield. The effect of this variable is expected to be negative: as interest rates in the traditional market are lower, shadow banking activities increase. The data source used to estimate the term spread for all twelve countries is the OECD database.

REG is included to measure the effects of bank regulation and for this study in

particular of capital stringency. More stringent capital requirements are a motivation for banks and institutions to engage in off-balance-sheet transactions. Regulation is included as a measure to test the first hypothesis (Pozsar et al., 2008). As bank regulation is relatively difficult to measure, the online database provided by Barth, Caprio and Levine (2013) proves to be very useful. The originators of this database have provided a measure of bank supervision and regulation in 180 countries.

The paper by the ECB concludes with recommendations for further research where it stresses the “particular importance of possible regulatory measures” (ECB, 2012). However in the research conducted by the IMF (2014b) regulatory variables7_{are founded to be} insignificant. The latter could be a limitation to this study but still one of the regulatory indices provided by Barth et al. (2013) is included because regulatory arbitrage is one of the main reasons for shadow banking growth (hypothesis 1). The capital requirements index is included as the explanatory variable because this type of regulation is identified as most significant in the comparable research done by the IMF (2014b). REG is measured per country on a scale from one to seven (low to high capital requirements respectively) as measured by the research of Barth et al. (2013). For this research the result of all four surveys (1999, 2003, 2007, 2011) is considered and the sign of the variable is expected to be positive, which indicates that as regulation is sharpened, the size of the shadow banking sector will grow.

TAX refers to the tax burden and fiscal freedom and reflects top marginal income tax

rates, corporate tax rates and the marginal tax rate for the average taxpayer. The tax burden variable is measured by the Heritage Index of Economic Freedom (2015) and the index ranges between 0 and 100. This variable is included to further test the first hypothesis on regulatory arbitrage. Adrian (2014) highlights the importance of the re-structuring of financial intermediation from traditional to shadow banking to avoid taxes (Adrian, 2014). HHI is the final explanatory variable on the right-hand side of the equation and

represents the Herfindahl-Hirschman-Index (HHI), which is a commonly excepted measure of

7_{The IMF uses indices of capital stringency, capital regulatory, supervision power, financial statement} transparancy and, global liquidity quantities in their panel regression.

(19)

market concentration8_{. The index ranges between 0 and 1; low and high market concentration} respectively, and the value of the index is found in the ECB database. HHI is included in the equation to measure the structure of the banking sector. In the Banking Structures Report by the ECB (2014) is shown that the concentration measure reflects a number of structural factors. Larger countries, such as Germany, have a more fragmented banking system and smaller countries tend to be more concentrated. Crisis countries, such as Greece and Spain, are more dominated by big banks (ECB, 2014). Various empirical studies have examined the effect of concentration and how competition affects banks risk taking. For example, Shy and Stenbacka (2014) show a positive relation between competition and risk taking but in contrast, Niinimaki (2004) finds that the effect depends on the structure of the market. Concentration therefore may have a positive or negative effect on the size of the shadow banking system.

In Table 3.1 the variables used are summarized. The next section will present the results of the regression analysis, including its assumptions and limitations.

Table 3. 1 Summary of panel regression on shadow banking growth

(SBS is measured by the flow of funds)

Variable Expected sign Data source

GDP + OECD Database

CRISIS - Laeven and Valencia (2012)

MFI + ECB Database

ICPF + ECB Database

TERM - OECD Database

REG + Barth, Caprio, and Levine (2013)

TAX + Heritage Index of Economic Freedom (2015)

HHI +/- ECB Database

8_{The Herfindahl-Hirschman-Index (HHI) is defined as the sum of the squares of the market shares of} the five largest credit institutions within the industry, where the market shares are expressed as fractions (ECB, 2014)

(20)

4 Regression analysis

4.1 Panel regression with fixed effects

To obtain an estimate of the explanatory variables, a panel regression with fixed effects is run. In this analysis fixed effects are added to the panel regression, because they use the information that certain sets of observations come from certain countries. The use of fixed effects makes it possible to look at general trends and to correct for country-specific effects that cannot be measured; it accounts for individual heterogeneity (Stock and Watson, 2012). To confirm the use of fixed instead of random effects, a Hausman test has been employed on the panel data. Lastly, the sample is perfectly balanced with 672 observations. The following Table summarizes the data characteristics:

Table 4.1 Descriptive statistics

Variable Acronym Mean Std. dev. Min Median Max

Shadow banking sector flow of funds SBS 1651.135 7333.361 -39637 1651.135 56215

Real GDP GDP 0.406 1.225 -6.900 .406 7.200

Systemic banking crisis CRISIS 0.196 0.398 0 .196 1

Banking sector flow of funds MFI 1067.271 5980.746 -31731 1067.271 38158

ICPF flow of funds ICPF 10.573 227.238 -1274 10.573 2879

Term spread TERM 1.807 2.536 -4.7 1.807 24.700

Regulatory variable REG 4.544 1.239 1 4.544 7

Tax burden TAX 59.195 14.793 24.100 59.195 82.600

Herfindahl-Hirschman-Index HHI 0.097 0.077 0.014 .0970 0.370

Note: 672 observations for each variable.

4.2 Assumptions

There are five assumptions to be made, which justify the use of this model (Stock and Watson, 2012).

First, the dependent variable should have a linear relationship with the independent variables. Natural logarithms are used to create this desired linearity in some of the parameters. Stock and Watson (2012) explain that economics time series are often analysed after the computation of their natural logarithms, one important reason for this is that many

(21)

economic series exhibit growth that is approximately exponential; such as GDP. The logarithm of the series grows approximately linearly so it is useful to transform the series so that changes are proportional. Another important reason is to make highly skewed variables less skewed; the use of natural logarithms may improve the fit of the variable in the model by altering its scale. In the regression equation of this research, the natural logarithm is used to transform the variables (excluding the indices, the term spread and the dummy variable). The transformed variables have a log-log relation with the dependent variable (the size of the shadow banking system): for example, a one percent change in real GDP is associated with a 𝛽𝑗! percent change in SBS (elasticity relation).

Second, the multiple linear regression requires all variables to have a normal distribution. The normality is checked and confirmed with a goodness of fit test: the Kolmogoroff-Smirnoff test.

Another assumption is that there is little or no multicollinearity in the data; the explanatory variables should be independent of each other. A correlation matrix is computed and little multicollinearity is confirmed because all the correlation coefficients are smaller than |0.8| (Stock and Watson, 2012).

Table 4.2 Correlation matrix

GDP CRISIS MFI ICPF TERM REG TAX HHI

GDP 1.000 CRISIS -0.217 1.000 MFI 0.109 -0.005 1.000 ICPF 0.019 -0.017 0.030 1.000 TERM -0.230 -0.179 -0.087 -0.036 1.000 REG -0.019 -0.061 -0.015 -0.033 0.131 1.000 TAX 0.041 0.061 -0.122 -0.022 0.044 -0.011 1.000 HHI -0.062 -0.077 -0.022 -0.024 0.034 -0.097 0.048 1.000

Fourth, the Woolridge test is derived to test the null hypothesis that the residuals are not linearly autocorrelated in the panel-data model (Woolridge, 2002). Autocorrelation is the linear dependence of a variable with itself in two points in time. The null hypothesis should not be rejected to justify the model. The Woolridge test gives a p-value of 0.4399, which means the null hypothesis is not rejected and the residuals are not autocorrelated.

Lastly, the multiple linear regression model makes the assumption that the error term is homoscedastic, which is the case if the variance of the error term is constant and does not

(22)

depend on the independent variables (Stock and Watson, 2012). To test for this assumption the modified Wald test for group wise heteroscedasticity in fixed effect regression models is run. The null hypothesis of constant variances (homoscedasticity) is rejected since the Wald test gives a p-value that is significant at the 99.99+ percent level. Therefore the model is corrected for heteroscedasticity by the use of robust standard errors.

4.3 Results and discussion

The extent to which the macro financial and regulatory variables contribute to the size of the shadow banking system is assessed for a set of twelve countries in the European Union (EU). In the flow of funds data regression, the size of the banking sector, the size of institutional investors and the tax burden have the expected signs. The sizes of the institutional investors, the tax burden, and the Herfindahl-Hischman-Index have a significant influence (benchmark signification, Table 3.4). Results differ from prior comparable research but it should be taken into consideration that this is one of the first studies based in Europe, the research of the IMF (2014b) focused on different economies worldwide. The regression results are presented in Table 3.4 below.

Table 4.3 Regression results

(Measure of the size of the shadow banking system: ln(SBS))

Variable Acronym Coefficient Estimate

Real GDP ln(GDP) -.0790

(.1426)

Systemic banking crisis CRISIS .7165**

(.3152) Banking sector flow of funds ln(MFI) .0159

(.0831)

ICPF flow of funds ln(ICPF) .2832***

(.0842)

Term spread TERM -.0636

(.1372)

Regulatory variable REG -.0982

(.1163)

Tax burden TAX .0977*

(.0369)

Herfindahl-Hirschman-Index HHI 13.8272**

(23)

Intercept -.4565

Number of observations 672

R squared 0.3086

Note: OLS regression = ordinary least squares regression. *, **, *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Natural logarithms are taken of SBS, GDP, MFI and ICPF. Standard errors are robust to heteroscedasticity and cross-country dependency. The estimation period is 1999-2012. Equations are estimated by fixed effects (within the regression). The countries in the sample are as follows: Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, and Spain.

GDP and CRISIS are included as control variables. A number of macro financial control variables are generally found to be significant, including GDP (Tieman and Maechler, 2009). However in this study, the size of real GDP has no statistically significant explanatory power for the size of shadow banking. This could be explained since the majority of the twelve countries in the sample have approximately the same real GDP, growing in at approximately the same pace, while they have significant different sizes of the shadow banking system.

The dummy variable that controls for systemic effects of banking crises is statistically significant but has the opposite effect of the expected sign. The reason for this could be that according to the index established by Laeven and Valencia (2012), the recent banking crisis has had effect from 2008 and onwards. In this period the size of the shadow banking sector first declined sharply as predicted by Krishnamurthy et al. (2014) but the total size remained higher than in the period before the crisis. Therefore it could be explained that the crisis dummy has a statistically significant positive effect at the 10 percent level. The presence of systemic effects of a banking crisis is associated with an average increase of 71.65 percent in the size of shadow banking activities.

According to the model, the size of the banking sector has a slightly positive effect on the size of the shadow banking system. This result is as expected by the ECB (2012), Pozsar et al. (2010) and Mandel, Morgan and Wei (2012). The coefficient in the model is positive but not significant and therefore no one-sided conclusions can be drawn.

The size of ICPF appears to have a highly significant (at the 1 percent level) positive effect on the size of the shadow banking sector. This result is in line with the research done by Yago and Mccarthy (2004), who argue that an expanding sector of institutional investors causes the growth for the financial assets provided by the shadow banking system. A one-point increase in the size of ICPF leads to a 0.2832 percent increase in the parallel banking

(24)

system.

The term spread has the expected sign, as interest rates in the traditional market are lower, shadow banking activities increase. The coefficient seems to have no statistically significant influence.

The regulatory explanatory variables REG and TAX are only slightly correlated so are both included in the model. The results indicate that the tax burden is the only regulatory variable that has a statistically significant (at the ten percent level) explanatory power. Expected is that the source of this result lies in the fact that the index that measures bank regulation, and in particular capital stringency, has limited variation in the covered period. Such a result is not uncommon in the literature (IMF (2014b), Pozsar (2008) or FSB (2011). The model assigns a positive significant effect to the tax burden coefficient as measured by the Heritage Index of Economic Freedom. A one-unit increase in the measure of tax burden leads to an average increase of 9.77 percent in the dependent variable. This result is supported by the literature, as Adrian (2014) states the importance of avoiding taxes through the shadow system.

Lastly, the findings suggest that the Herfindahl-Hirschman-Index (HHI) has a significant impact (at the five percent level) on the size of the shadow banking system. The HHI ranges from one to zero and a 0.01 increase in the HHI indicates an average increase of 13.83 percent in the size of OFIs. This means that higher market power, or lower market competition, leads to an increase in shadow banking activities. While this result may appear counter-intuitive, this need not be the case. Empirical literature shows mixed results. Shy and Stenbacka (2014) show a positive relation between market competition and risk taking but in contrast lower competition could lead to excessive risk taking as argued by Boyd and de Nicolo (2005). When confronted with increased competition among banks, moral hazard is worsened and banks intentionally take on more risk.

This thesis tests three hypotheses involving the effect of regulatory arbitrage, the increased demand by institutional investors, and the size of the complementary financial system. Applying the estimated results to the hypotheses, it is recognized that these results do not provide a sufficient basis to draw a definite conclusion about the relation between the size of shadow banking and the tested explanatory variables. Halve of the independent variables appear significant with signs that are supported in the literature. The other halve seems to have no statistically significant effect on the tested sample of the twelve countries in the euro zone. The regulatory arbitrage argument is supported by the significance of the effect of tax burden, therefore the first hypothesis is not rejected. Second, the size of ICPF has a highly significant effect on shadow banking, which supports the second and third hypothesis. Lastly, the effect of a systemic banking crisis has a significant effect supporting the last hypothesis.

(25)

4.4 Limitations

Although carefully performed, this study has some limitations and shortcomings. First, the shadow banking system is measured as the total of OFIs, which includes entities that may be regarded as engaged in shadow banking, but also intermediaries for which this view is questionable; such as regulated investment funds (ECB, 2012). The OFI measure used in this study is chosen based on the available data and most likely an underestimation of the real values because of the difficulties in gathering information on financial intermediaries (ECB, 2012). Second, to generalize for larger groups, the study should have involved more countries with economies at different levels; for example emerging market and advanced economies. Therefore the results of the research performed are only attributable to the countries in the euro area. Finally, a comprehensive assessment of all the determinants of the growth of shadow banking is beyond the scope of this thesis. This study is an attempt to identify the most important determinants in explaining shadow banking growth in EU countries.

(26)

5 Conclusion

The primary goal of this thesis is to perform a study on the dynamics in shadow banking across countries in the euro area by attempting to explain the key determinants of its size. Shadow banking is defined as “a system of credit intermediation consisting of financial institutions outside of the regular banking system engaged in activities related to securitization and derivatives”.

In this paper three hypotheses are formed based on existing literature and tested in the performed panel regression analysis. Earlier research identified regulatory arbitrage, the search-for-yield effect and the complementarities with the rest of the financial system as the main explanations for the development and growth of shadow banking. Findings of this study show that regulatory arbitrage has a significant influence: as the tax burden is higher, the incentive to engage in the parallel banking system is also higher. Second, the shadow banking system provides a solution for institutional investors seeking for safe investment as is shown by the positive highly statistically significant effect of the growth in size of ICPF on the size of shadow banking. Furthermore, market concentration seems to have a significant positive effect on the shadow banking system as is shown by the HHI. The last finding shows that when there are effects of a systemic banking crisis present (as estimated by the index of Laeven and Valencia (2012)), the size of shadow banking is higher. The term rate, real GDP and the size of the traditional banking sector seem to have no statistically significant influence on the growth of shadow banks. These last findings are interesting because they are in contrast with the comparable research performed by the IMF (2014b). Differences can be due to the sample of countries. The lack of comparable worldwide data limits the generalization of the evidence of this study for larger groups outside the euro area. Regarding the hypotheses no one-sided conclusion can be drawn but there is support for each of the hypotheses.

Concerning future research, special attention could be paid to the different components of shadow banking, such as MMMFs and the repo market. More investigation is useful to show whether such different vehicles of shadow banking have different determinants and to ascertain whether they are for example all motivated by regulatory arbitrage or by a certain degree or key element of the regulatory framework. Given the pace of research on the subject, such a distinction was beyond the scope of this thesis.

It should be emphasised that while this thesis represents an attempt of filling the gap of quantitative research on shadow banking in Europe, it is also believed that the study contributes to a better understanding of the causes of the growth of shadow banking.

(27)

References

Acharya, V., Philippon, T., Richardson, M. and Roubini, N. (2009): The financial crisis of 2007-2009: causes and remedies. Financial Markets, Institutions & Instruments, 18, 89–137.

Adrian, T. (2014): Financial Stability Policies for Shadow Banking. FRBNY Staff Reports, 664.

Barth, J. R., Caprio, G. and, Levine, R. (2006): Rethinking bank regulation: till angels govern. New york, NY: Cambridge University Press.

Boot, A.W.A. and Marinc, M. (2008): The evolving landscape of banking. Industrial and Corporate Change, 17, 1173-1203.

Bord, V., and Santos, J. (2012): The rise of the originate-to-distribute model and the role of banks in financial intermediation. Economic Policy Review, 18(2), 21-34.

Boyd, J.H. and De Nicolò, G. (2005): The Theory of Bank Risk Taking and Competition Revisited. Journal of Finance, 60(3), 343-1329.

Chang, W.W. (2011): Financial crisis of 2007-2010. Keio Economic Studies, 47, 25-56.

Committee of European Securities Regulators (2010): CESR’s guidelines on a common

definition of European money market funds. Retrieved from:

http://www.esma.europa.eu/system/files/10-049_cesr_guidelines_mmfs_with_disclaimer.pdf

Decamps, J.P., Rochet, J.C. and Roger, B. (2004): The three pillars of Basel II: optimizing the mix. Journal of Financial Intermediation, 13(2), 132-155.

Domian, D.L. (1992): Money market mutual fund maturity and interest rates. Journal of

Money, Credit and Banking, 24(4), 519-527.

European Central Bank (ECB) (2012): Shadow banking in the euro area. ECB Occasional

Paper, 133, April.

Edwards, F. and Mishkin, F. (1995): The decline of traditional banking: implications for financial stability and regulatory policy. Economic Policy Review, 1(2), 27–45.

(28)

European Commission (EC) (2012): Shadow Banking. Green Paper, Brussels, March.

Financial Stability Board (FSB) (2011): Shadow Banking: Strengthening Oversight and Regulation. Recommendations of the Financial Stability Board, October.

Financial Stability Board (FSB) (2013): Policy Framework for Strengthening Oversight and Regulation of Shadow Banking Entities. Consultative Document, Financial Stability Board, Basel. Retrieved from: http://www.financialstabilityboard.org/publications/r_130829c.htm

Financial Stability Board (FSB) (2014): Global Shadow Banking Monitory Report, November Retrieved from: http://www.financialstabilityboard.org/2014/11/global-shadow-banking-monitoring-report-2014/

Geithner, T. F. (2008): Reducing systemic risk in a dynamic financial system. Web. Remarks at The Economic Club of New York. New York, NY.

Gennaioli, N., Shleifer, A., and Vishny, R.W. (2013): A model of shadow banking. The Journal of Finance, 68(4), 1331-1363.

Gorton, G. (2009): Slapped in the face by the invisible hand: Banking and the panic of 2007. Yale School of Management and National Bureau of Economic Research (NBER). Retrieved from http://dx.doi.org/10.2139/ssrn.1401882

Gorton, G., and Metrick, A. (2010): Regulating the Shadow Banking System, Brookings Papers on Economic Activity, 1, 261–312.

Hördahl, P. and King, M.R. (2008): Developments in repo Markets during the financial turmoil. BIS Quarterly Review, (December), 37–53.

Heritage Foundation (2015): Heritage Index of Economic Freedom. Washington, DC.

Retrieved from http://www.heritage.org/index

IMF (2014a): Moving from liquidity- to growth-driven markets. Global Financial Stability

Report, April, 1-170.

IMF (2014b): Risk taking, liquidity, and shadow banking. Global Financial Stability Report: Risk, October, 65-104.

(29)

69(6), 2381-2417.

Laeven, L. and Valencia, F. (2012): Systemic Banking Crises Database. IMF Economic Review, 61(2), 225−70.

Lysandrou, P. (2011): Global inequality as one of the root causes of the financial crisis: a suggested explanation, Economy and Society, 40(3), 44-323.

Maechler, A.M. and Tieman, A.F. (2009): The real effects of financial sector risk. International Monetary Fund Working Papers, 9(198), 1-31.

Mandel, B.H., Morgan, D. and, Wei, C. (2012): The Role of Bank Credit Enhancements in Securitization. Federal Reserve Bank of New York Economic Policy Review, 34 (2), 225–54.

McCulley P. (2007): Teton Reflections. PIMCO Global Central Bank Focus, 2.

Mehrling, P (2009): Credit default swaps: the key to financial reforms. Time for a Visible Hand: lessons from the 2008 world financial crisis, Edited by S Griffith-Jones, JA Ocampo and JE Stiglitz, Oxford University Press.

Niinimaki, J.P. (2004): The effects of competition on banks’ risk taking, Journal of Economics, 81(3), 199–222.

Pozsar, Z. (2008): The rise and fall of the shadow banking system. Regional Financial Review, 13-15.

Sanches, D. (2014): Shadow banking and the crisis of 2007-08. Business Review, 2, 7-14.

Shy, O, and, Stenbacka, R. (2004): Market structure and risk taking in the banking industry. Journal of Economics, 82(3), 249–280.

Stock, J. and Watson, M.W. (2012): Introduction to Econometrics. New York: Prentice Hall. Woolridge, J.M. (2002): Econometric analysis of cross section and panel data. MIT Press, Cambridge.

Yago, G., and McCarthy, D. (2004): The U.S. Leveraged Loan Market: A Primer. Milken Institute, October.

The Shadow Banking System : an estimation of the determinants driving the growth of shadow banking in the euro area