The effect of Credit Default Swaps on a firm’s
leverage ratio
Faculty of Economics and Business
Abstract
This thesis aimed to find what kind of effect Credit Default Swaps (CDSs) have on the leverage ratio. This was studied for Standard & Poor’s 500 non-financial firms that have CDSs contracts traded on their debt. The researched period was from 2002 up to and including 2017. The dependent variable leverage ratio was defined as book and market leverage, whereas the explanatory variables were CDS and some firm characteristics control variables. According to the results, when firms have CDSs traded on their debt it increases their leverage ratio significantly for both book and market leverage. This implies that CDSs increase a firm’s credit supply. Thus, potentially stimulating economic growth and a firm’s investments opportunities.
MSc Economics, Monetary Policy and Banking Author: Maarten van Lieshoud
Student nummer: 10291253
Thesis supervisor: S.J.G. van Wijnbergen Date: 15𝑡ℎ August 2018
Statement of Originality
This document is written by Student Maarten van Lieshoud who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are 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.
Contents
1. Introduction
12. Literature review
42.1 Credit Default Swaps 4
2.2 Credit Default Swaps effect on a firm's leverage ratio 5
2.3 Credit Default Swaps informational role 6
2.4 Credit Default Swaps differences with respect to other derivatives 7
2.5 The empty creditor problem 8
2.6 The capital structure explanatory variables 9
2.7 Hypothesis 11
3. Methodology
133.1 Model 1: The effect of Credit Default Swaps on a firm's leverage ratio 13 3.2 Model 2: extended with Credit Default Swaps interaction terms 15
3.3 Endogeneity problem 16
4. Data and descriptive Statistics
174.1 Data sources 17
4.2 Explanatory and dependent variables 18
4.3 Summary statistics 18
5. Results
225.1 The effect of Credit Default Swaps on the leverage ratio 22 5.2 Extended model 2 with the Credit Default Swaps interaction 26
variables integrated
5.3 Robustness of the dataset 28
6. Conclusion
31References
331
1. Introduction
Ever since the recent credit crisis of 2008, Credit Default Swaps (CDSs) have been a subject for discussion among other subjects regarding financial innovations and derivatives. The CDS market has been particularly discussed in heated policy debates due to its sheer size and its potential effects on the economy. Looking at the size of CDSs trading, it has peaked to a notional value of $62.2 trillion. After the peak, a decline for CDSs was observable in 2007. Nonetheless, the disappearance of the market is far away according to the International Swaps and Derivatives Association (ISDA). The ISDA indicates that in 2010, the CDSs notional value of CDSs trading was $26.3 trillion, which is a greater amount outstanding than the recorded $12.4 trillion in 2005. According to Greenspan (2004), credit derivatives contributed in such a manner, in combination with other financial instruments, which led to the development of a financial system that is more efficient, flexible and resilient. Identifying what potential effects these CDSs may have is crucial for understanding the possible influence on the economy. After the 2008 financial crisis, questions arose about the benefits and downsides of CDSs and scholars speculated if CDSs contributed to the collapse of the economy. It did not help that CDSs are not traded in a centralized exchange but rather over the counter (OTC) which made it hard to regulate. Numerous scholars are of the opinion that CDSs may be contributing in a negative manner to the economy. This resulted from the situation that until 2009, trading in CDSs was unregulated in general. However, other scholars, such as Stulz (2010) argue that it was not the CDSs markets that contributed to the financial crisis, but it was rather a consequence of the decline in the housing market.
This means that there are different beliefs regarding what contributed the most to facilitate the financial crisis. However, this thesis is particularly interested in the effects of CDSs on the leverage ratio. This thesis aims to determine if there is a possible increase in the leverage ratio when firms have CDSs traded on their debt, as during the financial crisis of 2008 some evidence implied that this might be the case. For example, Duchin, Ozbas and Sensoy (2010) studied the effect of non-financial firms’ investments with respect to the credit crisis. They found that because of the credit crisis, investments of non-financial firms declined. The firms that especially were hurt, were firms that relied on external financing, such as loans, and firms that suffered from financial constraints. Ivashina and Scharfstein (2010), who studied the effect of expired credit lines, found that when a firm has no access to an active credit supply anymore, their investment activity declines in comparison to firms that still have active credit lines. In addition, Saretto and Tookes (2013) found that CDSs indeed contribute to an increase in the leverage ratio for the period 2002 up to and including 2009. These studies suggest that relaxation of the credit constraints through CDSs may indeed improve the investment opportunities for non-financial firms.
The influence of CDSs on the investment possibilities for firms can be addressed from the viewpoint of capital suppliers. If capital suppliers can hedge the credit risk on a loan, it increases their willingness to provide a loan and thus reducing the supply side frictions. This occurs for, for example, financial institutions that are frequent suppliers of loans and who are more willing to supply credit if they can protect themselves against the associated credit risk. Especially if the holding of the credit risk is not desired by the lenders party, they can shift the risk to a third
2
party and the necessary capital becomes available for non-financial firms (Ashcraft & Santos, 2009). This is the case when treasuries are diminished and are in short supply. So, the presence of CDSs will attract potential investors that are willing to supply the needed debt capital. Consequently, the ability to hedge the credit risk makes it more attractive for these investors to supply the necessary funds. Moreover, the increased provision of credit supply can have stimulating effects on the growth of the economy. Therefore, it is important that the possible positive effects of CDSs are taken into consideration by public policy makers, otherwise policy rules made about CDSs can be counterproductive for the economy. In addition, according to Weistroffer (2009), the extended investment universe created through the availability of CDSs, assists completing markets. However, the empirical evidence regarding the constraints that impede the capital supply to firms, which in turn may possible relax with the availability of CDSs, has not been strong.
Another study involving CDSs was done by Ashcraft and Santos (2009), who examined the influence of the introduction of CDS. However, they only looked at the debt financing of their respective firms. According to their results, the credit spreads for CDS firms are not lower when issued in the syndicated loan and bond market. Nevertheless, they did find that CDSs cause the loan and bond spreads of firms to be a little lower. However, this reduction in the loan and bond spreads is only found for firms that are relatively transparent and safe. Similarly, the S&P 500 non-financial firms that are studied in this thesis are most likely to be firms that are relatively safe and transparent. Hirtle (2007) argues that an increase of credit supply of banks is connected to an increase in usage of CDSs. However, his study focussed solely on banks as capital suppliers with limited evidence. Moreover, Saretto and Tookes (2013) focused their study on the influence of CDSs on the debt maturity and leverage ratio. Their results suggested an economically and statistically significant increase in the leverage ratio and the debt maturity. This study aims to find the effect of CDSs traded contracts on a firm’s debt on their leverage ratio. This thesis will try to provide an answer for non-financial firms that have CDSs contracts traded on their debt. Consequently, this enables these researched firms to attract more capital from credit suppliers by increasing their leverage ratio. Therefore, the research question for this thesis is: 'Does Credit Default Swaps contracts traded on a firm's debt increase the leverage
ratio of the studied non-financial firms in a significant manner?'
To determine the effects of CDSs on the leverage ratio a model is structured which incorporates certain capital structure determinants. These capital structure determinants are based on the identified firm characteristics variables of Faulkender and Petersen (2005) that have significant influence on the leverage ratio. These variables are included in the model to control for the researched non-financial firms leverage ratio. The information about these firm characteristics is retrieved from Standard & Poor’s 500 index. Also, the non-financial firms, that are used in this study, will be identified using the S&P 500 index during the period 2002-2017. Additionally, the researched period will be extended up until 2017. After determining the firm characteristics explanatory variables, the CDS dummy variable was added to this model to capture the influence of the CDSs on the leverage ratio, wherein leverage is defined as book and market leverage for robustness. The necessary data needed for this thesis research was retrieved from three main data sources. First, Compustat-Capital IQ from Standard & Poor’s
3
was used to retrieve information for all the identified firm characteristics. Second, the Center for Research Data Services Prices (CRSP) was used to retrieve information on the asset volatility for the researched firms. Finally, the Bloomberg terminal was used to identify the non-financial firms that have traded in CDSs.
The main variable of interest is CDS. The aim of the study is to determine the influence of traded CDSs contracts on a non-financial firm’s debt on the dependent book and market leverage ratio variable. Interpreting the results from the ordinary least squares (OLS) regressions will determine the influence of CDSs on the leverage ratio. Also, the effects of the explanatory variables on the leverage ratio are interpreted. As these CDSs are used for hedging credit risk by the credit suppliers, they possibly have a positive effect on these firm’s investment opportunities. The results of the present study suggest that non-financial firms that have CDS contracts on their debt, significantly increase the leverage ratio when controlled for the firm’s characteristics. This finding possibly indicates an increase in the available investment opportunities for firms and possibly stimulates economic growth.
The main limitations for this thesis are mainly the eliminating process of firms’ year observations from the data sample due to missing data on the researched variables and a possible reverse causality problem between the dependent variable leverage ratio and the explanatory variable CDS. Moreover, following Faulkender and Petersen (2005), replacing the missing R&D variables with a value of zero caused a measurement error in this research. Additionally, another limitation was that of omitted variable bias. These limitations possibly may have caused biased results. Therefore, inferences made from the results should be made with care.
From here this thesis is organized as followed. Chapter 2 will examine literature on CDSs, their effects on a firm’s leverage ratio, their informational role, the differences with respect to other derivatives and the empty creditor problem. The chapter will be concluded with the hypothesis and research question. Next, chapter 3 will describe the methodology used to construct model 1: the influence of Credit Default Swaps on a firm’s leverage and the extended model 2 with the Credit Default Swaps interaction terms. In addition, the endogeneity problem will be addressed. Chapter 4 describes the data and descriptive statistics of the data sources, explanatory and dependent variables and the summary statistics of the identified variables. Then, chapter 5 will present the OLS regressions results of the influence of CDSs contracts traded on a firm’s debt on a firm’s leverage ratio for model 1 and 2. Finally, in chapter 6 the findings of this thesis will be concluded.
4
2. Literature Review
This chapter will present literature about CDSs and the capital structure of firms. As a lot of research has been done on these subjects, the focus will be on discussing literature that is relevant for formulating a hypothesis. This hypothesis will be about the influence of CDSs on the leverage ratio of non-financial firms, controlled for identified capital structures variables. This chapter will address different aspects of CDSs and starts with what the purpose of a CDS is, which includes a definition for CDSs. Then it continues to explain what the possible effects of CDSs are on a firm’s leverage. After that, the informational role that CDSs can provide about less transparent firms and what the benefits for potential investors are is discussed. This is followed by differences between CDSs with respect to other credit derivatives. Then the empty creditor problem is shortly addressed followed by an explanation of the used capital structure explanatory variables. Finally, the hypothesis is stated which will address the main subject of interest for this research.
2.1 Credit Default Swaps
A credit default swap is basically an insurance contract on a firm's debt. The purpose of a CDS is to transfer the credit exposure of fixed income products between the underlying parties. The CDS buyer is insured in case the debt issuer defaults or experiences another credit event before repaying the loan it took from the CDS buyer. The CDS buyer is thus able to shift the risk of their investment to a CDS seller. In the case of default, the buyer receives a fixed payment, namely the security's premium including all the interest payments up until the maturity date, from the CDS seller. Furthermore, the CDS buyer receives credit protection from the CDS seller. In return, the CDS seller receives a premium from the CDS buyer until the maturity date of the contract. This means that the CDS seller guarantees the credit worthiness of the debt security. The benefit for the CDS buyer is that they are protected against a possible default on their investment and they receive the par value of the contract. Otherwise it costs the CDS buyer only a fraction of the investment paying the CDS seller the premium. However, the CDS buyer is not protected if the value of a firm bonds decreases due to, for example, a change in the interest rate (Saretto and Tookes, 2013).
The payment that is paid upon a default is usually the difference between the recovery value and the debts face value. This value is calculated over a prespecified period that typically consist of 30 days after the default took place and makes an estimation based on the market prices. The other estimation option is based on the settlement auction of the CDS. Whenever a payout is necessary, the payout of the contract to the CDS buyer will be a payment in cash, or the bond that has defaulted is exchanged for cash (Bolton and Oehmke, 2011). It can be the case that when a credit event happens, which leads to a payout from the CDS seller to the CDS buyer, the debtor is uncapable of paying the required debt payment (non-payment) and the debtor must file for bankruptcy. Moreover, there are CDS contracts that state that payment to the CDS buyer is required in case of a downgrade in credit rating or if the debt is restructured (Bolton and Oehmke, 2011).
5
Furthermore, a CDS is a credit derivative that involves mortgage-backed securities, emerging market bonds and municipal bonds. In addition, it is used for hedging or speculation on credit risk, whichever the CDS buyer prefers. Also firms that issue debt in the form of bonds or other securities cannot guarantee that at the end of the maturity date of the contract they will be able to repay their debt. The reason is that these kind of debt securities, in general, have long maturity dates. Also, the security can have poor ratings, which is an indication that the issuer of these debt securities is more likely to default. So, the firm does not know with certainty that at the end of the contract it will have a healthy financial position. Therefore, the time frame of the securities and the possibility of default results in an amount of risk associated with the different kind of debt securities issued. If the risk on a debt security is high, it is more favourable to get a CDS to hedge the investment against a possible default (Faulkender & Petersen, 2005). Thus, a CDS becomes more desirable when the opinion of the debt security holder is leaning towards a default of the debt issuer. In that case, the payment of the premium is a small sacrifice with respect to the possible loss of the investment.
2.2 Credit Default Swaps effect on a firm's leverage ratio
Firms determine their favoured leverage ratio by calculating the costs of financial distress, incentive effects of debt versus equity, mispricing and tax advantages. If it turns out that the net benefit of debt is positive these firms will reduce their equity and/or issue additional debt. Consequently, these firms will move toward their favoured leverage ratio. Therefore, the implicit assumption has been that a firm’s demand for debt is determining the leverage ratio of the respective firm. In other words, the supply of capital should be infinitely elastic when the price is correct and only the risk of the firm’s projects defines what the cost of capital is (Faulkender & Petersen, 2005). However, some scholars argued that some firms suffer from being significantly under-levered. Firms that suffer from being under-levered are firms that, for example, miss the opportunity to increase their value significantly when they do not take the opportunity to increase their leverage ratio. Moreover, this increase in leverage would lead to a tax payment reduction (Graham, 2000). In addition, it is possible that firms are not able to issue additional debt because of market frictions, investment distortions and information asymmetry. This results in that firms can find it difficult to find a suitable lender and are therefore limited in their growth possibilities (Stiglitz and Weiss, 1981).
For instance, the possible effects of hedging with the use of CDSs, is a relaxation of the credit supply constraints firms may be facing. For potential investors, it is important that an investment is profitable. So, the ability to hedge against risk can make it more attractive to invest in risky debt securities, especially when treasuries are in short supply. Frequent providers of capital for debt securities are financial institutions, such as insurance companies and banks, that face regulatory capital requirements. The benefit of holding CDSs for these buyers is the possibility to reduce these regulatory capital requirements. If these capital suppliers are willing to buy the debt securities and provide a loan to the debt securities sellers, the credit supply to firms can increase. From there on they can forward the holdings of credit risk to a third party that is willing to hold that credit risk for a premium fee (Saretto & Tookes, 2011). Since hedging opportunities are created by CDSs, it seems that the cost of corporate debt indeed can be
6
reduced. This is because the credit risk of firms has essentially been untradeable and many investors that purchase bonds hold these bonds until their maturity dates, causing limited liquidity in the bond market. Bank loans in general are also illiquid when there is not an alternative to hedge or transfer their risk exposure, associated with the borrowing firms, to a third party. Introducing CDSs is an opportunity for these possible credit lenders to a less expensive way for solving these problems and provide loans to firms. (Ashcroft & Santos, 2009). Furthermore, Petersen and Rajan (2000) found that the credit supply of small firms increased after the constraints on their loan supply experienced a relaxation. Credit Default Swaps can contribute to the relaxation of the credit supply constraints and thus potentially increasing the available credit investment for firms. Consequently, this will increase the leverage ratio of these firms.
In addition, numerous scholars addressed the issue of supply frictions in their studies. For instance, Faulkender and Petersen (2005) argue that firms that have a credit rating, which enabled these firms to access public debt markets, can attract more debt. In addition, Lemmon, Roberts and Leary (2009) studied the effect of events on the supply of credit, that simulate shocks to the supply of credit of firms, which influences the investment and financing. On the other hand, Massa, Yasuda and Zhang (2009) found that the effect on leverage depends on the certainty of capital supply. If this is uncertain, it has a negative effect on the leverage ratio. Moreover, Erel et al. (2012) study the ability of firms increasing their capital in a time-series providing evidence that capital supply has a greater influence on increasing the availability of capital for firms during a downturn. In addition, Morellec (2010) looked at the accessibility with respect to the credit of a firm. Specifically, to a firm’s financing and investment choice when that access is uncertain. Furthermore. Duffee and Zhou (2001) found that CDSs traded on a firm’s debt can provide these firm’s credit suppliers additional opportunities in which they can diversify the risk on their credit exposure. These derivatives can be used to hedge credit exposures of banks. The evidence that CDSs are used for hedging purposes on banks risk exposures is provided by, among others, Minton et al. (2009) and Hirtle (2009). Minton et al. (2009) studied the lending activity of U.S. banks and whether these banks used CDSs to hedge their credit exposure risks. On the other hand, Hirtle (2009) studied if CDSs indeed affected the lending decisions of banks. Estimating the leverage of a firm thus not only depends on the determinants of the demand side, namely the proposed explanatory variables, but it is also important to include the variables that influence the leverage ratio through the supply side. In this case, the independent variable on the supply side is CDS. Therefore, this thesis will research if the leverage ratio of non-financial firms will change with the introduction of CDSs, in which CDSs possibly increase the capital supply to firms.
2.3 Credit Default Swaps informational role
Firms that are informationally opaque are difficult to assess if they would be trustworthy borrowers. Banks or private lenders are useful when it comes to investigating these kinds of firms to assess whether they would be financially healthy borrowers. These financial intermediaries decide, depending on the information they retrieve, if firms can borrow from them. In due time, this can partially resolve the information asymmetry which causes the market
7
failure. This is reflected in the ability of firms to access debt markets. Opaque firms, but also firms that have more discretion in their investment opportunities, are usually firms that are credit constrained (Carey, Post & Sharpe, 1998).
Credit Default Swaps on a firm's debt can be interpreted as a warning that the firm's likelihood of financial distress is higher, which in turn may affect the total risk of a firm (Subrahmanyam et al., 2014). Nevertheless, the benefit of CDSs is that investors can assess a firm's default risk because of the clean measure of the spread in CDS prices and the source of new information. These CDSs offers investors the opportunity to be able to trade risk in the secondary bond or loan markets. The loans that are traded in the secondary market are not a good source of information for retrieving information on firms, mainly because only a small number of loans are traded in this market. Moreover, the secondary market is filled with corporate bonds, a diversity of coupon structures and suffers from illiquidity, making the information available in this market opaque. Therefore, the CDSs are used by investors to determine what the firm’s default risk is and offer these investors the opportunity to trade risk exposures (Ashcroft and Santos, (2009).
Moreover, the cost of private debt capital increases when the expenditure of monitoring an opaque firm is higher or the lender's organizational form suffers from tax disadvantages (Graham, 1999). The CDS market can help to reduce the cost of debt by anticipating the credit rating events and extract information of opaque firms. This information could contribute in reducing the cost for investors and banks. In addition, for investors the information premiums on bonds goes down and it allows banks to lower the interest rate for borrowers. Furthermore, the CDS market can provide informational advantage which reduces the cost of debt (Hull et al., 2004).
2.4 Credit Default Swaps differences with respect to other derivatives
The most obvious reason for differences is the way in which CDSs are traded. CDSs are financial derivatives that are traded over the counter (OTC). This decreases the transparency of CDSs with respect to other financial derivatives that are traded on a central exchange. This means that recognizing forms of systemic risk, such as contagion, will be difficult to assess due to the opacity created by OTC trading (Peltonen, Scheicher & Vuillemey, 2014). The Dodd-Frank Act addresses this issue and contains a provision that is meant for regulating derivatives such as CDS. In the Dodd-Frank Act, CDS trading is centralized to reduce the possibility of a defaulting counterparty. Additionally, it creates more transparency in the CDS markets as the aim is to increase the disclosure of information to the public. Key components of the Dodd-Frank Act are restricting the way banks can invest, eliminating proprietary trading unless they have sufficient “skin in the game” and limiting speculative trading, which is indicated as the Volcker Rule. Moreover, it is not allowed for banks to be involved with businesses that are considered too risky, such as private equity firms or hedge funds. In addition, the Volcker Rule regulates financial firms that are “too big to fail” from preventing these institutions of using derivatives in a manner that are accompanied with large risk and possibly have great negative effects for the economy. Moreover, the Volcker Rule is all about minimizing the possibility of
8
conflict of interest for financial firms. Implementing the Dodd-Frank Act should make CDSs similar in the way they are traded in comparison to other derivatives (Whitehead, 2011). Then there is a difference between the participants in the CDS with respect to other derivative markets. The CDS market participants consist of financial institutions, such as insurance companies and banks, that are the suppliers of debt capital for the non-financial firms. The buyers of CDSs, for hedging their loan portfolios, consisted of banks that accounted for 20% of the total purchases in the CDS market. This figure was estimated in a comprehensive survey done by the British Banker’s Association (BBA). Banks provide these loans to maintain a healthy relationship with their clients. Besides, the provisions of these loans are possible because CDSs allows these banks to mitigate the risk in their loan portfolios. (British Banker’s Association, 2006)
Finally, it is assumed that derivatives do not influence the value of the underlying assets. However, Subrahmanyam et al. (2012) argue that CDSs differ from other derivatives in this aspect. According to them, trading in derivatives does not influence the value of the respective firm, whereas a firm trading in CDSs does influence the probability of default. Namely, it increases when a firm is trading in CDSs, which in turn is reflected in the value of the respective firm going down (Subrahmanyam et al., 2012).
2.5 The empty creditor problem
The relationship between creditors and their respective debtors have changed since the creation and exponential growth of the credit insurance market. This has especially been the case with the introduction of CDSs. A consequence of CDS credit insurance is that the risk transfer to the seller of the insurance does not only involves a transfer of risk, but also changes the relation between the creditor and respective debtor. In case of a financial distress event, the creditor fully or partially loses control of their cash flow rights. Consequently, the concern is raised that CDSs may create empty creditors as an effect of such a separation. As a result, empty creditors, that are both in possession of debt and CDSs, possibly lose interest in the wellbeing of the debtor and even may contribute in guiding the debtor into liquidation or bankruptcy. Moreover, empty creditors can even stimulate inefficient policies that lead to defaults (Yavorsky, 2009). It is important to remember that CDSs are an insurance that, when purchased, protect the respective credit suppliers in case of a firm’s default. This causes these creditors to lose interest in monitoring the debtors. Consequently, the credit suppliers no longer suffer as much from a debtor’s default, withholding these creditors for supplying new credit, voluntary restructuring of the debt or give up on some of their debt to possibly save the financially distressed firm from bankruptcy. In addition, CDSs created the possibility of granting the owner contractual rights on the underlying security without being exposed to the associated risk of a debtor’s default. This protects the credit supplier in the event of the respective debtor’s default. Even if these firms currently are not in a financially distressed position, it is important to keep in mind that these firms can possibly be in a financial bad position in the future (Hu and Black, 2008).
9
Another aspect of CDSs that is favourable for CDSs buyers is that it can provide an incentive for borrowers to be committed to pay out the agreed cash flows to the lenders. Creditors that own CDSs have different incentives in comparison with other creditors that are not protected against a debtor’s default. This means that creditors do not have to rely on debt renegotiations to retrieve their investments, causing them to be less forgiving when a firm is in financial distress. The outcome is that creditors have a stronger bargaining position in debt renegotiations and can extract more cash to their advantage. For the borrowers on the other hand, it diminishes their incentives to strategically try to lower their debt renegotiations, because the creditors are insured for their potential losses. It is thus possible that CDS protected creditors are not interested in debt renegotiations with the borrowing party and let the default happen. As a result, the firm has no other option to file for bankruptcy even though it would be less costly and more efficient to renegotiate the debt exchange. A CDS contract on a firm’s debt is then only renegotiable if the creditors find the offered terms by the borrowing party attractive enough. This should stimulate the borrowing firm to prevent policies that increases the chances of default (Bolton & Oehmke, 2011).
2.6 The capital structure explanatory variables
Empirical literature has successfully identified explanatory variables that are correlated with the capital structure choices of firms. A firm’s capital structure is determined by these explanatory variables that are researched and identified by multiple papers for influencing the leverage ratio of these firms. In addition, this thesis will integrate Faulkender and Petersen's (2005) identified firm characteristics control variables. The influence of CDSs on the leverage can then be researched in combination with these control variables.
Faulkender and Petersen (2005) identified the following firm characteristic control variables for the leverage ratio: market to book ratio, profitability, return on equity from the year prior, size of the firm, tangible assets, research and development expenditure, and the volatility of the assets. Implementing these control variables in the model will make it possible to determine what the influence of CDSs is on the leverage ratio of the researched non-financial firms. The effect will be captured for the researched firms that have traded in CDSs and those that not have traded in CDSs, the so called non-CDS firms.
According to Hovakimian et al. (2001), the market to book ratio can capture the growth opportunities of firms. In addition, it is a proxy for the intangible assets that are not always visible, such as trademarks, copyrights and intellectual property, as investors will make assumptions based on these intangible assets and will use it for assessing the possible risk of these unpredictable assets. Consequently, this may influence their decision for supplying credit to firms. Firms that have high market to book ratios, which implies high growth opportunities, should be careful that they do not suffer from underinvestment. In other words, having lower leverage ratios will attract more investments for financing these growth opportunities. Additionally, Hovakimian et al. (2001) found that the profitability of a firm is reflected in the amount of debt a firm can attract. Compared to less profitable firms, firms that are more
10
profitable and therefore financially stronger, have more debt on their respective balance sheets. This means that these profitable firms find it easier to find credit suppliers.
Moreover, according to Hovakimian et al. (2001) the capital structure of firms does not change quickly as this is costly. This opens the possibility that the return on equity can influence the leverage of these firms. Return on equity from the year prior is an indication of whether a firm had a positive or a negative return in the previous year. This provides information if the firm’s equity value went up or down. For example, if the return on equity is negative, the debt to equity ratio and thus the leverage ratio of the respective firm increases. On the other hand, a positive return on equity causes deleveraging and the leverage ratio of the respective firms decreases. Furthermore, the size of the firm influences the leverage ratio in a positive manner according to numerous researchers, such as Graham et al. (1998), Hovakimian et al. (2001) and Faulkender and Petersen (2005). That is, a large firm suffers less from asymmetric information problems and is more diversified. Thus, potentially reducing their risk, which in return makes them more attractive to be supplied with credit. Consequently, these firms find it easier to access the equity markets which in return can decrease their debt.
Moreover, tangible assets can function as collateral for loans, increasing the possible supply of credit where the ability of a firm to lend money increases. Which additionally increases a firm’s debt capacity as well. If a firm default, with a high value of tangible assets, investors can secure a great part of their loan in the liquidation process of these assets making it safer for them to invest. In addition, the effect of tangible assets on the leverage ratio is found to be positive (Rajan & Zingales, 1995). Tangible assets are relatively safe, whereas intangible assets are relatively risky. An intangible asset is an asset that is not physical in nature, such as brand recognition and goodwill. The effect of these assets on the leverage ratio is found to be negative. As these assets are difficult to measure, research and development expenditure is used as a proxy for intangible assets (Graham, 2000).
Besides, a firm’s volatility of assets represents the vulnerability of the firm with respect to financial insecurity. When this volatility increases the accompanied distress costs also increases. Firms that suffer from high volatility of assets aim for lower debt ratios as distress costs increases with volatility. On average, firms that have a higher volatility of assets have lower leverage ratios (Cantillo and Wright, 2000).
Substantiated by the abovementioned literature, the present study will research the interaction of CDS with the following three explanatory variables: size, volatility and profitability. Integrating these interaction variables will be done to try and capture the possible influence of these explanatory variables on a firm’s leverage ratio. Furthermore, Ashcroft and Santos (2009) argue that the influence of CDSs is different between the researched firms. On average the borrowing cost does not change for the firms, but for less opaque firms the cost of debt is lower. Therefore, implementing the three interaction variables in the model will possibly provide additional information about the effect of CDSs on the researched non-financial firms leverage ratio.
11
2.7 Hypothesis
The main literature, which is used for determining the effect of CDSs on the leverage ratio, is discussed below and involves the studies done by Saretto and Tookes (2011), Ashcroft and Santos (2009) and Bolton and Oehmke (2011).
First, Saretto and Tookes (2011) state that the positive effect of CDSs on the leverage ratio can occur when the treasuries are in short supply and the availability to risk free bonds is diminished. Risk averse investors will probably not be willing to supply credit to firms that are identified as high-risk firms that, with all probability, will default on their debt. The introduction of CDSs enables these risk averse investors to hedge the default risk, associated with the respective firm, to a third party. Being protected from default on their loan, these investors may be willing to supply the necessary credit to firms, which will increase the possible credit supply for firms. Another beneficial component of CDSs is that it can provide investors insurance against the illiquidity risk in the corporate bond market where selling these bonds is rather difficult. The holding of these bonds becomes more attractive to investors if they can hedge, using CDSs, the different risks mentioned. Moreover, another aspect of CDSs is that it is beneficial for the bank’s client relationships with firms that are trading in CDSs. These banks can supply credit to these firms while hedging the possible default risk to other parties. This enables them to not exceed their exposure standards of risk.
Second, Ashcroft and Santos (2009) argue that CDSs provide revealing information about firms. Investors can assess a firm's default risk because of the clean measure of the spread in CDS prices. This information will cause the provision of credit by investors to increase to firms that are trading in CDSs, because these firms become more transparent. The illiquidity of the corporate bond market is the reason that this market provides little information about firms, where it is not helping in making opaque firms more transparent. For investors, information about the firm is important to be able to make investment decisions about supplying credit. Finally, according to Bolton and Oehmke (2011) CDSs can increase credit supply to firms even though the holders of CDS contracts are empty creditors. The CDSs protects these empty creditors in case of default and possibly loses incentive to monitor the firm. Therefore, they may be unwilling to negotiate for a possible restructure of the debt or rollover some of the debt, so the firm may survive. With this knowledge, firms that do not follow up on their debt payment have to consider that they will go bankrupt when there is no support by the insured investors. Moreover, CDSs can induce these firms to commit to pay the required debt payment in fear of going bankrupt otherwise. Consequently, investors are more likely to provide the necessary investment to firms that are committed to pay their debt payments.
Therefore, the hypothesis for this study is: ‘non-financial firms that have CDSs contracts traded on their debt find a significant increase in their leverage ratio’. The leverage ratio of a firm is not expected to solely depend on CDSs contracts traded on a firm’s debt. Thus, other determinants need to be identified that might influence the leverage ratio. To determine the possible effect of CDSs on the leverage ratio, it is also important to determine the explanatory variables that influence the leverage ratio of the researched non-financial firms. These identified explanatory variables or firm characteristics, are researched by many among which Faulkender
12
and Petersen (2005). The present study will use the firm characteristics discussed by them and implement them in the model as well. The firm characteristics are: market to book ratio (MTB), profitability (EBITDA), size of a firm (SIZE), tangible assets (TA), research and development expenditure (RD), volatility of assets (VOL) and the return of equity of the previous year (𝑅𝑂𝐸−1).
Hence, the research question of this thesis is: 'Does Credit Default Swaps contracts traded on
a firm's debt increases the leverage ratio of the studied non-financial firms in a significant manner?'
13
3. Methodology
This chapter will discuss the methodology that was used to answer the hypothesis discussed in the previous chapter. First, model 1 was constructed which tried to capture the influence of CDSs on a firm’s leverage ratio. Second, the extended model 2 was created to incorporate the interaction variables between the three discussed explanatory variables and CDS. Finally, the dependent and explanatory variables are explained.
3.1 Model 1: The effect of Credit Default Swaps on a firm's leverage ratio
First, it was necessary to create a linear model that grasps the concept of the possible positive influence of CDSs traded contracts on a firm's debt on the firm's leverage ratio. To notice a change in the leverage ratio, it was important to add explanatory variables that control for the leverage ratio. After the model was completed, an Ordinary Least Square (OLS) regression was run in Stata to conclude what the effect of CDSs are on the leverage ratio.
Both the dependent variables book and market leverage were included in all the empirical analyses for robustness. Book leverage was estimated as total debt–which was defined as debt in current liabilities plus the long-term debt–divided by the book value of assets of the respective firms. Market leverage was estimated by dividing the total debt–which was defined similar as the total debt for book leverage–by the firm’s value. Firm value was defined as the book value of deferred taxes, plus the market value of equity, plus the book value of assets, minus the book value of common equity.
CDSs were used in this model as a measure for credit suppliers to be able to hedge the accompanied risk of the respective loans. This ability to hedge for the credit suppliers was expected to have a relaxation effect on the credit supply constraints, which may increase the credit supply and possibly increase the leverage ratio. The CDS variable that was included in the empirical regression models was CDS traded. This variable was an indicator if the respective non-financial firm had a CDS contract traded on its debt at any time during the sample period from 2002 up to and including 2017. This variable for CDS was a dummy variable and equalled the value of 1 if the researched firms indeed had traded CDS contracts on their debt during the indicated period.
Model 1 was constructed with the firm characteristics as discussed by Faulkender and Petersen (2005) and integrated the CDS variable. Moreover, this CDS variable was included by other scholars as well, such as Ashcraft and Santos (2009) and Saretto and Tookes (2013). To summarise, model 1 had the following construction:
𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡 = 𝛼 + 𝛽1𝐶𝐷𝑆𝑖𝑡 + 𝛽2𝑀𝑇𝐵𝑖𝑡+ 𝛽3𝐸𝐵𝐼𝑇𝐷𝐴𝑖𝑡+ 𝛽4𝑅𝑂𝐸𝑖𝑡−1+ 𝛽5𝑆𝐼𝑍𝐸𝑖𝑡+ 𝛽6𝑇𝐴𝑖𝑡+ 𝛽7𝑅𝐷𝑖𝑡 + 𝛽8𝑉𝑂𝐿𝑖𝑡+ 𝛽9𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡−1+ 𝜀𝑖𝑡 (1)
14
The explanatory and dependent variables can be found in table 1. Table 1: Explanatory and dependent variables
Dependent variable
Definition
𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡 The leverage ratio is defined as book leverage or market leverage. Both
definitions of leverage are separately tested in the OLS regressions. Explanatory
variables
Definition
𝐶𝐷𝑆𝑖𝑡 Credit default swap is a dummy variable that equals 1 if CDSs contracts are traded on a firm’s debt at time t. If no CDSs contracts exist at time t, then the dummy variable has value 0.
𝑀𝑇𝐵𝑖𝑡 Market to Book ratio
𝐸𝐵𝐼𝑇𝐷𝐴𝑖𝑡 Earnings Before Interest, Taxes, Depreciation and Amortization 𝑅𝑂𝐸𝑖𝑡−1 Return of Equity of one year prior
𝑆𝐼𝑍𝐸𝑖𝑡 Natural logarithm of revenue
𝑇𝐴𝑖𝑡 Tangible assets in the form of property, plant and equipment 𝑅𝐷𝑖𝑡 Research and Development
𝑉𝑂𝐿𝑖𝑡 Volatility of assets
𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡−1 The lagged leverage ratio is defined as book leverage from the previous
year or market leverage from the previous year. Both definitions are separately added to the regressions for the respective leverage definitions.
𝜀𝑖𝑡 Error term
The null hypothesis is formulated to provide an answer for the following hypothesis: ‘non-financial firms that have CDSs contracts traded on their debt find a significant increase of their leverage ratio’. Therefore, the null hypothesis is: ‘CDSs traded contracts on a firm's debt have no significant effect on the respective firm's leverage ratio’. This null hypothesis stated that if there is no positive effect on the firms leverage ratio due to CDSs traded contracts on a firm’s debt, then 𝛽1 equals zero, otherwise 𝛽1 is greater than zero. The expected result is that, based
on previous literature, CDS trading does have a significantly positive effect on the leverage ratio. If that is the case, then the null hypothesis would be rejected. Also, the OLS Estimators in the linear regression model need to fulfil the Gauss-Markov theorem to be the Best Linear Unbiased Estimator (BLUE).
Thus, the null hypothesis is: 𝐻0: 𝛽1 = 0
15
A t-test was performed to test the significance and effect of 𝛽1 on the book and market leverage ratio. The test follows a t-distribution with n-k-1 degrees of freedom, where k is the number of variables used. The t-test was constructed as followed:
𝑡 =𝛽̂1− 𝛽1
𝑆𝐸𝛽̂1 ~𝑡𝑛−𝑘−1
3.2 Model 2: extended with Credit Default Swaps interaction terms
The extended model 2, which was based on model 1, integrated the three interaction terms with the variable CDS. The three variables of which the interaction with CDS was examined were: the size of the firm (SIZE), the volatility of assets (VOL) and the profitability of the firm (EBITDA). Model 2 was constructed as followed:
𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡 = 𝛼 + 𝛽1𝐶𝐷𝑆𝑖𝑡 + 𝛽𝑌𝑖+ 𝛽10𝐶𝐷𝑆 ∗ 𝑆𝐼𝑍𝐸+𝛽11𝐶𝐷𝑆 ∗ 𝑉𝑂𝐿𝑖𝑡+ 𝛽12𝐶𝐷𝑆 ∗ 𝐸𝐵𝐼𝑇𝐷𝐴𝑖𝑡+ 𝜀𝑖𝑡 (2)
As the explanatory and dependent variables are the same as presented in table 1, only the defined interaction variables were added in table 2.
Table 2: Interaction and dependent variables Dependent
variable
Definition
𝐿𝐸𝑉𝐸𝑅𝐴𝐺𝐸𝑖𝑡 The leverage ratio is defined as book leverage or market leverage. Both
definitions of leverage are separately tested in the OLS regressions. Interaction
variables
Definition
𝐶𝐷𝑆𝑖𝑡 Credit default swap is a dummy variable that equals 1 if CDS contracts are traded on a firm’s debt at time t. If no CDS contract exist at time t the dummy variable has value 0
𝑌𝑖 Explanatory variables that represent the firm characteristics in model 1 𝐶𝐷𝑆𝑖𝑡*𝑉𝑂𝐿𝑖𝑡 Volatility of assets multiplied by the dummy variable CDS
𝐶𝐷𝑆𝑖𝑡*𝑆𝐼𝑍𝐸𝑖𝑡 Natural logarithm of revenue multiplied by the dummy variable CDS 𝐶𝐷𝑆𝑖𝑡*𝐸𝐵𝐼𝑇𝐷𝐴𝑖𝑡 Earnings Before Interest, Taxes, Depreciation and Amortization
multiplied by the dummy variable CDS 𝜀𝑖𝑡 Error term
16
3.3 Endogeneity problem
Reverse causality between the dependent variable and an explanatory variable is also referred to as an endogeneity problem in econometrics. There was a potential problem of reverse causality between the dependent variable leverage ratio and the explanatory variable CDS. It means that when firms have traded CDSs on their debt, it will likely increase that firm’s leverage ratio. This, in turn, can imply that when a firm has a high leverage ratio it likely has CDSs traded on its debt. So, the reverse causality problem arises when a firm’s leverage ratio is higher because that firm has traded CDSs on their debt and for a firm with a high leverage ratio, it is likely to have CDSs traded on their debt.
Furthermore, the simultaneity between the assumed exogenous independent variable CDS and the dependent variable, both book and market leverage, caused the random error term to be correlated with the explanatory variables. Addressing the potential endogeneity issue can be done by defining an appropriate instrument for the CDS variable. Such an instrument should capture the changes in the explanatory variable without having an independent effect on the dependent variable. This would have allowed us to determine the causal effect of the explanatory variable on the dependent variable. In this case, the instrument should measure the effect of CDS on the leverage ratio without influencing the leverage ratio of the firm itself. In addition, the two-stage least square (2SLS) technique is frequently used by many researchers when an instrumental variable approach is used. A 2SLS technique can possibly solve the simultaneity problems that arises between the dependent and the independent variables. This will eliminate the endogeneity bias. Unfortunately, this research did not succeed in finding appropriate instruments that could be implemented in the model. Therefore, integrating instruments that solve this endogeneity problem was beyond the scope of this thesis. Thus, any inferences made from the results should be made with care for the possible presence of reverse causality.
17
4. Data and descriptive statistics
This chapter will present the data sources that were used for collecting the necessary data to run the introduced models. The sample size will be discussed before and after adjustments. Next, the explanatory and dependent variables are explained. After that, the summary of the statistics provides a descriptive statistic overview of the chosen variables.
4.1 Data sources
The Wharton Research Data Services provides access to different databases. These databases were used to retrieve the necessary data for answering the research question: 'Does Credit
Default Swaps contracts traded on a firm's debt increase the leverage ratio of the studied non-financial firms in a significant manner?'
There were two main data sources used from Wharton Research Data Services namely: Compustat-Capital IQ from Standard & Poor’s and the Center for Research and Security Prices (CRSP). Compustat-Capital IQ was used to retrieve the information for the identified explanatory variables, in other words the firm characteristics. In addition, CRSP was used to retrieve information on asset volatility for the researched firms.
The third database was Bloomberg and was used to retrieve information for which firm has traded in CDSs during the researched time frame from 2002 up to and including 2017. The year 2002 was the starting year of this thesis research as it was the first year in which Bloomberg identified CDS quote data. The information for CDSs traded contract on a firm’s debt was obtained from the Bloomberg terminal located at the Erasmus University. The Bloomberg quote data allowed us to identify firms that had CDSs contracts and firms that did not have CDSs traded contracts on their debt, or the so called non-CDS firms.
For identifying the non-financial firms, the databases were searched on basis of tickers, namely SIC and CUSIP codes that are designated to one specific firm. Financial firms were identified by SIC codes ranging from 6000-6999 and were removed from the dataset. Next, annual data was collected for the period when the CDS quotes started in 2002 up to and including 2017. After implementing and matching these codes in the databases, the information on all the variables was retrieved. This resulted in 7656 firms’ year observations. All variables need non-missing data for the researched regression, so that intuitive explanations could be made on the influence of these variables on the book and market leverage ratio. After the correction for the missing data on the variables, there was a final dataset that consisted of 3520 firms’ year observations.
After adjusting the dataset for non-financial firms and eliminating the missing data variables, a total of 434 non-financial firms was present in the dataset. The Bloomberg terminal provided the CDS trading history information for these non-financial firms and identified a total amount of 112 firms that had traded in CDSs contracts for the adjusted data set.
18
4.2 Explanatory and dependent variables
The dependent variable was the leverage ratio which was defined in two separate ways, namely book and market leverage. Both approaches of leverage were be tested and analysed for robustness. In addition, differences between both definitions for the leverage ratio were examined. The explanatory variables consisted of firm characteristics that controlled for both the definitions of the leverage ratio, the leverage ratio of the previous year and CDS. The chosen firm characteristics were: market to book ratio (MTB), profitability (EBITDA), size of a firm (SIZE), tangible assets (TA), research and development expenditure (RD), volatility of assets (VOL) and the return of equity of the previous year (𝑅𝑂𝐸−1). This research was particularly
interested in the effect of CDSs on the leverage ratio and it was therefore the explanatory variable of interest. To see if the influence of CDSs also differed in effect when it interacted with the size, volatility and profitability of a firm, three interaction variables were added to the first model. These interaction variables were CDS*SIZE, CDS*VOL and CDS*EBITDA. Appendix A can be consulted for a broader understanding of the variables used.
4.3 Summary statistics
Table 3 shows the summary statistics for the adjusted data sample for the non-financial firms in the S&P 500 index. It displays the original data and the winsorized data for the relevant variables. Winsorizing the original data helps to eliminate large outliers in the data sample. This limits the effect of extreme values that are present in the different variables. Moreover, large outliers can cause estimators to be under- or overestimated and can thus affect the possible effect of the different researched estimators. Eliminating these extreme values from the data sample restricts these unwanted effects. The dataset was winsorized at the 10𝑡ℎand 90𝑡ℎ
percentile to rectify the large outliers and keep most of the researched values intact. The information displayed below is for the non-financial firms that use CDSs and for non-financial firms that do not have CDS contracts on their debt.
19
Table 3: Summary statistics for CDS firms and non-CDS firms
Obs. Mean Median Standard
Deviation
Min Max
N Data Winds. Data Winds. Data Winds. Data Winds. Data Winds. Book leverage 3520 0.2676 0.2578 0.2463 0.2463 0.1687 0.1240 0.0001 0.0782 1.7048 0.4663 Market leverage 3520 0.6232 0.2679 0.1939 0.1929 11.0732 0.2186 0.0000 0.0348 645.434 0.7218 MTB 3520 1.6014 1.4468 1.2330 1.2330 1.4334 0.8260 0.0024 0.4816 20.0927 3.0826 EBITDA 3520 0.0115 0.2091 0.1971 0.1971 7.4728 0.0967 -421.4 0.0739 0.7826 0.3769 𝑅𝑂𝐸𝑡−1 3520 0.2283 0.16850 0.1579 0.1579 3.0984 0.1005 -35.60 0.0120 143.588 0.3493 SIZE 3520 8.9972 9.0286 9.0134 9.0134 1.3836 1.0723 -1.634 7.4034 13.0695 10.784 TA 3520 0.2706 0.2608 0.1848 0.1848 0.2245 0.1965 0.0010 0.0528 0.9304 0.6299 RD 3520 0.1119 0.0314 0.0042 0.0042 1.8512 0.0461 0.000 0.000 85.8909 0.1362 VOL 3520 0.3318 0.3242 0.3100 0.3100 0.1181 0.0881 0.1200 0.2080 0.9890 0.4826 Table 3 contains summary statistics for the sample of non-financial firms with and without CDSs traded contracts on their debt. The table provides information about the number of firms’ year observations, the mean, median, standard deviation and the minimum and maximum values for the variables indicated in the table. Missing values for research & development are set equal to zero. The data sample is winsorized at the 10𝑡ℎ and 90𝑡ℎ percentile.
Several important observations about the book and market leverage can be made from table 3. First, the table shows that the average mean of book and market leverage is substantial. The mean value of book leverage is lower than the mean value of market leverage (0.2676 < 0.6232) in the data sample. Second, the substantial difference between the book and market leverage mean value is due to large outliers. These outliers are especially present in the market leverage where a maximum value is found of 645.434. After winsorizing this maximum value of the market leverage drops to 0.7218. The difference is diminished to the values 0.2578 for book and 0.2679 for market leverage. Finally, the standard deviations of book and market leverage after winsorizing are 0.1240 and 0.2186, respectively. These standard deviations indicate substantial cross-sectional variation in the leverage variables.
The explanatory variables observations, from the summary statistics in table 3, will be discussed next. First, it is observable that the market to book ratio is greater than one on average and that the mean value, for the original and winsorized data, for market leverage is higher than the book leverage. The market to book maximum value is 20.0927, which is quite large for the original data set. After winsorizing this maximum value, the value drops to 3.0826 and the minimum value increases from 0.0024 to 0.4816. Thus, all the researched firms have growth opportunities with an average median of 1.2330.
20
Second, the profitability of the firms is captured with the EBITDA variable which has an original value of 0.0115, and after winsorizing a value of 0.2091. In addition, it can be observed that the minimum value of EBITDA is as low as -421.4, which indicates that the losses would be extremely high for the respective firm. Winsorizing helps eliminate this outlier which leads to a minimum value of 0.0739. The median for EBITDA is 0.1971, so in general the firms in S&P 500 index are profitable firms.
Next, the return on equity from the previous year (𝑅𝑂𝐸𝑡−1) has a median of 0.1579 and is thus a positive return on average. The original mean value of 𝑅𝑂𝐸𝑡−1 is 0.2283 and after winsorizing, this value changes to 0.1680. 𝑅𝑂𝐸𝑡−1has a large value for the minimum and
maximum value which are respectively -35.60 and 143.58. The minimum and maximum value changes after winsorizing to 0.0120 and 0.3493. Therefore, the return on equity from the previous period is positive. This is a logical outcome because the firms’ EBITDA is positive as well.
Fourth, the size (SIZE) of a firm is an indication of how big the firm is from a perspective of their total revenue that is log linearized. The minimum and maximum value of the size of the firms are respectively -1.634 and 13.069. After winsorizing, these values change to a minimum of 7.403 and a maximum value of 10.784 with a median of 9.013.
Following, tangible assets (TA) has an original mean value of 0.2706 and after winsorizing, this mean value changes only a little to 0.2608. The minimum and maximum value of the tangible assets are respectively 0.0010 and 0.9304. After winsorizing, the minimum and maximum value changes to 0.0528 and 0.6299 with a median of 0.1848.
Sixth, it is difficult to measure intangible assets that are non-physical assets, such as trademarks, intellectual property and goodwill. Research and development (RD) was therefore used as a proxy for intangible assets. The original mean value for RD is 0.1119 and after winsorizing this value changes to 0.0314 with a median of 0.0042. There are large outliers for the maximum value of RD with the maximum value of 85.891. After winsorizing, the maximum value changes to 0.1362. The minimum value of RD is 0 before and after winsorizing. This result follows most likely because the RD values equals zero if the data to calculate the RD value is missing. In this dataset there are numerous firms’ years where the data on RD is missing. Therefore, the possible effect that RD has on the leverage ratio and the inferences made from that outcome should be made with caution.
Finally, the volatility of assets (VOL) is an indicator of a certain risk associated with the respective firm. A high VOL increases the firm’s default probabilities. The maximum value of volatility decreases from 0.9890 to 0.4826 after winsorizing. Moreover, the minimum value of VOL increases from 0.1200 to 0.2080 after winsorizing. The median of VOL is 0.3100 and the mean value is 0.3242 after winsorizing.
21
Table 4: Summary statistics CDS firms only
Obs. Mean Median Standard
Deviation
Min Max
Data Winds. Data Wind s.
Data Winds. Data Winds. Data Winds.
Book leverage 1.035 0.308 0.300 0.276 0.276 0.156 0.123 0.011 0.143 1.106 0.524 Market leverage 1.035 0.565 0.380 0.305 0.305 1.633 0.248 0.007 0.112 28.09 0.902 MTB 1.035 1.042 0.993 0.887 0.887 0.641 0.459 0.021 0.416 4.365 1.859 EBITDA 1.035 -0.205 0.195 0.173 0.173 13.106 0.102 -421.43 0.0642 0.762 0.383 ROE 1.035 0.248 0.159 0.156 0.156 3.083 0.122 -35.60 -0.032 70.38 0.348 SIZE 1.035 9.768 9.785 9.610 9.61 1.202 0.945 -1.634 8.404 13.06 11.327 TA 1.035 0.340 0.333 0.274 0.274 0.248 0.227 0.001 0.062 0.904 0.701 RD 1.035 0.020 0.010 0 0 0.050 0.018 0 0 0.655 0.053 VOL 1.035 0.319 0.313 0.300 0.300 0.104 0.078 0.140 0.2079 0.970 0.450 Table 4 contains summary statistics for the sample of non-financial firms with CDSs traded contracts on their debt. All the non-CDS firms are removed from the data sample. The table provides information about the number of firms’ year observations, the mean, median, standard deviation and the minimum and maximum values for the variables indicated in the table. Missing values for research & development are set equal to zero. The data sample is winsorized at the 10𝑡ℎ and 90𝑡ℎ percentile.
Table 4 provides summary statistics information for only CDS firms and will be compared to table 3. Next, the differences between the summary statistics for CDS firms and non-CDS firms will be examined. When comparing the winsorized summary statistics dataset for non-CDS and CDS firms (table 3) with the winsorized summary statistics for only CDS firms (table 4), differences can be found in certain explanatory variables. The book and market leverage are both higher for only CDS firms, respectively 0.300>0.258 and 0.380>0.268. In addition, the market to book value has declined to a value 0.993 which is lower than the average data set value 1.447. Both the values for profitability and the return on equity from the previous period are lower for only CDS firms when compared to the average data set value, respectively 0.195<0.209 and 0.159<0.169. Tangible assets for only CDS firms have a value of 0.333 which is higher than the value of the average data set 0.261. Moreover, the size of only CDS firms is larger with a value of 9.785 compared to the average dataset’s value 9.029. Finally, the values of research and development and volatility of assets are both lower, respectively 0.010<0.0314 and 0.313<0.324.
22
5. Results
The following chapter discusses the results obtained from the constructed models previously mentioned in the methodology chapter. The used models incorporate both the book and market leverage in two separate equations. These models were tested on what the influence of CDSs is on the leverage ratio of non-financial firms, controlled for known identified firm characteristics that determine the firm’s capital structure. The models that are extended with three interaction variables, CDSSIZE, CDSVOL and CDSEBITDA, were also tested on both the book and market leverage in two separate equations for significance. These models were used to determine the possible influence on the leverage ratio.
5.1 The effect of Credit Default Swaps on the leverage ratio
The regression that was run for model 1 on the book and market leverage can be found in table 5. Regression Ⅰ, Ⅱ and Ⅲ shows the results solely for book leverage and the effect of CDSs on the respective leverage ratio. The variable of interest CDS has a positive influence on the book leverage ratio for non-financial firms that have CDSs traded contracts on their debt. It has an estimated value of 0.035 which accounts for a 3.5% increase in the book leverage ratio for firms that have traded in CDSs. This estimated value is very significant with a p-value of 0.000. The effect of market leverage ratio is calculated in the regressions Ⅳ, Ⅴ and Ⅵ. The estimate CDS effect is again positive with a value of 0.028. This indicates that non-financial firms that have CDSs traded contracts on their debt influences the market leverage ratio with a positive increase of 2.8%. Also, this estimated value is very significant with a p-value of 0.000.
The adjusted 𝑅2 shows how much of the model’s dependent variable is clarified by the explanatory variables. In case of the book (market) leverage it has a value of 0.2179 (0.3226) which is equal to an explanation of 21.79% (32.26%) for the dependent variable. This is an indicator that there are possibly more explanatory variables that influence the book (market) leverage ratio, which are not included in this model. When the lagged book (market) leverage variable is introduced in the regression of book (market) leverage, the 𝑅2 increases to 0.6305
(0.7162). This explains the leverage ratio for 63.05% (71.62%). Therefore, including the lagged book (market) leverage variable increases the percentage explained of the respective dependent variable. However, the adding of the lagged book (market) leverage variable impacts the other estimates in which they all decrease in value. The associate effects on the book (market) leverage thus diminishes. In addition, the significance of the explanatory CDS variable, for the book leverage regression, drops from significant at the 1% level to significant at the 5% level, where RD is not significant anymore. The adding of the lagged market leverage variable causes a reduction in significance for CDS, ROE and RD. The estimate CDS is now only significant at the 5% level, ROE and RD are not significant anymore. For this study, the effects of the explanatory variables that control for the book (market) leverage ratio and the influence of the explanatory variable CDS are interesting. In addition, this thesis aims to capture the effects of the interaction variables, CDSSIZE, CDSEBITDA and CDSVOL with CDS. Therefore, the extended model 2 with the interaction terms is without the lagged book (market) leverage variable as this reduces the significance of some explanatory variables. Furthermore, the
