Is there spillover risk between financial institutions after the crisis? : the evolution of systemic risk caused by hedge funds

(1)

MSc Business Economics, Finance Track

Master Thesis

Is there Spillover Risk between Financial Institutions after the

Crisis?

The Evolution of Systemic Risk caused by Hedge Funds

Author: Bence Dukát

Student Number: 11085274

Supervisor: Dr Jan Lemmen

June 29

th

_{, 2016}

(2)

PREFACE AND ACKNOWLEDGEMENTS

I would like to express my gratitude to my supervisor Dr Jan Lemmen for helping me achieve this MSc thesis.

Statement of Originality

This document is written by Student Bence Dukát 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

In this paper, I investigate the role of hedge funds in the risk transmission among four sets of financial institutions. I gather daily stock returns and create different indices which represent the commercial and investment banking, the insurance, and the hedge fund sector. I use a sample period from April 2003 to December 2015 which I divided into four sub-periods in order to examine the evolution of the degree of systemic risk posed by the hedge fund industry. I calculate the Value at Risk (VaR) for each index and regress the different VaR measures on the VaR of the other three sets of institutions. The results of this thesis underpin the growing concerns with regards to the hedge funds’ contribution to the systemic risk. The results suggest that during the recent financial crisis hedge funds were the most responsible for the risk transmission among the four sets of financial institutions. One percentage point change in the VaR of hedge funds is estimated to increase the VaR of investment banks and insurance companies by 1.9 and 1.4 percentage points, while decrease the VaR of commercial banks by 1 percentage point, respectively. After the crisis, the coefficients on hedge funds decreased remarkably but remained the highest among the financial institutions considered in the sample. One percentage point increase in the VaR of hedge funds leads to 0.61, 0.73, and 0.28 percentage point increase in the VaR of investment banks, commercial banks, and insurance companies, respectively. Even though these results may seem to be economically small, an upcoming crisis would shift these coefficients upwards. Since the risk of financial institutions does not appear to be independent of each other under normal periods, it can be concluded that the systemically important and large institutions continue to be highly interrelated. Therefore, new measures need to be implemented in order to prevent future crises.

Keywords: Hedge funds, systemic risk, failure contagion, spillover effect, financial regulation, financial interconnectedness, Value at Risk (VaR)

(4)

List of tables

Figure 1. Interconnectedness of financial institutions ... 17

Table 1. List the of institutions considered in the sample ... 25

Table 2. Summary statistics ... 27

Figure 2.The evolution of returns ... 29

Table 3. Return correlation matrix ... 32

Table 4. Summary for the variance inflation factor calculations in the return regressions ... 35

Table 5.Regressions on the returns ... 39

Table 6. Regressions on the VaRs ... 44

Table 7. Regressions on the VaRs of the equally-weighted indices ... 50

Table 8. Regressions on the VaRs of monthly indices ... 51

Table 9. Regression with standard deviations. ... 58

Table 10. Summary for the variance inflation factor calculations in the VaR regressions ... 59

(6)

6

1. Introduction

The crisis of 2007-2009 has made systemic risk a priority on the agenda of policymakers. Before the crisis, the microprudential approach was the common risk measure in which regulators tried to address the risk of individual financial institutions and prevent their collapse. The legislations under this approach aimed to regulate the entities separately but did not take into account the potential interconnectedness of these institutions. After the series of failures of financial institutions during the credit crunch, regulators and researchers got aware of the shortcomings of this approach. Nowadays, there is a growing agreement about the need for the reassessment of traditional regulatory practices. As opposed to the microprudential method, the macroprudential approach looks on the financial system as a whole. It recognises the threat posed by the risk exposure of financial institutions to each other. The essential of the macroprudential view is that a common shock can affect financial institutions simultaneously, thereby, they can fail at the same time. The difficulties of one institution can spread to other institutions and to the entire financial system which can accelerate the degree of impairment relatively quickly.

1.1. Systemic risk

While many factors can cause systemic risk, broadly, two different channels can be distinguished. First, since financial institutions are heavily interconnected, a shock to one part of the financial system can easily spill over to other parts. Second, financial institutions tend to undertake similar activities, or display homogeneity in other dimensions (such as their risk management systems), which may amplify the impact of common shocks. In order to avoid a repeat of the crisis, regulators are now redesigning financial regulation to address systemic risk.

A major challenge for this is the measurement of systemic risk. Since systemic risk can arise from several sources, a popular approach is to focus on an institution's overall systemic risk, as reflected in market prices. There exist now a variety of market-based measures, such as the Distressed Insurance Premium (Huang et al., 2012), the Conditional Value at Risk (CoVaR) by Adrian and Brunnermeier (2009), the Marginal Expected Shortfall (Acharya et al., 2010), the

(7)

7

State-dependent Sensitivity VaR (SDSR) by Adams et al (2014), or the Systemic Risk Indices (SRISK) by Brownlees and Engle (2015).

Many of these measures are essentially co-variances between a bank’s stock price and a stock (or banking sector) index, conditional on a tail realisation in the system. The conditional tail expectation measures the expected value of the loss that can take place above a given probability level.

1.2. Hedge funds

As well as other financial institutions, hedge funds have also been blamed for their contribution to the recent financial crisis. Hedge funds are alternative investments using pooled funds which are managed by the management company. In general, they have great flexibility with regards to the investment strategy. They make use of high leverage with the aim of generating higher returns than the market. The investments are illiquid, and only wealthy individuals or institutional investors are allowed investing in the funds (Kambhu et al., 2007).

Although the degree of the hedge fund sector’s role in the crisis is still blurred due to the lack of data availability on the industry, hedge funds have drawn the attention of policymakers. It is hard to disagree with the fact that hedge funds can cause adverse effects on other institutions. In general, hedge funds are highly leveraged, and their portfolios consist of illiquid assets which expose the sector to liquidity risk. As commercial and investment banks provide credit to hedge funds, these institutions bear a counterparty risk. If a shock occurs, hedge funds might face funding liquidity risk since banks are less willing to provide additional capital. Consequently, hedge funds have to liquidate some of their assets to meet margin requirements due to the deterioration in the value of the pledged collateral. In case a homogenous shock affects the industry as a whole the series of liquidations lead to sharp declines in the prices, and all the assets are required to liquidate at fire sales prices. Banks that are highly exposed to the hedge funds that have difficulties or already defaulted have to bear substantial losses. This process can lead to an overall credit shortage since banks’ willingness to lend to other market participants declines (Klaus et al., 2009).

(8)

8

Nonetheless, hedge funds have not been subject to any disclosure requirements and registration for an extensive duration. First of all, the original model of hedge funds consisted of only a limited number of investors, mainly wealthy individuals who were required to have a certain level of income and liquidity. As long as these accredited professionals met all of the requirements, they could freely invest in these hedge funds under the oversight of the Investment Advisers Act (Stulz, 2007). Also, as opposed to deposit-taking institutions such as the commercial banking sector, hedge funds are not exposed to bank runs thereby little attention had been given to the hedge funds’ potential contribution to the financial instability. Runs on deposit-taking institutions occur when their customers (depositors) withdraw their deposits simultaneously due to the fear of their banks’ default. It had been considered the greatest risk since the Great Depression. Consequently, policymakers tried to assess this risk directly by regulating mainly the banking sector. However, the crisis showed that runs could also happen on other institutions such as the run on Bear Stearns by its hedge fund clients. Finally, the maturity mismatch of the asset and liability side of the balance sheet, that used to typify the transformation function of the commercial banking sector, was absent in the hedge fund industry. As hedge funds were not systematically engaged in maturity transformation, their default did not cause systemic concerns.

On the contrary, the increasingly growing size and importance of the hedge fund sector made regulatory changes required. Over the past decades, more and more institutional investors joined the clientele of hedge funds, and gradually replaced the wealthy individuals. Institutional investors may force hedge funds to disclose more information about the portfolio that they hold. Due to the increased amount of capital in the pool of investments of hedge funds, hedge funds have to seek for new approaches and investment opportunities which may not be so lucrative. As a result, certain types of investment draw the attention of regulators. For instance, being an activist investor became popular among hedge funds. Activist investors acquire a substantial percentage of the shares of a company by which they can seize control over the board of directors. They engage in strategic decision-making with the aim of making major changes in a company to improve its profitability. However, investor activism is highly regulated in certain countries thus hedge funds that follow this investment strategy should be subject to regulations as well (King and Maier, 2009). Building on the ideas that the hedge funds’ assets under management have increased significantly and the business model has

(9)

9

been shifted towards institutional investors, hedge funds have got involved in more and more deals with systemically important institutions. As a consequence, hedge funds may pose a major threat to our financial system.

With these facts in mind, the need of implementing new legislations became apparent during the crisis. Hence, policymakers aimed to alleviate the contribution of hedge funds to the systemic risk and prevent potential costly failures in the future. Therefore, a number of changes have been made regarding the legislation of hedge funds in order to mitigate the riskiness of the industry. The most notable changes were introduced by the implementation of the Dodd-Frank Act in the US (Kaal, 2011). The act obliges hedge funds to register and disclose information. The act also tries to address systemic risk by means of distinguishing small and systemically-important institutions since the compulsory registration depends on

the size of assets under management (Claessens and Kodres, 2014).

In the European Union, the Alternative Investment Fund Managers Directive (AIFMD) as a common law was implemented in 2011 aiming to regulate hedge funds, private equities and other alternative investment funds. The law obliges these institutions to disclose information on a regular basis and possess all the required authorizations. Hedge funds are obliged to register with government agencies, advise investors and regulators as well as to meet certain capital requirements. The implementation of the law at a national level was due to 2013. The AIFMD took effect in the Netherlands in 2011. Managers must fulfil certain administration requirements which comprise the valuation of the portfolio, the controlling of disclosures, and the provision of consultancy services (EVCA, 2015).

Moreover, several researchers such as Chan et al. (2006), Lo (2008), Boyson et al. (2010), Dixon et al. (2012), Adams et al. (2014) and Zykaj et al. (2014) among others have tried to examine the hedge fund industry’s role as potential source and transmission channel of systemic risk. In line with previous findings, the spillover effect should be assessed in both “normal” and “shock” periods. The existing literature mainly focuses on the mortgage crisis period. However, the time frame between 2007 and 2009 was very irregular and therefore may not be relevant for “normal” market conditions. 2007 and 2008 were bubble years with very high and fast economic growth while the second half of 2008 and 2009 have been years of extreme

(10)

10

recession. Investigating the period prior to the crisis and the years that followed the recession could give better insights into the spillover effects of hedge funds during more regular market conditions as well as the evolution of systemic risk over time. As modern financial institutions became highly interrelated, the failure of hedge funds can affect other financial intermediaries, and amplify the meltdown of the global financial system (Billio et al., 2012).

1.3. Method and motivation

In this paper, I gather daily stock returns for commercial banks, insurance companies, investment banks, and hedge funds and create indices for each category. The next step is to calculate the Value at Risk (VaR) for each index. VaR is a widely used risk measure by regulators and financial institutions. It is able to express the amount of risk taken in one figure, and it is easily interpretable. I apply four sets of regressions in which the VaR of institution k becomes the dependent variable while the VaRs of the other three sets of institutions become the independent variables. This method allows measuring not only the extent to which financial institutions are exposed to each other but the direction of the spillover effects as well. Furthermore, I divide the sample into three sub-periods which allow assessing the evolution of spillover effects over time.

Since numerous analysts have already warned of an upcoming crisis, the scrutiny of the effectiveness of the existing regulation is essential. The results may provide a better understanding of the extent to which financial institutions are exposed to each other and the evolution of systemic risk over the past decades as well as find additional reasons to consider new regulations imposed on the hedge fund sector.

1.4. Results

The results suggest that spillover effects increase considerably when a common shock occurs such as the recent credit crunch. Moreover, the risk that arises from hedge funds is the highest among the four sets of institutions. Under volatile market conditions, one percentage point change in the VaR of hedge funds is estimated to increase the VaR of investment banks and insurance companies by 1.9 and 1.4 percentage points, while decrease the VaR of commercial banks by 1 percentage point, respectively. After the crisis, there is a drop in the coefficients on hedge funds which can be explained by the relatively stable market conditions. Yet the

(11)

11

magnitude of the coefficient on hedge funds remained the highest among the four sets of institutions. One percentage point increase in the VaR of hedge funds leads to 0.61, 0.73, and 0.28 percentage point increase in the VaR of investment banks, commercial banks, and insurance companies, respectively. Despite the economically small coefficients, the spillover effects would be higher during times of financial distress. Since the financial institutions appear to be exposed to each other during normal market times as well, the existing legislation should be reconsidered.

In the next section, I review the literature on systemic risk and hedge funds while presenting the different studies that aimed to measure the systemic risk arising from the hedge fund industry. In addition, I discuss the different channels through which hedge funds affect other financial institutions. In section 3, I present the data collection and the summary statistic of the variables used in the regressions. Section 4 describes the methodology used in this paper; in Section 5 I provide the empirical results and the conclusions. Sections 6 and 7 contain the robustness check for the results and the conclusion part of the paper, respectively.

(12)

12

2. Literature review

2.1. The concept of systemic risk

Systemic risk is a complex concept that several researchers have tried to define. Chan et al. (2006) describe systemic risk as a series of correlated defaults that occur over a short period which is purely connected to financial institutions. Billio et al. (2012) share the view of this finding by drawing a parallel between systemic risk and the financial sector but neglecting to analyse the effect on the aggregate economy.

De Bandt and Hartmann (2000) discuss the different components of systemic risk in order to develop a broad concept that can be used as a reference point for policymakers to maintain financial stability. The authors claim that the definition of systemic risk has to comprise failure contagion of the financial institution and the spillover effects in the financial markets as well as the related settlement and payment risk. Broadly speaking, the concept integrates collective shocks leading to contemporaneous financial instabilities. The paper concludes that the vulnerability of financial institutions may be higher than other industries. Beside the different characteristics of banks’ balance sheets and difficult financial contracts, the researchers highlight the interconnectedness of financial institutions that may make the sector more sensitive to systemic risk.

Archarya et al. (2009), apart from the impairment of financial sector, try to explain the importance of systematic risk by also highlighting the possible consequences for the real economy. They define it as an event that stems from the widespread defaults and losses of financial firms which are somehow connected to each other. Difficulties in one institution can spread to other organisations and the entire financial system leading to major disruptions in the overall economy. Firms might have difficulties in financing their operation due to lack of credit sources available in the market which may result in the slowdown of economy growth.

Lately, Smaga (2014) aimed to define the concept of systemic risk by analysing the existing literature on the topic. Also, the author points out the factors that play key roles in the evolution of systemic risk and the spreading of the spillovers among firms. Smaga suggests defining the concept as the risk that emerges as a significant realisation of disruptions after a

(13)

13

shock that damages the stability of the financial system leading to an adverse effect on the real economy. As the author explains, contagion cannot evolve by itself, but an initial shock is needed. Under a shock, these transmissions amplify between financial institutions to a level that one would not expect under normal market circumstances. Like many others, Smaga also stresses the vulnerable characteristic of the banking sector to systemic risk.

Khambhu et al. (2007) define it as a state which emerges under a common shock affecting negatively different financial organisations or other market participants simultaneously. The authors provide insight into the threats of systemic risk in terms of the whole economy. According to them, the loss of confidence in the entire financial system jeopardises the financial stability that might lead to a general welfare loss. They explain that regulators historically paid more attention to the banking sector to prevent costly bank runs. As runs on deposit-taking banks were the main reasons for the Great Depression, policymakers tried to reduce the risk of credit shortages by regulating the commercial banking sector. However, financial deregulation and the shift towards active trading strategies gave rise to the hedge fund industry whose emergence has also contributed to the need for new risk measures. The subprime mortgage crisis showed the significance of contagion through correlated exposures, contagion through asset prices and fire-sales, and direct exposures. (Brunnermeier, 2009).

2.2. The growing concerns about the hedge fund industry

Hedge funds as unregulated and highly leveraged financial institutions, their potential contribution to financial instability has drawn increased attention in the recent years. Hedge funds are historically made of investments provided by institutions and high-net-worth individuals. A special feature of the industry that only professional investors are allowed to invest in the funds, consequently hedge funds are not subject to any disclosure requirements which leads to the lack of information available in the industry.

Bernanke’s speech (2006) that was addressed to the Federal Reserve raised concerns about the limited information available on the hedge fund industry. According to him, authorities cannot properly assess the degree of systemic risk. In order to prevent a crisis such as the collapse of Long-Term Capital Management in 1998, authorities must implement future

(14)

14

legislations aiming to ensure the stability of the financial system by enforcing hedge funds’ disclosure requirements.

Although hedge funds play an indisputably useful and important role in the economy, the academic literature mainly agrees that the increasing importance of the sector may pose a threat in the future (Khambu et al. (2007); KPMG, AIMA, MFA (2015); Chan et al. (2006); Lo (2008), among others). Hedge funds usually provide liquidity to illiquid markets by purchasing securities that are momentarily less attractive for other investors. Thereby, they fill a gap in the market and improve market efficiency. Hedge funds also play a key role in diversification, and they somewhat represent an alternative universe of investment opportunities. For instance, they can take positions in equities both the long and short side or invest in fixed incomes without being exposed exceedingly to interest rate changes.

Having said that, the potential for market disruptions that may stem from hedge funds has enlarged over the past decades. Chan et al. (2006) investigate the contribution of hedge funds to systemic risk by examining their risk-and-return profile. Due to the unavailability of hedge fund data such as the degree of leverage, product mix, and credit exposures they cannot assess the magnitude of systemic risk. However, they find that the banking sector bears the risks posed by hedge funds. Furthermore, these risks are non-linear and hard to measure.

The recently published report by KMPG (2015) raises concerns by analysing future challenges and prospects of the industry and provide additional reasons to regulate the sector. As the report shows, the hedge fund industry has grown roughly by 10 percent since the crisis, and it is estimated to grow continuously in the upcoming five years, achieving USD 4 trillion total fund size by 2020.

Moreover, institutional investors such as endowments, corporate pension funds, etc. continue to take over the leading role of the larger hedge funds’ investor base from wealthy individuals since their contribution accounts for two-third of the committed capital. The shift towards institutional investors requires an entirely new approach of regulation imposed on the sector (E&Y, 2015).

(15)

15

According to the findings of Billio et al. (2012), who examined the interconnectedness of financial intermediaries, commercial and investment banks, hedge funds, and insurance companies have become more interrelated in the past decades since they tend to engage in related activities. For instance, hedge funds started to cooperate with financial intermediaries such as prime brokers. As hedge funds need to finance their trades through collateralized borrowing, they began to attract commercial and investment banks’ capital which exposed the banks to counterparty credit risk.

In contrast, several studies do not find any evidence that the hedge fund industry generates a major spillover risk. For instance, the research published by Dixon et al. (2011) shows that hedge funds did not contribute to the systemic risk in the last crises such as the Dotcom bubble or the mortgage crisis. Zykaj et al. (2014) examine the nature of the distribution of hedge funds returns. The findings of the research suggest that spillovers caused by exogenous liquidity shocks cannot explain the correlation across hedge fund returns entirely. Consequently, there are other factors that may influence hedge funds returns rather than liquidity shocks, and they may follow a heterogeneous pattern. These controversial findings require further studies to be done to clarify and assess the hedge funds’ impact on systemic risk under stressed market conditions.

Due to the fact that hedge funds are highly leveraged and not restricted with regards to trading strategies, they can potentially disrupt the financial markets. The paper published by Brunnermeier and Pedersen (2009) points out the significance of funding and market liquidity during a crisis. Hedge funds are particularly exposed to both. Market liquidity determines how easily hedge funds can trade since they tend to hold relatively illiquid portfolios. Also, they operate with a high level of leverage thus they rely strongly on the amount of credit supply. Market participants, especially traders play a crucial role in providing liquidity for capital markets. However, the amount of liquidity that they can provide depends on the available funding sources that they can obtain. Nevertheless, the level of funding is strongly dependent on market liquidity. The deterioration or drying up of one of these liquidity sources can lead to a liquidity spiral thereby causing a liquidity shortage in the markets.

(16)

16

In times of financial distress, hedge funds may face funding liquidity risk through increased margin calls or deterioration in their asset prices. If banks require hedge funds to liquidate additional positions to meet the margin requirements, they have to liquidate assets at a fire-sale price which stresses the prices further leading to liquidity spirals (Brunnermeier and Oehmke, 2012). Hedge funds also provide liquidity to illiquid markets. Hence, their failure may result in a liquidity dry-up for least attractive asset classes (Bernanke, 2006). Investment banks are also heavily exposed to the hedge fund industry due to their active presence in derivatives markets. Nonetheless, hedge funds tend to short these positions by funds obtained from investment and commercial banks which may increase the spillover risk among the participants (Brunnermeier and Pedersen, 2009).

2.3. The complex network of financial institutions

The first empirical study on systemic risk can be attributed to Chan et al. (2006) who measured the spillover effect using logistic regression analysis and nonlinear factor models and addressed the illiquidity risk exposure of different financial institutions. They found that the risk posed by the hedge fund industry had risen significantly causing contagion effect to the banking sector. Their findings also suggest that apart from the fact that hedge funds are systemically important, the significance of the sector is expected to increase in the future.

Gropp (2014) emphasises the increasing importance of the hedge fund industry and the central role that it played in the recent crisis. Like many other financial institutions, hedge funds also took an active part in investing in the complex instruments which made the financial system a complicated and opaque network. The study emphasises that the credit crunch shed light on the interconnectedness and spillover effects between different financial institutions in which hedge funds might have played a dominant role.

The interconnectedness can be illustrated perfectly by taking a closer look at the events during the credit crisis (Brunnermeier, 2009). The financial crisis started from the growing defaults in subprime mortgages. The majority of financial institutions started pursuing higher profits and more lucrative deals as they invested heavily in risky securities (the lower tranches of collateralized debt obligations (CDOs) or other asset-backed securities) in order to achieve higher returns. Banks were holding these risky securities through their special purpose

(17)

17

vehicles (SPV) and internal hedge funds which used credit default swaps to insure against the defaults of these securities. As the number of defaults rose sharply, the cost of insurance increased simultaneously triggering margin calls. Since hedge funds relied mostly on external capital, they had to borrow additional money to pledge it as collateral. As the financial situation worsened, banks were less willing to provide capital. In the same time, the market lost its confidence in the work of rating agencies. Due to the increased pressure, rating agencies started to downgrade the mortgage-backed securities’ (MBS) ratings which triggered further losses as the prices of securities begun to shrink gradually. The market of MBSs finally became totally illiquid where nobody wanted to buy these securities but sell.

Figure 1. Interconnectedness of financial institutions

Meanwhile, banks were extremely exposed to their “conduits” (since the risk with regards to subprime mortgages never left the banking industry) the failures of these SPVs affected the banking sector adversely. Hedge funds that experienced margin calls started to redeem the securities and withdraw the investments and capital that they had invested in their prime brokers, investment banks. After the rescue of Bearn Sterns, the government decided not to bail out Lehman Brothers which defaulted in September 2008. The default of one of the biggest investment banks triggered a series of negative events. While the banks’ write-offs and the downgrading of securities continued, one of the oldest Money Market Funds (MMF) which had significant exposure to Lehman, the Prime Reserve fund broke the buck first time

(18)

18

in its history. MMFs are considered to be one of the safest investment categories since they promise to maintain a net asset value of $1 per share.

At the same time, AIG one of the largest insurance companies announced its need for external capital in order to prevent failure. AIG was heavily exposed to the mortgage market through the issuance of insured derivatives against default (CDS) but did not purchase reinsurance to hedge that risk. It also traded MBSs actively by using collateral. When the events of defaults started, AIG was obliged to pay out the claims on CDSs and also to provide additional capital for banks to meet margin calls on MBSs. As leveraged traders like broker-dealers at investment banks and hedge funds could not obtain additional funds from the market, they had to liquidate their positions. Due to a lot of illiquid securities that emerged in the market over a short period of time the prices were depressed further. The term “broker-dealer” refers to the investment banking sector from now on.

The most common measure of risk used by financial institutions is the Value at Risk (VaR). VaR measures the potential loss of an investment with a given probability within a certain period such as a day. Nonetheless, the crisis showed its shortcomings since VaR focuses on the risk of the individual institutions but does not take into account the interconnectedness. In addition, VaR is not able to handle longer periods of distress such as a month. Since the crisis lasted almost two years, the VaR measures underestimated the risk and the losses that occurred in a prolonged distressed market. Several researchers such as Adrian and Brunnermeier (2009), Adams et al. (2014), etc., have tried to address these VaR problems recently by examining its behaviour during the recent crisis and improving it towards a measure which also considers the interconnectedness of the financial institutions.

2.4. Empirical studies on systemic risk

The methodology of the thesis is related to the state-dependent sensitivity VaR (SDSR) approach proposed by Adams, Füss and Gropp (2014). It aims to develop the conditional value at risk approach (CoR) of Adrian and Brunnermeier (2009) further. While CoR model tries to assess the distributions of the return of different financial institutions by using quantile regression, the SDSR shifts the focus to the distribution of VaR. One of the main differences between the two papers is that the CoR method uses low and fixed quantiles thereby it is not

(19)

19

able to model the different states of the economy. The VaR varies depending on the financial situation of a firm. If a shock occurs for a brief period of time but the market is under normal condition in general, institutions with strong financial background are able to mitigate losses. In this case, the VaR is less volatile and takes up positions in the higher quantiles of the distribution. Consequently, the usage of different quantiles allows investigating the behaviour of VaR under different scenarios.

The SDSR reveals the magnitude of spillover risks among financial institutions, thus enables to investigate the increased significance of hedge funds’ role in contagion. It distinguishes three different periods (tranquil, normal, volatile), and sets the quantiles of the VaR distribution in the regressions accordingly. The model measures statistical relationships using the stock returns of financial firms rather than risk exposures by analysing balance sheet items, given the absence of financial reports provided by the hedge fund industry. Based on their results, commercial banks and, mainly, hedge funds are responsible for shock transmission between different financial institutions. In addition, they find that the contagion effect is even higher during volatile markets such as the recent subprime mortgage crisis.

CoR and its extensions became a common way for researchers to measure systemic risk and the contagion effect. The study of Wong and Fong (2010) analyses the interconnectedness of the banking sector in eleven Asia-Pacific countries using the CoR method. Their findings suggest that there is a significantly increased sovereign risk if the banking sector is in distress in one of these economies.

Gauthier, Lehar, and Souissi (2012) assess the level of systemic risk that the Canadian banking system is exposed to. They also measure how the idiosyncratic risk of the amount of banks’ capital can affect aggregate systemic risk. The study is rather connected to research on capital requirements. Nonetheless, it finds evidence that capital requirements can reduce default probabilities thus decreasing the likelihood of a systemic crunch.

The use of quantile regression stems from the Conditional Autoregressive Value at Risk (CAViaR) model proposed by Engle and Manganelli (2004) which takes into account the time-varying evolution of quantiles using a Generalized Autoregressive Conditional

(20)

20

Heteroscedasticity (GARCH) model. The study aims to gauge the probability of contagion effect as opposed to the size of the effect, thereby it is not an adequate measure to identify the magnitude of the contribution of different sectors.

Boyson, Stahel, and Stulz (2010) also use quantile regression to gauge contagion effects of different hedge fund strategies. They find evidence of contagion among hedge fund styles. The contagion effect amplifies when a large adverse shock to credit spreads occurs, the market prices of stocks deteriorate, or an unexpected decline in the stock price of financial institutions occur.

Lately, Acharya, Pedersen, Philippon, and Richardson (2010) proposed the marginal expected shortfall (MES) as a new way to measure systemic risk. Rather that the contribution to systemic risk, the MES assesses the extent to which financial institutions are exposed to systemic risk. They show that the exposure increases in the level of leverage.

Brownlees and Engle (2015) extend the approach above and introduce the conditional capital shortfall measure (SRISK Index) to assess systemic risk. The method measures the capital shortage of a firm that it would experience when a crisis was to occur. The SRISK appears to provide a useful framework to predict future difficulties of different market participants under distress and enables to absorb signals of a potential upcoming crisis.

Billio, Getmansky, Lo, and Pelizzon (2012) measure correlations of market returns among banks, insurance companies, brokerage firms and hedge funds and apply a Granger causality test to assess the systemic risk of these four sectors. Their results are in line with previous studies and show that the liquidity of these industries has decreased, and they become more interconnected over the past decade. Although hedge funds are essential components of systemic risk, their result suggests that the contribution of banks, insurance companies, and brokerage firms are more significant as those of hedge funds.

Finally, Huang, Zhou, and Zhu (2012) measure systemic risk contribution of individual banks by assessing the insurance premium paid against financial distress by the financial sector.

(21)

21

Based on their results, the contribution to the systemic risk shows a linear relationship with regards to the default probabilities of the institutions but nonlinear with regards to the magnitude of firms’ assets.

2.5. Motivation for the thesis

Several efforts have been taken to gain a better insight of systemic risk arising from various sources. Systemic risk can be interpreted either in a broader or a narrower view, but most of the studies underline its increasing magnitude and importance due to the emergence of new financial products and the changing structure of the financial industry. Also, some studies tried to gain a better understanding of the hedge fund industry as a whole, and there are researches underway to provide some implications regarding its role in systemic risk. While the majority of experts acknowledge the potential disruptions that the hedge fund industry may cause, there is no commonly used methodology which can be applied to measure the contribution of different sectors to systemic risk.

The financial crisis undermined the confidence in the financial sector. In the aftermath of the crisis, existing legislation had to undergo major changes in order to rebuild investors’ and markets’ confidence. As authorities started to initiate fundamental reassessments, the hedge fund industry has also come under close scrutiny recently. Hedge funds, whose assets under management exceed the given thresholds under the Dodd-Frank Act, became subject to

disclosure requirements and leverage restrictions (Claessens and Kodres, 2014). Also, the EU

implemented the AIFMD, which obliges hedge funds to obtain authorization from governmental agencies and provide information with regards their investments (EVCA, 2015).

The thesis aims to contribute to these recent works on spillover effect and systemic risk, by assessing market events in a highly current time frame in the context of hedge funds. The results may elucidate the role of hedge funds in risk transmission and its evolution over the past decades along with providing additional proofs for the need of legislation imposed on the sector.

(22)

22

3. Data

3.1. Data collection

Daily stock returns for publicly traded insurance companies, commercial and investment banks in the US are collected by using the University of Chicago’s Center for Research in Security Prices (CRSP) database. Daily stock returns are required since spillover effects can occur over a short period a time. Moreover, it allows measuring the immediate effects which take place overnight. The period of study is from 2003Q2 to 2015Q4.

In this paper, the extended period can be viewed as a novelty since the existing literature mainly examines either the crisis or an earlier period. Meanwhile, policymakers have implemented a series of new regulations; the most current timeframe should also be investigated to assess whether these changes have a significant impact on the role of systemically important financial institutions in systemic risk. Lately, regulators recognised the importance of macroprudential regulation which desires to alleviate the risk of the financial system as a whole. Before the crisis, policymakers mainly put emphasis on the risk profile of individual institutions. Nonetheless, the crisis showed that risk factors are endogenous thus depend on the complex network within the financial system (Hanson et al., 2011).

The thesis aims to provide further insights into the evolution of systemic risk which stems from various financial institutions, for this reason, the dataset is divided into three periods. The years between April 2003 and November 2007 correspond to normal market conditions before the crisis, December 2007 – July 2009 represents the crisis with years of extreme recession. Finally, August 2009 – December 2015 is the recovery period after the crisis. Since daily hedge funds returns are available from April 2003, the dataset starts from this date. The crisis period is chosen according to the NBER based Recession Indicator Index. The latest time period is the main interest of the research which is supposed to demonstrate the effect of different legislations implemented on the financial sector as well as the realignment of the interconnectedness of financial institutions.

(23)

23

The downloaded dataset is restricted based on the Standard Industrial Classification codes. SIC 6211 and 6282 represent the investment bank industry; SIC 6020-6030 represent the commercial banks while SIC 6300-6400 represent the Insurance industry, respectively.

The next step is to collect daily hedge fund returns. Global Hedge fund indices are gathered from the HFRX hedge fund database. The HFRX Global Hedge Fund Index represents an overall composition of the hedge fund industry. The strategies are value-weighted based on the asset

size of each hedge fund. It involves all of the strategies used by hedge funds such as event

driven, equity hedge, convertible arbitrage, merger arbitrage, and so on. The index is constructed from the self-reported returns of the largest hedge funds with a minimum of $10 million in assets under management (AUM).

3.1.1. Potential bias in the hedge fund dataset

Even though the HFRX database is one of the most reliable databases, there are certain biases that have to be considered. One of the main concerns is connected to the method of index construction. Survivorship bias arises when dead funds or funds that stopped providing data are eliminated from the database. The problem is that these hedge funds tend to perform worse thereby removing them makes the dataset biassed upwards.

Instant history bias relates to new funds that joined the sample. Since managers tend to report the figures only when their hedge funds have entered a successful phase and had established a proven track record. In this case, managers do not report to the database as soon as they are established but wait until they can provide decent numbers.

Finally, selection bias relates to the fact that the numbers might be manipulated since managers are in charge of reporting directly. There is no existing legislation which obliges hedge funds to disclose their financial reports publicly. The voluntary reporting structure may lead to upward bias since hedge funds prefer disclosing more attractive figures in order to find more investors (Gropp, 2014). Yet given the limited data available on the hedge funds industry such as the assets that individual hedge funds own or the return of individual hedge funds, it requires using these databases which give some limitations to the research.

(24)

24

The motivation for the collection of the stock returns of these institutions stems from the underlying reasons that led to the recent crisis. During the credit crunch, losses spread from asset-backed securities and collateralized debt obligations (CDOs) to different financial institutions. The losses affected the returns of hedge funds, investment banks, and commercial banks. As a consequence, they triggered a series of events which lead to a loss spiral among the institutions whose portfolio was heavily exposed to deterioration of those securities. Although insurance companies historically did not engage in businesses with the hedge fund or investment banking industry, their business model has changed remarkably over the past decades. Even though they do not trade directly with these institutions, they started insuring against the complicated financial products such as the mortgage-backed securities and CDOs by issuing credit-default swaps. The shift of their core business models towards generating higher returns by taking higher risk gave rise to the potential implication for the contribution to systemic risk (Bllio et al., 2011).

By the same token, the traditional banking model of deposit-taking institutions was replaced by the originate-and-distribute model. In order to meet the capital requirements, they securitized the loans and passed them on to their special purpose vehicles which in turn sold these instruments to different institutional investors thereby creating a complex network of the representatives of the financial sector.

In addition, daily Nasdaq 100, MSCI US REIT Index returns are collected and used as control variables in the regressions. Nasdaq 100 represents the returns of non-financial institutions and controls for the market risk. As the crisis showed, financial institutions are highly sensitive to the movements in real estate prices hence the REIT index is supposed to control for these effects.

After this, the years of operation for each institution are calculated, and all of the observations with less than six years of consecutive operation are dropped. This step is required to filter out institutions which were established after the crisis. Since the spillover effects amplify during volatile market periods, the years of crisis should be included as well.

(25)

25

The next step is to calculate the total average market value by the four sectors. As systemic risk mainly stems from systemically important institutions which can be measured by the size of firms, all observations with a market value lower than the calculated average of total market value are dropped. After restricting the observations, several institutions are added to the sample based on the sample used by Acharya et al. (2010).

There are 78 institutions left in the sample, namely, 16 investment banks, 33 commercial banks and 29 insurance companies which are considered systemically important due to their size.

Table 1. List the of institutions considered in the sample

The table contains the names of the institutions used in the sample. The first column comprises the names of investment banks; the second comprises the names of commercial banks while the last column shows the insurance companies. In total, there are 78 institutions in the sample.

Investment banks (16) Commercial (33) Insurance (29) AMERIPRISE FINANCIAL INC B B & T CORP A F L A C INC AMVESCAP PLC BANK MONTREAL QUE ACE LTD BEAR STEARNS COMPANIES INC BANK NEW YORK INC AETNA INC NEW BLACKROCK INC BANK OF AMERICA CORP ALLSTATE CORP

BLACKSTONE GROUP L P BANK OF NEW YORK MELLON CORP AMERICAN INTERNATIONAL GROUP INC E TRADE GROUP INC BANK OF NOVA SCOTIA ANTHEM INC

EDWARDS A G INC CANADIAN IMPERIAL BANK COMMERCE AON CORP

FRANKLIN RESOURCES INC CITIGROUP INC BERKSHIRE HATHAWAY INC DEL GOLDMAN SACHS GROUP INC COMERICA INC C I G N A CORP

INVESCO LTD CREDICORP LTD C N A FINANCIAL CORP LEHMAN BROTHERS HOLDINGS INC DEUTSCHE BANK A G CHUBB CORP

MERRILL LYNCH & CO INC FIDELITY BANCORP INC DEL CIGNA CORP

MORGAN STANLEY DEAN WITTER & CO HSBC HOLDINGS PLC COVENTRY HEALTH CARE INC PRUDENTIAL FINANCIAL INC HUDSON CITY BANCORP INC GENWORTH FINANCIAL INC SCHWAB CHARLES CORP NEW HUNTINGTON BANCSHARES INC HARTFORD FINANCIAL SVCS GRP INC T ROWE PRICE GROUP INC ICICI BANK LTD HEALTH NET INC

J P MORGAN CHASE & CO HUMANA INC

KEYCORP NEW LINCOLN NATIONAL CORP IN M & T BANK CORP LOEWS CORP

MARSHALL & ILSLEY CORP MANULIFE FINANCIAL CORP NORTHERN TRUST CORP PRINCIPAL FINANCIAL GROUP INC P N C FINANCIAL SERVICES GRP INC PROGRESSIVE CORP OH

REGIONS FINANCIAL CORP PRUDENTIAL PLC ROYAL BANK CANADA MONTREAL QUE ST PAUL COS INC STATE STREET CORP SUN LIFE FINANCIAL INC SUNTRUST BANKS INC TRAVELERS COMPANIES INC SYNOVUS FINANCIAL CORP UNITEDHEALTH GROUP INC TORONTO DOMINION BANK ONT WELLPOINT INC

UBS AG X L GROUP PLC

U S BANCORP DEL

WACHOVIA CORP 2ND NEW WELLS FARGO & CO NEW ZIONS BANCORP

(26)

26

3.2. Index creation

After that, two types of indices are constructed as proxies aiming to represent each sector. First, equally-weighted indices are created which are rebalanced daily. The weights used to calculate the equally-weighted indices, add up to one. Thus, these indices do not take into account the market value of each institution.

Equation for indices

Ʃi Weigth(i, t) ∗ return(i, t) Ʃi Weigth(i, t)

Due to the fact that the hedge fund indices are asset-weighted, the next step is to create daily value-weighted indices by categories that are calculated as the sum of the daily market value of each institution in a given category times the return, divided by the sum of the market value (Chan et al., 2007).

3.3. Summary statistics

Table 2. represents the descriptive statistics of the variables. The table is divided into four sub-tables that illustrate the full sample and the three time periods that are investigated in the thesis. The average daily returns of the indices are reported in column 2, which contains both equally and value-weighted indices. The summary statistics of the full sample shows that investment banks generate the highest return on average among financial institutions which are followed by insurance companies, commercial banks, and hedge funds, respectively. It has to be mentioned that in some of the cases banks and insurance companies such as AIG, Wells Fargo, Citigroup or Bank of America received a capital investment from the government. Since the bailout was needed in order to prevent the failures some of the big institutions, the returns might be biassed which belongs to the few limitations of this research.

As expected, the tables illustrate that the value-weighted indices appear to be somewhat less volatile than the equally-weighted indices. The lower volatility of former indices is due to the fact that larger institutions are more diversified and have more capital which helps them to absorb shocks better and leads to the conclusion that the returns of smaller institutions are more volatile. In turn, the higher volatility of the returns of smaller institutions implies that they are not able to react in a smoother way to the changes in market conditions. As the

(27)

27

interest of the research lies in the spillover effect generated by systemically important institutions. Furthermore, the hedge fund indices are asset-weighted, more emphasis is put on the value-weighted indices.

Table 2. Summary statistics

The table represents the summary statistics of the variables. The table is divided into four sub-tables which represent the full period, the period prior to the crisis, the financial crisis, and the period after the crisis, respectively. The number of observations can be seen in the first column while the mean and standard deviation of the returns can be found in the second and third columns, respectively.

Variable Obs Mean Std. Dev. Min Max

Equally-weighted Investment 3213 0.00076 0.02718 -0.17730 1.11559 Equally-weighted Commercial 3213 0.00044 0.01793 -0.13249 0.31048 Equally-weighted Insurance 3213 0.00063 0.01833 -0.12000 0.64280 Value-weighted Investment 3213 0.00052 0.01858 -0.10519 0.26593 Value-weighted Commercial 3213 0.00036 0.01733 -0.15207 0.22002 Value-weighted Insurance 3213 0.00039 0.01195 -0.10304 0.13886 Hedge Fund index 3213 0.00005 0.00180 -0.01764 0.01655 Nasdaq 100 3213 0.00056 0.01353 -0.10519 0.12580 MSCI US REIT index 3213 0.00052 0.0213 -0.1974 0.1878

Equally-weighted Investment 419 -0.00104 0.03972 -0.17730 0.22628 Equally-weighted Commercial 419 -0.00076 0.03745 -0.13249 0.21407 Equally-weighted Insurance 419 0.00073 0.04468 -0.12000 0.64280 Value-weighted Investment 419 -0.00027 0.03737 -0.10519 0.26593 Value-weighted Commercial 419 -0.00023 0.03794 -0.15207 0.22002 Value-weighted Insurance 419 -0.00046 0.02466 -0.10304 0.13886 Hedge Fund index 419 -0.00045 0.00281 -0.01764 0.01655 Nasdaq 100 419 -0.00034 0.02408 -0.10519 0.12580 MSCI US REIT index 419 -0.00040 0.0490 -0.1974 0.1878

1 April 2003 - 30 November 2007

1 December 2007 - 31 July 2009

1 August 2009 - 31 December 2015 Full sample

(28)

28

Taking a closer look at the period before the financial crisis, the table reveals that smaller institutions performed better compared to larger ones. This fact can be seen by comparing the equally and weighted indices as the former is higher in every case than the value-weighted indices. Investment banks were extremely successful prior to the crisis which was one of the main reasons behind the bubble years. Investment banks took an especially active part in the creation and trade of complex financial products. Although this activity was highly lucrative before the crisis, it turned out to be harmful to the economy.

The returns during the recent credit crisis fell sharply and as it is known that all of the financial institutions suffered tremendous losses. Besides investment and commercial banks and insurance companies, hedge funds can be viewed as the biggest victims since the return of the industry changed from 0.021% to -0.045% on average. Nevertheless, it has to be noted that there is a big increase in the standard deviation of the returns of commercial and investment banks as well as the returns of insurance companies. It implies that some of the institutions managed to generate positive returns which somewhat offset the negative performance of the sectors as a whole. The equally-weighted insurance company index also reveals an interesting detail as the mean is positive during the crisis. Compared to the value-weighted index it suggests that a couple of smaller insurance companies were able to be profitable while larger ones were more exposed to the market.

After the crisis, the figures returned to be positive. However, the investment banking sector appears to be losing its competitive edge with regards to returns. The gap between the returns of these three set of institutions declined as they achieved almost the same level of returns. Investment banks are followed by commercial banks, insurance companies, and hedge funds, respectively. The volatility of the returns has also moderated since the crisis. The standard deviation of the value-weighted commercial and investment bank indices seems to be similar while the hedge fund and insurance firm indices have a standard deviation closer to zero which may imply that these sectors have become more stable since the credit crunch.

In conclusion, the returns of financial institutions are strongly dependent on the market conditions. Although investment banks are estimated to be the most profitable during the

(29)

29

whole period, the sector seems to be losing its leading position. Based on the figures, one could also conclude that the gap between returns somewhat narrowed.

Figure 2.The evolution of returns

The graph illustrates the evolution of returns of the financial institutions from April 2003 to December 2015. Each sector is represented by an index. The upper graph illustrates the returns of equally-weighted indices while the lower graph shows the returns of value-weighted indices.

The graph reveals another interesting picture. The value-weighted hedge fund index, which is supposed to represent the various strategies used by the hedge fund universe, shows a relatively balanced structure, that is, it has the lowest standard deviation among financial institutions. As it was mentioned before, hedge funds are not subject to any regulations with

(30)

30

strategies tend to take both long and short positions. Thus these hedge funds may be able to eliminate volatility. In contrast, the event-driven strategy focuses on special occasions such as large mergers, bankruptcies, and so on. The heterogeneous profiles of hedge funds may lead to more or less steady returns at an aggregate level whose volatility is offset by different strategies. This pattern can have important implications regarding investments in a fund of hedge funds. However, this aspect is out of the scope of this paper. Moreover, the returns of hedge funds may suffer from survival bias as it was mentioned before. In this case, defaulted hedge funds are removed from the sample thus the strongly negative returns are not reflected in the index. The graphs of the value-weighted indices also show that the returns of the other three sets of institutions became highly correlated as they moved together during the crisis and remained more correlated in comparison with the period prior to the crisis. Based on the graph, the use of the value-weighted indices also helps to avoid outliers.

3.4. Return correlation

The standard practice of gauging connectedness of firms is to measure return correlations. Nevertheless, this method is not sufficient to examine spillover effects among institutions since it is not able to detect the direction of spillovers, thereby, it cannot serve the purpose of this study. However, return correlation provides a general idea about the co-movements of returns and may strengthen the analysis beyond the regressions.

As a point of reference, Table 3. illustrates the correlations of daily returns of financial firms examined in this paper. As before, four tables are representing the full sample, the pre-crisis period, the crisis, and finally the years after the crisis.

The return correlation among financial institutions increased rapidly between 2003 and 2015. This underpins the existing literature on the interconnectedness of financial institutions, and it is consistent with the findings of Billio et al. (2012) who showed that the financial sector became more interrelated over the past decades. Although the return correlation of the other three sets of institutions increased steadily up to the crisis, the correlation coefficient of hedge funds dropped after 2007. Consequently, hedge funds may have been less responsible for the shock transmission over the period. After the credit crunch, all of the correlation coefficients

(31)

31

remained surprisingly high which may suggest that the institutions continue being dependent on each other. Even though some of the figures show a decreasing trend in co-movements of returns, insurance companies and commercial banks remained the most interconnected institutions among the four set of institutions considered in the sample.

The coefficient on hedge funds also unveils a staggering fact. The hedge fund industry, by its nature, should be kind of independent of other financial institution as they represent the world of unorthodox investment strategies, that is, one should be able to hedge against market downturns by turning to hedge funds. However, the tables imply that hedge funds engage in similar activities to the market in general.

Another line of thought on the correlation coefficient among financial institutionsis that the

strong relationship may lead to multicollinearity. In the case of multicollinearity, the coefficients can take opposite signs as well as the t-statistics can take lower values than the normal ones resulting from higher standard errors. As the method requires all of the variables to use in one regression, the presence of multicollinearity should be assessed. The problem can arise from multicollinearity that the results may not be significant. However, the estimates in the regression will not be biassed.

(32)

32 Table 3. Return correlation matrix

The table represents the correlation matrix of the returns. The table is divided into four sub-tables which represent the full period, the period prior to the crisis, the financial crisis, and the period after the crisis, respectively. The return correlation is computed for the value-weighted indices as the main interest of the research lies in the systematically large institutions.In addition, the HFRX Global Hedge Fund index, the Nasdaq 100, and MSCI US REIT index can be found in the table.

Investment Commercial Insurance Hedge Fund Nasdaq 100 MSCI REIT Value-weighted Investment 1

Value-weighted Commercial 0.8128 1

Value-weighted Insurance 0.7787 0.7831 1

Hedge Fund index 0.4912 0.4 0.4888 1

Nasdaq 100 0.5006 0.4499 0.4852 0.4717 1

MSCI US REIT index 0.4007 0.4219 0.3808 0.2203 0.6754 1 Value-weighted Investment 1

Hedge Fund index 0.5437 0.4993 0.4381 1

Nasdaq 100 0.4209 0.4091 0.3934 0.4395 1

MSCI US REIT index 0.4067 0.4128 0.3248 0.3847 0.4851 1 Value-weighted Investment 1

Hedge Fund index 0.4177 0.2962 0.4517 1

Nasdaq 100 0.493 0.4423 0.4643 0.3447 1

MSCI US REIT index 0.3845 0.4106 0.338 0.0592 0.777 1

Value-weighted Investment 1

Hedge Fund index 0.5924 0.5802 0.6207 1

Nasdaq 100 0.5734 0.5328 0.5891 0.6385 1

MSCI US REIT index 0.455 0.4628 0.5179 0.4454 0.6919 1

1 December 2007 - 31 July 2009

1 August 2009 - 31 December 2015 Full sample

(33)

33

4. Methodology

The objective of this paper is to estimate the spillover effects among financial institutions, thereby, a methodology should be implemented which is able to capture not only the extent to which the returns of financial institutions affect each other but its direction as well. For this reason, it aims to develop a framework by combining two different methodologies proposed by Chan et al. (2007) and Adams et al. (2014). Furthermore, the research examines an extended time period which is separated into three different categories. Assessing these three market conditions is accordance with the findings of Adams et al. (2014) who suggest analysing both regular and stressed economic settings.

Following the methodology proposed by Chan et al. (2007) who examined the systemic risk

by assessing the potential risk exposure of the bank sector to hedge funds, this paper tries to implement a similar regression model. While the authors develop a series of new measurements where they regress bank indices on hedge fund indices, this research topic requires extending the regressions done by Chan et al. (2007) with additional financial institutions. Chan et al. (2007) construct bank indices from stocks with SIC codes 6000 – 6199, henceforth, they do not include insurance companies and investment banks in the sample. The time frame used in their regression is also shorter as it does not cover the recent crisis. Also, it mainly concentrates on different hedge fund strategies rather than investigating the hedge fund industry’s role in risk transmission.

In contrast, the state-dependent VaR (SDSR) model proposed by Adams et al. (2014) uses a broad range of financial institutions to calculate the VaR of each company. The calculation of VaRs is followed by the construction of indices for the insurance, investment banking, commercial banking, and hedge fund sector. In the next step, the authors apply the quantile regression model to capture the spillover effects among the categories by running four sets of regressions. The paper mainly focuses on the recent crisis but does not assess the years leading to the crisis nor the most recent time frame. Analysing an extended time period can provide a better understanding of the evolution of systemic risk as well as enable to draw a conclusion with regards to the individual contribution of each sector. In addition, the periods

(34)

34

are investigated by using value-weighted indices to identify the risk exposure of systematically large institutions.

4.1. Regressions with returns

In order to estimate risk exposure of different sectors, four sets of regressions are created. The indices, representing the investment and commercial banking, insurance, and hedge fund sector, become the dependent variables in the regressions, respectively. The returns of dependent variables are regressed on their own lags and the returns of the other financial institutions. Additional control variables are also used to control for omitted variables bias.

The coefficients on the variables of interest change by regressions that is, R1: (β3, β4, β5), R2:

(β1, β4, β5), R3: (β1, β2, β5), and R4: (β1, β2, β3), respectively.

Regressions

1. R (invest,t) = α + β1 * R (invest,t-1) + β2 * R (invest,t-2) + β3 * R (com,t) + β4 *R (insur,t) + β5 * R(hedg,t))+ ϒ1 * R (control, t) + ut

2. R (com,t,) = α + β1 * R (invest,t) + β2 * R (com,t-1) + β3 * R (com,t-2) + β4 * R (insur,t) + β5 * R (hedg,t) + ϒ1 * R (control, t) + ut

3. R (insur,t) = α + β1 * R (invest,t) + β2 * R (com,t) + β3 * R (insur,t-1) + β 4 * R (insur,t-2) + β5 * R (hedg,t) + ϒ1 * R (control, t) + ut

4. R (hedg,t) = α + β1 * R (invest,t) + β2 * R (com,t) + β3 * R (insur,t) + β4 * R(hedg,t-1) + β5 * R(hedg,t-1) + ϒ1 * R (control, t) + ut

Where,

com: Commercial Banks invest: Investment Banks insur Insurance Companies hedg: Hedge funds

Based on the approach of the SDSR method, the additional simultaneous spillover effects from the other three institutions can be controlled by including the own lags of the dependent

variables. The underlying assumption is that the own lags of Rk have only an effect on Rk, but

do not affect the returns of other institutions. Consequently, two lagged values of each institution are used to address the bias that may arise from omitted factors. All of the lagged values are statistically significant at a one percentage significance level. Thus they serve as valid instruments.

Is there spillover risk between financial institutions after the crisis? : the evolution of systemic risk caused by hedge funds

MSc Business Economics, Finance Track

Master Thesis