1
Master Thesis
Stanislaw Cieslak
Renegotiation of financing agreements by distressed
entities.
Abstract
Using a sample for American public companies from 2004 until 2012, I examine factors determining the choice of distressed companies between in court bankruptcy and out of the court debt renegotiation. In accordance with my research results, I find strong confirmation of negative impact of executive change on probability of renegotiation. Moreover, I confirm that higher concentration of ownership has significant, although small impact on probability of out of the court workout. However, I cannot confirm impact of liquidity on probability of renegotiation. The findings are of use for distressed companies owners, financial policy regulators and banking system institutions.
Student number: 10397493
Supervisor: Dr. J.E. Ligterink
University of Amsterdam
2
Statement of Originality
This document is written by Student Stanislaw Cieslak who declares to take full responsibility for the content of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Introduction
The worldwide financial crisis of 2007 substantially increased the number of distressed companies facing bankruptcy. In legal terms, this leads to a choice between either in court bankruptcy (through Chapter 11) or out of the court debt renegotiation. The decision between the two entails serious consequences regarding the shape of restructuring or bankruptcy process and its costs. Besides direct costs, such as legal, accountant and other professional services fees associated with Chapter 11 that tend to consume around 3,5% of company value, there are also indirect costs, such as loss of credibility or loss of key employees. On the other hand, there are multiple factors important for the distressed company owners that result in selection of Chapter 11, like higher credibility and readability of the process.
The topic of renegotiation of financing agreements by distressed entities has been already researched in numerous academic papers. The most investigated factors impacting the choice between the two described bankruptcy processes include information asymmetry problems among all the stakeholders, risk shifting and control issue concerning company’s management during reorganization process, coordination problems during the bankruptcy process and insufficient economic motives to bargain for the debtholders.. I describe research papers regarding these topics in the literature review section.
In empirical part of my thesis I concentrate on companies based in the United States, although legal foundations allowing debtor to continue supervised restructuring after bankruptcy are prevalent present in many legislatures.
In my hypotheses I seek to supplement existing findings regarding a choice between Chapter 11 bankruptcy and out of the court workout. In my first hypothesis I investigate how a change of incumbent executives impact the chances of out of the court distress resolution. In my second hypothesis I look at the impact of ownership concentration at probability of workout. In my third hypothesis I try to contribute to discussion about importance of liquidity in distress resolution.
My research sample consists of companies that between 2004 and 2012 had defaulted on their debt obligations. Among 189 companies in my sample, 147 companies have subsequently renegotiated financing and 42 field for Chapter 11 bankruptcy.
I find strong confirmation of negative impact of executive change on probability of renegotiation. Moreover, I confirm that higher concentration of ownership has significant,
4 although small impact on probability of out of the court workout. However, I cannot confirm impact of liquidity on probability of renegotiation.
I begin with Section 1 describing evidence regarding distress debt renegotiation described in the existing research. This is followed by derivation of research hypotheses and description of methodology in Section 2. In Section 3 I present descriptive statistics and construction of my data sample. Results can be found in Section 4, followed by concluding remarks in Section 5.
5
1. Literature review
In US, debtors willing to continue operations after default unusually face the choice between entering bankruptcy under Chapter 11 Code of Bankruptcy or recontracting with debtholders out of the court. Chapter 11 bankruptcy is associated with significant direct costs, such as legal, accountant and other professional services fees that tend to consume around 3,5% of company value. At the same time there are also indirect costs of formal bankruptcy, such as loss of credibility or loss of key employees. Depending on industry indirect costs consume further 9% to 15% of company value. (Wruck (1990)). At the same time workout cost was fund to be negligible, with mean cost found by Gilson et al. (1990) to be only 0.65% of book value of assets.
In this context resolution of financial distress out of the court was found to be significantly more efficient, leaving more value for stakeholders. Franks and Torous (1994) document higher recovery rates for creditors under out of the court restructuring (80%) in comparison to Chapter 11 proceedings (51%). Similarly, position of equity holders in out of the court renegotiation was found to be preferable compared to a formal bankruptcy.
Early research in subject of corporate distress looks at costs of financial distress solely from perspective of company, assuming high or perfect efficiency of negotiations between stakeholders. Theoretical foundations of this assumption can be found in classical macroeconomics that is prevalently based on market efficiency assumption and effective rights allocation, where agents are expected to resolve the conflicts as to avoid any deadweight loss. This is reflected for example in the Coase Theorem (1960) that would imply that regardless allocation of rights, default should be resolved in effective way leading to Pareto optimum outcome.
Similarly, on grounds of Modiligani-Miller Capital Structure Irrelevance Theory (1958), company should be able to ex-ante accommodate effects of ex-post decisions between claimholders without any cost (Fama and Miller (1972), Stiglitz (1974), Haugen and Senbet (1978, 1988), making in court bankruptcy nonexistent due to higher costs. Even in case of liquidation company should cease to operate and distribute its assets between claimholders out of the court.
Yet in 2009 there were over 61 thousand formal bankruptcies in United States alone. Assuming that not every company choosing Chapter 11 instead of a workout is acting inefficiently, there must be rational factors impeding restructuring out of the court.
6 Existing literature in this subject points out not only at explicit costs of bankruptcy, but also at interrelation between creditors, shareholders and management, as well as number of other factors to play a pivotal role.
In next paragraphs I present these findings divided into three major groups; liquidity and performance, inefficient bargaining and institutional biases.
1.1. Liquidity and performance
The opinion about company’s future prospects, and in result the opinion about its ability to generate cashflows sufficient to service the debt obligations can be expected to be more heterogeneous for companies with lower liquidity and poorer performance. Since workout usually requires unanimous creditor consent, it is more likely that companies requiring longer and more costly turnaround will face more difficulties in convincing all debtholders to reorganization and will be more likely to file for Chapter 11 bankruptcy that allows for automatic creditor protection. ((Bulow, Jeremy and Shoven, (1978)).
Formal bankruptcy also allows companies to obtain additional liquidity by raising debtor in possession financing and by doing this secure resources needed in turnaround process. Debtor in possession financing is senior to all existing liabilities, allowing accessible liquidity for companies with poor credit rating that otherwise wouldn’t be able to obtain new financing. At the same time firms with better prospects and higher liquidity can be expected to save on both direct and indirect bankruptcy costs, especially if they operate in high market to book value industries generally characterized by higher going concern value that might be partially lost in formal bankruptcy proceeding.
However, this point was only partially supported by Asquish (1994), who finds no evidence of interrelation between financial performance in distress and probability of bankruptcy, attributing course of distress resolution mainly to capital structure and ability to obtain additional liquidity via asset sales.
1.2. Inefficient bargaining
Inefficient bargaining is a fundamental limitation of Coase Theorem, as well as other theories assuming perfect efficiency of negotiation between stakeholders. In imperfect world bargaining is costly what imply that Coase Theorem no longer holds. In this paragraph I
7 present the most important factors contributing to cost of bargaining and therefore impacting decision about renegotiation.
1.2.1. Information asymmetry
Information asymmetry, firstly described by Akerlof (1970), is one of the most important costs of bargaining. Formal bankruptcy proceeding under Chapter 11 Code of Bankruptcy in US, as well as similar regulations in many other countries, requires debtor to disclose a range of financial information under Disclosure Statement that needs to be accepted by court, before reorganization can take place. In contrast, information available to creditors during out of the court renegotiation varies, but can be generally expected to be far less comprehensive. For that reason claimholders are subject to risk of misrepresentation of relevant information by company insiders.
Among the most important reasons for company insiders to take advantage of information asymmetry it is important to notice their duty to shareholders. Since company executives are ultimately liable to shareholders they can be expected to represent their interest over interest of other stakeholders. Generally situation of shareholders under workout should be better in comparison to bankruptcy, what was documented by many researchers, such as Asquith et al. (1994), Betker (1998), Gilson et al. (1990) and Jorstrandt and Sautner (2009). It provides a possible reason for executives to use information asymmetry in order to protect shareholders which are residual claimants in bankruptcy.
As shown by Giammarino (1989) and Heinkel and Zechner (1993), company may also use private information to exaggerate scope of financial difficulties. In situations when cost of distress is big enough to impact the value of debt, company can expropriate debtholders by convincing them to partial debt forgiveness not required in order to avoid liquidation.
Regardless of the reason for a company to take advantage of private information, the formal bankruptcy offers creditors significant reduction of information asymmetry and thus can contribute towards a decision not to agree on renegotiation.
1.2.2. Risk shifting and control
In presence of asymmetric information, described in previous paragraphs, there is possibility of risk shifting by company’s management during reorganization process. Expecting increased probability of liquidation, managers may want to gamble at expense of debtholders by engaging a company in potentially profitable, but very risky projects.
8 As it was described by Ross (1977) and Gilson (1990) executives of bankrupt companies suffer significant personal costs. Annual turnover rate exceeds 50% in top management of distress companies, compared to 19% turnover for bottom 5% performing non-distressed corporations, and 12% for random sample of listed firms. What is more, none of 73 managers researched by Gilson held an executive position in listed company for at least three years after their departure, what suggest that downside risk is limited, whereas gain from risk shifting may be substantial.
Problem of risk shifting might be further amplified by diminution of control mechanisms during financial distress. Jorstrandt and Sautner (2009) document substantial decrease in ownership concentration of German companies in distress, what can potentially lead to increase in free-riders problem and diminution of control.
In contrast, formal bankruptcy can be expected to greatly decrease the risk shifting. Under Chapter 11 Code of Bankruptcy, management is required to implement previously agreed reorganization plan and acts under scrutiny of bankruptcy court. For that reason, expecting potential risk shifting, debtholders may be inclined towards formal bankruptcy.
1.2.3. Coordination problems
Assuming that a company has valuable intangible assets that should be protected from formal bankruptcy and stakeholders can effectively mitigate risks of information asymmetry, the interrelation between groups of stakeholders can become another obstacle in workout.
In contrast to out of the court renegotiation, formal bankruptcy facilitates coordination of lenders, especially when structure of claimholders is complex. It requires much less unanimity, allowing for involuntary imposition of reorganization plan against some classes of creditors.
Chapter 11 bankruptcy offers a company automatic protection from claimants until it is waived by the court decision and thus allows to dedicate more management’s focus on restructuring efforts. Company is required to prepare restructuring plan that needs to be accepted by only 2/3 creditors in each impaired claim class. Under specified circumstances, bankruptcy court can also force reorganization even without required 2/3 creditors consent (cram down).
Above described provisions are particularly important in context of evidence documenting impact of coordination problems resulting from different term structure, claim classes,
9 objectives and limitations of lenders on renegotiation probability. (Smith and Warner (1979) and Leftwich (1981)).
Intuitively private debt, as well as debt owed to banks was found to be more easily renegotiated in comparison to public debt (Hege (1991), Smith and Warner (1979)). More sophisticated lenders, such as banks, have more experience and resources to manage distress of borrower and hence it is easier for them to utilize higher company value during out of the court negotiations. In countries with strong banking system, such as Germany, they can even create ex-ante contractual coordination solutions such as bank-pools, as described by Jostarnd and Sautner (2009).
Higher amount of lenders leads also to what was described by Gilson (1990) as hold out problem, when creditors expect from each other to make concessions to a company in order to foster restructuring efforts, although no one creditor is willing to make concessions first. Eventually no concessions are made even though particular debt holders were willing to forgive part of their claims.
1.2.4. Insufficient economic motives to bargain
Ultimately, decision about method of distress resolution can be looked at from fundamental cost-benefit perspective. As outlined, out of the court renegotiation is unusually preferred by shareholders and management, what leaves a decision in hands of debtholders that carry the cost of bargaining. Debtholders can be expected to considerate renegotiation preferable to bankruptcy if it provides them with a perspective of higher value under workout. Franks and Torous (1993) describe higher recovery rates for creditors under out of the court distressed exchanges (80%) in comparison to Chapter 11 bankruptcy (51%) in US.
Although it suggests that overall out of the court solutions preserve more value that can be attributed to debtholders, highly collateralized debt can make creditors reluctant to undertake any risks. Asquith (1994) describes evidence of impact of collateralization and capital structure on firm’s ability to renegotiate financing agreements. Author finds that collateralization leads to poorer outlook on the workout. This is closely connected to coordination problem described in section 2.1.3, since even with large group of creditors willing to negotiate highly collateralized debtholders can block renegotiation efforts all together. Knowing that in Chapter 11 bankruptcy debtor can undermine the value of collateral and obtain more senior debtor-in-possession financing, fully protected lenders will favor liquidation over renegotiation, and renegotiation over Chapter 11 bankruptcy.
10 Seemingly, low going concern value measured by Tobin’s Q was found to impair possibility of workout. Low going concern value, often resulting from lengthy distress or highly tangible asset based industry leaves little intangible value to be protected from bankruptcy, at the same time limiting future prospects and ability to incentivize debtholders. (Jorstarndt and Sautner (2009), Gilson, John, Lang (1990), Jensen (1989))
1.3. Institutional biases against workout
Apart from inefficient bargaining, there is number of exogenous institutional factors that may affect the choice between workouts and bankruptcies. Although factors described below do not vary between companies and change relatively slowly, it is important to notice them in order to provide broader picture of the subject. Subsequently I shift a discussion to somehow less research topic of institutional biases connected with banking system, especially during financial crisis.
1.3.1. Legal environment
The most evident exogenous factor is the legal environment that governs formal bankruptcy and greatly impacts costs and benefits associated with choice between in and out of the court default resolutions. It is however important to notice the impact of regulations not directly connected with bankruptcy. As described by Jostrandt, Sautner (2009), there is impact of more creditor friendly law on lower probability of workout.
Authors compare very creditor friendly legislature of Germany to more shareholder friendly legislature of other countries and find relatively higher bankruptcy rates when position of debtholders is strong. The finding was confirmed using US data from before and after 2005 legislature change that strengthened creditor’s rights1
in US.
Authors also point out on impact of specific regulations imposing a legal requirement for some actions to be taken during distress. For example, under German legislation managers are legally required to file for bankruptcy when company is not able to cover its due payments as well as when liabilities value exceed value of assets. This may not give enough time for a management to renegotiate debt obligations and thus force a company into bankruptcy, what gives a potential explanation for lower renegotiation rates in this country.
1 The Bankruptcy Abuse Prevention and Consumer Protection Act of 2005 implemented, among others, the
11 Another factor that is taken into consideration is tax treatment of default. The US legislator framework may bias decision makers towards bankruptcy by offering creditors ability to utilize tax benefits on income resulting from forgiven liability, as well as preferable loss carryforward regulations, not otherwise available under out of the court solutions.
These were however proven to be of secondary importance since in sample used by Betker (1995), no single company undergoing Chapter 11 bankruptcy would have lost its net operating loss benefit, should it choose to restructure out of the court.
1.3.2. Banking system and credit crunch
There are reasons to believe that condition of banking system can significantly attribute to ability of distressed firms to renegotiate, especially during financial turmoil. Although I did not managed to identify research directly addressing this issue, the work of Ivashina and Schaffstein (2009) and Brunnermeier (2009) focused on credit availability and liquidity of banking system during financial crisis provides findings applicable in discussion about renegotiation processes after year 2007.
Ivishina and Schaffstein (2009) show that after approximately 30% spike in number of drawings in second quarter of 2007 total amount of new corporate loans decreased by over 75% until fourth quarter of 2008, with over 90% decrease in non-investment grade instruments. Authors also notice that banks with lower deposit to capital ratio, as well as those co-syndicating with Lehman Brothers cut lending to a greater extend.
Roberts and Sufi (2008) describe high dependence of renegotiations of not distressed claims (claims renegotiated for reasons other than default) on cost of capital. This interrelation can be believed to hold also with distressed claims, especially in times of substantial increase in cost of capital. In subject of renegotiation it suggests that ability to ensure additional liquidity sources must have been significantly limited, making creditors more willing to liquidate the assets regardless of potential losses and bargaining considerations.
2. Hypotheses and methodology
In following paragraph I derive my research hypotheses in context of existing literature. I also present methodology used in order to test hypotheses, which is followed by description of data sample in the next chapter.
12
2.1. Hypothesis I
Change of incumbent executives reduces probability of out of the court workout.
As I explain in Section 1.2.2, incumbent executives of distressed companies have vital interest in keeping their positions due to unfavorable career perspectives they would face if removed from office. Assuming presence of problems resulting from agency theory in distressed companies (Jensen and Meckling (1976)), it can be assumed that managers are at least to some extend self-serving, preferring solutions granting their job security, even if these are sub-optimal from corporate perspective.
Factually, change in the executive positions during financial distress in corporations in very prevalent, what was shown by number of researchers, such as Gilson (1990). Alike to my hypothesis in sample investigated by Gilson (1990) there is higher executive turnover in companies restructuring under Chapter 11 in comparison to out of the court workout. (Gilson – 60% vs 71%; Hotchkiss, Moorandi (1997) – 34,2% vs 81,2%)
Gilson (1989) suggests that self-serving managers can secure their position by proposing overly generous terms of workout, even though in some cases shareholders may be better off in Chapter 11 bankruptcy.
This theory is somewhat supported by findings of Brown, James and Mooradian (1993). Brown et al. (1994), where authors find significantly lower CEO turnover following asset sales in distressed firms when proceeds are used to repay debt. It can mean that managers responding to creditor’s pressure are retained by the firm.
For that reason I would hypothesize that change in executive position around distress period is associated with lower probability of resolving distress out of the court, because it limits number of executives that could be inclined to out of the court resolution for reasons other than corporate wellbeing.
On the other hand, changes of incumbent executives could improve probability of workout by motivating hesitant claimholders towards out of the court solutions. Sine Hotchkiss (1995) finds continued involvement of incumbent management in restructuring process to be strongly connected to poor restructuring outcomes it is possible that if company decides to introduce new board members it has better chances to improve its performance by replacing people that could be at least partially blamed for company’s poor performance in the past. At the same
13 time new executives appointed in place of dismissed board members can signal improvements in top level management and promise improvements in future financial performance.
Change of incumbent executives allows also for at least partial mitigation of information asymmetry and control risks I describe in sections 1.2.1 and 1.2.2. and to some extend can substitute additional control mechanisms that are placed on companies under Chapter 11 proceeding. Lower information asymmetry and lower control risk as well as potential for better management can convince hesitant claimholders to allow for out of the court resolution.
2.2. Hypothesis II
Higher concentration of ownership improves chances for out of the court distress resolution.
By analogy to existing research describing impact of creditors concentration (summary in Section 1.2.3), in my second hypothesis I investigate impact of a shareholders concentration on probability of out of the court distress resolution.
I hypothesize that higher ownership concentration allows for easier renegotiation of agreements due to higher involvement of shareholders in restructuring process. More concentrated shareholding makes owners more incentivized to actively facilitate the process. Gilson et al. (1990) shows that securities are offered in 74% of successful distressed exchanges. In 19% of cases explicit shareholder approval is in order to issue stock or sell specified assets. It can be expected that more concentrated ownership makes it easier to obtain such approval.
Higher concentration of shareholders can also improve management oversight by limiting the free-riders problem and limit deterioration of control mechanisms described in section 1.2.2. As I point out in previous chapter, companies with more dispersed shareholding can be subject to greater creditor pressure on self-serving managers that are willing to forgo shareholders interest and proceed to out of the court distress resolution on unfavorable terms. This mainly takes form of asset sales and shareholding dilution.
Whereas in healthy firms asset sales puts them to most productive use, liquidity constrains may lead distressed firms to sale assets at significant discount, liquidating upside potential for shareholders that won’t receive any additional value due to debt overhang. (Shleifer and Vichny (1992))
14 The subject of shareholders consent is especially important in light of Brown et al. (1994), which finds that in contrary to healthy firms, proceeds from asset sale in distressed enterprises are much more likely to paid to creditors rather than retain by the firm and cause negative security price reaction on announcement.
2.3. Hypothesis III
Higher liquidity improves chances for out of the court distress resolution.
Since Chapter 11 bankruptcy allows for automatic stay feature protecting company from seizing its assets by creditors, it can be expected that more distressed companies, with more constrained liquidity will try to utilize this benefit of formal bankruptcy proceeding. This hypothesis was previously researched and largely confirmed by number of scholars.
In contrary to this standpoint Asquish et al. (1994) do not confirm impact of company cash holdings on probability of renegotiation. Authors however do find companies obtaining additional liquidity through asset sales to be more likely to avoid bankruptcy. Selling assets in Chapter 11 bankruptcy is easier because buyer is free from legal challenger risk and if sales is prohibited by credit convenants court can allow for such operation. At the same time cash generated from operations may be often too small to save a company from bankruptcy, what is in contrary to findings of Jensen (1989).
In order to contribute to this discussion in my research I look at impact of cash held by company on probability of renegotiation.
2.4. Methodology
Similarly to Gilson et al. (1990), in order to test all my hypotheses I utilize a Probit regression model. Below I present its equation following with variable description.
15 where:
R probability of successful out of the court renegotiation
intercept
𝛽1, 𝛽2,𝛽3, 𝛾, 𝜃 coefficients
EChange explanatory variable of Hypothesis I – change in corporate management during distress period
Owners explanatory variable of Hypothesis II – percentage of shareholding structure owned by investors, each having more than 5% of common shares outstanding
C control variables – ROA, debt to capital and total assets
FE year dummies
residual
For each successful renegotiation in my data I assign a dependent variable of 1 if a company has successfully resolved default with private workout, or value of 0 if it entered in Chapter 11 bankruptcy. Following this logic, the greater coefficient I observe on given independent variable, the bigger is association of this variable with successful workout.
I use three explanatory (independent) variables: EChange, Owners and Liquidity. EChange is an explanatory variable of first hypotheses. I construct this variable assigning value of one to all companies in my data sample that have reported changes in composition of board of directors in time period between 30 days before default and 360 days after it, or until bankruptcy. I do not analyze reasons for given director leaving the office, assuming all changes to be disciplinary. According to Gilson (1989) this is not always true, although most disciplinary changes are reported as voluntary resignations due to personal reasons and it is difficult to separate them form truly voluntary decisions.
Owners is the explanatory variable of my second hypothesis. I construct it as ownership concentration proxy by summing number of shares kept by block holders in relation to total number of shares. One possible alternative would be to follow methodology of Philipp, Sautner (2007) and use Herfindahl Index in order to measure concentration of ownership. I
16 apply more simplified approach but it allows separating fraction of ownership subject to the highest risk of free riders problem.
Liquidity is my third hypothesis explanatory variable. In order to measure liquidity I use Quick Ratio (cash and marketable securities to total assets). Quick Ratio is widely used liquidity measure in financial analysis. One possible limitation of all accounting data is its availability only at the end of reporting period. I use information from report dated closest to debt default date.
In order to control for factors that were previously documented as determinants of workout probability, I introduce three control variables: return on assets, total assets and financial leverage.
Return on assets in included to account for company’s profitability. Jostrand, Sautner (2009) do not find significant differences in ROA between companies choosing bankruptcies (-0,28) and companies restructuring out of the court (-0,03), but the difference is statistically significant (t-statistic of -3,03).
Total assets is included to account for company size. According to Gilson et al (1990) firms restructuring with workouts are generally larger. Additionally, assets value allows to control also for debt structure complexity, since in sample of Gilson there is high correlation between book value of assets and number of debt contracts. (0,72, p-value 0.00).
Since Jostrand, Sautner (2009) hypothesize and confirm that higher pre-distress leverage has positive impact on probability of renegotiation, I control for leverage including ratio of debt to capital into the regression.
My data sample stretch over turbulent times of economic crisis in United States and in many other economies. I presume these events to have significant impact on renegotiation process in two primary ways.
Firstly, due to increasing volatility and risk aversion, in financial turmoil cost of capital increases, what can be proven in framework of market risk premium in Capital Assets Pricing Model. It would be reasonable to expect that higher cost of capital will make it more difficult for companies to obtain additional financing sources. As it was shown by Roberts and Sufi (2008) renegotiation events tend to be dependent on cost of capital, and thus financial crisis can be expected to impendent renegotiation processes.
17 Secondly, in response to illiquidity in financial sector, Federal Reserve System (FED) took actions named as Quantitative Easing, leading to inflating the monetary base of American dollar. In turn it can be expected that at least some of these funds were made available to distressed enterprises, fostering out of the court renegotiation. Regardless of which effect was stronger at what point in time, in order to capture the impact of economic crisis on my research, I use time fixed effects model with dummy year variables.
In order to avoid perfect multicollinearity arising from ”dummy variable trap” in my regression I skip a dummy variable for year 2014. Also, I skip a dummy variable for year 2012, because it perfectly predicts a renegotiation outcome.
Since my sample is well over 30 observations I assume Central Limit Theorem to apply, so that population standard deviation σ equals sample standard deviation s. For that reason I use Z statistics in order to test my hypotheses, implicitly assuming normal distribution of probability.
Although in all my hypotheses I can expect the direction in which explanatory variables impact probability of renegotiation, knowing controversy around one tailed tests, I decided to follow more standard approach of two tail test with significance level α of 0,05. Z Statistics required to reject null hypothesis under these assumptions is 1,96.
3. Data and descriptive statistics
I have obtained a list of US and Canadian companies that between 2005 and 2014 defaulted on their debt obligations, what is one available search criteria in S&P Capital IQ database. Capital IQ use very broad definition of default, incorporating all credit agreement violations, from technical defaults, such as broken convenants to delayed or missed payments.
Described begging of data sample results directly from availability of data in Capital IQ database. The end of sample period provides an interval in order to assure there will be no companies with renegotiation or bankruptcy process in progress. Gilson et al (1990) finds average time between default and bankruptcy to be 8 months.
Whereas described sampling method allows registering companies defaulting after beginning of renegotiation process, it does not allow incorporating companies completing their private workout without violating their credit agreements. Being aware of this limitation I did not find any theory that could point out to substantial differences between these two types of
18 companies in context of hypotheses I test. Nevertheless companies completing restructuring without default can be expected to be in better financial situation.
Subsequently, using Capital IQ and CRSP Compustat, the database has been supplemented with bankruptcy and financial statement information corresponding to financial year in which debt default occurred and a year before debt default.
In next steps I use Key Developments column to obtain information about workouts, bankruptcies and executive changes. The data presented in Key Developments is a textual log of all information registered from company reports, as well as information from news agencies. Having this data I have performed a search of key words including such terms as “renegotiation”, “workout”, “bankruptcy” and “chapter 11”.
Lastly, data on shareholding structure at debt default date was hand collected manually from information available in Capital IQ. Shareholding structure is updated quarterly, so I register shareholders at the end of last quarter before debt default date.
In order to improve quality of results, companies with insufficient financial information, as well as companies that defaulted on debt obligations or entered into bankruptcy more than once were excluded from data sample. Because timespan of the research covers period of financial crisis, companies operating in financial sector were also eliminated from data sample.
There are some data limitations arising from the way in which S&P Capital IQ database registers bankruptcies.
Firstly Capital IQ does not provide details about class of debt contract that company defaults on. Knowing that trade liabilities can be much more flexible than bank agreements, it would be particularly important to determine the origin of liability. Also in order to measure impact of coordination problems more accurately it would be helpful to include information about secured debt in regressions applied.
Secondly, there are cases where debt default is clearly registered, but there is no subsequent information about future development neither into bankruptcy nor into workout. Companies with uncomplete records were assumed to successfully restructure out of the court.
Also some companies renegotiate their debt obligations before default. Since debt restructuring is not available as separate information in Capital IQ, I had to begin my screening with information about debt defaults and subsequently manually find information
19 about renegotiation or bankruptcy. Following this methodology I do not grasp companies that successfully renegotiate before default. I do however collect information about companies that fail in renegotiation, default and end up in Chapter 11 bankruptcy.
Finally Capital IQ does not provide information on renegotiation terms, which could be informative to assess how shareholders and creditors distribute value between themselves and what is the impact of executive change on this distribution.
In subsequent paragraphs I present descriptive statistics of my data sample and key variables used in regressions.
Table A: Number of bankruptcy and workout cases for each year in data sample
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Total
Chapter 11 2 3 8 7 14 3 3 0 1 1 42
Workout 8 18 19 17 32 16 8 9 12 8 147
Total 10 21 27 24 46 19 11 9 13 9 189
My sample consists of 189 companies in total, with 42 bankruptcies and 189 successful renegotiations. The effects of financial crisis are clearly visible, with much higher default density in years 2007-2009. 78% of companies in my data sample have successfully renegotiated their financing agreements. This is substantially higher percentage than described by other researches (Gilson (1990) – 48%, Franks (1994) – 55%). Also it is difficult to notice substantial changes in success rates in years of financial crisis.
20 Table B: Descriptive statistics for key parameters
Executive
change* Owners** Revenues ROA Assets
Quick ratio Total debt Debt to capital
[#] [%] [USD M] [%] [USD M] [#] [USD M] [#]
Chapter 11 Median 22 18,40 44,65 (7,12) 75,79 0,25 36,85 0,90 Mean 52% 25,00 209,85 (22,62) 248,22 0,43 147,32 1,32 SD 25,69 537,11 41,41 460,51 0,48 276,04 1,03 Workout Median 29 34,00 42,63 (5,38) 42,86 0,33 14,89 0,81 Mean 20% 37,50 1 368,01 (33,07) 1 408,38 0,63 450,51 1,59 SD 28,03 7 627,98 76,67 6 426,61 1,05 1 743,59 3,52 Total Median 51 31,00 42,63 (6,63) 46,76 0,30 17,20 0,84 Mean 27% 34,72 1 110,64 (30,75) 1 150,57 0,58 383,13 1,53 SD 27,95 6 744,11 70,42 5 688,10 0,96 1 547,10 3,14
*Dummy variable; statistics given for sub group sum and frequency ** Total shareholding by block holders; 1-free float
27% companies in my sample change key executives, what is consistent with Philipp, Sautner (2007), where 19% of sample companies change executives. Only 20% of companies attempting successful out of the court workout have changed their top executives around distress period, in comparison to over 50% in companies proceeding in Chapter 11. Successful companies also tend to have more concentrated ownership with median of 34% versus 18%.
In terms of financial performance, companies restructuring out of the court have similar median revenues, although there must be significant outliers in data sample causing mean revenues to be much higher for workouts, which is similar to assets that on average are almost six times bigger in workout group, although the median is almost twice bigger in Chapter 11 group. Surprisingly, median companies in workout seem to be little less leveraged than Chapter 11 companies, although on average ineptness is a bit lower in Chapter 11. Outliers must be also present in ROA since median value is higher in workouts with lower average profitability.
As it can be expected, workout companies carry more cash than Chapter 11 enterprises. For all measures variance coefficient is higher in workout companies. It suggests that this group might be less homogenous and thus regression methodology should control for differences in this sub sample.
21
Table C:Correlation table for key explanatory and control variables
Executive change Owners ROA Assets Quick ratio Debt to capital
Executive change 1,0 Owners (0,2) 1,0 ROA (0,1) 0,1 1,0 Assets (0,1) 0,0 0,1 1,0 Quick ratio (0,1) (0,0) 0,2 0,1 1,0 Debt to capital 0,1 0,0 (0,6) (0,0) (0,1) 1,0
In order to test for perfect multicollinearity, in Table C, I present correlation coefficients for explanatory and control variables. I can conclude that overall correlation coefficients are rather low with exception of correlation between debt to capital and return on assets. This is rather expected since higher financial leverage, represented by debt to capital ratio, is intuitively highly correlated with greater risk and worse profitability in times of financial distress, shown in ROA. Since these are control variables, correlation between debt to capital and ROA should not impact overall model prediction, nor coefficients on explanatory variables.
It can be also noticed that there is low positive correlation between ROA and quick ratio what is also intuitively expected since more profitable firms should carry more cash.
Another possible explanation for this correlation is spurious correlation resulting from using ratios instead of absolute measurements. Although denominators of both ratios are different, current liabilities used as denominator in quick ratio and total assets used in ROA can be expected to be highly correlated.
In my sample companies with more concentrated ownership tend to change executives a bit less than companies with dispersed ownership. This is however more difficult to interpret with correlation table.
Overall, there is no significant correlation between explanatory and control variables. It suggests that there should be no perfect multicollinearity problem in assessment of explanatory coefficients. Naturally, correlation between any variable with itself is always 1.
22
4. Results
In Tables D and E I present results of my model. The Adjusted-R2 of the model is 22,6%, what is similar to fit achieved by other researchers, such as Sautner (2009) 17,3%-37,1% or Gilson (1990) 1,3%-5,1%. Since this is not predictive model, but rather a method to test impact of particular variables in question, low R2 is acceptable. Nevertheless model explains only around 23% of explained variable variance, so the impact of other, not included factors that determine probability of renegotiation is high. Value of Chi-squared allows to reject a hypothesis that all coefficients are not significantly different from 0.
Table D: Results margins
Variable Coefficient Std err Z P>|z| 95% confidence interval
Echange (0,259)*** 0,071 (3,660) 0,000 (0,398) (0,120) Owners 0,003** 0,001 2,810 0,005 0,001 0,005 Liquidity 0,090 0,058 1,550 0,121 (0,024) 0,205 ROA (0,001)* 0,001 (2,000) 0,046 (0,003) (0,000) Total assets 0,000 0,000 0,850 0,393 (0,000) 0,000 Debt to capital (0,016) 0,019 (0,810) 0,420 (0,053) 0,022 Dummy 2005 (0,042) 0,196 (0,210) 0,831 (0,426) 0,342 Dummy 2006 (0,042) 0,171 (0,250) 0,805 (0,377) 0,293 Dummy 2007 (0,229) 0,160 (1,440) 0,151 (0,542) 0,084 Dummy 2008 (0,265) 0,164 (1,620) 0,106 (0,587) 0,056 Dummy 2009 (0,219) 0,151 (1,450) 0,148 (0,515) 0,077 Dummy 2010 (0,082) 0,173 (0,470) 0,637 (0,420) 0,257 Dummy 2011 (0,110) 0,178 (0,620) 0,538 (0,458) 0,239 Dummy 2012 (omitted)
23
Dummy 2013 0,122 0,205 0,600 0,551 (0,279) 0,523
Dummy 2014 (omitted)
R-squared 22,6%
* p < 0,05; ** p < 0,01; *** p < 0,001
In order to give applicable interpretation to computed coefficients I calculate marginal effects for each regressor. Marginal effects are presented in Table D. They were computed investigating unit change in given variable, assuming all other variables are set at their respective means. In case of binary variables adopted unit change is from 0 to 1. In case of continuous variables it is instantaneous rate of change.
In regard to Hypothesis I, I find strong confirmation of negative impact of executive change on probability of renegotiation. The p-value of 0,00 is less than applied two tail significance level α of 0,05. Also, the whole scope of confidence interval is negative, what allows for less ambiguity in interpretation of results. According to Table D, changing top executives in distress period reduces chances for out of the court renegotiation by 26%.
This is consistent with results of Philipp, Sautner (2007) that find firms going bankrupt to be more likely to change executives, at 1% significance level (coefficient 1,03).
At the same time my results are not aligned with hypothesis of Giammarino (1989) that sees executive change as a method to reduce information asymmetry between creditors and managers, and improve changes of renegotiation.
I come up with two possible explanations for this observation. I) either executives of companies with poor prospects decide to leave in order to protect their reputation, before the financial situation deteriorates or II) executives indeed are prone to agency conflicts and try to settle with creditors regardless of terms, in order to protect their jobs.
If seconds is the case, my finding confirms possible corporate governance problem. Its implications are especially important in light of research documenting worse post-bankruptcy performance of companies restructuring with incumbent management (Hotchkiss 1995). Also Bogan, Sandler (2011) finds new management to be a factor in post-bankruptcy survival, improving probability of 5-year survival by 5%. New board members, likely not granted
24 golden parachutes are described as better suited to refocus on business lines and increase on post-reorganization performance.
In regard to Hypothesis II, I confirm that higher concentration of ownership has significant, although small impact on probability of out of the court workout. The coefficient on variable Owners is 0,01. Coefficient is calculated for instantaneous change in ownership concentration and can be accurately extrapolated only under assumption of lineal relationship between variables. For increase of one standard deviation from mean, ownership concentration value (35% to 62%) the probability of workout increases by 7,5%.
In that regard my finding quantifies hypothesis developed by Jorstrand, Sautner (2009) where authors find that probability of workout is greater for unlisted firms due to less dispersed ownership, what makes bargaining easier. As I describe in literature overview, higher ownership concentration not only streamlines negotiations due to lower coordination risks, but also mitigates risk shifting and control problems.
In regard to Hypothesis III, I cannot reject null hypothesis that coefficient on variable Liquidity is different from zero and thus cannot confirm impact of liquidity on probability of renegotiation. Nevertheless, this hypothesis would be confirmed with less strict assumptions of one-tail test and α of 0,10. Coefficient on quick ratio is 0,09. One standard deviation increase in quick ratio value (0,58 to 1,54) can be suspected to increase probability of workout by 7,3%. This is consistent with results obtained by other researchers that describe liquidity to have much smaller significance than other factors. Whereas higher liquidity allows company to operate longer without external sources of financing, it is not enough to reduce leverage or allow for structural changes in business model.
Insignificance of firms leverage and total assets is consistent with findings of Philipp, Sautner (2007), although in contrary to that research I find ROA to be highly significant, although marginally important determinant of workout. Increase in one-half standard deviation (-31% to 4,5%) in fact decreases changes for renegotiation by 3,5%)
Also, for period of financial crisis significant increase in coefficient values for time dummies can be observed. There is also parallel decrease in p-values associated with these coefficients. Last financial crisis demonstrated its effects in real economy the strongest in year 2008, when my model forecasts probability of renegotiation to decrease by 27%, with relatively low p-value of 0,10.
25 Constant term is positive, what could be rather expected taking into account possible overrepresentation of successful workouts in my data sample.
Table E: Results
Variable Coefficient Std err Z P>|z| 95% confidence interval
Echange -1,05*** 0,26 -3,97 0,00 -1,57 -0,53 Owners 0,01** 0,00 2,80 0,01 0,00 0,02 Liquidity 0,37 0,24 1,51 0,13 -0,11 0,84 ROA -0,01 0,00 -1,95 0,05 -0,01 0,00 Total assets 0,00 0,00 0,81 0,42 0,00 0,00 Debt to capital -0,06 0,08 -0,80 0,42 -0,22 0,09 Dummy 2005 -0,17 0,79 -0,21 0,83 -1,72 1,38 Dummy 2006 -0,17 0,69 -0,25 0,81 -1,53 1,19 Dummy 2007 -0,93 0,64 -1,44 0,15 -2,19 0,33 Dummy 2008 -1,07 0,66 -1,63 0,10 -2,37 0,22 Dummy 2009 -0,89 0,61 -1,46 0,14 -2,08 0,30 Dummy 2010 -0,33 0,70 -0,47 0,64 -1,70 1,04 Dummy 2011 -0,44 0,72 -0,62 0,54 -1,85 0,96 Dummy 2012 (omitted) Dummy 2013 0,49 0,83 0,59 0,55 -1,14 2,13 Dummy 2014 (omitted) Constant 0,93 0,62 1,50 0,13 -0,28 2,15 R-squared 22,6% * p < 0,05; ** p < 0,01; *** p < 0,001
26
5. Conclusion
Using a sample of 189 American companies that between 2004 and 2012 defaulted on their debt obligations, I examined factors determining the choice of distressed companies between in court bankruptcy and out of the court debt renegotiation. I utilize a Probit model to explain variance in ex-post renegotiation successfulness. In terms of explanatory variables I concentrate my attention on previously ambiguously described subjects of executive change, ownership concentration and liquidity.
I find that companies changing their executives around distress period are much more likely to success in renegotiation. A possible explanation for that is much higher board turnover rate under Chapter 11 bankruptcy, what may be a decision factor for executives in deciding between bankruptcy and workout. My finding signals a potential corporate governance problem, because in some cases Chapter 11 bankruptcy may be preferable from shareholders point of view. Of course other possible explanations for this finding may assume that I find correlation but no causality and board members are changing jobs when they anticipate company going bankrupt.
I also find that greater ownership concentration slightly improves chances of workout. I assume reason for that to be parallel to previously described positive impact of greater debt holding concentration. Since impact of this factor is limited (one standard deviation increase in concentration results in 7,5% improved workout probability) this finding is rather to confirm and quantify relationship expected, but not examined by other researchers.
In my research I also contribute to discussion about impact of liquidity, finding that its impact on probability of renegotiation is insignificant, similarly to results obtained by Asquish (1994).
Finally, controlling for financial performance of distressed companies I describe lower probability of workout in years of financial crisis (2008, -27%). Whereas it is not statistically significant, consistent pattern of increased significance and coefficient values in financial crisis dummies means that during recent financial crisis companies could have had problems with renegotiation arising from turbulences in banking system. The extend of these problems could have been greater if it was not for anti-austerity measures implemented by FED, although it is still an open question for further research in that subject.
27 References:
Akerlof, George A. (1970). "The Market for 'Lemons': Quality Uncertainty and the Market Mechanism". Quarterly Journal of Economics (The MIT Press) 84 (3): 488–500.
Asquith P., Gertner R., Scharfstein D., 1994, Anatomy of Financial Distress: An Examination of Junk-Bond Issuers, The Quarterly Journal of Economics 109 (1994), 625-658
Betker, Brian L., 1995. “An Empirical Examination of Prepackaged Bankruptcy,” Financial
Management 24: 3-18.
Betker 1998, The security price effects of public debt defaults, Journal of Financial Research 21 (1998), 17-35 Brunnermeier, 2008, Deciphering the Liquidity and Credit Crunch 2007–2008, Journal of Economic
Perspectives—Volume 23 (2009), 77–100
Bulow, Jeremy, Shoven, 1978, The bankruptcy decision, Bell Journal of Economics 9 (1978), 436-445
Coase R. H, (1960) “The Problem of Social Cost” Journal of Law and Economics, Vol. 3 (Oct., 1960), 1-44 Fama, E. F., and M. H. Miller, 1972, The Theory of Finance, Dryden Press, Hinsdale, Ill.
Franks J., W.Torous, 1993, A comparison of financial recontracting in distressed exchanges and Chapter 11 reorganizations, Journal of Financial Economics 35 (1994) 349-370
Franks J., Julian R. and Walter N. Torous, "A Comparison of Financial Recontracting in Distressed Exchanges and Chapter 11 Reorganizations", Journal of Financial Economics, Vol. 35, No. 3, (June 1994), pp. 349-370. Haugen, R. A., and L. W. Senbet, 1988, "Bankruptcy and Agency Costs: Their Significance”
Stiglitz, J. E., 1974, "On the Irrelevance of Corporate Financial Policy," American Economic Review, 64, 851-866.
Haugen, R. A., and L. W. Senbet, 1978, "The Insignificance of Bankruptcy Costs to the Theory of Optimal Capital Structure," Journal of Finance, 33, 383-393.
Hotchkiss, E. 1995, "Post bankruptcy Performance and Management Turnover”, The Journal of Finance, Vol. L, No.1
28
Ivashina, Scharfstein, 2009, Bank lending during financial crisis of 2008, Journal of Financial Economics 97 (2010), 319-338
Jostrand P, Sautner Z., 2009, Out-of-Court Restructuring versus Formal Bankruptcy in a Non-Interventionist Bankruptcy Setting, Review of Finance 14 (2010), 623-668
Miller, Merton H. and F. Modigliani. 1958. "The Cost of Capital, Corporation Finance and the Theory of Investment." American Economic Review, 48(3), 261-297
Roberts R., Sufi A., 2008, Renegotiation of Financial Contracts: Evidence from Private Credit Agreements,
