Effects of CDS spreads on M&A returns, does monitoring matter?
An empirical study about domestic US M&A deals after the start of the crisisAbstract
This thesis analyzes whether well-managed acquirers, that may have a lower default risk, have higher CARs or that the CAR is higher when the default risk of the acquirer is higher, since the acquirer may then be better monitored. To research this a dataset was used that consists of 281 US domestic M&A deals that took place between 15 September 2008 and 1 September 2014. The CDS spread is used as a proxy for the default risk. It is found that one standard deviation increase in the log of the CDS spread of the acquirer increases the CAR of the acquirer by approximately 2.43%. This effect is found to be significant at the 10% level.
Supervisor: Mark Dijkstra MSc
Student: Jasper Verwoerd
Student Number: 10357556 Bachelor: Economics and Business
Specialization: Finance and Organization Date: 25-01-2015
This document is written by Jasper Verwoerd 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.
Table of Contents
1. Introduction ... 4
2.Literature Review ... 6
2.1 Theoretical review ... 6
2.1.1 The negative effect of default risk on abnormal returns ... 6
2.1.2 The positive effect of default risk on abnormal returns ... 7
2.2 Empirical Review ... 9
2.3 Literature review conclusion ... 12
3. Methodology ... 13
3.1 Regressions and Hypothesis ... 13
3.2 Description of the variables ... 13
4. Data description ... 16
5. Results ... 19
5.1 Description and interpretation of the results ... 19
5.2 Robustness and alternative tests ... 23
5.2.1 Target default risk ... 23
5.2.2 10-day CAR ... 25
5.2.3 Other event windows ... 26
6. Conclusion ... 29
Reference list ... 31
Appendix ... 35
Appendix I Histograms of the CDS spreads of the acquirer ... 35
Appendix II Histograms of relative size ... 36
1. Introduction
Default risk may impact M&A deals in two ways. First, acquirers that are better managed do acquisitions with higher abnormal returns (Lang, Stulz, & Walking, 1989). Acquirers that are better managed may also have a lower default risk, which means a negative relationship between default risk and abnormal returns is expected. On the other hand, a higher default risk leads to more monitoring (Parlour & Winton, 2013). While, acquisitions wherein the acquirer is more monitored have higher abnormal returns (Kang, Shivdasani, & Yamada, 2000). Furthermore, managers of firms that have more free cash, are less afraid of default and perform more value-destroying acquisitions (Jensen, 1986) These firms might also have a lower default risk, so that acquirers with a lower default risk do worse acquisitions. It has been argued that these diversifying acquisitions should lead to higher abnormal returns
(Lewellen, 1971). Firms that have a higher default risk might also gain more by reducing their default risk through diversifying acquisitions. To research the possible effect of default risk on the abnormal returns the following research question is analyzed: Does the default risk of the acquirer have an effect on his CAR of M&A deals after the start of the crisis? Empirical tests of the effect of default risk on the CAR of the acquirer were not published before to the author’s knowledge, so this thesis contributes to the academic literature by testing this relationship.
To research this effect the CDS spread is used as a proxy for the default risk. The abnormal returns are calculated using the CAPM. To get the CAR the abnormal returns are cumulated over an event window starting five days before the announcement and ending at the completion date of the M&A deal. A fixed event window is not used since this might lead to biases because the full effect of the M&A deal might not be included. This could be the case because untill the completion date of the M&A deal the CAR also reflects uncertainty and this uncertainty might be related to the default risk of the acquirer1. OLS regressions are done with the CAR as the dependent variable and the log of the CDS spread as the main explanatory variable.
The dataset that is used in this research consists of 281 US M&A deals that were announced between 15 September 2008 and 1 September 2014. The mean CAR is -1.86% and not significantly different from zero, while the mean CDS spread of the acquirer is 149.24
1 Bradley, Desai, and Kim (1983) argue that around the first announcement date there might still be uncertainty about whether the merger or acquisition will actually go through, and this might influence the CAR. Chen, Harford and Li (2007) furthermore argue that withdrawal of bad bids is more likely to happen when the acquirer is more monitored, while Parlour and Winton (2013) argue that the default risk is a determinant of the amount of monitoring a firm receives. So firms with a higher default risk should have more uncertainty about whether the deal will actually go through or whether the bid will be withdrawn.
basis points.
In the complete model the coefficient on the log of the CDS spread of the acquirer is 2.926. This means that if the log of the CDS spread increases by one standard deviation (0.8302), then the CAR increases by approximately 2.43%. This relationship is significant at the 10% level. This effect cannot be fully explained by the diversification argument, however it is found that the higher the default risk of the acquirer the more the acquirer benefits by diversifying, though this effect is not significant.
The significance of the effect of default risk on the CAR of the acquirer doesn’t persist using fixed event windows, even though the effect remains positive. Adjusting the flexible event window of each M&A deal for its length2, leads to a more significant positive effect of the default risk of the acquirer on the CAR of the acquirer. The CDS spread of the target was also added to the regressions in a robustness test. However adding this reduced the sample to 30 observations so that no significant results were found and no conclusions can be drawn on the basis of the sample.
In the second section the theoretical literature and empirical evidence are discussed. Section three and four discusses the methodology and data. In section five the results and their robustness are discussed. Finally, section seven contains the conclusion.
2 This is done by by dividing the CAR of the acquirer by the number of days that the M&A deal took to complete.
2.Literature Review
This section starts with a discussion about M&A and the relationship of the default risk with the CAR of the acquirer. Subsequently a short overview is given of empirical findings that are related to the default risk and the CAR of the acquirer.
2.1 Theoretical review
In acquisitions an acquirer purchases the stock or assets of a target firm (Berk & DeMarzo, 2011). If investors expect that the NPV of the acquisition is positive then the stock price should increase, since the NPV is the expected dollar gain of the acquirer’s shareholders (Halpern, 1983). Central to research on M&A is the event study approach where abnormal returns are calculated by comparing realized returns to predicted returns (Halpern, 1983)3. A
research by Andrade, Mitchell and Stafford (2001) suggests that the shareholders of the target firms gain significantly, since the median premium that is paid to them is 37.9% over the time period 1973-1998. However, the shareholders of the acquiring firm earn on average a zero abnormal return (Fuller, Netter, & Stegemoller, 2002). Firms that are better managed have higher abnormal returns (Lang, Stulz, & Walking, 1989), and may also have a lower default risk. However firms that have a higher default risk should also incur more monitoring (Parlour & Winton, 2013), monitoring of the acquirer also leads to higher abnormal returns for that acquirer (Kang, Shivasani, & Yamada, 2000).
2.1.1 The negative effect of default risk on abnormal returns
Lang et al. (1989) argue that the abnormal returns are higher when well-managed acquirers do an acquisition and are highest when a well-managed acquirer takes over a poor-managed target. The authors argue that in this case well-managed acquirers force the management of the poor-managed target to make better use of its resources. Wheelock and Wilson (2000) also reason that well-managed banks have a lower likelihood of failure4. If this extends to other types of companies and their default risk then well-managed acquirers should have a lower default risk. This means that abnormal returns to the acquirers’ shareholders are higher when the acquirer’s likelihood of default is lower, since this is an indication that the acquirer is well managed.
3 If the abnormal returns are positive then this means that the shareholders of that firm benefit from the takeover. 4 Wheelock and Wilson (2000) do not provide a theoretical elaboration for this expectation, but instead test this expectation, and they do find that well-managed banks have a lower likelihood of failure.
2.1.2 The positive effect of default risk on abnormal returns
There are three different arguments for a positive effect of the default risk on the abnormal returns. Firstly, Jensen (1986) predicts that firms that have more cash have lower abnormal returns, however these firms might also have a lower default risk. Secondly, Banal-Estañol, Ottaviani and Winton (2013) argue that firms can reduce their default risk by doing a diversifying acquisition. This can create value by creating a higher borrowing capacity and thereby a higher tax shield, according to Brealey, Myers and Allen (2006). If this benefit is higher for firms with a higher default risk then those firms should create more value and have higher abnormal returns. Finally, Parlour and Winton (2013) argue that the higher the default risk the more monitoring that firm incurs, while Kang et al. (2000) argue that increased monitoring of the acquirer leads to higher abnormal returns.
Jensen (1986) states that large free cash flows induce firms to do value-destroying acquisitions, since the managers of those firms are not motivated to do good acquisitions by the fear of default, because a default is not likely to happen5. As such those managers tend to engage in more non-optimal acquisitions that benefit themselves, rather than the firm (Jensen, 1986). Managers can pursue their own interests by doing acquisitions with the goal of empire building6 or maximizing sales growth, according to Halpern (1983). So firms that have a
higher default risk, tend to perform acquisitions that create more value and should therefore have higher abnormal returns.
Banal-Estañol et al. (2013) reason that if firms with imperfectly correlated cash flows merge they can reduce their default risk and expected default costs and thereby increase their borrowing capacity. This may lead to a higher interest tax shield and the acquisition can thereby create value (Brealey, Myers, & Allen, 2006). This potential benefit of diversification may bigger for firms with a higher default risk7. However, it is unclear who gets the
diversification gain. So does Lewellen (1971) argue that the shareholders of the acquirer get this gain by an increase in the value of the firm, which should result in a higher equity value according to Lewellen and thus higher share prices. Nevertheless, Higgins and Schall (1975) argue that Lewellen didn’t account for the possibility that the debt could increase in value
5 A default is not likely to happen, because according to Jensen (1986) firms with more free cash have more cash available to pay the debt obligations. This makes a default less likely.
6 In empire building, managers take over firms to expand the size of the firm and not necessarily because it is a positive NPV project. The managers do these acquisitions because they think that with a bigger firm they will get more prestige and a higher wage.
7 This could be the case if the lower the default risk, the harder it is to reduce the default risk by diversifying and the lower the diversification benefit. For example if an acquirer with a low default risk takes over a firm with a medium default risk then the diversification benefits have to be greater to reduce the default risk then if an acquirer with a high default risk takes over a medium default risk firm.
instead of the equity. So do Higgins and Schall claim that bondholders of the acquirer may capture the benefits of the acquisition instead, if they can use covenants to renegotiate the debt when the M&A takes place so that the value of debt increases instead of the value of equity.
Another positive effect of default risk on abnormal returns comes from monitoring. Duchin and Schmidt (2013) claim that reduced monitoring by external analysts and investors leads to more agency-driven acquisitions. These acquisitions might destroy value leading to negative abnormal returns for the shareholders of the acquirer (Halpern, 1983). Chen, Harford and Li (2007) argue that more monitoring of the acquirer leads to higher abnormal returns for the shareholders of the acquirer, since value-destroying bids are more likely to be withdrawn under the pressure of the monitoring institution. The authors however argue that there will only be monitoring when the benefits exceed the costs, which is according to them only the case for large long-term independent institutional investors89 (2007). However, Chen,
Harford and Li (2007) only consider shareholders and don’t study the effect of monitoring by debt holders. In contrast, Kang et al. (2000) argue that monitoring by informed creditors reduces the number of acquisitions that are value-destroying. These creditors have the incentive to monitor according to Kang et al. to make sure that acquisitions that reduce the value of their debt, like the ones related to the asset-substitution problem10, don’t happen.
Since monitoring of the acquirer by its informed creditors leads to acquisitions that destroy less value, shareholders of monitored acquirers should have higher abnormal returns (Kang et al., 2000).
Parlour and Winton (2013) argue that even if there are large shareholders in a firm, there are limited options for those shareholders to affect firm policy without waging a costly proxy fight. According to Parlour and Winton this is in contrast to debt holders, who often have covenants that give them control over the borrowing firm through the threat of default. Furthermore, Parlour and Winton argue that the banks that are lending money to a firm have an incentive to monitor that firm even though they can sell the loan on the syndicated lending
8 Chen et al. (2007) define large long-term institutional investors as the top five institutional investors in an acquirer. Institutional investors are defined as those investors that were manually selected from all the investors who have a CDA 5 institutional classification. Large long-term institutional investors are those investors that were also in the top five institutional investors the year before.
9 According to Chen et al. (2007) the investors have to be large since the proportional cost of monitoring is lower for them. Secondly, the investors have to be long-term since they would otherwise not engage in monitoring but in trading according to the authors. Thirdly, Chen et al. argue that investors that are not independent may have a business-relationship with the firm and are therefore not willing to challenge and monitor the firm, since this could hurt that relationship.
10 In the asset-substitution problem shareholders choose to do projects that have a negative NPV, but increase the risk of the firm in such a way that the equity value increases, while the debt value decreases (Gavish & Kalay, 1983).
market. Banks have a reputation to protect, since buyers in the syndicated lending market take the reputation into account (Gopalan, Nanda, & Yerramilli, 2011). Those buyers specifically use the default rate of the lender’s loans as an indication of whether the lender has monitored its borrowers (Parlour & Winton, 2013). The participants in the syndicated lending market then don’t buy the syndicated loan if too much borrowers of the original seller of the loan have defaulted (Parlour & Winton, 2013) 11. Riskier firms are more likely to go bankrupt if they are not monitored, which implies that the lender is more likely to incur reputation damage in addition to possible losses on its loan (Parlour & Winton, 2013). So that the bank and other lenders have more to gain by monitoring borrowers that have a high default risk (Parlour & Winton, 2013). This means that the existence of credit risk transfers leads to more monitoring of riskier credits compared to safer credits (Parlour & Winton, 2013).
To summarize, Kang et al. (2000) argue that monitoring leads to acquisitions that create more value for the acquirer’s shareholders (and thereby higher abnormal returns), while Parlour and Winton (2013) argue that firms that have a higher default risk incur more
monitoring. As a result of this it is expected that acquirers with a higher default risk have higher abnormal returns.
However, Bolton and Oehmke (2011) warn for the empty creditor problem in which banks can obtain insurance against default by using credit default swaps, in which case they do not have an incentive to monitor. This could mean that in a research that only includes firms that have CDSs available there is no monitoring by banks, since banks can use CDS spreads to avoid monitoring. Yet, a default still leads to a worse reputation for the bank on the syndicated loan market (Parlour & Wilton, 2013). This might mean that the banks still have an incentive to monitor credit risk. Minton, Stulz and Williamson (2009) also notice that banks use credit default swaps on only 2% of their loans to hedge default risk, so the empty creditor problem may not be present in practice.
2.2 Empirical Review
To see the effect of default risk on the CAR a proxy for the default risk is needed. A proxy for this default risk is the credit default swap premium (Longstaff et al., 2005). A credit default swap (CDS) is a type of insurance contract; the party that buys the insurance pays a fixed premium, called the CDS premium or spread, each period till the moment that a default
11
If participants still buy the syndicated loan they might do this at a price for which the seller doesn’t earn any economic profit (Parlour & Winton, 2013).
occurs or the contract is matured (Longstaff et al., 2005). In exchange for this fee the seller of the CDS pays the par value of the bond when there is a credit event at the underlying firm12 (Longstaff et al., 2005).
2.2.1 The effect of default risk on abnormal returns
Park and Peristiani (2007) find a negative relationship between Tobin’s q and the likelihood of default from a bank by studying 1082 Tobin’s q ratios between 1993 and 200513. The higher the default risk14 the lower the share price of the shareholders, which leads to a reduction in the market value of equity and a decrease in the Tobin’s q (Park & Peristiani, 2007). This therefore provides some evidence for the assumption that well-managed firms have lower default risk.
Lang, Stulz and Walking (1989) use Tobin’s q as a proxy for how well managed firms are, when they study 209 US M&A deals that took place between October 1968 and
December 1986. Lang et al. (1989) find that when a high q acquirer takes over a low q target the CAR of the acquirer is 15.09% higher than when a low q acquirer takes over a high q firm. Servaes (1991) found when studying a sample of 704 mergers that took place between 1972 and 1987 that when an acquirer has a high q instead of a low q then the CAR of the acquirer is 6.36% higher. Both Lang et al. (1989) and Servaes (1991) use the reasoning of Lang et al. (1989), which can be seen in the literature section, to explain this.
Harford tested the free-cash flow theory of Jensen and finds that firms that have higher cash levels have lower abnormal returns15 by using a sample of 23686 observations from U.S.
corporations in the time period 1950-1994.
There is no consensus in empirical research on the effect of M&A with the goal to diversify. Morck, Shleifer and Vishny (1990) find that acquirers that do a diversifying
acquisition have, depending on the diversification measure used, between 9.66% and 22.22%
12 Those credit events include defaults, bankruptcy, (temporary) refusal to pay, and restructuring (Berndt, Jarro & Kang, 2007). However, the no structuring (XR) version of Credit Default Swaps only includes bankruptcy, default and (temporary) refusal to pay according to Amato and Gyntelberg (2005).
13 Park and Peristiani (2007) however only find this relationship till a probability of default of approximately 50%, if the probability of default is higher than 50% a positive relationship between Tobin’s q and default rate is expected, since if the default rate then increases the share prices fall but the put option values increases in such a way that the total equity value (including the option values) increases so that the Tobin’s q increases.
14 Till a probability of default of approximately 50%.
15 Harford (1999) finds that these acquirers destroy on average seven cents for every dollar of excess cash. Harford defines excess cash flows as the accumulated free cash flows. He uses cash deviation as an
approximation for this, which is defined as the cash that the firm has in excess of the average cash predicted for its industry divided by the total assets of that firm.
lower abnormal returns16. Both Elgers and Clark (1980) and Hornstein and Nguyen (2014) find17 that diversifying acquisitions do not have a significant effect on the CAR. So it is unclear whether diversification with the aim to reduce default risk can generate abnormal returns for the shareholders of the acquirer.
Chen et al. (2007) find that monitoring only happens by large independent long-term institutional investors18, by studying 2150 US M&A deals that took place between 1984 and 2001. However, monitoring didn’t have a significant effect on the three-day abnormal returns of the acquirer (Chen et al., 2007). Kang et al. (2000) use the leverage ratio as an
approximation for the amount of monitoring a firm should receive and find a significant 7.4% effect on the CAR for every one-point increase in the leverage ratio, by studying 154
nonfinancial Japanese acquisitions that took place between 1977 and 1993. However, Kang et al. show that this effect is because of bank leverage by showing that only the bank leverage has a significant effect. This implies that only the amount of bank debt has an influence on the CAR, Kang et al. (2000) argue that this is the case because the banks, and especially the main bank, are the informed creditors that monitor. It is worth noting that Kang et al. did their research in Japan, where main banks have closer ties to their borrowers and therefore may have more information about the borrowers than in the US (Kang et al., 2000)19.
Finally, Gopalan, Nanda and Yerramalli (2011) find that bankruptcies and default of the borrowers leads to reputation damage for the lender20, by studying 61278 loans in the US from the time period 1990 till 2006.
16 Morck et al. (1990) use two different diversification measures when they study 326 US M&A deals that took place between 1975 and 1987. The first one is a dummy variable that is equal to one if both firms share a SIC industry classification. The results of this diversification measure indicate that diversifying acquisitions have abnormal returns in the period 1980-1987 that are 11.07% or 22.22% lower. The second measure is a correlation coefficient of the monthly stock returns of the bidder and target that started three years before the takeover. These results indicate that diversifying acquisitions have abnormal returns in the period 1980-1987 which are 9.66% or 17.6% lower.
17 Elgers and Clark (1980) study 337 US mergers that took place between 1957 and 1975 and concluded that specifically mergers with the aim to diversify and thereby increase debt capacity do not have different abnormal returns. Hornstein and Nguyen (2014) studied 1030 US M&A transactions and concluded that in general diversifying acquisitions do not have different abnormal returns.
18 Cheng et al. (2007) add dummy factors for different types of investors, the number of analysts, and S&P 500 membership (to see if public scrutiny of big firms is already enough of a monitoring function) to see which of those variables an effect has on the long-term abnormal returns. They find that only the dummy of the large independent long-term shareholders an effect has on the CAR. The definition of Cheng et al. of large independent long-term institutional investors is the same as in the in the literature section.
19 Kang et al. give an example of the closer ties by noticing that banks often have executives on the boards of their clients in Japan.
20 Gopalan et al. (2011) look at the reputation damage by looking at the change in percentage of loans that are made by a lender and is subsequently retained by that lender. If this change is positive, then this means that the lender has incurred reputation damage, because he needs to retain a greater portion of his loan to signal to the possible buyers of loans that he will monitor more and that the quality of loans is stronger.
2.3 Literature review conclusion
A negative relationship between default risk and abnormal returns might be expected on the basis of the theoretical literature because well-managed acquirers may have a lower default risk. So are abnormal returns highest according to Lang et al. (1989) when well-managed acquirers take over a poor-managed target and force the target to make better use of its resources.
Instead of a negative effect there might be a positive effect of the default risk of the acquirer on the abnormal returns of the acquirer for three reasons. Firstly, firms that have more free cash, and that therefore may have a lower default risk, perform more acquisitions that destroy value, since a default is less likely so that the managers of those firms are not motivated to do good acquisitions by a fear of default (Jensen, 1986). Secondly, firms that have a higher default risk may gain more by diversifying and reducing their default risk. Reducing the default risk can result in a higher debt capacity and thereby in a higher tax shield (Banal-Estañol et al., 2013). However, it is unclear whether shareholders or debt holders of the acquirer benefit21. A third argument involves monitoring. Parlour and Winton (2013) argue that when a bank is a lender it can incur reputation damage on the syndicated lending market if a borrower defaults and as the result of this earn less. This is the reason why banks monitor borrowers with a high default risk more than firms with a low default risk (Parlour & Winton, 2013). Moreover, Kang et al. (2000) argue that bank monitoring leads to higher abnormal returns. So that firms with a higher default risk should have higher abnormal returns.
Looking at the empirical results; Park and Peristiani (2007) find that the lower the Tobin’s q of a bank the higher the default probability is of that bank, because a low Tobin’s q provides incentives for the shareholders to take risks. Furthermore, it is also found that a high Tobin’s q of the acquirer leads to higher abnormal returns for acquirers, because it means that the acquirer is well managed and can force the target to use its resources in a more efficient way (Lang et al., 1989). The market rewards this type of acquisitions with higher abnormal returns (Lang et al., 1989).
Harford (1999) finds that firms with more free cash, that may have a lower default risk, do worse acquisitions with lower abnormal returns. However, there is no consensus in the empirical literature about whether acquisitions that aim to diversify have higher abnormal
21 Lewellen (1971) argues that the shareholders of the acquirer get this gain, because the diversification acquisition creates values, which should result in a higher share price. However Higgins and Schall (1975) claim that bondholders of the acquirer get the gain, because they can use covenants to renegotiate the debt in such a way that the value of the debt increases by reducing the default risk.
returns (Morck et al., 1990; Hornstein and Nguyen, 2014).
Kang et al. (2000) find in Japan that the higher the leverage, and especially the leverage from the main bank the higher the abnormal returns to acquirers. This implies that banks are the informed creditors that monitor the acquirer, which leads to less
value-destroying acquisitions (Kang et al, 2000). Gopalan et al. (2011) find that bankruptcies and default of the borrowers leads to reputation damage for the lender, since bankruptcy and defaults signals to the market that the lender does not adequately monitor his borrowers.
3. Methodology
3.1 Regressions and Hypothesis
To see what the effect of default risk, measured by the CDS spread of the acquirer, is on the CAR of the acquirer, the following regression will be performed:
𝐶𝐴𝑅 ! = 𝛼 + 𝛽!𝐿𝑜𝑔𝐶𝐷𝑆 ! + 𝛽!𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒 ! + 𝛽!𝑇𝑜𝑏𝑖𝑛 𝑞 ! + 𝛽!𝑃𝑢𝑏𝑙𝑖𝑐𝑇𝑎𝑟𝑔𝑒𝑡𝐷𝑢𝑚𝑚𝑦! + 𝛽!𝐿𝑜𝑔𝑅𝑒𝑙𝑆𝑖𝑧𝑒! + 𝛽!𝐶𝑎𝑠ℎ𝐷𝑢𝑚𝑚𝑦! + 𝛽!𝑀𝑖𝑥𝑒𝑑𝑃𝑎𝑦𝑚𝑒𝑛𝑡𝐷𝑢𝑚𝑚𝑦! + 𝛽! ∗ 𝐶𝑎𝑠ℎ𝐷𝑢𝑚𝑚𝑦! ∗ 𝐿𝑜𝑔𝑅𝑒𝑙𝑆𝑖𝑧𝑒!+ 𝛽!∗ 𝑀𝑖𝑥𝑒𝑑𝑃𝑎𝑦𝑚𝑒𝑛𝑡𝐷𝑢𝑚𝑚𝑦!
∗ 𝐿𝑜𝑔𝑅𝑒𝑙𝑆𝑖𝑧𝑒! + +𝛽!"!!"𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝐹𝑖𝑥𝑒𝑑𝐸𝑓𝑓𝑒𝑐𝑡𝑠! + 𝛽!"!!"𝑌𝑒𝑎𝑟𝑙𝑦𝐹𝑖𝑥𝑒𝑑𝐸𝑓𝑓𝑒𝑐𝑡𝑠! + 𝜀!
With the results of this regression the following hypotheses will be tested: 𝐻!: 𝛽! = 0
𝐻!: 𝛽! ≠ 0
If the null hypothesis is rejected then it can be concluded that the default risk of the acquirer, measured by the CDS spread of the acquirer, has an effect on the CAR of the acquirer.
3.2 Description of the variables
The variable 𝐶𝐴𝑅!"#$%&'& ! is the dependent variable, and is the cumulative abnormal returns of acquirer i. The CAPM will be used to calculate the CAR of the acquirer; 𝐸 𝑟! = 𝑟!+ 𝑎 + 𝛽 ∗ 𝑟!− 𝑟! , following Malesta (1983) and Halpern (1983). The following model will then be regressed using stock prices from 240 trading days until six trading days before the
announcement of an offer; (𝑟! − 𝑟!) = 𝑎 + 𝛽 ∗ 𝑟!− 𝑟! + 𝜀!. Following Malesta (1983), the estimated parameters are then used to calculate the expected returns, using again the CAPM: 𝐸 𝑟! = 𝑟!+ 𝑎 + 𝛽 ∗ 𝑟!− 𝑟! .
Subsequently, the abnormal daily returns will be calculated by subtracting the expected returns from the realized returns. These abnormal returns are calculated and cumulated for a flexible event window starting five trading days before the announcement until the completion date of the M&A deal. The choice was made to use this flexible event window because of multiple reasons. So do Andrade, Mitchell and Stafford (2001) note that it is common to use an event window that starts several days before an announcement and ends at the closing of the M&A deal. According to Andrade et al. (2001) investors systematically fail to quickly assess the impact of corporate announcements, making inferences flawed that are based on event windows centered around the announcement dates. Secondly, Bhagat, Ding and Hirshleifer (2005) argue that using an event window that includes the successful completion of the transaction captures the market’s assessment of the full effect of the M&A deal, in contrast to using a shorter event window. In line with this Bradley, Desai and Kim (1983) notice that around the before the completion of the M&A deal there might still be uncertainty about whether the merger or acquisition will actually go through, and this might influence the CAR. This uncertainty might be related to the default risk of the acquirer, which means that using an event window that doesn’t include the completion date leads to biased results. So do Parlour and Winton (2013) argue that the default risk is a determinant of the amount of monitoring a firm receives, while Chen, Harford and Li (2007) argue that
withdrawal of bad bids is more likely to happen when the acquirer is more monitored. If the market recognizes that the uncertainty about whether an M&A deal will be completed is higher for higher default risk firms, then the actual effect of the default risk on the abnormal returns might become only clear using a flexible event window.
However in contrast to for example Ang and Cheng (2006) and Higson and Elliot (1998) who use a flexible event window starting one trading day before the announcement and ending at the completion date, the abnormal returns are calculated starting trading five days before the announcement to take into account rumors like Bradley et al. (1983) did. This makes the event period the same as the one Healy, Palepu and Ruback (1992) use; starting five trading days before the announcement and ending at the completion date.
The main explanatory variable 𝐿𝑜𝑔𝐶𝐷𝑆!"#$%&'& ! is the natural log of the CDS spread of acquirer i six trading days before the announcement. The CDS spread will be used as a proxy for the default risk, following Longstaff et al. (2005). The choice to take the CDS
spread six trading days before the announcement was made to avoid a possible effect of the rumor of the acquisition on the CDS spread. In the appendix it can be seen that the log distribution of the CDS spread of the acquirer seems to give a better approximation to the normal distribution than the distribution of the CDS spread of the acquirer itself, because the outliers have less of an impact.
Both a positive and a negative effect of the default risk on the abnormal returns may be expected. A negative effect of the default risk on the abnormal returns may be expected, because firms that are better managed have higher abnormal returns (Lang, Stulz and
Walking, 1989) and may have a lower default risk. A positive effect of the default risk on the abnormal returns may be expected, because firms with a higher default risk should be better monitored (Parlour & Winton, 2013; Kang et al., 2000).
The first control variable 𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒!"#$%&'& ! is the leverage of acquirer i 30 days before the announcement22. Kang et al. (2000) define leverage as total book value of
debt/(total book value of debt + total market value of equity). It is expected to have a positive effect on the CAR of the acquirer because a higher leverage leads to more monitoring (Kang et al., 2000). The variable 𝑇𝑜𝑏𝑖𝑛 𝑞!"#$%&'& ! is the Tobin’s q of acquirer i. Tobin’s q is the market value of assets divided by the replacement value of these assets and is approximated by (market capitalization + value of liabilities)/ (book value of equity + value of liabilities) (Chung & Pruitt, 1994). Servaes (1991) found that Tobin’s q has a positive effect on the CAR of the acquirer. 𝑃𝑢𝑏𝑙𝑖𝑐𝑇𝑎𝑟𝑔𝑒𝑡𝐷𝑢𝑚𝑚𝑦! is a dummy variable that equals one if the target is publicly listed, and zero if it is not. It is expected to have a negative effect on the CAR of the acquirer, since the abnormal returns for public targets are lower than for private target (Faccio, McConnel & Stolin, 2006). The reasoning behind this is that the shares of private firms are less liquid and therefore the acquirer has to pay a lower premium to the equity holders of private targets, because these equity holders cannot sell their shares on an exchange (Fuller et al., 2002). The control variable 𝐿𝑜𝑔𝑅𝑒𝑙𝑆𝑖𝑧𝑒! measures the natural logarithm23 of the
final transaction value divided by the equity value of the acquirer 30 days before the first announcement of an M&A transaction 24. The relative size of the acquirer is expected to have
22 Following Bouwman et al.(2009) all the information on control variables that are related to the debt and equity of a firm are taken 30 days before the announcement of an acquisition.
23 The log distribution of the relative size provides a better approximation of the normal distribution than the relative size distribution (appendix figure II and III). Furthermore, Asquith, Bruner, and Mullins (1982) and Fuller et al. (2002) also use a log distribution.
24 The variables final transaction value and equity value of the acquirer are not included separately for two reasons. First there is a risk of multicollinearity because of the -0.5955 correlation between the log of the equity of the acquirer and the log of the CDS spread of the acquirer. Secondly, Bouwman et al. (2009) and Fuller et al. (2002) also use this log of the relative value, instead of two separate logs wherein there is a separate log for the equity of the acquirer and a separate log for the transaction value.
a positive effect on the CAR (Asquith et al., 1982). If the target is relatively big compared to the acquirer then the M&A deal is more important and should therefore be better monitored so that higher abnormal returns are achieved (Fuller et al., 2002). Furthermore, there are cash and mixed payment dummies included. The cash dummy equals one if the acquisition is fully paid in cash or related methods, for example non-convertible debt and nonconvertible
preferred stock, and zero otherwise. The mixed payment dummy equals one if the acquisition is paid in a mixture of cash related methods and stock related methods, and zero otherwise. Stock related payment methods include; convertible debt, stock options and shares (Bouwman et al., 2009). Travlos (1987) found that acquisitions that are fully paid in cash have a positive effect on the CAR compared to acquisitions that are paid in stock, because paying by stock may signal to the market that the share price is overvalued (Myers & Majluf, 1984).
Furthermore, like Bouwman et al. (2009) and Fuller et al. (2002) the interaction variable of the payment method and relative size are included, because Fuller et al. find that these interaction effects have a higher positive effect on the abnormal returns for cash offers when the relative size increases. The underlying reason is that for acquisitions paid with stocks there is a higher negative effect when relative size increases, because the release of adverse information might be bigger (Fuller et al., 2002). For mixed payments there is little effect on the abnormal returns (Fuller et al., 2002). Finally following Bouwman et al. (2009) yearly fixed effects and 17 Fama-French industry fixed effects are included. The industry fixed effects are calculated using the SIC codes from a company and matching it to the 17 industry classifications from French (2014). The Fama-French industry classifications were designed to address the problems of the SIC codes by creating industry classifications based on common risk characteristics, in contrast to the definitions of the SIC codes which were last updated in 1987 (Bhojrai, Lee, & Oler, 2003).
4. Data description
The focus of this research is on domestic M&A deals that were announced after 15 September 2008 and took place in the United States of America. The choice for domestic M&A deals is made, because for example Cakici, Hessel and Tandon (1996) found that when the acquirer is a foreign company then this positively influences the CAR of this acquirer25. A sample is taken of deals that were announced between 15 September 2008, when Lehman Brothers went
25 However Cakici et al. (1996) do not find evidence for possible explanations of this phenomenon, like tax differences.
bankrupt, and 1 September 2014. The choice was made to start the data sample after the start of the crisis, since Galil, Shapir, Amiram and Ben-Zion (2014) found that after the start of the crisis CDS spreads are structurally differently priced, by studying CDS spreads for 764 US firms between January 2002 and February 2013. If CDS spreads from before the start of the crisis are included in the data sample then these CDS spreads might have a different effect on the CAR, which might lead to biases. The last constraint comes from Bouwman et al. (2009) who state that the acquirer must be listed on one of the following exchanges: the AMEX, NYSE or the NASDAQ national market. 7996 deals fulfill the criteria and were extracted from Zephyr. These deals are furthermore selected on whether data is available for the CDS spreads of the acquirer and all the control variables. The CDS is taken as the five-year credit default swaps on senior unsecured debt, following Corò, Dufour and Varotto (2013). The choice was made to use XR as the type of CDS, since XR CDSs exclude restructuring events as credit events26. The data on the CDS spreads and the market and book value of debt and equity are taken from Datastream. The data on the payment method, transaction value, SIC codes of acquirers and targets, dates of the deals and whether the target is a public target are taken from Zephyr. The deals for which not all the necessary information can be found are excluded, which results in 281 observations. Furthermore, to calculate the CAR, stock returns and market returns are taken from Datastream, while the risk-free rate27 and the industry
classifications are taken from the website of French (2014). Following Harris, Marston, Mishra and O’Brien (2003), the S&P 500 is used as the market portfolio. After all the constraints are fulfilled 281 deals are left, with the following characteristics:
26 XR means no restructuring (Wei, 2012). The no structuring (XR) version of Credit Default Swaps only
include as credit events; default, bankruptcy and (temporarily) refusal to pay (Amato & Gyntelberg, 2005). Restructuring events are excluded according to Berndt, Jarro and Kang (2007) and include postponements of dates, change in priority of payment ranking, currency and a reduction in the interest rate or principal. These events are not defaults and it is therefore useful that they are excluded.
Table 1
Characteristics of deals, acquirers and targets
Data sample (n=281) Mean Median Std. Dev. Min Max
CAR (in %) −1.8608 −0.7974 16.6085 −66.4864 103.0497
CDS acquirer (in basis points) 149.2377 99.98 144.853 22.7200 815.1599
Leverage acquirer 0.3079 0.2534 0.2080 0 0.9171
Tobin’s Q acquirer 1.6171 1.5238 0.5672 0.7532 3.577
Relative size acquirer 0.2045 0.0373 0.7244 0.0002 11.1602
The mean CAR is not significantly different from zero, which Li (2013) also found to be the case when he studied 1430 M&A deals that took place in the US between 1981 and 2002. The minimum CAR is from the Wells Fargo acquisition of Wachovia securities and may be the result of the long-event window, since the acquisition was completed on 31 December 2009 while it was announced on 17 June 2009. The maximum CAR occured when Best Buy Company bought Best Buy Mobile. This may also be the result of the long-term event window in the dataset because the acquisition was announced on 7 November 2011 but was completed on 31 March 2014. The minimum CDS spread is from Pfizer, a pharmaceutical company and the maximum CDS spread is from Transdigm Group, an aerospace components manufacturer.
Figure 1 plots the CAR of the acquirer against the default risk, and shows a positive relationship between default risk of the acquirer and the CAR of the acquirer.
Figure 1
Scatterplot of the CAR of the acquirer with respect to the log of his CDS spread
5. Results
5.1 Description and interpretation of the results
To estimate the effect of the default risk of the acquirer on the CAR of the acquirer OLS was used. The results of the regressions are summarized in table 2:
Table 2
The effect of the CDS spread of acquirers on the CAR of acquirers (Standard errors in parentheses)
CAR
(1) CAR (2) CAR (3) CAR (4) CAR (5) CAR (6) Log (CDS spread acquirer) 1.096
(1.20) 2.220* (1.25) 3.157* (1.64) 3.084* (1.64) 2.981* (1.66) 2.926* (1.66) Leverage acquirer −6.873 (6.73) −10.312 (7.37) −8.422 (6.84) −11.393 (7.44) Tobin’s q acquirer −2.501 (2.18) −2.207 (2.18) Public target −0.515 (1.97) −0.629 (1.97) −1.757 (2.15) −1.809 (2.15)
Log (relative size) 0.043
(1.81)
0.021 (1.81)
Cash offer 1.522
(3.29) 1.300 (3.30) 3.523 (6.66) 3.297 (6.66)
Mixed Payment offer 4.248
(3.82) 4.012 (3.83) 9.280 (6.70) 8.993 (6.71)
Cash*Log (relative size) 0.597
(1.88)
0.591 (1.88) Mixed payment* Log
(relative size) 2.897 (2.21) 2.842 (2.21) Constant −6.949 (5.64) −12.681* (7.59) −16.096* (8.41) −10.408 (9.76) −13.405 (10.35) −8.560 (11.40)
Year Fixed Effects No Yes Yes Yes Yes Yes
Fama-French industry fixed effects
No Yes Yes Yes Yes Yes
N 281 281 281 281 281 281
R2 0.0030 0.2297 0.2368 0.2408 0.2517 0.2548
Adjusted R2 −0.0006 0.1673 0.1620 0.1631 0.1686 0.1687
*, **, *** Denotes significance at the 10, 5, and 1 percent level.
In all of the six models it can be seen that the default risk, measured by the log of the CDS spread, has a positive effect on the CAR of the acquirer. This in accordance with the argument that higher default risk leads to more monitoring (Parlour & Winton, 2013), while monitoring leads to better acquisitions with higher abnormal returns (Kang et al., 2000). Secondly, this positive effect is in accordance with the free cash flow theory, in which firms that have less free cash, do better acquisitions (Jensen, 1986), while they might also have a lower default risk because of the higher cash levels. This positive effect is significant at the 10% level, over all models except model one. Model six has the highest adjusted 𝑅!, in this model the
coefficient is 2.926, which means that if the log of the CDS spread increases by one standard deviation (0.8302), then the CAR of the acquirer increases by approximately 2.43%. It can be concluded with 10% certainty that the default risk of the acquirer has a significant effect on the CAR of the acquirer in M&A deals. This effect of the default risk of the acquirer on the CAR of the acquirer and its significance is bigger than control variables like Tobin’s q and the public dummy variable and is 14.63% of one standard deviation of the CAR.
The coefficients on both leverage and Tobin’s q in the models are insignificant and negative. However, on the basis of research from Kang et al. (2000) a positive relationship of leverage to the CAR of the acquirer is expected. Also, Servaes (1991) found that the Tobin’s q of the acquirer has a positive effect on the abnormal returns. However, in other research that use Tobin’s q and leverage as control variables and the CAR of the acquirer as independent variable an insignificant negative effect of those variables on the CAR of the acquirer in US M&A deals is found28.
The other control variables are also all insignificant. The public target dummy is negative over all the regressions, in line with expectations (Faccio et al., 2006). The log of the relative size has a positive effect in models five and six, in line with findings by Fuller et al. (2002). The payment variables show that cash offers have higher cumulative abnormal returns than stock offers; this effect becomes bigger when the relative size increases, in line with findings by Fuller et al. (2002).
A third explanation for the positive effect of default risk of the acquirer on the CAR of the acquirer is that firms acquire other firms to diversify and thereby increase their tax shield as argued by Brealey et al. (2006). To test this the variables diversification and the interaction variable of diversification with the log of the CDS spread of the acquirer are added to the model. The dummy variable diversification equals one if the industry classification29 of the acquirer, is not equal to the industry classification of the target. Using the same methodology as before an OLS regression leads to the following results, which are summarized in table 3.
28 Golubov, Petmezas and Travlos (2012) use leverage of the acquirer as a control variable, where they find a negative insignificant effect of leverage on the CAR. While Moeller, Schlingemann and Stulz (2005) and Masulis, Wang and Xie (2007) use Tobin’s q of the acquirer as a control variable and they also found a negative insignificant effect of Tobin’s q on the CAR.
Table 3
The effect of the CDS spread of acquirers on the CAR of acquirers (Standard errors in parentheses)
CAR (7) CAR (8) CAR (9) CAR (10) CAR (11) CAR (12) Log (CDS spread acquirer) 2.225*
(1.25) −0.738 (4.30) 2.707* (1.63) 0.219 (4.50) 2.908* (1.65) 1.275 (4.55) Diversification −5.484* (3.24) −20.530 (21.14) −6.499* (3.51) −19.037 (21.42) −6.259* (3.52) −14.504 (21.70) Diversification*Log (CDS spread acquirer) 3.184 (4.42) 2.655 (4.47) 1.748 (4.54) Leverage −11.687 (7.42) −11.293 (7.46) −11.446 (7.41) −11.199 (7.45) Tobin’s q −2.282 (2.17) −2.143 (2.19) −2.139 (2.17) −2.054 (2.18) Public −2.873 (2.25) −2.820 (2.26) −2.954 (2.25) −2.922 (2.26)
Log (relative size) 1.157
(1.07) 1.140 (1.8) −0.197 (1.81) −0.111 (1.83) Cash offer 3.853 (6.65) 3.451 (6.74)
Mixed Payment Offer 9.240
(6.68)
8.789 (6.79)
Cash*Log (relative size) −0.313
(0.93) −0.300 (0.93) 3.853 (6.65) 0.624 (1.90) Mixed payment* Log (relative
size) 0.586 (1.34) 0.657 (1.35) 9.240 (6.68) 3.024 (2.23) Constant −8.093 (8.07) 6.306 (21.56) 3.629 (8.41) 15.311 (22.46) −3.140 (11.86) 4.883 (23.98)
Year Fixed Effects Yes Yes Yes Yes Yes Yes
Fama-French industry fixed effects Yes Yes Yes Yes Yes Yes
N 281 281 281 281 281 281
R2 0.2382 0.2397 0.2559 0.2569 0.2639 0.2644
Adjusted R2 0.1732 0.1717 0.1732 0.1711 0.1756 0.1728
*, **, *** Denotes significance at the 10, 5, and 1 percent level.
The Diversification dummy has a negative effect, which is in line with research from Morck et al. (1990). However, when the interaction effect of the log of the CDS spread of the acquirer with the diversification effect is added the significance of this effect disappears and the adjusted 𝑅! falls. The interaction effect is positive indicating that the higher the default
risk the more acquirers benefit from diversifying, but it doesn’t result in an significant effect on the CAR. So the conclusion can be made that the positive effect of the default risk on the abnormal returns of the acquirer cannot be explained by the diversification effect.
5.2 Robustness and alternative tests
5.2.1 Target default risk
Lang et al. (1989) argue that the abnormal returns of the acquirer are highest when the
acquirer is well managed, while the target is poorly managed. Again using that poor-managed firms may have a higher default risk, additional regression analyses are done in which the log of the CDS spread of the target is included. From the 281 deals from before there are 30 deals for which the default risk of the target is also available. This is the case because the target companies often were holding companies for which there were no CDS spreads available. It is then not clear whether the CDS spread of the underlying company reflects the default risk of the holding company, because the holding companies may have different ownership or legal organizational structure. The mean CAR is -1.55%, while the mean CDS spread of the acquirer is 165.28 basis points and the mean CDS spread of the target is higher at 476.14 basis points30.
This data is regressed with OLS and the results are summarized in Table 4.
Table 4
The effect of the CDS spread of acquirers and their targets on the CAR of acquirers (Standard errors in parentheses)
CAR
(13) CAR (14) CAR (15) CAR (16)
Log (CDS spread acquirer) −2.394
(5.25) −0.489 (4.61) −8.145 (6.79) −43.979 (67.75)
Log (CDS spread target) −0.042
(3.66) −1.767 (2.85) −0.105 (3.12) 9.437 (12.61) Leverage 41.333 (29.92) 177.391 (280.87) Tobin’s q 1.414 (10.81) -10.045 (54.76) Public 23.434 (18.51)
Log (relative size) 44.121
(95.53)
Cash offer −66.959
(182.99)
Mixed Payment offer −91.508
(221.03)
Cash*Log (relative size) −43.776
(110.82)
Mixed payment* Log (relative size) −28.134
(77.59) Constant 10.230 (23.00) 26.905 (23.62) 31.525 (27.78) 171.192 (409.68)
Year Fixed Effects No Yes Yes Yes
Fama-French industry fixed effects No Yes Yes Yes
N 30 30 30 30
R2 0.0104 0.7938 0.8362 0.8967
Adjusted R2 −0.0629 0.5401 0.5681 0.4011
*, **, *** Denotes significance at the 10, 5, and 1 percent level.
None of the variables above have significant coefficients, which might be explained by the small sample. Furthermore, in contrast to the results of the models as described in section 5.1, the coefficients on the log of the CDS spread of the acquirer are negative in all of the models. This might be because of the small sample and not because of the addition of the log of the CDS spread of the target, since in the models with the dataset of 30 observations where the default risk of the target is not included the coefficient on the log of the CDS spread
of the acquirer is still negative31. The coefficient on the log of the CDS spread of the target is negative in all models except in model 16 where it has a positive effect.
5.2.2 10-day CAR
To test the robustness of using the flexible event window to calculate the CARs a fixed event window is used that starts five trading days before the announcement and ends five trading days after the announcement of an M&A transaction. The results of these regressions are summarized in table 5.
Table 5
The effect of the CDS spread of acquirers on the 10-day CAR of acquirers (Standard errors in parentheses)
CAR (17) CAR (18) CAR (19) CAR (20) CAR (21) Log (CDS spread acquirer) 0.687
(0.46) 0.736 (0.45) 0.910 (0.58) 0.802 (0.61) 0.802 (0.61) Leverage acquirer −1.336 (2.51) −0.836 (2.50) −0.980 (2.60) Tobin’s q acquirer −0.170 (0.72) −0.156 (0.75) Public target −0.080 (0.79) −0.080 (0.79)
Log (relative size) −0.293
(0.66) −0.299 (0.66) Cash offer 2.291 (2.43) 2.271 (2.44)
Mixed Payment offer 1.768
(2.45)
1.741 (2.45)
Cash*Log (relative size) 0.315
(0.69)
0.314 (0.69) Mixed payment* Log
(relative size) 0.276 (0.81) 0.264 (0.81) Constant −3.524 (2.19) −3.382 (2.74) −3.392 (3.16) −5.116 (3.78) −4.832 (4.02)
Year Fixed Effects No Yes Yes Yes Yes
Fama-French industry fixed
effects No Yes Yes Yes Yes
N 281 281 281 281 281
R2 0.0078 0.3337 0.3344 0.3382 0.3384
Adjusted R2 0.0043 0.2797 0.2749 0.2648 0.2620
*, **, *** Denotes significance at the 10, 5, and 1 percent level.
In table 5 it can be seen that the sign of the coefficient on the log of the CDS spread of the acquirer remains positive. However the significance of this effect disappears, even though it stays close to a 10% significance. This might be because as Bradley et al. (1983) argue there might still be uncertainty about the M&A deal five days after the announcement. If this uncertainty is related to the default risk then the full significance of default risk may not appear unless a flexible event window is used, leading to insignificant results using a 10-day event window.
5.2.3 Other event windows
If the uncertainty about the effect of the M&A deal disappears the closer the completion date of the M&A deal gets, then it may be the case that longer fixed event windows give a more significant effect of the CDS spread of the acquirer on his CAR then the 10-day event window. Therefore in model 22 till 25 longer fixed event windows will be used. In model 26 only the M&A deals will be included that were completed within 30 days, which leaves 52 observations. Finally, in model 27 the CAR of acquirer i is divided by the number of days from the announcement till the completion date. This gives an average daily abnormal return (AAR) for each acquirer i. OLS regressions are used for these models and the results are summarized in Table 6.
Table 6
The effect of the CDS spread of acquirers on the CAR of acquirers (Standard errors in parentheses)
CAR (-5,10) (22) CAR (-5,20) (23) CAR (-5,30) (24) CAR (-5,40) (25) CAR (-5,close) (26) AAR (-5, close) (27) Log (CDS spread acquirer) 0.507
(0.68) 0.259 (0.852) 0.622 (0.98) 0.588 (1.089) 4.854 (3.03) 1.089** (0.440) Leverage acquirer −2.887 (2.90) −4.051 (3.65) −6.526 (4.22) −8.313* (4.67) −15.770 (14.17) −1.791 (1.89) Tobin’s q acquirer −0.654 (0.83) −0.414 (1.05) −1.182 (1.21) −1.265 (1.34) 1.394 (3.41) −0.520 (0.542) Public target −0.665 (0.88) −1.835* (1.11) −1.821 (1.28) −2.157 (1.42) 0.298 (4.77) 0.528 (0.57) Log (relative size) −0.175
(0.74) −0.057 (0.93) −0.416 (1.08) −0.465 (1.19) -0.484 (3.89) -0.170 (1.81) Cash offer 0.898 (2.72) 1.952 (3.43) 3.054 (3.96) 2.715 (4.38) -9.070 (15.83) -2.101 (1.77) Mixed Payment offer 1.490
(2.74) 1.129 (3.45) 2.918 (3.99) 2.604 (4.41) -0.275 (15.67) 0.026 (1.78) Cash*Log (relative size) 0.226
(0.770) 0.555 (0.97) 1.084 (1.12) 1.287 (1.24) -2.082 (4.04) -0.322 (0.50) Mixed payment* Log
(relative size) 0.195 (0.91) -0.190 (1.15) 0.397 (1.32) 0.522 (1.46) 3.971 (5.14) 0.443 (0.59) Constant −1.492 (4.49) 5.541 (5.66) 5.502 (6.53) 6.864 (7.23) −20.600 (24.40) −4.58 (2.92)
Year Fixed Effects Yes Yes Yes Yes Yes Yes
Fama-French industry fixed effects
Yes Yes Yes Yes Yes Yes
N 281 281 281 281 52 281
R2 0.4349 0.5036 0.5638 0.5978 0.3712 0.0802
Adjusted R2 0.3696 0.4462 0.5135 0.5514 −0.1058 −0.0261
*, **, *** Denotes significance at the 10, 5, and 1 percent level.
The effect of the CDS spread of the acquirer on the CAR of the acquirer is positive in all the models above, but is only significant in model 27. This means that the effect of the CDS spread of the acquirer on his CAR does not become significant if a longer event window is used compared to the 10-day event window. In model 26 the M&A deals that took longer than 30 days to be completed are excluded and the effect of default risk on the abnormal returns is not significant. However this is probably because of the smaller sample since the effect of the CDS spread of the acquirer on his CAR is close to significance. Finally, in model 27 the
effect of the CDS spread of the acquirer on the CAR of the acquirer has the highest
significance from all of the 27 models, with an effect that is significant at the 5% level. This indicates that the flexible event window gives more significant results if the CARs are
adjusted for the length of the event window. However the adjusted 𝑅! in model 27 is negative in contrast to models where the CAR is not averaged over the number of days it took the merger to complete.
6. Conclusion
This thesis analyzes whether the default risk of the acquirer an effect has on his CAR of M&A deals after the start of the crisis. The answer to this question is affirmative with 10% certainty. A positive relationship between the default risk of the acquirer and the CAR of the acquirer is found by doing OLS regressions on the data sample where the CDS spread was used as a proxy for the default risk. The CAPM was used to calculate the abnormal returns, these abnormal returns were cumulated for a flexible event window starting five days before the announcement of an acquisition and ending at the closing date of the acquisition. This was done because in a fixed event window the full effect of default risk on the CAR might not have been realized, because the uncertainty about whether the M&A deal will be completed might be related to the default risk of the acquirer. So using a fixed event window might have introduced biases. The data sample consisted of domestic US M&A deals that were
announced and completed between 15 September 2008 and 1 September 2014.
From the theoretical review there are three possible explanations that support this positive relationship. In the monitoring argument it is argued that firms with higher default risk incur more monitoring and therefore do better M&A deals. In the free-cash flow theory argument firms that have more free cash, and might therefore have lower default risk, do more wasteful acquisitions since the probability of default is lower. In the third argument it is argued that firms with a higher default risk have more to gain from diversifying acquisitions, since they can reduce their default risk by combining cash flows that are imperfectly
correlated. In this research there was no significant evidence for this diversification argument found.
A negative effect of the default risk on the CAR could have been explained by the argument that well-managed acquirers might have a lower default risk, and do better acquisitions.
To test the predicted relationships OLS regressions were done on a dataset. In a complete model the coefficient on the log of the CDS spread of the acquirer was 2.926. This means that if the log of the CDS spread increases by one standard deviation (0.8302), then the CAR increases by approximately 2.43%. This relationship is significant at the 10% level. Next robustness test were done by adding the CDS spread of the target and testing different event windows. Adding the CDS spread of the target reduced the sample to 30 observations. This was also the explanation for the insignificant and negative relationship between default risk of the acquirer and the CAR of the acquirer that followed. Subsequently different relationship between default risk of the acquirer and fixed event windows CARs
were tested. The effect of default risk remained positive in all of the fixed event windows, but was not significant. When only M&A deals that were completed within 30 days were
analyzed in the flexible event window the effect was positive but not significant. Finally, dividing the CAR of each acquirer by the number of days that the M&A deal took to
complete, the effect of default risk became significant at the 5% level. This provided evidence that adjusting the flexible event window for its length results in a higher significance of the effect of default risk on the CAR of the acquirer.
Reference list
Amato, J., & Gyntelberg, J. (2005). CDS index tranches and the pricing of credit risk correlations. BIS Quarterly Review, (March), 73-87.
Andrade, G., Mitchell, M., & Stafford, E. (2001). New evidence and perspectives on mergers. Journal of Economic Perspectives, 15(2), 103-120.
Ang, J. S., & Cheng, Y. (2006). Direct evidence on the market‐ driven acquisition theory. Journal of Financial Research, 29(2), 199-216.
Asquith, P., Bruner, R. F., & Mullins, D. W. J. (1983). The gains to bidding firms from merger. Journal of Financial Economics, 11, 121-139.
Banal-Estañol, A., Ottaviani, M., & Winton, A. (2013). The flip side of financial synergies: Coinsurance versus risk contamination. The Review of Financial Studies, 26(12), 3142-3181.
Berk, J., & DeMarzo, P. (Eds.). (2011). Corporate finance (Global edition of 2nd revised edition ed.). Edinburgh Gate: Pearson Education Limited.
Berndt, A., Jarrow, R. A., & Kang, C. (2007). Restructuring risk in credit default swaps: An empirical analysis. Stochastic Processes and their Applications, 117(11), 1724-1749. Bhagat, S., Dong, M., Hirshleifer, D., & Noah, R. (2005). Do tender offers create value? new
methods and evidence. Journal of Financial Economics, 76(1), 3-60.
Bhojraj, S., Lee, C., & Oler, D. (2003). What's my line? A comparison of industry classification schemes for capital market research. Journal of Accounting Research; J.Account.Res., 41(5), 745-774.
Bolton, P., & Oehmke, M. (2011). Credit default swaps and the empty creditor problem. Review of Financial Studies, 24(8), 2617-2655.
Bouwman, C., Fuller, K., & Nain, A. (2009). Market valuation and acquisition quality: Empirical evidence. Review of Financial Studies, 22(2), 633-679.
Bradley, M., Desai, A., & Kim, E. H. (1983). The rationale behind interfirm tender offers: Information or synergy? Journal of Financial Economics, 11(1), 183-206.
Brealey, R., Myers, S., & Allen, F. (2006). Corporate finance / richard A. brealey, (8th ed ed.). New York, NY etc.: McGraw-Hill/Irwin.
Cakici, N., Hessel, C., & Tandon, K. (1996). Foreign acquisitions in the united states: Effect on shareholder wealth of foreign acquiring firms. Journal of Banking & Finance, 20(2), 307-329.
Chen, X., Harford, J., & Li, K. (2007). Monitoring: Which institutions matter? Journal of Financial Economics, 86(2), 279-305.
