An event study : the directional effect of M and A announcements on the acquiring firms’ CDS spread

(1)

An Event Study: The Directional Effect of M&A

Announcements on the Acquiring Firms’ CDS Spread

Author: Alex van de Haterd Student number: 10212671 Supervisor: Rafael Matta

(2)

2

Statement of Originality

This document is written by Alex van de Haterd 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)

3

1. Introduction

Financial weapons of mass destruction; that is how world’s most famous investor Warren Buffet had named credit default swaps (CDS) in 2003 in the midst of the recession. Reading this in The Economist (Financial WMD, 2004) made me curious about these financial derivatives. After all, CDS were developed in the 1990s by Goldman Sachs bankers to transfer credit risk from one party to another (O’Kane & Turnbull, 2003). Financial

institutions and banks could now manage their credit portfolios more actively, this way CDS contributes to the economy. Especially if you consider the billions dollar of losses resulting from the collapse of Enron and WorldCom.

As CDS increasingly were used as a trading instrument the CDS market matured. Investors exploited mispricing and took speculative positions. Hence the CDS market got more liquid and enhanced pricing efficiency, and grew from $630 billion notional outstanding in 2001 to $62 trillion notional outstanding in 2007. During the same period the mergers and acquisition (M&A) market was at its peak with a total deal value in 2007 of $3500 billion. This made me wonder about the relationships between the CDS and M&A markets. My first research showed there was not much literature available on this particular subject. I

formulated my research question as follows: what is the directional effect of M&A announcements on the acquirers’ CDS spread?

I found one study on the relationship between leverage buyouts and CDS from the targets’ perspective (Eisenthal, 2009). From the acquirers’ perspective I could not find any available research. On the other hand, there was much work to be found on the relationship between CDS and bond and equity markets. Also much theoretical and empirical research was available on M&A and bond and equity markets. I wanted to investigate the relationship between the two markets by doing an event study on the effects of M&A announcements on CDS spreads.

The thesis has the following structure: I start with discussing the existing literature on CDS and its implication on corporations to get an understanding of CDS. Next, the

relationship of CDS and M&A on bond and equity markets will be discussed. After that, briefly the valuation of CDS spreads is shown in order to gain more insight in the pricing process. Then the methodology, an event study, will be described. The data used for this

(5)

5

thesis will be presented. Finally the results of the event study will be shown and analysed followed by the conclusions that can be drawn for this research will be presented.

2. Literature Review

In this section I will provide some general knowledge concerning Credit Default Swaps, starting with the origin and basic concept of CDS. Next I will examine what kind of implication the issuance of CDS has on the probability of default for corporations.

Furthermore, the literature about the relationship of M&A announcements and CDS with the equity and bond markets will be discussed. Finally I will present the CDS spreads pricing mechanism.

2.1 Credit Default Swaps

A CDS is a contract that allows the protection buyer, the one that buys the CDS, to hedge the risk that an underlying entity will enter a credit event. The three most common types of credit events are the following (Brennermeier, 2008):

 bankruptcy: insolvency for the reference entity or unable to repay debts;

 failure to pay: When a reference entity fails to make payment of a principal or interest; and

 restructuring: Refers to a change in the terms of debt obligations that are adverse to creditors.

The protection buyer pays a premium, usually quarterly, to the protection seller until the contract matures or a credit event occurs. In the case of a credit event the protection seller is obliged to compensate the protection buyer for the incurred losses suffered. The incurred loss is the difference between the face value and the recovery proceeds of the underlying security. At the time the CDS contract is settled, the recovery value of a firm in case of a credit event is and remains an estimate. Physical delivery of the underlying security has been the primary choice to overcome this issue of uncertainty. The protection seller receives the underlying security in return for the face value of the bond for example to the protection buyer

(6)

6

(Weistroffer, 2009). In figure 1 a typical contractual relationship between the parties involved are illustrated. For example, a bank issues bonds worth $100 million for ten years at a 10% interest rate to a corporation. The corporation will make yearly interest payments of $10 million to the bank. By paying a premium to the protection seller, which can be paid off by the interest payments, the bank can insure itself against a possible loss of $100 in case of a possible bankruptcy. The protection seller can make a profit by paying off the losses caused by one triggered CDS contract with the premiums collected from other contracts.

Figure 1. Typical payout scheme

CDS usually have a market value close to zero quoted as an annual percentage in basis points of the face value of the bond on which protection is bought. The market value, or spread, is calculated by comparing the present value (PV) of the premiums paid in respect to the

expected losses of a default (Weistroffer, 2009). As the default risk of a firm can change over time the CDS spread can differ from the risk premium initially agreed on. This results in a positive market value for one of the parties involved, and a negative market value its contracting party, since a CDS is a zero-sum contract.

2.2 Implications of CDS for corporations

In this section I will discuss the effects of CDS on corporate management and the probability to bankruptcy.

As discussed before, CDS were designed to transfer risk from one party to another. The entrance of CDS on the derivatives market led to substantially better risk allocation and

(7)

7

liquidity into the economy. Banks were now able to avoid concentration risk – which means having a too large debt position in one firm – by buying CDS. It freed up capital in banks and drove the liquidity in the market upwards (Weistroffer, 2009).

However, this increase in liquidity did not really lead to a reduction of cost of capital for average type of firms. For stable and transparent firms Ashcraft and Santos (2009, p. 26) found a small decrease in the spreads of bonds and loans as a result of the rising market liquidity. Unexpectedly, riskier firms were affected adversely by the trade of CDS on their loans, this due to their dependency on syndicate loans. The option for lead banks - in syndicate loans - to hedge their corporate exposure leads to a reduction of incentive

alignment. The lead bank has, in the case of a credit event, less incentive to cooperate in an out-of-court renegotiation. Therefore such an investment becomes riskier.

Saretto and Tookes (2012, p. 1) found some substantial differences between firms that were trading CDS on their debt and firms that were not. The first difference found was that firms trading CDS on their debt could maintain higher leverage ratios than firms that were not trading CDS on their debt. Another difference is, CDS traded firms were able to attract debt with longer maturities. It is inherent that higher leverage ratios, and lower liquidity within firms, will result in a greater chance of default.

Alternatively, findings of Subrahamanyan (2014, p. 27) show that firms trading CDS on their debt perform a significantly more conservative liquidity policy. Investors protected by CDS tend to be tougher negotiators. This phenomena is called the empty creditor problem (Bolton & Oehmke, 2011). Creditors protected by CDS have less incentive to cooperate in the continuation or restructuring process in case of a credit event. As a consequence, firms

reserve more cash in order to manage their future liquidity position better. Furthermore, it is proved that firms with more financial expertise hold a bigger cash position when CDS are traded on their debt. On average the cash position of CDS traded firms increased by 2%, to 8.2%.

Moreover, CDS can lead to managers taking risky investments when insured against a credit event. The losses incurred due to a credit event will be limited. This shift of risk, triggered by CDS, will lead to a higher probability of default for firms (Campello & Matta, 2012).

(8)

8 2.3 CDS relation to bond and equity markets

In theory CDS spreads should be closely related to bond yield spreads (Hull, Predescu & White, 2004). If we define y as the yield on a bond, r as the yield on a risk-free asset and s the spread of a CDS all with the same maturity than the following relationship should

approximately hold: 𝑠 = 𝑦 − 𝑟 (Hull, Predescu & White, 2004). In a situation this relationship doesn’t hold market neutral arbitrage opportunities will exist. So in theory little new

information could be derived from CDS spreads since bonds already reveal the

creditworthiness of bonds. However in practice the two show significant differences. This may be a result from bond yields being influenced by other factors apart from credit risk, interest rate risk and liquidity risk.

According to Hull, Predescu and White (2004, p. 2809) an advantage of CDS spreads is that no further adjustment is required compared to bond yields. These need to be corrected for risk-free rates to obtain the credit spreads. Assumptions are need to be made to decide which benchmark risk-free rate is the appropriate one. Likewise, recovery value and counterparty risk are uncertain factors for CDS pricing, but, it excludes the risk-free rate as external factor (Weistroffer, 2009).

Furthermore, in general corporate bonds are held until maturity (Blanco, Brennan and March, 2005). The secondary bond market is as an effect relatively illiquid compared to others. The bond and equity market are also subject to heavy regulations - like short-sale restrictions - in contradiction to the CDS market. CDS spreads are therefore more responsive in the short run to new credit information. Studies also show that for riskier credit the CDS market is more liquid than its underlying reference entity, taken the lower bid-ask spreads in consideration. Consequently, due to higher market liquidity and lead in the pricing

mechanism process, it is generally accepted that CDS spreads are used to determine the ‘market implied rating’ (Blanco, Brennan and March; 2005).

Forte and Pena (2009, p. 1) concluded that equity price developments precedeCDS and bond market prices more frequently than vice versa. In addition, equity volatility explains 14% to 18% of the total volatility of credit spreads after controlling for firm specific factors in the regressions (Blanco, Brennan and March, 2005).

Bystrom (2005, p. 10) did an investigation on the correlation between the European iTraxx - CDS index containing the 125 most liquid firms – and the stock market. He found

(9)

9

that the CDS spreads widened as stock prices were decreasing and tightening spreads as stock prices were increasing.

As can be expected CDS spreads appear to be negatively related to its credit ratings (Hull, Predescu & White, 2004). CDS spread levels appear to have significant effect on downgrades of credit ratings. However, for positive rating events there was clearly less evidence found.

2.4 M&A announcements’ relation to bond and equity markets

Unquestionably M&A have a significant impact on the credibility and liquidity of firms. In this section it is examined what the effects of a M&A announcement will be on the bond and equity markets.

Studies show that target firm bonds generate excess returns at the announcement of a merger, this indicates a higher credit spread. For acquiring firms, however, there is no exclusive effect found. Outcomes vary from significantly positive to significantly negative and insignificant results. Clear is however that the wealth effects for target equity is more visible than acquirer firms wealth effects (Billet, Dolly King & Mauer, 2004).

There is also theoretical literature that shows how mergers can have positive wealth effects for bondholders of both the acquiring and target firm when their cash flows are imperfectly correlated. This, so called, coinsurance results in a lower default risk for both parties (Billett, Dolly King & Mauer, 2004).

The use of leverage can have a negative effect on the acquiring firm bonds. It both increases the probability of default and deadweight costs. Mergers can also lead to the reordering of priority of claims in bankruptcy which will have a negative effect (Eisenthal, 2009).

Higher leverage is only likely to have positive wealth effects for acquirer bond holders if the benefits of increased leverage are sufficiently high. The most evident advantage of leverage are tax shield gains by the reduction of taxable income through higher interest costs.

The effect of M&A on shareholders wealth is controversial. On the one hand M&A can exploit imperfections in the market which creates wealth for the acquiring firms, just as synergy. Sympathizers of this theory conclude that M&A in general are profitable. A different perspective is the one from a modern corporation where ownership is separated from the

(10)

10

control of a firm. From this management point of view firms are seeking for size instead of profit maximization, often acquired at the expense of the equity holder. As a consequence shareholders are expected to earn a lower than normal return (Mandelker, 1974). The last is supported my most of the modern empirical research.

2.5 Valuation of CDS spreads

As has been told earlier the premium paid on a CDS is based on the discounted cash flows of the premiums paid and the expected probability of a credit event and the corresponding losses. A fair spread will equal the value of the premium leg and the default leg. For the following model used we must assume that the period of default and the interest rates are independent factors (O’kane & Turnbull, 2003).

The expected value of the premium leg consist out of two components. The first component is the PV of the sum of all premium payments with survival probability 𝑝_𝑡,𝑖 taken into account. Where 𝐷_𝑡,𝑖 is the discount factor, 𝑆 the CDS spread and 𝑑 the accrual days - which is usually 0.25 since premiums are paid quarterly- at time 𝑡 for firm 𝑖. This gives:

∑𝐷_𝑡,𝑖∗ 𝑝_𝑡,𝑖∗ 𝑆_𝑡,𝑖 ∗ 𝑑 (1)

The second component consists of the PV of all expected accrued payments:

∑ 𝐷𝑡,𝑖∗ [𝑝𝑡−1,𝑖− 𝑝𝑡,𝑖] ∗ 𝑆𝑡,𝑖 ∗𝑑₂𝑖 (2)

Next the default leg is the PV of the transfer made from the seller of the CDS to the buyer when in the situation of a credit event. With 𝑅 as the recovery rate it gives the following:

(1 − 𝑅) ∗ ∑ 𝐷𝑡,𝑖∗ [𝑝𝑡−1,𝑖− 𝑝𝑡,𝑖] (3)

The initial point from where we derive the CDS spread is where the premium leg and default leg are equal, so 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 1 + 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 2 = 𝑒𝑞𝑢𝑎𝑡𝑖𝑜𝑛 3. This results in:

(11)

11

∑ 𝐷𝑡,𝑖∗ 𝑝𝑡,𝑖∗ 𝑆𝑡,𝑖∗ 𝑑 + ∑ 𝐷𝑡,𝑖 ∗ [𝑝𝑡−1,𝑖− 𝑝𝑡,𝑖] ∗ 𝑆𝑡,𝑖∗𝑑₂𝑖= (1 − 𝑅) ∗

∑ 𝐷_𝑡,𝑖∗ [𝑝_{𝑡−1,𝑖}− 𝑝_𝑡,𝑖] (4)

If we solve for 𝑆 we arrive at the formula that gives the CDS spread value: (5)

𝑆𝑡,𝑖 =

(1 − 𝑅)∗∑𝐷𝑡,𝑖∗[𝑝𝑡−1,𝑖− 𝑝𝑡,𝑖]

∑𝐷𝑡,𝑖∗ 𝑝𝑡,𝑖∗ 𝑑 +∑𝐷𝑡,𝑖∗[𝑝𝑡−1,𝑖− 𝑝𝑡,𝑖]∗ 𝑆𝑡,𝑖∗

𝑑𝑖

2

According to O ‘Kane and Turnbull (2003) the precise definition of R is the expected price of ‘the cheapest to deliver’ obligation of the protection in the case of a notified credit event. The recovery rate 𝑅 and probability of survival 𝑝𝑡,𝑖 are uncertain variables that may change over

time. Since M&A directly affect the probability of survival and recovery rates it can be assumed that the CDS spreads are sensitive to such announcements.

(12)

12

3. Research Design

In the third section the hypothesis will be introduced and a general outline of the research will be given. First the assumptions that needed to be made for this research will be discussed, followed by the research method that has been used. Afterwards the data used in this thesis will be described.

3.1 Assumptions

Until 1970 there was not yet a clear theory on the reflection of informational content in stock prices and capital markets. Fama (1970, p. 414) started using event studies to look for

evidence on how stock prices reflect information introduced to the public. His Market

Efficiency Hypothesis takes the degree and pace of informational content absorbed by capital markets into consideration. There would be an efficient market if the prices fully reflect all information available at a given moment (Fama, 1970).

Fama (1970, p. 414) makes a distinction between three different types of efficient markets; the weak form, semi-strong form and the strong form. The weak form suggests that historical prices and returns is the only information available on the market. The semi-strong form assumes that all publicly available information is reflected in the data. The most

restrictive category, the strong form, has the assumption that both publicly available information as inside information are incorporated in the stock prices. The strong form implies that the outperformance of the stock market based in informational knowledge, like insider trading, is unattainable (Fama, 1970). Taken earlier research into consideration the semi-strong market hypothesis is assumed in the thesis.

There has been a revision of the Market Efficiency Hypothesis that was challenged by literature on long-term return anomalies. It has been concluded that these long-term return anomalies in the economy can be assigned to chance. The divergence in the market by an overreaction occurs as often as an underreaction. The same counts for abnormal returns in an event study. The continuation of abnormal returns, which will be discussed next section, after an event is as likely as the opposite (Fama, 1998).

Perfect market liquidity needed to be assumed in the CDS markets that will be examined. Market liquidity allows new informational content to be immediately adopted in

(13)

13

the market prices. Investors should be able to alter their portfolios instantly at a relatively low cost. As been stated before, CDS market are more liquid compared to bond markets due to the absence of regulation and a more efficient pricing mechanism (Blanco, Brennan and March; 2005). This therefore seems a valid assumption.

3.2 Event Study

The use of event studies is already proven in many fields of research, from law to economics. The methodology was initially developed to evaluate the impact of a given event, like

earnings announcement or issuance of new debt, on common stocks. MacKinlay (1997, p.13) introduces this technique to the application of debt securities with little modifications.

Therefore, his methodology will be used as a guidance for this study on testing whether there is a significant effect visible of M&A announcements on CDS spreads. The M&A

announcements will act as event in this study.

In order to perform the analysis I needed to make a clear distinction between two different time periods; the estimation window and the event window. As the name already tells the M&A announcement will occur in the event window at t = 0. Below a schematic overview is shown of the different timeframes. A post event window is not included in this thesis since the outcome of the announcement, success or failure, is not relevant for the research question.

Figure 2. Time scheme event study

For our analysis I will make use of the market model instead of the constant mean return model. The first assumes a stable linear relation between the examined CDS spreads and market return whereas the second assumes a constant relation between the market and CDS.

(14)

14 For any CDS 𝑖 at time 𝑡 the market model is given as:

𝑅𝑖,𝑡 = ∝𝑖+ 𝛽𝑖∗ 𝑅𝑚𝑘𝑡,𝑡+ 𝜖𝑖,𝑡

With 𝐸(𝜖_𝑖,𝑡 = 0) and 𝑉𝑎𝑟(𝜖_𝑖,𝑡) = 𝜎_𝜖2_𝑖

The corresponding parameters of the model for each firms CDS are calculated over the

estimation window, from 𝑡 = −140 up until 𝑡 = −11. If the event window would be included in the estimation of the corresponding parameters these could be soiled by the returns around the M&A announcement. This would lead to an inaccurate calculation of the normal and abnormal returns which is essential for this methodology.

Next the normal returns for the event window 𝑡 = −10 up until 𝑡 = 10 are calculated with use of the corresponding market model 𝐸(𝑅𝑖,𝑡|𝑅𝑚𝑘𝑡). The normal returns are needed to

derive the abnormal returns (AR), which is essential in this methodology. The AR are the indication for a possible effect of M&A announcements on CDS spreads and will eventually be tested for significance.

The AR is derived by calculating the difference between the actual and the normal returns: 𝐴𝑅𝑖,𝑡 = 𝑅𝑖,𝑡− 𝐸(𝑅𝑖,𝑡|𝑅𝑚𝑘𝑡) = 𝑅𝑖,𝑡− (∝𝑖+ 𝛽𝑖 ∗ 𝑅𝑚𝑘,𝑡)

According the MacKinlay (1997, p.18) the market model is a potential improvement over the constant mean return model. It has the ability to reduce the variance of the abnormal returns. A lower variance results in a more accurate calculation of the M&A announcement effects. The gain of the application of the market model over the constant mean return model is greater as the R-squared of the market model regression increases. If the length of the estimation window is of sufficient size the variance of 𝐴𝑅_𝑖,𝑡 will become equal to 𝜎2_(𝐴𝑅

𝑖,𝑡) = 𝜎𝜖𝑖

2_{. Under 𝐻}

0 the distribution becomes 𝐴𝑅𝑖,𝑡 ~ 𝑁⌊0, 𝜎2(𝐴𝑅𝑖,𝑡)⌋ as the daily

returns are expected to be close to zero (Fama, 1997).

The following step is to aggregate the data over two dimensions, time and M&A announcement observations. For the aggregation we must assume independence across

(15)

15

observations, so no overlap in event windows was allowed to occur (MacKinlay, p. 14). For 𝑁 observations the sample aggregated abnormal returns is given by:

𝐴𝑅 ̅̅̅̅_𝑡₌1

𝑁∑ 𝐴𝑅𝑖,𝑡 𝑁

𝑖=1 and 𝑉𝑎𝑟(𝐴𝑅̅̅̅̅𝑡) =_𝑁12∑𝑁𝑖=1𝜎𝜖2𝑖

Secondly the aggregation over event windows is done by cumulating the sample aggregated abnormal returns (SACAR):

𝑆𝐴𝐶𝐴𝑅(𝑡₁, 𝑡₂) = ∑𝑡2 𝐴𝑅̅̅̅̅̅_𝑡

𝑡=𝑡1 and 𝑉𝑎𝑟[𝑆𝐴𝐶𝐴𝑅(𝑡1, 𝑡2)] = (𝑡2− 𝑡1+ 1) ∗ 𝑉𝑎𝑟(𝐴𝑅̅̅̅̅𝑡)

The variance of the SACAR is found by taking the square root of the summation of the variances of abnormal returns. The variance is needed to test the null hypothesis for significance. The hypotheses are stated as:

𝐻₀: 𝑁𝑜 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛 𝑎𝑓𝑡𝑒𝑟 𝑎 𝑀&𝐴 𝑎𝑛𝑛𝑜𝑢𝑛𝑐𝑒𝑚𝑒𝑛𝑡 𝐻₁: 𝐴𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛 𝑎𝑓𝑡𝑒𝑟 𝑎𝑛 𝑀&𝐴 𝑎𝑛𝑛𝑜𝑢𝑐𝑛𝑒𝑚𝑒𝑛𝑡

With the use of the t-test statistic it can be tested whether there is sufficient evidence to reject the null hypothesis. The t-statistic is calculated by the following formula and is standard normally distributed:

𝜃 = 𝑆𝐴𝐶𝐴𝑅(𝑡1, 𝑡2) √𝑉𝑎𝑟[𝑆𝐴𝐶𝐴𝑅(𝑡1, 𝑡2)]

~𝑁(0,1)

For this thesis a two-sided t-test will be performed to confirm whether the M&A

announcement have an effect; and to determine whether this is positive or negative. The null hypothesis can be rejected with a certainty level of 95% if |𝜃| > 1.645 in a one-sided test. And with a 99% certainty level if the t-value exceeds |𝜃| > 2.33.

(16)

16 3.3 Data

The data used in this thesis has been acquired from multiple sources. In this section the data - where this research rests upon - will be discussed. It will become clear where the data is acquired and the reasoning behind some choices of particular data.

The CDS securities used in this thesis are collected from the Depository Trust & Clearing Corporation (DTCC). Firms with the highest CDS notional outstanding between October 2008 and March 2013 were gathered to establish a dataset, consisting of 1215 firms. The corresponding CDS spreads were acquired from Thomas Reuters’s Datastream. The CDS for every individual firm have different maturities like bonds. The spreads of CDS with a maturity of 5 years have been used because of their relatively high liquidity and volume.

The M&A announcements were gathered from Compustat, a database for worldwide financial and statistical information. I have searched for the 150 largest in value

announcement from January 2010 up until December 2014 for listed firms. The observations are all listed on a U.S. stock exchange market, either S&P 500 or NYSE. The completion or success of the merger or acquisition was not a criteria for this research since we are only interested in the announcement. The announcement dates fell in weekends or holidays were adjusted to the first following working day to match the CDS spread data.

After the M&A announcements and database for CDS spreads were gathered we had to merge the two datasets. From the 150 M&A announcements a mere 91 had to be dropped because of the absence of the involved corporation in the CDS database. From the 59 left another 14 had to be dropped because of the non-existence of sufficient market data - needed for the Market Model Event Study – or unusable prices in the CDS database. It was noticed that some collected CDS spreads did not move during the whole period, these were dropped out as well. The observations used in this thesis can be found in the appendix.

In order to calculate the abnormal returns in the event study the market model has been used. With the use of this model normal expected returns – based on the market return – for the particular CDS security can be calculated. A higher R-squared will lead to a greater benefit of the market model of the constant mean return model, it reduces the standard deviation of the errors. As a result this will lead to more accurate t-statistics. Since all the observed corporations are listed in the U.S. I have used an U.S. CDS index, the CDX.NA.IG index mainly based on the investment grade of firms. This index consists of 125 firms and is

(17)

17

composed by three criteria. The firms average spread over the last 90 must not exceed five times the average CDS spread of the index, have a higher than BBB- rating and must adhere to some liquidity rules (Market index rules, 2013). In figure 3 the course of the CDX index is illustrated.

Figure 3. Markit CDX-NAIG 5Y OTR index – Model price %

94 95 96 97 98 99 100 101 102 103 11-6-2010 11-6-2011 11-6-2012 11-6-2013 11-6-2014 11-6-2015

(18)

18

4. Empirical Results

After the performance of the event study we can analyse the outcome of research. I will first present and discuss the results that are found. Afterwards we will test whether the results are statistically significant. With the help of the outcomes it will be decided whether the null hypothesis can be rejected.

4.1 Descriptive Analysis

For the 45 observations I performed regressions between the CDX index and the

corresponding CDS spreads over the estimation window. Next, the normal returns could be predicted and compared with the realized returns to find the abnormal returns. In figure 4 the average abnormal returns over time is displayed just as the cumulative abnormal returns (SACAR). It is interesting to monitor the SACAR closely, trends are here clearly observable. What can be seen is a decreasing trend from t = -140 till t = -85. It gets followed by a period of increase which stops just before the announcement date t = 0. At first sight an abrupt decrease in AR can be spotted after the announcement date t = 0 which implies a relative tightening of the CDS spreads.

Figure 4. Average Abnormal Returns for period t = -140 to t = 10

-6,00% -5,00% -4,00% -3,00% -2,00% -1,00% 0,00% 1,00% 2,00% 3,00% 4,00% -1 40 -1 35 -1 30 -1 25 -1 20 -1 15 -1 10 -1 05 -1 00 _-95 _-90 _-85 _-80 _-75 _-70 _-65 _-60 _-55 0_-5 _-45 _-40 _-35 _-30 5_-2 _-20 _-15 _-10 -5 0 5 10 R eturn Time AR SACAR t=-140 to t=10

(19)

19

Figure 5. Average Abnormal Returns for period t = -10 to t = 10

When we take a closer look into the course of AR and SACAR at the event window clearly a reversal at t = 0 can be spotted. After a series of positive returns, three consecutive days with negative return. Seemingly, this in turn gets followed up by a price correction at t = 3. From the theoretical background it can be suggested that the drop in CDS spread, and accordingly default risk, may be caused by positive wealth effects expected form the M&A for the

shareholders as stock prices are negatively correlated with CDS spreads. Gains by imperfectly correlated cash flows, synergy or tax shield benefits may be credible explanations (Billett, Dolly King & Mauer, 2004). However, most modern empirical research concluded that the wealth effects from M&A were more frequently negative due to the separation of control and ownership of the firms (Mandelker, 1974). Therefore, a drop in CDS spread is different than was anticipated.

The sharp increase of the spreads just before the announcement date is from theoretical point of view also difficult to explain. It may be due to uncertainty caused by speculative rumours prior to the announcement. The relatively high responsiveness of CDS to such events, compared to bonds, could be an explanation for the volatility around the

-3,00% -2,00% -1,00% 0,00% 1,00% 2,00% 3,00% 4,00% -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 R eturn Time AR SACAR t=-10 to t=10

(20)

20

announcement date (Blanco, Brennan and March; 2005). Expected negative wealth effects, as a result of the M&A, for the shareholders does not seem a credible argument since this is not observable after the announcement. From studies investigating the effect of M&A’s on acquirers’ bond yields no exclusive conclusion could be drawn.

Whether these price changes are actually a result of the M&A announcement will be tested in the following section.

4.2 Statistically Analysis

I will test the outcome of our event study based on a one-sided statistical t-test. As been shown before, the t-value is calculated with the use of its standard deviation and has a standard normal distribution. The null hypothesis can be reject respectively at a 99% or 95% certainty level for absolute t-values of t = 2.33 and t = 1.64. In table 2 the abnormal returns and corresponding t-values in the event window are given.

Table 2 & 3. Daily abnormal returns in event window / t-test

t -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 0 AR 1,53% -0,13% -0,15% 0,85% 0,01% 0,14% -0,07% 0,10% 0,17% 0,42% -1,12% t-stat 3,94** -0,33 -0,38 2,19** 0,02 0,37 -0,18 0,25 0,45 1,08 -2,87** t 0 1 2 3 4 5 6 7 8 9 10 AR -1,12% -1,13% -2,16% 0,60% 0,05% -0,18% -0,08% -0,41% -0,25% 0,08% 0,05% t-stat -2,87** -2,89** -5,54** 1,55 0,14 -0,46 -0,22 -1,06 -0,65 0,20 0,12 * & ** indicates a significance level of 5% or 1% with a one-sided t-test

From table 2 can be observed that t = -10 and t = -7 have significantly positive t-values. They are mainly responsible for the increase of SACAR before the announcement date and the indication for an upward trend. When I started digging deeper into the abnormal returns of the individual firms for these time periods I noticed that mainly two firms were responsible for the given abnormal returns. At t = -10 Talisman Energy Inc had an AR of 12.7% and BHP Billiton LTD 59%. For t = -7 Talisman Energy Inc and Dish Network Corporation were mainly accountable for the AR with respectively 13.3% and 30.3%.

(21)

21

Table 4. SACAR periodically / t-test

* & ** indicates a significance level of 5% or 1% with a one-sided t-test

If we look at table 3 it can be seen that for t = 0, t = 1 and t = 2 significant results are found with a certainty level of 99%. Also for these periods I have looked into the AR of the

individual firms. Such an accountability for the AR, as in t = -10 and t = -7, by a small group of firms was not found.

The statistical significance over different time frames is given in table 4. As can be observed there is a slight decrease over the period from t = -140 to t = 10. However, this is proven not to be significantly different from zero. For the SACAR around the M&A announcement, between t = -10 and t = 10, there is sufficient proof to reject the null

hypothesis. It can be concluded that due to the M&A announcement the CDS spreads between t = -10 and t = -1 are rising, and after the announcement are tightening. The null hypothesis can therefore be rejected in this study.

t -140 to 10 -10 to 10 -10 to -1 0 to 1 0 to 3 0 to 10 SACAR -1,671% -1,671% 2,880% -2,246% -3,799% -4,550%

Var(SACAR) 0,228% 0,030% 0,015% 0,003% 0,006% 0,017%

(22)

22

5. Conclusion and Discussion

In this last section I will summarize the conclusions that can be drawn from this study. Next, the limitations of the research method and data will be discussed. Hereafter the practical and theoretical implications will be proposed. Finally possible recommendations for future research on this topic will be given.

At a significance level of 1% for the one-sided null hypothesis test, it can be concluded that the directional effect of an M&A announcement on its acquirer’s CDS spread prior to the announcement, between t = -10 and t = -1, is positive. Alternatively, from t = 0 till t = 10 there is sufficient proof for tightening CDS spreads due to the announcement.

There is theoretical literature that supports both the rise and tightening of the CDS spreads around the M&A announcement, however, they are not undisputed. It can be argued that the relative responsiveness of CDS spreads, to new credit information on the short run, and expected rise of default risk are accountable for the rise in spreads prior to the announcement (Blanco, Brennan and March; 2005). However, if the rise would originate from leakage of information this trend would be expected to continue after the announcement, this is not observable. From the announcement onwards it is statistically proven that the spreads are tightening, indicating a decrease of credit risk. A decrease in credit risk could be explained by positively expected wealth effects for shareholders due to synergy, coinsurance or tax shield gains. But multiple studies that investigated the effect of M&A on acquirers bond yields, which should be closely related to CDS spreads, did not exclusively found a positive or negative effect (Hull, Predescu & White, 2004). Therefore, further research is needed to specify the drivers behind the rising and tightening CDS spreads.

This study carried some limitations that would suggest for further investigation. The

relatively small sample size, due to difficulty of merging and collecting data, is more sensitive to price movement of individual firms. It was specified that the rise of CDS spreads prior to the announcement could be related to three individual firms, although showing significant returns.

Next, for this research only the largest M&A announcement over the last five years were taken into consideration. The results of this thesis can therefore not be extended to M&A

(23)

23 announcement of a smaller degree.

Furthermore, as in the methodology was discussed a higher R-squared between the market index and CDS spread could lead to more accurate prediction of the AR. The R-squared found in this study varied from 62% to less than 1%. Due to lack of availability to CDS index data it could not be concluded whether the CDS index used was the most appropriate one.

The results of this study could be insightful for speculators and CDS traders. The practical implications of this study could help traders anticipating on M&A announcements more profitable. By either hedging risk more appropriate or incorporating the outcome of this thesis in pricing models which helps in trading strategies.

To my knowledge, this is the first study that investigates the directional effect of M&A announcement on its acquirers CDS spread. My thesis will contribute to a better comprehension of the interaction between bond and equity markets and CDS. Furthermore, the evidence for an effect of M&A announcement on CDS spreads could encourage

researcher for further investigation.

During my thesis I came across some possible recommendations for future research. It could be investigated to what degree the financing structure of the M&A has an impact on the CDS spreads. And, are the M&A effects for the acquirers and targets spread significantly different? In this thesis I only made use of the largest M&A announcement over the last five years, it could also be researched to what extent the size of a M&A has an influence on the CDS spreads.

(24)

24

References

Ashcraft, A. B., & Santos, J. a C. (2009). Has the CDS market lowered the cost of corporate debt? Journal of Monetary Economics, 56(4), 514–523.

doi:10.1016/j.jmoneco.2009.03.008

Billett, M. T., King, T. H. D., & Mauer, D. C. (2004). Bondholder Wealth Effects in Mergers and Acquisitions: New Evidence from the 1980s and 1990s. Journal of Finance, 59(1), 107–135. doi:10.1111/j.1540-6261.2004.00628.x

Blanco, R., Brennan, S., & Marsh, I. W. (2005). An empirical analysis of the dynamic relation between investment-grade bonds and credit default swaps. Journal of Finance, 60(5), 2255–2281. doi:10.1111/j.1540-6261.2005.00798.x

Bolton, P., Oehmke, M., 2011. Credit default swaps and the empty creditor problem. Review of Financial Studies 24, 2617–2655.

Byström, H. (2005). Credit Default Swaps and Equity Prices: The iTRAXX CDS Index market. Knowledge Creation Diffusion Utilization, 14(4), 1–14.

doi:10.3905/jfi.2005.491109

Campello, M., & Matta, R. (2012). Credit default swaps and risk-shifting. Economics Letters,

117(3), 639–641. doi:10.1016/j.econlet.2012.08.013

Eisenthal, Y. (2009). Pricing of LBO Risk in Credit Spreads. World, (July).

Fama, E. F. (1970). JSTOR: The Journal of Finance, Vol. 25, No. 2 (May, 1970), pp. 383-417. Journal of Finance.

Fama, E. F. (1998). Market efficiency, long-term returns, and behavioral finance. Journal of

Financial Economics, 49(3), 283–306. doi:10.1016/S0304-405X(98)00026-9

Forte, S., & Peña, J. I. (2009). Credit spreads: An empirical analysis on the informational content of stocks, bonds, and CDS. Journal of Banking and Finance, 33(11), 2013–2025. doi:10.1016/j.jbankfin.2009.04.015

Hull, J., Predescu, M., & White, A. (2004). The relationship between credit default swap spreads, bond yields, and credit rating announcements. Journal of Banking and Finance,

28(11), 2789–2811. doi:10.1016/j.jbankfin.2004.06.010

International Swap and Derivatives Association market survey (2010), Notional amounts

outstanding, semi-annual data, all surveyed contract, 1987-2010.

MacKinlay, A.C. (1997). Event Studies in Economics and Finance. Journal of Economic Literature, Vol. 35, 13-19

(25)

25

Mandelker, G. (1974). Risk and return: The case of merging firms. Journal of Financial

Economics, 1(4), 303–335. doi:10.1016/0304-405X(74)90012-9

Markit CDX High Yield & Markit CDX Investment Grade Index Rules. (2013), (March). O’Kane, D., & Turnbull, S. (2003). Valuation of Credit Default Swaps. Lehman Brothers

Quantitative Credit Research, (April). Retrieved from

http://iscte.pt/~jpsp/Teaching/Credit_MMF/Handouts/Okane and Turnbull, Lehman Brothers 2003, Valuation CDS.pdf

Saretto, A., Tookes, H., Becker, B., Denis, D., Esty, B., Faulkender, M., Schwert, W. (2012). Corporate Leverage, Debt Maturity and Credit Supply : The Role of Credit Default Swaps ∗ University of Texas at Dallas, (972).

Subrahmanyam, M. G., & Wang, S. Q. (2013). Credit Default Swaps and Corporate Cash.

Working Paper.

Weistroffer, C. (2011). Credit Default Swaps: heading towards a more stable system, 177– 198. doi:10.1002/9781118266403.ch13

(26)

26

Appendix

M&A announcement overview

Date Announced Acquiror Name Target Name

15-12-2014 Talisman Energy Inc Repsol SA

8-12-2014 Merck & Co Inc Cubist Pharmaceuticals Inc

24-9-2013 Applied Materials Inc Tokyo Electron Ltd

18-5-2014 AT&T Inc DirecTV Inc

17-8-2010 BHP Billiton PLC Potash Corp of Saskatchewan

5-10-2014 Becton Dickinson & Co CareFusion Corp

13-2-2014 Comcast Corp Time Warner Cable Inc

4-5-2011 ConAgra Foods Inc Ralcorp Holdings Inc

11-7-2014 Reynolds American Inc Lorillard Inc

15-4-2013 DISH Network Corp Sprint Nextel Corp

16-6-2014 Level 3 Communications Inc tw telecom inc

15-6-2014 Medtronic Inc Covidien PLC

15-6-2014 Williams Partners LP Access Midstream Partners LP

29-5-2014 Tyson Foods Inc Hillshire Brands Co

22-4-2014 Novartis AG GlaxoSmithKline PLC-Oncology

30-7-2013 Community Health Systems Inc Health Management Assoc Inc

29-7-2013 Elan Corp PLC Perrigo Co

14-11-2014 Halliburton Co Baker Hughes Inc

21-7-2011 Hewlett Packard Co Hewlett Packard Co

30-6-2013 Amgen Inc Onyx Pharmaceuticals Inc

27-5-2013 Bausch & Lomb Inc Valeant Pharmaceuticals Intl

15-4-2013 Thermo Fisher Scientific Inc Life Technologies Corp

12-2-2013 Comcast Corp NBCUniversal Media LLC

25-2-2013 Lowe's Cos Inc Lowe's Cos Inc

29-6-2012 Bristol-Myers Squibb Co Amylin Pharmaceuticals Inc

(27)

27

19-6-2012 Walgreen Co Alliance Boots GmbH

21-5-2012 Eaton Corp Cooper Industries PLC

2-2-2012 Glencore International PLC Xstrata PLC

16-10-2011 Kinder Morgan Inc El Paso Corp

21-9-2011 United Technologies Corp Goodrich Corp

21-7-2011 Express Scripts Inc Medco Health Solutions Inc

20-7-2011 Nalco Holding Co Ecolab Inc

28-4-2011 Exelon Corp Constellation Energy Group Inc

18-4-2011 Johnson & Johnson Synthes Inc

14-3-2011 Lubrizol Corp Berkshire Hathaway Inc

25-1-2012 Roche Holding AG Illumina Inc

18-7-2013 Schlumberger Ltd Schlumberger Ltd

21-2-2011 Reliance Industries Ltd-21 Oil BP PLC

26-1-2011 ProLogis AMB Property Corp

10-1-2011 Duke Energy Corp Progress Energy Inc

1-1-2014 Time Warner Inc Time Warner Inc

15-11-2010 Caterpillar Inc Bucyrus International Inc

7-9-2010 Enterprise Products Partners Enterprise GP Holdings LP

An event study : the directional effect of M and A announcements on the acquiring firms’ CDS spread