Understanding the Recovery Rates on Defaulted Securities1

Viral V. Acharya London Business School Sreedhar T. Bharath University of Michigan

Anand Srinivasan University of Georgia Keywords: Recovery, Default, Credit risk, Distressed securities, Bankruptcy. J.E.L. Classification Code: G33, G34, G12. First Draft: February 2003, This Draft: April 20042
1 Viral

Acharya:

Department of Finance, London Business School and is a Research Affiliate of the Centre for Economic Policy Ree-mail:

search (CEPR). Address: London Business School, Regent’s Park, London - NW1 4SA, UK. Phone: +44(0)20 7262 5050 x 3535.

vacharya@london.edu. Sreedhar T. Bharath: University of Michigan Business School, Department of Finance, D6209, Davidson Hall, Ann Arbor, MI-48103, U.S.A Phone: (734) 763-0485. e-mail: sbharath@umich.edu. Anand Srinivasan: University of Georgia, Department of Banking and Finance, Brooks Hall, Athens, GA 30602-6253, U.S.A. Phone: (706) 542-3638. e-mail: asriniva@uga.edu.

2 We thank Sergei Davydenko, Amy Dittmar, Denis Gromb, Rohit Guha, Julian Franks, Herbert Rijken, Ilya Strebulaev, and Per Stromberg for
useful discussions, the seminar participants at Bank of England, Dynamic Corporate Finance Workshop at Copenhagen Business School, European Finance Association Meetings (EFA) 2003, Federal Reserve Board at Atlanta, Lehman Brothers, London Business School, New York University, University of Michigan Business School, Anderson School of Business – UCLA, Wharton School of Business, and the 14th Annual Financial Economics and Accounting Conference (2003), for comments, and Deepak Bhandari, Karthik Balakrishnan, Jugal Parekh, and Anupam Sharma for excellent research assistance. The authors acknowledge the help of Edward Altman, Brooks Brady, and Standard and Poors for the provision of data employed in the paper and its documentation. The authors are grateful to the Institute for Quantitative Investment Research (INQUIRE), UK for its financial support for the project. Acharya is grateful to the Research and Materials Development (RMD) grant from London Business School. Srinivasan is grateful for financial support from the University of Georgia Research Foundation research grant and the Terry-Sanford research grant. The views expressed and errors that remain in the paper are our own and should not be attributed to these supporting institutions.

Understanding the Recovery Rates on Defaulted Securities

Abstract We document the empirical determinants of the recovery rates on defaulted securities in the United States over the period 1982–1999. In addition to seniority and security of the defaulted securities, industry conditions at the time of default are found to be robust and important determinants of the recovery rates. Recovery in a distressed state of the industry (median annual stock return for the industry firms being less than −30%) is lower than the recovery in a healthy state of the industry by 10 to 20 cents on a dollar. A better liquidity position of the peers of the defaulted firm also implies higher recovery at emergence from bankruptcy. Furthermore, recovery is negatively correlated with asset-specificity of the industry when the industry is in distress, but not otherwise. Our findings are thus consistent with the industry equilibrium hypothesis of Shleifer and Vishny (1992). In contrast to perceived wisdom and recent literature, we find that recovery is not affected by the aggregate supply of defaulted securities, once the industry distress effect is accounted for.

1

Introduction

Since the seminal work of Altman (1968) and Merton (1974), the literature on credit risk has burgeoned especially in modeling the likelihood of default of a firm on its debt. Extant credit risk models, such as the industry standard KMV (www.kmv.com), Litterman and Iben (1991), Jarrow and Turnbull (1995), Jarrow, Lando, and Turnbull (1997), Madan and Unal (1998), Sch¨ onbucher (1998), Duffie and Singleton (1999), Das and Sundaram (2000), and Acharya, Das, and Sundaram (2002), employ varying choices for modeling the likelihood of default. The credit spreads or prices of risky bonds and loans determined in these models depend also on the loss given default or inversely on the recovery rates on the bonds under consideration.1 However, many of the extant models and their calibrations assume that recovery is deterministic.2 Some preliminary evidence, see e.g., Carty and Hamilton (1999), Carty, Gates, and Gupton (2000), Brady (2001), and Altman (2002), suggests, however, that recovery rates on defaulted instruments exhibit substantial variability. This paper studies the empirical determinants of recovery risk - the variability in recovery rates over time and across firms - using the data on observed prices of defaulted securities in the United States over the period 1982–1999. We measure recovery rates using the prices of defaulted securities at the time of default or bankruptcy and at the time of emergence from default or bankruptcy. We focus our investigation on the contract-specific, firm-specific, industry-specific, and, finally, macro-economic determinants of recovery rates. We document systematically the impact of these factors on recovery rates and study whether they are different from the determinants of the likelihood of default for these securities. Among contract-specific characteristics, we find that seniority and security are important determinants of recovery rates: Bank Loans recover on average 20 cents more on a dollar than Senior Secured and Unsecured instruments, which in turn recover 20 cents more than Subordinated instruments. In terms of security, the difference in recovery at emergence between instruments backed by Current Assets and Unsecured instruments is about 25 cents on a dollar.3 Recoveries at emergence are found to be affected adversely by the length of time the firm spent in bankruptcy.
For instance, under the specific recovery assumption of Duffie and Singleton (1999), called the Recovery of Market Value (RMV) assumption, the credit spread to be added to the risk-free rate in order to obtain the discount rate applicable for defaultable securities is precisely equal to the risk-neutral hazard rate of default multiplied by the loss given default. 2 Notable exceptions which model recovery risk are Das and Tufano (1996), Frye (2000a, 2000b), Jokivuolle and Peura (2000), Bakshi, Madan, and Zhang (2001), Jarrow (2001), and Guntay, Madan, and Unal (2003). We discuss some of these models in Section 7. See also the Introduction in Altman et al. (2003) for a survey of the literature on recovery risk. 3 Our data-sets do not provide collateral information matched with instruments for which we have recoveries at default.
1

1

we test the following four hypotheses: (1) Poor industry or poor macroeconomic conditions when a company defaults should depress recovery rates. receive a lower price for these aircraft than companies that sell aircrafts at other times.25 cents on a dollar. or the anticipation of low prices for asset sales gives greater bargaining power to equityholders in writing down creditor claims.Among firm-specific factors. Surprisingly. The effect of tangibility is captured entirely by the Utility industry dummy: Utility industry firms recover in default 30 to 40 cents on a dollar more than other industries. and in turn. and leverage ratio measured as Long Term Debt/Assets of the defaulting firms in default year minus one. To the best of our knowledge. In particular. Pulvino (1998) examines data on asset sales in the airline industry. no study has examined directly the implications of this theory for recoveries on defaulted instruments. in the year preceding default) increases recoveries at default by 0. Asquith. we find the tangibility of assets (measured as Property. Andrade and Kaplan (1998) find supporting evidence as well that poor industry and economic conditions adversely affect company performance or value. a 1% marginal increase in the profitability of assets (measured as EBITDA/Sales. Firm-specific characteristics thus determine some variation in recovery rates but not in a consistent manner across different measures of recovery. and outstanding maturity of the instruments. Our most striking results concern the effect of industry-specific and macroeconomic conditions in the default year. 5 Using airline industry data. 4 2 . and Equipment/Assets in the year preceding default) does not affect either recoveries at default or at emergence. or companies that sell aircrafts when the industry is doing poorly. Therefore. the empirical implications of Shleifer and Vishny (1992) model can be translated into implications for debt recoveries. (2) Companies that operate in more concentrated industries should Also.5 A lower asset value in liquidation should translate into a lower debt value: Either bankruptcy entails liquidation of assets to pay creditors. we do not find any explanatory power for recoveries in coupon. Plant. none of the firm-specific characteristics appear to be significant determinants of recoveries at emergence. Gertner and Scharfstein (1994) find in their study of “junk” bonds during the 1970s and 1980s that the use of asset sales in restructuring of distressed firms is limited by industry conditions such as poor performance and high leverage. However. consistent with the findings of Altman and Kishore (1996).4 Finally. Using a sample of 39 highly levered transactions. consistent with the theory that predicts greater coordination problems among creditors in these cases. He finds evidence supportive of the Shleifer and Vishny (1992) model: Companies that sell aircrafts when they are financially constrained. Shleifer and Vishny (1992) develop a theoretical model where financial distress is more costly to borrowers if they default when their competitors in the same industry are experiencing cash flow problems. past stock return of the firm positively affects recovery at default whereas stock return volatility negatively affects recovery at default. Their findings however suggest though industry and economic factors may be important determinants of recovery rates. These studies however do not study prices of bonds. issue size. greater incidence of costly liquidations. Recoveries at default are also negatively affected by greater number of defaulted issues and by greater debt dispersion.

Third. Then. this effect of industry distress is always non-linear: Industry return employed as a continuous return variable has no explanatory power for recoveries. We identify the industry of the defaulted firm using the 3-digit SIC code of the company. the ratio of Current Assets minus Inventories to Current Liabilities. we correlate industry-level recoveries with industry-level asset-specificities for industry-year pairs when the industry is in distress and when it is not in distress. we calculate the specificity of an industry using the mean (or the median) of Specific Assets of all firms in that industry. we define Specific Assets of a firm as the book value of its machinery and equipment divided by the book value of total assets. Furthermore. and another variant based on the median interest coverage ratio (EBITDA/Interest Expense). (4) Industries that have more specialized assets (those that have few alternative uses) should have lower recoveries in the event of default. we define an industry to be “distressed” if the median stock return for this industry in the year of default is less than or equal to −30%. Defaulting companies whose industries also have suffered adverse economic shock thus face significantly lower recoveries. we calculate a sales-based Herfindahl index for the industry of the defaulted firm as a measure of its concentration. the magnitude of the effect is about half the relative effect of seniority of the instrument (Bank Loans vs. The effect is robust to this definition and for recoveries at default it is in fact twice as large as that under the previous definition of distress. and Lang (1990) and Opler and Titman (1994). As in Berger. First. Ofek. We find that the correlation between recoveries at emergence 3 . Controlling for the industry distress condition. both for recoveries at default as well as emergence. We find that when the defaulting firm’s industry is in “distress. John. we construct a measure of industry liquidity based on the median Quick ratio. Finally. Median Q controls for the fact that industry distress may capture worsening of industry’s future prospects. Importantly.” its instruments recover about 10–12 cents less on a dollar compared to the case when the industry is healthy. we find that poor industry liquidity negatively affects the recovery rates at emergence. Second. be associated with low economic worth of assets. However. We also consider an alternative definition of industry distress that employs stock returns and sales growth in the two years prior to default. The effect is both statistically and economically significant. and also controlling for Median Q of the industry. in the spirit of Gilson. Subordinated debt).have lower recoveries due to the lack of an active market of bidders. In fact. this effect is found after controlling for contract-specific and firm-specific characteristics. and in turn. poor industry liquidity does not appear to depress the recovery rates at the time of default. We do not find any evidence supporting the hypothesis that industry concentration lowers the recovery rates. (3) Poor liquidity position of industry peers when a firm defaults should lower the recovery on its debt. and Swary (1996) and Stromberg (2001). Senior Debt vs.

and Energy and Natural Resources) experience a significant drop in debt recoveries (about 30 cents on a dollar) when they are in distress relative to their no-distress levels. Does this imply that a credit-risk model must incorporate factors over and above the ones that determine default risk? We provide preliminary evidence that models incorporating recovery risk would likely also need to model industry conditions as a separate state variable. We also test the hypothesis that poor macroeconomic conditions at time of default should depress recoveries.6 We find that the aggregate default rate in the default year and the aggregate supply of defaulted bonds measured mid-year in the default year (Altman et al. Resti. However. the industry distress effect is present for recoveries at emergence even when 1990. It is difficult to conclude that illiquidity in the financial markets for trading defaulted instruments causes lower recoveries: The more likely candidate is in fact the illiquidity in the market for asset sales. These findings are in contrast to those of Altman et al. Transportation.76) for mean (median) recovery and mean (median) asset-specificity. their effect becomes insignificant. Brady. except for the hypothesis concerning industry concentration..g. in that there is no effect of bond market conditions over and above the industry conditions. The linkage between bond market aggregate variables and aggregate recoveries stressed by Altman et al. and of the Fama and French factors (SMB and HML). In particular. as arising due to supply-side effects in segmented bond markets appears to be a manifestation of omitted industry conditions. or the Frye (2000a. 6 4 . the variables motivated by Shleifer and Vishny (1992) explain well the time-series variation in recovery rates. our results demonstrate that recovery risk exists and its magnitude is economically significant. or the credit-score of Zmijewski (1984).63 (−0. variables BDR and BDA.and asset-specificity when industries are not in distress is statistically and economically insignificant. 2000b) and Hu and Perraudin (2002) also show that aggregate quarterly default rates and recovery rates are negatively correlated. is excluded from the data. once industry conditions are employed. and Sironi (2002) who study aggregate recovery rates and find that aggregate recovery rates on defaulted debt are negatively related to aggregate default rates and to aggregate supply of defaulted bonds. The findings are complementary to theirs in that the effect of industry conditions is robust to inclusion of bond market conditions and macroeconomic conditions. Industries with highly specific assets (e. the NBER recession year in our sample. The correlation is however significantly negative when industries are in distress: −0. More broadly. Thus. is generally insignificant. respectively) significantly lower recovery rates when employed in the absence of industry conditions. We employ macroeconomic data from Altman. 2000). GDP growth rate in the default year. The effect on recovery of S&P 500 stock return. Introducing an interaction term between industry distress dummy and asset-specificity also confirms this finding in the regression framework. We include in the regression specification a Z–score based on the credit-scoring models of Altman (1968.

and pre-packaged bankruptcies or distressed exchanges which involve no Chapter 11 filing. Institutional and legal differences between the bankruptcy codes in the U. 7 5 . but not perfectly. and 465 firm defaults and 1. but the effect of contract-specific and industry-specific characteristics remains unaffected. Section 3 discusses the data we employ. a separate literature examines the recovery rates of different classes of creditors in the event of distress. Also. Franks and Torous (1994) examine the recovery rates of different classes of creditors in the event of a distressed exchange of securities or a bankruptcy. Thorburn (2000) looks at the recovery rates for debt in a set of bankruptcy auctions in Sweden. The remainder of the paper is structured as follows. Hotchkiss (1995) documents that firms that emerge from Chapter 11 tend to default again subsequently. The recovery rates in their sample are largely based on book values of securities received in a reorganization or bankruptcy. our data set has bankruptcies with a Chapter 11 filing and also cases of distress and cure (where there is a default and a rapid resolution).Distance to Default as computed by the KMV adaptation of the Merton (1974) model. As explained in the paper.S. our data enables us to test hypotheses concerning fall in defaulted debt prices in anticipation of strategic write-downs by equityholders who exploit their bargaining power The median recovery for this sub-sample was significantly lower than their overall sample. recovery rates based on book values are likely to overstate true recoveries. We find that the default-risk proxies show up as significant. Franks and Torous (1994) do report recovery rates for a sub-sample of 10 cases of distressed exchanges and 12 cases of bankruptcy based on market values for all securities received by the given creditor class. The determinants of risk of default and the risk of recovery thus seem positively. Section 4 presents summary descriptive statistics. Furthermore. 2 Other Related Literature In addition to the related papers discussed above. Our proposed study uses the market prices of debt after the default event thereby circumventing this problem.511 instruments for recoveries at emergence.S. Section 7 presents implications for future credit risk models. Section 8 concludes. Chapter 11 gives the incumbent management a substantial advantage in the reorganization process in the U. covering 379 firm defaults and 645 instruments for recoveries at default. Section 6 compares the determinants of recovery risk with the determinants of default risk. correlated. and Sweden may make some of her results and conclusions inapplicable to the U. Therefore.S.7 Our sample of market-value based recoveries is more comprehensive. Section 2 discusses additional related literature. Section 5 presents the regression analysis. providing a further justification for using market-based recovery measures.

and Carty and Gupton (2000) for recoveries on defaulted bank loans. and industries. a focal point in the analysis of this paper. Being a proprietary model. Somewhat surprisingly. he finds a negative effect of industry concentration on recoveries. a variable that is unavailable in our data set and somewhat cumbersome to obtain given the exhaustive nature of our data set (over 1500 instruments). and on the average length of time to default-resolution. 8 6 . Izvorski finds a positive effect on recoveries from fixed to total assets and from industry growth.when asset sales are expected to fetch low prices. The key differences between our work and Izvorski’s arise from the facts that (i) we examine both recoveries at default and at emergence. Note that in the spirit of our findings. Izvorski finds seniority and type of industry to be the major cross-sectional determinants of recovery. Consistent with our results. and macroeconomic factors. securities.. and Thorburn (2000) documents a significant effect on recoveries of a binary variable signifying if bankruptcy occurred during an economic downturn (year 1991) or not.9 These studies focus on description of average recoveries on loans and bonds across different seniorities. 9 See. firm-specific. Carty and Hamilton (1999) for recoveries on corporate debt in general. tangibility of assets has no significant effect on recoveries and neither does industry concentration. A study of recoveries on bank loans and corporate debt can be found in reports by Moody’s Investors Service (Global Credit Research). More important. 2002) explains the variation in recoveries using contract-specific. industry-specific. Since these are precisely the variables whose variation we attempt to explain. In contrast. but do not study the cross-sectional variation of recoveries across firms and through time. Finally. e. Perhaps our paper is closest to the work of Izvorski (1997) who examines the recovery ratios for a sample of 153 bonds that defaulted in the United States over the period 1983– 1993. in contrast to our industry and macroeconomic variables. (ii) our data enables us to study effect of collateral on recoveries. the choice of these variables in lagged forms may be desirable from a numerical-fitting perspective but is not very useful from an economic perspective.g. the LossCalc model employs the moving average of industry recoveries and an index of prices of bankrupt bonds. we examine in detail the effects of industry distress and industry liquidity on recoveries. Izvorski also finds the type of restructuring attempted after default to be a significant determinant. Carty and Lieberman (1996). Finally. Moody’s LossCalcT M model (Gupton and Stein. our tests provide support for industry distress and industry liquidity effects. and (iii) crucially.8 At an industry level. Carty (1998). The industry growth effect is similar to the effect of median industry Q in our results. and Moody’s LossCalc model by Gupton and Stein (2002). its description does not disclose the exact point estimates or their standard errors. Thorburn (2000) does not consider the effect of industry conditions on debt recoveries. we find that once the industry is controlled for. Franks and Torous (1994) do find in their sample a positive relationship between the past performance of the overall stock market and recoveries.

While the sample of prices at default from S&P database is smaller (perhaps due to greater illiquidity of corporate bonds at the time of default). the recovery rates for which likely behave differently from the instruments we examine. coupons. In addition. by and large the information in one is not replicated in the other. Note that our data does not contain any trade credit or project finance instruments. The S&P database has a total of 646 observations. price at default. the S&P database and the Portfolio Management Data (PMD) database. inventory and/or receivables. real estate. 1999. The PMD data includes recovery information in the form of price at default and also price at emergence on more than 1. the database provides company names. Collateral for each secured instrument is specifically identified and grouped into several categories including all assets. The information is derived from more than 300 non-financial. press releases.S. The database consists of two component databases. At the issue level. and default dates. Although there is some overlap between the two databases. Out of the 1. The PMD database contains recovery data developed by Portfolio Management Data (a part of S&P). the PMD database also provides information on collateral backing the instruments in default. The PMD database measures recoveries at emergence (henceforth. The coverage becomes extensive after 1987. Securities and Exchange Commission filings. (3) Value for illiquid settlement instruments at the time of a “liquidity event” – the first 7 . and wherever available. The S&P database provides detailed information on all companies that have defaulted between Jan. denoted as P e) using three separate methods: (1) Trading prices of pre-petition instruments at the time of emergence. the sample of prices at emergence from PMD database is close to being exhaustive and captures well the set of defaults in the United States over the period 1981 to 1999. (2) Earliest available trading prices of the instruments received in a settlement. For debtors that have emerged from bankruptcy. The source data was obtained by S&P from bankruptcy documents: reorganization and disclosure statements. and their internal rating studies on the issuer.0) developed by Standard and Poor’s (S&P).511 observations in the PMD database 399 are found in the S&P database. seniority rankings. and other.3 Data The data source is the Credit Pro database (version 4. and other debt instruments.200 bank loans. The S&P database contains information only about bonds and does not include collateral information. non-current assets. issue sizes in dollars. totaling over $100 billion. companies that have defaulted until the end of 1999. the CUSIP and SIC codes. public and private U. industry codes. we obtained our overall sample of bank loans and corporate bonds. the database provides bond names. emergence dates also are furnished. 1. press articles. 31. equipment. 1981 and Dec. high-yield bonds. At the issuer level. Combining these two data sets.

Therefore.) that prices of bonds of a corporation with different maturities and coupons but the same seniority differ substantially before bankruptcy. cross-acceleration clauses typically cause all its other claims to also file for default. we hand-matched the list of defaulted companies in this data set with the CRSP–COMPUSTAT database. or distressed exchange. it is to be noted that collateralized instruments continue to accrue post-petition interest (after filing for bankruptcy) and thus the amount payable can exceed par. denoted as P d). 4 Recovery Rates: Summary Before discussing the general patterns observed in recovery rates data. and WorldCom Inc. once the bankruptcy is announced however. the last trading price at the end of the month in which default took place is recorded in the database. since bond covenants contain a provision that makes the principal amount immediately payable upon default. Several of the defaulted companies were private at the time of default as they had undergone leveraged buy-outs prior to the default event. Furthermore. The S&P database measures recoveries at default. We have confirmed this finding by also examining prices on defaulted instruments provided by LoanX. Guha documents with examples (Enron Corp. For 399 instruments. These were obtained from Edward Altman. 10 8 . In the case of price at default (henceforth. the prices of these bonds converge to identical or close to identical values. we are unable to obtain accounting or stock market data for these firms around the time of default or even one year prior to default. Since each defaulted firm’s bankruptcy procedure and reorganization take a different period of time. a new data service started jointly by a group of the largest investment banks. such as the subsequent acquisition of the company. 11 The S&P database does not always have the CUSIP of the issuing firm. we have both the price at default and the price at emergence.11 We also use the macroeconomic variables identified by Altman. Our matching procedure is conservative in that we assign a match only when we can absolutely confirm the identity of the defaulted company. refinancing. Both these measures of recovery are given in nominal terms and should be interpreted as Recovery of Face Value or Recovery of Par. However. Resti. This way of measuring recovery is often used in practice and is justified by the fact that when a firm defaults on any one of its obligations.10 We obtain the firm and industry variables for our analysis by cross-matching the CUSIPs of these firms with the CRSP–COMPUSTAT merged database. we present the adjustments we make to the recovery prices at emergence. significant ratings upgrade.date at which a price can be determined. Brady. an event that would make prices of these bonds identical. Our private communication with “distress” hedge-fund managers also revealed that they often engage in convergence trades between bonds of the same firm with differing maturities in order to bet that the firm will default. subsequent bankruptcy. the time-period between default Guha (2003) discusses the institutional underpinnings of Recovery of Face Value as the appropriate measure of recovery. the amount payable on defaulted instruments once the firm is in bankruptcy is usually close to par. Therefore. In particular. and Sironi (2002) in our analysis.

Issues included in the index have maturities of one year or more and have a credit rating lower than BBB–/Baa3. most of these firms had not yet emerged from bankruptcy when our data was collected. The mean (median) recovery rates at emergence for These indices are total return indices.34. and t is the default date (in years). However. For P ehyld. in addition to raw emergence prices (P e). In particular. in most of the tables we report the results only with P ehyld as the emergence recovery rate. and Merrill Lynch highyield indices since none of these indices were available for use over the entire sample period. they do not appear in the recovery prices at emergence. we adjust them for the time between emergence and default dates. but are not in default. The number of defaults is quite small during the period from 1982 to 1986 (under ten in terms of firm defaults). The number of defaults for which we have recovery prices at default rises steeply again in 1999 (64 firm defaults). including deferred interest bonds and payment-in-kind securities. picks up rapidly reaching its maximum during the recessionary phase of 1987 through 1992.96 (38. we describe the time-series behavior of recovery prices at default and at emergence. which includes reinvestment of income. 12 9 . Merrill Lynch High Yield Master II index is a market value-weighted index comprised of 1. T is the emergence date (in years). Hence. We employed Lehman Brothers.800 domestic and yankee high-yield bonds. For example. we used the formula P ehyld = P e ∗ Id . (1 + c)(T −t) (2) where c is the coupon rate on the defaulted instrument under consideration. Time-series behavior: In Table 1. the economic magnitudes are similar too. emergence prices discounted at high yield index (P ehyld) and emergence prices discounted at coupon rate (P ecoup).date and emergence date is not identical for different default instances. and reduces somewhat in the mid-1990s. respectively.12 For P ecoup. and in most cases.00) cents on a dollar with a sample standard deviation of 25. Ie (1) where Id and Ie are the level of a high-yield index at default date and at emergence date. we construct two other measures. Hence. The index is a fully invested index. Our results are qualitatively robust across these measures. we used the formula P ecoup = P e ∗ 1 . In order to compare emergence recovery prices for different default instances. Salomon Brothers. Note that the default recovery prices P d are all measured within one month of default and such time adjustment is not as important. The mean (median) recovery rates at default are 41.

There were 64 firm defaults in 1999 as compared to a similar number of defaults over the entire period from 1994 to 1998. Consumer and Service Sector.07 and 31. This recovery is 8% to 10% below the 15–year mean for bank loan recoveries of 81. The number of firm defaults in 1990 and 1991 were respectively 41 and 52.94).99). High Technology and Office Equipment. The mean (median) recovery at default is 68. the relationship being particularly strong for the period starting 1987.3 billion and simultaneously bank loans achieved their lowest recovery rate of 72% (in terms of value of instruments received at emergence).” Standard and Poor’s (www. the recent evidence on recoveries is a point in case for the negative correlation between aggregate default intensity and recovery levels.”13 Effect of Industry: In Table 2. the highest number of firm defaults have been for the Consumer and Service sector. we find the recovery rates are the highest for the Utility sector.09) and 52. For example. Aerospace and Auto and Capital Goods.90 for the year. These levels are statistically different from mean recoveries for other industries (at 5% level using the Scheffe. In 2002.50. Financial Institutions. 13 10 . in 1999.53).S. respectively.58 and 36. test). Leisure Time and Media. with sample standard deviations of 36.00) and at emergence is 74. with a small standard deviation of 20. Consistent with the evidence of Altman and Kishore (1996) who examine 696 defaulted bond issues over the period 1978 to 1995. 62. Leisure Time and Media sector. with respective values of 26. Figure 1 plots the time-series variation in the number of firm defaults (corresponding to P d series) and median recovery price at default (P d) in each year. Insurance and Real Estate. it should be noted that while the number While we do not have complete data on recoveries after 1999. we present the industry-based summary for recovery prices at default and at emergence.00. the mean and the median P d are lowest in 1990. respectively. See “Unsecured Bondholders Hit Hardest in 2002 Amidst Declining Recovery Rates. 1953.11 (49.90. the numbers being 97. unsecured bondholders have recovered even less: 28% in 2002 and 22. and Energy and Natural Resources.49 (76. Forest and Building Production and Homebuilders. Telecommunications.1% in 2001 compared to the 15–year mean of 46%. respectively. There is a strong negative relationship between P d and aggregate default intensity (correlation of −0.6%. and Aerospace. However. and 50.risksolutions. This lends justification for examining the time-series variation in recovery rates and its correlation with aggregate default intensity and macroeconomic conditions as a potential dimension of “recovery risk. the mean and the median P d values are 32. There is a clear and a substantial variation in these recovery rates through time. Based on default date recovery data.com).37 (77. Healthcare and Chemicals.27 (49. This coincides with a period of deep recession in the U. Transportation.standardandpoors.82 and 19. Indeed. We include only P ehyld since the patterns are similar for P ecoup and P e. In another instance of this correlation. Auto and Capital Goods industries. a year that again coincides with the sharp increase in aggregate default intensity. global defaults hit a record amount of $157.P ehyld and P ecoup are 51. Our data divides the defaulting firms into 12 industries using S&P’s classification: Utility.

and Junior Subordinated. Though we do not have any bank loans in the recovery data at default (S&P database). Effect of Seniority: In Table 3a. and Junior Subordinated instruments. such DIP loans are not included in our data.41 (58. Senior Subordinated.. the number of firm defaults in this sector has been quite low (8 and 9 for P d and P e data. Comparing the mean recoveries across these different seniority categories and for both types of recoveries. petroleum. respectively). In level terms. and further down to 6. The mean recoveries are not statistically different across the other 11 industries. respectively).25 cents for Junior Subordinated instruments. we find that 11 out of 15 pair-wise means are statistically different at 5% confidence level using Scheffe’s test. especially given the lack of much prior evidence on recovery rates on bank loans and given the somewhat smaller relative variation in recoveries across other seniority classes.78 cents for Senior Subordinated instruments. A notable exception to the lack of research 14 11 .e. we have 358 defaulted loans from 219 defaulting firms in the recovery data at emergence (PMD database). Senior Secured. Senior Unsecured. the Utility sector appears as being different from other industries (perhaps partly due to regulatory issues). to 26. Their sample however does not contain data on recoveries at emergence for Bank Loans. Bank Loans earn on average 30 cents more on a dollar than the next class of seniority. The categories in decreasing seniority are: Bank Loans. it is possible that a part of the higher recovery at emergence on Bank Loans arises from a superior ability of banks and financial institutions to coordinate a reorganization plan for the firm and from their greater bargaining power in the bankruptcy proceedings compared to the dispersed bondholders. Senior Secured instruments. Subordinated. The median recoveries at default decline from 48 cents on a dollar for Senior Secured instruments to around 30 cents on a dollar for Senior Subordinated.14 Note that recoveries at emergence in our sample for public bonds of different seniorities are roughly similar to the numbers reported in Altman and Kishore (1996) and in Izvorski (1997). and related products had average recoveries of 63% in their sample. i. median recoveries at emergence decline from 91. Though banks often provide supra-priority debtor-in-possession (DIP) financing to firms in bankruptcy. Thus.of instrument defaults is large for the Utility sector (55 and 82 based on P d and P e data. but the simple industry classification by itself does not have much power in explaining the cross-sectional variation of defaults. This underscores the importance of seniority of a defaulting instrument as a determinant of its recovery. In terms of recoveries at emergence. In addition to their highest seniority.80). Subordinated. though the Energy and Natural Resources sector does stand out with mean (median) recoveries at emergence of 60. we classify defaulted instruments by seniority. This latter result also is consistent with the findings of Altman and Kishore (1996) who find that chemical. This difference is striking.55 cents on a dollar for Bank Loans.

Note that about two-thirds of our sample (1.0 cents. no collateral data is available for recoveries at default (S&P database). in particular from Moody’s Investors Service (Global Credit Research) cited in Section 2. Instruments backed by All or Most Assets have the second-highest mean (median) recovery of 80. The category of instruments where no information is available on collateral has the lowest mean (median) recovery of 38.19 (98. Current Assets. It is. Other Assets.511).511) and those backed by PPE (83 of 1. that have the highest mean (median) recovery of 94. 12 . across these collateral categories. All or Most Assets. We have verified that most of these instruments are in fact un-collateralized bonds.Effect of Collateral: Finally. For recoveries at emergence (PMD database).05 (89.005 of 1.0 to 72. the instruments backed by liquid. This descriptive summary of our data suggests that contract-specific characteristics such as seniority and security (collateral). Real Estate. Real Estate. Our findings are consistent with this research. In order to develop a more formal model of factors that determine recovery rates.511 defaulting instruments) has no collateral information.91). We conclude that the relevant collateral categories for determining recovery rates are thus sufficiently captured by just liquid Current Assets and Unsecured (un-collateralized bonds) categories. however. Table 3b documents the behavior of P ehyld. we document the behavior of recovery rates as a function of the collateral backing the defaulting instruments. and Secured) have mean recoveries ranging from 63. Unfortunately. and macroeconomic condition (aggregate default intensity).16) cents. Although there is a certain amount of cross-sectional variation in recoveries across these categories. instruments are classified into eight categories depending on type of collateral: Current Assets. Similar numbers are reported for the effect of collateral on recoveries of just bank loans by Carty and Lieberman (1998). the recoveries at emergence. Studies such as Altman and Kishore (1996) and Carty and Lieberman on bank loans is the study by rating agencies on prices of defaulted loans. only the mean recovery for instruments collateralized with Current Assets and the mean recovery for instruments where no collateral information is available are statistically different from the mean recoveries on other collateral categories at 5% statistical significance level using a Scheffe’s test. The Unsecured category corresponds to un-collateralized loans. are likely to play an important role in explaining variation in recovery rates. The other collateral categories (PPE. Among the collateralized instruments. there is substantial variation in recoveries around the means.64 (30.81) cents on a dollar. Secured. Unsecured. Within all categories. Other Assets. we proceed to a multi-variate regression analysis where we exploit firm-specific characteristics as well as industry-specific conditions at time of default. Plants and Property and Equipment (PPE). Unsecured. most common are those backed by All or Most Assets (228 of 1. and Unavailable Information. industry of defaulting firm (utility sector or other sector).

(1998) contain summary data on recoveries with reported magnitudes similar to those in our data. emergence. Cherokee. however. firm. we consider each firm’s debt instruments as a single cluster. etc. That is. statistical analysis of the explanatory power of different variables. In the tables.5 and only eight firms in the whole sample have multiple firm default observations. we report summary statistics of firm-specific. which is one minus the recovery rate. 15 13 . The rest of the firms have defaults of different securities within a 3-month period of the first default and these are counted as being part of a single firm default observation. Heileman. the price of defaulted instruments right after default is a more appropriate measure of recovery. we report only the coefficient on the Utility dummy since other industry dummies do not show up as being significant. We assume the price at default of each instrument is an unbiased estimate of its actual recovery. New Valley Corp. reorganization. Furthermore. TWA. The difference of about 3-month period arises due to the fact that bank loans typically default first followed by bonds. many credit risk models do not explicitly capture the bankruptcy proceeding.. we use ordinary least squares estimates. For these models. In Table 4. Defaults on instruments of the same firm that are separated by more than one year are counted as being parts of separate firm default observations. and macroeconomic characteristics for pooled data that combines the entire time-series and the cross-section of recoveries on defaulted instruments. in their framework. This helps us address the issue that a single bankrupt firm may have multiple defaulted instruments and all these instruments show up in our data as separate observations. The average number of defaulted instruments per firm in our sample is about 4. An alternative interpretation is that for investors who sell their instruments once default occurs. and macroeconomic variables we employ in our regressions. and Zale. and standard errors of these estimates are adjusted for heteroscedasticity using the White (1980) estimator and also adjusted for the existence of firm-level clusters as described in Williams (2000) and Wooldridge (2002). Crucially.15 All our regressions include industry dummies using the classification employed by S&P. they do not examine the industry conditions of the defaulting firm that we show below to be robust and economically important determinants of recovery rates. do not undertake a comprehensive. Greyhound. We visit these summaries in sub-sections corresponding to each set of variables. Caldor. 5 Determinants of Recovery Rates Our primary tests relate the price at default and at emergence to the contract. In all our tests. These studies. this is indeed the relevant measure of recovery. They simply assume a loss given default. industry-specific. The eight firms experiencing such multiple defaults are: Ballys. industry.

14 . then the coupon on the bond will affect the accelerated amount payable to bondholders in bankruptcy. If a bond was issued at par.5. and collateral of the instrument (dummies for Current Assets and Unsecured) or the extent of Collateralized Debt of the firm as a fraction of its total debt. that is. Senior Unsecured. We estimate the specification Recovery = α + β1 ∗ Coupon + β2 ∗ Log(Issue Size) + β3 ∗ Dummy(Bank Loans) + β4 ∗ Dummy(Senior Secured) + β5 ∗ Dummy(Senior Unsecured) + β6 ∗ Dummy(Senior Subordinated) + β7 ∗ Dummy(Subordinated) + β8 ∗ Time in Default + β9 ∗ Maturity Outstanding + [β10 ∗ Dummy(Current Assets) + β11 ∗ Dummy(Unsecured) OR β10 ∗ Collateralized Debt] + β12 ∗ Private Debt + .1 Contract-specific characteristics We first consider the effect of contract-specific characteristics. However. its coupon equals the original issue yield and the accelerated amount would be the par value. Recall that collateral information is not available for data with P d information and there are no Bank Loans in P d data. if a bond was issued at discount or premium. We also include the fraction of the firm’s debt that is Private Debt. some data is lost upon requiring that Coupon. Larger issues may earn higher recoveries as a larger stakeholder may be able to exert greater bargaining power in the bankruptcy proceedings. P d. the Time in Default is included since protracted bankruptcies may result in (or capture) lower debt recoveries. Issue Size. (3) The specification considers seniority of the instrument (dummies for Bank Loans. P ehyld. then the discounted value of remaining promised cash flows would also depend on the Maturity Outstanding of the instrument. Senior Subordinated. a variable that we also include. Senior Secured. Finally. Subordinated). log of the issue size to which the instrument belongs. the yield at which the bonds were issued. we include Log (Issue Size) in our tests. Hence. Though most bonds get issued at par. and recovery at emergence. Furthermore. Table 5 reports the point estimates for some variants of this specification with Recovery being proxied by recovery at default. we also include the time in default (years). Some of these were found to be relevant for recoveries in the summary statistics of Section 4. For recoveries at emergence. and its outstanding maturity. In addition. A common clause in bond indentures is that the accelerated amount payable to bondholders in bankruptcy equals its remaining promised cash flows discounted at the original issue yield. we include the coupon rate on the instrument. we include the Coupon on the instrument to allow for such an effect. Note that if the bond was indeed issued at discount or premium.

and Maturity Outstanding information be available for defaulting instruments. As the point estimates reveal, seniority is statistically and economically an important contract-specific characteristic of recovery rates at default (Columns 1–2) and at emergence (Columns 3–7). The coefficients on dummies for seniority, β3 through β7 , decline monotonically as seniority declines. The coefficients are positive and statistically significant, typically at 1% confidence level (except for the coefficients on Subordinated dummies). The differences between the coefficients for Senior and Subordinated seniorities have also been verified as being statistically significant. On average, the bank loans recover at emergence 20 cents on a dollar more than senior bonds which in turn earn 20 cents more than senior subordinated bonds.16 The extent of private debt in the capital structure of the firm does not appear to have any effect for either the bank loans or the public bonds (Columns 6 and 7). The coefficients on collateral dummies for Current Assets and Unsecured are both statistically significant at 1% confidence level (Column 5). These coefficients are positive (about 14 cents on a dollar) and negative (about 11 cents on a dollar), respectively, mirroring the descriptive evidence that instruments that are collateralized with liquid, current assets recover more than other instruments. The extent of collateralized debt of the firm does not appear to affect the recoveries on defaulted instruments (Column 6), either for secured or unsecured instruments (Column 7). The sign on Coupon is always negative, but the effect is not robust in its statistical significance. This is potentially consistent with higher coupon instruments being more likely to be issued at discounts and thus being discounted at higher yields to obtain the amounts payable in bankruptcy. Another possibility is that coupon is in fact an endogenous variable and higher coupons reflect issuance by weaker credits, which, in turn, gives rise to lower recoveries. Log (Issue Size) and Maturity Outstanding are usually insignificant statistically. In fact, the sign on Log (Issue Size) is not robust across specifications. The fact that Maturity Outstanding is insignificant is consistent with the provision of cross-default clauses leading to payment amounts that are close to par for all obligations of a firm. Since maturity information is available for a small subset of our data (about one-fourth for recoveries at emergence), we exclude this variable in tests that follow. The Time in Default is always significant at 1% confidence level. An additional year in bankruptcy reduces recovery at emergence by about 5 cents on a dollar. Finally, distressed utility firms recover on average about 25–35 cents more at default and at emergence compared to other industries (whose
Another proxy for seniority commonly employed in industry is the extent of debt below or above (in terms of seniority) a specific issue in the capital structure of the firm. In regressions not reported here for sake of brevity, we find this measure of seniority to be statistically and economically significant when it is included as the only proxy for seniority: a greater debt above the issue at hand in firm’s capital structure reduces recoveries on that issue. For sake of parsimony, we do not include in our specifications this additional variable that captures seniority of the instruments in firm’s capital structure in a continuous manner.
16

15

dummy coefficients are statistically not different from zero). Two other observations are in order based on Table 5. First, the explanatory power of comparable specifications (without collateral dummies) for recoveries at emergence is substantially greater than for recoveries at default. The adjusted-R2 ’s are of the order of 40% for P ehyld, whereas they are close to 30% for P d. Second, for all specifications and for all recovery measures (except the fourth column), the intercept term is between 20 to 50 cents on a dollar and is usually significant. This suggests that seniority and security, even though important in determining recoveries on defaulted instruments, are not sufficient in explaining the time-series and the cross-sectional variation in recoveries.

5.2

Firm-specific characteristics

As a first step toward explaining the residual variation in recovery rates, we examine the role played by characteristics of defaulting firms. Since accounting data is difficult to obtain in the year of default, we follow prior literature by using firm level accounting data one year prior to default. This potentially biases our tests against finding any predictive power from firm-specific characteristics. We consider the specification ˆ ∗ Contract Characteristics + γ1 ∗ Profit Margin + Recovery = α + β γ2 ∗ Tangibility + γ3 ∗ Debt Ratio + γ4 ∗ Log(Assets) + γ5 ∗ Q Ratio + γ6 ∗ No. of Issues + γ7 ∗ Debt Concentration + γ8 ∗ Firm Return + γ9 ∗ Firm Volatility + . (4)

In this specification (and in specifications that follow), Contract Characteristics employed are exactly as in the specification of equation (3), except that Maturity Outstanding, Collateralized Debt, and Private Debt are not employed. The corresponding vector of coefficients ˆ. is denoted as β The first five firm-specific characteristics considered in the specification are: Profit Margin, defined as EBITDA/Sales for the defaulting company; Tangibility, proxied by the ratio of Property, Plant and Equipment (PPE) to Assets; Leverage, measured as Long-Term Debt to Assets ratio - we report results only with book leverage as employing market leverage yielded similar results; Log (Assets), the natural log of total assets; and Q Ratio, the ratio of market value of the firm (estimated as book value of total assets − book value of equity + market value of equity) to the book value of the firm (estimated as book value of total assets). From Table 4, the median values of these characteristics for firms in our sample are: 7% profitability, 33% tangible assets, 48% leverage ratio, 6.05 for log of asset size which corresponds to $424 million of assets, and a Q ratio of 0.76 (all in the year preceding default). 16

The next four characteristics employed are: No. of Issues, the total number of issues defaulting for the defaulted company; Debt Concentration, the Herfindahl index measured using the amount of the debt issues of the defaulted company; Firm Return, the stock return of the defaulted company; and Firm Volatility, measured as the standard deviation of daily stock returns of the firm. The median values of these characteristics are: 3.5 number of defaulted issues, 0.40 Herfindahl index of debt concentration among defaulted firms, and a gross (net) stock return of 0.30 (−70%), and a daily firm return volatility of 7% (again, all in the year preceding default). There is substantial cross-sectional variation in all firm-specific characteristics as revealed by the standard deviation and quantiles reported in Table 4. The recoveries on defaulted bonds should depend on the expected value at which the firm gets acquired or merged in a reorganization or on the expected value fetched by the assets of the firm in liquidation. The profitability of a firm’s assets should thus positively affect recoveries: The greater the profitability, the more a potential buyer would be willing to pay for it (all else being equal). Furthermore, many firms default due to liquidity problems and not economic problems per se, so the profitability of assets of defaulted firms prior to default does exhibit substantial cross-sectional variation. We include the firm’s Q to proxy for the growth prospects of the assets which also should affect recovery rates positively. The tangibility of assets also is expected to enhance recovery rates: Intangible assets may be less easily transferrable to acquiring firms and may fetch little or no value in liquidation. We include the leverage of the firm to capture the possibility that bankruptcy proceedings of high-leverage firms may be more difficult to resolve: Higher leverage may be associated with greater dispersed ownership requiring greater coordination among bargaining parties. In a similar spirit, firms with greater number of issues and more dispersed creditor base, that is, lower debt concentration, may experience greater coordination problems and in turn lower recovery rates for creditors.17 We also consider the size of the firm, its total asset base, in order to allow for potential economies or diseconomies of scale in bankruptcy. On the one hand, a part of bankruptcy costs may be fixed in nature giving rise to some scale economies; on the other hand, larger firms may be difficult in terms of merger and acquisition activity giving rise to greater ongoing concern value and lesser value in a reorganization or a liquidation. We also include the firm’s stock return to proxy for its financial health: Firms with poorer stock returns prior to default, controlling for firms’ Q ratios, should be closer to
Shleifer and Vishny (1992) show that the expected recovery rates should affect the ex-ante debt capacity of firms and industries. Bolton and Scharfstein (1996) argue how the number of creditors can be optimally chosen by a firm to trade off strategic defaults by management (equityholders) against the costs of default resulting from liquidity shocks. We do not model this endogeneity aspect of firm leverage. Building a model that simultaneously explains design of leverage and recovery rates of firms is indeed a worthy goal to pursue but beyond the scope of the current paper.
17

17

The magnitudes of coefficients on these characteristics are mostly not different from zero at 5% confidence level. measured as Cash and Cash Equivalents.25 to 0.binding financial constraints. do not appear to be important determinants of recoveries at default. A few firm-specific characteristics show up as being significant in affecting recoveries at default (Table 6a). not a single firm-specific characteristic shows up as a significant determinant of the recoveries at emergence (Table 6b). Firm leverage. however. There is little improvement in the We employed the cash position of the firm. Consistent with our hypothesis.35 cents on a dollar. The greater the volatility of firm’s total asset value. respectively. Overall. is significantly enhanced with the addition of firm-specific variables. 18 18 . The economic magnitude of these effects is however small: For example.18 Somewhat surprisingly. debt design measures show up as significant and with signs that are consistent with starting hypotheses. All else being the same. Instruments of firms with greater number of issues in debt structure and with more dispersed ownership of debt experience lower recoveries suggesting greater bankruptcy and/or liquidation costs arising from coordination problems between creditors. It should be noted however that stock return volatility is an imperfect proxy for firm volatility. This variable did not have any explanatory power for recoveries either. in the specification. and Q. but available upon request. size. there is a mechanical increase in equity volatility as the firm approaches default even if firm volatility remains unaffected. The adjusted-R2 for P d specifications in Table 5 is about 30% whereas in Table 6a it is close to 50%. we include the firm’s stock return volatility. We also employed the following variable: Cash minus the amount of debt in the capital structure that is senior to the defaulted instrument. we find that firms with greater profitability of assets have greater P d. P d. Finally. then the effect of tangibility of assets is positive and statistically significant (results are available from the authors upon request). Finally. Noticeably. the explanatory power of the regression for recoveries at default. In particular. These results are excluded for sake of parsimony. the marginal effect of a percentage point increase in EBITDA/Sales margin one year prior to default is between 0. The qualitative nature of effects for contract-specific characteristics is similar to the results in Table 5. it does not show up as a significant determinant of recovery. tangibility of assets has little effect on recoveries at default. an additional issue depresses recoveries by less than a cent on a dollar. If the Utility dummy is not included. In contrast. Table 6a and 6b report the point estimates and standard errors for the specification in equation (4) for recoveries at default and emergence. and thus have lower recovery rates. After controlling for the Utility dummy. neither as Cash itself nor as Cash by Assets. a greater stock return for the firm in the year prior to default is associated with better recoveries. and a greater stock return volatility with lower recoveries. the lower would be the anticipated value of creditor claims when bankruptcy is resolved.

and the intercept term in most specifications of Table 6b is positive and statistically significant. Andrade and Kaplan (1998). suppose the debtors can make a take-it-or-leave-it offer to creditors (Anderson and Sundaresan.explanatory power for P ehyld. (3) Poor liquidity position of industry peers when a firm defaults should lower the recovery on its debt. Gertner and Scharfstein (1994). and Mella-Barral and Perraudin. 5. It should be noted though that for many firms in our sample the emergence and default year are the same (median Time in Default is about 1. In particular. acquired. However. The studies of Asquith. Such strategic behavior would manifest itself as 19 . 1996. We believe that this approach has some advantages over studies that look only at actual asset sales. the firm would be liquidated.S. but the creditors receive simply their expected value from the scenario where the firm is liquidated. The firm may eventually get reorganized. and Pulvino (1998) test the implication of the Shleifer-Vishny model on asset sales and firm values. Asset sales through cash auctions when most of the industry is in distress would fetch prices that are lower than the economic worth of assets if assets cannot be deployed easily by firms outside the industry. (2) Companies that operate in more concentrated industries should have lower recoveries due to the lack of an active market of bidders. 1997). these are “stale” by the time of emergence. To the best of our knowledge. even when such asset sales do not actually occur. something not tested in the current literature. or merged in the bankruptcy. we test the implications of the model for debt recoveries. A lower asset value in liquidation should translate into lower debt recoveries.5 years). Thus.. the first-mover advantage enables the debtors to strategically offer to creditors only the value that the creditors would receive if the firm’s assets were liquidated. looking at debt recoveries enables us to understand the magnitude of losses expected from distressed asset sales. Then. How would this happen? In Chapter 11 in the U. the Shleifer and Vishny model gives rise to the following implications for debt recoveries: (1) Poor industry or poor macroeconomic conditions when a company defaults should depress recovery rates. For sake of argument. debtors have the first-mover advantage in filing a reorganization plan. If the take-it-or-leave-it offer is rejected by creditors. (4) Industries that have more specialized assets (those that have few alternative uses) should have lower recoveries in the event of default.3 Industry characteristics Shleifer and Vishny’s(1992) model provides the theoretical insight that financial distress is more costly to borrowers if they default when their competitors are experiencing cash flow problems. no study has examined directly the implications of this model on debt recovery rates in the event of default. One plausible explanation is that since these characteristics are measured one year prior to default.

and δ5 < 0. United’s advisers argue it is in a strong negotiating position because of the weak market for used aircrafts. our sample of defaults covers more than 600 instrument defaults (for emergence prices). we include only Profit Margin and Debt Concentration as the firm-specific characteristics. According to the hypotheses.violations of absolute priority rule (APR). December 13.. 2002). general creditors have even less bargaining power than lessors do. by Robert Clow in New York. Financial Times. we expect that δ1 < 0. and such effects should magnify their losses. the industry of a defaulted firm is identified as the set of firms with the same 3-digit SIC code as the defaulted firm. The US airline believes it can slash its costs by renegotiating its $8bn of aircraft leases that are spread among 300 companies. δ4 > 0.20 The defaulted firm is excluded from calculation of industry variables. are facing billions of dollars of losses on United Airlines leases. (5) Note that γ ˆ is the vector of coefficients on firm-specific characteristics. ranging from Walt Disney to Pitney Bowes and DaimlerChrysler. Firms in industries with high leverage may be closer to being financially constrained or may find taking on more leverage costly. Following the literature. We believe the comprehensive nature of our data has the potential to shed additional light on the effect of industry conditions on defaulted firms. If the 3-digit SIC code of a defaulted firm does not include at least five other The recent example of the bankruptcy of United Airlines illustrates this point well: “Some of the US’s leading companies.. documented in Lopucki and Whitford (1990). and Franks and Torous (1994). Weiss (1990).. It plans to send revised terms to leaseholders over the next three days. Based on the results in Table 6. All industry variables are computed using data from CRSP and COMPUSTAT and are measured contemporaneous to default. including Ford and Philip Morris. In order to test the first three hypotheses. Hence. more than 200 instrument defaults (for default prices). we include only a representative specification where Firm Volatility is also included. δ3 < 0. 19 20 . that is. 20 We were not able to classify any industry as distressed if we measured industry conditions at the time of emergence of the defaulted firm.” (US groups face UAL lease losses. Requiring stock market data reduces sample size which is detrimental to finding any effect from industry distress (which occurs in only 10% of our data). Indeed. we estimate the following specification: ˆ ∗ Contract Characteristics + γ Recovery = α + β ˆ ∗ Firm Characteristics + δ1 ∗ Dummy(Industry Distress) + δ2 ∗ Median Industry Q + δ3 ∗ Industry Concentration + δ4 ∗ Industry Liquidity + δ5 ∗ Median Industry Leverage + . and spans a period of about two decades from 1982 to 1999. Tangibility is included to verify that the result of Table 6 that tangibility of assets does not affect recoveries is not due to omitted variable bias. Note that median industry leverage also could be employed as a proxy for difficulty in raising new financing.19 Finally. in the year of default..

We employ only one of Distress1 or Distress2 at a time in a specification. Table 7b contains the results for recoveries at emergence. defaulting companies whose industries also have suffered adverse See. In Table 7a. 3. and Asquith. This dummy is Distress1. the number being 3% when classified on the basis of Distress2.91 for Q ratio. Industry Concentration is proxied by the sales-based Herfindahl index of all other firms in the industry. First. these industry distress dummies take on the value of 1 for about 10% of the sample size in terms of defaulting instruments. an alternative definition of industry distress: In addition to Distress1 being one. We find that when the defaulting firm’s industry is in “distress. As Table 4 reveals.21 Finally. Median Industry Leverage is the median Long-Term Debt to Assets for all firms in the industry.Inventories] to Current Liabilities. We construct two versions of dummy variables for Industry Distress. Both measures are frequently employed in empirical corporate finance to proxy for industry liquidity conditions.” as defined by Distress1 (based on median stock return for industry). First. Distress2 requires that one year or two year median sales growth for the industry in the year of default or the preceding year (based on data availability) be negative. Second.17 for Industry Concentration. and 18% industry leverage. 0. as one would expect for defaulting firms. the median values of these variables for industry-year pairs in our sample are: 0.99 for Quick ratio (Industry Liq1). we define an industry to be “distressed” if the median stock return for the industry of the defaulting firm in the year of default is less than or equal to −30%. Industry Liq1 is the median Quick ratio. 0. defined as the ratio of [Current Assets . Industry Liq2 is the median Interest Coverage ratio. the median being taken over all other firms in the industry. The median firm Q (leverage) is substantially smaller (greater) compared to the respective medians at industry-level. measured as EBITDA/Interest. Median Industry Q is the median of the ratio of market value of the firm (estimated as book value of total assets − book value of total equity + market value of equity) to the book value of the firm (estimated as book value of total assets). Distress2 is thus based on stock market performance of the industry as well as on the book measure of industry performance. its instruments recover about 10–12 cents less on a dollar compared to when the industry is healthy. 13% of industry-year pairs have distressed industries classified on the basis of Distress1. We employ either Industry Liq1 or Industry Liq2 in a given specification. This statement holds for both recoveries at emergence as well as recoveries at default. In our data.firms. we report the results from estimation of equation (5) for recoveries at default.08 for interest coverage ratio (Industry Liq2). then we do not include the observation in the tests. From Table 4. for example. in the spirit of Gilson. Gertner and Scharfstein (1994) and Andrade and Kaplan (1998) for the use of Interest Coverage ratio. John. We also consider Distress2. The median is taken over all other firms in the 3-digit SIC code of the defaulted firm. Stromberg (2001) for the use of Quick ratio. Industry Liquidity is proxied using two measures. and Lang (1990) and Opler and Titman (1994). 21 21 . Thus.

Thus. This measure is less likely to embed expectations of future profitability. In this case. the coefficient on Profit Margin is positive and significant for P d. and generally statistically significant. it recovers between 16–20 cents on a dollar less at default compared to the case when the industry is healthy. We address this issue along three dimensions: First. Third. there is a residual effect of industry distress in the Distress2 dummy. as the dummy variable for industry distress. and in Table 7b. then the positive linkage between industry conditions and recoveries should be symmetric: When industry is doing well. This is further confirmed when we use Distress2. An alternative hypothesis is that a very high negative median stock return for the industry may arise also if assets of this industry are not expected to be profitable in the future. Senior Debt vs. Second. the coefficient of Median Industry Q is positive and at least marginally significant in most specifications. controlling for Profit Margin and Median Industry Q. debt recoveries should be higher. Distress2 dummy is based not only on stock returns but also on the sales growth for the industry which is a book measure of the industry’s condition. when industry is not doing 22 . The coefficients on distress dummies are negative in both Tables 7a and 7b. the median stock return for industry may in fact proxy for expected profitability of assets of the defaulting firm. The effect of Distress2 on recoveries at emergence is weaker statistically. and mostly at 1%. The effect of Distress2 on P d is even stronger than that of Distress1. Our assumption is that median industry Q proxies for future growth prospects of the industry and in turn of the defaulted firm’s assets. though still of the order of 7–10 cents on a dollar. Subordinated debt). based also on median sales growth for the industry. Thus. This counterargument would generate a negative coefficient on Industry Distress dummies without any role for conditions of peer firms of the defaulting firm. We have also included in the specification Median Industry Q. if the defaulting firm’s industry is in distress. we have employed Profit Margin of the defaulting firm one year prior to default. examining the coefficient of distress dummies in the presence of firm-level and industry-level profitability measures helps isolate the effect on recoveries due to financial and economic distress of the peers of the defaulted firm. That is. if industry conditions merely capture the expectation of future growth prospects. We believe this provides support for the first hypothesis that poor industry conditions when a company defaults depress recovery rates on the defaulting company’s instruments. The effects discussed above are all statistically significant at 5%. specifically about twice as large.economic shock face significantly lower recoveries. The magnitude of the effect is as high as the relative effect of seniority of the instrument (Bank Loans vs. This assumption is borne out in the estimates: In Table 7a. even if one were to attribute the entire effect of Distress1 dummy to expectations about future growth prospects. a very high negative return generating lower recoveries simply because defaulting firm’s assets are fundamentally not worth much.

14.6 for P e and P ehyld. all differences being statistically significant with p-values close to zero. we find that the difference in recoveries between no industry distress years and industry While we do not report these numbers. the NBER recession year. 1990. Auto and Capital Goods in 1990. Aerospace. and 6b). This evidence implies that the effect of industry distress on recovery rates is orthogonal to that of contract-specific and firm-specific characteristics. The alternative z-statistics for Wilcoxon rank sum test between no industry distress and distress samples also have a p-value close to zero. 1994. and other industry-specific characteristics. Financial Institutions in 1987. 1990 and 1991. and 1998. firm-specific. we find that the standard deviation of P d and P ehyld are of the same order between no industry distress and industry distress years.22 Interestingly. Building Products and Homebuilders in 1990. we identify in Table 8a the industries that experience distress based on Distress1 and the year in which they do so. the effect of industry return on recoveries is always non-linear and restricted to situations where the industry is in distress. This is more consistent with the Shleifer and Vishny hypothesis than with the alternative hypothesis. Consumer and Service Sector in 1990. We find that controlling for Median Industry Q. This is to confirm that our results are not driven by just one year of economy-wide distress in which recoveries were skewed toward zero. 1995. 1994 and 1995. in which nine out of the twenty-three industry distress events occur. In Columns 7 and 8 of Tables 7a and 7b.6 for P ecoup. 6a. In Panel A of Table 8b. we drop industry distress dummies and instead include median stock return for the industry as a continuous. Transportation in 1984 and 1990. the level of median industry return has no explanatory power at all for debt recoveries. and Energy and Natural Resources in 1986. This year was also special in that a large number of defaults occurred in the aftermath of the leveraged buy-out (LBO) phenomenon. un-truncated variable. and 12. 1993. Forest.5 cents on a dollar for P d. High Technology and Office Equipment in 1990. the coefficients on contract-specific and firm-specific characteristics in these regressions are of similar magnitudes as in the earlier tables (Tables 5. debt recoveries should be lower. That is. Furthermore. we examine the difference in recoveries between no industry distress years and industry distress years where we exclude 1990. Healthcare and Chemicals in 1987. the magnitudes of the differences based on non-parametric tests are close to the ones implied by the coefficients on Distress1 and Distress2 in the parametric regressions of Table 7 where we employ contract-specific. In Panel B of Table 8b.well. Leisure Time and Media in 1990. 1995 and 1996. The difference is 19. To examine this effect of industry distress with a microscope. The table shows the 23 industry-year distress pairs: Insurance and Real Estate sector experienced distress in 1990 and 1994. 22 23 . Except for P d. we examine non-parametrically the difference in recoveries (measured in different forms) between no industry distress years and industry distress years.

Furthermore. It must be observed though that excluding 1990 leaves only 10 instrument observations in the P d sample of industry distress years. the coefficient on Industry Liquidity is mostly insignificant for P d in terms of statistical confidence. Note that the difference in P d. The effect is.23 Continuing our examination of Tables 7a and 7b. If an industry is in distress. What is crucial is whether the industry of the defaulting firm is itself in distress or not. but finds strong support in Table 7b for recoveries at emergence. In terms of explanatory power.1 for P e. This result is in contrast to the findings of Izvorski (1997) who documents a positive relationship between industry concentration and recoveries for a set of 153 bonds that defaulted in the United States between 1983–1993. across firms. and 12. both industry liquidity proxies have positive and significant effect on P ehyld. The coefficient on revenue-based Herfindahl index for the defaulting firm’s industry is always negative for recoveries at emergence as well as at default. 13. The coefficient is negative for Industry Liq1 and positive for Industry Liq2.distress years (excluding 1990) is of similar magnitude as in Panel A which includes 1990 in industry distress years: 17. Finally. Nevertheless. the recoveries for firms defaulting in this industry are significantly depressed even when the overall economy is not in distress or recession. however. finds little support in Table 7a for recoveries at default. This illustrates that it is not per se the existence of an economy-wide recession year which depresses recoveries at emergence. the sign of the coefficient is not robust. The liquidity hypothesis is thus supported to some extent.4 for P ecoup. The total variability in recoveries that is to be explained consists of cross-sectional variability (i. we find little evidence supporting the second hypothesis that industry concentration lowers the recovery rates.05 for all the specifications. We find that controlling for the industry distress condition and industry concentration. the price at default. the coefficient on Median Industry Leverage is insignificant for both P d and P ehyld.e. that poor liquidity position of industry peers when a firm defaults should lower the recovery on its debt. In contrast.1 for P ehyld. we find the effect of lowering industry illiquidity by one standard deviation is to depress the recovery at emergence by about 5 cents on a dollar. all differences being statistically significant with p-values close to zero. 23 24 .. The smaller incremental power for industry-specific effects is an artifact of the pooling of data in the cross-section as well as in the time-series. Izvorski considers his finding a “puzzle” since it is opposite to the theoretical priors based on Shleifer and Vishny (1992). Using Table 4 and the coefficient estimates. between no industry distress and distress years is close to zero when 1990 is excluded from distress years. never statistically significant. but the effect is not as strong or robust as the effect of industry distress. The third hypothesis. F-tests performed to check that industry-specific effects are jointly significant have p-values less than 0. note that the incremental explanatory power of industryspecific characteristics (over and above the explanatory power of contract-specific and firmspecific characteristics) is relatively small.

and Swary (1996) and Stromberg (2001). we employ industry-level asset-specificity. since industry distress dummy takes the value of one in only 10% of our sample. Since firm-level asset-specificity is available for only a small fraction of our sample. We confirm this result non-parametrically as well. which is small in magnitude compared to the cross-sectional variability in recoveries. We calculate the specificity of an industry using the mean (or the median) of Specific Assets of all firms in that industry and over the entire sample period. Panels A and B of Table 8c provide the time-series mean and median of Specific Assets for the twelve industries. creditors in an industry that is more asset-specific than a benchmark industry by 10% experience recoveries that are lower by 16 cents on a dollar compared to the creditors of the benchmark industry. When in distress. The correla25 . there is little power to identify the interaction with the sample of recoveries at default (about 250 data points). Panel B reports P ehyld. and Financial Institutions. Finally. Consistent with the starting hypothesis. this requires interacting the industry distress variable with firm-level or industry-level asset-specificity. In a regression specification. for industry-year pairs with no distress and with distress. As in Berger. The correlation coefficients have unstable signs and are not statistically significant. The estimates with the interaction term are reported in Columns 8 and 9 of Table 7b. Ofek. and collateral classes) as well as time-series variability. Telecommunications (30%). the interaction term is negative and statistically significant. Hence. So far we have not included a measure of asset-specificity to test the fourth hypothesis that firms in industries that have more specific assets (with few alternative users outside of the industry) should have lower debt recoveries. we define Specific Assets of a firm as the book value of its machinery and equipment divided by the book value of total assets. The four most asset-specific industries are Transportation (44% median Specific Assets).seniority. Also. where the industry dummies are replaced by industry-level asset-specificities. P e. Panel C correlates industry-level recovery with industry-level asset-specificity for industry-year pairs when industry is in distress in a year and when it is not in distress. Panel C illustrates that there is no correlation between recovery and asset-specificity in industry-year pairs when the industry is not in distress. The effect is of similar magnitude whether we employ industry’s asset-specificity calculated as mean of firm-level asset-specificity ratios (Column 8) or as median (Column 9). Energy and Natural Resources (25%). have close to zero asset-specificity. Panel A of Table 8c also reports mean and median recovery at emergence. Insurance and Real Estate. we focus on recoveries at emergence (about 750 data points). and Consumer and Service sector (24%). The effect of whether the industry is in distress or not explains well the time-series variability in recovery rates. whereas all other industries have moderate asset-specificity lying between 14% and 17%.

Computers. the industry variables motivated by Shleifer and Vishny (1992) explain well the time-series variation in recovery rates.g. We believe this finding is difficult to reconcile with the alternative hypothesis that debt recoveries are low during periods of industry distress simply because the fundamental worth of assets has gone down: it is not clear why this effect should be sensitive to the assetspecificity of industries. and Energy and Natural Resources) experience a significant drop in debt recoveries (about 30 cents on a dollar) when they are in distress relative to their no-distress levels. 25 This is consistent with the regression results in Columns 8 and 9 of Table 7b where the effect of industry distress is seen to be positive at low levels of asset-specificity. 24 26 . the correlation between mean (median) P ehyld and mean (median) asset-specificity is −0. Industries with highly specific assets (e. Exodus and PSI–Net) were unable to sell most of their core telecom assets and their creditors recovered value only from the sale of office-space and other such non-specific assets.24 It should be noted that for Financial Institutions. and Office Equipments.25 This is due to the small sample size of defaulted firms in distress years for these industries. Transportation. From Table 4.90 (−0. when Financial Institutions and High Technology. there is no data in our sample period for Utilities and Telecommunications when these industries are in distress. The corresponding correlations for no industry distress years are insignificantly different from zero. the total face value amount of defaulted bonds in a year measured at mid-year and in trillions of dollars. We conclude our pursuit by examining macroeconomic conditions which if poor also may depress recoveries. Brady. In particular.. In particular. except for industry concentration. These variables are: BDR.76) for mean (median) recovery and mean (median) asset-specificity. Table 8c). It is striking to note though that 1999–2002 constituted years of industry distress for Telecommunications sector and they were also characterized by extraordinarily low debt recoveries for defaulted telecom firms. and for High Technology.69) for industry distress years.tion is however significantly negative in industry-year pairs when the industry is in distress: −0. Overall.g. are excluded. and Sironi (2000) to be significant in explaining the time-series of average annual recoveries. and BDA. we see that both BDA and BDR are highly skewed Unfortunately. and Office Equipments.4 Macroeconomic and bond market conditions We examine the bond market variables shown by Altman. the aggregate weighted average default rate of bonds in the high yield market where weights are based on the face value of all high-yield bonds outstanding in the year. 5.. recoveries are actually higher when these industries are in distress compared to when they are not in distress. many firms (e. Computers. Resti. The correlation patterns are qualitatively unaffected by the exclusion of these two industries (Panel D.63 (−0. consistent with the first hypothesis.

is associated with a lowering in aggregate recoveries of 4. a response to increased supply meeting a generally shallow.5bln in 1998 and $63. Similary. in the Altman et al. the aggregate defaulted amount was $7. In particular. the aggregate default rate. hypothesis) that causes recoveries to be low during periods of economywide distress. illiquid market. median face value of defaulted bonds in a year is about 4 billion dollars with a maximum of 23. the latter being driven by large number of defaults in telecom. the supply of defaulted bonds. median aggregate default rate is about 2% reaching a maximum value of 10%. present another hypothesis which is that such a negative effect may capture supply conditions in the defaulted bond market: The set of investors participating in the defaulted bond market is segmented and limited to vulture funds. For example. Risk Solutions. Size (Small Minus Big). in fact do find such an effect. January 2002) 26 27 . the aggregate default rate is 1. Similarly. Similarly.6% in 1998 and 9.6% in 2002.3: Poor macroeconomic conditions reduce the ability of potential buyers to pay high prices for these assets. hedge funds. Market. (2002) data. Altman et al. and the three Fama and French factors. the S&P 500 stock return for the year.5 billion dollars (in 1999). USD increase in BDA. 27 This is also the perceived wisdom in some industry literature concerning the depressed prices of defaulted securities in 2001–2002 period (a NBER recession period): “As the huge volume of defaulted debt floods the market. Standard & Poors. They find that BDR and BDA (and their logarithms) affect average annual recoveries significantly and negatively. we estimate the specifications The time–series variation in BDR and BDA is also quite large.27 We attempt to disentangle whether it is the illiquidity in the market for real assets (Shleifer and Vishny hypothesis) or the illiquidity in the market for defaulted securities (Altman et al.5%. then the negative effects of these variables on recovery rates would be consistent with the first hypothesis discussed in Section 5.6bln in 2002.26 If we interpret high values of BDR and BDA as capturing adverse macroeconomic conditions. the annual Gross Domestic Product growth rate. In addition to BDR and BDA. trading prices for distressed debt have become depressed. we capture comprehensively the extent of macroeconomic risk in the year of default by examining the effect of SR. and Book-to-Market (High Minus Low). However. Dropping for Poorly Structured Debt. a 1% increase in BDR. and steel sectors. airlines. GDP. and high net-worth individuals.variables.” (Ultimate Recovery Remains High for Well-Structured Debt. is associated with a reduction of 4% in aggregate recoveries. high-yield desks of banks and financial institutions. A greater supply of defaulted bonds for a limited demand could imply that the prices on defaulted bonds must fall in order to clear the markets. a 10 bln. obtained from the Web-site of Ken French and computed using the procedures described in Fama and French (1993). Using these variables. Altman et al.

significant usually at 5% level. For ease of reporting. is in striking contrast to their results. stock market return. While SR and GDP do not show up as significant also in the results of Altman et al. recoveries at default. Distress1. and economy-wide risk factors. the macroeconomic and bond market conditions are not significant determinants of recoveries. in Table 9b we run the same specifications as in Table 9a but without any industry variables (industry dummies are still included). on P d and P ehyld. USD increase in BDA in Columns 1 and 3 gives rise to a reduction in recoveries of about 7%. examine annual average recovery rates. GDP growth rate. a 10 bln. using the NYU Salomon Center data on the closing “bid” prices for about 1300 bonds (as close to the default as possible) over the period 1982–2002. the coefficients on seniority dummies are not reported in this table. SR. have no explanatory power for either P d or P ehyld. For example. BDR and BDA. As the table reveals. GDP. However. P ehyld (Columns 3–5). are rendered insignificant. (2000) results are for P d. and the effect of aggregate default intensity BDR is negative and significant for P ehyld. P d (Columns 1–2). the insignificance of bond market conditions. once industry distress effects are controlled for in Table 9a. and the Fama and French factors do not appear to have any incremental explanatory power either. To examine this issue further. Even in the absence of industry variables. and Industry Liq1) are controlled for. In contrast. It is important to recognize that in our regressions. Median Industry Q. once industry dummies and industry conditions (Distress1. we control for the effect of industry conditions and also control for industry dummies. in Table 9a the bond market conditions do not drive out the effect of industry distress. The magnitude of these effects in Table 9b is also comparable to those in Altman et al.ˆ ∗ Contract Characteristics + γ Recovery = α + β ˆ ∗ Firm Characteristics + ˆ ∗ Industry Conditions + δ θ1 ∗ SR + θ2 ∗ GDP + θ3 ∗ BDA + OR θ1 ∗ BDA + OR θ1 ∗ Market + θ2 ∗ SMB + θ3 ∗ HML + . the effect of aggregate defaulted bond supply BDA is negative and statistically significant for both P d and P ehyld. we report the point estimates and the standard errors from estimation of equation (6) for recoveries at default. Our sample of P d data is much smaller. such conditioning is not possible.28 In contrast. and of the One possibility we must entertain is the following. Since Altman et al. Altman et al.. The effect of Distress1 in the presence of macroeconomic and bond market conditions is negative. and recoveries at emergence. these coefficients fall by between 20% to 40%. and in particular. (6) In Table 9a. 28 28 .

We conclude that industry conditions are an essential ingredient of a specification that explains well the recoveries at emergence. The effect of industry liquidity. the intercept terms for P ehyld are not statistically different from zero. It is difficult to conclude that illiquidity in the financial markets for trading defaulted instruments causes lower recoveries: The more likely candidate is in fact the illiquidity in the market for asset sales. our results suggest that the linkage between bond market conditions and recoveries stressed by Altman et al. First. The Z–score we employ is as defined in Altman (1968.order of 10–13 cents on a dollar as before. 6 Are Determinants of Recovery Risk and Default Risk Identical? We examine whether ex-ante measures of likelihood of default of a firm. and each set has incremental power in explaining observed recovery rates in the United States over the period 1982–1999.2 ∗ Working Capital)/Assets.3 ∗ EBIT + Sales + 1. we examine three predictors of default risk of a firm employed in the literature and in practice. Seniority and security of defaulted instruments help explain the cross-sections of recoveries at default as well as at emergence. Recoveries at default and at emergence are both affected significantly by the industry condition when a firm defaults (distressed or healthy).4 ∗ Retained Earnings + 1. Finally. as arising due to supply-side effects in segmented bond markets may be a manifestation of omitted industry conditions. we employ the Z–score employed in credit-scoring models by rating agencies. and by the asset-specificity of the industry when that industry is in distress. 2000) and as modified by Mackie-Mason (1990): Z = (3. Profitability of assets of defaulting firms and concentration of their debt structure also help explain the cross-section of recovery rates at default. affect recovery rates or not. recoveries at emergence are affected adversely by the illiquidity of peer firms in industry of the defaulted firm and by the length of time the firm spends in bankruptcy. by the type of industry (utility or not). 5. the intercept terms for P ehyld regressions in Table 9b are positive and statistically significant. found to be important by extant empirical literature. In particular. Industry Liq1. whereas in Table 9a. Finally.5 Summary We conclude from the results of Tables 5–9 that the following factors play an important role in explaining recovery rates. on P ehyld is also unaffected in magnitude and significance.(7) 29 . These sets of factors do not subsume each other. Also.

7 Implications for Credit Risk Models Our results from Sections 5 and 6 put together show that while determinants of default risk and recovery risk are correlated. (8) Finally. We estimate the specification ˆ ∗ Contract Characteristics + γ Recovery = α + β ˆ ∗ Firm Characteristics + ˆ ∗ Industry Conditions + δ ω ∗ Z–Score or Zmijewski Score or Distance to Default + . and asset-specificity are factors that seem to affect recoveries over and above factors that affect default risk. based on the Merton (1974) model.7 ∗ Total Debt/Total Assets − 0. industry liquidity. How do these factors affect inputs of recovery rates in existing credit risk models? Seniority and collateral.kmv. Zmijewski Score for P ehyld.com) using stock returns and stock return volatility of a firm. and also concentration of debt structure.004 ∗ Current Assets/Current Liabilities. they are not perfectly correlated. the Zmijewski Score. (9) Note that since the determinants of default risk are also based on firm-specific characteristics. a variant of the Distance to Default measure. We have employed (but do not report the results for) the Expected Default Frequency (EDF). we also employ the Distance to Default as computed by KMV (www. as defined in Zmijewski (1984): Zmijewski Score = −4. The estimates are reported in Table 10 for recovery at default (Columns 1–3) and at emergence (Columns 4–6). industry distress. debt concentration. time in default.3 − 4. we consider another credit-scoring model from the accounting literature. The determinants of default risk are in general also significant as determinants of recoveries: Z–Score for both P d and P ehyld. and Distance to Default for P d.5 ∗ Net Income/Total Assets + 5. could be captured by allowing a constant recovery rate but one that varies depending on the firm and the 30 .Second. Seniority and collateral. This lets us capture cleanly whether determinants of likelihood of default are also determinants of recoveries or not. concentration of debt structure. we only include Debt Concentration among these variables. The effects of seniority of instruments. The exact computation of these measures is described in Appendix A. time in default. and industry conditions remain overall unaffected from the presence of the determinants of ex-ante default risk.

in contrast to our finding that these risks are positively correlated but not perfectly so. In a recent contribution to the limited literature that considers modeling recovery risk and associated risk premium. Their model however limits recovery risk to being determined completely by interest rates. It is not fully clear that uncertainty about time in default is a systematic risk and thus may also be reasonably captured in an average recovery rate. Our results thus underscore the need for modeling recovery risk as stemming from firm-specific factors as well as systematic. (2000) find that 1–year and 10–year treasury rates do not help explain aggregate annual recovery rates. To the best of our knowledge. distressed or healthy. the state of the industry of the defaulted firm. it is the risk of an industry recession rather than an economy-wide recession which is the primary driver of recovery risk. and the level of senior debt appear to be significant determinants of the price of recovery. Their model however assumes that default risk and recovery risk are independent. In this regard. As our tests reveal. industry-specific factors. firm tangible assets. Jarrow (2001) also allows likelihood of default and recovery to be correlated and based on the state of the economy. Another possibility is to analyze in general equilibrium or asset-pricing frameworks the risk-premia arising from the industry effect. Guntay. Such an exercise would be valuable in understanding the implications of industry-driven recovery risk for prices of credit risky instruments. 2000b) capture in a reduced form the correlation of default risk and recovery risk as potentially arising from the risk of recessions. It constitutes a dimension of recovery risk that may in fact carry a risk-premium to it given its systematic nature.instrument (assuming debt structure does not change dramatically during the life of the instrument). They demonstrate that interest rates. Building a next generation of credit risk models that embed industry-specific factors thus appears to be a fruitful goal to pursue. is certainly a systematic risk factor. Das and Tufano (1996)’s assumption where interest rate risk affects credit risk as well recovery risk is more consistent with our findings. Altman et al.29 The models of Frye (2000a. Madan and Unal (2003) propose an approach to infer the risk neutral density of recovery rates implied by prices of debt securities of a firm of differing seniority. However. such an integrated credit risk model does not yet exist either in the structural class of credit risk models or in the reduced form variety. He then uses equity prices to identify separately these two components of credit risk. 29 31 . from the risk of low recovery or the risk of asset fire-sales when firms receive common shocks. that is. and our empirical work should provide guidance concerning the additional factors to introduce in these models.

and Martin (2001). 32 . Finally. our results on industry-level asset-specificity and recoveries could also serve as useful benchmarks for future research attempting to link ex-post recovery outcomes to ex-ante capital structure of firms. such as Collin-Dufresne. along the lines suggested by Shleifer and Vishny (1992). and to understand the variation in leverage across industries and over the business cycle. For example. Our main finding is that whether the industry of the defaulted firm is in distress or not is a robust and economically important determinant of recovery rates. Goldstein. A more complete analysis of credit spread changes is perhaps called for. studies that examine the determinants of credit spread changes. have not accounted for the rich cross-sectional and time-series variation in recoveries found in our study. especially one that employs some of the industry factors identified by us as important determinants of recovery rate changes. We hope our research will serve as the empirical benchmark for recoveries on different kinds of debt and in different conditions.8 Conclusions We have provided a comprehensive empirical analysis of recovery rates on defaulted loans and bonds in the United States over the period 1982–1999.

Almost Everything You Wanted to Know about Recoveries on Defaulted Bonds.. Kishore. 2002. Madan. Quarterly Journal of Economics. New York University. Edward. Scharfstein. 1996. Nov/Dec. Financial Ratios. 2000. 57–64. Working paper. Andrea Resti and Andrea Sironi. How costly is Financial (Not Economic) Distress? Evidence from Highly Levered Transactions that Became Distressed. Understanding the Role of Recovery in Default Risk Models: Empirical Comparisons and Implied Recovery Rates. The link between Default and Recovery Rates: Theory. Kaplan. 37–68. Altman. 2003. Edward. Review of Financial Studies 9(1). Sundaram.. 2002. August. Altman. Altman. Salomon Smith Barney United States Corporate Bond Research High Yield. forthcoming. Anatomy of Financial Distress: An Examination of Junk-Bond Issuers. Stern School of Business. 589–609. Altman report on Defaulted Bonds and Bank Loans. and D. Sundaresan. 2002.References Acharya. 2001. Altman. 625–658. V. 28–44. Washington D. The Link between Default and Recovery rates: Implications for Credit Risk Models and Procyclicality. R. Anderson. R. Andrade. R. Design and Valuation of Debt Contracts. New York University. 33 . Gregor and Steven N. G. 53. Bakshi. Journal of Finance. 1998. Edward. Edward I. 1996. Vol. Andrea Resti and Andrea Sironi. 1443-1491. 2001–37. Edward I. V. Edward I. Federal Reserve Board of Governors. Analyzing and Explaining Default Recovery Rates. Pricing Credit Derivatives with Rating Transitions. Andrea Resti and Andrea Sironi. 2002. Altman. and S. Financial Analysts Journal. Financial Analysts Journal... Sep. Journal of Business.C. Working paper. Asquith. and R. Zhang. Gertner. Discriminant Analysis and the Prediction of Corporate Bankruptcy. Edward. Journal of Finance. New York University. Das. and F. Empirical Evidence and Implications. v58(3). Predicting Financial Distress of Companies: Revisiting the ZScore and Zeta Models. D.. Altman. 1968. P. Finance and Economics Discussion Series. 1994. Brooks Brady. and Vellore M. Working Paper. S.. Altman. Brooks Brady.

. E. 91–94. V. Vol. Bolton. 1998. 161198. 1994. and R. L. Optimal Debt Structure and the Number of Creditors. Darrell J. 35. November issue. 2000. 33. Carty. J. J. 257–287. Journal of Financial Engineering. S. Journal of Political Economy. Bank Loan Loss Given Default. Vol. pp. 3–56. L. Investor Valuation of the Abandonment Option. Defaulted Bank Loan Recoveries.. The Determinants of Credit Spread Changes. R. P. Moody’s Risk Management Services. Working Paper. 687-720. V. 1–25. E. Pierre.. 2001. Spencer Martin. Collin-Dufresne. B. A Discrete-Time Approach to Arbitrage-Free Pricing of Credit Derivatives. and D. and P. Swary. and Kenneth Singleton. 1996. Scharfstein. 2000a. Review of Financial Studies. 2000. 1999. New York University. Frye. Lieberman. Credit Ratings and Credit Spreads are Stochastic. Duffie. Sundaram. Debt Recoveries for Corporate Bankruptcies. and Walter N. 1996. 1999. R. Risk. Carty. and D. Tufano. Stern School of Business. Golstein.. 2000b. S. and K. French. Brady. Frye. Torous. Modeling Term Structures of Defaultable Bonds. Ofek. Moody’s Risk Management Services. Journal of Finance. v46(1). and D. Moody’s Risk Management Services. Franks. and J. Risk.. v5. Gates. Journal of Financial Economics. Common Risk Factors in the Returns on Stocks and Bonds. 1996. Robert S.. Journal of Financial Economics. Carty. L. P. Vol. Collateral Damage. 1993. Journal of Financial Economics.. Pricing Credit Sensitive Debt when Interest Rates. 12.. 349-370.Berger. and I. Das.. and D. V. April issue. 34 . 2177–2207.. Default Rate and Recovery Rate Correlation: Evidence and Implications for Modeling Corporate Bond Loss Rates and Loan Loss Rates. Bankrupt Bank Loan Recoveries. 104(1). Carty. L. A comparison of financial recontracting in distressed exchanges and Chapter 11 reorganizations. 2001. V. Moody’s Risk Management Services. Julian R. Lieberman. 42. Depressing Recoveries. Das. Management Science. 56(6). 46-62. Fama. 1996.

1995. Peura. D. 139. Troubled debt restructurings: An empirical investigation of private reorganization of firms in default. Number 3. and S.. 315-354. 35 . 1997. Journal of Financial Economics 27. Washington. Bargaining over Equity’s Share in the Bankruptcy Reorganization of Large. Litterman. Lando. Jarrow. Birkbeck College. The Dependence of Recovery Rates and Defaults. Review of Financial Studies. 1471–1493. Yen-Ting and William Perraudin. Edith. University of Pennsylvania Law Review. I. Pricing Options on Financial Securities Subject to Default Risk. Journal of Finance. 1990. Recovery Ratios and Survival Times for Corporate Bonds. Stuart.. International Monetary Fund. Lopucki. Kose John and Larry Lang. R. R. Turnbull. Forthcoming. Working Paper. Working Paper. Jokivuolle. Hotchkiss. Whitford. 57(5). 2000. Levent. A Model for Estimating Recovery Rates and Collateral Haircuts for Bank Loans. v10. Moody’s Risk Management Services.. 45(5). Default Parameter Estimation Using Market Prices. Izvorski. 1990. Vol. and S. 3-22. 53-86. Bank of Finland. Madan. Dilip and Haluk Unal. Pricing the Risk of Recovery in Default with APR Violation.Gilson. 125–152. Jarrow. 2002. 50. 1997. Journal of Banking and Finance. 1995. 1991. Journal of Finance. London Business School. Journal of Finance. E. A Markov Model for the Term Structure of Credit Spreads. Working Paper. v50. 2001. 52-64. Gupton. 2003. Stein. 2002. R. and S. LossCalc: Moody’s Model for Predicting Loss Given Default (LGD). Jarrow. Robert and Thomas Iben. Journal of Portfolio Management. Publicly Held Companies. Working Paper. Volume 17. D. 1990. Guha. G. L. Jeffrey K. Financial Analysts Journal. 481-523. and R. and W. 2003. Guntay. MacKie-Mason. Do Taxes Affect Corporate Financing Decisions?.. Turnbull. Post-bankruptcy performance and management turnover.. R. Hu.. Corporate bond valuation and the term structure of credit spreads. Recovery of Face Value at Default: Empirical Evidence and Implications for Credit Risk Pricing.C. 75–92.

1980. J.. Weiss. Review of Derivatives Research 2.. Bankruptcy auctions: costs. 419–444. Do Asset Fire Sales exist: An Empirical Investigation of Commercial Aircraft Sale Transactions. Unal. 2000. P. A Heteroscedasticity-Consistent Covariance Matrix and a Direct Test for Heteroscedasticity. 1998. Liquidation values and debt capacity: A market equilibrium approach. Pricing the Risks of Default. P. R. Tim and Sheridan Titman. and W. 1974. 55. 29. Journal of Finance. Strategic Debt Service. Journal of Finance. Biometrics. Merton. Stromberg. Vol. 161–92.Madan. 1992. 2002. The MIT Press. and H. 47. 939-978. Supplement. debt recovery and firm survival. Vol. White. 1998. Journal of Finance. Cambridge. J. 337-358. 817–838. 59–82. Mark E. Vol. Journal of Financial Economics 27. 49. 1984. The Journal of Finance. Scheffe. (1990) Bankruptcy Resolution: Direct Costs and Violation of Priority of Claims. 1953. Perraudin. 48. Todd C. P. L.. D. Zmijewski. 36 .. 53. 2001. Massachusetts. Opler. 1343-1366. 121–60. Methodological Issues Related to the Estimation of Financial Distress Prediction Models. Andrei and Robert Vishny. pp. 449-470.. Journal of Finance 52. 1015-1040. 645-646. Williams. Journal of Accounting Research. 2000. Pulvino. Sch¨ onbucher. 1997. H. Review of Derivatives Research 2. 1998.. Shleifer. Financial Distress and Corporate Performance. Wooldridge. Thorburn. Term-Structure Modeling of Defaultable Bonds. 56. M. 58. 1994. 40. A note on robust variance estimation for cluster-correlated data. Econometrica. Journal of Finance Vol. Robert C. ”On the Pricing of Corporate Debt: The Risk Structure of Interest Rates”. A Method for Judging All Contrasts in the Analysis of Variance. 531–566. Biometrika. Mella-Barral. Journal of Financial Economics. Conflicts of Interest and Market Liquidity in Bankruptcy Auctions: Theory and Tests. Karin S... Vol.. 87–104. 2641–2691. Halbert. Econometric Analysis of Cross Section and Panel Data.. L.

5σV )T √ .6) 37 .1) and (A. to translate the value and volatility of a firm’s equity into an implied probability of default.4) The KMV-Merton model uses the two nonlinear equations. E is the market value of firm’s equity. σV T 2 ln(V /F ) + (r − 0. (A.3) ∂E In the Black-Scholes-Merton model. d2 = d1 − σV T . The second relates the volatility of the firm to the volatility of its equity. and.1) where V is the value of firm’s assets. so it follows directly from Ito’s lemma that σE = (V /E ) ∂E σV .A Distance to Default in the KMV-Merton Model Symbolically. (A.4) for V and σV . the volatilities of the firm and its equity are related by σE = (V /E )N (d1 )σV .2) The KMV-Merton model makes use of two important equations.1) and (A. so that under the Merton model’s assumptions. expressing the value of a firm’s equity as a function of the value of the firm.5σV )T √ . The value of a firm’s equity. (A. We can then solve (A. E . N (·) is the cumulative standard normal distribution function. the distance to default and the expected default frequency (EDF) are computed as 2 ln(V /F ) + (r − 0. (A. The first is the BlackScholes-Merton equation (A.5σV )T √ EDF = N − σV T DD = (A. F is the face value of the firm’s debt (assumed to be zero-coupon) maturing at date T . r is the instantaneous risk-free rate continuously compounded. σV T (A. d1 and d2 are given by d1 = 2 √ ln(V /F ) + (r + 0.5) . is easy to observe in the marketplace by multiplying shares outstanding by the firm’s current stock price. The estimate of σE is obtained from either the stock returns data or the implied volatility of the options written on the stock. Under Merton’s assumptions the value of equity is a function of the value of the firm and time. Using these. ∂V (A.1). it can be shown that ∂V = N (d1 ). the Merton (1974) model stipulates that the equity value of a firm satisfies E = V N (d1 ) − e−rT F N (d2 ).4).

Profit Margin is the ratio of EBITDA to Sales. . SR. Ind. Med. Pecoup. Distress2 is a dummy variable that takes on a value 1 if distress1 is 1 and if the median sales growth of all the firms in the 3-digit SIC code of the defaulted firm is negative in any of the 2 years before the default date. Ind. Note that one firm could be classified in multiple collateral classes based on instruments that defaulted. All firm-specific variables are measured as of the last fiscal year before the default and data is obtained from COMPUSTAT. Pe is the price observed at emergence. Pd is the price observed at default. Note that one firm could be classified in multiple industries based on the division that defaulted. BDA. Pd is the price observed at default.Q is the median. Pe is the price observed at emergence. Pecoup is the price observed at emergence discounted by the coupon rate of the instrument in default for the period between default and emergence. Firm return is the stock return of the firm that has defaulted in the year before default. Table 2: Recovery Prices at Default and at Emergence based on Industry. Q ratio is the ratio of Market Value of the firm (estimated as Book Value of total assets . Ind. Debt concentration is the Herfindahl index measure by amount of the debt issues of the firm that are in default. GDP is the annual GDP growth rate in percentage. Pd is the price observed at default. Tangibility is the ratio of Property Plant and Equipment to Total Assets. There was only one default in 1981. 11 of the 15 pairwise means test for difference is significant at 5% level or lower using a Scheffe’s test. All Industry variables are measured in the year of default.Table 1: Time-series Pattern of Recovery Prices at Default (Pd) and at Emergence (Pehyld. Pehyld is the price observed at emergence discounted by the high yield index for the period between default and emergence. BDR is the weighted average default rate on bonds in percentage in the high yield bond market. Note that one firm could be classified in multiple seniorities based on instruments that defaulted. Pecoup is the price observed at emergence discounted by the coupon rate of the instrument in default for the period between default and emergence. Distress1 is a dummy variable that takes a value 1 if the median stock return of all the firms in the 3-digit SIC code of the defaulted firm is less than −30% and 0 otherwise. Table 3b: Recovery Prices at Default and Emergence based on Collateral. of all the firms in the 3 digit SIC code of the defaulted firm. No. Pehyld is the price observed at emergence discounted by the high yield index for the period between default and emergence. Pe is the price observed at emergence. of all the firms in the 3 digit SIC code of the defaulted firm. ∗∗ means significantly different from other group means at 5% level using a Scheffe’s test. of the ratio of Market value of the firm (estimated as Book Value of total assets − book value of equity + market value of equity) to the book value of the firm (estimated as book value of total assets). leverage is the median Long term debt to total assets. of issues is the total number of debt issues of the firm that is currently under default. Pehyld is the price observed at emergence discounted by the high yield index for the period between default and emergence. Liq2 is the median Interest coverage ratio (EBITDA/Interest). Pe). Table 3a: Recovery Prices at Default and at Emergence based on Seniority. Liq1 is the median Quick ratio (ratio of (Current Assets−Inventories) to Current Liabilities). Log (Assets) is the natural logarithm of the total assets. Med. All recovery prices are measured in cents per dollar. Note that one firm could have defaulted in multiple years. BDA is the total amount of high yield bonds defaulted amount for a particular year (measured at mid-year in billions of dollars) and represents the potential supply of defaulted securities. ∗∗ means different from other group means is significant at 5% level using a Scheffe’s test. Note that the number of data observations are different for each variable due to data availability. Pd is the price observed at default. Industry and Macro Characteristics. BDR are the macro variables used by Altman et. Pecoup is the price observed at emergence discounted by the coupon rate of the instrument in default for the period between default and emergence. Table 4: Summary Statistics of Firm.book value of equity + market value of equity) to the book value of the firm (estimated as book value of total assets). This table documents the time series pattern of recovery prices in terms of cents per dollar. GDP.al (2002). All recovery prices are measured in cents per dollar. Weights are based on the face value of all high yield bonds outstanding each year. Pecoup is the price observed at emergence discounted by the coupon rate of the instrument in default for the period between default and emergence. Pe is the price observed at emergence. Pehyld is the price observed at emergence discounted by the high yield index for the period between default and emergence. SR is the annual return on the SandP 500 stock index. Debt Ratio is the ratio of Long-Term Debt to Total Assets. Herfindahl Index is the industry concentration measure based on sales.Ind. Firm volatility is the standard deviation of daily stock returns of the firm in the year before default. This table documents the patterns of recovery prices in terms of cents per dollar across various industries.

Firm return is the stock return of the firm that has defaulted in the year before default. Senior Secured. Time in default is the time in years between the emergence and default dates. . Table 6a: OLS Estimates of Regression of Recovery Prices at Default on Contract and Firm Characteristics. Senior Unsecured. and 10% respectively. ∗ ∗ ∗. Utility is a dummy variable if the firm belongs to the utility industry. Debt concentration is the Herfindahl index measure by amount of the debt issues of the firm that are under default. Senior Subordinated. All regressions have industry dummies (the coefficients are not reported except for utility dummy). 5%. Subordinated are dummy variables that take on a value of 0 or 1 depending on the seniority of the instrument. ∗ represent significance levels at 1%. All regressions have industry dummies (the coefficients are not reported except for utility dummy). No. Firm return is the stock return of the firm that has defaulted in the year before default.Table 5: OLS Estimates of Regression of Recovery Prices at Default and at Emergence on Contract Characteristics. Utility is a dummy variable if the firm belongs to the utility industry. Debt Ratio is the ratio of Long-Term Debt to Total Assets. Pehyld is the price observed at emergence measured in cents per dollar for each debt instrument and discounted by the high yield index for the period between default and emergence.book value of equity + market value of equity) to the book value of the firm (estimated as book value of total assets). Log (Issue size) is the natural logarithm of issue size (in millions of dollars). Debt concentration is the Herfindahl index measure by amount of the debt issues of the firm that are under default. Pehyld is the price observed at emergence discounted by the high yield index for the period between default and emergence. Log (Issue size) is the natural logarithm of issue size (in millions of dollars). Subordinated are dummy variables that take on a value of 0 or 1 depending on the seniority of the instrument. Table 6b: OLS Estimates of Regression of Recovery Prices at Emergence on Contract and Firm Characteristics. Coupon is the coupon rate of the instrument. Q ratio is the ratio of Market value of the firm (estimated as Book Value of total assets . Cluster (based on each firm’s debt instruments as a single cluster) and heteroscedasticity corrected standard errors are reported in parentheses. All regressions have industry dummies (the coefficients are not reported except for utility dummy). Private Debt is the ratio of dollar value of bank debt for the defaulting firm to the dollar value of non-bank debt of the defaulting firm. Coupon is the coupon rate of the instrument. ∗ represent significance levels at 1%.book value of equity + market value of equity) to the book value of the firm (estimated as book value of total assets). Collateralized Debt is the ratio of dollar value of debt backed by collateral for the defaulting firm to the dollar value of all debt of the defaulting firm. No. Profit Margin is the ratio of EBITDA to Sales. Pd is the price observed at default. ∗∗. Cluster (based on each firm’s debt instruments as a single cluster) and heteroscedasticity corrected standard errors are reported in parentheses. ∗ represent significance levels at 1%. Bank Loans. 5%. Debt Ratio is the ratio of Long-Term Debt to Total Assets. Tangibility is the ratio of Property Plant and Equipment to Total Assets. Senior Secured. Maturity O/s is the remaining time to maturity of the instrument that has defaulted. Q ratio is the ratio of Market value of the firm (estimated as Book Value of total assets . Senior Unsecured. Coupon is the coupon rate of the instrument. ∗∗. Tangibility is the ratio of Property Plant and Equipment to Total Assets. ∗ ∗ ∗. of issues is the total number of debt issues of the firm that is currently under default. Both recoveries are measured in cents per dollar for each debt instrument. and 10% respectively. and 10% respectively. Cluster (based on each firm’s debt instruments as a single cluster) and heteroscedasticity corrected standard errors are reported in parentheses. Firm volatility is the standard deviation of daily stock returns of the firm in the year before default. Log (Assets) is the natural logarithm of the total assets. Bank Loans. Senior Unsecured. Senior Subordinated. All firm-specific variables are measured as of the last fiscal year before the default and data is obtained from COMPUSTAT. Current Assets and Unsecured are dummy variables that take the value of 0 or 1 depending on the collateral for the instrument. Log (Issue size) is the natural logarithm of issue size (in millions of dollars). 5%. Log (Assets) is the natural logarithm of the total assets. ∗∗. Subordinated are dummy variables that take on a value of 0 or 1 depending on the seniority of the instrument. Current Assets and unsecured are dummy variables that take the value of 0 or 1 depending on the collateral for the instrument. Senior Subordinated. All firm specific variables are measured as of the last fiscal year before the default and data is obtained from COMPUSTAT. of issues is the total number of debt issues of the firm that is currently under default. Firm volatility is the standard deviation of daily stock returns of the firm in the year before default. Time in default is the time in years between the emergence and default dates.Senior Secured. Utility is a dummy variable if the firm belongs to the utility industry. Pd is the price observed at default measured in cents per dollar for each debt instrument. Profit Margin is the ratio of EBITDA to Sales. ∗ ∗ ∗.

and 10% respectively. All firm-specific variables are measured as of the last fiscal year before the default and data is obtained from COMPUSTAT. Specificity of assets is defined as the ratio of machinery and equipment as percentage of total assets and follows Stromberg(2000) for all firms in Compustat over the sample period and using the S and P industry classification. Ind. Herfindahl Index is the industry concentration measure based on sales. Log(Issue size). Senior Subordinated. Tangibility is the ratio of Property Plant and Equipment to Total Assets.Q is the median. of all the firms in the 3 digit SIC code of the defaulted firm. ∗ ∗ ∗. Log(Issue size) is the natural logarithm of issue size (in millions of dollars). Coupon is the coupon rate of the instrument. Ind. Cluster (based on each firm’s debt instruments as a single cluster) and heteroscedasticity corrected standard errors are reported in parentheses. All regressions have industry dummies except specifications eight and nine (the coefficients are not reported except for utility dummy). Senior Unsecured.Ind. Distress1. Med. Senior Secured. Liq1 is the median Quick ratio (ratio of (Current Assets−Inventories) to Current Liabilities). Med. of all the firms in the 3 digit SIC code of the defaulted firm. Log(Issue size) is the natural logarithm of issue size (in millions of dollars). the description of the industry. Time in default is the time in years between the emergence and default dates. Ind. Herfindahl Index is the industry concentration measure based on sales. Senior Subordinated. ∗∗. Senior Unsecured. of the ratio of Market value of the firm (estimated as Book Value of total assets − book value of equity + market value of equity) to the book value of the firm (estimated as book value of total assets). ∗ represent significance levels at 1%. Liq2 is the median Interest coverage ratio (EBITDA/Interest). Senior Secured. Ind. Med. of all the firms in the 3 digit SIC code of the defaulted firm. ∗ ∗ ∗. Utility is a dummy variable if the firm belongs to the utility industry. Leverage is the median Long term debt to total assets. Distress2 is a dummy variable that takes on a value 1 if distress1 is 1 and if the median sales growth of all the firms in the 3-digit SIC code of the defaulted firm is negative in any of the 2 years before the default date. Liq1 is the median Quick ratio (ratio of (Current Assets−Inventories) to Current Liabilities). Bank Loans. Industry Distress. of all the firms in the 3 digit SIC code of the defaulted firm. All Industry variables are measured in the year of default.Ind.Ind.Ind. Med.Q is the median. Table 7b: OLS Estimates of Regression of Recovery Prices at emergence on Contract. Table 8a: Industries in Distress. All Industry variables are measured in the year of default. ∗∗. Pd is the price observed at default measured in cents per dollar for each debt instrument. ∗ represent significance levels at 1%. Firm and Industry Characteristics. Subordinated are dummy variables that take on a value of 0 or 1 depending on the seniority of the instrument. Coupon is the coupon rate of the instrument. Subordinated are dummy variables that take on a value of 0 or 1 depending on the seniority of the instrument. Med. Pehyld is the price observed at emergence measured in cents per dollar for each debt instrument and discounted by the high yield index for the period between default and emergence. Med.Table 7a: OLS Estimates of Regression of Recovery Prices at Default on Contract. Utility is a dummy variable if the firm belongs to the utility industry. Cluster (based on each firm’s debt instruments as a single cluster) and heteroscedasticity corrected standard errors are reported in parentheses. and 10% respectively. Tangibility is the ratio of Property Plant and Equipment to Total Assets. Ind. Distress1 is a dummy variable that takes a value 1 if the median stock return of all the firms in the 3-digit SIC code of the defaulted firm is less than −30% and 0 otherwise. Ind. of the ratio of Market value of the firm (estimated as Book Value of total assets − book value of equity + market value of equity) to the book value of the firm (estimated as book value of total assets). Current Assets and unsecured are dummy variables that take the value of 0 or 1 depending on the collateral for the instrument. and the year in which it was classified as distressed using the above criterion. 5%. All firm specific variables are measured as of the last fiscal year before the default and data is obtained from COMPUSTAT.Return is the median stock return of all the firms in the 3-digit SIC code of the defaulted firm. 5%. All regressions have industry dummies (the coefficients are not reported except for utility dummy). The following table lists the S and P Industry Code. Tangibility are not reported to conserve space. Profit Margin. Profit Margin is the ratio of EBITDA to Sales. Distress2 is a dummy variable that takes on a value 1 if distress1 is 1 and if the median sales growth of all the firms in the 3-digit SIC code of the defaulted firm is negative in any of the 2 years before the default date. . Profit Margin is the ratio of EBITDA to Sales. is a dummy variable that takes a value 1 if the median stock return of all the firms in the 3 digit SIC code of the defaulted firm in the default year is less than −30% and 0 otherwise.Return is the median stock return of all the firms in the 3-digit SIC code of the defaulted firm. Coefficients on Coupon. Firm and Industry Characteristics. leverage is the median Long term debt to total assets. Liq2 is the median Interest coverage ratio (EBITDA/Interest). Distress1 is a dummy variable that takes a value 1 if the median stock return of all the firms in the 3-digit SIC code of the defaulted firm is less than −30% and 0 otherwise.

5%. Pecoup. BDR are the macro variables used by Altman et. The table lists the recoveries as average and median values over the entire sample when there is no industry distress. Cluster (based on each firm’s debt instruments as a single cluster) and heteroscedasticity corrected standard errors are reported in parentheses. The z–statistic tests for differences in medians (B)-(C) using the Wilcoxon rank sum test.Table 8b: Pattern of Recovery Prices at Default (Pd) and at Emergence (Pehyld. BDA is the total amount of high yield bonds defaulted amount for a particular year (measured at mid-year in trillions of and represents the potential supply of defaulted securities. ∗ ∗ ∗. ∗ ∗ ∗. average over the sample whose industry is in distress in a given year. GDP. Senior Subordinated. Correlations are calculated separately for years when the industry in in distress and when it is not. Tangibility is the ratio of Property Plant and Equipment to Total Assets. Industry and Macro Characteristics. Profit Margin is the ratio of EBITDA to Sales. ∗ ∗ ∗. ∗∗. ∗ represent significance levels at 1%. Utility is a dummy variable if the firm belongs to the utility industry. Coupon is the coupon rate of the instrument. Debt concentration is the Herfindahl index measure by amount of the debt issues of the firm that are under default. Med. The medians are shown within brackets. Pe is the price observed at emergence. of all the firms in the 3 digit SIC code of the defaulted firm. ∗ represent significance levels at 1%. Time in default is the time in years between the emergence and default dates. 9b: OLS estimates of regression of Recovery Prices at default and emergence on Contract. Pehyld is the price observed at emergence discounted by the high yield index for the period between default and emergence. and 10% respectively. SR. The t–statistic tests for difference of means (B)-(C). Current Assets and unsecured are dummy variables that take the value of 0 or 1 depending on the collateral for the instrument. Pe) and Asset Specificity within each Industry. HML are the Fama-French factors in the 3 factor model. Pehyld is the price observed at emergence discounted by the high yield index for the period between default and emergence. Industry Distress is a dummy variable that takes a value 1 if the median stock return of all the firms in the 3-digit SIC code of the defaulted firm in the default year is less than −30% and 0 otherwise. and average over the remaining sample. ∗∗.5%. Market.Ind. Panel C displays the correlation between specific assets and one of the recovery measures (mean Pe.Q is the median. Pd is the price at default. ∗∗. Panel A is for Pe and Panel B is for Pehyld. median Pe. of the ratio of Market value of the firm (estimated as Book Value of total assets − book value of equity + market value of equity) to the book value of the firm (estimated as book value of total assets). 5% . Pe) for Distressed and Non-distressed industries.al (2002). All Industry variables are measured in the year of default. and 10% respectively for the test that correlation is not equal to zero. To conserve space these variables are not reported in the table. Panel A is for the entire sample while Panel B excludes 1990 defaults. . All recoveries are measured in cents per dollar for each debt instrument. Table 8c: Pattern of Recovery Prices at Emergence (Pehyld. average over the sample whose industry is in distress in a given year. All firm specific variables are measured as of the last fiscal year before the default and data is obtained from COMPUSTAT. BDA. Weights are based on the face value of all high yield bonds outstanding each year. The table lists the recoveries as average over the entire sample. Subordinated are dummy variables that take on a value of 0 or 1 depending on the seniority of the instrument. Log(Issue size) is the natural logarithm of issue size (in millions of dollars). Both recoveries are measured in cents per dollar on each debt instrument. Pe is the price observed at emergence. Distress1 is a dummy variable that takes a value 1 if the median stock return of all the firms in the 3 digit SIC code of the defaulted firm is less than −30% and 0 otherwise. SMB. Ind. mean Pehyld or median Pehyld). Bank Loans. Industry Distress is a dummy variable that takes a value 1 if the median stock return of all the firms in the 3-digit SIC code of the defaulted firm in the default year is less than −30% and 0 otherwise. BDR is the weighted average default rate on bonds in the high yield bond market. All regressions have industry dummies (the coefficients are not reported except for utility dummy). Senior Unsecured. Table 9a. Pd is the price observed at default. SR is the annual return on the S and P 500 stock index. Liq1 is the median Quick ratio (ratio of (Current Assets−Inventories) to Current Liabilities) of all the firms in the 3 digit SIC code of the defaulted firm. Firm. ∗ represent significance levels at 1% . All recoveries are measured in cents per dollar for each debt instrument. Pecoup is the price observed at emergence discounted by the coupon rate of the instrument in default for the period between default and emergence. Specific assets is defined as the ratio of machinery and equipment as percentage of total assets and follows Stromberg(2000) for all firms in Compustat over the sample period and using the S and P industry classification. GDP is the annual GDP growth rate.and 10% respectively. Pehyld is the price observed at emergence discounted by the high yield index for the period between default and emergence. Senior Secured.

Pd is the price at default. Med. ∗ represent significance levels at 1%. All firm specific variables are measured as of the last fiscal year before the default and data is obtained from COMPUSTAT.Table 10: OLS estimates of regression of Recovery Prices at default and emergence on risk factors that explain default. Senior Subordinated. Bank Loans. Cluster (based on each firm’s debt instruments as a single cluster) and heteroscedasticity corrected standard errors are reported in parentheses. Subordinated are dummy variables that take on a value of 0 or 1 depending on the seniority of the instrument. of the ratio of Market value of the firm (estimated as Book Value of total assets − book value of equity + market value of equity) to the book value of the firm (estimated as book value of total assets). Ind. Senior Unsecured. Log(Issue size) is the natural logarithm of issue size (in millions of dollars). of all the firms in the 3 digit SIC code of the defaulted firm.and 10% respectively. All regressions have industry dummies (the coefficients are not reported except for utility dummy). All Industry variables are measured in the year of default. Both recoveries are measured in cents per dollar on each debt instrument. Time in default is the time in years between the emergence and default dates.5%. Distance to default is the measure obtained by solving the Merton(1974) model for each firm.Score is the Zmijeswki (1984) Score. Liq1 is the median Quick ratio (ratio of (Current Assets−Inventories) to Current Liabilities) of all the firms in the 3 digit SIC code of the defaulted firm.Ind.Q is the median. Senior Secured. Z–Score is the Altman Z–score as modified by Mackie-Mason(1990). Debt concentration is the Herfindahl index measure by amount of the debt issues of the firm that are under default. ∗ ∗ ∗. Coupon is the coupon rate of the instrument. . Distress1 is a dummy variable that takes a value 1 if the median stock return of all the firms in the 3 digit SIC code of the defaulted firm is less than −30% and 0 otherwise. Utility is a dummy variable if the firm belongs to the utility industry. Zmij. Current Assets and unsecured are dummy variables that take the value of 0 or 1 depending on the collateral for the instrument. Pehyld is the price observed at emergence discounted by the high yield index for the period between default and emergence. ∗∗.

31 25.30 25.02 34.00 35.00 54.40 73.40 41.96 35. Firm defaults 465 5 4 3 8 16 11 24 29 69 81 53 36 25 35 27 11 16 12 Pehyld Average Overall 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 645 16 6 11 19 43 23 55 52 77 111 42 11 10 23 17 14 21 94 Firm defaults 379 9 5 6 12 20 14 25 28 41 53 25 9 9 18 13 9 19 64 .00 43.71 38.50 30.73 37.59 43.47 37.62 62.80 64.64 61.91 31.88 45.50 35.63 1511 12 5 6 12 37 56 101 110 245 326 137 103 60 97 75 38 49 42 51.11 33.07 38.89 19.84 66.94 48.50 38.27 29.57 10.50 45.50 32.99 44.75 50.50 15.55 40.58 16.48 58.09 51.18 36.06 33.18 49.53 55.11 44.25 53.57 34.43 16.17 50. Year 41.22 60.31 44.00 34.91 30.88 50.24 48.31 20.64 36.71 21.69 67.58 49.Table 1 : Time-series Pattern of Recovery Prices at Default (Pd) and at Emergence (Pehyld.00 40.54 68.78 35.02 53.34 56.00 36.75 38.34 15.82 47.80 55.76 80.86 46.50 52.23 36.54 19.82 15.67 26.01 20.00 25.49 35.20 56.30 62. Pe).82 23.66 35.70 19.00 55.96 27.Dev.18 38.Dev.09 49.70 36.89 38. Defaults Median St.00 52.13 23.32 27.19 Defaults Pd Average Median St.77 19.67 20.25 41.59 56. Pecoup.02 63.96 23.65 53.55 43.00 31.97 58.96 36.09 82.00 37.14 47.80 49.76 41.70 21.95 26.49 36.90 29.27 28.

99 68.78 67.73 44.20 Defaults Pecoup Average Median St.26 34.43 14.00 65.16 65.06 24.77 36.57 32.04 42.98 33.25 36.04 89.30 42.61 71.15 45.35 38.57 37.54 37.83 39.46 15.88 68.03 72.51 37.12 81.71 36.56 72.39 50.73 54.76 80.05 53.61 49.50 33.35 93.46 58.35 27.63 38.28 29.15 73.85 35.58 8.66 75.88 36.31 52.05 42.27 61.93 60.90 23.44 59.31 44.63 53.09 52.23 36.69 72.93 30.21 60.00 42. Defaults Median St.50 38. Year 52.30 35.48 36.27 31.Dev.50 80.Dev.81 41.44 38.32 34.96 43. Firm defaults 475 5 4 3 8 16 11 24 29 70 85 54 38 26 36 27 11 16 12 Pe Average Overall 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 1510 12 5 6 12 37 56 101 110 245 326 137 103 60 97 74 38 49 42 Firm defaults 464 5 4 3 8 16 11 24 29 69 81 53 36 25 35 26 11 16 12 .46 37.02 40.96 70.28 40.03 58.17 65.83 44.92 66.31 29.71 51.67 54.57 42.01 42.83 30.48 41.18 38.70 36.05 45.28 63.90 58.02 78.99 68.07 51.03 63.44 28.27 66.25 27.57 80.81 78.81 42.21 59.06 57.15 1541 12 5 6 12 37 56 101 110 244 344 138 111 61 100 75 38 49 42 62.70 59.45 68.Table 1 (continued) : Time-series Pattern of Recovery Prices at Default (Pd) and at Emergence (Pehyld. Pecoup.32 67.12 44.75 54.53 32.13 20.75 64. Pe).43 62.13 19.13 77.45 65.07 88.

09 48.79 27.50 22.00 29.29 40. 36.70 9.99 26.58 18.41 40.Table 2 : Recovery Prices at Default and at Emergence by Industry.09 76.25 26.98 29.05 35. S&P Code Overall 1 2 3 4 5 6 7 8 9 10 11 12 646 55 33 16 43 24 22 16 82 43 165 87 60 41.94 49.11 74.68 1511 82 77 26 99 76 111 63 138 114 472 167 86 Industry Description Overall Utility Insurance and Real Estate Telecommunications Transportation Financial Institutions Healthcare / Chemicals High Technology/ Office Equipment Aerospace / Auto / Capital Goods Forest.88 45.18 32.67 47.69 51.38 40.79 55.13 53.99 39.82 60.00 28.25 38.70 36.Dev.35 35.22 51.43 53.00 33.33 36. Def Firm defaults 424 9 23 6 20 24 35 22 46 30 126 54 29 Pehyld Avg 51.49 18.59 29.11 48.10 24.94 27.38 25.90 25.08 53.96 68.13 38.82 26.01 38.69 37.33 42.79 30.92 49.34 20.96 44.50 47.13 38.33 22.63 26.41 Mdn 49.00 77.57 36.80 St.25 49.76 42.41 .46 21.75 36.00 37.50 58.07 37.37** 39.33 41.92 58.51 46.05 52.31 36.97 21.00 35.07 38.Dev.78 41.49** 37.Building Prod / Homebuilders Consumer / Service Sector Leisure Time / Media Energy and Natural Resources Def Firm defaults 365 8 15 5 24 15 18 11 50 24 97 62 36 Pd Avg Mdn St.

07 18.81 77.Dev.07 37.58 30.70 48.74 77.79 67.36 71.05 60.00 42.26 30.Dev.94 63.48 .07 43.58 26.19** 71.63 26.00 32.00 42.25 Seniority Code Overall 1 2 3 4 5 6 Seniority Description Overall Bank Loans Senior Secured Senior Unsecured Senior Subordinated Subordinated Junior Subordinated Firm defaults 418 Pd Avg Mdn St.78 16.06 34.11 81.00 28.11 94.83 80.11 33. Collateral Code Overall 1 2 3 4 5 6 7 8 No data Collateral Description Overall Current Assets PP and E Real Estate All or Most assets Other Unsecured Secured Information Not available 1511 52 83 38 228 33 32 40 1005 Firm defaults 644 46 44 23 126 20 25 17 343 51.43 33.51 31.71 63.42 30.92 34.75 25.13 24.Dev.07 26.90 23. 36.16 53.Dev.77 89.35 31.96 27.67 63.59 38.01 36.21 33.50 29.37 27.64** 49.58 15.28 49.33 51.17 1511 358 267 236 266 346 38 51. Def Firm defaults 829 219 119 98 172 186 35 Mdn St.18 34.12 59. Def Pehyld Avg Mdn St.39 30.91 36.11 Pehyld Avg Table 3b : Recovery Prices at Default and at Emergence by Collateral. Def 598 No Data 183 107 153 148 7 110 57 133 113 5 48.14 55.55 61.37 27.66 6.09 91.89 23.10 26. Def Firm defaults Pd Avg Mdn St.09 98.Table 3a : Recovery Prices at Default and at Emergence by Seniority.99 54.48 36.

02 0.17 0.08 1.50 0.93 10.97 0.33 0.84 0.03 -2.29 4.18 0.63 2.D.07 1.49 0.43 0.32 0.40 0.05 10.13 0.91 0.00 1.25 -0.00 3.08 0.00 0.23 0.06 25th Percentile 0.30 0.13 0.17 0.03 0.86 0.14 0.29 1.48 6.35 0.93 1.46 2.17 0.35 1.99 3.00 0.01 3.20 0.03 0.00 3.49 3.69 2.00 0.07 0.56 0.57 0.61 0.54 4.10 0.23 4. 0.24 0.65 0.34 2.84 2.03 1.20 0.43 0.18 1.05 0.17 3.25 0.04 0.00 0.98 5.00 0.70 0. S.07 0. Industry and Macro Characteristics.43 3.47 3.08 0.43 0.09 0.27 23.23 5.38 7.01 0.33 0.00 0.17 7.77 0.08 0.09 0.Table 4 :Summary Statistics of Firm.50 0.25 75th Percentile 0.16 0.20 1.18 0.12 1.22 6.52 0.45 33.86 4.30 0.00 0.09 0.05 0.94 0.07 0.81 0.79 0.82 -0.02 0.00 0.18 0.76 3.00 1.34 0.27 Min Median Max Variable n Mean Profit Margin Tangibility Debt Ratio Log(assets) Q ratio No of issues Debt Concentration Firm Return Firm Volatility Distress1 Distress2 Med Ind Q Herfindahl Index Ind Liq1 Ind Liq2 Med Ind Leverage SR GDP BDA BDR 163 163 165 168 113 196 196 86 149 94 94 94 91 86 87 90 18 18 18 18 0.99 6.00 0.00 0.00 0.49 6.28 0.00 1.27 1.50 .36 0.53 10.01 0.43 0.69 7.10 0.01 0.27 0.38 0.00 1.18 3.83 0.67 2.27 0.81 1.08 2.01 -0.

11 (0.19∗∗∗ (6.43 370 0.21) 34.09) ∗∗∗ 12.22∗∗∗ (9.Table 5 .22) Pehyld (6) 25.19) 12.61) 34.38 (8.22) (7.91) -0.96 ∗ -0.31 1375 0.03 (1.15 ∗∗∗ ∗∗∗ 47.14 (0.79) ∗∗∗ (7.39) -1.27 (0.76) ∗∗∗ 34.01) -0.82 14.63) (6.91) 0.27∗∗∗ (3.54∗∗∗ (5.22 (0.98) 7.17 (1.65) ∗∗∗ (7.82 (8.90) ∗∗∗ (5.17 (0.08) 4.66) 59.39 23.34) 26.94) 12.06) (7.84) 2.37 34.48 1375 0.Pd and Pehyld Pd (1) 24.10∗∗∗ (8.10∗∗ (1.15∗∗∗ (5.03 ∗∗∗ ∗∗∗ 62.98) -5.28) -0.44∗∗∗ (1.30 370 0.46) -0.99) Pehyld (5) 37.19 ∗∗∗ 58.86 14.01 (8.56∗∗∗ (6.36∗∗ (4.46) 7.62 38.93∗∗∗ (6.28 34.63∗∗∗ (7.88 40.41∗∗∗ (0.25∗∗∗ (10.21) 19.55∗∗∗ (5.05) ∗∗∗ 26.54 (8.20∗∗∗ (0.27 (0.73 (7.09) (5.10) Const.43 .06) Pehyld (4) 9.48) (6.36 34.26) -11.38) -0.39) -0.95 14.16 (5.87) (7.55) (0.36) (6.43 1375 0.30∗∗ (11.80 (8.11 (0.99) -3.07 (5.03) -2.95) 33.07) (8.22) -5.27∗∗∗ (9.03 (7.07) (5.09 ∗∗ 14.93) ∗∗∗ (5.38) Pehyld (7) 27.66) ∗∗∗ 2.22) ∗∗∗ (6.45 (12.79 ∗∗ 7.11 ∗∗∗ 37.74) (5.92) (1.44 1375 0.27) 13.73 23.23) ∗∗∗ (8.12 (5.10) 34.52 27.84) -0.83) -3.28 ∗∗∗ 31.43 (7.28 39.23∗∗∗ (6.93) -0.10) 373 0.20 4.54) (6.27) 7.52) -2.82) (5. R2 Pd (2) 25.92 ∗ -0.91) ∗∗∗ (5.55) (0.23) 37.48∗∗∗ (6.39) (0.02) -5.90 ∗∗∗ 37.49∗∗∗ (9.70) -1.21 (0.14 (0.43) Pehyld (3) 27.16) (8.46∗∗ (11.56) 26.33∗∗∗ (1.23) 14.27 (5. Coupon Log(Issue Size) Bank Loans Senior Secured Senior Unsecured Senior Subordinated Subordinated Time in Default Maturity O/s Current Assets Unsecured Collateralized Debt Collateralized Debt * Unsecured Collateralized Debt * (1-Unsecured) Private Debt Private Debt * Bank Loans Private Debt * (1-Bank Loans) Utility Obs.98 (6.46) -5.22) 7.87 ∗ -0.07) 11.09 (0.79∗∗∗ (6.77 (8.Contract Characteristics .

55 ∗∗ 22.26) -3.81∗∗∗ (9.45∗∗ (18.94 (9.Pd Pd (1) 13.36 ∗∗∗ 35.38) 28.22) (2.37) 30.74∗∗∗ (8.21 ∗∗ -0.11 ∗∗∗ 25.95 (19.94) 4.42∗ (14.14) (12.20 (7.54 (0.05) 27.64 (0.80 (12.52) 27.09) 1.03 4.39) 3.41) (8.81 ∗∗∗ 30.51 241 0.52∗∗ (10.00 (12.57) -1.17) (2.22) 4.69∗∗ (8.of Issues Debt Concentration Firm return Firm Volatility Utility Obs.81∗∗ (7.21) -0.51 166 0.19) Const.06) (2.93) 23.48) Pd (4) 18.14 (2.91∗∗ (0.49 (18.Table 6a .27∗∗∗ (69.65) (9.84 (9.29 ∗∗∗ 17.14) 24.58 .40 (7.32) -0.33) 20.58 ∗∗∗ 38.66) (17.06∗∗∗ (9.47 241 0.25 ∗ 4.03 ∗ 3.70) 1.32∗∗∗ (7.66 ∗ -0.22 19.27) (1.24) ∗∗ 11.84) 25.21∗∗ (14.43∗∗∗ (9.77) Pd (7) 43. R2 Pd (2) 7.72) (9.26) -5.33 (8.97) 7.52) 28.94 (0.32) 3.43) 3.91 (9.10∗∗∗ (8.68) -1.78 ∗∗ 42.07 (9.98) 241 0.Contract.53) -37.58) -4.89 (11.63) -0.58 (17.50 241 0.10) 8.38 (11.02 (8.47 190 0.44) 23.76 ∗ 3.05) Pd (6) 15.62) (6.55∗ (19.27 (2.38) 11.66) -13.15 ∗∗∗ 23.74 (8.86∗∗∗ (7.56) -0.38 (8.23 ∗∗ 27.60) -1.04) Pd (5) 8.27) 10.81 (16.86 ∗ 15.49) (2.28∗∗ (0.53 (0.40) 0.18) (8. Coupon Log(Issue Size) Senior Secured Senior Unsecured Senior Subordinated Subordinated Profit margin Tangibility Leverage Log(assets) Q Ratio No.82) 12.84) 26.27) (7.78) 27.60) (0.06) -13.5) (10.20∗∗∗ (9.33 (7.09 (8.32 (8.81 ∗ 12.25∗∗∗ (12.08) (11.38 ∗∗∗ 28.18) 26.15) (9.22) -235.32 ∗∗ -7.70∗∗∗ (6.03) -29.15) Pd (3) 32.97∗∗ (9.28∗ (17.62) (9.69 (6.32) (10.92) 32.60) ∗∗ 3.79 (2.52∗ (8.68) 17.60 180 0. and Firm Characteristics .83) -1.88∗∗∗ (8.39∗∗∗ (10.43) (10.81 ∗ 7.32 (6.11) (7.67) (7.28 (9.47) 36.44) 4.31 0.75 (12.35) (7.33 ∗ -29.51) 21.79 ∗∗∗ 27.24 (23.14 ∗∗∗ 18.38) 26.13) -11.57) (0.69) (16.

18∗ (1.34 (0.83 (4.01) (10.60 670 0.41 ∗∗∗ 16.74 (8.55) (13.34 (8.86 (0.74 ∗∗∗ 19.67 ∗∗∗ 25.84 ∗∗∗ -7.43 (17. R2 Pehyld (2) 30.58) 0.88 (12.15) 1.67 ∗∗∗ 18.60) 11.78 ∗∗∗ 48.43 ∗∗∗ -7.86∗∗ (9.92) 20.69 (1.80) -7.16) 20.08) 19.08) 2.24) -1.15 (0.32) Pehyld (5) 38.53 (4.of Issues Debt Concentration Firm return Firm Volatility Utility Obs.15) (1.38 (16.14) -6.09 (9.20 (8.71 ∗∗∗ -8.56 ∗∗∗ 18.31) -5.89) -4.58) -55.68) 24.83) -0.09) 25.31) -0.47 (5.00 ∗∗∗ 41.19∗∗∗ (14.04 (0.79 (12.14 (9.08) -4.86∗∗∗ (8.60) (9.83) -1.08) 0.25) (6.47) 0.53) 18.55 (16.18 ∗∗∗ 36.40) -20.92) (5.40 (16.98) 1.28) 25.37) -4.85∗∗∗ (18.32) -12.31∗∗ (12.84) -4.23) (5.45) -0.42) (8.59 (1.89 ∗∗∗ -6.47) 0.22) Pehyld (7) 51.19) (6.29 (4.22) Const.11∗∗ (20.97∗∗∗ (8.10) 19.09) (9.96∗∗ (21.47) 0.15 (16.94 (11.57 670 0.54) -20.32) -4.06) (7.79 ∗∗∗ 25.72 (16.31) 43.35) -22.75) 24.40) (1.51) -7.45∗∗ (9.Table 6b .53 ∗∗∗ 18.37) -3.30) -4.82 (5.00) 5.62∗∗∗ (1.22) 37.35) 2.52) Pehyld (3) 49.46) 7.55) 0.5 (11.05 (0.57 468 0.40) -6.98) (8.13 (12.01 (0.76) -2.32) Pehyld (6) 49.17 (1.25) (1.21 (0.19∗∗ (9.49 (8.86 ∗∗∗ 41.21) 7.58 (9.57 348 0.52) 10.69) -9.40 (11.64 (5.40) (10.78) -6.03 (0.81 (6.81) 24.37) 5.95) (8.57 670 0.02 ∗∗∗ 33.73 ∗∗∗ 37.16) (1.97) 0.86∗∗∗ (8.91) Pehyld (4) 41.85) -15.09 (16.17 ∗∗∗ 46.01 (5.37) -6.78 (5.66 (4.51 ∗∗∗ -7.76 (12.35 (8.49) -22.85 (16.43∗∗∗ (14.07) 1.32) 5.29) -3.55) -0.66 .40 ∗∗∗ -8.61 (8.30) 5.85) -1.49∗ (17.11∗∗ (9.78) -4.52) -14.82 (9.60 (8.Pehyld Pehyld (1) 39.06) 670 0.45 ∗∗∗ 45.45) 1.54 (1.81) 0.05 (64.54∗∗∗ (5.49) (1.22) (5.07 (5.43) 24.37) -2.13) 7.39) -0.64 (11.78 (9.64 (9.17 (8.29 (8.85 (1.49) (9.55 (1.67 396 0. and Firm Characteristics .90∗∗∗ (8.44) (5.25 ∗∗∗ 43.07 (9.32 (8.99) -1.33) -9.81) 14.70) 5.14 (7.17) (1.16) 4.53 ∗∗∗ 38.70∗∗∗ (14.64 (16.63) 20.95) -3.29 (17.50) (9.72 (5.31) 4.Contract.74 (8.40∗∗∗ (8.13∗ (8.21 (0.78 (5.14) -4.75 (9. Coupon Log(Issue Size) Bank Loans Senior Secured Senior Unsecured Senior Subordinated Subordinated Years in Default Current Assets Unsecured Profit margin Tangibility Leverage Log(Assets) Q ratio No.52) 1.

72) 24.61 243 0.16) 21.14 (8.57) Pd (8) 5.71 (15.27 ∗∗∗ 25.17) 20.59) -0.58) ∗∗ 5.83) 1.42) 26.86 (8.87 (9.73∗∗∗ (8.48 (20.76) 26.49 (13.72 (9.42) -19.25 ∗∗ 24.45∗∗∗ (7.11 (8.06 (5.39 ∗ -0.60 254 0.53 ∗ 18.36) (11.53) (8.79) 28.15 (0.46 (8.00) (2.35) Pd (6) 33.Pd Pd (1) 11.56) 3.63) (8.07 (7.64) -11.06 ∗∗ -10.66 (12.84) -0.16 (1.09 (5.33) (8.65∗∗∗ (6.94 (18.29) 6.24) 21.48) 0.18) 29.44 ∗∗ 2.58 ∗∗ -22.34) 8.Table 7a .66 (9.20 (14.79) 8.09) 22.15∗∗∗ (9.53∗∗ (1.12 (8.85) 2.52∗∗∗ (7.92 (0.47) -0. Coupon Log(Issue Size) Senior Secured Senior Unsecured Senior Subordinated Subordinated Profit margin Tangibility Debt Concentration Firm Volatility Distress1 Distress2 Med Ind Q Herfindahl Index Ind Liq1 Ind Liq2 Med Ind Leverage Med Ind Return Utility Obs.85) (2.70∗∗∗ (8.53 ∗ 18.44∗∗∗ (8.51 (11.86) 29.19 (5.25) 7.97) -0.76 (8.58) 19.39∗∗∗ (8.06 (13.80) (13.85 ∗∗∗ 24.87) (10.97) -1.95) 28.56) (7.74 4.14) 7.28 ∗∗∗ 24.67 (8.60) -13.59) ∗∗ (0.94) 25.66) -0.76 (7.29) 25.82 (8.62 254 0.78 (14.39) (1.22) 20.84 (16.47) (11.88 (8.18) 22.32) 2.13) (2.76) Pd (7) 11.14 (12.37∗∗ (9.13 (13.68) Pd (4) 10.30∗∗∗ (8.66) 1.57) -1.98∗∗ (7.71 (5.37) 2.30 ∗ 4.55) 2.19) Pd (5) 4.43) (10.97) 0.00 (7.90) -3.75 ∗ -1.04) 24.30 (14.45) ∗∗ (0.98) (2.41) (1.70) -5.09) 25.40) -5.17) 19.63 (8.25∗∗ (16.55∗∗∗ (9.34) (8.14 243 0.20 (7.23) (4.15) -0.21) 6.49) -1.84∗ (10.29∗∗∗ (7.57) -1.98∗∗∗ (7.80) (8.64 ∗∗ (4.72) -15.63) 30.26) -2.57 ∗ -0.47) (10.60∗∗∗ (7.71) 28.53 ∗ 18.27) -10.30 (14.70 (0.64 (7.65) -4.57 (0.75 (7.17 ∗∗∗ 25. R2 Pd (2) 6.70) -1.88 -12.03∗∗∗ (8.22) -0.00) -18.47) (0.31 (8.68 ∗∗∗ ∗∗∗ (87.45) -5.33∗∗∗ (8.89) 6.48∗∗∗ (7.80 (8.71 ∗∗∗ 18.79) -0.89) 23.93) 0.38∗∗∗ (8.88 (13.95 5.98) (2.14 (13.81) Pd (3) 13.49) -19.01 243 0.61 ∗∗ 18.33∗∗∗ (7.83) 29.74) -0.58 254 0.16 (14.11) 262 0.23 (15.87 (8.83 (0.17) -276.Contract.26) 27.10) 23.80) -4.87∗∗ (9.52 186 0.47 ∗∗ 4.16) (4.51∗∗∗ (9. Firm.19) -1.04∗∗∗ (6.50) -4.20) -9.52 .19∗∗∗ (7.83∗∗∗ (8.00 (14.45 (6.16) 24.00 5.28 (18.96) 2.17) 8.66∗∗ (8.21) 24.53 ∗ 24.11 ∗∗∗ 24.03 (7.47) 0.85 (8.75 (14.35) Const.33 (7.02∗∗∗ (9.53) (8.03 (11.04 (13.21) (1.59∗∗ (2.29) -20.25) -21. and Industry Characteristics .90) 25.53∗∗∗ (8.94) (10.71) 0.96 (12.31) 3.51 ∗ 17.71 ∗∗ 5.33 ∗∗ 4.14) -5.

03 (14.58 729 0.39) (9.82) -5.Contract.41) 8.41 (8.62∗∗∗ (4.06) 1.62 ∗∗ 32.37) Pehyld (6) 8.45 (16.22 (5.37 (7.60 (6.46) (8.19) -3.66) 12.24) 2.12 (13.09) 19.05∗∗∗ (1.65) -6.13) ∗∗∗ (8.39) (7.82∗∗∗ (1. R2 Pehyld (2) 12.07 (16.58 754 0.52) 4.29) 11.38) (8.49 ∗∗∗ 51.14∗ (5.87) -3.20 (14. Firm.82) -5.09) -1.54) (8.07 ∗∗∗ 42.99∗∗ (7.56 (14.67) (8.88 (8.54) -24.72) -9.99) -4.88) 2.05) (7.89) ∗∗∗ (9.12 (0.08 (6.78∗∗ (4.49 (7.35 (8.33 (4.24 (0.74) 44.74) Pehyld (5) 12.70 (5.36) -3.09) 1.00) -0.91∗∗ (7.85) -7.40∗∗∗ (1.27 ∗∗∗ 42.56) 0.32) 9.80 (6.50 729 0.32 (8.69∗∗∗ (0.83 (7.71) ∗∗∗ 35.99 ∗∗ 22.Ind Leverage Med.66 (14.61∗∗∗ (0.88) ∗∗∗ (8.03 (7.35 (4.83∗∗∗ (8.68 25.24 (8.86) -6.Ind Return Specificity (mean) Specificity (median) Distress1 * Specificity (mean) Distress1 * Specificity (median) Utility Obs.98) (7.39) 16.40) 16.32∗∗∗ (9.48∗∗ (7.09 ∗∗∗ 42.81∗ (6.65 (14.67∗∗∗ (8.45∗∗ (7.10∗∗∗ (1.83) 1.85) -6.28) Pehyld (8) 3.69) (8.70) 11.15) -6.60∗ (5.32) Pehyld (3) 20.26) -1.28) 2.38) -1.58 754 0.39) 18.28) 31.27) Pehyld (4) 9.74 ∗∗ 20.13∗∗∗ (7.73) (9.25 (8.48∗∗∗ (1.25∗∗∗ (0.83) 14.25 (7.8 ∗∗∗ 42.13) -6.18 (4.65∗ (5.36) 4.31) 11.18) -5.54 (8.67 ∗∗∗ 42.58 765 0.25) -10.66) -23.11) -9.92∗ (5.37) 41.78∗∗ (4.19) ∗∗∗ (8.03 ∗∗ 19.13 32.03∗∗∗ (1.19∗ (5.77 (8.58 754 0.46) 12.70∗∗∗ (10.12 (7.93∗∗∗ (4.80∗ (5.95) -3. Bank Loans Senior Secured Senior Unsecured Senior Subordinated Subordinated Years in Default Current Assets Unsecured Distress1 Distress2 Med Ind Q Herfindahl Index Ind Liq1 Ind Liq2 Med.16 ∗∗∗ 42.38) 729 0.25) -1.86) 7.99) (9.53 (4.94 (3.35) (8.23∗∗∗ (11.90) -5.34) 2.65) 12.50 .89∗∗ (4.88 ∗∗∗ 51.01 (8.16) -6.14) -5.93) -2.26 ∗∗∗ 44.31) 8.16) 1.60 (15.33) 10.90 (4.84) ∗∗∗ (9.01∗∗ (6.95) ∗∗∗ (9.87 (8.42 (13.86) 2.40∗∗∗ (1.38) -10.41 ∗∗∗ 40.74) (8.64 (7.65∗∗∗ (1.86 ∗∗ 21.02 ∗∗∗ (6.89) -4.58 729 0.15) 3.91) Pehyld (9) -0.30 26.00 25.86 (14.85 (4.Table 7b .39) 16.90 (14.16) -6.16 (15.34 (15.87 27.17) ∗∗∗ (8.49) 11.70) -0.63) (9.94∗∗ (5.69∗∗ (7.18 ∗∗∗ 39.59∗∗ (7.59) -6.85) -6.13∗∗ (0.06 ∗∗∗ 43.18∗∗ (0.94 (5.65) -11.25 35.14) 15.39) (8.02) (7.29 (16.96∗∗ (6.31∗∗∗ (4.21) 10.76) 1.51) 10.18) -3.69) Pehyld (7) 11.22) 7.13) 18.74) Const.16) 17.20) 8.16 (8.86) -3.76) (8.59) (9.20 (15. and Industry Characteristics .97) -3.31) -8.04∗∗∗ (1.80) -3.24 ∗∗ 19.81 25.38 26.58 729 0.20 (4.01) -2.93) ∗∗∗ 22.94 (17.57 (15.76∗ (6.13) -2.98 25.36) (9.17) 28.25) (7.36) (7.81 (4.89 ∗∗∗ 42.85) -4.60 ∗∗ 21.64) (9.Pehyld Pehyld (1) 8.06 (17.25) -0.58) (9.49 ∗∗∗ 42.

8 52. S and P Code 4 12 5 6 2 4 5 6 7 8 9 10 11 5 10 2 6 11 6 10 11 10 6 Description Transportation Energy and Natural Resources Financial Institutions Healthcare/Chemicals Insurance and Real Estate Transportation Financial Institutions Healthcare/Chemicals High Technology/Office Equipment Aerospace/Auto /Capital goods Forest.0) (32.0) (65. Building Products/Home Builders Consumer/Service Sector Leisure Time/Media Financial Institutions Consumer/Service Sector Insurance and Real Estate Healthcare/Chemicals Leisure Time/Media Healthcare/Chemicals Consumer/Service Sector Leisure Time/Media Consumer/Service Sector Healthcare/Chemicals Year 1984 1986 1987 1987 1990 1990 1990 1990 1990 1990 1990 1990 1990 1991 1993 1994 1994 1994 1995 1995 1995 1996 1998 Table 8b: Industry Distress Behavior of Recovery Prices at Default (Pd) & Emergence (Pehyld.21)*** (4.statistic) 0.4 (41.85)*** (4.0 (39.2) (52.4 53.3) Obs 350 1312 1285 1285 Obs 320 1194 1167 1167 Distress Mean(Median) 26.9) Obs 37 161 158 157 Obs 10 43 42 42 t-statistic (z.9 (42.4 63.statistic) 4.07*** 4.5) (61.5) (51.4) Distress Mean(Median) 42.4 52.0 40.30** (2.0) (35.1 53.Table 8a: Industries in Distress. Pe).0 63.24)*** Full sample Mean (Median) 47. Recovery rates (Panel A) Pd Pe Pehyld Pecoup Recovery rates (Panel B) Pd Pe Pehyld Pecoup Full sample Mean (Median) 44.4 (41.15)** .0) Obs 387 1473 1443 1442 Obs 330 1237 1209 1209 No Industry Distress Mean(Median) 46.62) 2.5) (27.2 53.77*** 4.15** (2.5 47.52*** 4.9) (27.5 48. Pecoup.57*** (2.8 50.0) (66.62 (0.5 52.2 61.3) t-statistic (z.2 41.92)*** (4.5 64.8 53.8) (51.8 (19.0) (50.7) (65.33)** 2.0) (50.7) (48.0) (24.8 37.3) (51.03*** (4.8 40.62)*** 2.5) (34.4) (49.5 (38.3) No Industry Distress Mean(Median) 47.

37 28.1% 1.89 67.3% 19.49 47.75 38.74 14.9% 23.00 14.5% 16. Pe (mean) (median).00 53.9% 8.8% 16.4% 36.91 47.6% 15.188 -0.67 55.31 53.01 47.04 69.00 no data 5.08 52.51 no data 14.0% 14.6% 16.167 0.81 36.3% 25.93 40.3% 19.579∗∗ -0.60 44.30 80.59 63.30 72.0% 73.17 49.55 75.04 57.7% 16.6% 18.10 25. Pehyld (median) -0.54 Specific Assets Mean 14.8% 16.Excluding High Tech and Financial Institutions Specific Specific Specific Specific Assets Assets Assets Assets (mean).77 36.066 0.85 75.91 23.35 33.00 16.1% 2.224 -0.8% 18.14 100.4% 36.1% 2.5% 46.6% 15.9% 23.63 100.74 50.800∗∗∗ -0.42 no data 9.56 Panel D (Correlations).826∗∗ -0.57 41.086 0.8% 18.28 20.71 61.25 58.1% 1.52 55.99 60.896∗∗∗ -0.98 68.29 72.96 34.46 81. Pehyld (mean) (median).51 61. Pe (median) (mean).0% 14.566 -0.6% 30.80 no data 42.626∗∗ Industry Distress -0.80 26. Pe (mean) (median).80 39.63 28. Pehyld (mean) (median).79 58.65 28.35 50.46 100.688∗ 14.37 56.99 36.75 36.5% 19.06 53.36 Median 103.42 60.80 76.02 100.85 43.86 15.73 no data 3.26 82.00 51.0% 44.0% 44.9% 8.763∗∗∗ -0.67 74.54 52.6% Median 14.6% 16.75 52.Table 8c: Industry Distress Behavior of Recovery Prices at Emergence (Pehyld. Panel A (Pe) Description Utility Insurance and Real Estate Telecommunications Transportation Financial Institutions Healthcare / Chemicals High Technology / Computers / Office Equipment Aerospace / Automotive / Capital Goods / Metals Forest and Building Products / Homebuilders Consumer / Service Sector Leisure Time / Media Energy / Natural Resources Panel B (Pehyld) Utility Insurance and Real Estate Telecommunications Transportation Financial Institutions Healthcare / Chemicals High Technology / Computers / Office Equipment Aerospace / Automotive / Capital Goods / Metals Forest and Building Products / Homebuilders Consumer / Service Sector Leisure Time / Media Energy / Natural Resources Panel C (Correlations) No Industry Distress Specific Specific Specific Specific Assets Assets Assets Assets (mean).8% 29.25 20.5% 46.73 34.35 71. Pe) and Asset Specificity.3% 25.200 No Industry Distress -0.88 no data 30.7% 26.96 89.04 Industry Distress Mean no data 66. Pehyld (median) .45 55.09 56.7% 16.2% 0.00 57.10 35.6% 18.59 50.2% 0.04 48.53 47.00 47.072 Industry Distress -0.7% 26.6% 14.40 32. Pe (median) (mean).5% 16.5% 19.8% 29.64 42.0% No Industry Distress Mean 109.44 Median no data 39.015 0.6% 30.

17) 22.77) -11.77) -4.Table 9a .87) -167.13) -62.03 (9.31 (12.98 (0.61∗∗∗ (7.51) 14.57 242 0.14) Const.07 (0.59 709 0.26) 14.97 (11.80) -10.00 (12.46) 0.72) -3.00∗∗ (5.57) (10.14) 5.17) 14.08∗∗ (8.05) -8.54 ∗ 21.50) -44.56 (4.77∗∗∗ (7.73) -2.25) 3.10∗∗ (4.80 ∗ -13.11) 2.69 (9.26) Pehyld (4) 16.28) (8. Coupon Log(Issue Size) Time in Default Current Assets Unsecured Profit margin Tangibility Debt Concentration Distress1 Med Ind Q Ind Liq1 SR GDP BDA BDR Market SMB HML Utility Obs.85 (80.13 ∗∗ 17.27∗∗∗ (7.72) 26.31 (5.89 (7.06 (14.44) -17.81) -558.19) 12.70 (5.44) 5. R2 Pd (2) 14.05 (12.25 (18.30∗∗∗ (5.04 (150.91 (0.22) (7.22) 20.12) 1.Industry and Macro Characteristics Pd (1) 5.48 (9.60) 0.68) 25.18) 8.14 (0.04∗∗ (4.35) (7.31) 1.79) 5.60∗∗ (4.96) 18.47 (26.48∗∗∗ (7.10 (0.62) -0.47) 5.14) 14.57) (5.58 702 0.79) Pehyld (3) 22.18 (9.58 .14) -16.11) 1.20) Pehyld (5) 10.51) -11.08∗∗ (1.08) 2.41 (139.92 (300.58 (1.79) -4.98∗∗∗ (1.03∗ (294.97) -9.02∗∗∗ (7.13) -5.56 (43.74 (15.31) -3.03 (5.83 (11.73) 0.17 (9.61 (12.34) -0.25 ∗∗ 41.18) (10.18 (8.52 (9.57) 12.15) 242 0.04) -0.27 (4.39 (4.11 (18.98∗∗ (2.28 (1.Firm.54∗∗ (2.63) 23.Contract.09) 12.12) -4.46) 0.29 (14.00) -381.64) -0.69 (5.13) -5.31 (1.48) -17.52 709 0.19∗∗∗ (4.43∗∗∗ (1.07) -6.41∗∗ (5.42 ∗∗ 19.78 ∗∗∗ 41.77) (7.69) 8.48 ∗∗∗ 39.78 (9.

92 ∗ -0.57 (7.44) 41.32 (1.56 753 0.73 (13.61) Pehyld (5) 36.43) (2.04) 1.95∗∗∗ (1.55) 1.66∗∗ (7.77∗∗ (317.30∗ (4.75∗∗ (44.79∗∗ (7.92) 12.49) -1.13) 18.57) -0.18) -83.71 (0.91 (8.79 (8.10∗∗∗ (7.88) -101.37∗∗ (10.15∗∗∗ (2.56 (145.04) Pehyld (3) 49.89∗∗∗ (13.61∗∗∗ (7.65) 16.97) Const.07) -678.06) -39.26 (0.32 ∗ 1.98 (12.44 760 0.95∗∗ (9.27) 18.49) -0.30) 266 0.34 (1.67) -15.37) -717.60∗∗∗ (1.59) -14.00 (4.87 3.84 ∗ 1.07) (2.13) -3.91 (12.42 (140.47 (18.Table 9b .52∗∗∗ (260.56 .67 (10.54 (4.53) ∗ 4.91∗∗∗ (13.47 (8.42) Pehyld (4) 42.32) (1.05) -7.42) 21.15∗∗∗ (3.51) 42. R2 Pd (2) 18.13) -6.94) -5.82∗∗ (7.82) -4.49) -0.57 760 0.88∗∗∗ (6.73∗∗∗ (14.49) 13.23) 20.04) 17.36) 42.63) 15.16) -6.12) -0. Coupon Log(Issue Size) Time in Default Current Assets Unsecured Profit margin Tangibility Debt Concentration SR GDP BDA BDR Market SMB HML Utility Obs.69 (11. Firm and Macro Characteristics Pd (1) 18.75) 8.92) 9.34∗∗∗ (3.94) -7.50) (0.11 (0.39 (12.46 266 0.23∗ (4.04∗∗∗ (7.19 (0.31∗∗∗ (3.85 (4.45) 7.88) 6.80) 11.90) -181.66) -7.49) -0.70 (16.44) 13.61 (70.Contract.38 (11.

03 ∗∗∗ 47.36) 0.85 (5.44) 16.62∗∗∗ (10.90) 27.67 (11.64) (6.63 (14.96 (8.65 598 0.14) -6.77) 5.17) Pehyld (5) 15.60 ∗ 15.65 (1.87 (18.44) 2.57 ∗∗∗ 43.79 ∗∗ 9.18) 4.18) 36.89) -5.Table 10: OLS estimates of regression of Recovery Prices at default and emergence on risk factors that explain default.03) -0.79 (5.15) -14.81 (5.56 (13.94) 5.62) 5.33) 1.32 -0.51 (5.48 (8.69) -14.12) (8.58∗∗∗ (2.54 (3.13 (8.18) 211 0.20 (0.98∗∗∗ (10.45 ∗∗∗ -13.40∗∗∗ (8.36) Pd (3) 8.68 .31 ∗∗∗ 22. Score Distance to Default Utility Obs.03 (0.10) 44.73) -1.94) 28.72 (7.94) 4.39) 15.59) (7.43) 13.50) 0.76 (9.43 (17.15∗∗∗ (7.71) -4.48) 3.32∗∗∗ (1.19) -5.63 (8.96 (1.36) ∗∗∗ (1.01 ∗∗∗ Pd (2) 12.51 212 0.09) 5.19 (0.26) 6.96 (4.16) -14.16∗ (0.24) (5.14 ∗∗∗ 48.14∗∗∗ (2.18) -1.57) -4.22∗ (9.74∗∗ (9.19) (4.92∗∗∗ (0.76) 2.10) (0.60) 1.70∗∗ (8.15∗ (0.16) -1.85 (1.30 ∗∗∗ -20.84∗∗∗ (8.20) -0.26 (5.14∗ (8.14∗∗∗ (1.47) 6.48) 0.57) -17.24 (15.16) 11.65 (0.67) -0.28) 0.16) Pehyld (4) 13.22 (9.68 ∗∗ 18.90∗∗∗ (14.20) 28.99 (0.09) 22.28∗∗∗ (8.61∗∗∗ (5.59 (14. R2 32.88) 22.30∗∗ (5.78) 3.46∗∗∗ (9.42) 24. Coupon Log(Issue Size) Bank Loans Senior Secured Senior Unsecured Senior Subordinated Subordinated Time in Default Current Assets Unsecured Debt Concentration Distress1 Med Ind Q Ind Liq1 Z-Score Zmij.50) 44. Pd (1) Const.95 (13.63) -1.33 ∗∗ 25.71 (8.88) -14.87) 31.94 ∗∗ 3.76) 6.22∗∗∗ (1.09) 24.11) 0.11) -1.54) -0.09) (8.65 (14.79 (7.89) -0.80 ∗∗∗ 21.50 165 0.18) 16.38) 6.72) -8.27 (5.55 (5.23) -6.91 ∗∗∗ ∗∗ 0.84) 6.26∗ (1.48 26.94) 13.50) (7.49) 18.37) Pehyld (6) 10.19) (8.55∗∗∗ (6.12 (15.36) (4.01) -7.15) 2.71∗∗ (13.59 609 0.52 (8.86∗∗∗ (6.67 (8.10 (7.77 (11.57 395 0.01) 20.78) (8.47) (9.07 (14.25∗∗∗ (5.39) (1.50∗∗∗ (4.84 (19.92) -6.87) (8.62 (8.79) (9.

00 10.Time-series behavior of Recoveries at Default (Pd) and at Emergence (Pehyld) 90. median recovery price at default (Pd) in each year.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Years .00 50.00 40. .1982 to 1999 10 0 # Defaulted firms 50 40 30 20 70 60 Median Pd Median Pehyld #Firm Defaults Figure 1: Time-series behavior of Recoveries at Default (Pd) and at Emergence (Pehyld) This figure plots the time-series variation in the number of firm defaults (corresponding to Pd series).00 80.00 0.00 70.00 30.00 Cents per dollar 60. and median recovery price at emergence (Pehyld) in each year.00 20.

Sign up to vote on this title
UsefulNot useful

Master Your Semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master Your Semester with a Special Offer from Scribd & The New York Times

Cancel anytime.