Out-of-court renegotiations of financial contracts in financially distressed firms : inside the black box : an empirical analysis in the determinants of out-of-court workouts in the United States

(1)

1

Out-Of-Court Renegotiations of Financial

Contracts in Financially Distressed Firms

Inside the black box: an empirical analysis in the determinants of out-of-court

workouts in the United States

Master Thesis

July 2014

Geert-Jan Morskate

6081657

geertjan.morskate@gmail.com

MSc Business Economics

Finance track

Supervisor: Dr. Vladimir Vladimirov

(2)

2 Abstract

This paper tries to open the black box of out-of-court restructurings. Little is known about how out-of-court workouts are renegotiated. This paper gathered a sample of out-of-court workouts between 1999 and 2013.The recent popularity of distressed exchanges, a form of an out-of-court workout, indicates the increasing popularity of out-of-court workouts. Between 1999 and 2005 the total debt restructured was merely USD 6 billion while after 2005 the total debt restructured became roughly USD 60 billion. In an out-of-court debt renegotiation there are two key issues: what kinds of securities are exchanged for the current debt and how many? Using the gathered sample on workout specifics both aspects are explored. In both cases the most important factor is the leverage. When the firm is more leveraged it is more likely to exchange equity for debt and the recovery rate for investors is lower. Furthermore, intangible assets make a debt-to-equity more likely when the firm has a high leverage. The recovery rate for creditors increases when the firm has more R&D spending and high leverage. However, when restricting the sample to after BAPCPA in 2005 the dynamics seem to change. Hence, much research is still needed to better understand workouts.

(3)

3

1. Introduction

The recent financial crisis has been an economic difficult time for many firms. In such times more firms struggle to stay afloat. The last two decades there have been two major recessions and in the height of these recessions the number of defaults soared (see Appendix A Figure 1). Furthermore, the total number of defaults during recessions seems to increase and during the most recent financial crisis there were more defaults and distressed exchanges than ever (Altman and Karlin, 2008; also see appendix A Table II). Defaulted firms or firms struggling to repay their creditors are in financial distress. However, financially distressed firms (or their creditors) do not automatically need to file for chapter 11. Firms can renegotiate their debt out-of-court as well, of which a distressed exchange is an example when the restructured debt is publicly listed. The U.S. saw a surge of these distressed exchanges, which seems to indicate that out-of-court workouts (“Workouts”) are becoming more popular and important (see Appendix A Table II). This is potentially due to the regulatory change in 2005 when BAPCPA was signed, making the bankruptcy process less favorable for managers (Miller, 2007; Hotchkiss et al., 2008). The regulatory change is long-term and therefore will keep stimulating workouts and therefore a better understanding of renegotiation process is important.

The aim of this paper is to open the black box of workouts and to investigate with what kind of securities successful workouts are restructured and how much of a haircut investors take. To my knowledge this has not been researched before, most likely due to the fact that firms disclose only limited information on their workout. There is to my knowledge also no database which contains this information. This paper composes a sample of workouts by identifying workouts in the U.S. between 1999 and 2013, which consists of both public and private firms. The workouts are identified by (1) looking at distressed exchanges; (2) looking at defaults that are not followed by a bankruptcy. The distressed exchanges were in the recent financial crisis a popular tool to avoid bankruptcy and to restructure public debt out-of-court. These distressed exchanges generate a wealth of detailed information about workouts because these firm needs to make a public offer for the old debt. Using both Moody’s and S&P identified distressed exchanges workouts are gathered. Furthermore Moody’s database is used to identify defaults without a bankruptcy following. Subsequently, by going through the firm’s 10-k details about the financial restructuring are gathered. The advantage of this dataset is the detailed information about the restructuring. However, the

(5)

5

limitation of this dataset is that it contains only large firms that need to file 8-k and 10-k’s at the SEC.

There is only a limited amount of current research on out-of-court workouts which is in general relatively old and also mostly focusses on when firms restructure out-of-court. For example Gilson et al. (1990) find that firms with intangible assets are more likely to restructure out-of-court. Opler and Titman (1994) find that firms that operate in a concentrated market, or have more R&D spending are also more likely to restructure out-of-court. A more recent paper by Jostrandt and Sautner (2010) adds that firms with more leverage are also more likely to restructure out-of-court. This paper tries to open the black box of out-of-court restructurings and to look at how successful workouts are structured, e.g. debt-to-equity exchanges or debt-to-debt exchanges? Furthermore, the composed dataset contains specific workout information about how much the creditors get. Nevertheless, to determine the recovery rate is difficult because the repayment is through several kind securities. Hence, when a firm pays the creditors in cash, equity, or debt the liquidity will be different and the same payment with a different liquidity can mean that the value of the same amount can differ. Furthermore, to determine how large of a haircut creditors take the market value of debt would be best but this is not known for non-public listed debt. Therefore, the analysis of the recovery rate explores the gathered sample but the findings only give an indication of what matters for the creditor’s recovery rate.

The rest of the paper is structured as follows: In section two the current literature on both the method of restructuring and bargaining power of creditors are reviewed. Then section three describes in detail how the dataset is composed and what kind a workouts are in the sample. Section four reviews the research method and hypotheses. Section five presents the results and discusses potential issues. Finally section six gives an overview of the main findings and suggestions for future research.

2. The Workout Renegotiation Process

When renegotiating the financial contracts in financially distressed firms there are two main aspects. First it matters with what kind of securities the current contracts are going to be restructured. Debt contracts can be exchanged for equity, new debt, or cash. In the first section the current literature on capital structure is examined and its potential implications for the workout renegotiations and the choice of the restructuring method. The second matter is how much of a haircut do investors take. Previous research identified factors making a

(6)

6

workout more likely but not to whom it gives an advantage in the negotiations. The literature review is structured as following. First the method of restructuring is examined using the general capital structure theory. Secondly, the effects of firm characteristics on the bargaining power are discussed.

2.1.1 The Method of Restructuring

Modigliani and Miller (1961) theorize that when there are perfect capital markets, rational market participants, and there is perfect certainty, that then the capital structure of the firm should be irrelevant for the firm. In a perfect capital market participants among others have equal and costless access to information, there is no asymmetric information, and there are no transaction costs. In practice investors have asymmetric information about the performance of the firm. Managers have a better view on the performance than for example potential new investors or current creditors. Additionally, when firms become financially distressed they need to restructure, either through a bankruptcy or a workout, which entails transaction costs (e.g. Gilson et al., 1990). With imperfect capital markets the capital structure does matter for the firm. When raising new capital firms indeed do take into account the impact on their financial flexibility (Graham and Harvey, 2001). The financial flexibility becomes important when the firm becomes financially distressed. Firms that need to raise new funding or restructure their debt under asymmetric information should have used their debt capacity and now need to attract equity. When the firm faces low asymmetric information during the renegotiations then they should build up debt capacity (Inderest and Vladimirov, 2014).

When a firm becomes financially distressed it has trouble honoring all the debt requirements of the existing capital structure. Hence, either through a workout or through a formal bankruptcy procedure the firm can alter their capital structure. The capital structure can be altered by exchanging existing debt for new debt, equity, or for cash. To my knowledge there are no papers that research how or why firms and creditors chose one over the other in a workout. However, there is literature on the general choice of debt or equity financing. Rajan and Zingales (1995) investigate factors such as tangible assets, profitability, investment opportunities, and firm size and their effect on leverage. Firstly, when a firm has a considerable amount of tangible assets, and therefore can offer collateral, have in general a higher leverage ratio. Secondly, when a firm has many investment opportunities they need a lower leverage (the M/B is used as a proxy for investment opportunities). A firm with high leverage sometimes will need to forgo also profitable opportunities, which is costly for firms with many investment opportunities (Myers, 1977). Thirdly, larger firms more often issue

(7)

7

debt while smaller firms issue more often equity (Frank and Goyal, 2005). Rajan and Zingales (1995) argue that larger firms less often go bankrupt and therefore size is a proxy for the probability of a failure, making larger firms less risky and able to take on more debt. Finally, profitability decreases the leverage because managers prefer to first use internal funds.

A firm in financial distress has too much debt and needs to find a new leverage ratio or needs to adjust the repayment schedule. The previous identified factors that correlate with leverage can also be applied on workouts and the choice to restructure debt for debt, equity, or cash. When the total assets of a firm in a workout consist of relatively many intangible assets and the firm has a high leverage then it is more likely that the workout is restructured via a debt for equity exchange. The intangible assets are difficult to liquidate, also in the future. Hence, a capital structure with relatively low leverage is more sustainable and by executing an equity exchange the leverage is reduced the most making future financial distress less likely. Furthermore, when the firm has many investment opportunities high leverage impedes the capability to seize these opportunities. The firm benefits from a sharp reduction in debt and will be able to generate value when debt is either sharply reduced or exchanged for equity. Both debt-to-debt and debt-to-equity exchanges reduce debt but a debt-to-equity is more likely because the creditors will reduce the firm’s leverage and also benefit from the value created. Thirdly, when the firm is larger, even though it is currently in distress, it is most likely safer than smaller firms in distress and therefore will have easier access to debt. Finally, when a firm is more profitable the chance on debt for cash exchange increases. This makes it more likely that just the debt is restructured through a debt repurchase at a discount.

2.2.1 The Formal Bankruptcy Process vs. Workouts

The bankruptcy procedure is a legal framework for recontracting debt contracts when the involved parties are not able to agree on the debt restructuring of the firm in default out of court. Therefore, within an out-of-court workout the bankruptcy procedure is an alternative and influences the renegotiation. First of all in a court supervised bankruptcy the bargaining power is formally set for both the creditors and the firm. Secondly, the bankruptcy process involves both direct and indirect costs of financial distress making it more expensive than a workout. The direct costs are due to legal and administrative fees which include expenses related to lawyers, accountants, and other professionals. Weiss (1990) estimates that on average these direct costs are around 3% of the firm’s market value, while Altman (1984) estimates the direct costs at 7.5% of the firm’s value. The difference between Weiss (1990) and Altman (1984) are due to different firm’ sizes in the sample, the direct costs are relatively

(8)

8

large for smaller firms. Regarding the direct costs of a workout, corporations are not obligated to disclose these costs. However, corporations do need to estimate these costs in an exchange offer for publically traded debt. Using these estimates Gilson et al. (1990) finds that the workout costs are relatively small in economic terms, on average 0.6% of the book value of the assets. Therefore, workouts seem to be relatively less costly. Hence, based on the direct costs an efficient workout process will be preferable over a formal bankruptcy proceeding. Furthermore, for smaller firms the direct costs are relatively costly and therefore will benefit more from a workout by mitigating these direct costs.

A bankruptcy also has indirect costs, which can be more severe than the direct costs. These costs accrue for example because stakeholders are reluctant to continue doing business because contracts might get renegotiated, or due to the loss of market share (Opler and Titman, 1994), or reputation damage. Altman (1984), Andrade and Kaplan (1998) estimate the indirect costs to be between 10 and 20%, while Davydenko et al. (2012) argue that default costs are even higher, on average 21.7%. While the direct costs are more equal across firms the indirect costs are more heterogeneous. Financial distress costs can be as high as 20% of firm value or as low as 1% (e.g. Bris et al., 2006; Opler and Titman, 1994). The indirect costs depend on firm characteristics while the direct costs are more standard. Therefore the pressure on a successful workout depends on the expected total costs of a formal proceeding which can be substantial and are heterogeneous across firms due to the indirect costs.

2.2.2 The Out-of-Court Renegotiation Mechanism

The relative bargaining power of the creditors and debtors is set by the outside bankruptcy option. The court sets the relative bargaining power in the formal procedure but can’t determine the true value accurately, therefore leaving room for bargaining also under formal procedures (Giammarino, 1989; Bebchuk, 1992; Carapeto, 2005). However, the bankruptcy procedure sets the legal rules and gives bargaining power to each party. Next to this the bankruptcy costs matter as well. The bargaining power of both creditors and shareholders depends on the possible costs of the bankruptcy proceeding, these costs give them an incentive to settle out-of-court. High bankruptcy costs gives a high incentive to settle but to determine who wants to settle, it matters who pays for these costs. Initially shareholders are the residual claim holders and therefore accrue these costs first. The theoretical framework of John et al. (2013) supports this and predicts that the creditors’ payoff increases when the indirect costs are larger because shareholders want to settle and avert the costs. The direct costs are between 3 and 7.5%, depending on the size, showing that formal bankruptcy

(9)

9

procedures are relatively costly for smaller firms making a workout more desirable. The indirect costs are more heterogeneous and can be more substantial. Therefore, the size of the indirect costs depend more on the firm’s characteristics.

Firstly, firms that own a large amount of intangible or firm-specific assets have higher indirect costs. Intangible and firm-specific assets are worth most to the current firm when it continues its business. However, in the bankruptcy procedure forced assets sales are likely to repay more senior debt but liquidating these intangible assets is costly (Gilson et al., 1990; Garlappi et al., 2008). Secondly, the size of the R&D spending indicates larger indirect costs. When a firm has high R&D spending and is in financial distress the firm has to forgo good projects. This can result in a loss of market share because competitors do invest and the firm gets behind on the competition. Therefore, having high R&D expenditures gives shareholders a disadvantage in the negotiation process (Garlappi et al., 2008; Opler and Titman, 1994). Therefore, firms with a high amount of intangible assets and R&D expenses have a weaker bargaining position and want to settle out-of-court giving creditors a stronger bargaining position.

The shareholders are the residual claimants and incur the costs first. However, when the leverage is larger high indirect costs also affect creditors. Therefore, when the leverage is high the dynamics change and high indirect costs give shareholders a better bargaining position. In this case not only the shareholders lose in a bankruptcy but also the creditors. Jostrandt and Sautner (2010) find that German firms with extreme high leverage are more likely to settle. Additionally when the financial distress is more severe it is more likely that creditors incur costs as well. Therefore, higher leverage and more severe financial distress increases the bargaining power of shareholders. Furthermore, when firms want to attract external capital their firm characteristics can generate financial constraints. These financial constraints effect the capital structure decisions because it limits the firm’s options. In order to research the effect of financial constraints a measure is needed. There are several measures, including the Hadlock-Pierce (HP) index, the Whited-Wu (WW) index, and the Kaplan-Zingales (KZ) index. All measures rely on empirical and theoretical assumptions, making the indices proxies and caution is needed when interpreting the results (Farre-Mensa and Ljungqvist, 2013). For a complete overview on how the measures are constructed see appendix A, Table IV.

(10)

10

3. The Workout Sample

The novelty of this paper is the sample and is discussed extensively to get an oversight of the workouts and potential future research. A detailed sample of details on workouts is gathered manually because workout details are not in an existing database. Workouts happen outside the court and firms do not have to file a separate report specifying the details of a workout. This sample is gathered by first identifying workouts and to subsequently find the workout specifics. To identify workouts three search methods were used. First when S&P announces that a firm executed a distressed exchange. Secondly, workouts are identified when the Moody’s database indicates that a firm executed a distressed exchange or when the firm defaulted without going into bankruptcy. Thirdly, the new EDGAR search tool is used. The EDGAR search tool allows a word search through all of the last four years of filings. After identifying the workouts the details about the workouts are gathered through searching the 10-k and 8-10-k filings. It is straightforward what constitutes a ban10-kruptcy but this is not the case for an out-of-court workout. Workouts do not have a clear beginning and ending date. Sometimes the firm restructures the same debt several times or restructures different classes of debt at different points in time. In the simple cases debt got restructured at one moment in time and that is the end date of the workout. In the more complex cases the last restructuring date is used, and if exchanges were executed throughout the year the fiscal year end is the end date. Furthermore, when several consecutive years the firm restructured debt then the last year is the end date. The sample of this research includes workouts in the United States between 1999 and 2013.

3.1 Identifying Workouts

The workouts are identified by using S&P, Moody, and the EDGAR search tool. Firstly, the S&P published an article in which they named all distressed exchanges between 2007 and 2011 (Standard & Poor's, 2011). This time period had more distressed exchanges than all years before combined (Altman and Karlin, 2009). Distressed exchanges are one form of a workout and are necessary when the restructured debt is publicly listed. When the restructured debt is publicly listed then the firm needs to make an offer for the old debt and the firm mostly only commences the deal when a certain threshold accepts. Due to the fact that the firm needs to make a public offer the firm needs to disclose how the debt is restructured and how large the haircut is for the creditors. Hence, this is used to gather more detailed information about workouts. However, some firms do not have to publish 10-k’s or 8-k’s and therefore there is not always detailed information on the firm itself. The S&P article names all

(11)

11

the distressed exchanges between 2009 and May 2011, going through the list 51 distressed exchanges were identified and had information about the workout publically available.

Secondly, the Moody’s sample is used to identify workouts between 1999 and 2009. The sample consists of defaulted firms between 1990 and 2009. First all distressed exchanges that occurred during this time period and which were not already covered by the S&P article are selected. Secondly all firms that defaulted but that did not go into bankruptcy where examined if they had a workout by looking at the firm’s 10-k. This way another 52 workouts are identified. Thirdly, the new EDGAR search tool allows searching the full text of the last four years of filings. The search criteria “debt-to-equity”, “debt-to-debt”, “debt restructuring”, and “financial restructuring”, lead to four additional workouts that were not covered by the previous methods.

3.2 Workout Specifics

First from all identified workouts detailed information is gathered on which debt is restructured and for what it is exchanged. It is straightforward what constitutes a bankruptcy but this is not the case for out-of-court workouts. Workouts do not have a clear beginning and ending date. In the case of distressed exchange there is a clear ending date but sometimes several exchanges take place making it more difficult to specify an end date. In this sample the last exchange is the end date, or when there were several including debt repurchases then the fiscal year end date is used as the end date for all restructurings. Furthermore, to either classify a workout as a debt-to-equity exchange or as a debt restructuring is not straight forward. In this sample when a firm repurchases debt with cash by issuing new equity the exchange is marked as a debt-to-equity exchange. Furthermore, when there are both debt and equity exchanges it is marked as an equity exchange when the current creditors receive a majority in the firm.

Secondly, firm characteristics are gathered in the year of the restructuring and the previous year. From the 107 workouts 92 were covered by COMPUSTAT. Nevertheless some do not cover the right time period because firms stopped filing publicly. For the other firms the SEC filings were used to manually complement the sample but only added limited observations. Furthermore, the industry variables, the industry growth and average industry market-to-book ratio, are computed by searching the entire database of COMPUSTAT. The sum all revenues in the same year is assumed to be the industry revenue and the average of all market-to-book ratios is assumed to be the industry market-to-book ratio (see Appendix A Table IV).

(12)

12 3.3 A Description of the Sample

The sample consists of 107 workouts executed between 1999 and 2013 (see Table I, Appendix A). Consistent with the number of defaults the amount of workouts increases when recessions prolong, as in 2002 and 2003 respectively one and two years after the internet bubble and in 2009 two years after the start of the financial crisis the number of workouts spiked (see Appendix A, Table II). However, the amount of workouts has spiked enormously with 44 workouts in 2009 against 17 workouts in 2002 and 2003 combined. Furthermore, the total dollar amount of debt that is restructured is much higher during the most recent financial crisis. In 2005 the BAPCPA was signed and changed the bankruptcy law, potentially making workouts more attractive. This might help explain the spike in workouts and also the spike in total value of debt restructured. From 1999 till 2005 the total debt restructured was merely USD 6.6 billion while since 2006 the total amount of debt restructured is USD 59.9 billion, roughly nine times as large (see Appendix A, Table II). Furthermore, it seems that overtime the debt-to-debt exchange has become more popular than the debt-to-equity exchange.

On an industry level the most workouts occurred in the manufacturing sector (see Appendix A: Table III). Within industries workouts occur mostly only a few times per industry code (SIC) but the publishing sector and related industry for example saw five workouts. Furthermore, chemicals and allied products also saw five workouts in their sector. This might indicate that some workouts are not solely financially distressed but also economically distressed.

On the firm level (see Appendix A: Table III) the age of the firms start at the IPO date, which is assumed to be when COMPUSTAT first has data available on the stock price. The average age is around 18 years and the firms after 2005 are in general a bit older. The number of observations is limited to 52 due to the fact that not all workouts are executed by publicly listed firms. Regarding the profitability firms that restructure are in general profitable, indicating that they indeed are financially distressed and not economically distressed. Nevertheless there are also 16 unprofitable firms able to restructure. Additionally, it seems that previous to 2005 the firms that executed a workout were more profitable. Regarding the leverage of the year before the restructuring this is on average 72 percent. There is a difference between before and after 2005, before 2005 the average leverage was 22 percent higher. The firms that restructure after 2005 seem to have on average twice as much intangible assets. The size of the firms did significantly increase over time. Previous to 2005 large firms with around USD 595 million in sales restructured their debt while after 2005 the

(13)

13

average firm size shoots up to USD 4.3 billion. The recovery rate remains quite steady around 70 percent. The lagged value of R&D intensity shows that only 5 firms in total invest in R&D at all before 2005. Those firms spend on average 11.5 percent of their sales on R&D before 2005, and 3.8 percent after 2005. The lagged value is used because it gives a better indication what it normally spends on R&D, when it is not in financial distress. The market-to-book ratios are all larger than one indicating that the market value is larger than the book value. However, it seems that the market value is more positive for firms after 2005. This indicates that the market sees the firms as having more potential. Finally, the industry growth is positive for firms prior to 2005 but strongly negative for firms after 2005. This might be biased because of the huge amount of workouts in 2009 when there was a deeper recession. Nevertheless, this indicates that firms also in more difficult industries were able to restructure out-of-court.

Finally, all firms are in financial distress and therefore I also look at the level of financial constraint. The more financially constraint the firm is the more difficult it will be to attract external capital. To measure the financial constraint three measures are used: the KZ index, WW index, and the HP index. All three indices show a similar phenomenon, after 2005 firms going into default and restructure through a workout are less financially constraint (see Table IV).

Financial constraints

N Mean S.D. N Mean S.D. N Mean S.D.

Till 2005 Least constraint 9 0.0 (1.9) 9 -0.36 (0.04) 5 -3.8 (0.6) Middle constraint 9 3.0 (0.7) 9 -0.30 (0.01) 4 -3.2 (0.1) Most constraint 9 10.6 (11.3) 8 -0.24 (0.04) 4 -2.8 (0.2) After 2005 Least constraint 16 -2.1 (4.1) 17 -0.45 (0.05) 13 -4.2 (0.3) Middle constraint 16 2.0 0.70 16 -0.35 (0.02) 13 -3.6 (0.1) Most constraint 16 4.5 (1.3) 16 -0.29 (0.02) 13 -3.2 (0.1) Table IV.

The Financial Constraint of Firms with a Succesful Workout

HP index WW index

KZ index

This table gives an overview about the level of financial constraint firms faced during the workout. The KZ index, WW index, and the HP index are all different measures for financial constraint. For the exact calculations see Appendix A -Variables.

(14)

14

4. Research Method

The research method section is structured as follows. In the first part the hypotheses are formulated regarding the research question. First the hypotheses regarding the choice of debt restructuring are described and subsequently the hypotheses regarding the recovery rate. The second part describes how the variables constructed. Finally, the methodology used is described.

4.1 Hypotheses

The aim of this research is to better understand how out-of-court workouts are renegotiated in financially distressed firms. First, this papers looks at the structure of a successful workout. Secondly this research looks at the bargaining power of creditors in the workout process by looking at the recovery rate. The hypotheses on how workouts are structured are based on previous research regarding the capital structure. To my knowledge there is no previous research done on what increases the likelihood of for example a debt-to-equity workout. The hypotheses regarding the bargaining power are based on previous research that looks at the likelihood of a successful workout (e.g. Gilson et al., 1990) and the theoretical model of John et al. (2013). By gathering detailed information about the workouts and firm specific characteristics this research aims to increase the understanding of the dynamics of the bargaining within workouts.

4.1.1. Method of Restructuring.

There is no previous research done on the choice of debt-to-debt, debt-to-equity or cash exchanges. The novelty of this dataset is that it gathered detailed workout information on how the workout was structured. One of the most important factors in a debt restructuring is how leveraged the firm is. When the leverage is high then to restructure the debt for new debt might not solve the problem. Hence, when the leverage becomes larger the need to restructure more debt becomes imminent. The equity exchange will be mostly used as a last resort for extreme cases. By exchanging the debt for equity the leverage is reduced most.

Hypothesis 1. The likelihood of a debt-to-equity exchange should increase when the leverage increases.

Rajan and Zingales (1995) find four factors affecting the capital structure of firms. First, the tangibility of the assets increases in general the leverage. Hence, when the intangibility of the assets is high then the firm is less capable of taking on high leverage. Therefore, when both leverage ratio and intangibility ratio increase the likelihood of a

(15)

debt-to-15

equity restructuring increases. The firm is less able to sustain the high leverage due to the intangibility of its assets.

Hypothesis 2. The likelihood of a debt-to-equity exchange should increase when the firm has more intangibles, the effect is the strongest when the leverage is higher.

Secondly, Rajan and Zingales (1995) identify investment opportunities as a determinant for leverage. When the firm has many investment opportunities high leverage impedes the capability to seize these opportunities. Therefore, when the leverage is high and the firm has a high market-to-book ratio then the firm will need to forgo more profitable investment opportunities. Hence, the firm would benefit from a reduction in leverage. A debt-to-equity is more likely because in this case the creditors reduce the firm’s debt sharply and will be able to benefit from the value created.

Hypothesis 3. The likelihood of a debt-to-equity exchange should increase when there are more investment opportunities, which is more imminent for firms with a high leverage.

The third and fourth factor Rajan and Zingales (1995) identify are the profitability of the firm and the size of the firm positively correlates with the leverage firms have. More profitable and larger firms can take on more leverage. When the firm is larger it is still safer than a smaller firm, even though it is in financial distress. Therefore, larger firms can more easily attract debt. This is supported by Frank and Gayul (2005) that find that smaller firms more often use equity. Furthermore, more profitable firms are also safer and can attract more debt. Furthermore, when a firm is profitable it can restructure its debt also by repurchasing at a discount, restructuring solely the debt on the balance sheet.

Hypothesis 4. When firms increase in size and/or become more profitable a debt-to-debt exchange becomes more likely.

4.1.2. Creditor’s Bargaining Power.

Previous papers by e.g. Gilson et al. (1990), Frank and Torous (1994), Jostrandt and Sautner (2010) look at determinants that increase the likelihood of a workout but do not look to whom that gives an advantage in the negotiations. This research uses these previously found determinants for a workout to examine if they increase the creditor’s payoff. Additionally, the expected effects are based on the theoretical model of John et al. (2013) and the empirical findings of Garlapi et al. (2008). The most critical component of a debt restructuring is the current leverage, which made the firm financially distressed to begin with. Furthermore, the level of financial constraints shows how limited the firm is in attracting new external financing.

(16)

16

Hypothesis 5. The creditors’ payoff decreases when firms have more leverage and/or are more financially constraint.

The bankruptcy process generates direct and indirect costs that are larger than with a workout (Gilson et al., 1990). To determine who gets the bargaining power it is important to know who pays for these costs if they would be materialized in a formal bankruptcy. Shareholders are the residual claim holders and therefore accrue these costs first. In the theoretical model of John et al. (2013), the payoffs to bondholders increase when the indirect costs of financial distress are larger. The R&D expenditure is a proxy for indirect costs (Opler and Titman, 1994) and Garlapi et al. (2008) finds that R&D expenditure indeed reduces the shareholder’s advantage in the negotiation. When the firm is R&D intensive it most likely is limited in competing with competitors, who can invest in new technology, when it goes into bankruptcy. With limited R&D investments the firm needs to forgo positive NPV projects, diminishing the future value of the firm. When the value of the firm is largely in the future as well, which will be adversely affected when it cannot invest in R&D, shareholders will prefer to settle out-of-court to prevent this loss of value. Other determinants of indirect costs are intangible assets (Gilson et al. 1990). Intangible assets are costly to liquidate in therefore in a forced assets sale the firm loses value by selling it under its value.

Hypothesis 6. The creditors’ payoff increases when the indirect bankruptcy costs are larger.

Furthermore, there are firm characteristics that most likely affect the recovery rate. Firstly, when firms are more profitable the creditors’ payoff increases because the firm has internal liquid funds to pay off the creditors. Secondly, the size of the firm can impact the recovery rate because for smaller firms bankruptcy is relatively costly due to the direct costs. This gives the residual claim holders an incentive to settle out-of-court.

Hypothesis 7. The creditors’ payoff increases when the firm is more profitable and bigger.

4.2 Variables

First the dependent dummy variable is constructed to be one if the workout was a debt-to-equity exchange. A workout is a debt-to-equity exchange when: (1) the firm solely exchanges debt for equity; (2) when a firm repurchases debt with cash by issuing new equity; (3) both debt and equity are exchanged for the existing debt but the received equity stake constitutes a majority. Regarding the investment opportunities the market-to-book ratio is used. However, not all firms are publicly listed and therefore following Farre-Mensa &

(17)

17

Ljungqvist (2014) the industry market-to-book ratio is used instead for those that are private (see Appendix A Table IV). The other independent variables and control variables are constructed using balance sheet items gathered form the COMPUSTAT database (see Appendix A Table IV).

Secondly, the dependent variable for the creditor bargaining power regression is the recovery rate. The denominator is the book value of debt. The market value of debt would be better because it shows how much of a concession creditors are willing to make in order to have a workout. However, the market value of debt is missing for most debt and therefore the book value is used. The nominator is the securities received in return for the old debt. The payment to existing creditors can be in three kinds, cash, debt, and equity. What the creditors recover is the sum of the new debt, received equity value, and cash. The received equity value is calculated by the stake creditors get times the value of the equity in that fiscal year (See Appendix A Table IV). The fact that there are three different methods of payment also hinders the measurability of the recovery rate. An equity stake mostly can be more easily sold while with new debt the creditor might need to hold it longer. Therefore the same amount in received securities might have different value due to the liquidity differences. Both the fact that the book value of debt is used and the difference securities received makes the recovery rate potentially inconsequent. Therefore, the results should be interpreted with care and only serve as an indication for further research. The intangible assets and R&D spending are used as a proxy for the size of the indirect costs (see Appendix A Table IV for the operationalization of these variables). For a proxy of indirect costs of a bankruptcy the fraction of intangible assets and R&D spending are used (see Appendix A Table IV). The other independent variables and control variables are constructed using balance COMPUSTAT data (see Appendix A Table IV).

Finally, three proxies are constructed to measure the financial constraint of a firm. According to Farre-Mensa and Ljunquist (2014) there are three popular indices: The Kaplan-Zingales index, Whited-Wu index, and the Hadlock-Pierce index. Following their paper and convention I construct all three indices (see Appendix A Table IV). All three indices used the coefficients from previous structural models. However, the three differ in what determines how financially constraint a firm is. The KZ index used firm characteristics cash flow, market value, debt, dividends, and cash holdings, each scaled by the total assets. The WW index looks at the cash flow to assets, dummy for paying dividend, leverage, size, sales growth, and industry sales growth. Finally, the HP index simple looks at the size and the age of the firm.

(18)

18

All three are proxies and Farre-Mensa and Ljunquist (2014) shows that all three proxies limitedly indicate how financially constraint a firm is. Hence, results with using these proxies should be interpreted with care.

4.3 Methodology

Two different methods are used. First, for the choice of restructuring, which is a dummy variable, a logit regression is used. Logit regressions are designed for binary dependent variables. The standard estimation method is the maximum likelihood. In general the coefficients show the effect of a unit change in de independent variable on the dependent variable. However, in the logit regression the coefficient shows the change in log probability that the dependent variable is one (Stock & Watson, 2011). The independent variables expecting to affect the likelihood of a debt-to-equity exchange are: Leverage, profitability, investment opportunities (“IO”), size, and intangibility (see equation 1). The second method is a simple OLS regression using cross-sectional data. The OLS regression is used to determine what affects the creditors’ recovery rate. The independent variables expecting to affect the recovery rate are: Intangibility, R&D intensity, leverage, profitability, and the level of financial constraint (“FC”) (see equation 2). The t is the fiscal year in which the workout was executed. (1) (2)

A threat of the logit regression is that the sample is relatively small. Therefore, the coefficients should be interpreted carefully. Furthermore, the way dependent dummy variable is constructed is limited. Workouts are classified as either a debt-to-equity exchange or as a debt restructuring. Therefore, forcing cases where both occur to become one of them. This might make the dummy less consistent. Furthermore, there might be multicolinearity. Variables such as leverage and the intangibility ratio most likely are correlated. This since the fact that Rajan and Zingales (1995) find that more tangible assets means a higher leverage, vis-à-vis more intangible assets will mean a lower leverage. However, the threat of endogeneity is reduced by including the interaction term. The interaction term has an

(19)

19

economic meaning, e.g. higher leverage in combination with intangibles assets means more likely an equity exchange, but also remedies finding insignificant results due to the correlation between variables.

There are also threats regarding the OLS regression. First of all, as mentioned before, the recovery rate is computed in a simple fashion which takes not into consideration illiquidity, market value of debt or adjustment in the interest rate making it prone to measurement errors. Furthermore, the proxies for financial constraint are according to Farre-Mensa and Ljunquist (2014) poor proxies. Additionally, when introducing the leverage as an independent variable in combination with profitability, intangibility, Market-to-Book ratio, and size, this most likely correlates. This due to the fact that Rajan and Zingales (1995) that these factors influence the leverage of the firm.

5. Results

The section is divided in two main parts. The first shows the results of the regressions in section 4.3, equation one and two. First, the workout probability via a debt-to-equity exchange is described. Secondly, the results from the regression looking at the recovery rate are discussed. The variables are described in section 4.2 and a summary can be found in Appendix A Table IVa-IVb.

5.1 The Probability of a Debt-to-Equity Exchange

In section 4.1.1 the hypotheses regarding the method of restructuring a workout are described and are tested in this section. Table V consists of four regressions and provides an overview of the main results of this analysis. The first regression looks solely at the leverage and the following regressions include factors that according to the three hypotheses should affect the restructuring choice. The second regression includes the fraction of intangible assets the firm has. Subsequently, the third regression adds the investment opportunities and the fourth adds the firm characteristics profitability and the firm’s size.

The first hypothesis expects that when the leverage increases the likelihood of a debt-to-equity exchange increases as well. The first regression in Table V shows that the leverage following the workout has a negative sign and the leverage prior to the workout has a positive sign and both variables are strongly significant. The leverage after the workout shows the new capital structure. The negative sign of the height of leverage after the

(20)

20

workout indicates that when the leverage level remains relatively high the restructuring is less likely to have been a debt-to-equity exchange. This is in line with the expectations because a debt-to-equity exchange will reduce the leverage sharply and thus after such an exchange the leverage will be lower. The log odds ratio is -0.0685 and converting to odds by exponentiation results in -6.6 percent, meaning that when the leverage after the restructuring is one per cent higher the likelihood that the exchange was a debt for equity becomes 6.6 percent lower. Furthermore, the positive sign of last year’s leverage indicates that when a firm is more leveraged and needs to restructure, the likelihood of debt-to-equity exchange increases. The Leveraget Leveraget-1 Profitabilityt-1 Firm sizet Intangibilityt Leveraget-1×Intangibilityt IOt Leveraget-1×IOt Constant Likelihood Ratio Number of workouts 78 78 77 (0.70) (1.5102) (-0.03) -36.6337 -35.5464 -36.2986 (1.41) 0.6311 1.4489 -0.054 (-1.24) 0.0223 -1.9854 (-1.80) 0.0004* (1.56) (0.52) -0.0601* (-1.69) (1.67) (-0.67) 0.0006 0.1203 -0.0568* 0.0394*** 0.0303 (2.53) (1.40) (2.66) 0.0395** -0.0164 -37.9092 79 -0.2085 (-0.34) -0.0645*** -0.0678*** -0.0667*** (-3.87) (-3.43) (-3.58) 0.0517*** (3.58) -0.0685*** (-3.84) Table V.

The Logit Regression of the Method of Restructuring

(1) (2) (3) (4)

This tables shows the results of the logit regression. The regression looks at the likelihood of a debt-to-equity exchange. All four hypotheses are tested in column (1) till column (4). In column (1) only the leverage influences the method choice. Gilson et al. (1990) found that intangible assets increased the likelihood of a workout and in the column (2) the effect of intangible assets on the method of restructuring is examined. Rajan and Zingales (1995) find that in general firms with more investment opportunities ("IO") have less leverage and in column (3) the effect of this on the method of restructuring is examined. Finally, the factors firm size and profitability, which affect the leverage according to Rajan and Zingales (1995), are examined in column (4). The coefficients are the without parentheses, the robust z-statistic are in parentheses. The statistical significance is denoted by *, **, ***, which indicate significance at 10%, 5%, and 1%, respectively.

(21)

21

log odds ratio is 0.0517 meaning that an increase of one per cent in leverage prior to the restructuring increases the likelihood of a debt for equity exchange with 5.3 percent.

The second hypothesis expects that when the firm has more intangibles and is more leveraged a shaper leverage reduction is needed and therefore increases the likelihood of a debt-to-equity exchange. The second regression in Table V shows that the leverage coefficients remain relatively stable and strongly significant. The intangibility coefficients are less straightforward to interpret. Furthermore, the intangibility coefficient is weakly significant and the interaction term is not significant. The intangibility coefficient is -0.0568, and the interaction coefficient is 0.0006. This means that when the firm has for example a 50 percent leverage a one per cent increase in the intangibility ratio decreases the likelihood of a debt-to-equity exchange with 2.5 percent (-5.5 percent plus ). However, the hypothesis expects that the combination of high leverage and intangibility increases the likelihood of debt for equity exchange. When the leverage of the firm is 100 percent then the likelihood of a debt for equity actually increases with 0.7 percent (-5.5 percent plus

_{). When firms have more than 95 percent (0.0568/0.0006) leverage then the}

likelihood of a debt to equity exchange increases. On average the leverage prior to the workout is 72 percent (Table III, Appendix A) and therefore intangible assets in most cases will not benefit a debt-to-equity exchange. Nevertheless these findings should be interpret with care because they are weakly significant or not significant at all. An explanation for this might be the limited workouts available for the regression. Only 78 observations remain in the sample.

The third hypothesis expects that the investment opportunities will increase the likelihood of a debt for equity exchange. However, the third regression of Table V indicates that both the coefficient of the investment opportunities, as well as the interaction term with leverage is insignificant. Nevertheless the signs have the expected value, when the investment opportunities and the leverage are low a debt for equity exchange is less likely but when the leverage increases in combination with the investment opportunities the likelihood of debt-to-equity increases as well. A potential explanation is that the variable investment opportunities is constructed poorly. When the firm was private the industry market-to-book ratio was used, which might be a poor indicator of the firm’s market-to-book ratio. Nevertheless, this analysis cannot confirm that the market-to-book ratio matters in the workout process.

(22)

22

Finally, the last hypothesis looks at the effect of size and the profitability and their effect on the method of restructuring. In this regression the investment opportunities variables are left out due to their lack of explanatory power. In this regression the leverage coefficient remains relatively stable compared to the first two regressions. The intangibility variables remain also quite stable but are both weakly significant. Finally, the size and the profitability of the firm are both not significant. The size also has an unexpected positive sign indicating that larger firms more often have a debt for equity exchange. The sign of the profitability is negative and therefore more profitable firms are less often taken over by their creditors. Nevertheless the last hypothesis is not confirmed. Again this can be due to the limited amount of workouts but the firm size also has a different than expected sign.

In terms of the method of restructuring this analysis finds that the leverage before and after the restructuring, as one expects in financial distress and the resolution of it, plays the main role in choosing for either a debt-to-equity exchange or a debt restructuring. Furthermore, highly leveraged firms with intangible assets seem to be more likely to have a debt-to-equity exchange while firms with low leverage but a sizeable amount of intangibles are less likely to have a debt-to-equity exchange. Nevertheless this should be interpret with care because the intangible variables are weakly significant in regression (2) and (4) of Table IV, while in (2) the interaction term is even insignificant. The other hypotheses could not be confirmed, which might be due to a lack of observations, troublesome proxy for investment opportunities, or they do not play a role at all in the renegotiation process.

5.2 The Factors Determining the Recovery Rate of Creditors

In section 4.1.2 the hypotheses regarding the recovery rate in a workout are described. Table VI consists of four regressions and provides an overview of the main results of this analysis. The first regression looks solely at the leverage and the financial constraint of the firm while the following regressions include factors expecting to affect the recovery rate. The second and the third regression include the proxy variables for indirect cost intangibility and R&D expenses. Subsequently, the fourth regression adds the firm characteristics profitability and the firm’s size. The Kaplan-Zingales index is used in the results section as the measure for financial constraint. The other two indices are in Appendix B Table VIa-VIb.

The fifth hypothesis expects a lower recovery rate when firms have more leverage or are more financial constraint. The first regression in Table VI shows that both the leverage prior to the workout and after the workout negatively impacts the recovery rate. When the

(23)

23

leverage prior to the workout is one per cent higher, then the recovery rate decreases with 6.3 percent. This is expected because high leverage most likely means that the financial distress is larger and the firm needs a steeper discount to be able to survive. Furthermore, when the leverage after the workout remains high this means that the haircut the creditors took was larger. A possible explanation is that when the most distressed firms with the highest leverage restructure their reduction will be large but most likely also the remaining debt. Finally, when the firm is less constraint then the recovery rate was positively affected. All signs are as expected and significant.

The sixth hypothesis expects that when the indirect costs of a bankruptcy is higher the recovery rate increases. The second regression contains the indirect cost proxy R&D intensity and the third regression contains the intangibility variable as the indirect cost proxy. The coefficients of the leverage variable and financial constraint index remain relatively the same. The R&D variables shows that the effect depends on how leveraged the firms is and are significant. When the leverage is for example 50 percent then a 1 per cent increase means a 2.9 percent lower recovery rate (-9.4 percent plus 50×0.13 percent). However, when the leverage as high as 100 percent then the recovery rate increases with 3.6 percent. Hence, it seems that the bargaining power for creditors is larger when the firm is more leveraged and relies on R&D spending. The third regression looks at indirect cost proxy intangibility. Again the coefficients of the leverage and the Kaplan-Zingales index remain stable. However, the coefficients of the intangible assets are insignificant. The signs of coefficient are similar to the R&D coefficients but the leverage needs to become extremely high 270 percent (0.081/0.0003) in order to give the creditors an advantage. Concluding, the hypothesis is partially confirmed, R&D expenses seem to give creditors an advantage when the firm has more than 72 percent leverage (0.0941/0.0013). The other proxy intangibility is insignificant and therefore does not support the hypothesis. The final hypothesis expects that the firms’ size and the profitability affect the recovery rate. The coefficients do have the expected sign in regression (4) of table VI, larger firms and more profitable firms have a larger recovery rate but the coefficients are not significant.

(24)

24

Concluding, in the recovery rate analysis the most important and constant factors are the leverage variables and the level of financial constraint. This seems logical because the leverage is the main issue when the firm restructures. Furthermore, the possibility to attract external capital can help repay creditors. Secondly, only the indirect bankruptcy cost proxy R&D expenses seems to affect the recovery rate, as expected, and is significant. Nevertheless, these results should be interpreted with care. First of all the recovery rate variable is difficult to correctly compute and in this research a too simple calculation is which does not take into Leveraget

Leveraget-1

Kaplan-Zingales Indext

R&D Intensityt-1

Leveraget-1×R&D Intensityt-1

Intangibilityt Leveraget-1×Intangibilityt Profitabilityt-1 Firm sizet Constant Number of workouts R² 0.73 0.7312 0.7307 72 71 72 (2.76) (1.72) (0.43) 6.2725*** 6.2392*** 4.99* (0.14) 0.1791 0.0003 (0.57) 0.0156 (1.26) -0.0081 (-0.15) -0.0723 (-1.73) (-0.73) 0.0013* 0.0011 (-2.49) (-2.26) 1.5854*** 1.5910*** 1.5828*** (3.34) (3.31) (3.28) -0.0497* -0.0477* (-1.92) (-2.03) (-1.9) -0.0697** -0.0721* -0.0689** -0.0465* (-2.28) -0.0941* (1.75) (3.24) Table VI.

The Recovery Rate Regression.

(1) (2) (3) (4)

This tables shows the results of the recovery rate regression. The regression looks at what determines the recovery rate for creditors and all three hypotheses are tested in columns (1) till (4). The reason why a firm is in financial distress, the leverage and financial constraint to attract new capital, are examined in column (1). To test the effect of indirect costs on the recovery rate columns (2) and (3) separately test two proxies, intangible assets and R&D expenses. Both include an interaction term with leverage because when the firms are more leveraged the indirect costs are larges as well as it compensate for the strong correlation between the dependent variables. Finally, the hypothesis regarding firm characteristics is examined in the column (4) where the profitability and the firm size is added to the regression. The coefficients are without parentheses, the robust t-statistic are in parentheses. The statistical significance is denoted by *, **, ***, which indicate significance at 10%, 5%, and 1%, respectively.

0.7282 -0.0493** (-2.06) -0.0631** (-2.22) 1.58254*** (3.37) 5.9899*** (3.3) 72

(25)

25

account for example liquidity, change in maturity, and/or change in interest. Additionally, there is no good working measure for how financially constraint a firm is (Farre-Mensa and Ljunquist, 2014). I also compute the same regressions with the Hadlock-Pierce index and Whited-Wu index (see Appendix B Table VIa-VIb). As shown in Appendix B, all regression using these indices turns out to be insignificant. A potential explanation is that Farre-Mensa and Ljunquist (2014) find that in contrast to the Kaplan-Zingales index the Hadlock-Pierce index and Whited-Wu index label younger, smaller, and fast growing firms as financial constraint. As shown by regression four in Table V with firm size and profitability, which in general resemble the difference between old and young firms, is not significant. Hence, it makes sense that the indices identifying these traits are not significant. Therefore, in this research the Kaplan-Zingales index might be the best but is still questionable. Furthermore, the limited number of observations with enough data might impede finding significant results.

6. Robustness Checks

To test the robustness of the results additional tests are executed. The sample so far has been used as a cross-sectional dataset. However, due to the BAPCPA and other economic factors there potentially are also external effects. To look if the BAPCPA changed the dynamics the regressions are redone on a restricted sample, workouts after 2005. Furthermore, the macro economic situation is measured by the S&P 500 stock return, and the U.S. GDP growth. 6.1 Macro Variables

Every regression encounters the possibility of omitted variables and in the case of workouts there might be omitted variables regarding the economic situation. Besides firm characteristics economic variables might also influence workouts. To test for this the regressions for both the recovery rate and the method of restructuring are examined by using the S&P 500 return and the GDP growth for the economic situation. The S&P 500 shows the stock market sentiment and the GDP growth the underlying economic strength (see Table VII for the used values).

Table VIII in Appendix B shows the results from including the macro economic variables. The first column unsurprisingly shows that the S&P 500 annual return affects the choice to exchange debt for equity. When the S&P 500 performs one per cent better then it is less likely that there will be a debt for equity exchange. However, the GDP growth is insignificant. The second column adds the S&P stock return to the first regression of Table IV

(26)

26

in section 5.1. The leverage coefficients are relatively unaffected and remain significant. Therefore, it seems that leverage coefficient is indeed robust and the macroeconomic variable only complements the regression. However, adding the S&P 500 return to the second regression of Table IV does not affect the coefficients much but does makes the intangibility variable insignificant. Therefore, the previous weak evidence found for the effect of intangible assets on the debt for equity exchange does not hold and seems not robust.

The last three columns of Table VIII in Appendix B are regarding the recovery rate. The fourth column shows the recovery rate regressed on the macroeconomic variables. It seems that in this case they do not matter. However, the GDP growth is close to being weakly significant and is used for the other regressions. Combining the GDP growth with the first regression of Table VI of section 5.2 does not affect its coefficients and GDP growth remains insignificant. Finally, adding GDP growth to the second regression of Table VI of section 5.2 makes the R&D variables significant but the GDP growth remains insignificant. Therefore, it seems that for the recovery rate the macroeconomic variables GDP growth and S&P 500 return do not help predict its outcome.

6.2 The BAPCPA

The BAPCPA was signed in 2005 and changed the rules of the bankruptcy code giving more power to creditors. The bankruptcy procedure is the most important outside option during the renegotiations. The decreased bargaining power of managers might stimulate to settle out-of-court. Hence, firms that otherwise would filed for a bankruptcy might now work out their debt restructuring out-of-court. This change might also affect the results from the regression making it more difficult to find significant results. To determine if the dynamics has changed the sample is restricted to solely included restructurings that were executed after the signing of the BAPCPA in 2005. However, between 2005 and now there has been a severe financial crisis. Therefore, changes before and after 2005 might reflect that as well.

In the column (1) - (3) from Table IX in Appendix B the results from the regressions on the method of restructuring are presented. Previous results showed that the leverage and the intangible assets determined the choice of securities exchanged. In regression (1) and (2) leverage no longer significantly predicts a restructuring through equity. Furthermore, the intangible assets coefficient in regression (2) is also no longer significant. However, regression (3) shows a regression on the leverage, the Rajan & Zingales (1995) factors, firm characteristics, and S&P returns, which are all significant. However, the interaction variables

(27)

27

are no longer significant and some signs changed. A higher leverage in the main regression increases the likelihood of a restructuring through equity. In the period after 2005 a higher leverage is either insignificant, regressions (1) and (2), or decreases the likelihood of a restructuring through equity, regression (3). A potential explanation is the changed dynamics due to the BAPCPA, and the firms that workout their debt. As show in the descriptive section 3 the firms after 2005 are different. The firms in general have less leverage and are much larger. Jostrandt and Sautner (2010) find that firms with more leverage are more likely to restructure out-of-court. It seems that before BACPCA only the extremely distressed firms had a workout but after the regulatory change they restructure also at lower levels. Furthermore, the observations in the regression are limited and might be biased by 2009, in which year the most workouts occurred, and creditors had other reasons to settle as well.

In the column (4) - (7) from Table IX in Appendix B the results from the regressions on the recovery rate are given. Here as well the results change. The leverage alone no longer significantly predicts the recovery rate. However, in regression (6) and (7) as expected a higher leverage still leads to a lower recovery rate for the creditors. Subsequently, in the previous results the intangible assets did not affect the recovery rate. In regression (5) – (7) intangible assets indeed affect the recovery rate, which was expected beforehand. When a firm has a leverage of roughly 75 percent then intangible assets start increasing the recovery rate of creditors. This is as expected according to the theory, the combination of high a leverage and intangible assets means more potential indirect costs in a bankruptcy due to forced asset sales. In the previous results the R&D spending influenced the recovery rate. However, here this is no longer significant. This is most likely due to the fact that already very few firms have R&D spending and limiting the sample only 13 firms remain. Hence, it seems non robust but this might be due to a lack of observations. The previous result did not find other significant results. Additionally, regression (7) looks at the other factors that were previously insignificant. The R&D without interaction term, investment opportunities, and profitability does seem to matter. When the R&D spending increases the recovery rate becomes lower. This indicates that R&D spending unexpectedly reduces the creditors’ bargaining power. An alternative explanation is that firms with high R&D spending will need to restructure more when they become financially distressed and that it has nothing to do with the bargaining power. Then the investment opportunities gives the creditors a higher recovery rate when the leverage is low. This is unexpected because when the leverage is high the firm will need to forgo more opportunities, giving the creditors a stronger bargaining position. A

(28)

28

potential explanation is that the investment opportunities is not measured correctly, for example the market opportunities are used in many cases. Finally, as one would expect more profitable firms indeed have a higher recovery rate.

7. Discussion and Conclusion

The total number of defaults during recessions seems to increase and during the most recent financial crisis there were more defaults and distressed exchanges than ever (Altman and Karlin, 2008). The U.S. saw a surge in these distressed exchanges, which seems to indicate that out-of-court workouts are becoming more popular and important. This paper tried to open the black box of out-of-court workouts. When negotiating out-of-court two aspects matter: what kind of securities do the creditors receive and how much do they get. The novelty of this paper is the gathered sample. There is to my knowledge no database that has information about out-of-court restructurings. This paper identified 107 workouts and gathered detailed information about the debt restructuring by going through the 10-k and 8-k filings. Hence, the sample contains information about the kind of securities that are exchanged in a workout and furthermore how much of those securities the creditors get.

The sample is analyzed by using two main regressions. First, this paper tries to determine what securities are offered in successful exchanged, a logit regression was used determine the likelihood of an equity offer. Therefore, for each workout it is determined if it was either a debt or an equity exchange offer. The most important factor, the current leverage, in a financially distressed firm is also the most important in determining what kind of securities are offered. When a firm has more leveraged prior to the restructuring it becomes more likely that the creditors receive equity. Furthermore, when the firm’s is less leveraged after, then the firm is more likely to have had a debt for equity exchange. This indicates that the debt for equity method is used for high leveraged firms that need to restructure much. Furthermore, the fraction of intangible assets compared to the total assets seems to affect the likelihood of a debt-for-equity exchange. When the firm is highly leveraged, which is at 95 percent, an increases in the intangibility ration results into an increase in the likelihood of a debt-to-equity exchange. When firms have less than 95 percent leverage then it becomes less likely that it will be a debt-for-equity exchange. The other expected determinants, the investment opportunity, profitability, and size do not seem to affect the choice for equity securities. However, when the sample is restricted to workouts after the BAPCPA was signed

Out-of-court renegotiations of financial contracts in financially distressed firms : inside the black box : an empirical analysis in the determinants of out-of-court workouts in the United States