Does bankruptcy law sweeten the taste of distressed debt? : coordination problems and the presence of vulture investors

(1)

1

Does bankruptcy law sweeten the taste of distressed

debt?

Coordination problems and the presence of vulture investors

August 11, 2014

Eva de Beet (6123775) Supervisor: Dr. Jeroen Ligterink

Master Thesis

Business Economics, specialization Finance

Faculty of Economics and Business University of Amsterdam

(2)

2

In ancient Greek civilization, there were no bankruptcy laws. A man and his family were forced into physical labour to repay a creditor, when he was not able to repay his debt. In the absence of bankruptcy laws, very little changed for centuries and defaulting debtors failing to repay their debts were considered criminals and sent to prison. When the first US bankruptcy laws were introduced in the 19th century, they were focused on recovery of the creditor’s investment. Over the following decades, development of the bankruptcy law changed the relative bargaining positions of creditors and debtors. Chapter 11, an important code in the US bankruptcy law, enables the reorganization of debts and assets, however conflicts of interests can hinder negotiation. Doubts concerning asset valuation can add to this conflict. The increased importance of intangible assets such as innovation, intellectual property and human capital can add to this uncertainty and therefore raises questions as to the relevance of current US bankruptcy practice. Distressed debt investors, also known as vulture funds, appear precisely at this complex situation. Previous research has shown evidence that involvement of vulture investors promotes negotiation to a satisfactory agreement and settlement. This is found in higher debt recovery and positive valuation effects after involvement of distressed debt investors (Hotchkiss and Mooradian, 1997 & Jiang et al., 2012). Evidence of balance in the coordination problem, contributed by vulture investors, raises the question whether or not their appearance is a market reaction to supplement the existing bankruptcy law, which due to the increased importance of intangible assets does not seem to be as appropriate as it once was. This research attempts to answer this question by investigating whether the presence of large coordination problems in chapter 11 has a positive effect on the probability of vulture involvement. If vulture investors indeed target firms with large debt resolution issues, then would this be an indication that their involvement is a market evolution caused by a dated bankruptcy law?

To answer the question this research primarily investigates the magnitude of the

coordination problem and the effect on the probability of vulture involvement. Further it investigates whether the stock market’s reaction on vulture involvement differs when a firm has large debt resolutions.

While most literature focuses on the effects of vulture funds on bankruptcy outcomes, post reorganization performance and announcement returns, this research extends these studies by deepening in the environment that enables vulture to become involved (Hotchkiss and Mooradian, 1997, Jiang et al. 2012).

Despite evidence of vulture investors’ contribution to the negotiation process in achieving more economically efficient outcomes (Jiang et al., 2012), this research does not find that vultures are likely to step in when the coordination problem is large. Only the size of

(4)

4

the firm is found to attract vulture involvement. So vulture investors’ appearance is not a consequence of inappropriate bankruptcy law. Further this research finds no evidence that the stock market appreciates vulture involvement more, when it is with firms plagued by large coordination problems.

This paper continues with section 2 which describes the literature regarding the bankruptcy process, how coordination problems arise during financial distress and vulture fund activity. The section ends with the hypothesis development. The third section describes the sample selection and research method, it finalized with the results of this study. The fourth section contains the conclusions. At the end of this paper, the appendix and references can be found.

2. LITERATURE AND HYPOTHESIS DEVELOPMENT

2.1 Coordination problems in financial distress

2.1.1 Bankruptcy

When a company is financing its investments with debt, it will always have to take into account the possibility of bankruptcy. This occurs when a corporation is no longer able to pay interest on its debts. Huge conflicts of interest exist during bankruptcy between shareholders and creditors. When a default dooms for a company, creditors want their investment

returned as quickly as possible. Shareholders in contrast with creditors, have huge interest in continuing existence of the firm, to prevent them from ending up empty handed.

Whenever a firm defaults on its debt, the amount of debt exceeds the amount of collateral. In the absence of bankruptcy law, this could lead to a run from creditors on the distressed firm to be the first to seize available collateral (Smith & Strömberg, 2003).If upon default secured creditors start procedures to recover their secured assets, unsecured creditors and equity holders having claims on other assets suffer significant losses (Giammarino, 1986). The withdrawal of fundamental production factors goes at the expense of the total value of the firm, hence the total value of the creditors’ claims. Creditors capturing the freezers of an ice cream factory, for example, cause immediate discontinuance of operations and a huge drop in valuation of the firm. This example illustrates how snatching behavior by creditors can decrease the total value of the claims of the claimholders. Inefficiencies in operations that are the result of such a run on assets are called the common pool problem. Jackson (1986) argues that the purpose of bankruptcy law is to minimize the negative effects of this common pool problem. Bankruptcy law prevents runs on assets and mitigates the common pool problem by imposing an automatic stay on all creditor actions. This stay means that the assets continue to operate and can’t be seized by the creditors, until a reorganization

(5)

5

plan is formulated and ultimately confirmed by the courts. Bankruptcy law stimulates a collective procedure over individual procedures and seeks for the most efficient economic solution to maximize the value for the entire group of creditors.

This section described the concept of bankruptcy, the conflicts of interests that come with it and why laws and rules are necessary to regulate this process. Hereafter this paper will continue with the development of the US bankruptcy law in order to adapt to a continuously changing environment.

2.1.2 The development of the US bankruptcy law

The bankruptcy rules and practices in the United States from before the 20th_century protected the investments of creditors from inefficient investment behavior of debtors. The first US bankruptcy laws in the 19th_{century took a very harsh stance towards debtors. They} allowed very little discharge of debt and were mainly focused on recovery of the investments of creditors. Such creditor-friendly bankruptcy law stimulates liquidation of firms. But the early creditor-friendly procedures did not really fit the environment. Railroad corporations in the 19th _{century illustrated this misfit. During frequent financial panics in the 1800s, railroads} were the first large industrial corporations. These organizations often failed to pay their bills when the markets crashed. Because railroad companies were worth more alive than dead, ways were figured out to reorganize, rather than shut them down (Fox, J., 2008). As a result, corporations started to move around this bankruptcy law because the prescribed procedure was not appropriate for their situation. After the 20th_{century the focus of bankruptcy law} shifted in favor of debtors. This new legislation laid emphasis on the reorganization. The bankruptcy act formed in 1978, shaped the way bankruptcy codes in the United States look today. This act introduced Chapter 11 next to Chapter 7 and made it easier for individuals and firms to file for bankruptcy and reorganize. (Scott J.A. & Smith T.C., 1986)). Chapter 11 authorized managers to continue operating the business during bankruptcy and they were given the exclusive right to propose a reorganization plan. This shifted the bankruptcy process in favor of debtors.

When Creditor-orientated bankruptcy law stimulates liquidation of firms, managers and shareholders gain more from survival of the firm or at least postponement. This can motivate managers to make inefficient high-risk high-return investments (A.W.A. Boot & J.E. Ligterink, 2000). Reorganization, stimulated by debtor-orientated bankruptcy law

diminishes postponement of bankruptcy filings. But it also leads to too much continuation, and this is visible in the amount of firms which become financially troubled again after reorganization. (Hotchkiss 1995).

(6)

6

2.1.3 Reorganization in Chapter 11

Bankruptcy reorganization provides a remedy for capital market inadequacy, it protects firms whose value cannot be realized through sale or fragmentation (Doherty & Lopucki, 2007).

When a corporation is in financial distress, it may reorganize the business, sell it as a going concern or liquidate the assets. The reorganized businesses or sold going concern should solve the lack of operating profits, excessive debt and illiquidity. When a corporation is bankrupt, a firm can file under chapter 7 or chapter 11. Chapter 11 is a title in the US

Bankruptcy code which permits reorganization for corporations and partnerships. Chapter 7 is the code for liquidation.

A plan of reorganization is a proposal to exchange the firm’s existing financial claims for a new basket of claims, reducing the amount of debt in the capital structure (Gilson, 1995). During the reorganization, the debtor remains in control of the business operations and has the exclusive right to present the first reorganization plan. Typically after a year debtors are allowed to propose a reorganization plan. All claim classes must vote for approval of the plan. If the plan is not approved, the firm files for chapter 7, which implies liquidation of the assets. A majority of claimants and at least two third of the value within each claim class must approve the reorganization plan. Claim classes can be categorized in secured, senior and junior claim classes. (Kaley et al. 2007). Every priority level must be paid in full before the next highest priority level receives payment. So secured debt will have to be paid before unsecured senior claim classes and so on.

Changes in the bankruptcy law changed relative bargaining positions of creditors and debtors. But the environment in which corporations operate changed. Since the

technological movement in the 20th_{century, the value of many firms depends more and more} on innovation, knowledge and human capital, instead of just land equipment and buildings. Knowledge generated by an organization adds a lot of value but it is very difficult to sell. (Boot, 2009). Intangible assets are very difficult to value. Uncertainties about the value of a company increase the trouble which the different claim classes have to commonly and quickly find and execute the plan which maximizes the total firm value.

This section described how development of the bankruptcy law adjusted according to the needs of the market. The following paragraphs describe what problems exist causing difficulties for the coordination of claimants with different claims. Underinvestment and increased risk taking by the debtor, the bias towards liquidation and different perspectives with respect to the valuation of the firm.

(7)

7 2.1.4 Coordination problems

It is important to note that different claim classes position themselves to achieve the most economically efficient outcome for their particular class. Different priorities within claim classes cause conflicts of interests. These conflicts of interest complicate the road towards reorganization. The next paragraph describes behavior of debtors which can be motivated by their low priority after bankruptcy. This is followed by a description of the liquidation and continuing biases, and the role of valuation.

2.1.4.1 Risk-shifting and underinvestment

Consider a firm high in debt and in financial distress, where the owners have limited liability. The managers, acting on behalf of the shareholders, will be motivated to take larger risks at the expense of creditors. They will even invest in negative NPV investments as long as the upward potential is high enough. This behavior increases the odds that the distressed firm becomes solvent again, which means that some value remains for the debtors. However, the risky investments negatively affects the value of the claims of the creditors. This is the well-known risk shifting (or asset substitution) problem of a debt-overhang (Jensen & Meckling, 1976).

Myers (1977) showed that there is also an underinvestment problem; investors will not invest in marginally profitable investments. When there is risky debt and the expected return is equal or less than the interest payment, shareholders will not invest in a positive NPV investment. This is due to the fact that the NPV only accrues to the creditors without any benefit for the shareholders. This behavior goes at the expense of the shareholders. It reduces the firm-value which limits the amount of debt a firm can issue. For example, troubled

airlines tend to avoid further investing in maintenance because the benefits of doing this disproportionately accrue to the bondholders. Rational customers expect this and will avoid doing business with such a firm, resulting in diminishing total value of the firm (Schroeck, 2002).

2.1.4.2 Liquidating bias

The liquidation bias arises because of different priorities among creditors. Due to priority rulings, creditors with high priority have a tendency towards liquidation because this provides them with a quick and certain return. Pressure to liquidate the assets too quickly can destroy value for creditors with lower priority and debtors. Junior creditors and debtors with impaired claims gain more when the firm ‘survives’ and thus have a continuing bias (Boot, Ligterink, 2000). Bebckuck (1991) describes the purpose of the reorganization process as follows: The first objective is to maximize the total value of the assets and the second

(8)

8

objective is optimal division of this total value based on initial contracts. These objectives cause coordination problems between debtors and creditors in the bargaining process. The liquidation bias causes creditors with high priority to have interest with a low valuation. A continuation bias however, ensures creditors and debtors with low priority have interest with a high valuation. This complicates the consideration between liquidation and continuation during the reorganization.

Even if the considerations are straightforward, the complexity of valuing the firm complicates the strategy and execution of the plan. Valuation of the firm’s assets is difficult because the present information of a firm in financial distress is not representative of future earnings. Therefore, discounted future cash flows must be estimated, which are established using many uncertain factors. A problem is that the information around future cash flows is unequally spread around creditors, the management and shareholders. For example, if due to a lack of information one group of claimants attach less value to the firm’s going concern value lower than to the liquidation value, they might unjustly reject the plan and raise costs by delaying or causing inefficient liquidation.

Valuation is also important when the debt is conversed to equity after the

reorganization. Debtors and creditors then have to come to an agreement over the value of the shares (Doherty & Lopucki, 2007). Research by Brown et al. (1992) confirms the

asymmetric information problem which occurs during the valuation of the firm. He finds that debtors have the incentive to signal negative information which would make creditors more likely to accept a lower pay-off from the debt to equity conversion.

The complexity of the valuation of the firm’s assets increases divergence among claimants’ estimations of the value of the firm. This becomes more complex when a large fraction of those assets are intangible because intangible assets are more likely subject to information asymmetry resulting in more incentives for private information acquisition (Barth, Kasznik and McNichols, 2001). The more intangible assets there are, the more

difficult it becomes to estimate the value of the firm, which in turn increases the coordination problem. Another proxy for the complexity of the renegotiation process is the amount of claim classes (Kalay, 2007).

Problems during the reorganization also arise with the division of value. Debtors have the exclusive right to propose the first reorganization plan, disapprove, block, or delay a plan. This gives debtors a disproportional bargaining position, when the debt to equity ratio is very high. This is reinforced by the high costs for creditors to make their own reorganization plan due to expensive appraisals. Weiss (1990) confirms that when the creditors are not paid in full, equity holders still receive significant value. The literature shows that reorganization during Chapter 11 has several features causing coordination problems between debtors and

(9)

9

creditors. The following section describes the phenomenon of vulture investors, who often appear during Chapter 11 reorganization.

2.2 Vulture funds

Occasionally during financial distress, distressed debt investor appear. Because these funds ‘attack’ firms which are in financial stressed situation, the resemblance is often made with a vulture circling around a weak or sick prey. The charge of this nickname demonstrates the negative public image of these investors.

Vulture investors invest in securities of firms who find themselves in a situation of distressed debt. Among vulture investors, there are institutional investors, hedge funds and money investors as well as individuals (Gilson, 1995). Due to the uncertainty of recovery of the debt, distressed debt is traded at a discount. In 2006 the average market to face value ratio of distressed debt was 0.75 (Altman, 2007). According to Gilson, publicly traded bonds from a firm in financial distress trade at about 30% of their face value. The more junior a claim, the smaller the value at which the distressed bond is traded.

There are several ways vulture involvement takes place. For example, by investing in debt claims or by buying equity stakes (Jiang et al. 2012). Investment in debt claims and equity could respectively secure vulture investors a place in the unsecured creditors

committee or in the equity committee. Investment in debt could also be part of a loan-to-own strategy, the vulture then acquires debt with the intention to convert this debt into a

controlling equity stake when the firm emerges from Chapter 11. According to Gilson (1995) good access to information, skills to process this information, a good understanding of the risk and the skill to accurately value assets are factors contributing to the generation of profits by vulture funds. These skills are necessary to have a superior position in the bargaining and negotiating process.

Involvement of vulture investors in firms with distressed debt has increased in the last few decades. Evidence of vulture involvement during chapter 11 was found in 60% of the sample between 1980 and 1993 (Hotchkiss & Mooradian, 1997). This paper will now describe the different approaches which vulture funds undertake during Chapter 11 to generate profits.

2.2.1 Strategies of vulture investors

The different ways to make profits by vulture funds can be separated in active and passive strategies (Gilson 2002, Altman 2008). With the active strategies, vulture investors influence the restructuring process, in order to enlarge the value for all claimholders or by enlarging the value of claims in their specific claim class. The passive strategies involve identification

(10)

10

of undervalued debt claims and speculating that they appreciate (Hotchkiss & Mooradian, 1997).

In an active strategy, vultures make use of the influence gained by the purchase of distressed debt. Voting to influence the restructuring process or submitting a reorganization plan are examples of influence. One way for vulture funds to make a profit is by taking an active role in the management of the firm and deploying its assets more efficiently (Gilson, 2002). The purchased debt can be converted to equity for control of the assets after the reorganization. After the vulture seizes control, it can cause the company to be run more profitably by

directly influencing investment and operating policies. (Gilson, 1995) This increases the total value for all claimholders.

Alternatively, vulture investors can choose to frustrate the restructuring process, with the purpose of attaining a more favorable position in the priority and eventually often leading to the liquidation of the firm. This way, vulture funds make a profit at the cost of other

claimants. An example of frustrating the restructuring process is through bondmail. By buying the majority of distressed debt within a claim class, the vulture investor can threaten to hold up the reorganization unless he is distributed higher recovery. However, there is the risk of a so-called cram down. This entails that the reorganization plan may be confirmed in court despite the objection of any impaired class that votes against it. Such a forced

confirmation only holds up in court if the plan does not unfairly discriminate against the member of that class, is fair and equitable with respect to the members of that class. (Anson, 2000). Besides a passive buy-hold strategy, distressed bond arbitrage is another passive strategy for investors who think the stock is undervalued (Anson, 200). This strategy entails buying distressed debt and shorting the underlying stock.

2.2.2 Empirical evidence on the effects of the presence of vulture funds Table 1: This table summarizes literature about effects of vulture involvement on firm-value Thomas

(2003)

Due to inconstant creditor identities caused by distressed debt trading, relations have to be rebuilt to achieve reorganization. This delay costs money.

H&M (1997)

Positive abnormal returns two days before and after purchase of public debt by vulture funds.

Post-restructuring operating performance relative to pre-default is greater when the vulture investor gains control on the target firm.

Jiang et al. (2012)

A higher debt recovery benefitting junior claims and positive announcement effect.

(11)

11

The previous section described which strategies vulture investors can undertake to make profit. The generated profit generally originates from an increase in the value of the total firm, or from an increase in the value of a particular claim class. The following paragraph describes several researches which investigated the effect of vulture involvement in Chapter 11 on the value and performance of the firm and the respective claims.

Despite the bad public image of vulture funds, existing literature and theories are inconclusive about the value vulture funds add or extract (Table 1).

Thomas (2003) for example, argues that inconstant creditor identities create confusion, due to the sale and purchase of distressed debt. The exit and entrance of new creditors makes it necessary that new relations have to be built in order to reorganize. This delay in negotiations costs money and causes the debtor to languish in bankruptcy longer than necessary.

Hotchkiss & Mooradian (1997) investigated vulture involvements in the period between 1980 and 1993. This study finds that the post-restructure performance is significantly positively related to a vulture being active in the management of the

restructured firm. They also find significant positive abnormal stock returns of 8.64 % for two days surrounding the announcement-date of the purchase of public debt. Jiang et al. examine hedge funds involvements in distressed firms from 1996 to 2007. They find negative stock returns 5 and 10 days around the bankruptcy announcement date for the complete sample of firms filing for Chapter 11. Compared to abnormal returns of firms without hedge fund involvement, these returns where respectively 16.8% and 24.7% higher in 10 and 20 days surrounding the event date. The announcement effect indicates how the stock market initially perceives the involvement of the vulture. In another article, Jiang et al. (2012) investigated the contribution of hedge funds to coordination by testing whether these hedge funds help push the outcome in the direction of survival or liquidation. An outcome in-between hereby indicates a balance in-between debtors and creditors. They find that the involvement of hedge funds leads to higher debt recovery, and a higher positive effect on the stock market. Another noteworthy conclusion is that they found a more favorable

distribution to the junior class of claims. The presence of hedge funds on the side of unsecured creditors leads to higher debt recovery for this class and their presence in the debtor’s committee led to more favorable distributions to the debtor’s claims. Finally this paper concludes that hedge funds participation leads to a more neutral debt restructuring process. This process was usually dominated by either the (secured) creditors or the management, (Harner, 2008). Where managers facilitate and implement the restructuring plans but do not control the restructuring process. This is shown by a larger chance of loss of the exclusive right to file a reorganization plan by the debtor due to successful objections from the side of the creditors.

(12)

12

involvement of hedge funds in firms in Chapter 11. They find that the participation of hedge funds is positively correlated with firm size. Furthermore, they find that hedge funds are more likely to participate as unsecured creditors when secured debt relative to total assets are low. In addition, they are more likely to participate in the equity side when the equity is high.

2.2.3 Dysfunction of bankruptcy law and vulture funds

The literature shows that under current bankruptcy law, conflicts of interests in combination with uncertainty about the firm value increase the distance between different claimholders. Their coordination problems in Chapter 11, lead to inefficient and costly reorganizing. Vulture funds target firms in financial distress to squeeze themselves in this complex

situation. Jiang et al. (2012) concluded that involvement of hedge funds in Chapter 11 pushes the outcome to a more neutral debt restructure process (Jiang et al., 2012). This raises the question whether vulture funds only serve their own interests or whether their appearance, like the rail-road reorganizations, is a market reaction to supplement the misfit of bankruptcy law to the current business environment. The aim of this research is to find out whether vulture involvement occurs because the current bankruptcy law has not yet been adjusted in a response to the complexity of valuing intangible assets. This paper will therefore investigate whether vulture funds target firms with high uncertain values, where coordination problems are large.

2.3 Hypothesis

In this section the hypothesis that will be tested in this paper is presented.

Previous research found that liquidation biases, risk shifting and the complexities related to valuing a firm lead to coordination problems in firms in financial distress and during chapter 11 (Doherty, Lopucki 2007, Brown et al, 1992, Jensen & Meckling, 1976, Goldschmidt, 2005). The increasing importance of intangible assets in the current business environment causing coordination problems arises the question whether the current US bankruptcy law dysfunctions in the current environment. Jiang et al. (2012) concluded that involvement of hedge funds in chapter 11, help push the outcome to a more neutral debt restructure process characterized by a more favorable distribution to the junior class of claims. The contribution of vulture investors to the coordination problem leads to the following hypothesis:

Hypothesis 1: The size of the coordination problem is a positive significant determinant for the probability of vulture involvement in chapter 11.

(13)

13

Intangible assets will be used as a proxy for complexity and with that, for more likely coordination problems. Intangible assets are very difficult to value. The relative amount of intangible assets increases the complexity of estimating the value of a firm for all claim classes. This provokes the coordination problem. Disagreement on the value of the firm increases differences between preferences of debtors and creditors because it enlarges the liquidation bias and at the same time, the continuing bias.

The first hypothesis states that the presence of large coordination problems attract vulture funds. The second hypothesis relates to the reaction of the stock-market to the presence of Vulture investors in firms with a large coordination problem. Jiang et al (2012) find higher debt recovery for the junior class of claims after involvement of hedge funds and more positive returns around the announcement of the Chapter 11 filing when vulture funds were involved. Hotchkiss & Mooradian find positive announcement returns of 14.7% surrounding the announcement date of the debt purchase. Therefore expected is, that when a large coordination problem is present, the presence of vulture funds can mean more for the junior class of claims. This leads to the following hypothesis:

Hypothesis 2: Vulture involvement in combination with larger coordination problems has a positive impact on the abnormal returns around the

announcement of the vulture involvement.

3. EMPIRICAL ANALYSIS

3.1 Sample Selection

A list of all companies that went bankrupt was requested from the Lo Pucki Bankruptcy Research Database. From this database I extracted all the public companies that went bankrupt between 1995 and 2010. In this sample, only firms which filed an Annual Report (10-k or 10) not less than three years prior to the filing of the bankruptcy case are included. Further, the annual report must have reported assets worth over 100 million measured in 1980 dollars to be included in the Lo Pucki sample. 19 Cases were dropped from the sample because they were Chapter 7 filings. Graph 1 shows the number of bankruptcies and the amount of vulture involvement every year. In 2001 and 2009 the most filings occur, this can be explained by the two crises which occurred in those periods.

(14)

14

on the basis of a list of known vulture investors.1_{Involvement of a vulture fund is identified} by searching through the form 13G and 13D filings in the Edgar database, which has to be submitted when anyone acquires more than 5% of publicly traded equity. When debt is converted to equity, SEC requires a file informing the amount of and date of the purchased debt. These filings were searched until three years prior to the bankruptcy filing date.2_After this, the amount of claim classes (both impaired and unimpaired) for each company was manually collected by searching through the reorganization plans. These plans were found using BamSec.com and info.sec.com, and sometimes to be found on the bankruptcy website of the firm. Reorganization plans were found for 468 firms of 761 firms, of which 427 gvkey numbers are known. The amount of claim classes is the number of claim classes which were distinguished in the reorganization plan and excludes unclassified claims which are

administrative claims and priority tax claims.

Then, the gvkey numbers of the residual sample were used to find intangible assets, total assets, research and development expenses, total expenses, the value of property plant and equipment and the SIC code using the Compustat database. This information is reported from the last year in which the firm filed an annual report, which is till prior three years from the bankruptcy. However, some gvkey numbers were missing from the original sample, this led to a sample of 599 firms. Compustat didn’t have information on intangibles for every firm. This led to 520 observations of firms from which the intangible assets are known. Finally, for the event study, the CUSIP-codes of firms and the debt purchase date reported by the 3-G or 3-D file were used to compute the cumulative abnormal returns. This resulted in 70 events where vulture investors purchased debt from 59 distressed firms, involved and intangible assets are known.

1_{This list was composed in previous studies by Schoen (2009) and Dr. Ligterink. Through} public data (LEXIS NEXIS) firms were identified as vulture fund if they were connotated in such a role in the financial press, or if they were listed in Altman et al. (2008). One vulture investor, ‘Sopris’, is added to this list in response to their described vulture activities in the news.

2_{I extended the list in which Schoen (2009) identified vulture involvement for the period} 2000-2005 for firms with a minimum total asset value of 500 million dollar. I extended this list for the period 1995-2010 and included all firms with reported assets worth over $100 million.

(15)

15

Table 1 in the appendix shows the first results. It shows the total amount of vulture

involvement found for all firms which filed for Chapter 11 between 1995 and 2010. Besides the vulture involvement over the complete list, it shows the vulture involvement over the sample which is also shown in Graph 1. The vulture involvement in the sample (27.55%) is not very different from the vulture involvement in the complete list (26.43%). Graph 1 shows that the involvement of vulture investors varies from year to year with a slight increasing trend. The vulture involvement found is slightly smaller than the amount which Hotchkiss and Mooradian (1997) found in public debt purchases between 1980 and 1993 (29.2%), and also smaller than the amount Jiang et al (2012) found between 1997 and 2010 (60.7%). Possible explanations for this could result could be different time period and the more extensive search for vulture firms through news stories by Hotchkiss and Mooradian (1997). The extensive list of Jiang et al. (2012) using broad criteria to identify hedge funds

involvement, could be an explanation for their findings.

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Total Total 17 13 16 30 39 66 86 75 52 28 22 13 13 31 72 26 599 Vulture involvement 3 3 5 8 10 12 21 14 10 7 9 4 6 9 35 9 165 % Vulture involvement 18% 23% 31% 27% 26% 18% 24% 19% 19% 25% 41% 31% 46% 29% 49% 35% 27,55 0% 10% 20% 30% 40% 50% 60% 0 100 200 300 400 500 600 700 Axis T itle

Graph 1: This graph shows the number of firms which filed for Chapter 11 and the amount of vulture involvement by year from 1995 untill 2010.

(16)

16

Table 2 presents the summary statistics of the firm characteristics of the sample. The sample includes 599 observations. The table shows that apart from the total assets variable, the amount of observations differ from the sample size. The median of the sample of 762.2 $million is similar in magnitude to the median of 702 $million which Jiang et al. (2012) found in their sample. The median of the secured debt ratio in the sample is very low compared to the median Jiang et al (2012) found. The small amount of observations for which this variable is found, could be an explanation. The median of the amount of claim classes is exactly the same as the sample of Jiang et al. (2012). Table 3 presents the sample statistics for firms where vulture involvement is identified and for firms where no vulture involvement is identified. It shows that the medians and means of the total amount of assets is larger for the sample where vulture involvement is found. This is consistent with Hotchkiss and Mooradian (1997) who find means and medians for firms where vultures are involved of more than twice the value of the total assets. The table also shows a higher mean value of intangible assets for firms where vulture funds are involved compared to firms where no vulture involvement was observed.

A table containing sample statistics by industry is included in the appendix (table 2).

Table 2: Sample summary statistics for the complete sample. This table shows summary statistics for key

variables for a sample 599 companies that filed for Chapter 11 between January 1995, and December 31, 2010. The data are from the Compustat annual and Capital IQ Compustat databases. PPENT is the net value of the property, plant and equipment. SD stands for the secured debt to total asset ratio. Claim classes is the number of claim classes. The secured debt ratio is computed as secured debt divided by total assets and the Intangible ratio is computed by dividing the value of intangible assets by the value total assets. Claim classes were manually collected from reorganization plans. All accounting variables are measured in millions of dollars.

Variable Number Minimum Lower

Quartile

Median Upper

Quartile

Maximum Mean Standard

deviation Total assets (in

$millions)

599 0.6 374.7 726.2 1803.8 410063.0 4028. 4

23957.8

Intangible assets (in $millions)

520 0.0 2.6 60.6 252.8 17016.0 368.0 1386.5

PPENT (in $millions) 587 0 59.2 178.4 455.2 59862 792.8 3249.9

R&D expenses (in $millions)

209 0 0 1.9 18.284 6600 70.8 490.6

Sale (in $millions) 596 -4234.5 229.9 552.38 1353.1 207349.0 1984.

3

9298.2

Secured debt ratio

182

0% 0% 0.002% 0.009% 27.252% 0% 2%

Claim classes 427 4 7 9 12 56 10.7 6.6

Intangible ratio 520 0 0% 7.9% 27.4% 90.5% 16.4% 19.7%

(17)

17

This table shows that the manufacturing industry is largely represented in the sample and that relatively most vulture involvement appear in ‘Transportation, Communications, Electric, Gas’, Finance, Insurance, and Real Estate’ and ‘Manufacturing’ industry.

Table 3: Sample summary statistics for the sample split in the observations were vulture and no vulture

involvement is founds. This table shows summary statistics for key variables for a sample of in total 599 companies that filed for Chapter 11 between January 1995, and December 31, 2010. The data are from the Compustat annual and Capital IQ Compustat databases. PPENT is the net value of the property, plant and equipment. SD stands for the secured debt to total asset ratio. Claim classes is the amount of claim classes. The secured debt ratio is deﬁned as secured debt divided by total assets and the Intangible ratio is computed by dividing the value of intangible assets by the value total assets. The claim classes were manually collected from reorganization plans. All accounting variables are measured in millions of dollars.

No Vulture involvement

N Minimum Median Maximum Mean Standard dev Total assets(in $millions) 434 0.63 666.38 410063.0 3258.88 20860.77 Intangible assets (in

$millions ) 371 0 59.54 14314.3 341.41 1325.02

PPENT (in $millions ) 427 0 159.36 28618 639.00 2127.04

R&D expenses (in $millions )

151 0 1.30 170.94 15.94 33.83

Sale (in $millions ) 432 0 523.69 37120 1403.89 3583.01

Secured debt ratio 115 0 0 0.27 0.00 0.03

Claim classes 290 4 9 56 10.82 7.24 Intangible ratio 371 0 0.08 0.89 0.17 0.20 Vulture involvement

N Minimum Median Maximum Mean Standard dev Total assets(in $millions) 165 30.67 866.00 343573 6052.60 30631.19 Intangible assets (in

$millions ) 149 0 65.23 17016 434.33 1531.65

PPENT (in $millions ) 160 0 248.19 59862 1203.24 5154.792

R&D expenses (in $millions )

58 0 4.04 6600 213.55 920.11

Sale (in $millions ) 164 -4234.47 682.89 207349 3513.25 16685.44

Secured debt ratio 67 0 0 0.08 0 0.01

Claim classes 137 4 9 35 10.53 4.80

Intangible ratio 149 0 0.06 0.91 0.16 0.20

(18)

18 3.2. Research method

3.2.1 Probit Regression

In order to test the first hypothesis which states that the probability of involvement of vulture funds, in Chapter 11, is higher when there is a large coordination problem, a regression is constructed to estimate the coefficients of the explanatory variables of the maximum

likelihood function. A probit model is appropriate when the dependent variable can only take on two values. In this case, the outcomes are limited to vulture involvement or no vulture involvement. A probit model usually takes the following form:

𝑃(𝑌 = 1|𝑋) = Φ(𝑋′_𝛽)

Where P denotes the probability of the event, Φ is the standard normal cumulative distribution and 𝛽 is the coefficient of the independent variable estimated by maximum likelihood. In this thesis ‘Stata’ is used to perform the probit regression.

To decide whether vulture funds target financially distressed firms with a large coordination problem, the two equations below are constructed. The difference between the equations is the inclusion of the secured debt-ratio in equation 2. The observations of this variable is very limited, therefore both equations are estimated on this sample, to conclude the effect of exclusion of this variable. Hereafter equation 1 is estimated with use of the larger sample. 𝑷(𝒀 = 𝟏|𝑿) = 𝚽(𝛼 + 𝛽(𝐶𝑃) + (𝑓𝑖𝑟𝑚𝑠𝑖𝑧𝑒) + ∑nt=1995𝜖𝑡𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡) ) + ∑nt=1995𝜖𝑡𝑦𝑒𝑎𝑟𝑡 (1) 𝑷(𝒀 = 𝟏|𝑿) = 𝚽(𝛼 + 𝛽(𝐶𝑃) + 𝛿(𝑓𝑖𝑟𝑚𝑠𝑖𝑧𝑒) + 𝜄(𝑆𝑒𝑐𝑢𝑟𝑒𝑑𝐷𝑒𝑏𝑡) ∑nt=1995𝛾𝑡𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡) ) + ∑nt=1995𝜖𝑡𝑦𝑒𝑎𝑟𝑡 (2) Where: 𝑌 = 1 𝑖𝑓 𝑉𝑢𝑙𝑡𝑢𝑟𝑒 𝑖𝑛𝑣𝑜𝑙𝑣𝑒𝑚𝑒𝑛𝑡 𝑌 = 0 𝑖𝑓 𝑁𝑜 𝑉𝑢𝑙𝑡𝑢𝑟𝑒 𝑖𝑛𝑜𝑙𝑣𝑒𝑚𝑒𝑛𝑡

And where ∑ 𝜖n𝑖 𝑖𝐼𝑛𝑑𝑖 contains 8 dummy-variables to indicate 9 industries, which are

manufacturing, services, retail, finance insurance and real estate, transportation, mining, agricultural, whole sale trade and construction.

(19)

19

The Coordination problem, CP, is approximated by either the dummy variable ‘Coordination problem’, which indicates the rank according to the intangible asset ratio, or by the

continuous variable ‘Intangible Ratio’.

The literature revealed that complexity of valuing intangible assets decreases the ability of claimholders in all classes to coordinate themselves in order to achieve the most efficient outcome of the reorganization process (Doherty & Lopucki, 2007). In order to create proxies for the size of the coordination problem, the observations were ranked on the basis of the size of the intangible asset ratio which is computed as the value of the intangible assets, divided by the value of total assets. These ranked observations are subsequently divided in five quintiles containing an equal amount of observations. Observations within the quintile containing firms with the highest intangible asset ratios are indicated to have a large

coordination problem. Because these ratios and the amount of intangible assets vary among and in the quintiles, the intangible asset ratio itself will be used as a proxy as well.

Table 4 shows the descriptive variables by quintile. It shows twice the amount of vulture involvement for firms within the fifth quintile, relative to firms within the fourth quintile. It also shows that the average value of total assets is significantly higher for these firms

compared to the other quintiles. The reported intangible asset values in the first quintile are zero on average. This appears strange because some firms in quintile 1 do report Research and Development costs. The tangibility of the first quintile however, is very high compared to quintile 5. This makes it unlikely that the firms who reported zero intangible assets belong actually in quintile 5. To check robustness the regression using the intangible asset ratio as a proxy is run on the sample where observations with zero intangible assets are omitted. Several control variables are included in this regression. The first one is firm size which is measured by the logarithm of total assets. Vulture funds buy debt at extreme discounts (Altman, 2007), the size of intangible and tangible assets are related to the value of

liquidation or the going concern which is the source where vultures try to extract money from (Gilson 2002, Altman 2008). So because firm size is positively related to opportunities for Vulture funds to gain profits, it is expected that they target relatively larger firms. Hereafter dummies indicating the SIC-industry are included because the success of vulture involvement partially depends on the expertise and industry-related knowledge (Gilson, 1995). Another control variable is the secured debt to total asset ratio, low ratios are found to be positively related with unsecured creditor participation, a low ratio of secured debt to total assets implies that the senior debt is more likely to be overcollaterized, which creates a more active role for unsecured creditors (Jiang et al., 2012). Finally the varying amount of bankruptcies and vulture involvements through the years (Graph 1) leads to the inclusion of dummy variables indicating the year in which the firm filed for chapter 11.

(20)

20

Table 4: Descriptive statistics by quintile

Descriptive statistics for key variables by quintile. The observations are ranked according to Intangible Asset Ratio and divided into quintiles. This sample consists out of 520 observations, which are the observations for which Compustat reported intangible asset values The data are from the Compustat annual and Capital IQ Compustat databases. PPENT is the net value of the property, plant and equipment. Tangibility is PPENT divided by Total assets, SD stands for the secured debt to total asset ratio, Claim classes is the amount of claim classes SD and the secured debt ratio is deﬁned as secured debt divided by total assets and the Intangible asset ratio is computed by dividing the value of intangible assets by the value total assets. The claim classes were manually collected from reorganization plans. All accounting variables are measured in millions of dollars.

Quintile VI Total Intangible PPENT Tangibilit

y

R&D Sale SD Claim Intangibl

e le-

Assets Assets classes asset

tratiotio ratio 1 N 30 104 104 98 26 101 27 74 104 mean 1684.16 0.00 584.41 35% 22.36 1205.7 8 0.00 11.28 0.00 sd 4348.00 0.00 1942.36 68.79 4179.4 9 0.00 7.56 0.00 p50 0.63 0.00 0.00 0.00 -4234.4 7 0.00 4.00 0.00 max 33234.00 0.00 17818.00 347.74 35925. 00 0.00 48.00 0.00 min 508.67 0.00 188.19 0.00 255.58 0.00 9.00 0.00 2 N 30 104 104 103 36 104 13 60 104 mean 901.60 10.62 209.91 23% 9.65 505.76 0.00 11.25 0.04 sd 2050.90 9.48 445.38 25.88 645.17 0.00 7.00 0.05 p50 30.67 0.00 0.00 0.00 0.00 0.00 5.00 0.00 max 19051.94 31.42 3161.00 126.60 4446.0 0 0.01 35.00 0.39 min 383.44 7.81 88.17 0.00 336.40 0.00 8.00 0.02 3 N 26 104.00 104.00 104.00 39.00 104.00 26.00 73.00 104.00 mean 1641.22 62.78 376.20 23% 12.53 1072.6 8 0.01 9.60 0.14 sd 5409.58 22.45 665.62 27.61 2094.0 7 0.05 4.05 0.13 p50 73.80 31.59 0.00 0.00 15.30 0.00 5.00 0.00 max 52705.05 111.57 4345.00 161.20 16918. 00 0.27 31.00 0.76 min 511.32 60.64 136.59 2.38 568.43 0.00 9.00 0.11 4 N 21 104.00 104.00 103.00 38.00 104.00 41.00 77.00 104.00 mean 1596.40 198.49 419.31 26% 14.51 1009.8 9 0.00 10.18 0.30 sd 3267.27 64.25 1868.92 29.10 1493.6 8 0.00 7.74 0.18 p50 254.60 112.99 9.96 0.00 0.00 0.00 4.00 0.01 max 23226.99 334.10 18850.97 131.25 11597.6 9 0.00 56.00 0.78 min 663.26 181.74 146.65 1.55 589.62 0.00 8.00 0.29 5 N 42 104.00 104.00 101.00 35.00 104.00 46.00 89.00 104.00 mean 15286.81 1568.29 2364.63 15% 323.11 6233.3 3 0.00 11.39 0.34 sd 55078.19 2799.72 6527.53 1173.0 4 20892. 87 0.01 6.38 0.22 p50 457.78 335.35 5.10 0.00 30.68 0.00 4.00 0.01 max 410063.00 17016.00 59862.0 0 6600. 00 207349 .00 0.08 42.00 0.91 min 2318.16 625.62 620.03 5.50 2137.27 0.00 10.00 0.31

(21)

21

Claim classes is also a proxy for the coordination problem, as it measures the complexity of the

renegotiation process (Kaley, 2007). In order to check robustness, a regression is performed on the number of claim classes to check whether the probability of vulture involvement increases when a large number of claim classes is observed. Quintiles are created based on ranking according to the number of claim classes. An indicator variable is created when an observation falls in the highest quintile. The sample differs from the sample in the previous regressions. Observations where intangible assets were not observed are included and observations were claim classes were not

observed were excluded. This results in a sample of 427 observations. The descriptive statistics of these quintiles are included in table 10 in the appendix.

𝑷(𝒀 = 𝟏|𝑿) = 𝚽(𝛼 + 𝛽(𝐷𝑒𝑏𝑡 𝐶𝑙𝑎𝑠𝑠𝑒𝑠 𝑞𝑢𝑖𝑛𝑡𝑖𝑙𝑒) + 𝛿(𝑓𝑖𝑟𝑚𝑠𝑖𝑧𝑒) + ∑nt=1995𝜖𝑡𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡) ) + ∑nt=1995𝜖𝑡𝑦𝑒𝑎𝑟𝑡 (3)

3.2.2 OLS regression using CAR

According to hypothesis 2, investors welcome involvement of vulture investors more, whenever coordination in Chapter 11 is difficult. To test this hypothesis, an event study is performed at the announcement date of the debt purchase. Cumulative abnormal returns (CAR) resulting from the event study can then be used to examine whether a dummy variable indicating a large coordination problem, positively affects the cumulative abnormal return. As in the study of Jiang et al. (2012), CAR’s are calculated during a period of 10 and 20 days surrounding the event date. These periods are indicated by (-5, 5) and (-10, 10). These periods were used to exclude biases caused by rumors leakages and slow market reactions around the event date. In contrast to Jiang et al. (2010), the event is not the Chapter 11 filing date. But as in Hotchkiss and Mooradian (1997), the CAR’s are calculated around the

announcement date of distressed debt purchases of vulture funds, which were found in the SEC documents filed by vulture investors.

Daily abnormal returns are calculated to investigate the reaction of the market to the debt purchase. The abnormal return 𝐴𝑅𝑖𝑡 is the difference between the expected normal return

and the actual return is computed by subtraction of the ‘normal return’, from the daily return, shown by the following equation.

𝐴𝑅𝑖𝑡 = 𝑅𝑖𝑡− 𝐸(𝑅𝑖,𝑡)

Here 𝑅𝑖𝑡 is the daily return and 𝐸(𝑅𝑖,𝑡) is the normal return. The normal return is computed

(22)

22

and the return of the market portfolio. In this study the CRSP value-weighted index is used as the market portfolio. In the value-weighted index, security are weighted according to the fraction of the firm’s value of outstanding shares relative to the value of the market index. By using this index, very large movements in the market index caused by very large firms, are prevented.

𝐸(𝑅𝑖|𝑅𝑀𝑡) = 𝛼 + 𝛽𝑅𝑚,𝑡+ 𝜀

Then, depending on the event window, 𝐶𝐴𝑅𝑖𝑡, the cumulative abnormal return can be

computed, by adding all daily abnormal returns;

𝐶𝐴𝑅𝑖𝑡 = ∑ 𝐴𝑅𝑖,𝑡 𝑡2

𝑡=𝑡1

After the cumulative abnormal returns are calculated, a test is necessary to infer whether these are significantly different from zero. When there is no evidence that the CAR is significantly different from zero, the outcome of the regression is not very meaningful. When the sample has significant cumulative abnormal returns, a regression can be constructed to examine whether vulture involvement in firms with a large coordination problem, leads to more positive stock-market reactions. The computed individual CARs are used as the dependent variable. As independent variable, the proxy for the coordination problem will be used. SIC-industry and year variables are included as explaining variables. This will be examined using an OLS-regression with the following equations:

𝐶𝐴𝑅𝐼(−5, +5) = α(𝐶𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑖𝑜𝑛 𝑝𝑟𝑜𝑏𝑙𝑒𝑚)+ ∑ 𝛽𝑡(𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡 n t=1995 ) +∑nt=1995𝛾𝑡(𝑦𝑒𝑎𝑟𝑡) (4) 𝐶𝐴𝑅𝐼(−10, +10) = α(𝐶𝑜𝑜𝑟𝑑𝑖𝑛𝑎𝑡𝑖𝑜𝑛 𝑝𝑟𝑜𝑏𝑙𝑒𝑚)+ ∑ 𝛽𝑡(𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡 n t=1995 ) +∑nt=1995𝛾𝑡(𝑦𝑒𝑎𝑟𝑡) (5) Similar to the probit regression, the coordination proxy is based on the size of the intangible

asset to total asset value. A firm is determined to have a large coordination problem, when the ratio exceeds or is equal to 0.316, which is the minimum observation of the fifth quintile in the total sample. Another influence on the stock-market reaction could be the amount of purchased distressed debt by vulture funds. However, the results I found were different from the results listed in the dataset which I extended so due to inconsistency I decided not to extend this information.

(23)

23 3.3 Results

3.3.1 Probit Regression

In order to determine whether the presence of coordination problems increases the

probability of vulture involvement, the coefficients of the explanatory variables are calculated by means of a probit model using a maximum likelihood analysis. To see if there is a linear relationship between the size of the intangibles and the probability of vulture involvement, the intangible assets to total assets value ratio is used as an explaining variable. Because a linear relation between the intangible asset to total asset ratio and the presence of a large coordination problem is unlikely, the coordination problem as explaining variable is also approximated by an indicator variable. Observations are ranked according to the intangible asset to total asset ratio, and then divided in quintiles. Observations in the quintile with the highest intangible asset to total asset ratios are indicated to have a large coordination problem.

According to the literature, the probability of vulture involvement is expected to be larger for firms with a large coordination problem, than for firms with a small coordination problem. A positive significant sign would confirm this expectation. Firm size is also expected to be a positive determinant for vulture involvement because the size of the firms is expected to be related to the opportunities there are for vulture funds to generate profits within a financially distressed firm (Gilson 2002, Altman 2008). Jiang et al. (2012) found a positive relation between hedge fund participation and the firm size. Because low secured debt to total asset ratios imply that there is more to cover unsecured debt, negative coefficient is expected for secured debt ratio (Jiang et al. 2012).

The results of the regressions are presented in table 5 (Tables 3 and 4 in the appendix include the Stata output of regression (3) and (6)). The first three and the last three regressions use respectively, the coordination problem indicator and the intangible asset to total asset ratio as proxies for the coordination problem. Columns (1) and (4) show an insignificant negative effect of secured debt ratio. The effect of including the secured debt ratio as control variable is negligible when the results are compared with columns (2) and (5). The results in columns (3) and (6) omit this control variable, which increases the size of the sample.

(24)

24

Table 5: Probit regression

The table shows results from probit regressions of vulture involved in Chapter 11 from January 1995 to December 2010 on the probability of vulture involvement. Regressions 1, 2 and 3 use Coordination problem, which indicates the firms belonging to the highest quintile, based on intangible asset to total asset value-ratio rank. As control-variables, firm size, secured-debt-ratio, SIC-industry variables and year variable are used. In regressions 4 ,5, 6 the coordination problem dummy is replaced by the Intangible asset ratio. Which is computed as the value of total intangible assets divided by the value of total assets. Regressions 1, 2, 4 and 5 use the observations where secured debt is observed. Regressions 3 and 6 are performed on the whole sample. Firm size is measured by total assets. A log transformation is applied to firm size. The low amount of observations in the Agricultural production crops results in multicollinearity, therefore three observations belonging to this industry were omitted. Standard errors are reported in brackets. Statistical signiﬁcance at the 1%, 5%, and 10% level is indicated by ***, **, and *, respectively.

(1) (2) (3) (4) (5) (6)

VARIABLES Vulture Vulture Vulture Vulture Vulture Vulture Constant -0.863 -0.866 -1.493*** -0.763 -0.768 -1.484*** (0.768) (0.768) (0.493) (0.784) (0.783) (0.493) Coordination problem 0.157 0.157 -0.0818 (0.280) (0.280) (0.159) Intangible Ratio -0.346 -0.345 -0.278 (0.564) (0.564) (0.331) Firm size 0.145 0.146 0.103** 0.144 0.145 0.105** (0.0944) (0.0942) (0.0471) (0.0951) (0.0949) (0.0472) Secured debt -0.873 -0.910 (6.747) (6.511) Mining 0.902 0.904 0.610* 0.775 0.778 0.581* (0.696) (0.695) (0.328) (0.699) (0.699) (0.331) Construction 0.783 0.783 0.186 0.681 0.682 0.174 (0.806) (0.806) (0.415) (0.813) (0.813) (0.416) Agricultural production crops -0.294 -0.298 -0.0278 -0.279 -0.284 -0.0291 (0.289) (0.288) (0.177) (0.288) (0.286) (0.177) Transportation, Communications, Electric, Gas -0.552 -0.545 (0.400) (0.400) Wholesale trade -0.0301 -0.0314 0.103 -0.0726 -0.0739 0.0956 (0.423) (0.423) (0.205) (0.427) (0.427) (0.205) Retail -0.182 -0.185 -0.434* -0.234 -0.238 -0.464** (0.592) (0.592) (0.222) (0.603) (0.602) (0.227) Finance, Insurance &

Real estate 0.0308 0.0261 -0.0961 0.0252 0.0201 -0.0860 (0.369) (0.368) (0.201) (0.370) (0.368) (0.201) Services 0.175 0.182 (0.546) (0.547) 1996 0.395 0.403 (0.505) (0.506) 1997 0.419 0.425 (0.459) (0.459) 1998 0.126 0.133 (0.443) (0.444) 1999 -0.903 -0.902 0.0561 -0.924 -0.922 0.0645 (0.789) (0.788) (0.430) (0.790) (0.790) (0.431) 2000 -0.381 -0.381 0.295 -0.370 -0.370 0.305 (0.495) (0.495) (0.417) (0.498) (0.498) (0.418) 2001 -0.844* -0.845* -0.0362 -0.852* -0.853* -0.0301 (0.492) (0.492) (0.427) (0.496) (0.496) (0.428) 2002 -0.859 -0.867* -0.0805 -0.838 -0.846 -0.0728 (0.527) (0.524) (0.440) (0.527) (0.523) (0.441) 2003 -0.572 -0.572 0.152 -0.563 -0.563 0.173 (0.542) (0.542) (0.476) (0.542) (0.542) (0.477) 2004 0.139 0.138 0.566 0.0709 0.0701 0.577 (0.599) (0.599) (0.469) (0.606) (0.606) (0.470) 2005 -0.315 -0.317 0.412 -0.277 -0.280 0.432 (0.690) (0.690) (0.538) (0.691) (0.690) (0.539) 2006 -0.00614 -0.00728 0.855* -0.0525 -0.0536 0.862* (0.982) (0.982) (0.516) (0.985) (0.985) (0.516) 2007 -1.094* -1.095* 0.323 -1.046 -1.048 0.333 (0.642) (0.642) (0.449) (0.646) (0.645) (0.451) 2008 -0.0228 -0.0269 0.762* 0.0438 0.0394 0.781* (0.498) (0.497) (0.412) (0.495) (0.494) (0.413) 2009 0.446 0.461 (0.464) (0.466) 2010 Observations 149 149 517 149 149 517

(25)

25

Because the form of the function is nonlinear, the interpretation of the estimated coefficients is different from interpretation in the case of a linear probability model. To obtain the

required relationship between changes in the dependent variable and the probability, the functions need to be differentiated with respect to the variables. Table 6 shows the impact of changes in the explanatory variables is evaluated by setting them to their mean values, for regressions 3 and 6. (The stata output of these regressions are shown in tables 5 and 6 in the appendix).

Table 6: Margins: This table shows the marginal effects of the coefficients of the

explaining variables which were computed for column 3 and 6 in Table 5. (3) (6) Coordination problem -0.026 (0.051) Intangible Ratio -0.089 (0.105) Firm size 0.033 0.033 (0.015) (0.015) Mining 0.195 0.185 (0.104) (0.105) Construction 0.059 0.056 (0.132) (0.133)

Agricultural production crops 0 0

(omitted) (omitted) Transportation, Communications, Electric, Gas -0.009 -0.009

0.057) (0.056)

Wholesale trade -0.176 -0.174

(0.127) (0.127)

Retail 0.033 0.030

(0.065) (0.065) Finance, Insurance & Real estate -0.139 -0.148 (0.070) (0.071) Services -0.031 -0.027 (0.064) (0.064) 1996 0.056 0.058 (0.174) (0.174) 1997 0.126 0.128 (0.161) (0.161) 1998 0.134 0.135 (0.146) (0.146) 1999 0.040 0.042 (0.141) (0.141) 2000 0.018 0.021 (0.137) (0.137) 2001 0.094 0.097 (0.133) (0.133) 2002 -0.012 -0.010 (0.136) (0.136) 2003 -0.026 -0.023

(26)

26 (0.140) (0.140) 2004 0.048 0.055 (0.152) (0.152) 2005 0.180 0.184 (0.149) (0.149) 2006 0.131 0.138 (0.171) (0.171) 2007 0.273 0.275 (0.163) (0.163) 2008 0.103 0.106 (0.143) (0.143) 2009 0.243 0.249 (0.130) (0.130) 2010 0.142 0.147 (0.148) (0.148)

The regression using the highest quintile as a proxy predicts that the probability of vulture involvement for a firm with a large coordination problem is 2.6% lower than the probability of vulture involvement for another firm. When intangible asset ratio is used as a proxy, the model predicts that for each 0.10 increase of the intangible asset ratio, the probability of vulture involvement decreases with 0.089. In contrast with the expectation, negative

coefficients are estimated. However, none of these coefficients are significantly different from zero. This means that we have to reject the first hypothesis.

Firm size has a positive significant effect on the probability of vulture involvement. Each increase of 1 percent of the firm size, increase the probability of vulture involvement with 3.3 percent. This result is statistically significant at the level of 5% and in line with the

expectations and similar to the finding of Jiang et al. (2012). Further the results show a positive effect when the bankruptcy was filed in 2006 and 2008 and positive effects for the retail and mining industry at the 10% statistical level.

These results prove that the probability of vulture involvement increases with the size of the firms but they do not confirm the expectation that the size of the coordination problem is a positive significant determinant for the probability of vulture involvement in chapter 11.

3.3.2 Goodness-of-fit

In order to analyze how well the models predict whether a vulture fund buys distressed debt from a particular firm in financial distress, two test statistics are calculated in this section. Methods regularly used for linear dependent variable models such as the residual sum of squares are not appropriate to measure the fit of the non-linear probit model. These models measure how tight the model fits to the data. The objective of maximum likelihood however,

(27)

27

is to find the values of parameters which maximize the probability of observing a particular sample. Therefore different methods have to be used.

A measure of fit that is commonly reported for limited dependent variable models is the percentage of values correctly predicted. This would imply calculating the percentage of correctly predicted outcomes, in this case the involvement of vultures (Y=1) and the absence of vulture involvement. However, Kennedy (2003) points out that this is not a good measure, when the amount of zeroes and ones are unbalanced. When there are 100 observations for example, and there are 70% zeroes observed, a predictor predicting only zeroes will have predicted 70% correctly. Instead, he suggests to add the fraction of correctly predicted ones and the fraction of correctly predicted zeroes. This measure is algebraically shown by the following equation: 𝑃𝑒𝑟𝑐𝑒𝑛𝑡 𝑐𝑜𝑟𝑟𝑒𝑐𝑡 𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑖𝑜𝑛𝑠 = 100𝑥 [∑ 𝑦𝑖𝐼(𝑃̂) ∑ 𝑦𝑖 + ∑(1−𝑦𝑖)(1−𝐼(𝑃̂)) 𝑁−∑ 𝑦𝑖 ] (6) Where I (𝑦̂) = 1 𝑖𝑓 𝑦𝑖 ̂ > 𝑦̅ 𝑎𝑛𝑑 0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒 𝑖

The classification statistics, using the fraction of vulture involvement as a threshold (𝑦̅ ), have been used to compute the percentage correct predictions. (Table 7 in the appendix shows the classification statistics).

Table 7 shows the computed percentages correct prediction for regressions (2), (3), (5) and (6). It shows that the correct predictions in the restricted sample, are slightly higher than in the unrestricted sample. The regressions using both samples also show that the difference between the correct predictions using the indicator or the ratio proxy is very small. The likelihood ratio test examines the hypothesis that all the coefficients are jointly zero. Acceptance of this hypothesis would mean that inclusion of the coefficients add no

explanatory power in comparison with a model without the explanatory variables (Brooks, 2008). The test statistic is computed as follows:

𝐿 = −2[𝑙𝑜𝑔𝐿 − 𝑙𝑜𝑔𝐿0] (7)

Table 7: The table shows the percent correct predictions for all regressions. Regressions 2 and 3 use Coordination problem, which indicates the firms belonging to the highest quintile, based on intangible asset ratio

ranking. In regressions 5 & 6 the coordination problem dummy is replaced by the Intangible asset to total value ratio. (2) (3) (5) (6) N 149 517 149 517 Threshold 0.396 0.288 0.3959 0.288 Percent correct predictions 133.30% 125.35% 131.08% 126.70%

(28)

28

Where LogL is the log-likelihood of the regression and Log𝐿0 is the log-likelihood that would

have been obtained without the coefficients in the regression. This statistic follows the chi-square distribution with k degrees of freedom. K is the number of coefficient tested to be zero. Table 8 shows the computed statistics for both models.

Table 8: This table shows the computed 𝜒2_{-statistics following equation 7. Regressions 2 and 3 use} Coordination problem, which indicates the firms belonging to the highest quintile, based on intangible asset to total asset value-ratio rank. In regressions 5 & 6 the coordination problem dummy is replaced by the Intangible asset to total value ratio.

(2) (3) (5) (6)

LogL -100 -310.5 -100 -310.5

Log𝐿₀ -89.4 -290.9 -89.4 -290.9

𝜒2_-statistic _21.18 _39.13 _21.24 _39.57

Prob>chi 0.1058 0.0265 0.2675 0.0238

The p-value shows the probability of computing a statistic as extreme as calculated, under the hypothesis that the coefficients are jointly zero. From the p-values it can concluded that coefficients of variables in the unrestricted sample (columns 3 and 6) are significantly different from zero using the ratio or the indicator as proxy for the coordination problem. The p-values of column (2) and (5) provide no evidence that coefficients estimated using the restricted sample are different from zero.

3.3.3 Robustness-check

In order to check the robustness of the probit model, this part investigates the effect of complexity measured by the amount of debt-classes on the probability of vulture

involvement. Further it is important to check whether the coefficients behave different when the observations reporting zero intangible assets are omitted. The descriptive statistics by quintile, in table 4 show that some of these firms report R&D-costs and simultaneously zero intangible assets. This raises doubts about the trustworthiness of these reported variables. The high values of property plant and land relative to the total firm-assets suggest that the value of intangible assets of these observations may not be significant in magnitude. It is then unlikely that these observations would be assigned to have a coordination problem according to the quintile proxy. But it is still important to check the effect of omitting these

(29)

29

For the first check, the quintile containing firms with the largest amount of Claim classes is used as a different proxy for the coordination problem. Claim classes is a different measure of the complexity of the reorganization process (Kalay, 2009). It is checked whether the probability of vulture involvement increases when there is a large amount of claim classes. Column 1 in Table 9 shows a positive coefficient for the claim classes’ proxy, but it is statistically insignificant, therefore it can’t be concluded that the probability of vulture involvement increases when complexity, measured by the amount of Claim classes, is large. Similar as the regression using the intangible asset ratio proxy as explaining variable, the results show a positive effect when the bankruptcy was filed in 2006 and 2008 and positive effects for the retail industry at the 10% statistical level

For the second check, the coefficients calculated using the sample omitting observations reporting zero intangible assets is compared to the original sample including these

observations. The results are listed in the second panel in Table 9. Omitting the observations reporting zero intangible assets do not seem to have a large effect on the coefficient for the Intangible ratio. But the results do show differences in the magnitudes and signs coefficients for the control-variables. This is probably caused by the shrinkage of the sample, altering the composition of firm-characteristics.

Does bankruptcy law sweeten the taste of distressed debt? : coordination problems and the presence of vulture investors

Does bankruptcy law sweeten the taste of distressed

debt?

Table of contents