Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Essays on corporate defined benefit pension plans and Chapter 11 bankruptcy Dimitrova, Milka 2015

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2015_november_dimitrova_milka.pdf [ 863.72kB ]
Metadata
JSON: 24-1.0132736.json
JSON-LD: 24-1.0132736-ld.json
RDF/XML (Pretty): 24-1.0132736-rdf.xml
RDF/JSON: 24-1.0132736-rdf.json
Turtle: 24-1.0132736-turtle.txt
N-Triples: 24-1.0132736-rdf-ntriples.txt
Original Record: 24-1.0132736-source.json
Full Text
24-1.0132736-fulltext.txt
Citation
24-1.0132736.ris

Full Text

Essays on Corporate Defined BenefitPension Plans and Chapter 11BankruptcybyMilka DimitrovaB.A., Ramapo College of New Jersey, 2008A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty of Graduate and Postdoctoral Studies(Business Administration)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)August 2015c© Milka Dimitrova, 2015AbstractIn this thesis, I present two essays on corporate defined benefit pension claimantsand Chapter 11 bankruptcy. First, defined benefit claimants are related to a lowerlikelihood that the firm files for Chapter 11 bankruptcy. Second, defined benefitclaimants influence the bankruptcy reorganization process beyond the role playedby the firm’s traditional creditors.In the first essay, I examine the role of defined benefit claimants in times lead-ing up to bankruptcy. Defined benefit claimants are less diversified and face highercosts of Chapter 11 bankruptcy than traditional lenders. I show that these differ-ences have implications for the likelihood that firms file for Chapter 11 bankruptcy:the higher the share of defined benefit liabilities relative to overall liabilities, thelower the likelihood of Chapter 11 bankruptcy. These results indicate that definedbenefit claimants’ incentives to keep the firm as a going concern matter for thefirm’s decision to file for Chapter 11 and should be considered in studies of debtrenegotiation between the firm and its creditors.In the second essay, I focus on defined benefit claimants in bankruptcy and theirimpact on the reorganization process. I provide evidence that pension claimantsinfluence the Chapter 11 restructuring beyond the impact of traditional lenders.In particular, defined benefit claimants play a role in the decision to terminate apension plan in bankruptcy, in the likelihood that firms refile for bankruptcy, and inthe amounts that unsecured creditors recover in bankruptcy. These results highlighta role for pension claimants in bankruptcy restructuring beyond that of traditionalcreditors. Additional tests indicate that one channel through which defined benefitclaimants influence the Chapter 11 process and its outcomes is by accepting cutsin their pension liabilities which cannot be explained by the average reductionsexperienced by other creditors. These findings highlight the role of defined benefitclaimants as an important player in bankruptcy restructuring.iiPrefaceThis dissertation is original, unpublished, independent work by the author, MilkaDimitrova.iiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Corporate Defined Benefit Pension Plans and Chapter 11 Bankruptcy 82.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82.2 Defined Benefit Claimants Prior to Bankruptcy . . . . . . . . . . 122.3 Data Sources and Summary Statistics . . . . . . . . . . . . . . . 162.3.1 Sample . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.3.2 Control Variables . . . . . . . . . . . . . . . . . . . . . . 182.3.3 Summary Statistics . . . . . . . . . . . . . . . . . . . . . 232.4 Determinants of the Likelihood of Bankruptcy . . . . . . . . . . . 242.4.1 Regression Specification . . . . . . . . . . . . . . . . . . 242.4.2 Determinants of a Chapter 11 Filing . . . . . . . . . . . . 252.5 Determinants of Bankruptcy for All Compustat Firms . . . . . . . 272.5.1 Sample of All Compustat Firms . . . . . . . . . . . . . . 282.5.2 Regression Specification . . . . . . . . . . . . . . . . . . 30iv2.5.3 Results for the Sample of All Compustat Firms . . . . . . 312.6 Robustness Tests . . . . . . . . . . . . . . . . . . . . . . . . . . 342.6.1 Funding Status . . . . . . . . . . . . . . . . . . . . . . . 352.6.2 Alternative Measure of Defined Benefit Liabilities . . . . 362.6.3 Event Study Around Chapter 11 Filing . . . . . . . . . . 382.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393 Corporate Defined Benefit Pension Plans in Bankruptcy Reorgani-zation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 543.2 Defined Benefit Claimants in Bankruptcy . . . . . . . . . . . . . 583.3 Sample and Data . . . . . . . . . . . . . . . . . . . . . . . . . . 633.3.1 Additional Datasources . . . . . . . . . . . . . . . . . . . 643.3.2 Summary Statistics . . . . . . . . . . . . . . . . . . . . . 683.3.3 Univariate Comparisons . . . . . . . . . . . . . . . . . . 713.4 Bankruptcy Duration . . . . . . . . . . . . . . . . . . . . . . . . 723.4.1 Regression Specification . . . . . . . . . . . . . . . . . . 723.4.2 Determinants of Bankruptcy Duration . . . . . . . . . . . 753.5 Emergence from Chapter 11 . . . . . . . . . . . . . . . . . . . . 773.5.1 Determinants of the Likelihood to Emerge . . . . . . . . . 773.6 Defined Benefit Plan Termination in Chapter 11 . . . . . . . . . . 793.6.1 Regression Specification . . . . . . . . . . . . . . . . . . 793.6.2 Determinants of Pension Plan Termination . . . . . . . . 813.7 Likelihood to Refile for Chapter 11 . . . . . . . . . . . . . . . . . 843.7.1 Regression Specification . . . . . . . . . . . . . . . . . . 853.7.2 Determinants of the Likelihood to Refile . . . . . . . . . . 863.8 Unsecured Creditors’ Recovery Rates . . . . . . . . . . . . . . . 883.8.1 Regression Specification . . . . . . . . . . . . . . . . . . 883.8.2 Determinants of Unsecured Creditors’ Recovery Rates . . 893.9 Changes in Pension Benefits During Bankruptcy . . . . . . . . . . 903.9.1 Regression Specification . . . . . . . . . . . . . . . . . . 903.9.2 Determinants of Changes in Pension Benefits in Bankruptcy 923.10 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94v4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119Appendix A: Accounting Standards for Defined Benefit Pension Plans . 119Appendix B: Employees’ Legal Rights in Bankruptcy . . . . . . . . . . 122Appendix C: Similarities and Differences Between Defined Benefit Obli-gations and Financial Liabilities . . . . . . . . . . . . . . . . . . 127viList of TablesTable 2.1 Variable Definitions . . . . . . . . . . . . . . . . . . . . . . . 43Table 2.2 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . 44Table 2.3 Determinants of Chapter 11 Likelihood . . . . . . . . . . . . . 47Table 2.4 Descriptive Statistics for Sample of All Compustat Firms . . . 48Table 2.5 Determinants of Chapter 11 Likelihood for All Compustat Firms:Pension Sponsors . . . . . . . . . . . . . . . . . . . . . . . . 50Table 2.6 Determinants of Chapter 11 Likelihood for All Compustat Firms:Pension Liabilities . . . . . . . . . . . . . . . . . . . . . . . . 51Table 2.7 Determinants of Chapter 11 Likelihood for Underfunded Plans 52Table 2.8 Determinants of Chapter 11 Likelihood Under ABO . . . . . . 53Table 3.1 Variable Definitions . . . . . . . . . . . . . . . . . . . . . . . 97Table 3.2 Sample Selection . . . . . . . . . . . . . . . . . . . . . . . . 98Table 3.3 Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . 99Table 3.4 Changes in Firm Variables- Before vs. After Bankruptcy . . . 102Table 3.5 Determinants of Bankruptcy Duration . . . . . . . . . . . . . . 103Table 3.6 Likelihood to Emerge from Bankruptcy . . . . . . . . . . . . . 104Table 3.7 Distressed Pension Plan Terminations . . . . . . . . . . . . . . 105Table 3.8 Likelihood to Refile for Chapter 11 . . . . . . . . . . . . . . . 107Table 3.9 Determinants of Unsecured Creditors’ Recovery Rates . . . . . 108Table 3.10 Changes in Pension Obligations in Bankruptcy . . . . . . . . . 109viiList of FiguresFigure 2.1 U.S. Corporate Debt from 1987 to 2012 . . . . . . . . . . . . 41Figure 2.2 Event Study Around Chapter 11 Filing . . . . . . . . . . . . . 42Figure 3.1 Changes in Pension Variables Before and After Bankruptcy . . 96viiiAcknowledgmentsI am profoundly indebted to my supervisor, Murray Carlson, and to Ron Gi-ammarino for their encouragement and constant support during the completionof my thesis and the job market process. They generously shared their time andknowledge with me.I am deeply grateful to the members of my committee Joy Begley, Hernan Ortiz-Molina, and Elena Simintzi for their help and constructive comments. I also thankthe faculty members in the Finance Division at the Sauder School of Business.Each of them helped me in different ways throughout the PhD program.I thank my fellow PhD students for many long and invaluable discussions. I amespecially indebted to my colleague and long-term friend Pablo Moran for his end-less support, help and advice, and for always being there for me. I am also gratefulto Alejandra Medina for all her help.I thank the University of British Columbia, the Sauder School of Business, and theResearch Award in the Economics of Pension Plans for providing financial supportto complete the PhD program.I would not have completed my thesis without the relentless support, encourage-ment, and love of my family. My parents, Nataliya and Valentin Dimitrovi, mysister Diana, and my grandmothers, Milka and Diana, always believed in me andstood by me throughout this journey. Without my family’s generosity and uncon-ditional faith in me I would not have achieved what I have.Milka DimitrovaVancouver, June 2015ixDedicationTo my family.xChapter 1IntroductionCorporate defined benefit pension obligations are sizable firm liabilities in theeconomy1. In 2013, U.S. corporations held $5 trillion in aggregate defined benefitliabilities and $18 trillion in financial liabilities2. The large amount and relative sizeof pension liabilities is not a recent phenomenon: Corporate defined benefit obli-gations have consistently represented close to a third of financial liabilities over thepast 20 years3. Despite pension claimants’ sizable liabilities, little is known aboutdefined benefit claimants and their role as firm creditors.While defined benefit liabilities are off-balance sheet obligations, the financeliterature has long recognized that these pension obligations resemble firms’ bal-ance sheet liabilities. Defined benefit pension obligations are similar to firms’ otherliabilities along several dimensions. First, firms are liable for the pension benefitsthey promise, just like for other corporate liabilities. Like other creditors, definedbenefit claimants are promised a fixed payout regardless of the financial perfor-mance of the assets that are set aside to meet the pension promises. Firms mustregularly contribute to the defined benefit plan to cover the pension promise and1In general, there are two types of corporate pension obligations: defined benefit and definedcontribution liabilities. In a defined benefit plan, the firm promises to pay participants a certain levelof retirement income, which is often based on employees’ years of service, age, and salary. Suchplans may distribute benefits in the form of a life annuity payable at a specified retirement age. In adefined contribution plan, the firm does not guarantee a particular level of income upon retirement.Instead, the employer and employer make certain contributions to an individualized account duringthe course of the worker’s employment. The focus of this thesis is on corporate defined benefitpension plans.2Note that in this dissertation, the term “defined benefit liability” is not limited to the accountingdefinition of the pension liability but instead refers to the entire projected benefit obligation reportedin the footnotes to firms’ financial statements. Therefore, the terms “defined benefit liability” and“defined benefit obligation” are used interchangeably.3I will use the terms “defined benefit” and “pension” as synonymous hereafter.1these contributions resemble bond interest (and principal) payments. Like inter-est payments, pension contributions are tax deductible at the corporate level. Inaddition, failure to contribute to the pension plan can trigger bankruptcy, just likemissing interest payments would. In default, pension liabilities can be senior or atpar with unsecured financial liabilities, and they are never junior to financial debt.Overall, the defined benefit pension commitment is similar to a firm’s legal promiseto pay off conventional debt on its balance sheet.However, defined benefit pension liabilities may not be perfect substitutes fortraditional debt. For example, the existence of government guarantees by the Pen-sion Benefit Guarantee Corporation sets aside defined benefit obligations fromother firm liabilities. As a result, pension obligations may encourage risk-takingbehavior since the firm may not bear the full costs of imposing high risk on de-fined benefit claimants. In addition, the off-balance sheet treatment and reportingdiscretion of pension liabilities differentiate them from traditional debt obligations.Even further, defined benefit claimants are less diversified than traditional lenders.Banks or hedge funds, for instance, have well diversified portfolios and one firm’sbankruptcy would rarely have a sizable effect on their wealth. Defined benefitclaimants, on the other hand, have their pension wealth invested in their employerand the firm’s bankruptcy would have sizable wealth effects for pension lenders.Altogether, defined benefit obligations differ from traditional liabilities along vari-ous dimensions.In this thesis, I exploit the wedge between defined benefit obligations and otherfirm liabilities to study how pension claimants influence firm decisions. I use Chap-ter 11 bankruptcy as a testing laboratory for my experiments. Chapter 11 is theideal setting to investigate the role of defined benefit claimants because bankruptcyis the only time when defined benefit plans can be terminated. As a result, Chap-ter 11 is particularly costly for pension beneficiaries. Although defined benefitclaimants may have to make concessions even in private negotiations, they bearthe additional risk of having their plan terminated in Chapter 11. Pension plantermination may lead to losses of pension benefits and future pension coverage,thus making Chapter 11 the costliest form of reorganization for defined benefitclaimants. Traditional lenders do not face such a loss in bankruptcy. Due to thehigher costs associated with Chapter 11 bankruptcy relative to other forms of reor-2ganization, defined benefit claimants’ role will be most pronounced prior to and inbankruptcy.To capture the role of pension claimants in bankruptcy, I control for firms’overall indebtedness and measure pension claimants’ influence relative to that ofother creditors. I account for all firm obligations by adding together firms’ totalbalance sheet liabilities and two of the largest off-balance sheet obligations, definedbenefit liabilities and operating leases. Since all liabilities become important inbankruptcy (Denis and Rodgers, 2007), I control for the role of the firm’s overallindebtedness and I capture any additional effect that defined benefit claimants mayhave by using the ratio of defined benefit liabilities to all firm liabilities. This ratiocaptures pension claimants’ influence relative to that of the firm’s other creditors.By controlling for firms’ overall indebtedness and focusing on the composition offirm lenders, I am able to investigate the incremental impact that defined benefitclaimants have beyond other lenders on the Chapter 11 reorganization process andits outcomes.Using these measures, I explore the impact of defined benefit claimants on thelikelihood of Chapter 11 bankruptcy and on the outcomes of bankruptcy restruc-turing. In Chapter 2, I investigate whether defined benefit claimants are related tothe likelihood that firms file for bankruptcy. Due to their lack of diversification andthe option to have their pension plan terminated, defined benefit claimants have ahigher incentive than traditional lenders to avoid Chapter 11 by negotiating withthe firm privately. In line with this prediction, I find that defined benefit claimantsare associated with a lower likelihood of bankruptcy: firms are less likely to file forChapter 11 whenever more of their liabilities are comprised of defined benefit obli-gations. The effect is economically significant: a one standard deviation increasein the contribution of defined benefit liabilities to overall liabilities is associatedwith a 6% decrease in the likelihood of bankruptcy with respect to the sample av-erage. These results extend to different specifications and alternative measures ofthe pension liabilities. My findings confirm that defined benefit claimants are morelikely to agree to concessions to keep the firm out of Chapter 11 than traditionalcreditors. Altogether, I provide evidence for a role of defined benefit claimants intimes leading up to bankruptcy.Since defined benefit claimants matter prior to bankruptcy, in Chapter 3 I also3consider if pension claimants influence the bankruptcy restructuring process. Thesimilarities and differences between defined benefit claimants and other lendersmay lead pension claimants to exert an influence beyond that of traditional lendersin bankruptcy. In support of this idea, I provide evidence that defined benefitclaimants matter for certain features of the reorganization process. In particular,I show that firms with more pension obligations relative to overall liabilities aremore likely to terminate a pension plan in bankruptcy. This result supports the firstessay’s argument that the expectation of plan terminations in bankruptcy providesan incentive for defined benefit claimants to avoid Chapter 11. I also documentthat pension claimants are associated with a lower likelihood that firms refile forChapter 11 post-reorganization. This finding is consistent with the first essay’s re-sults that defined benefit claimants are less likely to file for bankruptcy. In addition,I find some indication that pension claimants are associated with higher recoveryrates for unsecured creditors upon bankruptcy emergence. In light of the collectiveevidence that defined benefit claimants may play a role in bankruptcy above andbeyond the influence of traditional lenders, I try to identify the channels throughwhich pension claimants influence the reorganization process. I focus on benefitconcessions that pension claimants may agree to as one action that these claimantsmay undertake to impact bankruptcy restructuring. I find that the reductions thatpension claimants agree to are largely determined by pension claimants with higherunfunded liabilities relative to overall liabilities. My results indicate that definedbenefit claimants take deliberate actions to impact the bankruptcy restructuringprocess beyond the influence of traditional firm lenders in Chapter 11.In the literature, Ippolito (1985a) is one of the first authors to identify corporatedefined benefit claimants as firm bondholders. In his model of pension liabilities,Ippolito (1985a) shows that corporate pensions represent an implicit contract be-tween the firm and its employees because the firm promises workers a stable levelof income upon retirement in exchange for lower upfront compensation. This im-plicit contract gives pension claimants strong incentives to remain with the firmand to save in the firm by contributing more heavily to the pension4. By under-funding its defined benefit pension plan, the firm makes its employees long-term4Workers contribute to defined benefit plans by accepting lower wages upfront and higher pen-sions in the future.4bondholders in the firm. Thus, apart from their human capital investment, pensionclaimants have a direct financial stake in the firm in the form of an unsecured long-term bond. Ippolito (1985a) questions the optimality of turning employees intobondholders given that firm employees are less diversified than traditional lendersand do not have the level of sophistication of outside creditors. In a separate study,Ippolito (1985b) concludes that such a strategy may be optimal for firms that wantto avoid a potential hold-up problem created by powerful unions.As unsecured firm lenders, defined benefit claimants may influence variousfirm decisions, such as firms’ capital structure choice, for example. Arnott andGersovitz (1980) develop one of the first models that integrates corporate financialstructure and defined benefit liabilities. Corporate pensions are modeled as an in-strument which allows firms to simultaneously defer compensation and influencerisk-sharing between capital and labor. Arnott and Gersovitz (1980) show thatwhen firms cannot diversify risk completely, financial structure and employmentcontracts are determined simultaneously and are therefore interdependent. Shiv-dasani and Stefanescu (2010) provide empirical evidence in support of the viewthat firms make capital structure decisions with defined benefit pension liabilitiesin mind. In particular, they find that a 1 percentage point increase in the pensionliability to total assets ratio is associated with a 0.36 percentage points decrease inthe leverage ratio. Shivdasani and Stefanescu (2010) conclude that firms considerpension assets and liabilities in determining their leverage ratios and managers par-tially substitute pension-related deductions for interest deductions in capital struc-ture decisions.Apart from capital structure choice, the literature has also considered how de-fined benefit liabilities impact various other firm decisions. Rauh (2006) docu-ments that defined benefit pensions impact corporate investment. In particular, theauthor finds that pension sponsors decrease spending on capital expenditures wheninternal resources are reduced due to required pension contributions. Chang, Kang,and Zhang (2011) investigate the role of defined benefit claimants in firms’ invest-ment decisions as measured by mergers and acquisitions. The authors argue thatpension deficits serve as a control mechanism that limits managers’ discretionarypower and find that companies with unfunded pension liabilities are less likely toundertake diversifying mergers, they experience higher merger announcement re-5turns, pay lower premiums for targets, and use more cash as a method of payment.Cocco and Volpin (2012) study the role of pension liabilities from the perspectiveof target firms in mergers and acquisitions. They show that defined benefit pensionsponsors in the United Kingdom are less likely to be acquisition targets and if tar-geted, the deal is less likely to be completed. As a result, corporate defined benefitpension obligations are found to serve as a takeover deterrent.Lin, Liu, and Yu (2014) examine the role of pension liabilities in determiningfirms’ debt maturity structure. The authors argue that defined benefit obligationsincrease agency costs because the uncertainty in their valuation increases firm risk-iness to the benefit of shareholders. Lin et al. (2014) examine if firms use short-maturity corporate debt to mitigate these agency costs and show that there is apositive relationship between short-term debt levels and pension obligations.As unsecured firm creditors, defined benefit claimants influence firm deci-sions and may be relevant for firms in default. A substantial literature examinesbankruptcy and the renegotiation of creditors’ claims. Bulow and Shoven (1978)develop a model of the likelihood of bankruptcy which focuses on the conflicts ofinterest among three classes of claimants: bank lenders, bond holders, and stock-holders. In addition, James (1995) models the conditions under which bank lendersagree to concessions. The author shows that banks are less willing to scale downtheir claims in exchange for an equity position if they are not the only firm lender.Beyond negotiations with bank lenders in bankruptcy, the literature has in-vestigated the role of other firm lenders in Chapter 11 reorganization. For ex-ample, Hotchkiss and Mooradian (1997) consider vulture investors in financiallydistressed companies. The authors find that when vulture investors get involved inbankrupt companies, they purchase a significant amount of debt claims to influencethe terms of the restructuring and become active on boards and in management ofthe target companies. In addition, Hotchkiss, Smith, and Stro¨mberg (2012) doc-ument a role for private equity investors in bankruptcy. In default, firms backedwith private equity financing are more likely to restructure their debt claims inprivate negotiations, they restructure faster and are more likely to emerge as anindependent company following default. Last, Jiang, Li, and Wang (2012) provideevidence that hedge funds bring about efficiency gains in Chapter 11 bankruptcy.The authors show that hedge funds choose strategic positions in distressed compa-6nies to have the strongest impact on the reorganization process. In addition, hedgefunds are related to a higher likelihood that firms emerge from bankruptcy andthat unsecured creditors recover a higher portion of their claims. Overall, previousstudies have documented the importance of alternative firm lenders in bankruptcyproceedings.The current thesis makes several specific contributions to the literature. Thefirst essay contributes in two fronts. First, to the best of my knowledge, my studyis the first empirical work to explicitly consider the role of defined benefit pensionclaimants as firm lenders. I control for the overall financial position of the firm andI measure the role of defined benefit claimants as the relative contribution of theirliability to the overall firm obligations. Thus, I am able to focus on the compositionof firm creditors and to study one specific firm lender, defined benefit claimants,whose bargaining incentives in distress differ from those of traditional lenders.Second, I provide evidence that defined benefit claimants influence an importantfirm choice: the decision to file for bankruptcy. I find that pension claimants arerelated to a lower probability that firms file for Chapter 11. Therefore, my resultshighlight the need to account for defined benefit claimants in studies of bankruptcyprediction.The second essay contributes by showing that defined benefit claimants influ-ence the bankruptcy reorganization process above and beyond the impact of tradi-tional lenders. My results indicate a role for pension claimants in plan terminationsin bankruptcy and relate defined benefit claimants to lower probabilities of subse-quent bankruptcy and higher recovery rates for unsecured creditors. These resultsfurther reinforce the first study’s findings. Moreover, I document that defined ben-efit claimants’ willingness to accept concessions represents one channel throughwhich pension claimants can influence reorganization. I find that reductions in de-fined benefit obligations are determined mostly by the unfunded portion of pensionliabilities relative to other liabilities, after controlling for the expected losses forall creditors in bankruptcy. Therefore, the second essay provides insights about thespecific actions undertaken by defined benefit claimants in bankruptcy reorganiza-tion.7Chapter 2Corporate Defined Benefit PensionPlans and Chapter 11 Bankruptcy2.1 IntroductionCorporate defined benefit pension obligations are sizable firm liabilities. As Figure2.1 shows, in 2012 Compustat pension sponsors owed close to $5 trillion in definedbenefit liabilities, compared to $16 trillion in financial liabilities. The firm owesthese pension liabilities to its employees, the defined benefit pension claimants.While similar to traditional lenders, defined benefit claimants differ from the firm’sother creditors in the higher costs of bankruptcy that they face. In Chapter 11, de-fined benefit claimants stand to lose some of the benefits earned to date5 and possi-bly all future benefits if the plan is terminated6, along with any future salary lossesthey may incur. Despite their sizable liabilities and the high costs of bankruptcy,defined benefit claimants and their role in times leading to bankruptcy have re-5According to the Employee Retirement Income Security Act of 1974 (ERISA), retirees’ pen-sion income is partly guaranteed by a government entity, the Pension Benefit Guarantee Corporation(PBGC). In 2014, the maximum benefit guaranteed by the PBGC for a 65-year-old retiree is about$4,940 per month. For employees who are younger or have fewer years of service, the amount guar-anteed by the PBGC is lower. Therefore, if their defined benefit plan is terminated in bankruptcy andtransferred to the PBGC, employees whose pensions exceed the amount guaranteed by the PBGCor who do not meet the age requirement will receive an amount equal to the greater of the mini-mum PBGC guarantee and the amount funded by assets. Therefore, these employees will lose someunfunded benefits: the difference between what they were promised and the higher of the PBGCguarantee or the funded amounts, whichever is higher.6Under ERISA, fully funded and overfunded defined benefit pension plans may be terminatedat any time, but underfunded defined benefit pension plans can only be terminated in Chapter 11bankruptcy.8mained largely unexplored7.In this paper, I study defined benefit obligations in the context of Chapter 11bankruptcy. My conceptual framework is based on the idea that defined bene-fit claimants are less diversified than traditional lenders because these claimants’human capital and pension wealth are invested in the firm. As a result, definedbenefit claimants are more averse than traditional creditors to bad states of theworld such as bankruptcy. Bankruptcy is the perfect setting to study defined ben-efit claimants since the option to terminate pension plans in Chapter 11 makesbankruptcy a costlier form of reorganization than other types of default for pensionclaimants. Therefore, defined benefit claimants’ role will be most pronounced priorto and in bankruptcy. Due to their aversion to bad states, defined benefit claimantshave the incentive to avoid Chapter 11 by negotiating with the firm privately. Inline with this prediction, I find that defined benefit claimants are associated with alower likelihood of Chapter 11 bankruptcy.I study the role of defined benefit claimants in a sample of 481 defined ben-efit pension sponsors from 1987 to 2012, with 244 bankrupt sponsors matched tonon-bankrupt sponsors by industry, year, and size. I test whether conditional on thefirm’s overall indebtedness, the contribution of defined benefit liabilities to over-all liabilities impacts the likelihood of bankruptcy. To best capture firms’ overallindebtedness, I account for all of the firms’ balance sheet liabilities and two ofthe largest off-balance sheet obligations, defined benefit obligations and operatingleases. I find that firms are less likely to file for bankruptcy when more of theiroverall liabilities are comprised of defined benefit liabilities. The effect is econom-ically significant: a one standard deviation increase in the contribution of definedbenefit liabilities to overall liabilities is associated with a 6% decrease in the likeli-hood of bankruptcy with respect to the sample average. These results suggest thatdefined benefit claimants are more likely to make concessions in order to keep thefirm out of Chapter 11 than traditional creditors.While these findings indicate a role for defined benefit obligations in Chapter7Beyond the pension losses incurred in bankruptcy, defined benefit claimants stand to lose theirjobs, their future salaries, health insurance, and any benefits they earned through seniority, amongothers. As Ippolito (1985a, 2004) notes, if employees are paid above the competitive level to remainin the firm, a bankruptcy filing may lead to substantial losses for the firm’s employees.911, they apply to a subsample of defined benefit sponsors. To determine whetherthe results can be extended to all public firms, I re-estimate the main tests in theCompustat universe of all public firms from 1987 to 2012 which meet the origi-nal sampling criteria. I find that defined benefit sponsors are less likely to file forbankruptcy than firms without a defined benefit plan. Moreover, I find that thehigher the contribution of defined benefit liabilities to overall liabilities, the lowerthe likelihood of bankruptcy: a one standard deviation increase in the ratio of de-fined benefit obligations to overall liabilities is associated with a 1% decrease in thelikelihood of bankruptcy. Overall, these findings confirm that the results presentedin the main sample extend to the Compustat universe as well.In the main tests, I proxy for defined benefit claimants’ role in bankruptcy byconsidering the entire pension liability and not just the part of the liability that is notcovered by pension assets. Since defined benefit liabilities which are not coveredby assets may give pension claimants higher incentives to influence the bankruptcydecision, I re-estimate the main tests in a subsample of firms with underfunded de-fined benefit plans. Once again, I find that a higher contribution of defined benefitliabilities to overall liabilities is associated with a lower likelihood of bankruptcy,but the effect is not as statistically significant as in the main sample. Such a resultis consistent with the idea that the pension plan’s funding status represents onlya snapshot of defined benefit claims in a given year and ignores future promisesthat are important for pension claimants and influence their actions. Thus, usingonly the plan’s funding level may mask the role of defined benefit claimants inbankruptcy.As another robustness test, I consider an alternative measure of defined bene-fit obligations to the one used in the main specification. Throughout the paper, Imeasure pension liabilities as the pension benefit obligation (PBO) which equalsthe present value of benefits earned to date, assuming that the plan continues inthe future and employees’ salaries increase. In bankruptcy, firms are only liablefor the amount of benefits employees already earned and not for future benefits,so a measure capturing only earned benefits may be more relevant from the firm’sperspective. Therefore, I repeat the main tests with defined benefit liabilities mea-sured as just the benefits earned by employees to date assuming that the plan isdiscontinued (the adjusted benefit obligation, ABO). I confirm that my results hold10under this alternative definition of pension obligations8.My paper contributes to the literature on the role of defined benefit liabilitiesin corporate decisions. Prior work has documented that defined benefit obligationsinfluence firms’ capital structure decisions (Shivdasani and Stefanescu, 2010), cap-ital expenditures (Rauh, 2006), debt maturity (Lin et al., 2014), and mergers andacquisitions (Chang et al., 2011; Cocco and Volpin, 2012)9. I contribute to theliterature by showing that defined benefit claimants influence the decision to filefor bankruptcy, another important corporate decision. I provide evidence that de-fined benefit claimants are associated with a lower likelihood of bankruptcy. Myresults indicate that defined benefit claimants’ strong incentives to keep the firmalive matter for the firm’s decision to file for Chapter 11.Moreover, this essay is related to the literature on bankruptcy and the renego-tiation of creditors’ claims in default. In a study of bargaining in distress, James(1995) shows that banks are less willing to renegotiate the debt contract if theyare not the only firm lender. These results are reinforced by Colla, Ippolito, andLi (2013) who examine debt specialization and show that borrowing from mul-tiple lenders increases bankruptcy costs. In terms of accounting for alternativefirm lenders, prior studies have investigated the role of vulture investors in distress(Hotchkiss and Mooradian, 1997), private equity firms in default (Hotchkiss et al.,2012), and hedge funds in Chapter 11 (Jiang et al., 2012). My study contributesto the literature by considering the role of a set of claimants whose incentives toavoid bankruptcy and to keep the firm alive differ from those of the firm’s otherlenders. My findings show that defined benefit obligations provide explanatorypower beyond measures typically used in the literature to explain firms’ choiceto file for bankruptcy. Hence, accounting for defined benefit claimants is relevantwhen studying the renegotiation of firms’ claims in default.My paper is also closely related to the work of Benmelech, Bergman, and En-8Note that the PBO differs from the alternative measure, the ABO, only for plans which are stillactive and have not already been frozen. For frozen plans, the PBO will be approximately equal tothe ABO (Begley, Chamberlain, Yang, and Zhang, 2015).9Even further, prior studies have shown that defined benefit obligations are reflected in firms’market valuation (Franzoni and Marin, 2006), equity beta (Jin, Merton, and Bodie, 2006), debt rating(Carroll and Niehaus, 1998), and the pension funds’ asset allocation (Rauh, 2009). These studiesreinforce the importance of accounting for pension obligations in corporate finance studies.11riquez (2012) and Duan, Hotchkiss, and Jiao (2013). Benmelech et al. (2012) studythe role of defined benefit plans’ funding status and the threat of plan terminationin bankrupt airline companies when management bargains with labor. Duan et al.(2013) study defined benefit and defined contribution plans in distress. The au-thors document that defined benefit sponsors underfund their plans and reduce theamounts of own stock in the plans prior to debt default and that these actions areassociated with a higher probability of debt default. My paper is related to thesestudies as it shows that the composition of the firm’s creditors matters and account-ing for defined benefit lenders is crucial in studies of bankruptcy and renegotiation.The rest of the paper proceeds as follows. Section 2.2 discusses the role ofdefined benefit claimants prior to bankruptcy and develops the hypotheses understudy. Section 2.3 describes the data and presents summary statistics. Section 2.4outlines the results for the main sample used in this study. Section 2.5 considersan alternative sample and Section 2.6 presents several robustness tests. Finally,Section 2.7 concludes.2.2 Defined Benefit Claimants Prior to BankruptcyDefined benefit obligations represent firm liabilities which are largely notrecorded on the balance sheet but appear in the footnotes to firms’ financial state-ments10. Despite defined benefit claims being off-balance sheet obligations, thefinance literature has long recognized that pension liabilities resemble firms’ bal-ance sheet liabilities (Treynor, 1977). Ippolito (1985b) is one of the first papers toexplicitly model workers as long-term bondholders of the firm. Through the de-ferred pension compensation that employees accept, Ippolito (1985b) shows thatlabor holds a direct financial stake in the firm which is equivalent to an unsecured10Depending on the time period and the specific accounting rules effective at the time, somedefined benefit pension variables, such as the pension expense, prepaid/accrued pension cost, orthe plan funding status, among others, are reported on firms’ financial statements. However, thetotal amount of defined benefit obligations, the PBO, is only reported in the footnotes to financialstatements. For a more detailed discussion on the different reporting requirements over time, refer toAppendix A at the end of the thesis.12long-term bond11.While unreported on the balance sheet, defined benefit pension obligationsresemble firms’ other liabilities in various ways12. For example, defined bene-fit liabilities are similar to debt obligations because workers are promised a fixedpayout regardless of the financial performance of the assets that are set aside tomeet these promises (Shivdasani and Stefanescu, 2010). Firms are liable for thepension benefits they promise, just like for other corporate liabilities. Defined ben-efit pension sponsors must regularly contribute to the plan to cover pension obli-gations and these contributions resemble bond interest (and principal) payments.Like interest payments, failure to meet minimum pension contributions can triggerbankruptcy13. In default, pension liabilities can be senior or at par with unsecuredfinancial liabilities and they are never junior to financial debt. Overall, the definedbenefit pension commitment is similar to a firm’s legal promise to pay off conven-tional debt on its balance sheet.At the same time, defined benefit claimants differ from traditional lenders alongvarious dimensions. Of particular interest in this paper are two specific differences:the lack of diversification and the high costs of bankruptcy that distinguish definedbenefit claimants from traditional lenders (Ippolito, 2004). Traditional creditors,such as banks, have well diversified portfolios and one firm’s bankruptcy rarelyhas a sizable effect on their wealth. Defined benefit claimants, on the other hand,have their human capital and pension wealth invested in their employer. Therefore,defined benefit claimants are more averse to bad states of the world and thus havelarger incentives to avoid them than traditional lenders. While there are many def-initions of bad states of the world, such as bond downgrades, covenant violationsand debt default, among others, Chapter 11 is the ideal setting to test whether de-fined benefit claimants act differently from other creditors because bankruptcy is11For a detailed description of labor’s legal rights in bankruptcy, refer to Appendix B at the endof the thesis.12Appendix C presents a more in-depth discussion of the similarities and differences betweendefined benefit claimants and the firm’s other lenders than the discussion in the text.13A notable example is the case of LTV Corp which was forced into bankruptcy by the PBGC in1987. When ERISA was enacted, LTV’s pension plan was seriously underfunded and the companystopped contributing to its pension fund. As the size of the unfunded pension obligations grew, thePBGC increased its monitoring efforts. Finally, in 1987 the PBGC forced LTV in bankruptcy andtook over its defined benefit pension plan.13the costliest form of reorganization for pension claimants. Although defined ben-efit claimants may have to make concessions even in private renegotiations, theybear the additional risk that in Chapter 11, their pension plans can be terminatedand they may lose their jobs. Upon a distressed plan termination, pension claimantsstand to lose all pension benefits that exceed the minimum government guaranteeand are not covered by pension assets. These losses could be substantial for definedbenefit claimants (Ippolito, 2004). In addition, the promised pension amount is afunction of employees’ years of service, salary levels and mortality rates, amongothers. If employees have to switch between employers, both the years of serviceand salary amounts used to calculate the pension obligation will be lower at thenew employer compared to the existing one. Thus, the pension amount employeeswill receive will be reduced if they have to switch to a new job. Therefore, thedefined benefit obligations gives employees an incentive to salvage the firm frombankruptcy and to retain their job. Traditional lenders such as banks or hedge fundsdo not face such losses and incentives in bankruptcy. Due to their lack of diversifi-cation and high costs of bankruptcy, defined benefit claimants have high incentivesto renegotiate with the firm privately and to avoid the uncertain bankruptcy process.In light of these similarities and differences between defined benefit claimantsand other firm creditors, whether defined benefit claimants influence the likelihoodof Chapter 11 bankruptcy is an empirical question. To address this question, Icontrol for the role of traditional lenders in bankruptcy and capture any additionaleffect that defined benefit claimants may have. Firm leverage is the most commonmeasure used in the literature to account for the role of firm creditors in bankruptcy.However, the traditional leverage measure includes only short-term liabilities andlong-term debt and ignores all other firm obligations. Since firms consider theiroverall indebtedness when they decide whether to file for bankruptcy or not, I aminterested in accounting for all firm obligations. I capture all firm liabilities byaccounting for short-term liabilities and long-term debt, all other balance sheetobligations as well as two of the largest off-balance sheet liabilities: defined benefitobligations and operating leases. Thus, I am able to control for the influence of allmajor firm obligations on the decision to file for bankruptcy.To capture the role of pension claimants in times leading to bankruptcy, I needa measure of their influence relative to that of the firm’s other creditors. I proxy for14defined benefit claimants’ role with the ratio of defined benefit liabilities to overallliabilities as measured in the last fiscal year prior to the bankruptcy filing. Theratio captures pension claimants’ influence relative to that of the firm’s other cred-itors. In my tests, I account for the impact of the firm’s overall indebtedness andI study how the composition of firm lenders, and pension claimants in particular,influences the likelihood of bankruptcy. In this way, I am able to investigate the in-cremental impact that defined benefit claimants have on the bankruptcy probabilitybeyond that of the firm’s other lenders.Following the above reasoning, the main hypothesis under study in this paperis as follows:Hypothesis: Defined benefit pension liabilities are associated with a lower like-lihood of bankruptcy as compared to non-defined benefit liabilities (i.e. substitut-ing a dollar of financial liabilities for a dollar of defined benefit pension liabilitiesleads to a lower likelihood of bankruptcy).Overall, the paper’s main idea can be summarized with the following example:consider two hypothetical firms that are identical in all respects except for the sizeof their defined benefit obligations. Both firms have $100 of debt, but the composi-tion of their liabilities differs: one firm has $50 in financial debt and $50 in definedbenefit pension debt, and the other firm has $75 in financial debt ant $25 in pensiondebt. The question is, are these two firms equally likely to file for bankruptcy?In the main tests, I proxy for the role of defined benefit claimants in bankruptcywith the ratio of defined benefit pension liabilities relative to overall liabilities. Iconsider the full defined benefit liabilities rather than the unfunded portion of thepension obligation since focusing on the plan’s funding status has some importantlimitations. In particular, the funding status in a given year reflects the plan’s claimon cash flows for that year but ignores the impact the plan may have in futureyears. For example, for the same percentage decline in asset values, a large definedbenefit plan will experience a higher level of underfunding than a small plan andthus, the large plan will create a different level of bankruptcy risk for the sponsoringcompany. Arnott and Gersovitz (1980) and Shivdasani and Stefanescu (2010) build15a strong case for consolidating defined benefit assets and liabilities with firms’balance sheets assets and liabilities instead of focusing on the plan’s funding status.Following their reasoning, I study the full amount of the defined benefit pensionobligation and how it relates to the firm’s overall liabilities.2.3 Data Sources and Summary Statistics2.3.1 SampleThe sample in this paper is constructed to aid identification. Since the decisionto sponsor a defined benefit plan is endogenous, I focus only on firms with definedbenefit pension plans. In this way, the choice to sponsor a pension plan is takenfor granted and I compare the bankruptcy likelihood and outcomes across definedbenefit sponsors only. Moreover, previous studies on Chapter 11 have documentedthat bankrupt firms differ from non-bankrupt companies along various dimensions.Following the literature, I account for these differences by matching bankrupt com-panies to non-bankrupt firms in the same industry, closest in size and in the lastfiscal year prior to the bankruptcy filing (Altman, 1968; Chava and Jarrow, 2004).By design, I take the choice of sponsoring a defined benefit plan as given, andI compare bankrupt defined benefit firms to similar non-bankrupt defined benefitfirms. Thus, any marginal variation in defined benefit pension liabilities should beplausibly exogenous14.To create the sample, I start with all defined benefit sponsors in the CompustatPension Annual file from 1987 to 2012 with non-missing information on the pen-sion variables of interest15. The choice of 1987 as the starting year is warranted14The main underlying assumption behind the matching procedure is that the endogenous differ-ences between the two sets of firms are driven by observables which I am controlling for.15Apart from Compustat, pension data is also available from firms’ Form 5500 filings with theDepartment of Labor (DOL). However, Form 5500 data are difficult to link to Compustat sinceemployer identification number and firm name are the only firm identifiers. Even further, subsidiariesthat are more than 80% owned by the parent can report their pension obligations separately from theparent firm. Therefore, it becomes crucial to aggregate subsidiaries’ information with the parentfilings to obtain a full representation of the firm’s pension assets and liabilities from Form 5500filings (Rauh, Stefanescu, and Zeldes, 2012). Overall, use of Form 5500 data significantly constrainsthe sample size and also limits the accuracy with which I can identify defined benefit pension assets16by changes in accounting standards: until 1987, the accounting rules on definedbenefit pension disclosure were not standardized and thus, defined benefit pensionvariables are not comparable across firms or over time prior to this year. Moreover,significant pension-related assets and obligations prior to 1987 are not recognizedon firms’ financial statements. Therefore, my sample begins in 1987 when the ac-counting rules became standardized. The sample ends in 2012 since this is the lastyear for which I have information on Chapter 11 bankruptcies.Next, I obtain financial data for the sample of defined benefit pension sponsorsfrom Compustat. To alleviate concerns that Compustat pension data consolidatesdomestic and international pension plans, I select firms which are domiciled inthe U.S16. I restrict the sample to firms with total assets worth $100 million ormore, measured in 1980 dollars that file Form 10-K with the Securities and Ex-change Commission (SEC) not less than three years prior to the bankruptcy case.These filters are necessary because later I identify Chapter 11 cases from a samplewhich was constructed based on these criteria. In addition, I exclude all firms withmissing information for the main variables used in this study, as well as firms inthe utilities and financial industries. Industry affiliation is based on the Fama andFrench 48-industry classification. All firm-level variables are expressed in constant1987 dollars. To mitigate the influence of outliers, I winsorize all variables at the1% level. The sample consists of 2,007 unique defined benefit plan sponsors and20,546 sponsor-year observations.Out of the sample of defined benefit sponsors, I identify Chapter 11 filings us-ing Lynn M. LoPucki’s Bankruptcy Research Database (BRD). The BRD containsinformation on all bankruptcy filings (Chapter 11) and liquidation filings (Chapter7) from 1980 to 2012 for U.S. firms with assets of $100 million or more, measuredin 1980 dollars. From the original set of 961 filings on the BRD, I select all Chap-ter 11 filings by defined benefit sponsors in my sample17. The sample of bankruptand liabilities for some companies. For these reasons, I rely on Compustat for pension data.16While focusing on Compustat firms domiciled in the U.S. alleviates the concern about domesticand international plan consolidation to some extent, some U.S. firms can have foreign subsidiarieswhich sponsor a defined benefit pension plan in the foreign country. For those firms, the pensionvariables on Compustat will include both the domestic and the U.S. pension amounts.17In the sample period, there are 19 Chapter 7 filings. Of those, only 1 firm is not in the financialindustry, and there is no pension information for that firm. Thus, my sample is constrained to Chapter11 filings only.17defined benefit sponsors consists of 244 Chapter 11 filings by 219 unique pensionsponsors from 1987 to 2012.Next, I match bankrupt defined benefit sponsors to non-bankrupt defined ben-efit firms in the same industry, closest in size (defined as the logarithm of adjustedassets), and in the last fiscal year prior to the Chapter 11 filing. The matching pro-cedure is motivated by existing literature and helps mitigate the large difference inthe industry and size distribution across sample firms. In the final sample, there are481 matched firm-year observations representing 417 unique defined benefit plansponsors and 244 bankruptcy cases from 1987 to 2012.2.3.2 Control VariablesIn this paper, I am interested in controlling for the firm’s financial position andstudying the composition of firm creditors. While most studies focus on firms’short-term and long-term debt scaled by total assets as an estimate of firm indebt-edness, Welch (2011) points out that this measure captures only half of firms’ li-abilities. As a result, the author recommends using all balance sheet liabilities tobetter represent firms’ financial position. However, accounting for total balancesheet liabilities still ignores some large off-balance sheet obligations. In order tomore accurately measure firms’ indebtedness, I combine total balance sheet lia-bilities with two of the largest off-balance sheet firm obligations: defined benefitpension obligations and operating leases.Although a defined benefit pension plan is legally a separate entity from thesponsoring firm, pension benefits represent an integral part of a firms financialliabilities from an economic perspective (Treynor, 1977). As sponsors need to usetheir financial resources to fulfill pension obligations, defined benefit assets andliabilities should be analyzed in the context of the sponsor’s consolidated balancesheet. Following Shivdasani and Stefanescu (2010)’s consolidating approach, I addpension obligations to balance sheet liabilities and pension assets to firm assetsto better capture firms’ true financial position. Furthermore, a number of studieshave documented that operating leases are an important consideration for firms. Inparticular, Eisfeldt and Rampini (2009) and Rampini and Viswanathan (2013) showthat ignoring operating leases leads to understating firms’ true degree of leverage.18Furthermore, Rauh and Sufi (2012) provide evidence that accounting for operatinglease commitments improves the ability to explain capital structure variation inthe cross section. In light of these findings, I consolidate operating leases withbalance sheet assets and liabilities and defined benefit assets and obligations tobetter represent firms’ financial position and to account for all firm liabilities. Next,I turn to explaining in more detail how pension variables, operating leases, and allother control variables are constructed.Defined Benefit Assets and LiabilitiesThe accounting standards which govern defined benefit pension plan reportingon firms’ financial statements have changed over time18. In light of these changes,firms report different pension variables on their financial statements and in thefootnotes to these statements over time. Accordingly, the way I measure pensionassets and liabilities varies depending on the accounting reporting requirements.In particular, prior to 1998 firms reported pension variables separately for fundedand underfunded plans. Therefore, for these fiscal years, I measure defined bene-fit obligations as the sum of the funded pension obligations and the underfundedpension obligations from the footnotes to financial statements. Similarly, definedbenefit assets equal the sum of funded pension assets and underfunded pensionassets for fiscal years before 1998. After 1998, firms no longer had to differenti-ate between plans based on funding status but instead reported aggregated pensionassets and obligations from all defined benefit plans they sponsored. As a result,my measure of defined benefit obligations after 1998 equals the total value of allplan obligations reported in the footnotes to financial statements. In turn, definedbenefit assets after 1998 equal the value of all plan assets. Thus, I have a measureof defined benefit assets and obligations. I also measure a plan’s funding statusas defined benefit assets less defined benefit obligations scaled by defined benefitobligations.18Appendix A provides a more detailed discussion of the specific changes in accounting rulesover time.19Operating LeasesAlong with defined benefit assets and liabilities, I also consolidate operatingleases with balance sheet assets and liabilities in order to account for the firms’overall capital and liabilities19. Although operating leases are reported off the bal-ance sheet, numerous papers have demonstrated that operating leases are an impor-tant consideration for firms (Eisfeldt and Rampini, 2009; Rampini and Viswanathan,2013; Rauh and Sufi, 2012). Both leases and secured debt represent cash flow com-mitments the firm must make to continue using the asset. Given the similarity be-tween leases and secured debt, I integrate leases by capitalizing the operating leasecommitments and considering them as both an asset and a debt secured against theasset. Following Rampini and Viswanathan (2013), I use 10×the rental expense asa measure of leased capital20.Balance Sheet ConsolidationNext, I consolidate the off-balance sheet defined benefit assets and liabilities andoperating leases with balance sheet variables. While the total amounts of definedbenefit assets and obligations are only reported in the footnotes to financial state-ments, some other pension variables are reported on firms’ balance sheet and in-come statements. To avoid double-counting these variables, I subtract them whenI consolidate defined benefit assets and obligations with balance sheet amounts.Once again, different variables appear on financial statements depending on thetime period and I adjust my consolidated measures accordingly over time. Priorto 1998, if the pension expense21 exceeded the cash contributions the firm made19For accounting purposes, there are two different types of leases: capital leases and operatingleases. In a capital lease, the lessee treats the leased assets as if they have been acquired, so acapitalized asset and liability appear on the balance sheet. In an operating lease, the lease paymentsare expensed and no asset or liability is recorded on the balance sheet. Instead, operating leases arereported in the footnotes to the balance sheet.20If we assume that a firm borrows at a rate of 6%, the approximation of 10×the rental expenseused to calculate operating leases assumes a 15-year life of the lease.21The pension expense for a defined benefit pension plan is the employer’s annual cost of main-taining the pension plan. The pension expense depends on a number of future events, such as esti-mates of employees lifespan, employee tenure, and employees’ pay level just prior to retirement.20to the defined benefit plan, a pension liability was reported on the balance sheet.In turn, if the pension expense was lower than the firm’s cash contributions to theplan, a pension asset appeared on the balance sheet. However, as mentioned above,prior to 1998 all pension variables were reported separately for funded and under-funded plans, so firms’ balance sheets would contain a pension asset or liability forfunded plans and a pension asset or liability for underfunded plans. When I addtotal defined benefit assets and liabilities to balance sheet assets and liabilities, Isubtract the pension assets or liabilities which already existed on the balance sheetto avoid counting them twice22.Between 1998 and 2006, firms still reported an asset (a liability) on the balancesheet if the pension expense was higher (lower) than the cash contributions, butin this period firms aggregated pension variables across plans regardless of theirfunding status. Once again, for these fiscal years I subtract the pension amountswhich already appear on the balance sheet when I consolidate defined benefit assetsand obligations with the firm’s assets and liabilities. Lastly, for fiscal years after2006, firms no longer report the difference between the pension expense and cashcontributions on their financial statements. Instead, firms’ balance sheet include theplan’s funding status, defined as the difference between defined benefit assets andobligations from the footnotes. Therefore, when I consolidate defined benefit assetsand obligations with balance sheet amounts after 2006, I subtract the funding statusfrom consolidated assets when it is positive (i.e. when the plan in overfunded) andI subtract the funding status from consolidated liabilities when it is negative (i.e.when the plan is underfunded) since those amounts represent assets and liabilitieswhich already existed on the balance sheet.After accounting for the pension variables which already appear on financialstatements, I can consolidate the off-balance sheet amounts with balance sheet as-sets and liabilities. I construct two new measures which I refer to as adjusted assetsand adjusted liabilities. Adjusted assets include the book value of firm assets plusthe values of operating leases and defined benefit assets, less the pension assetsalready included on the balance sheet. I use adjusted assets as a measure of firmsize which accounts for both balance sheet and off balance sheet assets. Moreover,22If a firm’s balance sheet reports both a pension asset and a pension liability, I subtract thepension asset from the balance sheet asset and the pension liability from the balance sheet liability.21I scale all variables which are usually scaled by balance sheet assets by adjustedassets instead. Adjusted liabilities equal the sum of total balance sheet liabilitiesplus defined benefit obligations and operating leases, less the pension liabilitiesalready included on the balance sheet. I use adjusted liabilities as a measure offirm liabilities. In addition, I scale defined benefit obligations by adjusted liabili-ties to capture the role of pension claimants relative to all claimants, i.e. the sizeof their defined benefit pension plan in relation to all liabilities. Last, I define ad-ditional leverage as adjusted liabilities less short-term and long-term debt, scaledby adjusted assets. Additional leverage is a measure of firm debt which combinesbalance sheet liabilities apart from the traditional short-term and long-term debtwith the largest off-balance sheet obligations: pension obligations and operatingleases. In turn, financial leverage equals short-term and long-term debt scaled byadjusted assets. Therefore, the measure of leverage used in this thesis differs fromthe traditional measure used in prior studies in that it is scaled by adjusted assetsrather than just those assets reported on the balance sheet.Other Control VariablesFollowing the literature, I control for several variables that have been found tobe related to the likelihood that firms file for Chapter 11 bankruptcy. In particular,I include cash to adjusted assets and cash flow volatility as proxies for the firms’ability to meet short-term commitments and the potential to generate working cap-ital funds. Moreover, since less profitable firms are likely to become less liquidand more highly geared, I control for profitability, or return on assets, measuredas operating income before depreciation over adjusted assets. I use the market-to-book ratio as a measure of firm value. I control for firms’ dividend policy, R&Dexpenses and S&P credit ratings as those variables proxy for financial constraintsand firms’ overall health.Furthermore, to control for firms’ financial strength and the likelihood of de-faulting on debt agreements, I include firms’ S&P credit rating in all specifications.To use the rating in the main tests, I convert it to a numerical scale with values be-tween zero and one. Following the methodology outlined in Rauh (2009), if thefirm has a AAA S&P credit rating, the credit rating variable equals 0.929; if the22firm has a D rating, the credit rating variable equals 0.042; and each of the ratingsteps in between raises the credit rating variable by 0.042. In my sample, roughlyhalf of defined benefit sponsors have a bond rating, which is twice the number offirms in the Compustat universe. Observations with no credit rating receive a valueof zero for that data item. To account for the substitution of missing values withzeros, I include an indicator variable for observations with no credit rating.All other variables are measured as is standard in the literature and are definedin Table 2.1.2.3.3 Summary StatisticsTable 2.2 reports descriptive statistics for my sample of 481 defined benefit spon-sors. Panel A presents the industry distribution of sample firms. Most definedbenefit sponsors in my sample are in the Retail, Transportation, and Steel Worksindustries, with 15%, 8% and 7% of all sponsors, respectively. By construction,bankrupt defined benefit sponsors are matched to non-bankrupt sponsors in thesame industry so the industry distribution across bankrupt sponsors is identical tothat of all defined benefit firms in the sample presented in Panel A.Panel B of Table 2.2 presents summary statistics for firm and pension plancharacteristics for the full sample and for two subsamples of firms: bankrupt de-fined benefit sponsors and the matched defined benefit sponsors that are solvent.Defined benefit sponsors in the full sample have a median size of $511 million.Median firm size increases to $584 million when defined benefit assets and op-erating leases are accounted for. Tangible assets represent approximately 33% ofadjusted assets. Firms hold on average 7% of adjusted assets in cash and the aver-age return on assets equals 9%. In terms of off-balance sheet obligations, operatingleases represent on average 30% of adjusted assets. Defined benefit assets make upon average 15% of adjusted assets, whereas defined benefit liabilities are on aver-age 18% of assets. In addition, defined benefit obligations are on average 14% ofsample firms’ adjusted liabilities, which include both balance sheet and off-balancesheet obligations. Defined benefit firms in the full sample are 36% levered on av-erage. In turn, additional leverage which includes all balance sheet liabilities otherthan short-term and long-term debt, as well as defined benefit obligations and op-23erating leases equals on average 89% of adjusted assets.Bankrupt defined benefit sponsors differ from non-bankrupt sponsors acrossnearly all firm characteristics, which necessitates the matching procedure describedabove. As the last part of Panel B shows, matching eliminates the difference in sizeamong bankrupt defined benefit sponsors and control firms. In addition, severalother differences in firm-level characteristics become insignificant across the twotypes of firms. Altogether, the subsample of non-bankrupt defined benefit sponsorsmatched to bankrupt sponsors reduces the differences between defaulting firms andcontrols.Panel C in Table 2.2 reports the pairwise correlation coefficients for the mainvariables used this essay’s tests. As the coefficients in the table indicate, none ofthe variables are strongly correlated.2.4 Determinants of the Likelihood of BankruptcyIn this section I ask whether defined benefit claimants are related to the like-lihood that firms file for Chapter 11 bankruptcy. I control for the firm’s overallindebtedness and consider if the ratio of pension obligations to overall liabilitiesprovides explanatory power for the decision to file for bankruptcy. The premise ofthis design is that once I account for all firm liabilities, the contribution of pensionobligations to adjusted liabilities will capture the role of defined benefit claimantsin times leading up to bankruptcy.2.4.1 Regression SpecificationPrevious bankruptcy papers have implemented different specifications to studythe likelihood of Chapter 11, such as discriminant analysis (Altman, 1968), lo-gistic regression in event time (Lo, 1986), and logistic regression in panel data(Shumway, 2001). Due to endogeneity concerns, I restrict my attention to definedbenefit sponsors in bankruptcy and their matched counterparts. Therefore, the mostappropriate specification to study the role of defined benefit claimants on the prob-ability of Chapter 11 in this cross sectional sample is a logistic regression of the24likelihood to file for bankruptcy, estimated as follows:Bankruptit = α0 +α1(DB LiabilitiesAd justed Liabilities)it−1+α2Leverageit−1++α3Additional Leverageit−1 +α4Cashit−1 +α5CF Volatilityit−1++α6Tangibilityit−1 +α7ROAit−1 +α8Market to Bookit−1++α9Dividend Payerit−1 +α10R&Dit−1 +α11S&P Ratingit−1++α12No Ratingit−1 + εit−1(2.1)where Bankrupt is a dummy variable that equals one if a firm files for Chapter11 and zero otherwise and ε is the residual. The primary variable of interest is theratio of defined benefit liabilities to adjusted liabilities. The coefficient α1 is ofparticular interest as it elicits the fraction that defined benefit liabilities contributeto the firm’s overall indebtedness. Thus, the coefficient proxies for the influence ofdefined benefit claimants among all firm creditors.While size is an important determinant of bankruptcy, I do not control for firmassets in the regression specification because size is one of the dimensions on whichI match bankrupt sponsors to non-bankrupt controls. As the summary statistics inPanel B, Table 2.2, showed, the matching removed the size differences across thetwo subsamples. In addition, I do not include industry or year fixed effects in thisspecification since I also match firms on these two dimensions.2.4.2 Determinants of a Chapter 11 FilingThe results from estimating equation (2.1) are reported in Table 2.3. The first twocolumns report results in terms of marginal effects from a logistic regression. Thelast two columns report the coefficients from a linear probability model. Standarderrors are robust and are clustered at the industry level23.Model 1 in Table 2.3 considers only three determinants of the likelihood ofbankruptcy: the two measures of firm leverage and the ratio of defined benefit obli-23The results are robust to different clustering specifications.25gations to adjusted liabilities. This specification tests if the proportion of pensionliabilities relative to all firm liabilities influences the likelihood of bankruptcy whenthe total firm indebtedness is held constant. Consistent with previous studies, Ifind a positive relationship between firm leverage and the likelihood of bankruptcy.Moreover, additional leverage which includes all remaining balance sheet liabilitiesand the two largest off-balance sheet obligations is also a positive and significantdeterminant of the probability of bankruptcy. While higher leverage is associatedwith a higher probability of bankruptcy, when defined benefit liabilities constitutea larger share of overall liabilities, firms are less likely to file for bankruptcy. Inparticular, increasing the ratio of defined benefit liabilities to adjusted liabilitiesis associated with a marginal effect of -0.31 on the likelihood that a defined ben-efit sponsor files for bankruptcy. The interpretation of the marginal effects is asfollows: A one standard deviation increase in the measure of adjusted leverage isassociated with a 4% increase in the likelihood of bankruptcy. While firm leveragein terms of both financial liabilities and additional liabilities is associated with ahigher likelihood of bankruptcy, when these liabilities are composed of more de-fined benefit obligations, the likelihood of bankruptcy declines. Thus, the resultsfrom model 1 confirm that the composition of firm liabilities matters and that de-fined benefit claimants play a distinctive role in the likelihood of Chapter 11.The second model in Table 2.3 tests the full specification from equation (2.1)which includes other known determinants of bankruptcy. In the main specification,higher leverage and additional leverage are positively related to the likelihood ofChapter 11. At the same time, larger cash holdings, a higher share of tangibleassets and higher profitability are associated with a lower incidence of bankruptcy.In addition, firms that pay dividends and those that have high credit ratings are lesslikely to file for bankruptcy. The signs of the coefficients of these control variablesare in line with prior work on the determinants of Chapter 11 bankruptcy. The onlyexception is the negative sign on the indicator variable of no credit rating, whichsuggests that firms without a credit rating are less likely to file for bankruptcy.Holding the known determinants of bankruptcy constant, the results in model2 show that defined benefit claimants are related to a lower likelihood of Chapter11 bankruptcy. In particular, a one standard deviation increase in the proportionof defined benefit liabilities to adjusted liabilities is associated with a 3% decrease26in the likelihood of bankruptcy. The effect is economically meaningful: given theunconditional sample average of bankruptcy of 50%, defined benefit claimants areassociated with a a 6% decline in the probability of Chapter 11.The last two models in Table 2.3 estimate equation (2.1) in a linear probabilitymodel framework. The linear probability model results are included to gauge thelogistic regression estimates and their magnitudes. As the last two columns show,the results from the linear probability model support the findings reported in therest of the table. All signs and magnitudes in the linear model are close to identicalwith the marginal effects in the logistic model. Once again, the ratio of definedbenefit liabilities to adjusted liabilities is negatively related to the probability ofbankruptcy.In summary, I find that the higher the contribution of defined benefit obligationsto adjusted liabilities, the lower the likelihood that the firm files for bankruptcy.These findings are consistent with the idea that defined benefit claimants are morelikely to avoid reorganization in Chapter 11. Pension claimants have high incen-tives to avoid bankruptcy due to their lack of diversification and because of thehigher costs of Chapter 11 they face compared to traditional lenders.2.5 Determinants of Bankruptcy for All CompustatFirmsThe main sample used in this study is constructed with the intent to addressendogeneity concerns. However, it is not clear to what extent the main sample re-sults apply to the broader universe of Compustat firms. The main sample focuseson defined benefit sponsors only, with bankrupt sponsors matched to comparablenon-bankrupt sponsors. To determine the extent to which the main sample resultsextend to all firms, in this section I consider the role of defined benefit claimantsin the full Compustat universe. In the sample of all Compustat firms, I estimatetwo different specifications to determine if defined benefit claimants influence thelikelihood of bankruptcy. First, I control for the role of defined benefit claimantsby using a dummy variable that identifies defined benefit sponsors. Second, I con-trol for the role of pension claimants with the ratio of defined benefit liabilities to27overall liabilities, as in equation (2.1).2.5.1 Sample of All Compustat FirmsIn order to construct the sample of all Compustat firms, I start with all companieson Compustat from 1987 to 2012 which meet the original BRD sampling criteria:have assets worth $100 million or more in 1980 dollars, and file Form 10-K withthe SEC within three years prior to the bankruptcy case. I identify defined benefitplan sponsors among all firms from the Compustat Pensions Annual file. Next, Iidentify Chapter 11 cases from the BRD. I delete all firms with missing data forthe main control variables used in this study, as well as firms not domiciled inthe U.S and those in the utilities and financial industries. All firm-level variablesare expressed in constant 1987 dollars and are winsorized at the 1% level. Thefinal sample used in this section consists of 4,708 unique firms (41,225 firm-yearobservations) and 461 Chapter 11 filings by 430 unique firms from 1987 to 2012.Table 2.4 provides summary statistics for the sample of all Compustat firms.Panel A outlines the industry distribution of three groups of firms: all companiesthat do not sponsor a defined benefit plan, all defined benefit sponsors, and allbankrupt firms from the Compustat universe. Most firms without a pension plancome from the Business Services, Retail and Petroleum industries, constituting16%, 12% and 7% of all companies, respectively. Alternatively, among definedbenefit sponsors, Machinery, Retail, and Chemicals represent the industries withhighest firm concentration, with 7%, 7%, and 6%, of all firms, respectively. Theindustry distribution of defined benefit sponsors in the full Compustat databaseis comparable to that in the main sample of defined benefit sponsors only, wherethe most represented industries are the Retail, Transportation, and Steel Works.Overall, Panel A of Table 2.4 shows that in the sample of all Compustat firms,defined benefit sponsors are largely comparable to firms without a defined benefitplan in terms of industry representation. The last column of Panel A shows theindustry distribution of bankruptcy cases in the sample of all Compustat firms.Most bankruptcies in both the full Compustat sample and the main sample occurin the Retail and Transportation industries.Panel B of Table 2.4 presents summary statistics for firm and defined benefit28plan characteristics for all Compustat firms and for two subgroups of firms- thosethat sponsor a defined benefit plan and those that do not. In the sample of allCompustat firms, average firm size is $569 million in terms of balance sheet assetsand $600 million in terms of balance sheet and off-balance sheet amounts. Onaverage, 32% of adjusted assets are invested in property, plant and equipment andthe average return on assets is 13%. Firms hold 12% of adjusted assets in cash andcash flow volatility is close to 2% on average. Out of all Compustat firms, close to50% sponsor a defined benefit pension plan. Sample firms hold on average 27% ofadjusted assets in financial leverage and another 61% in additional leverage. Last,1% of all Compustat firms file for Chapter 11 bankruptcy.The next part of Panel B shows that defined benefit sponsors have a median sizeof $962 million which increases to $1,074 million when off-balance sheet assets areincluded. Tangible assets represent 33% of adjusted assets on average and returnon assets is 14%. Defined benefit sponsors hold an average of 8% of adjustedassets in cash and the volatility of their cash flows is 1%. Pension sponsors have anaverage market-to-book ratio of 1.8 and invest 2% of adjusted assets in research anddevelopment. In terms of non-balance sheet obligations, median operating leasesto adjusted assets equal 14%. Defined benefit assets represent on average 15% ofadjusted assets, whereas defined benefit liabilities are on average 16% of adjustedassets. In addition, defined benefit obligations equal 15% of adjusted liabilitieson average. Defined benefit firms are 27% levered in terms of financial leverageand 58% levered in terms of adjusted leverage. In addition, 7% of defined benefitsponsors file for Chapter 11 bankruptcy. Overall, defined benefit sponsors in thefull sample of all Compustat firms are comparable to defined benefit sponsors inthe main sample.In comparison, non-defined benefit firms differ from pension sponsors along allfirm characteristics. In particular, firms that do not sponsor a defined benefit planare smaller, less profitable, have less tangible assets, are less likely to pay dividendsor have a credit rating than defined benefit sponsors. Moreover, firms withouta pension plan are less levered in term of both financial leverage and additionalleverage, yet they are more than twice as likely to file for bankruptcy as definedbenefit sponsors. Overall, the summary statistics for all Compustat firms highlightvast differences among firms which sponsor a defined benefit plan and those that29do not.2.5.2 Regression SpecificationIn this subsection, I study the impact of defined benefit claimants on the likeli-hood of bankruptcy in the sample of all Compustat firms. To proxy for the influ-ence of pension claimants, I first use a dummy variable equal to one if a firm in thesample sponsors a defined benefit pension plan, and zero otherwise. I estimate thefollowing specification:Bankruptit = α0 +α1DB Sponsorit−1 +α2Leverageit−1++α3Additional Leverageit−1 +α4Sizeit−1 +α5Cashit−1++α6CF Volatilityit−1 +α7Tangibilityit−1 +α8ROAit−1++α9Market to Bookit−1 +α10Dividend Payerit−1 +α11R&Dit−1++α12S&P Ratingit−1 +α13No Ratingit−1 + γ j + γt + εit−1(2.2)where DB Sponsor is an indicator variable which equals one if firm i sponsorsa defined benefit pension plan in year t and zero otherwise, all other control vari-ables are defined as previously24, and γ j and γt are industry and year fixed effects,respectively. The DB Sponsor indicator variable captures the effect of defined ben-efit claimants on the likelihood of Chapter 11 bankruptcy.Since the dependent variable in equation (2.2) is binary, the natural specifica-tion of choice would be to estimate a logistic model. However, econometric esti-mators in binary response models with fixed effects may fail to converge on con-sistent estimators as the number of observations becomes large, a problem knownas the incidental parameters problem (Neyman and Scott, 1948). Several solutionshave been proposed in the literature to deal with this problem including estimat-ing a linear probability model or a conditional logistic regression. Therefore, forall specifications, I report regression results from both the logistic model as well24Variable definitions are also provided in Table 2.1.30as linear probability model and a conditional logistic regression. Since the condi-tional logistic regression results are not reported as marginal effects, the coefficientestimates are not comparable with those in the other specifications. Nevertheless,the conditional logit results are included for comparison.Along with the pension sponsor indicator variable, I also consider the ratio ofpension liabilities to overall liabilities as a measure for the role of defined benefitclaimants. As a second specification, I replicate the estimation procedure in equa-tion (2.1) on the sample of all Compustat firms. To retain non-defined benefit firmsin the estimation, I interact the ratio of defined benefit liabilities to overall liabil-ities with the defined benefit sponsor dummy. Therefore, firms without definedbenefit liabilities will have a value of zero for the ratio of defined benefit liabilitiesto adjusted liabilities. In particular, I estimate the following variant of equation(2.1):Bankruptit = α0 +α1(DB LiabilitiesAd justed Liabilities)it−1× (DB Sponsor)it−1++α2DB Sponsorit−1 +α3Leverageit−1 +α4Additional Leverageit−1++α5Sizeit−1 +α6Cashit−1 +α7CF Volatilityit−1 +α8Tangibilityit−1++α9ROAit−1 +α10Market to Bookit−1 +α11Dividend Payerit−1++α12R&Dit−1 +α13S&P Ratingit−1 +α14No Ratingit−1 + γ j + γt + εit−1(2.3)All variables are defined in Table 2.1. In this specification, including the DBSponsor indicator variable accounts for the substitution of missing values for pen-sion liabilities with zeros. Once again, the dummy captures the effect of sponsoringa defined benefit plan on the likelihood of bankruptcy. Moreover, equation (2.3) isestimated in a logistic, linear probability, and conditional logistic frameworks.2.5.3 Results for the Sample of All Compustat FirmsIn this subsection, I discuss the results from investigating whether the negativerelationship between defined benefit claimants and the likelihood of bankruptcy31which is found in the main sample of this study is also present in the full Compustatuniverse.First, I capture the impact of pension claimants on the probability of bankruptcywith a dummy variable for whether a firm in the sample sponsors a defined benefitplan or not. The results from estimating equation (2.2) are reported in Table 2.5.The dependent variable in all specifications equals one if a firm files for Chapter 11bankruptcy in a given year, and zero otherwise. All specifications include industryand year fixed effects and standard errors clustered at the industry level.Model 1 in Table 2.5 estimates the relationship between the likelihood ofbankruptcy and the defined benefit sponsor dummy variable. Such a test capturesthe likelihood of bankruptcy for firms which sponsor a defined benefit plan ver-sus those that do not. As the results show, defined benefit sponsors are negativelyand significantly related to the likelihood of Chapter 11 bankruptcy. In particu-lar, switching from not sponsoring a defined benefit plan to sponsoring a definedbenefit plan is related to a 1% decline in the probability of bankruptcy. In model2, I control for the pension dummy as well as financial leverage and additionalleverage. This specification considers if defined benefit claimants are related tobankruptcy after controlling for overall firm indebtedness. Higher leverage in termsof both financial liabilities and additional liabilities is a positive and significant de-terminant of the likelihood of Chapter 11 bankruptcy. Holding leverage constant,defined benefit sponsors are less likely to file for bankruptcy than non-defined ben-efit firms.Model 3 presents the results for the main specification in equation (2.2) whichincludes all firm-level controls. In line with the literature and with the results in Ta-ble 2.3, higher leverage is positively related to the likelihood of bankruptcy whereaslarger cash holdings, higher profitability, higher market-to-book ratios, dividendpayouts, and higher credit ratings are all negatively related to the incidence ofbankruptcy. The negative and statistically significant coefficient of the DB Spon-sor variable holds when firm characteristics are accounted for. The marginal effectfrom the logistic specification shows that switching from not sponsoring a definedbenefit plan to sponsoring a defined benefit plan is related to a 1% decline in theprobability of bankruptcy. Overall, the results from estimating equation (2.2) showthat in the full Compustat sample, defined benefit sponsors are less likely to file for32bankruptcy than firms without pension plans.The rest of Table 2.5 reports results of the same three specifications under dif-ferent estimation procedures. Models 4 through 6 present estimates from linearprobability regressions and models 7 through 8 report coefficients from conditionallogit estimations. These estimations are included in the table to ensure that the re-sults presented in the main specification are consistent and not influenced by theincidental parameter problem discussed above. Overall, the linear and conditionallogistic models confirm the main results presented in the first three columns. De-fined benefit sponsors are less likely to file for Chapter 11 bankruptcy than firmswithout defined benefit claimants under all specifications. All other explanatoryvariables have the same signs and significance across the alternative models.Second, I capture the impact of pension claimants on the probability ofbankruptcy with the ratio of defined benefit liabilities to overall firm liabilities.The results from estimating equation (2.3) are reported in Table 2.6. The depen-dent variable in all models equals one if a firm files for Chapter 11 bankruptcy ina given year, and zero otherwise. All specifications include industry and year fixedeffects and standard errors clustered at the industry level.Model 1 in Table 2.6 reports estimation results from a logistic regression ofthe probability of bankruptcy on overall leverage and the ratio of defined benefitliabilities to adjusted liabilities. Higher financial leverage and additional leverageare both associated with a higher likelihood of Chapter 11. However, controllingfor overall firm indebtedness, defined benefit claimants are related to a lower in-cidence of bankruptcy. In particular, a higher ratio of defined benefit liabilities toadjusted liabilities is associated with a -0.04 marginal effect on the likelihood ofbankruptcy. Overall, the results from the first model indicate that while firm lever-age is associated with a higher likelihood of bankruptcy, when more of the firm’sliabilities are composed of defined benefit obligations, the likelihood of bankruptcydeclines. Therefore, the composition of firm creditors matters in the full sample aswell.Results from the main specification from equation (2.3) including all firm-levelcontrols are presented under model 2 in Table 2.6. In the full specification, morelevered firms are more likely to file for bankruptcy. At the same time, healthierfirms with larger cash holdings, higher asset returns, more growth options, divi-33dend payouts and higher credit ratings are less likely to file for Chapter 11. Therole of defined benefit claimants in the full specification remains unchanged: firmswith more pension claimants relative to all other creditors are less likely to gobankrupt. Most importantly, the coefficient for the ratio of defined benefit liabil-ities to adjusted liabilities retains the same magnitude and significance when thefirm controls are included. These results confirm that the contribution of definedbenefit liabilities to adjusted liabilities remains a significant determinant of thelikelihood of Chapter 11 bankruptcy even after controlling for known predictors ofbankruptcy in the sample of all Compustat companies.The remainder of Table 2.6 re-estimates the first two models under a linearprobability and a conditional logit frameworks. The main variable of interest, theratio of defined benefit liabilities to adjusted liabilities, retains the same sign andsignificance through these alternative specifications. In addition, the coefficientsand significance of all other firm controls remain largely unchanged. Therefore,the results presented in the first two models are robust to the specification used.Altogether, the results for firms in the Compustat universe confirm the mainsample findings that defined benefit claimants are associated with a lower likeli-hood of bankruptcy. Among all Compustat firms, companies with defined bene-fit claimants are less likely to file for Chapter 11 than firms with only traditionallenders. Even further, the ratio of pension obligations to overall liabilities whichcaptures the relative role of defined benefit claimants compared to other firm cred-itors is related to a lower likelihood of bankruptcy in the full Compustat sample.These results confirm that the main sample findings extend to the larger universeof firms.2.6 Robustness TestsIn this section, I perform several robustness tests. In particular, I consider the re-lation between defined benefit claimants and bankruptcy under an alternative proxyfor their influence and under a different measure of defined benefit liabilities. I alsocheck whether investors incorporate the role of defined benefit claimants in timesleading to bankruptcy as reflected in the stock market returns around bankruptcy34announcement.2.6.1 Funding StatusThe main proxy for the impact of defined benefit claimants on the likelihood ofbankruptcy used in this study is the ratio of defined benefit pension obligations tooverall firm liabilities. While defined benefit claimants care about the full amountof their promised future income, it is possible that the portion of their liabilitieswhich is not covered by any assets may give them stronger incentives to influencethe bankruptcy decision. In this subsection, I test whether this argument holdsin my sample. If defined benefit claimants’ actions are driven by the unfundedportion of their liabilities, then the effect of the ratio of pension obligations tooverall liabilities will be even stronger for the subsample of firms with underfundedpension plans. To test this conjecture, I re-estimate equation (2.1) in the set of 351firms with underfunded plans only. Table 2.7 reports these estimation results. Thedependent variable in all models equals one if a firm files for Chapter 11 bankruptcyin a given year, and zero otherwise. Standard errors are robust and clustered at theyear level.Model 1 in Table 2.7 reports results for the likelihood of bankruptcy whenfirm indebtedness and pension claimants are considered. The positive relation-ship between financial leverage and additional leverage and bankruptcy confirmsthat leverage is a significant determinant of the likelihood of Chapter 11 for un-derfunded plans. Moreover, the ratio of defined benefit obligations to adjustedliabilities is negatively related to the probability of bankruptcy in the sample offirms with poorly funded pension plans. However, the marginal effect of the proxyfor defined benefit claimants in this model is only significant at the 10% level.The next model in Table 2.7 considers the role of defined benefit claimantswhen all firm-level controls are accounted for. In line with the results presentedthus far, higher leverage positively predicts the likelihood of bankruptcy while cashholdings, profitability, dividends, and high credit ratings are negatively related tothe probability of bankruptcy. More importantly, the ratio of defined benefit obli-gations to adjusted liabilities is negatively related to the likelihood of bankruptcyin the sample of underfunded plans when firm controls are included in the specifi-35cation.The last two models in Table 2.7 report estimates from a linear probabilitymodel for the same specifications used in the first part of the table. The linearmodels’ results reinforce the findings from the logistic regression and indicate arole for defined benefit claimants in firms with underfunded pension plans, albeit aless significant one.In untabulated tests, I compare the results for the group of firms with under-funded pension plans to those of firms with fully funded and overfunded plans.Such a comparison helps determine whether defined benefit claimants’ bargainingpower prior to bankruptcy is driven solely by the underfunded portion of their obli-gations or whether the full amount of the pension obligation matters to pensionclaimants. There are 130 firms in the sample whose defined benefit plans are fullyfunded or overfunded. I estimate equation (2.1) on this subset of firms and findresults comparable to those for the sample of underfunded pension plans. Overall,I find no heterogeneity across the two subsamples in terms of pension claimants’impact on the likelihood of bankruptcy. These results provide some support forthe idea that underfunding does not drive the main results presented in this paper.However, given the small number of firms with fully funded defined benefit plans,the data may not be rich enough to disentangle these effects.2.6.2 Alternative Measure of Defined Benefit LiabilitiesIn the main tests, I measure defined benefit liabilities as the projected benefitobligation (PBO), which is the present value of all benefits earned by employees todate for service rendered prior to that date, plus the present value of projected bene-fits attributable to future salary increases. An alternative measure of defined benefitliabilities is the accumulated benefit obligation (ABO), which is the present valueof benefits earned by employees for services rendered to date assuming the pensionplan is terminated in the same year. In bankruptcy, the firm is only liable for theABO portion of the pension obligation as bankrupt firms are held responsible forpaying for services already rendered and not for future expected benefits.In this subsection, I study the role of defined benefit claimants as measuredby the ratio of ABO liabilities to adjusted liabilities, instead of the PBO liabilities36to adjusted liabilities used in the main specifications. Once again, I account forthe different accounting standards over time and consolidate underfunded ABOobligations whenever necessary. Due to data issues, only 260 of the 481 samplefirms report ABO liabilities, which does not allow me to compare the main sampleresults to these small subsample results. Therefore, I study the impact of usingthe ABO in the sample of all Compustat firms. In the Compustat universe, 14,951firm-year observations out of the 41,225 observations have ABO data. Results fromestimating equation (2.1) with the ABO measure replacing the PBO are reported inTable 2.8. The dependent variable in all specification is the bankruptcy indicator.The first model in Table 2.8 controls for leverage and the role of defined benefitclaimants as measured by the ratio of ABO liabilities to adjusted liabilities. In thefull Compustat sample, higher leverage leads to higher incidence of bankruptcybut when firm liabilities are comprised of more pension obligations, the likeli-hood of Chapter 11 declines. In particular, a higher ratio of ABO liabilities toadjusted liabilities is associated with a -0.02 marginal reduction in the probabilityof bankruptcy. Hence, the composition of firm lenders matters even when definedbenefit liabilities are measured as earned pensions instead of earned and futurepensions.The results from the first model are robust to the inclusion of the remaining firmcontrols. As the second column in Table 2.8 shows, the other firm-level controlsenter the regression specification with similar magnitudes and the same signs asin the previous tables. The main specification indicates a negative relationshipbetween pension claimants and Chapter 11 bankruptcy. These results are furtherconfirmed in the linear probability model results presented in the last two columnsof Table 2.8.Overall, the results in Table 2.8 confirm that defined benefit claimants are re-lated to a lower likelihood of bankruptcy regardless of the proxy used to estimatetheir role. The PBO is the main measure of the role of defined benefit claimantsused in this essay’s tests for several reasons. While the firm is only liable for theABO portion of defined benefit obligations upon bankruptcy, the ABO ignores anyprojected benefit and salary increases beyond the estimation date. In addition, fromthe standpoint of pension claimants, both the amounts they currently accumulatedand future increases are relevant since that was the pension amount promised to37them by the firm. Ippolito (1985a) makes a similar observation that under theimplicit contract view of pensions, defined benefit claimants care about the fullpension amounts promised to them, rather than just the ABO amount. Therefore,while the main results hold under either specification, the main tests focus on thePBO as the closest proxy for the role of defined benefit claimants in times leadingto bankruptcy25.2.6.3 Event Study Around Chapter 11 FilingIf defined benefit claimants are associated with a lower likelihood of bankruptcy,investors’ valuations of the firm may reflect this relationship. Previous studies havedocumented that investors incorporate pension plan assets and liabilities in variousmeasures of firm performance, such as the firms’ market valuation (Franzoni andMarin, 2006), equity beta (Jin et al., 2006), and debt rating (Carroll and Niehaus,1998), among others. In this subsection, I consider whether investors considerthe role of defined benefit claimants, as measured by cumulative abnormal returns(CARs) earned by pension sponsors around Chapter 11 announcements.In my sample of 244 bankrupt defined benefit sponsors, there are 89 bankrupt-cies with available stock market return information for the filing firm around theChapter 11 bankruptcy announcement date. For these firms, I estimate the medianratio of defined benefit obligations to overall liabilities and separate firms into twogroups based on whether they fall above the median (45 firms) or below the median(44 firms)26. Figure 2.2 plots the two groups’ cumulative abnormal returns calcu-lated with the CRSP value-weighted return as the benchmark in the [-5,+5] windowwith day 0 as the date of the Chapter 11 filing. The stock market reacts negatively25As Begley et al. (2015) discuss, the PBO and the ABO are close to identical in frozen definedbenefit pension plans. If firms decide to no longer sponsor a defined benefit plan, they can freezetheir plan. As a result, future pension payouts will be estimated based on current salary levels andnot on future salaries, effectively equalizing PBO and ABO. Therefore, for firms in my sample withfrozen pension plans, my measure of defined benefit obligations, PBO, coincides with the alternativemeasure of pension obligations, ABO.26In the sample of 244 bankrupt defined benefit sponsors, the median ratio of defined benefitobligations to adjusted liabilities equals 0.087. This amount falls between the median ratio of definedbenefit obligations to adjusted liabilities of 0.096 in the sample of 481 pension sponsors in the mainsample and that of 0.077 in the sample of all pension sponsors in Compustat.38to bankruptcy filings in general which is evident from the overall negative returnsin the period. However, the returns for firms with higher ratios of defined benefitobligations to adjusted liabilities flatten out prior to the filing date, whereas the re-turns for firms with smaller ratios of defined benefit liabilities to overall liabilitiescontinue to fall throughout the days leading up to the bankruptcy filings. Thesestock price reactions suggest that investors may expect that firms in which definedbenefit claimants are stronger may be less likely to file for bankruptcy. However,given the small number of firms presented in Figure 2.2, these stock price reactionsprovide anecdotal support for the main findings in this paper at best.2.7 ConclusionIn this paper, I study the impact of defined benefit claimants on the firm’s de-cision to file for Chapter 11 bankruptcy. Using the proportion of defined benefitobligations to overall liabilities as a proxy for the impact of pension claimants oncorporate affairs, I find that defined benefit claimants are associated with a lowerlikelihood of Chapter 11 bankruptcy. These findings are robust to firm-level con-trols and to different specifications. Moreover, the results do not depend on the spe-cific proxy for the role of defined benefit claimants and are consistent in alternativesubsamples. Overall, the results presented in this essay indicate a role for definedbenefit claimants in one important firm decision: the choice to file for bankruptcyreorganization. This essay’s findings point to the importance of accounting forfirms’ defined benefit claimants in studies of creditor negotiations prior to Chapter11 bankruptcy.More broadly, my analysis raises a question about the composition of firms’creditors and the role of non-traditional lenders, such as defined benefit claimants,in times of financial distress. Defined benefit obligations represent a firm liability.However, defined benefit claimants differ from traditional creditors along manydimensions, and in their lack of diversification and the higher costs of bankruptcythat they experience, in particular. Thus, defined benefit claimants differ fromtraditional lenders in their willingness to avoid court reorganization. The collectiveevidence from previous studies and this paper suggests that the overall effect of39defined benefit claimants among the firms’ creditors is largely positive.40Figures and TablesFigure 2.1: U.S. Corporate Debt from 1987 to 2012This figure shows the amount of corporate debt for U.S. companies from 1987 to 2012. Theblue bars report the total dollar amount of corporate financial debt, defined as short-term lia-bilities plus long-term debt, from firms’ financial statements on Compustat. The dark blue linereports the total dollar amount of corporate defined benefit pension obligations, measured asthe pension benefit obligation (PBO), from the footnotes to financial statements on Compus-tat. All dollar amounts are adjusted for inflation and are presented in trillions of 2012 dollars. 41Figure 2.2: Event Study Around Chapter 11 FilingThis figure shows the cumulative abnormal returns (CARs) adjusted by the CRSP value-weightedreturn from five days before to five days after a Chapter 11 filing by defined benefit pension sponsors.The solid line depicts CARs for 45 firms whose ratio of defined benefit obligations to adjusted lia-bilities (PBO/AL) is higher than the median ratio of pension obligations to adjusted liabilities in thesample. The dashed line represents CARs for 44 firms whose ratio of defined benefit obligations toadjusted liabilities is lower than the sample median ratio of pension obligations to adjusted liabilities. -25-20-15-10- 505- 5 - 4 - 3 - 2 - 1 0 1 2 3 4 5Cumulative abnormal return Days relative to Chapter 11 filing  PB O/AL below median PB O/AL above median42Table 2.1: Variable DefinitionsThis table provides definitions for the variables used in this study. Compustat data items are includedin brackets. Note that due to the Compustat naming convention, the same variable name, pcppao, refersto different variables over time. Prior to 1998, pcppao refers to the pension cost for overfunded plans.Between 1998 and 2006, a positive value of pcppao refers to the prepaid pension cost, while a negativevalue of pcppa refers to the accrued pension cost. After 2006, pcppao refers to plan funding status.Variable DefinitionAdditional Leverage [Adjusted Liabilities - short-term debt (dlc) - long-term debt (dltt)]/ AdjustedAssetsAdjusted Assets If 1987 <= fiscal year <= 1997 and pcppao > 0, use Total assets (at) +DB assets + Operating lease - Prepaid pension cost (pcppao) - Underfundedprepaid pension cost (pcppau)If 1998 <= fiscal year <= 2006 and pcppao > 0, use Total assets (at) + DBassets + Operating lease - Prepaid pension cost (pcppao)If 2007 <= fiscal year and funded status > 0, use Total assets (at) + DBassets + Operating lease - Funded status (pcppao)Adjusted Liabilities If 1987 <= fiscal year <= 1997 and pcppao < 0, use Total liabilities (lt)+ DB liabilities + Operating lease - abs(Accrued pension cost (pcppao)) -abs(Underfunded accrued pension cost (pcppau))If 1998 <= fiscal year <= 2006 and pcppao < 0, use Total liabilities (lt) +DB liabilities + Operating lease - abs(Accrued pension cost (pcppao))If 2007 <= fiscal year and funded status < 0, use Total liabilities (lt) + DBliabilities + Operating lease - abs(Funded status (pcppao))Asset Tangibility Net property, plant, and equipment (ppent) / Adjusted AssetsBankrupt Dummy = 1 if a firm files for Chapter 11 bankruptcy in a given year, and 0otherwiseBV Equity Stockholders’ equity (seq or ceq + pstk or at - lt) + Taxes and investment taxcredit (txdb+itcb) - Book value of preferred stock (pstkrv or pstkl or pstk)Cash Holdings Cash and short-term investments (che) / Adjusted AssetsCF Volatility Standard deviation of quarterly operating income (oibdpq) over previous 12quarters scaled by adjusted assetsDB Assets (PPA) If 1987 <= fiscal year <= 1997, Pension plan assets (pplao) + Underfundedpension plan assets (pplau)If fiscal year >= 1998, Pension plan assets (pplao)DB Liabilities (PBO) If 1987 <= fiscal year <= 1997, Pension benefit projected obligation(pbpro) + Underfunded pension benefit projected obligation (pbpru)If fiscal year >= 1998, Pension benefit projected obligation (pbpro)DB Pension Sponsor Dummy = 1 if a firm sponsors a DB pension plan, and 0 otherwiseDividend Payer Dummy = 1 if common stock dividends (dvc) are positive, and 0 otherwiseLeverage (Short-term liabilities (dlc) + long-term debt (dltt)) / Adjusted AssetsMarket-to-Book [Adjusted Assets - BV equity + MV equity (prcc f*csho)] / Adjusted AssetsNo S&P Credit Rating Dummy = 1 if a firm does not have a credit rating, and 0 otherwiseOperating Leases Rental expense (xrent) × 10, as in Rampini and Viswanathan (2013)Return on Assets Operating income before depreciation (oibdp) / Adjusted AssetsR&D Research and development expenses (xrd) / Adjusted AssetsS&P Credit Rating A numeric variable between 0 and 1 indicating a firm’s credit rating, with0.042 corresponding to a D rating, 0.929 corresponding to a AAA rating,and each rating increment increasing the firm’s rating with 0.042, as in Rauh(2009)43Table 2.2: Descriptive StatisticsThis table reports descriptive statistics for the sample of 481 Compustat definedbenefit pension sponsors over the period from 1987 to 2012 which meet the originalBRD sampling criteria: (i) have assets worth $100 million or more, measured in1980 dollars, and (ii) file Form 10-K with the Securities and Exchange Commission(SEC) not less than three years prior to the bankruptcy case, . The 244 bankruptdefined benefit sponsors in the sample are matched to 237 non-bankrupt definedbenefit sponsors on industry, year, and size.Panel A: Industry DistributionThis panel presents an industry break-down for the sample firms, including theindustry name, the number of firms in that industry, and the percent of firms in theindustry relative to the 481 sample firms. Industry affiliation is determined usingthe Fama and French 48-industry classification. Industries are ranked from highestrepresentation in the sample to lowest.Industry N % Industry N %Retail 73 15% Construction 10 2%Transportation 38 8% Fabricated Products 10 2%Steel Works 34 7% Electrical Equipment 10 2%Autos/Trucks 30 6% Petroleum/Gas 6 1%Textiles 27 6% Computers 6 1%Consumer Goods 22 5% Electronic Equip 6 1%Apparel 22 5% Coal 4 1%Chemicals 22 5% Personal Services 4 1%Machinery 22 5% Other 4 1%Rubber/Plastic Prdcts 18 4% Recreation 2 0%Business Services 18 4% Healthcare 2 0%Business Supplies 18 4% Non-Metallic Mining 2 0%Printing and Publishing 14 3% Measuring/Control Equip 2 0%Construction Mtrls 14 3% Shipping Containers 2 0%Wholesale 14 3%Food Products 12 3%Entertainment 12 3% All 481 100%44Panel B: Summary StatisticsThis panel presents summary statistics for the main sample of 481 Compustat defined benefit pension sponsors from 1987 to 2012. The first fourcolumns under the heading All DB Sponsors report summary statistics for all 481 sponsors in the sample. The next four columns under the headingBankrupt DB Sponsors report summary statistics for the set of 244 bankrupt defined benefit sponsors. The last four columns under the headingMatched Non-Bankrupt DB Sponsors report summary statistics for the sample of 237 non-bankrupt defined benefit sponsors. All variables are definedin Table 2.1. All dollar values are expressed in constant 1987 dollars. All continuous variables are winsorized at 0.5% in each tail to reduce theimpact of outliers. Test statistics of the t-test and the Wilcoxon-test of the differences in firm and pension plan characteristics between bankrupt andnon-bankrupt defined benefit sponsors are given in superscript and denote statistical significance of the difference at the 1%(***), 5%(**), and 10%(*)levels, respectively.All DB Sponsors Bankrupt DB Sponsors Matched Non-Bankrupt DB SponsorsN Mean Median Stdev N Mean Median Stdev N Mean Median StdevAssets 481 1,914 510.8 5,759 244 1,757 492.8 5,584 237 2,075 533.8 5,941Adjusted Assets 481 2,235 584.0 6,891 244 2,144 545.9 6,884 237 2,330 593.0 6,912Ln(Adjusted Assets) 481 6.868 6.596 1.180 244 6.844 6.573 1.179 237 6.892 6.626 1.184Asset Tangibility 481 0.356 0.331 0.197 244 0.353 0.337 0.195 237 0.359 0.326 0.199Return on Assets 481 0.086 0.089 0.079 244 0.046 0.051 0.068 237 0.127*** 0.126*** 0.069Cash Holdings 481 0.065 0.033 0.078 244 0.058 0.032 0.066 237 0.072** 0.035 0.089Cash Flow Volatility 481 0.016 0.012 0.013 244 0.019 0.015 0.014 237 0.013*** 0.010*** 0.011Market-to-Book 481 1.378 1.157 0.746 244 1.273 1.104 0.598 237 1.485** 1.202** 0.860R&D 481 0.009 0 0.022 244 0.008 0 0.022 237 0.010 0 0.021Dividend Payer 481 0.370 0 0.483 244 0.164 0 0.371 237 0.582*** 1 0.494S&P Credit Rating 481 0.188 0 0.235 244 0.129 0 0.171 237 0.249*** 0*** 0.273No S&P Credit Rating 481 0.509 1 0.500 244 0.504 1 0.501 237 0.515 1 0.501Oper. Leases / Adj. Assets 452 0.298 0.171 0.337 228 0.337 0.195 0.358 224 0.258*** 0.141*** 0.310DB Pension Sponsor 481 1 1 0 244 1 1 0 237 1 1 0DB Assets / Adj. Assets 481 0.155 0.090 0.193 244 0.178 0.096 0.211 237 0.132** 0.079** 0.170DB Liab. / Adj. Assets 481 0.181 0.098 0.219 244 0.219 0.121 0.252 237 0.142*** 0.082** 0.172Funding status 463 -0.112 -0.126 0.354 237 -0.155 -0.161 0.288 226 -0.067*** -0.105*** 0.408Funding dummy 481 0.270 0 0.445 244 0.234 0 0.424 237 0.308* 0* 0.463DB Liab. / Adj. Liab 481 0.138 0.096 0.134 244 0.133 0.087 0.133 237 0.143 0.100 0.136Leverage 481 0.359 0.347 0.240 244 0.418 0.418 0.272 237 0.298*** 0.291*** 0.185Additional Leverage 481 0.894 0.704 0.568 244 1.107 1.014 0.623 237 0.675*** 0.590*** 0.403Bankrupt 481 0.507 1 0.500 244 1 1 0 237 0 0 045Panel C:: Pairwise CorrelationThis panel reports pairwise correlation coefficients among the main variables of interest for the sample of 481 Compustat defined benefit pension sponsors from1987 to 2012. All variables are defined in Table 2.1. p-values are reported in brackets.1 2 3 4 5 6 7 8 9 10 11 12 13 141 Bankrupt 1.002 DB Liabilities / Adj. Liabilities -0.04 1.00(0.41)3 Leverage 0.25 -0.23 1.00(0.00) (0.00)4 Additional Leverage 0.38 0.25 -0.25 1.00(0.00) (0.00) (0.00)5 Ln (Adj.Assets) -0.02 0.08 -0.08 0.21 1.00(0.66) (0.09) (0.08) (0.00)6 Cash Holdings -0.09 0.07 -0.33 0.18 0.04 1.00(0.04) (0.13) (0.00) (0.00) (0.38)7 Cash Flow Volatility 0.22 -0.10 0.04 0.15 -0.21 0.10 1.00(0.00) (0.02) (0.41) (0.00) (0.00) (0.03)8 Asset Tangibility -0.02 -0.04 0.07 0.00 0.17 -0.22 -0.15 1.00(0.71) (0.36) (0.10) (0.98) (0.00) (0.00) (0.00)9 Return on Assets -0.51 -0.11 -0.02 -0.30 0.02 -0.06 -0.22 0.03 1.00(0.00) (0.01) (0.69) (0.00) (0.66) (0.21) (0.00) (0.46)10 Market-to-Book -0.14 -0.02 -0.06 0.18 0.03 0.23 0.08 -0.14 0.39 1.00(0.00) (0.61) (0.17) (0.00) (0.51) (0.00) (0.09) (0.00) (0.00)11 Dividend Payer -0.43 0.05 -0.19 -0.29 0.15 -0.04 -0.20 0.02 0.31 0.01 1.00(0.00) (0.32) (0.00) (0.00) (0.00) (0.40) (0.00) (0.74) (0.00) (0.80)12 R&D -0.05 0.08 -0.00 -0.07 -0.02 0.15 0.07 -0.18 -0.02 0.13 -0.05 1.00(0.32) (0.07) (0.93) (0.12) (0.65) (0.00) (0.15) (0.00) (0.70) (0.01) (0.26)13 S&P Credit Rating -0.26 -0.04 0.17 -0.20 0.42 -0.11 -0.17 0.13 0.24 0.05 0.23 -0.01 1.00(0.00) (0.39) (0.00) (0.00) (0.00) (0.02) (0.00) (0.47) (0.00) (0.29) (0.00) (0.90)14 No S&P Credit Rating -0.01 0.05 -0.24 0.01 -0.42 0.07 0.14 -0.10 -0.07 -0.02 -0.04 0.01 -0.82 1.00(0.82) (0.23) (0.00) (0.75) (0.00) (0.11) (0.00) (0.03) (0.13) (0.74) (0.38) (0.78) (0.00)46Table 2.3: Determinants of Chapter 11 LikelihoodThis table reports regression results on the role of defined benefit claimants onthe likelihood of Chapter 11 filing for 481 defined benefit sponsors from 1987to 2012. Bankrupt defined benefit sponsors are matched to non-bankrupt definedbenefit firms in the same year, industry and closest in size. Columns (1) and (2)report the marginal effects from a logistic regression whereas columns (3) and (4)present coefficients from a linear probability model. All variables are defined inTable 2.1. t-statistics are reported in parentheses and are statistically significantat the 1%(***), 5%(**), and 10%(*) levels. Standard errors are robust and areclustered at the year level.Dependent Variable: Bankruptit (1) (2) (3) (4)DB Liabilities / Adj. Liabilities -0.31** -0.23** -0.30** -0.25**(-2.39) (-2.46) (-2.37) (-2.22)Leverage 0.76*** 0.44*** 0.74*** 0.48***(8.14) (5.60) (7.76) (5.24)Additional Leverage 0.47*** 0.22*** 0.43*** 0.23***(13.21) (5.63) (11.27) (5.57)Cash Holdings -0.71** -0.69***(-2.55) (-2.89)Cash Flow Volatility 1.27 2.01(0.86) (1.44)Asset Tangibility -0.10* -0.17**(-1.75) (-2.45)Return on Assets -1.75*** -1.93***(-6.65) (-7.79)Market-to-Book -0.01 -0.02(-0.38) (-0.86)Dividend Payer -0.08** -0.13***(-2.42) (-3.67)R&D -0.55 -1.04(-0.64) (-1.14)S&P Credit Rating -0.99*** -0.86***(-7.13) (-9.34)No S&P Credit Rating -0.33*** -0.32***(-6.86) (-5.80)Intercept -0.10* 0.83***(-1.87) (7.99)Model Logit Logit LPM LPMN 481 481 481 481R-squared 0.25 0.50 0.28 0.5147Table 2.4: Descriptive Statistics for Sample of All Compustat FirmsThis panel reports descriptive statistics for the sample of all Compustat firms over the period from 1987 to 2012 which meet the originalBRD sampling criteria: (i) have assets worth $100 million or more, measured in 1980 dollars, and (ii) file Form 10-K with the Securitiesand Exchange Commission (SEC) not less than three years prior to the bankruptcy case, .Panel A: Industry DistributionThis panel presents an industry break-down for all Compustat firms from 1987 to 2012. The first three columns under the heading Non-Defined Benefit Firms report the industry affiliation for all Compustat firms without defined benefit plans. The next three columns under theheading Defined Benefit Sponsors report the industry affiliation of all Compustat firms with defined benefit plans. The last three columnsunder the heading Bankruptcies present the industry affiliation for all Chapter 11 bankruptcies among Compustat firms. Industry affiliationis determined using Fama and French’s 48-industry classification. Industries are ranked from highest to lowest in terms of the number offirms that contribute to each sample.Non-Defined Benefit Firms Defined Benefit Sponsors BankruptciesIndustry N % Industry N % Industry N %Business Services 3369 16% Machinery 1405 7% Retail 66 14%Retail 2421 12% Retail 1396 7% Transportation 34 7%Petroleum/Gas 1484 7% Chemicals 1290 6% Business Services 32 7%Electronic Equip 1465 7% Business Services 1056 5% Wholesale 30 7%Computers 1246 6% Petroleum/Gas 1056 5% Entertainment 21 5%Wholesale 1193 6% Wholesale 1052 5% Steel Works 20 4%Transportation 944 5% Steel Works 943 5% Construction 19 4%Pharmaceutical Prdcts 832 4% Transportation 926 5% Petroleum/Gas 18 4%Healthcare 825 4% Electronic Equip 869 4% Autos/Trucks 17 4%Restaurants/Hotels 766 4% Business Supplies 842 4% Textiles 15 3%Entertainment 758 4% Construction Mtrls 819 4% Construction Mtrls 14 3%Construction 624 3% Food Products 776 4% Healthcare 13 3%Personal Services 462 2% Autos/Trucks 720 4% Chemicals 13 3%Medical Equipment 409 2% Consumer Goods 649 3% Personal Services 13 3%Apparel 378 2% Printing and Publishing 527 3% Computers 13 3%Machinery 339 2% Electrical Equipment 497 2% Consumer Goods 12 3%Other 327 2% Pharmaceutical Prdcts 461 2% Apparel 12 3%Measuring/Control Equip 305 1% Medical Equipment 432 2% Business Supplies 12 3%Consumer Goods 287 1% Computers 421 2% Other 12 3%omittedAll 20679 100% All 20546 100% All 461 100%48Panel B: Summary StatisticsThis panel reports summary statistics for the sample of all Compustat firms from 1987 to 2012. The first four columns under the heading All CompustatFirms report summary statistics for the full sample of Compustat firms. The next four columns under the heading Defined Benefit Sponsors reportsummary statistics for all Compustat defined benefit sponsors. The last four columns under the heading Non-Defined Benefit Firms report summarystatistics for all Compustat firms without defined benefit plans. All variables are defined in Table 2.1. All dollar values are expressed in constant 1987dollars. All continuous variables are winsorized at 0.5% in each tail to reduce the impact of outliers. Test statistics of the t-test and the Wilcoxon-testof the differences in firm and pension plan characteristics between defined benefit sponsors and firms without a pension plan are given in superscriptand denote statistical significance of the difference at the 1%(***), 5%(**), and 10%(*) levels, respectively.All Compustat Firms Defined Benefit Sponsors Non-Defined Benefit FirmsN Mean Median Stdev N Mean Median Stdev N Mean Median StdevAssets 41225 2,087 568.6 4,614 20546 3,133*** 962.4*** 5,778 20679 1,049 388.6 2,665Adjusted Assets 41225 2,332 600.3 5256 20546 3,615*** 1074*** 6666 20679 1,057 388.6 2769Asset Tangibility 41225 0.320 0.261 0.237 20546 0.332*** 0.292*** 0.205 20679 0.308 0.216 0.264Return on Assets 41225 0.131 0.130 0.092 20546 0.138*** 0.133*** 0.073 20679 0.124 0.125 0.107Cash Holdings 41225 0.124 0.061 0.154 20546 0.081*** 0.046*** 0.096 20679 0.166 0.089 0.186Cash Flow Volatility 41225 0.016 0.010 0.019 20546 0.013*** 0.009*** 0.012 20679 0.019 0.012 0.023Market-to-Book 41225 2.078 1.548 1.736 20546 1.813*** 1.457*** 1.192 20679 2.342 1.675 2.112R&D 41225 0.024 0 0.046 20546 0.017*** 0.001*** 0.031 20679 0.031 0 0.056Dividend Payer 41225 0.464 0 0.499 20546 0.647*** 1*** 0.478 20679 0.283 0 0.451S&P Credit Rating 41225 0.241 0 0.281 20546 0.333*** 0.422*** 0.299 20679 0.150 0 0.228No S&P Credit Rating 41225 0.540 1 0.498 20546 0.406*** 0*** 0.491 20679 0.673 1 0.469Operating Leases 37720 0.275 0.140 0.435 18675 0.226*** 0.131*** 0.307 19045 0.324 0.155 0.526DB Pension Sponsor 41225 0.498 0 0.500 20546 1 1 0 20679 0 0 0DB Assets / Assets 20604 0.147 0.090 0.172 20103 0.150 0.093 0.172 0 0 0 0DB Liab. / Assets 20604 0.157 0.099 0.177 20218 0.160 0.103 0.177 0 0 0 0DB Liab. / Adj. Liab 41225 0.077 0 0.120 20546 0.154 0.119 0.131 20679 0 0 0Leverage 41225 0.273 0.252 0.212 20546 0.288*** 0.266*** 0.187 20679 0.259 0.227 0.234Additional Leverage 41225 0.605 0.497 0.438 20546 0.677*** 0.582*** 0.399 20679 0.534 0.390 0.464Bankrupt 41225 0.011 0 0.105 20546 0.007*** 0*** 0.084 20679 0.015 0 0.12349Table 2.5: Determinants of Chapter 11 Likelihood for All Compustat Firms: Pension SponsorsThis table reports regression results for the role of defined benefit claimants on the likelihood of Chapter 11 filing in the sample of all Compustatfirms from 1987 to 2012. Columns (1) through (3) report marginal effects from a logistic regression, columns (4) through (6) report coefficientsfrom a linear probability model, and columns (7) through (9) report coefficients from a conditional logistic regression. All variables are definedin Table 2.1. t-statistics are reported in parentheses and are statistically significant at the 1%(***), 5%(**), and 10%(*) levels. Standard errorsare robust and are clustered at the year level.Dependent Variable: Bankruptit (1) (2) (3) (4) (5) (6) (7) (8) (9)DB Pension Sponsor -0.01*** -0.01*** -0.01*** -0.01*** -0.02*** -0.01*** -0.72*** -0.87*** -0.56***(-8.50) (-9.92) (-5.45) (-5.25) (-6.56) (-5.23) (-8.28) (-10.23) (-5.07)Leverage 0.04*** 0.02*** 0.05*** 0.02*** 3.14*** 2.01***(19.82) (10.13) (7.43) (4.84) (22.64) (10.33)Additional Leverage 0.02*** 0.01*** 0.03*** 0.03*** 1.23*** 1.13***(17.33) (10.95) (8.38) (8.33) (14.52) (8.91)Size -0.00 0.00** -0.10(-0.30) (2.14) (-1.37)Cash Holdings -0.04*** -0.03*** -4.66***(-5.27) (-4.69) (-6.40)Cash Flow Volatility -0.02 0.07 -4.43*(-0.67) (1.62) (-1.76)Asset Tangibility 0.01** 0.02*** -0.05(2.07) (3.35) (-0.20)Return on Assets -0.08*** -0.12*** -8.29***(-10.27) (-7.16) (-10.00)Market-to-Book -0.00*** 0.00** -0.37***(-2.79) (2.73) (-2.89)Dividend Payer -0.01*** -0.01*** -1.00***(-6.33) (-3.59) (-5.19)R&D -0.03 -0.05 -3.41(-1.09) (-1.66) (-1.33)S&P Credit Rating -0.11*** -0.09*** -10.83***(-15.29) (-4.21) (-15.44)No S&P Credit Rating -0.04*** -0.04*** -3.64***(-18.58) (-3.89) (-16.81)Intercept 0.01 -0.02** 0.03**(1.11) (-2.25) (2.19)Model Logit Logit Logit LPM LPM LPM CLogit CLogit CLogitN 38,159 38,159 38,159 41,225 41,225 41,225 41,225 41,225 41,225Year FE Yes Yes Yes Yes Yes Yes Yes Yes YesIndustry FE Yes Yes Yes Yes Yes Yes No No NoR-squared 0.09 0.18 0.36 0.01 0.03 0.05 0.01 0.12 0.3250Table 2.6: Determinants of Chapter 11 Likelihood for All Compustat Firms: Pen-sion LiabilitiesThis table reports regression results for the role of defined benefit claimants on the likelihood of Chapter11 filing for the sample of all Compustat firms from 1987 to 2012. Columns (1) and (2) report marginaleffects from a logistic regression, columns (3) and (4) report coefficients from a linear probability model,and columns (5) and (6) report coefficients from a conditional logistic regression. All variables are definedin Table 2.1. t-statistics are reported in parentheses and are statistically significant at the 1%(***), 5%(**),and 10%(*) levels. Standard errors are robust and are clustered at the year level.Dependent Variable: Bankruptit (1) (2) (3) (4) (5) (6)DB Pension Sponsor -0.01*** -0.01*** -0.01*** -0.01*** -0.67*** -0.32**(-6.10) (-3.35) (-5.63) (-4.48) (-5.55) (-2.28)(DB Liab/Adj. Liab) * DB Sponsor -0.04*** -0.04*** -0.04*** -0.03*** -1.41* -1.89**(-4.13) (-4.01) (-5.55) (-4.44) (-1.92) (-2.14)Leverage 0.03*** 0.02*** 0.05*** 0.02*** 3.09*** 1.94***(18.75) (9.14) (7.27) (4.54) (22.74) (9.65)Additional Leverage 0.02*** 0.01*** 0.03*** 0.03*** 1.24*** 1.15***(16.86) (11.30) (8.56) (8.51) (14.64) (9.16)Size -0.00 0.00** -0.10(-0.01) (2.18) (-1.37)Cash Holdings -0.04*** -0.03*** -4.79***(-5.39) (-4.76) (-6.43)Cash Flow Volatility -0.02 0.07 -4.80*(-0.78) (1.58) (-1.89)Asset Tangibility 0.01** 0.02*** -0.03(2.13) (3.48) (-0.13)Return on Assets -0.08*** -0.12*** -8.42***(-10.27) (-7.20) (-9.78)Market-to-Book -0.00*** 0.00** -0.37***(-2.87) (2.75) (-2.90)Dividend Payer -0.01*** -0.01*** -0.96***(-6.04) (-3.26) (-4.91)R&D -0.03 -0.05* -3.16(-0.97) (-1.71) (-1.26)S&P Credit Rating -0.11*** -0.09*** -10.92***(-14.73) (-4.17) (-15.09)No S&P Credit Rating -0.04*** -0.04*** -3.69***(-17.46) (-3.84) (-16.51)Intercept -0.02** 0.02*(-2.43) (2.04)Model Logit Logit LPM LPM CLogit CLogitN 38,159 38,159 41,225 41,225 41,225 41,225Year FE Yes Yes Yes Yes Yes YesIndustry FE Yes Yes Yes Yes No NoR-squared 0.19 0.37 0.03 0.05 0.12 0.3251Table 2.7: Determinants of Chapter 11 Likelihood for Underfunded PlansThis table reports regression results for the role of defined benefit claimants onthe likelihood of Chapter 11 filing for the 351 defined benefit sponsors with un-derfunded pension plans from 1987 to 2012. In an underfunded defined bene-fit plan, pension liabilities exceed pension assets. Columns (1) and (2) reportmarginal effects from a logistic regression whereas columns (3) and (4) reportcoefficients from a linear probability model. All variables are defined in Table2.1. t-statistics are reported in parentheses and are statistically significant at the1%(***), 5%(**), and 10%(*) levels. Standard errors are robust and are clusteredat the year level.Dependent Variable: Bankruptit (1) (2) (3) (4)DB Liabilities / Adj. Liabilities -0.26* -0.25** -0.27* -0.25*(-1.78) (-2.08) (-1.77) (-1.84)Leverage 0.73*** 0.43*** 0.69*** 0.48***(5.95) (4.45) (5.60) (4.56)Additional Leverage 0.47*** 0.24*** 0.42*** 0.22***(11.12) (4.55) (10.25) (4.57)Cash Holdings -0.95*** -0.87***(-2.62) (-3.02)Cash Flow Volatility 1.07 2.07(0.72) (1.49)Asset Tangibility -0.10 -0.20**(-1.26) (-2.38)Return on Assets -1.68*** -1.87***(-5.67) (-6.22)Market-to-Book -0.01 -0.02(-0.31) (-0.78)Dividend Payer -0.09* -0.15***(-1.89) (-3.50)R&D -0.43 -1.31(-0.42) (-1.41)S&P Credit Rating -1.25*** -1.03***(-5.32) (-8.59)No S&P Credit Rating -0.36*** -0.36***(-6.47) (-5.06)Intercept -0.07 0.92***(-0.99) (7.28)Model Logit Logit LPM LPMN 351 351 351 351R-squared 0.23 0.50 0.26 0.5052Table 2.8: Determinants of Chapter 11 Likelihood Under ABOThis table reports regression results for the role of defined benefit claimants on thelikelihood of Chapter 11 filing for all Compustat firms from 1987 to 2012. In thistable, defined benefit liabilities are estimated using an alternative measure of liabili-ties, the accumulated benefit obligation (ABO). Columns (1) and (2) report marginaleffects from a logistic regression, columns (3) and (4) report coefficients from a lin-ear probability model, and columns (5) and (6) report coefficients from a conditionallogistic regression. All variables are defined in Table 2.1. t-statistics are reported inparentheses and are statistically significant at the 1%(***), 5%(**), and 10%(*) levels.Standard errors are robust and are clustered at the year level.Dependent Variable: Bankruptit (1) (2) (3) (4)(ABO Liab/Adj. Liab) * ABO Sponsor -0.02** -0.02** -0.03*** -0.03***(-2.04) (-2.48) (-5.48) (-5.07)ABO Pension Sponsor -0.01*** -0.00 -0.01*** -0.00**(-3.29) (-1.34) (-4.09) (-2.21)Leverage 0.03*** 0.02*** 0.05*** 0.02***(17.56) (7.96) (7.19) (3.82)Additional Leverage (ABO) 0.01*** 0.01*** 0.02*** 0.02***(11.92) (8.09) (8.85) (9.00)Size -0.00 0.00*(-0.84) (1.72)Cash Holdings -0.04*** -0.03***(-5.30) (-4.85)Cash Flow Volatility -0.01 0.08*(-0.32) (1.88)Asset Tangibility 0.01** 0.02***(2.33) (3.56)Return on Assets -0.08*** -0.12***(-11.57) (-7.28)Market-to-Book -0.00*** 0.00***(-2.65) (2.85)Dividend Payer -0.01*** -0.01***(-6.82) (-3.83)R&D -0.03 -0.04(-0.88) (-1.46)S&P Credit Rating -0.11*** -0.09***(-15.36) (-4.21)No S&P Credit Rating -0.04*** -0.04***(-17.53) (-3.95)Intercept -0.02** 0.03**(-2.30) (2.57)Model Logit Logit LPM LPMN 38,171 38,171 41,238 41,238Year FE Yes Yes Yes YesIndustry FE Yes Yes Yes YesR-squared 0.15 0.35 0.03 0.0553Chapter 3Corporate Defined Benefit PensionPlans in Bankruptcy Reorganiza-tion3.1 IntroductionCorporate defined benefit pension plans hold a claim against the firm’s assets inbankruptcy. However, little is known about what role defined benefit claimants playin the bargaining among claimants in bankruptcy and the outcomes of the reorgani-zation process, if any. The main goal of this essay is to determine whether pensionclaimants influence the restructuring process under Chapter 11. In bankruptcy,defined benefit plans typically become members of the unsecured creditors’ com-mittee and can vote on the proposed plan of reorganization27. Therefore, pensionclaimants will play some role in bankruptcy because they are unsecured creditorswith voting power. Rather than capturing this mechanical effect, I am interested inwhether pension claimants influence the reorganization process above and beyondthe traditional lenders’ influence. Whether defined benefit claimants should play arole above that of other creditors is a priori unclear.On the one hand, defined benefit claimants may influence the bankruptcy re-structuring more than other lenders because pension claimants are different fromthe firm’s traditional creditors. In particular, pension claimants are less diversi-27Defined benefit claimants get a vote in bankruptcy proportional to the unfunded portion of theirobligation.54fied and stand to lose more than other lenders in bankruptcy so they have strongincentives to influence the Chapter 11 process beyond what traditional creditorswould do28. On the other hand, defined benefit claimants may not influence thebankruptcy restructuring more than other firm lenders if they have already givenup a lot in private negotiations prior to bankruptcy, if their claim is not sizable rel-ative to that of the other unsecured creditors, or if it is in their best interest not todo so.To understand whether defined benefit claimants influence the bankruptcy pro-cess beyond the impact of traditional creditors, I use a comprehensive sample of236 bankrupt defined benefit pension sponsors from 1987 to 2012. I control forthe role of traditional lenders in bankruptcy by accounting for the firm’s overallindebtedness and I capture any additional effect that pension claimants may haverelative to all other claimants with the ratio of defined benefit liabilities to over-all liabilities. By controlling for firms’ overall indebtedness and focusing on thecomposition of firm lenders, I am able to investigate the incremental impact thatdefined benefit claimants have beyond other lenders on the reorganization processand its outcomes.I present novel evidence on the role of defined benefit claimants in bankruptcy.First, pension claimants do not influence the time firms spend reorganizing or thelikelihood that firms emerge from bankruptcy more than other creditors. How-ever, defined benefit claimants impact the decision to terminate a pension plan inbankruptcy. In particular, firms are more likely to terminate a pension plan inbankruptcy whenever defined benefit obligations represent a higher proportion ofthe firm’s overall liabilities. This effect is significant and robust to different specifi-cations. The strong association between defined benefit obligations and distressedplan terminations implies a role for defined benefit claimants above and beyond theinfluence of the firm’s other lenders.Next, I examine the extent to which defined benefit claimants and the actionsthey take in bankruptcy predict the post-reorganization firm survival as measured28For example, consider a hedge fund, such as Bridgewater Associates or BlackRock, whoseportfolio is well diversified. To these creditors, one company’s bankruptcy will likely not have amaterial impact on the fund’s financial position. In contrast, defined benefit claimants stand to losea sizable portion of their wealth upon their employer’s bankruptcy because their jobs, salaries, andpensions are dependent on the company’s performance.55by the likelihood that firms refile for Chapter 11. I document that the more definedbenefit obligations contribute to the firm’s overall liabilities, the less likely the firmis to refile for bankruptcy. These results provide further support for the findings inthe previous chapter that firms with more defined benefit claimants relative to otherlenders are less likely to file for Chapter 11.If defined benefit claimants play a role beyond that of traditional lenders, theymay influence unsecured creditors’ recovery rates because pension claimants them-selves are typically unsecured creditors in bankruptcy. In a subsample of firms withrecovery data, I show that defined benefit claimants impact the recovery rates forunsecured creditors. In particular, a higher ratio of pension liabilities to overallliabilities is associated with higher recovery rates for unsecured creditors. Whilethe effect is marginally significant, it suggests that defined benefit claimants mayplay a role in bankruptcy beyond the influence of traditional lenders.In light of the evidence that defined benefit claimants impact some aspects ofthe restructuring process beyond the influence of traditional lenders, I next considerone potential channel through which pension claimants may impact Chapter 11reorganization: bargaining about their pension benefits. In particular, if definedbenefit claimants influence bankruptcy restructuring through accepting benefit cuts,their role will be evident in explaining extra variation in these benefit cuts beyondwhat is expected in bankruptcy. After controlling for the creditors’ expected lossesin Chapter 11, I show that defined benefit claimants influence the changes in theirliabilities in bankruptcy. This effect is present only for the unfunded portion ofpension liabilities relative to all firm liabilities. Thus, pension claimants agree toconcessions whenever they stand to lose the most.While not the main focus of the essay, the results in this paper provide novelevidence that unions facilitate negotiations between creditors in bankruptcy. Firmswhose employees are represented by unions reorganize faster under Chapter 11bankruptcy. In particular, switching from not having union representation to hav-ing unions is associated with a 26% reduction in the time firms spend in bankruptcy.Moreover, unions are positively related to the likelihood that firms reorganize suc-cessfully under Chapter 11 bankruptcy. These findings indicate a role for unions inaiding the bankruptcy reorganization process.My paper contributes to the literature on the outcomes of Chapter 11 bankruptcy.56Several studies have examined the factors influencing the time spent in bankruptcyand the probability that a firm successfully emerges from Chapter 11. Hotchkiss(1993) shows that firm size is the most important characteristic of emergence.Bryan, Tiras, and Wheatley (2002) document that solvency risk and liquidity riskmatter for the likelihood of bankruptcy emergence. Denis and Rodgers (2007)show that the duration, the outcome, and the post-reorganization process afterChapter 11 are related to firms’ operating and financial characteristics. This es-say contributes to the literature by showing how defined benefit claimants influ-ence the bankruptcy restructuring process. In terms of accounting for alternativefirm lenders, prior studies have considered how the resolution of bankruptcy is im-pacted by debtor-in-possession financing (Dahiya, John, Puri, and Ramirez, 2003),private equity firms (Hotchkiss et al., 2012), and hedge funds (Jiang et al., 2012).My study contributes to these works by showing that defined benefit claimants playa role in the resolution of bankruptcy beyond that of traditional lenders.My paper is also related to the literature on the determinants of defined benefitplan terminations. Most papers in this literature focus on a firm’s decision to ter-minate its overfunded defined benefit plans outside of bankruptcy (Mittelstaedt andRegier, 1993; Stone, 1987; Thomas, 1989)29. Ippolito (1985a) is one of the firstpapers to study the determinants of underfunded plan terminations at the firm level.Rauh (2009) provides a more recent analysis of bankrupt firms’ defined benefit planterminations at the plan level. I extend Rauh (2009)’s findings to show that definedbenefit claimants impact the decision to terminate the pension plan through the sizeof their obligation relative to the firm’s overall indebtedness. Thus, my study doc-uments an important role for defined benefit claimants in Chapter 11 bankruptcy.The rest of the paper is organized as follows. Section 3.2 discusses the role ofdefined benefit claimants in bankruptcy and develops the hypotheses under study.Section 3.3 describes the data and presents summary statistics. Sections 3.5 to 3.9present results for the Chapter 11 outcomes under study. Section 3.10 concludes.29In addition, Petersen (1992) provides a thorough analysis of firms’ financial, tax, and implicitcontract motives to terminate overfunded pension plans.573.2 Defined Benefit Claimants in BankruptcyIn Chapter 11, distressed firms are given a chance to restructure their operationswithout pressure from creditors. Firms have 120 days from filing for bankruptcyto come up with a reorganization plan that outlines the actions needed to be takento emerge from bankruptcy, including how much of their pre-bankruptcy claimdifferent claimants will recover upon emergence. Claimants whose positions areimpaired are allowed to vote on the proposed plan and their voting power is usu-ally proportionate to their claims’ size. While the firm drafts the reorganizationplan, it negotiates with different claimants on the plan’s acceptable terms. Thesenegotiations are needed to ensure the plan is approved by vital claimants30. Theproposed plan can be amended in creditors express concerns and if the firm’s planis not approved, creditors are allowed to put their alternative plan for reorganiza-tion to a vote. Altogether, the Chapter 11 restructuring process involves extensivenegotiations among the firm and its creditors. Different claimants participate differ-ently in the negotiations that ensue in bankruptcy. While secured creditors usuallyrecover their entire claim and equity claimants only receive what is left after allcreditors are paid off, unsecured creditors fall in the middle of active negotiationswith the firm to determine how much of their claim they will recover and whatthe bankruptcy outcomes will be. Defined benefit claimants are typically part ofthe unsecured creditors’ committee and they participate in the bankruptcy negotia-tions31.The main goal of this paper is to determine whether defined benefit claimantsinfluence the reorganization process under Chapter 11 bankruptcy. As members ofthe unsecured creditors’ committee, pension claimants will play a role in bankruptcysimply because they have voting power. Instead of capturing this mechanical effect,I am interested in whether pension claimants influence the reorganization processabove and beyond the traditional lenders’ influence.30An advantage of the bankruptcy process is the debtor’s ability to bind individual creditors toa repayment plan despite their dissent. Under 11 U.S.C §§1129(a) and 1126(c), a plan may beconfirmed as long as a sufficient number of creditors in a given class who hold a minimum amountof claims vote in favor of the plan.31Note that the part of defined benefit obligations which is covered by pension assets is akin to asecured claim and the unfunded portion of pension obligations is an unsecured claim.58There are several reasons why defined benefit claimants may impact the reor-ganization process and its outcomes beyond the influence that traditional lendershave. First, the previous chapter of this thesis documented that defined benefitclaimants are related to the likelihood that firms file for bankruptcy. If pensionclaimants matter for firm decisions prior to Chapter 11, they may also be relevantwhen the firm is in bankruptcy proceedings.Second, defined benefit claimants differ from traditional lenders in bankruptcy32.While all unsecured creditors typically incur losses in bankruptcy, defined ben-efit claimants face higher losses than other unsecured lenders because pensionclaimants are less diversified than traditional creditors. For traditional lenders, thebankrupt firm is one of many investments so the losses these creditors incur willbe at least partially offset by their other investments. Defined benefit claimants’pension wealth is entirely dependent on the bankrupt firm because the firm decideshow to manage the plan and it controls all plan assets until employees retire. More-over, even if firm creditors lose some portion of their investment in bankruptcy,they usually remain a creditor after emergence and have a chance to further re-cover their position. Defined benefit claimants may remain a creditor to the firmafter bankruptcy, but they may also be removed as a creditor if their pension planis terminated in Chapter 11 and their claim is transferred to the PBGC. Even fur-ther, defined benefit claimants are firm employees who stand to lose not only theirpensions but also their jobs and future salaries. Overall, pension claimants are aunique member of the unsecured creditors’ committee who is not as diversified andfaces higher losses than the other unsecured creditors in bankruptcy. As a result,defined benefit claimants in Chapter 11 may influence the reorganization processbeyond the influence of the firm’s other lenders.Last, pension claimants have a larger set of strategic actions they can undertakein bankruptcy to negotiate with the firm. In particular, pension claimants may ac-cept benefit cuts or even benefit freezes for a certain period to keep the firm alive.Pension claimants may agree to freeze their defined benefit plan to new employ-ees, thus reducing the share of employees covered by the plan. At the extreme,32Appendix B at the end of the thesis provides a detailed description of employees rights inbankruptcy. Appendix C provides a more in-depth comparison of defined benefit claimants and thefirm’s other lenders than the one presented in the text.59defined benefit claimants may agree to have their plan terminated to keep the firmas a going concern and to preserve their jobs. In addition, defined benefit claimantsmay seek the help of the PBGC in the bankruptcy negotiations33. The PBGC hasthe same incentives as pension claimants to see the firm cover as much of the pen-sion liability as possible. Moreover, the PBGC has expertise and experience indealing with bankruptcy proceedings and may influence the bargaining betweenthe firm and its creditors. Given these actions that pension claimants may under-take, defined benefit claimants may play a role beyond that of traditional lenders inbankruptcy.While defined benefit claimants may differ from traditional lenders, thesedifferences do not guarantee that pension claimants will impact the bankruptcyprocess beyond the influence that other lenders exert. One reason why pensionclaimants may not have a higher impact than traditional lenders is purely mechan-ical: if their claim is not large enough relative to the firm’s other lenders, pensionclaimants will not have significant power to influence negotiations and the out-comes of bankruptcy.In addition, defined benefit claimants may not have a stronger influence inChapter 11 if they are unable to provide further concessions needed to impact thereorganization. In private negotiations prior to bankruptcy, pension claimants ac-tively bargain with the firm and likely agree to wage and benefit cuts to preventthe firm from defaulting. Graham, Kim, Li, and Qiu (2013) provide evidence thatemployees agree to wage concessions prior to Chapter 11 so it is likely that pen-sion benefits are also reduced prior to the bankruptcy filing. Once in Chapter 11,defined benefit claimants may not have any leeway to make further concessionsbeyond the cuts they accept prior to bankruptcy34.Yet another reason why defined benefit claimants may not exert an additionalinfluence on the bankruptcy process could be because it is in their best interest toside with the rest of the unsecured lenders. If the firm does not fund the pension33As previously noted, the PBGC stands for the Pension Benefit Guarantee Corporation, a gov-ernment entity created under ERISA to protect corporate defined benefit pension plans.34Defined benefit claimants may agree to concessions prior to Chapter 11 if they believe that thesecuts will allow the firm to avoid bankruptcy. For example, prior to United Air Lines’ bankruptcy in2002, pension claimants agreed to $5 million in wage and benefit cuts over 5 years in hopes thatthese cuts would keep their employer solvent.60plan and invests its assets recklessly, defined benefit claimants will not take extraactions to aid the reorganization process beyond what the firm’s other lenders aredoing. Therefore, defined benefit claimants may choose not to play a role beyondthat of traditional lenders in bankruptcy.In light of these conflicting views, whether defined benefit claimants influencethe reorganization process above and beyond the traditional lenders’ impact re-mains an empirical question. To address this question, I control for the role oftraditional creditors in bankruptcy and capture any additional effect that definedbenefit claimants may have. The typical measures used in the literature to accountfor creditors’ role in bankruptcy include firm leverage and the number of classeswith a claim to the firm’s assets. However, all liabilities are important in bankruptcyrather than just the traditional short-term liabilities and long-term debt which com-prise the standard leverage measure. Since I am interested in the role of definedbenefit claimants relative to that of all firm creditors, I control for the firm’s over-all indebtedness by accounting for short-term and long-term debt as well as for allother balance sheet liabilities and two of the largest off-balance sheet obligations:pension liabilities and operating leases.To capture the role of pension claimants in bankruptcy, I use the ratio of definedbenefit obligations to overall liabilities. This ratio proxies for pension claimants’influence on the bankruptcy reorganization process relative to that of the firm’sother creditors. I account for the impact of the firm’s traditional lenders by control-ling for overall indebtedness and I measure how the composition of firm lenders,and pension claimants in particular, influences the restructuring process. In thisway, I am able to investigate the incremental impact that defined benefit claimantshave on the reorganization process beyond other lenders.Following the above reasoning, my main tests focus on the following hypothe-sis:Hypothesis H0 : Substituting a dollar of financial liabilities for a dollar of de-fined benefit pension liabilities does not impact the Chapter 11 restructuring pro-cess.61In general, defined benefit claimants care about the total size of the pensionpromise. Once in bankruptcy, plan terminations become an imminent threat andpension claimants may also care about the part of their obligations which is notcovered by assets, i.e. the unfunded portion of their obligation. Since pension planassets in bankruptcy are an estate of the pension trust and cannot be accessed by thefirm’s other lenders, the portion of unfunded liabilities is more closely related tounsecured lenders’ claims in Chapter 11. Therefore, the unfunded pension liabili-ties can be considered an alternative measure of defined benefit claimants’ role inbankruptcy. In particular, the more underfunded the pension liability is, the moredefined benefit claimants stand to lose, and hence, the higher the incentive willbe for pension claimants to influence the restructuring. In light of this reasoning, Ialso test whether substituting a dollar of financial liabilities for a dollar of unfundedpension liabilities impacts the bankruptcy reorganization process.Defined benefit claimants may influence various characteristics and outcomesof the reorganization process. Two common measures of bankruptcy restructuringinclude the length of time firms spend in Chapter 11 and the likelihood that firmssuccessfully emerge from bankruptcy. Along with these measures, I also considerwhether defined benefit claimants are related to the likelihood that firms terminatetheir pension plans in bankruptcy and to the likelihood that firms refile for Chapter11 after having successfully reorganized once. If defined benefit claimants play arole in bankruptcy, their influence may be evident in the amounts that other credi-tors can recover. Therefore, I test for a relationship between unsecured creditors’recovery rates and the ratio of defined benefit obligations to overall liabilities.As previously discussed, defined benefit claimants have a large set of strategicactions they can undertake to influence the bankruptcy process. These actions in-clude negotiations with the firm, agreement to benefit cuts, and agreement to planfreezes, among others. Ideally, I would measure the specific actions that definedbenefit claimants take over the course of the bankruptcy proceedings and com-pare them to the actions of traditional lenders. Then, I will be able to determinethe channels through which pension claimants influence the reorganization processbeyond other lenders’ influence. However, some actions, such as negotiations, arehard to measure. Even further, few firms provide information during bankruptcyproceedings which makes it difficult to pin down the exact channel through which62defined benefit sponsors influence the outcomes of bankruptcy. Nevertheless, ina subsample of firms which provide restructuring information, I study one of thechannels through which defined benefit claimants may influence the Chapter 11bankruptcy process: their willingness to agree to pension benefit cuts.3.3 Sample and DataIn this paper, I study the role of corporate defined benefit pension claimants inbankruptcy. My sample consists of all Chapter 11 filings by defined benefit spon-sors from 1987 to 2012. The sample selection procedure is outlined in Panel A ofTable 3.2. To construct the sample, I begin by identifying all firms that sponsor adefined benefit pension plan from the Compustat Pension Annual file. In the sam-ple period, there are 8,874 pension plan sponsors (85,750 firm-year observations).Next, I identify all sponsors with non-missing data on pension assets and liabili-ties, which reduces the number of firms to 5,695. I merge the firm-year data forthese remaining defined benefit sponsors to the Compustat Fundamentals Annualfile. Requiring that firms have non-missing data for all variables used in this studyleaves 2,005 firms with defined benefit plans in the sample.To identify Chapter 11 filings, I use Lynn M. LoPucki’s Bankruptcy ResearchDatabase (BRD) which is commonly used in bankruptcy studies. The BRD con-tains information on all bankruptcy filings (Chapter 11) and liquidation filings(Chapter 7) from 1980 to 2012 for U.S. firms with assets of $100 million or more,measured in 1980 dollars. Out of the 961 bankruptcy cases in the BRD, there are244 Chapter 11 filings by defined benefit sponsors from 1987 to 2012. In turn, 236out of the bankrupt defined benefit sponsors have data at most two fiscal years priorto the Chapter 11 filing35. Thus, the main sample used in this paper consists of 236defined benefit sponsors who filed for Chapter 11 bankruptcy between 1987 and2012.Panel B of Table 3.2 reports the industry distribution of my sample of bankrupt35The number of Chapter 11 cases drops from 244 to 236 because the sampling criteria in theBRD is for firms to have filed Form 10-K with the SEC at most three years prior to bankruptcy. Eightof the defined benefit sponsors had last filed three years prior to bankruptcy, so they are excludedfrom my sample. However, the results presented in the paper remain unchanged if those companiesare used as well.63defined benefit sponsors. Industry affiliation is based on the Fama and French 48-industry classification and 31 of the 48 industries are represented in my sample.As the table shows, 16% of bankruptcies in my sample come from defined benefitsponsors from the Retail industry. Defined benefit sponsors in the Transportation,Steel Works, and Automobile industries experience the next highest rates of de-fault, with 8%, 7% and 7% of sample firms, respectively. The results in Panel Bindicate that there is a significant dispersion in the sample firms in terms of theirindustry affiliation.As in the previous essay, defined benefit assets and obligations and operatingleases are consolidated with balance sheet assets and liabilities. The same regardto changes in accounting rules over time was paid when off balance sheet amountswere brought back on the balance sheet. In particular, to measure firms’ finan-cial position, I use two new measures of firm assets and liabilities: adjusted assetsand adjusted liabilities. Adjusted assets include the book value of firm assets plusthe values of operating leases and defined benefit assets, less the portion of pensionvalues included on the balance sheet. Adjusted liabilities equal the sum of total bal-ance sheet liabilities plus defined benefit obligations and operating leases, less thepension values already included on the balance sheet. Defined benefit obligationsare scaled by these additional liabilities to capture the role of pension claimantsrelative to all other creditors. Additional leverage equals adjusted liabilities lessshort-term and long-term debt, scaled by adjusted assets and financial leverageequals short-term and long-term debt scaled by adjusted assets. These two lever-age measures are used to control for the firm’s indebtedness. Appendix A at theend of the thesis provides a detailed discussion of the accounting rule changes andhow they influenced the variables used in both essays. Section 2.3.2 in the previouschapter describes how the main control variables are constructed.3.3.1 Additional DatasourcesSome of the variables used in this paper come from sources other than Compus-tat and the BRD. In this subsection, I discuss where information on these variableswas obtained from. All pension and financial variables in this study are measuredat the last fiscal year prior to the Chapter 11 filing. All variables and the databases64from which they are obtained are presented in Table 3.1.Pension VariablesAt the firm level, data on pension plan assets (PPA) and pension liabilities (pen-sion benefit obligation, PBO) come from Compustat. To identify whether firmemployees are represented by a union, I check firms’ Form 5500 filings with theDepartment of Labor (DOL)36 and Form 8-K and 10-K filings with the Securi-ties and Exchange Commission (SEC) available on Edgar37. Using these sources,I identify the unionization status for 95 firms from Form 550038, 57 firms frombankruptcy reorganization plans (8-K filings), and 24 firms from 10-K filings atthe year prior to Chapter 11 bankruptcy. Thus, I am able to classify 176 out of the236 bankrupt sponsors as being unionized or not.Another variable of interest in this study is the decision to terminate a definedbenefit pension plan in bankruptcy. To determine if a plan termination occurred, Ifirst check the PBGC website for mentions of any plan terminations for of the 236sample firms. From PBGC reports, I find that 70 of the firms in my sample termi-nated at least one defined benefit plan in bankruptcy, and 45 emerged from Chapter11 with their plans intact, for a total of 115 events. Next, I identify 74 pension planterminations from Form 550039. Out of those terminations, 42 cases overlap withthe events identified from the PBGC website, so 32 terminations contributed addi-tional events to my sample. Last, I count all firms that liquidated in bankruptcy ashaving terminated their pension plans, resulting in 25 additional terminations. Fol-36Pension plan sponsors with more than 100 employees have to annually file Form 5500 with theInternal Revenue Services (IRS) and the DOL for each of the pension plans that they sponsor. In theform, firms provide information on the pension plan such as the plan’s assets and liabilities and theactuarial assumptions used to evaluate different plan variables, among others.37Publicly traded firms file Form 10-K annually with the SEC at their fiscal year end and Form8-K whenever a material event, such as a bankruptcy filing, a merger, or a change in control, occurs.38Since Form 5500 data is reported at the plan level and I use firm-level data, I sum across firms’plans every year to convert the plan information to firm-level data. As a result, a firm is classified ashaving a union if at least one of its pension plans is covered by a collective bargaining agreement. Inturn, a firm is classified as not having a union if none of its pension plans are represented by a union.39Following the consolidation procedure described in the previous footnote, a firm is classifiedas having terminated a pension plan in bankruptcy if it terminated at least one of its defined benefitplans while in Chapter 11. In turn, a firm is classified as not having terminated a pension plan inbankruptcy if none of its defined benefit plans were terminated in Chapter 11.65lowing these procedures, I am able to identify whether 172 out of the 236 samplefirms terminated at least one of their pension plans in Chapter 11.While most tests in this study are based on pension data at the firm level, onespecification calls for the use of pension information at the plan level. Compus-tat consolidates across firms’ pension plans to provide firm-level measures of thepension variables. Form 5500, however, provides pension information at the planlevel: firms file Form 5500 for each pension plan that they sponsor40. Therefore,for these tests I use pension plan information from Form 5500 filings. Form 5500data covers the period from 1990 to 2007 and firms on it are identified by employerinsurance number (EIN) and firm name. I take several steps to link Form 5500data to Compustat data. First, I use the EIN, name and fiscal year to link the twodatabases. Following this procedure I am able to match 110 Form 5500 firm-yearsto Compustat. For the remaining companies without a valid EIN link, I match Form5500 firms to Compustat firms based on firm name and fiscal year41. In this way,I identify 43 additional firm-year matches. Altogether, I am able to match Form5500 and Compustat data for 153 of the 179 sample firms in the same time pe-riod. Imposing that firms have pension and financial data at most two years prior tobankruptcy reduces the sample to 94 firm-years. In addition, requiring all firms tohave non-missing values for the Form 5500 variables used in the tests brings sam-ple size to 83 firms with all the necessary data from Compustat and Form 5500.Thus, 83 out of the 179 sample firms from 1990 to 2007 are included in the plan-level tests, for a total of 176 plan-level observations. As a comparison between thetwo pension databases, the aggregate defined benefit plan assets for the 83 firms onForm 5500 data equal approximately $27,000 million whereas the pension assetsfor these firms on Compustat equal $36,500 million42.40As Rauh et al. (2012) emphasize, subsidiaries that are more than 80% owned by the parent canreport their pension obligations in separate Form 5500 filings from the parent firm. Therefore, it isnecessary to aggregate subsidiaries’ information with the parent filings to obtain the firm’s pensionassets and liabilities from Form 5500 filings. Form 5500 data used in this paper does not consolidatesubsidiaries’ pension plans with the parent firm.41In the matching procedure, names are cleaned of any symbols, such as hyphens, quotationmarks etc., and abbreviations in firm names are accounted for. After the names are cleaned, onlyidentical name matches are used to reduce matching error.42The sum of all defined benefit plan assets is lower when Form 5500 data is used instead ofCompustat data. Such a difference in plan assets could be due to the fact that Form 5500 does notconsolidate subsidiaries’ plans with the parent’s plan. However, a discrepancy between the variables66Bankruptcy VariablesFrom the BRD, I collect data on the date of the bankruptcy filing, the durationof the reorganization, the state in which Chapter 11 was filed, and whether thebankrupt firm refiled for bankruptcy after emerging. While an older version of theBRD provides data on the outcomes of Chapter 11, the updated database no longerreports bankruptcy outcomes. Therefore, I re-construct the outcome variable usingthe following steps. First, I identify bankruptcy outcomes for 185 of the 236 firmsfrom Form 8-K filings. Then, I compare the outcomes from the 8-K filings tothe outcomes on the BRD. In 150 cases, the outcomes from the two databases areidentical. For the remaining 35 bankruptcies with contradicting outcomes and the51 bankruptcies with no data on Form 8-K, I manually search Factiva, a researchtool created by Dow Jones & Company, to determine the Chapter 11 outcome. Asa result, I am able to identify the outcomes of all bankruptcy cases in my sample.Previous research has shown that it is important to control for firm capital struc-ture complexity in bankruptcy (Gilson, John, and Lang, 1990). Following the lit-erature, I control for this complexity by using the number of classes with claims tothe firm’s assets in bankruptcy. Upon bankruptcy filing or exit, firms disclose thenumber of claim classes in Chapter 11 in Form 8-K filings with the SEC. I readfirms’ 8-K filings from the bankruptcy filing date through the resolution date toidentify the number of claim classes in Chapter 11. Since reporting informationon 8-K filings is not mandatory and as regulated as 10-K filings, I am only able toobtain information on the number of claim classes in bankruptcy for 107 of the 236Chapter 11 cases in my sample.Financial VariablesFinancial data for the control variables used in this paper come from Compustat.Data on unsecured creditors’ recovery rates come from Capital IQ, a relativelyin the two datasources can also be due to the fact that Compustat consolidates international anddomestic plans, whereas firms file Form 5500 only for domestic plans. According to the Form 5500instructions, most pension plans maintained outside the U.S. are exempt from filing Form 5500.Those foreign plans that have to report to the IRS file Form 5500-EZ, which is not a part of thesample forms used in this study.67novel database provided by the Standard and Poor’s. Using the Capital IQ database,I identify recovery rate data for 32 of the 244 firms in the sample. The unsecuredcreditors’ recovery data covers the period from 2006 to 2012.3.3.2 Summary StatisticsIn this subsection, I discuss the summary statistics for the main sample used inthis study, as well as for several subsamples of firms for which results are alsoestimated.Panel A of Table 3.3 provides summary statistics for firm, pension plan, andbankruptcy characteristics for the full sample of 236 bankrupt defined benefit spon-sors. The top part of the panel presents firm characteristics for the sample firms.The average firm in the sample has assets worth $1,701 million and firm size in-creases to $2,139 million when defined benefit assets and operating leases are ac-counted for. Sample firms are 45% levered on average. Additional leverage, mea-sured as total balance sheet liabilities, defined benefit obligations and operatingleases less short-term and long term obligations is 57% of adjusted assets on aver-age. Moreover, on average sample firms hold 4% of their assets in cash, tangibleassets constitute 25% of adjusted assets, and the average return on adjusted assetsis 4%. The mean age of sample firms is 25 years old, with firms being as youngas 3 years old and as old as 62 years. The average firm employs 13,000 workers.In addition, of the 176 sample firms with data on unionization, 60% have at leastsome of their employees covered by a union.In terms of pension plan characteristics, pension assets constitute 11% of over-all firms assets, on average, and pension liabilities amount to 13% of adjusted assetson average. In addition, the average ratio of defined benefit pension obligations toadjusted liabilities equals 13%. The ratio exhibits substantial variation across firmsin the sample. For example, the median contribution of pension obligations to ad-justed liabilities is close to 9% but the pension liabilities can be as high as 59% ofoverall liabilities. In addition, the average ratio of unfunded pension liabilities toadjusted liabilities in the sample is 2%. While the minimum ratio can be negative,indicating that some plans in the sample are overfunded, the maximum ratio ofunfunded pension obligations to adjusted liabilities is 48%.68The bottom part of Panel A, Table 3.3, reports summary statistics for sam-ple firms’ bankruptcy case characteristics. Out of the 172 cases with informa-tion on distressed terminations, 58%, or 100 firms, terminate at least one pensionplan in bankruptcy. A quarter of the sample firms agree with all creditors on thereorganization plan prior to filing for Chapter 11. Close to 40% of the samplebankruptcy cases are filed in the state of Delaware. In addition, the average dura-tion of bankruptcy in the sample is 20 months. As a comparison, the average dura-tion of 20 months in my sample is comparable to the average duration of 17 monthsin Jiang et al. (2012)’s sample over the period 1996 to 2007. On average, 73% ofsample firms emerge from Chapter 11 bankruptcy, and 19% of those companiesrefile for bankruptcy after having successfully reorganized once. In comparison,the likelihood of emergence in samples of all Compustat firms was 60% on aver-age over both the period 1979-2005 (Bharath, Panchapegesan, and Werner, 2009)and 1996-2007 (Jiang et al., 2012). For the sample of 107 cases for which I wasable to obtain information on the number of claim classes, the average firm faced10 classes of different claimants in Chapter 11. The number of claimants varies forthe sample firms, with 3 classes being the minimum and 62 classes the maximumnumber of separate claimants listed on the bankruptcy reorganization plan43. Thesummary statistics for the number of claim classes in bankruptcy are largely in linewith those reported in Jiang et al. (2012).The remainder of Table 3.3 reports summary statistics for the different subsam-ples of firms used in various specifications throughout this essay. I do not reportseparate summary statistics for the set of 176 firms with data on unionization be-cause these statistics are close to identical to the statistics in the main sample offirms.Panel B of Table 3.3 provides statistics for the 107 firms from the main samplefor which I could identify the number of claim classes in bankruptcy. Firms in thissubsample are slightly larger and more levered than companies in the main sample.In addition, only 40% of firms in this set terminate a pension plan in bankruptcy,43The bankruptcy case with 62 claim classes is that of Magna Entertainment Corp which filedfor Chapter 11 in 2009. In the joint plan of reorganization, the company lists 36 classes of creditorclaims and 26 classes of equity claims. Apart from that case, the next highest number of claim classesis 24.69compared to close to 60% of firms in the main sample. Firms in the sample withclaim classes information are more likely to emerge from bankruptcy, with 86%of firms emerging in this sample, compared with 73% in the main sample. Thisresult is not surprising as information on the number of claim classes was obtainedfrom firms’ plans of reorganization. While firms that liquidated in bankruptcy alsofile a plan of liquidation with the SEC, they do not always report the number ofclaimants in bankruptcy. As a result, the sample firms with claims informationrepresents more firms that emerge from bankruptcy than the main sample. Allother characteristics are comparable to those in the main sample.The next panel in Table 3.3 reports summary statistics for the 176 plan-levelobservations of 83 unique firms with information on Form 5500 data. There are90 firm-year observations in this sample as firms that refile for bankruptcy appearin the sample more than once. On average, sample firms sponsor 2 defined benefitpension plans. The median firm sponsors 1 plan, but there are firms which sponsoras many as 26 defined benefit plans in a year. The average plan age is 29 yearsand the oldest plan in the sample is 78 years old. These statistics indicate that mostdefined benefit pension plans were established a long time ago. In comparison, theaverage firm in this sample is 32 years old and the oldest sponsor is 56 years old.Such a differences in the maximum ages is possible since firm age is measured asthe first year in which the firm appears in the Compustat database, whereas planage is calculated by using the first date in which the pension plan became effective.As firms may have started sponsoring a defined benefit pension plan while theywere a private company, firm age and plan age do not have to coincide.In terms of pension variables, plan-level assets and liabilities represent a smallerportion of firm assets and liabilities which can be expected given that the mainsample consolidates all plans into a single firm observation. Moreover, 43% ofplans in the sample are represented by a union, compared to the sample averageof 60%. The average plan covers close to 6,000 employees of which 32% are ac-tively working and earning benefits under the defined benefit plan. The remainderof the covered employees are either retirees, employees who are working but havenot reached the vesting requirement, or deceased beneficiaries whose spouses earnsome of the promised benefits. Last, 44% of the sample pension plans are termi-nated in bankruptcy, compared to 60% in the main sample.70Panel D of Table 3.3 presents summary statistics for the 32 firms with dataon unsecured creditors’ recovery rates. Firms in this subsample are larger thanfirms in the full sample, with the median adjusted assets in the subsample equal to$1,264 million and the median firm size in the main sample equal to $589 million.With an average financial debt of 57% and additional debt of 53%, the 32 firmshave more financial liabilities and fewer additional liabilities than firms in the mainsample. In addition, firms in the subsample have somewhat higher ratios of pensionliabilities and unfunded pension liabilities to adjusted liabilities. In terms of theother firm characteristics, the 32 firms in the subsample are largely comparableto the main companies under study. In addition to firm characteristics, the tablepresents summary statistics for the variables used in equation (3.4). Firms in thesubsample have a market value of equity of 9% relative to adjusted assets, a defaultbarrier of 44% relative to adjusted assets and they issue 56% of debt in long-termdebt instruments on average.In addition, Panel E in Table 3.3 reports the pairwise correlation coefficientsfor the main variables used in the tests in this paper. As the coefficients in the tableindicate, none of the variables are strongly correlated.3.3.3 Univariate ComparisonsAlong with the summary statistics in the last fiscal year prior to Chapter 11, Iconsider how the main variables of interest change as a result of the bankruptcyrestructuring. Out of the 236 defined benefit sponsors in the sample, 64 companieshave non-missing data on all variables of interest in the year prior to bankruptcy andin the year after emergence. I compare the median values of these variables beforeand after bankruptcy and present results from these univariate comparisons in Table3.4. The results in the table show that while firm size declines only marginallyupon emergence for the sample firms, financial leverage is reduced almost in halfin the bankruptcy restructuring. In addition, firms’ cash holdings almost doubleafter reorganization relative to their levels prior to bankruptcy. At the same time,none of the pension plan variables appear to change significantly throughout thereorganization period.In Figure 3.1, I also plot the evolution of defined benefit pension obligations71and pension assets in event time for these 64 firms. As the figure shows, the valuesof the pension liabilities is relatively similar in the years right before and right afterbankruptcy, but both pension assets and pension obligations decline significantlyin the years after reorganization. These trends could indicate benefit reductions asa result of bankruptcy. In the next sections, I turn to test these changes and theirdeterminants in a multivariate setting.3.4 Bankruptcy DurationIn this section, I ask whether defined benefit claimants are related to the dura-tion of the bankruptcy restructuring process. Previous studies have documenteda relationship between various alternative lenders and bankruptcy duration, so Iconsider if pension claimants also influence the speed of reorganization. I controlfor the firm’s overall indebtedness and consider if the ratio of pension obligationsto overall liabilities provides explanatory power for the time spent in bankruptcy.The premise of this design is that once I account for all firm obligations, the con-tribution of pension liabilities to adjusted liabilities will capture the role of definedbenefit claimants in the reorganization process.3.4.1 Regression SpecificationFollowing the literature, I account for several variables that have been foundto be related to the time spent in bankruptcy. Denis and Rodgers (2007) identifyfirm size and leverage as key variables that influence the duration of bankruptcy.Moulton and Thomas (1993), Hotchkiss (1993) and Bryan et al. (2002) providesimilar evidence relating size and leverage to bankruptcy outcomes. Therefore, Icontrol for firm size, measured as the logarithm of adjusted assets, and leverage,measured as short-term and long-term debt scaled by adjusted assets, in the spec-ifications. While this definition of leverage is commonly used in the literature,it largely understates firms’ true financial position (Eisfeldt and Rampini, 2009;Shivdasani and Stefanescu, 2010; Welch, 2011)44. To account for firms’ overall44The traditional definition of leverage scales the sum of short-term and long-term debt by totalassets. My measure differs from the standard leverage measure as it scales the sum of short-term and72indebtedness, I also include additional leverage, measured as the remaining part oftotal balance sheet liabilities and two of the largest balance sheet liabilities, definedbenefit obligations and operating leases. The underlying assumption of includingall firm liabilities is that all claimants are important in a bankrupt firm (Denis andRodgers, 2007).Along with size and indebtedness, I control for several additional firm char-acteristics that have been found to influence Chapter 11 outcomes. I use cash toadjusted assets and cash flow volatility as proxies for the firms’ liquidity, the abil-ity to meet short-term commitments and the potential to generate working capitalfunds. I control for the share of tangible assets relative to adjusted assets as ameasure of liquidity and debt capacity. Moreover, since less profitable firms arelikely to become less liquid and more highly geared, I control for profitability byincluding return on assets divided by adjusted assets.In addition to firm characteristics, I account for some bankruptcy featureswhich have been found to matter in the Chapter 11 reorganization. Prepackagedbankruptcies are typically accompanied by a plan of reorganization that has beenaccepted by all existing claim classes. As a result, firms that file a prepackagedbankruptcy will spend time less in bankruptcy than firms that do not file a prepack-aged bankruptcy. Hence, I control for prepackaged filings by including an indicatorvariable equal to one if the reorganization plan was preapproved by creditors in theregression specifications. In light of evidence that firms may strategically choosethe state in which they file for bankruptcy, I also include a dummy variable equal toone if a firm files for bankruptcy in the state of Delaware (Ayotte and Skeel, 2004).To determine the influence of defined benefit claimants on the duration ofbankruptcy, I use the ratio of defined benefit obligations to adjusted liabilities. Af-ter controlling for the firm’s overall indebtedness, the ratio of pension obligationsto overall liabilities captures the impact of defined benefit claimants relative to thefirm’s other lenders. Thus, I focus on the composition of the firm’s creditors and onwhether pension claimants exert any influence beyond that of traditional lenders.Motivated by Campbell (1996) who finds that bankruptcy outcomes vary byindustry and Dahiya et al. (2003) who note that the costs of financial distress maydiffer from one industry to another, all regressions include industry fixed effects.long-term debt by adjusted assets.73Since bankruptcy cases vary over time (Bharath et al., 2009) and because firms thatfile for bankruptcy in a recession could be intrinsically different from firms thatdefault during normal times (Ivashina, Iverson, and Smith, 2013), I also includeyear fixed effects to capture time trends in the data. Including all of these variables,the main specification is estimated as follows:Durationit = α0 +α1(DB LiabilitiesAd justed Liabilities)it−1+α2Ln(Ad j. Assets)it−1++α3Leverageit−1 +α4Additional Leverageit−1 +α5Cashit−1++α6CF Volatilityit−1 +α7Tangibilityit−1 +α8ROAit−1++α9Prepackagedit +α10Delawareit + γ j + γt + εit−1(3.1)where Duration is the natural logarithm of the number of months spent inChapter 11 and γ j and γt are industry and year fixed effects, respectively. Themain coefficient of interest is α1 as it captures the role of defined benefit claimantsrelative to the firm’s other creditors.Prior studies have documented at least two additional variables that could berelevant determinants of bankruptcy outcomes: firm unionization levels and thenumber of classes that hold claims to the firm’s assets in bankruptcy. As docu-mented by Chen, Kacperczyk, and Ortiz-Molina (2011), unions represent a power-ful stakeholder who is actively involved in bankruptcy negotiations. Moreover,some defined benefit pension plans are established under collective bargainingagreements with unions. Therefore, unions are likely relevant for the outcomesof bankruptcy. At the same time, Gilson et al. (1990) suggest that capital structurecomplexity may play a role in the reorganization process. Following the litera-ture, I capture the capital structure complexity by considering the number of claimclasses in bankruptcy. While these two variables may be important to include inthe main tests, data on firm unionization levels and on claim classes is sparse. Forthat reason, I do not include these variables in the main tests but I report results forthe subsamples with data on unionization rates and claim classes.Next, I repeat the above estimations by substituting the ratio of defined benefit74obligations to adjusted liabilities with the ratio of unfunded pension obligations toadjusted liabilities. In that way, I consider whether the portion of pension liabilitieswhich is not secured by pension assets predicts defined benefit claimants’ actions.3.4.2 Determinants of Bankruptcy DurationIn Table 3.5, I estimate equation (3.1) in the main sample of 236 bankrupt de-fined benefit sponsors as well as two subsamples with union and claim classes data.The dependent variable in all models is the logarithm of the number of monthsspent in Chapter 11 and the independent variables are the firm and deal character-istics defined above. All specifications include industry and year fixed effects andstandard errors clustered at the industry level45.Model 1 in Table 3.5 presents results for the benchmark regression with thevariables commonly used in studies of duration in the literature. In line with pre-vious studies, I find that larger firms spend longer in bankruptcy reorganizationwhereas more levered firms resolve the Chapter 11 process faster. To the extentthat higher leverage leads firms to become bankrupt more quickly, firms with higherpre-bankruptcy leverage may be less economically distressed and, therefore, morelikely to reorganize faster in bankruptcy. However, the leverage results are limitedto the traditional measure of leverage: additional leverage does not influence theduration of bankruptcy. Moreover, firms with volatile cash flows spend more timein bankruptcy, as do more profitable firms. The results in Model 1 also confirmthat prepackaged bankruptcies and filings in Delaware are associated with an ac-celerated resolution of bankruptcy. Since in a prepackaged bankruptcy, the plan ofreorganization has typically been approved by all claim classes, firms that file suchbankruptcies spend significantly less time reorganizing under Chapter 11.In Model 2 I examine the role of defined benefit claimants on bankruptcy du-ration by including the ratio of defined benefit obligations to the firm’s adjustedliabilities in the regression specification. As the results show, the variables fromthe benchmark specification retain their signs and significance in Model 2. Theratio of pension liabilities to adjusted liabilities does not influence the time firms45The results remain unchanged under different clustering specifications.75spend reorganizing in my sample.Model 3 considers the impact of accounting for firms’ unionization status onthe role of defined benefit claimants in bankruptcy. Including the union variable inthe specification does not alter the benchmark results and pension claimants do notinfluence the duration of bankruptcy in this subsample. However, the union dummyenters the specification with a negative and significant coefficient. In particular,switching from not having a union to having employees represented by a union isassociated with a 26% reduction in the duration of bankruptcy. Thus, while definedbenefit claimants do not impact duration, unions facilitate negotiations in Chapter11 and accelerate the resolution of bankruptcy reorganization.In the fourth specification, I consider whether accounting for firms’ capitalstructure complexity changes the main results on the determinants of bankruptcyduration. Due to data availability, the tests on the importance of accounting fordifferent claim classes are restrained to the 107 companies with such data. Inthis reduced sample, prepackaged bankruptcies are the only control variable withsignificant explanatory power for bankruptcy duration. None of the other controls,including the benchmark variables, the measure of pension claimants’ role, or thenumber of claim classes are significant determinants of the time spent in Chapter11 in Model 4.The last three models in Table 3.5 present results for tests using underfundedpension obligations as a proxy for the role of defined benefit claimants in bankruptcy.The results in the last three columns are in line with the rest of the findings: higherleverage, prepackaged bankruptcies, and Delaware filings are associated with fasterbankruptcy reorganization, whereas size, cash flow volatility and profitability arerelated to longer bankruptcy duration. The unfunded portion of defined benefitobligations relative to the firm’s overall liabilities does not impact the time firmsspend reorganizing in my sample. Unions, on the other hand, are related to an ex-pedited restructuring under Chapter 11. Therefore, unions appear to facilitate thebargaining among creditors in bankruptcy reorganization.Overall, the results from Table 3.5 indicate that defined benefit claimants donot influence the duration of bankruptcy reorganization. Controlling for the firm’stotal indebtedness, pension claimants do not impact the length of Chapter 11 pro-ceedings beyond the influence of the firm’s other lenders in my sample.763.5 Emergence from Chapter 11In this section, I examine whether defined benefit claimants are related to thelikelihood that firms emerge from bankruptcy reorganization. As the likelihood ofsuccessful reorganization is a common outcome of interest in studies of bankruptcy,I consider whether pension claimants influence the probability of emergence. I usethe same regression specification and control variables as in the tests of bankruptcyduration since both duration and emergence are bankruptcy outcomes likely deter-mined by similar firm characteristics. Once again, I control for overall indebted-ness and I measure the relative role of defined benefit claimants using the ratio ofpension obligations to overall liabilities.3.5.1 Determinants of the Likelihood to EmergeTo test for a relationship between defined benefit claimants and the likelihoodthat firms emerge from bankruptcy, I estimate equation (3.1) in the sample of 236bankrupt defined benefit sponsors. The results on the likelihood of emergenceare presented in Table 3.6. The dependent variable is an indicator variable equalto one if a firm successfully emerges from bankruptcy, and zero otherwise, andthe independent variables are the firm and deal characteristics defined above. Allspecifications include industry and year fixed effects and robust standard errorsclustered at the industry level.Model 1 in Table 3.6 reports results for the benchmark regression using thevariables commonly used in the literature to explain the likelihood that firms emergefrom bankruptcy. Consistent with previous studies, I find that larger firms aremore likely to emerge from bankruptcy. In addition, higher leverage is associatedwith a higher likelihood of successful reorganization. If higher leverage leads tobankruptcy more quickly, firms with higher pre-bankruptcy leverage may be lesseconomically distressed and, therefore, more likely to be able to reorganize andemerge from bankruptcy than firms with lower leverage. Once again, the leverageeffect is only significant for the traditional leverage measure and not for additionalleverage. Not surprisingly, firms that file a prepackaged bankruptcy are more likely77to emerge as a new entity since their plan of reorganization was accepted by allclaimants prior to the filing.Models 2 to 7 in Table 3.6 modify the benchmark model by including the twoproxies of the role of defined benefit claimants. Model 2 considers the effect of theratio of defined benefit obligations to overall liabilities on the probability of emer-gence. The results indicate that the portion of adjusted liabilities that is comprisedof defined benefit obligations is not related to the likelihood that firms emerge fromChapter 11 in my sample. Therefore, defined benefit claimants do not appear to in-fluence the likelihood of successful reorganization in bankruptcy above the impactof traditional lenders.The next two specifications consider the relationship between emergence andthe role of pension claimants when unionization and claim classes are considered.As the results from Model 3 show, accounting for whether firm employees arerepresented by a union does not change the results from the two previous specifi-cations. Firm unionization status, however, is positively related to the likelihoodof successful reorganization in bankruptcy. In particular, switching from no unionrepresentation to union representation is related to a 14% increase in the probabilityof bankruptcy emergence, which is sizable given the sample average likelihood ofemergence of 73%. Therefore, unions appear to positively impact the bankruptcyreorganization process in terms of the likelihood of successful reorganization underChapter 11. This finding supports the results from Table 3.5 for a role for unionsin aiding negotiations in bankruptcy. However, the unionization effect is only sig-nificant at the 10% level. Last, when the complexity of firms’ capital structure isconsidered (Model 4), firm size and the Delaware indicator are the only signifi-cant predictors of the likelihood that firms emerge from Chapter 11. The ratio ofpension liabilities to adjusted liabilities is not significant in either subsample.Models 5 to 7 in Table 3.6 consider the relationship between emergence anddefined benefit claimants as proxied by the ratio of the unfunded pension obliga-tions to adjusted liabilities. The results in these models are identical to those inModels 2 to 4. Once again, pension claimants do not seem to influence emergencefrom bankruptcy in these models. Unions, on the other hand, are related to a higherprobability of bankruptcy emergence. In sum, the results in Table 3.6 suggest thatdefined benefit claimants do not influence the likelihood that firms reorganize suc-78cessfully in Chapter 11 beyond traditional lenders’ impact.3.6 Defined Benefit Plan Termination in Chapter 11In this section, I consider whether defined benefit claimants influence firms’decision to terminate a defined benefit pension plan in bankruptcy. Chapter 11bankruptcy is a particularly costly form of reorganization for pension claimantsbecause it is the only time in which underfunded defined benefit plans can be ter-minated. Having discussed the threat of plan termination as a possible driver ofpension claimants’ actions to avoid Chapter 11 bankruptcy in the previous essay,I now test whether defined benefit claimants influence the termination decision inbankruptcy. The core insight for this test is that if pension claimants are related todistressed plan terminations, they may anticipate this outcome before bankruptcywhich in turn gives them an incentive to avoid bankruptcy, as discussed in the firstessay.3.6.1 Regression SpecificationWhile the specification in equation (3.1) is useful in predicting bankruptcy andits outcomes, the decision to terminate a pension plan in bankruptcy is likely drivenby different factors. Therefore, I turn to the literature for appropriate determinantsof the plan termination decision. The literature on defined benefit terminations inbankruptcy is largely focused on the termination of overfunded pension plans out-side of bankruptcy. Rauh (2009) is a notable exception as he considers distresspension terminations in his study of corporate pension plans’ investment policies.I use Rauh (2009)’s model of distressed terminations as the benchmark specifi-cation and I extend it to include my measures for the role of pension claimants inbankruptcy. Defined benefit claimants may influence the decision to terminate theirpension plan by actively opposing the proposed termination and related reorgani-zation plan or by agreeing to the termination without many objections.Following Rauh (2009), I study defined benefit plan terminations at the planlevel. As a result, the sample of firms for which this test can be carried out is79constrained to those companies with data on Form 5500. Rauh (2009) controlsfor the pension funding status, defined as pension assets minus pension obligationsdivided by pension obligations, and the share of employees covered by the pensionplan who are currently working (as opposed to retirees). The model also includesthe pension plan assets and the logarithm of the pension plan assets as explanatoryvariables. Lastly, the author controls for the return on pension plan assets and yearfixed effects in the regression. Rauh (2009) finds that better funding status, a largershare of active employees, a larger size of the plan assets, and higher investmentreturns are all associated with a lower likelihood that the pension plan is terminatedin bankruptcy.I extend the model used by Rauh (2009) to account for the role of defined ben-efit claimants in bankruptcy. I am interested if after controlling for the firm’s finan-cial condition, defined benefit claimants are related to the decision to terminate thepension plan in bankruptcy. Therefore, I extend Rauh (2009)’s model to include thefirm’s overall indebtedness, measured by financial and additional leverage. Then, Iadd the ratio of defined benefit obligations to the firm’s overall liabilities to proxyfor defined benefit claimants’ bargaining power in bankruptcy relative to the firm’sother lenders. I estimate the following model:Terminateit = α0 +α1(DB Liab.Ad justed Liab.)it−1+α2(DB Assets−DB Liab.DB Liab.)it−1++α3Active Share o f Employeesit−1 +α4DB Assetsit−1++α5Ln(DB Assets)it−1 +α6Leverageit−1++α7Additional Leverageit−1 + γt + εit−1(3.2)where Terminate is an indicator variable equal to one when a specific definedbenefit pension plan is terminated in bankruptcy, and zero otherwise, DB Assetsequals the pension plan’s assets (as opposed to the consolidated pension assets atthe firm level) and all defined benefit pension variables come from Form 5500.Following Rauh (2009), all specifications include year fixed effects and standarderrors clustered at the firm level.80The main results of estimating equation (3.2) focus on the sample of firms withplan-level data from Form 5500. In addition, I re-estimate the above model byaccounting for unionized plans and capital structure complexity. Plan-level datais more appropriate than firm-level data in studying pension plan terminations be-cause firms often sponsor multiple pension plans and may not terminate all of themin bankruptcy. At the firm level, it is more difficult to pin down the determinantsof plan termination because the termination variable captures whether at least oneof the firm’s plans is terminated. Nevertheless, I also report results for the modelin equation (3.2) estimated at the firm level46.3.6.2 Determinants of Pension Plan TerminationTable 3.7 provides results from estimating equation (3.2) at the plan level in thesubsample of firms with Form 5500 data. The dependent variable in all specifica-tions is an indicator variable of whether a specific defined benefit pension plan isterminated in bankruptcy or not.Panel A of Table 3.7 reports results for the 176 plan-level observations withdata on Form 5500. Model 1 presents the benchmark model which includes thevariables suggested by Rauh (2009) as well as my two leverage controls. WhileRauh (2009) finds all variables but the plan assets to be negative and statisticallysignificant determinants of distressed plan terminations, none of the variables aresignificant in my sample. Firm leverage is the only significant variable in Model 1:higher leverage is associated with a lower likelihood of pension plan terminationin the benchmark specification.Model 2 examines defined benefit claimants’ influence on the pension plantermination decision. Once again, higher firm leverage is related to a lower prob-ability of pension plan termination. In addition, I find evidence that the ratio ofdefined benefit obligations to adjusted liabilities provides explanatory power to thebenchmark model. In particular, pension plans with higher obligations relative tothe firm’s overall liabilities are more likely to be terminated in bankruptcy. The46Untabulated tests show that the results from estimating equation (3.2) at the plan and firm levelsremain unchanged if I only use the variables identified by Rauh (2009) and my proxies for the roleof defined benefit claimants in bankruptcy (i.e. if I do not control for firm leverage).81effect is economically significant: a one standard deviation increase in the ratio ofpension obligations to adjusted liabilities is associated with a 10% increase in thelikelihood of pension plan termination in bankruptcy. The effect is economicallymeaningful given that the sample average likelihood of a distressed pension plantermination is 44%. These findings imply that whenever defined benefit claimantsare more sizable relative to the firm’s other lenders, and hence, more likely to ne-gotiate with the firm, pension plans are more likely to be terminated. As agreeingto termination is not in the best interest of defined benefit claimants, this resultindicates that plan terminations are one concession that defined benefit claimantshave to accept in bankruptcy.An alternative explanation might be that firms are more likely to terminate pen-sion plans whenever it is more profitable for them to do so, i.e. when the pensionliabilities are large relative to the firm’s other liabilities. I return to this argumentin the last few columns of Table 3.7 (discussed below) and find no evidence tosupport the claim. Overall, the relationship between defined benefit obligationsand distressed plan terminations indicates that defined benefit claimants exert aninfluence above the impact of the firm’s other lenders.The positive and significant relation between plan terminations and the ratioof defined benefit obligations to adjusted liabilities remains after controlling forplan unionization rates. In Model 3, the union variable shows whether a specificplan is subject to a collective bargaining agreement and represented by a union inbankruptcy. Controlling for unionization does not impact the main results. There-fore, the relationship between plan terminations and relative pension liabilities can-not be explained by union representation.The next specification presents estimation results for the subsample of 103plan-years with data on the number of claim classes in bankruptcy. The resultsfrom Model 4 show that there is a positive and significant relationship betweenplan termination and the ratio of defined benefit liabilities to adjusted liabilitiesafter controlling for firms’ capital structure complexity.If the Model 2 results are driven by firms which terminate those pension plansthat are more beneficial to end, the effect between termination and defined benefitobligations should be even stronger when the unfunded portion of defined benefitliabilities is considered. Firms will gain the largest benefit from terminating their82most underfunded plans because by doing so, firms can offload a larger liability tothe PBGC than if they terminate a better funded plan. The results in the last threecolumns of Table 3.7 provide no support for this story. In particular, the ratio ofunfunded pension obligations to overall liabilities is not significantly related to planterminations. The other explanatory variables retain the same sign and significanceas in the main specification.As a consistency check, I also study the relation between defined benefitclaimants and termination decisions at the firm level. Panel B of Table 3.7 presentsresults for the 171 firms from the main sample for which information on pensionplan termination was obtained. The dependent variable in the firm-level specifi-cations equals one when at least one of the firm’s pension plans is terminated inbankruptcy, and zero if none of the firm’s plans are terminated. The active shareof employees covered by a pension plan is no longer included as a control variablebecause this variable is not available on Compustat47.In the benchmark specification at the firm level, the logarithm of pension planassets is a negative and statistically significant determinant of the likelihood thatfirms terminate at least one of their pension plans in Chapter 11. Higher firmleverage is also associated with a lower frequency of pension plan terminationsin bankruptcy. Moreover, higher additional leverage is positively related to planterminations, but the effect is significant only at the 10% level. The firm fundingstatus is not related to the likelihood that firms terminate a defined benefit plan inChapter 11.Models 2 to 4 in Panel B of Table 3.7 consider the role of defined benefitclaimants in distressed plan terminations at the firm level. Similarly to the plan-level results, at the firm level the ratio of pension obligations to adjusted liabil-ities provides positive and significant explanatory power beyond the benchmarkcontrols. Once again, the effect is economically meaningful. In particular, a onestandard deviation increase in the ratio of pension obligations to overall firm lia-bilities is related to a 17% increase in the likelihood of a distressed pension plantermination. This effect is robust to accounting for firm unionization. In Model4, the sample of 78 firms with information on claim classes is considered. In that47The results presented in Panel B of Table 3.7 remain unchanged if the logarithm of total firmemployees from Compustat is included as an additional control variable.83specification, additional leverage is a positive determinant of plan terminations,while defined benefit claimants are not related to plan terminations. Since the re-sults for Model 4 are based on less than half of the observations in the benchmarkspecification, these results should be interpreted with caution.The last three models in Panel B present results using unfunded pension obli-gations as a proxy for the role of defined benefit claimants in bankruptcy. Model 5shows that the unfunded portion of pension obligations relative to overall liabilitiesis related to a higher likelihood of distressed pension plan terminations. However,the effect is only significant at the 10% level and disappears when I include theunion dummy. The results at the firm level are consistent with those at the planlevel: a higher ratio of defined benefit liabilities to adjusted liabilities is associ-ated with a higher likelihood that the firm terminates at least one pension plan inbankruptcy.Altogether, the results in Table 3.7 provide evidence of a relationship betweendefined benefit claimants and pension plan terminations in bankruptcy. The strongassociation between defined benefit obligations and distressed plan terminationsimplies a role for defined benefit claimants above and beyond the influence of thefirm’s other lenders. In addition, the results suggest that plan terminations maybe one concession defined benefit claimants may have to agree to in bankruptcy.This finding is consistent with the previous essay’s argument that defined benefitclaimants have high incentives to avoid bankruptcy because of the option to havetheir plans terminated in Chapter 11. Pension claimants expect that they may haveto bargain with the firm about the plan’s survival and eventually may have to agreeto plan terminations, so defined benefit claimants have stronger incentives to avoidbankruptcy than traditional lenders.3.7 Likelihood to Refile for Chapter 11My next question of interest is the extent to which defined benefit claimants andthe actions they take in bankruptcy predict the firm’s post-reorganization survival.The motivation for this test comes from the previous essay’s findings. In particular,in Chapter 2 I documented that defined benefit claimants are less likely to file for84bankruptcy. As an extension, it can be expected that defined benefit claimants willalso be related to a lower likelihood that firms refile for Chapter 11, since this isjust the likelihood of bankruptcy after having emerged from reorganization once.3.7.1 Regression SpecificationI cannot measure post-Chapter 11 performance for firms which are liquidated oracquired in bankruptcy. For that reason, when I study the probability of refilingI examine only the firms which emerge from Chapter 11 as independent publiclytraded companies. I test the extent to which the likelihood that firms refile forbankruptcy is driven by pension claimants and their actions.To capture the determinants of refiling for Chapter 11, I use the same vari-ables as the controls for the other outcomes of bankruptcy, duration and emer-gence. However, I include both the variables as measured in the year prior tobankruptcy and their changes during Chapter 11. Changes in reorganization aremeasured as the difference between firm characteristics in the first fiscal year afterthe firm emerges from bankruptcy and characteristics at the last fiscal year prior tobankruptcy. In particular, I consider whether changes in firm size, leverage, andpension obligations following the Chapter 11 reorganization influence the prob-ability of refiling for bankruptcy. In this way, I capture whether pre-bankruptcycharacteristics and the actions taken in bankruptcy influence the likelihood thatfirms end up in Chapter 11 again. In particular, I test the following model:Re f ilei = α0 +α1∆DB LiabilitiesAd justed Assets i+α2∆Ln(Ad justed Assets)i++α3∆Leveragei +α4∆Additional Leveragei +βiXi + εi(3.3)where X is the vector of control variables included in equation (3.1) measuredin the last fiscal year prior to bankruptcy and ∆ represents the change in the vari-ables of interest from the year after Chapter 11 emergence to the year prior tobankruptcy.853.7.2 Determinants of the Likelihood to RefileThe results on the likelihood of refiling for bankruptcy are presented in Table3.8. I estimate equation (3.3) in the sample of 64 bankrupt defined benefit sponsorswith information at both the last fiscal year prior to bankruptcy and the first fiscalyear after the firm emerges from bankruptcy. Fifteen of the firms in this subsamplerefile for Chapter 11 after emerging. The dependent variable is an indicator variableequal to one if a firm refiles for bankruptcy after having emerged from Chapter 11,and zero otherwise. The independent variables include the firm and bankruptcycharacteristics and changes in those characteristics after emerging relative to priorto bankruptcy.The first model in Table 3.8 presents results for the benchmark regression. Themain determinants of whether firms refile for Chapter 11 include changes in ad-ditional leverage and cash flow volatility. In particular, firms which reduce theirindebtedness in bankruptcy, as measured by additional liabilities, are less likely torefile for Chapter 11. Moreover, firms with more volatile cash flows prior to theoriginal Chapter 11 filing are also less likely to file for bankruptcy in the future.Models 2 to 5 in Table 3.8 add the ratio of pension obligations to overall lia-bilities and changes in this ratio to the benchmark specification. The results fromModel 2 show that beyond the benchmark controls, firms with more defined benefitclaimants relative to all lenders are less likely to refile for bankruptcy. In partic-ular, a one standard deviation increase in the ratio of defined benefit obligationsto overall liabilities is associated with a 15% decrease in the likelihood that thefirm refiles for Chapter 11. The effect is economically meaningful given that theaverage likelihood of refiling for bankruptcy in the sample is 23%. Accounting forfirm unionization levels (Model 3) does not alter the conclusions about the role ofdefined benefit claimants on the likelihood of refiling.Model 4 accounts for changes in the pension obligation after emergence rel-ative to prior to bankruptcy. As the results from that specification show, whiledefined benefit claimants play a role in the likelihood of refiling for bankruptcy,changes in the pension obligations do not influence the probability of refiling. Onceagain, accounting for unionization status does not change the findings discussedthus far (model 5). These results are consistent with the findings from the previous86chapter that firms with more defined benefit claimants relative to all lenders areless likely to file for bankruptcy.To determine if the unfunded portion of defined benefit liabilities relates to thelikelihood of refiling for bankruptcy, the last four columns in Table 3.8 focus on theratio of unfunded pension obligations to overall liabilities. The results from Model6 indicate that firms with more unfunded defined benefit obligations relative toall firm liabilities are less likely to refile for bankruptcy. A one standard deviationincrease in the ratio of unfunded pension obligations to adjusted liabilities is relatedto a 12% decline in the probability of refiling for bankruptcy. Once again, unionrepresentation does not impact the likelihood that firms refile for Chapter 11 (model7).The last two models in Table 3.8 report results from including both the levelof unfunded pension obligations and the change in unfunded obligations to thespecification. The unfunded portion of pension obligation relative to overall liabil-ities remains a negative and significant predictor of the probability of refiling forbankruptcy. At the same time, firms that emerge with more unfunded liabilities rel-ative to when they entered bankruptcy are less likely to refile for Chapter 11 in thefuture. While this result may seem counterintuitive, it could imply that regardlessof whether their pension obligations are funded or not, defined benefit claimantsare less likely to refile for bankruptcy. The last specification confirms that firms’unionization status does not influence the probability of refiling for Chapter 11.Overall, the results presented in this subsection indicate that defined benefitclaimants influence firms’ post-reorganization performance. Whenever pensionobligations represent a larger portion of the firm’s overall liabilities upon filingfor bankruptcy, firms are less likely to refile for Chapter 11 in the future. Evenfurther, the results in Table 3.8 provide additional support for the earlier tests thatshowed that firms with larger defined benefit claimants relative to other lenders areless likely to file for Chapter 11 bankruptcy. In the tests presented in this section,I find that firms with more defined benefit claimants are less likely to refile forbankruptcy.873.8 Unsecured Creditors’ Recovery RatesOne additional outcome of the bankruptcy process that defined benefit claimantsmay influence is the unsecured creditors’ recovery rates. Studying recovery rates isrelevant because these rates indicate how much of their investment creditors gaveup in bankruptcy. Unsecured creditors’ recovery rates are of particular interest inthis essay because defined benefit claimants are typically members of the unse-cured creditors’ committee. As a result, pension claimants will have the strongestimpact on the amounts recovered by unsecured creditors. As unsecured lenders inbankruptcy, pension claimants may negotiate with the firm and take actions whichinfluence the overall unsecured creditors’ recovery rates. At the same time, pensionclaimants may not impact creditors’ recoveries if they are not powerful enough orif they do not negotiate with the firm more than the other lenders. Therefore, inthis section I investigate whether defined benefit claimants influence the ability ofthe firm’s unsecured creditors to recover their investment in the bankrupt firm.3.8.1 Regression SpecificationThe literature has documented various determinants of creditors’ recovery rates.Following Jankowitsch, Nagler, and Subrahmanyam (2014), I account for firms’financial condition by using the market value of equity over adjusted assets. Struc-tural credit risk models use the value of equity to infer the company’s asset valueand to define the leverage. In addition, I calculate the firm’s default barrier asdefined by Moody’s Analytics. As in Jankowitsch et al. (2014), I control for long-term debt issuance by the ratio of long-term debt to total debt. Moreover, I includeasset tangibility, profitability, firm size and the number of employees in the modelas those variables have been found to influence recovery rates. Along with thesedeterminants of creditor recovery rates, I consider the ratio of defined benefit lia-bilities to adjusted liabilities. Thus, I estimate the following specification:88Recovery Rateit = α0 +α1(DB LiabilitiesAd justed Liabilities)it−1+α2Equityit−1++α3De f ault Barrierit−1 +α4LT D Issuanceit−1++α5Asset Tangibilityit−1 +α6Pro f itabilityit−1++α7Ln(Ad j. Assets)it−1 +α8Ln(Employees)it−1 + εit−1(3.4)Once again, I also consider whether the unfunded portion of defined benefitobligations impacts unsecured creditors’ recovery rates.3.8.2 Determinants of Unsecured Creditors’ Recovery RatesI estimate equation (3.4) on the sample of 32 firm-year observations with dataon creditor recovery rates from Capital IQ. The results are presented in Table 3.9.The dependent variable equals the total unsecured creditors’ recovery rate uponbankruptcy emergence.Model 1 presents regression results including the variables commonly used toexplain creditor recovery rates in the literature. The results indicate that equityvalue is the strongest predictor of unsecured lenders’ recovery rates in my sam-ple: the higher the equity value, the lower the unsecured creditors’ recovery rate.In Model 2, I test whether defined benefit claimants impact the recovery rates forunsecured creditors. The results confirm the existence of a relationship betweendefined benefit claimants and recovery rates: a higher ratio of pension obligationsto adjusted liabilities is associated with higher recovery rates for unsecured cred-itors. In terms of economic significance, a one standard deviation change in theratio of pension liabilities to overall liabilities is associated with a 6% increase inrecovery rates. Such an effect is meaningful given that the sample average unse-cured creditors’ recovery rate is 40%. The results from Model 3 indicate that therelationship does not carry over to the unfunded portion of pension liabilities toadjusted liabilities48.48Untabulated tests confirm that the results presented in Table 3.9 remain unchanged when firmunionization rates are considered. Firm unionization does not impact unsecured creditors’ recoveryrates and accounting for unions does not influence the role of defined benefit claimants on unsecured89There are several possible explanations for the positive relationship betweenunsecured creditors’ recovery rates and the ratio of defined benefit liabilities tooverall liabilities. One possibility is that firms with a high proportion of pensionobligations relative to adjusted liabilities are more likely to terminate their pensionplan, thus leaving more for other unsecured claimants to recover. Another expla-nation is that a higher ratio of pension obligations to adjusted liabilities indicatesthat there are fewer claimants to bargain with, and improved negotiations may leadto higher recovery rates for unsecured claimants. While the few data points in mysample prohibit me from differentiating among these channels, the results in Table3.9 suggest that defined benefit claimants may play a role in creditors’ recoveriesbeyond the influence of traditional lenders.3.9 Changes in Pension Benefits During BankruptcyOne possible channel through which defined benefit claimants may influence thebankruptcy restructuring process is through bargaining about their pension benefits.To test if this is one of the mechanisms at play, I consider whether defined benefitclaimants are related to changes in the pension obligation during bankruptcy. Areduction in the defined benefit obligation may indicate that pension claimants ac-cepted concessions to help the firm reorganize. However, all unsecured creditorsusually suffer losses in bankruptcy, so at least part of the decline, if not the entirereduction in pension benefits, may be explained by the concessions taken by allfirm creditors. For that reason, when I study changes in pension liabilities through-out Chapter 11, I control for the expected reductions in bankruptcy and considerwhether pension claimants have any additional role in explaining the reductions intheir liabilities beyond the expected losses.3.9.1 Regression SpecificationTo account for the expected reductions that creditors experience in bankruptcy,I include the change in financial liabilities over the bankruptcy period in my re-creditors’ recovery rates.90gression specifications. The change in leverage is measured as the difference infinancial liabilities from the year after emerging to the year prior to bankruptcyscaled by adjusted assets in the year prior to Chapter 11. The change in leverage ismeant to approximate the average cut in liabilities that a firm’s creditors experienceas a result of bankruptcy. After controlling for the expected reduction in liabilities,I consider whether defined benefit claimants exert any additional influence on thechanges in their obligation in bankruptcy. The regression specification is estimatedas follows:∆DB LiabilitiesAd justed Assets= α0 +α1(DB LiabilitiesAd justed Liabilities)it−1++α1Ln(Ad j. Assets)it−1 +α2Leverageit−1 +α3∆Leverage++α4Employeesit−1 +α5∆Employees+ εit−1(3.5)where the dependent variable is the change in pension liabilities over thebankruptcy process, measured as the difference between defined benefit obliga-tions upon bankruptcy emergence and the pension obligations prior to filing forbankruptcy, scaled by adjusted assets in the year prior to bankruptcy. All changesin the other control variables presented are measured in the same way. The im-plicit assumption in this model is that financial liabilities are similar to definedbenefit obligations and reductions in the financial obligations capture the expectedreduction in pension obligations.One reason why pension claimants may explain changes in pension obligationsabove and beyond creditors’ expected losses in bankruptcy is that defined benefitclaimants influence the decision to terminate a pension plan, as discussed above. Inparticular, if a pension plan is terminated in bankruptcy, defined benefit obligationsare expected to decline. Therefore, pension claimants may influence the reductionin pension obligations through their relationship to plan terminations. To ensure Iam not capturing the termination effect when I study changes in pension obliga-tions, I re-estimate equation (3.5) by including a control for plan terminations. Theinclusion of the termination dummy reduces the sample size to 39 observations91because of missing information on the plan termination variable.3.9.2 Determinants of Changes in Pension Benefits in BankruptcyI study pension benefit concessions as one of the channels through which de-fined benefit claimants may influence various characteristics and outcomes of thebankruptcy process. Table 3.10 reports the results from estimating equation (3.5)on the sample of 62 bankrupt defined benefit sponsors with data after bankruptcyemergence49. The dependent variable in all specifications is the difference betweendefined benefit pension obligations upon bankruptcy emergence and the pensionobligations prior to filing for bankruptcy, scaled by adjusted assets in the year priorto bankruptcy.In the first model of Table 3.10 I consider whether firm size, leverage, changesin the leverage, the number of employees and changes in the number of employeesdetermine changes in the defined benefit pension liabilities during bankruptcy. Ascolumn (1) shows, none of the control variables are significantly related to changesin pension obligations. Model (2) introduces the ratio of defined benefit obligationsto adjusted liabilities as an additional control. The ratio does not explain changes inthe pension liabilities over the course of bankruptcy either. The lack of significanceof the ratio suggests that defined benefit claimants do not explain changes in thepension obligations throughout the Chapter 11 process.Model (3) in Table 3.10 tests whether the unfunded portion of pension liabili-ties relative to overall liabilities can explain changes in pension obligations duringbankruptcy. Several of the control variables in the specification provide explana-tory power. First, high firm leverage in the year before bankruptcy helps explainchanges in pension obligations in bankruptcy. In particular, firms that are morelevered upon bankruptcy filing are less likely to reduce their pension obligations inreorganization. Second, changes in firm financial leverage are related to changesin pension liabilities. Not surprisingly, firms that reduce their financial liabilitiesin bankruptcy also reduce their defined benefit liabilities in reorganization. Third,changes in the number of firm employees are related to changes in the pension49The sample size in this estimation drops from 64 in the earlier specification (Table 3.8) to 62 inthis model because of 2 firms with missing information on the number of employees.92obligations. Firms that reduce their workforce upon bankruptcy emergence alsoexperience a decline in their defined benefit obligations.Controlling for firms characteristics and their changes, I find that the unfundedportion of defined benefit liabilities to overall liabilities is a significant determinantof changes in pension obligations in bankruptcy. In particular, higher unfundedpension obligations relative to all firm liabilities are related to higher reductions inthe defined benefit obligations after bankruptcy. Since I already control for the ex-pected reductions in liabilities due to the bankruptcy reorganization, the significantratio of unfunded pension liabilities to adjusted liabilities suggests an influence ofpension claimants above that of traditional lenders.The last three models in Table 3.10 present results for the 39 bankrupt de-fined benefit sponsors with information on plan termination and with data afterbankruptcy emergence. In these specifications, I check if the results presented inthe first three column are driven by plans which were terminated in bankruptcy. Aplan termination will mechanically reduce the pension liabilities after bankruptcyrelative to those before Chapter 11. This effect could be captured by the proxy forthe role of defined benefit claimants as the ratio of defined benefit obligations tooverall liabilities is related to the likelihood of plan terminations. As the resultsin Table 3.10 show, the ratio of unfunded pension obligations to overall liabili-ties remains a significant determinant of changes in the pension obligations aftercontrolling for plan terminations. Therefore, plan terminations do not explain theentire reduction in defined benefit obligations in bankruptcy.Overall, the results from Table 3.10 provide evidence that defined benefitclaimants influence bankruptcy reorganization beyond the impact of traditionallenders. One of the mechanisms through which pension claimants impact the reor-ganization is by accepting cuts in their liabilities. The change in leverage variableis meant to capture the expected reduction in liabilities as a result of the bankruptcyreorganization. Therefore, any remaining explanatory power that pension obliga-tions have is an impact above the average reduction in liabilities that traditionalcreditors take in bankruptcy. The reductions that pension claimants agree to canbe explained in part by the expected liability reductions in bankruptcy, but they arelargely determined by the ratio of pension claimants with higher unfunded liabili-ties relative to overall liabilities.93In this essay, I document that one channel through which defined benefitclaimants influence the restructuring process is through agreeing to pension cuts.The fact that this mechanism is only at work for unfunded pension obligations isconsistent with previous findings that pension claimants give up the most whenthey stand to lose the most. In a study of bargaining between management and la-bor in distressed airlines, Benmelech et al. (2012) show that management threatenslabor with plan terminations and obtains the highest concessions from those laborunions which would incur the highest losses of plan terminations in bankruptcy- thepilots. While I cannot provide such detailed evidence as Benmelech et al. (2012),my findings are consistent with their results as I find that pension beneficiarieswith unfunded pension plans stand to lose the most and also give up the most inbankruptcy.3.10 ConclusionIn this essay, I investigate whether defined benefit claimants in Chapter 11 influ-ence the bankruptcy restructuring process. Given the similarities and differencesbetween pension claimants and traditional firm creditors, defined benefit claimantsmay influence the reorganization process but they may also negotiate with the firmsimilarly to other lenders, thus not playing a discernible role in bankruptcy. To de-termine if pension claimants impact the reorganization process above and beyondthe role of traditional lenders, I control for all firm liabilities and study the ratioof pension obligations to overall liabilities as a measure of the relative influence ofdefined benefit claimants. While they do not impact the bankruptcy duration or thelikelihood of emergence differently from other lenders, defined benefit claimantsinfluence other aspects of the Chapter 11 reorganization. In particular, defined ben-efit claimants are positively related to the likelihood that a pension plan is termi-nated in bankruptcy. This result supports the argument in the first essay that definedbenefit claimants foresee negotiations about plan terminations in bankruptcy andwant to avoid them by negotiating with the firm outside of Chapter 11.Along with the termination decision, defined benefit claimants influence thelikelihood that firms refile for Chapter 11 bankruptcy. Consistent with the first94essay’s results, I find that once they are out of bankruptcy, pension claimants areagain associated with a lower likelihood of filing for Chapter 11. I also providesome indication of a positive relationship between defined benefit claimants, whoare typically unsecured creditors in bankruptcy, and unsecured creditors’ recoveryrates.Last, I try to shed light on the possible channels through which defined benefitclaimants influence the bankruptcy reorganization process. One natural candidatefor the mechanism through which defined benefit claimants influence the reorgani-zation process is their willingness to accept reductions in pension benefits. How-ever, since most unsecured creditors experience some loss in bankruptcy, I attemptto control for the expected change in the pension liabilities and consider the role ofpension claimants above the predicted losses. I find that defined benefit claimantsinfluence changes in their obligations. This effect is present for the unfunded por-tion of pension liabilities relative to overall liabilities, which suggests that definedbenefit claimants experience the largest cuts in bankruptcy whenever they stand tolose the most in Chapter 11.The collective results in this essay indicate a role for defined benefit claimantsabove and beyond the influence of the firm’s other lenders. While defined benefitclaimants may not impact all bankruptcy outcomes, they are relevant for certainimportant features of the reorganization process. Therefore, the role of definedbenefit claimants should be considered when bankruptcy outcomes are examined.95Figures and TablesFigure 3.1: Changes in Pension Variables Before and After BankruptcyThis figure shows the evolution of the average defined benefit pension obligations (PBO)and the average pension assets (PPA) from two years prior to bankruptcy to two years afteremerging from Chapter 11. Thus, year 1 in the figure refers to the first year after Chap-ter 11 emergence. The statistics are based on data from 64 defined benefit sponsors withinformation in both the last year before bankruptcy and the first year after emergence. 2003004005006007008009001000- 2 - 1 0 1 2$ millions Years relative to Chapter 11 filing and emergence  PB O PP A96Table 3.1: Variable DefinitionsThis table provides definitions for the variables used in this study and the data sources from which variables were obtained. When available, Compustat data items are included in brackets.Note that due to the Compustat naming convention, the same variable name, pcppao, refers to different variables over time. Prior to 1998, pcppao refers to the pension cost for overfundedplans. Between 1998 and 2006, a positive value of pcppao refers to the prepaid pension cost, while a negative value of pcppa refers to the accrued pension cost. After 2006, pcppao refers toplan funding status.Variable Definition SourceActive Share of Employees Standard deviation of quarterly operating income (oibdpq) over previous 12 quarters scaled by total assets (at) Form 5500Additional Leverage [Adjusted Liabilities - short-term debt (dlc) - long-term debt (dltt)]/ Adjusted Assets CompustatAdjusted Assets If 1987 <= fiscal year <= 1997 and pcppao > 0, use Total assets (at) + DB assets + Operating lease - Prepaid pension cost (pcppao) -Underfunded prepaid pension cost (pcppau)CompustatIf 1998 <= fiscal year <= 2006 and pcppao > 0, use Total assets (at) + DB assets + Operating lease - Prepaid pension cost (pcppao)If 2007 <= fiscal year and funded status > 0, use Total assets (at) + DB assets + Operating lease - Funded status (pcppao)Adjusted Liabilities If 1987 <= fiscal year <= 1997 and pcppao < 0, use Total liabilities (lt) + DB liabilities + Operating lease - abs(Accrued pension cost(pcppao)) - abs(Underfunded accrued pension cost (pcppau))CompustatIf 1998 <= fiscal year <= 2006 and pcppao < 0, use Total liabilities (lt) + DB liabilities + Operating lease - abs(Accrued pension cost(pcppao))If 2007 <= fiscal year and funded status < 0, use Total liabilities (lt) + DB liabilities + Operating lease - abs(Funded status (pcppao))Asset Tangibility Net property, plant, and equipment (ppent) / Adjusted Assets CompustatAssets Total assets (at) CompustatCash Cash and short-term investments (che) / Adjusted assets CompustatCash Flow Volatility Standard deviation of quarterly operating income (oibdpq) over previous 12 quarters scaled by adjusted assets CompustatClaim Classes The number of claim classes identified on the bankruptcy reorganization plan Form 8KDefault Barrier (Short-term debt (dlc) + 0.5*long-term debt (dltt)) / Adjusted Assets CompustatDelaware Dummy = 1 if the Chapter 11 case was filed in the state of Delaware, and 0 otherwise BRDDuration Number of months in bankruptcy, from the date of filing to the date of plan confirmation BRDEmerge Dummy = 1 if the bankrupt firm emerges from bankruptcy, and 0 otherwise 8K, BRD, FactivaEmployees The number of employees (emp) CompustatEquity Market value of equity / Adjusted Assets CompustatFirm Age The number of years since the firm first reports on Compustat CompustatLeverage (Short-term debt (dlc) + long-term debt (dltt)) / Adjusted Assets CompustatLTD Issuance Long-term debt (dltt) / Short-term debt (dlc) + long-term debt (dltt) CompustatDB Liabilities (PBO) If 1987<=fiscal year<=1997, Pension benefit projected obligation (pbpro)+Underfunded pension benefit projected obligation (pbpru) Compustat,If fiscal year >= 1998, Pension benefit projected obligation (pbpro) Form 5500Plan Age The number of years since the defined benefit plan became effective Form 5500DB Assets (PPA) If 1987 <= fiscal year <= 1997, Pension plan assets (pplao) + Underfunded pension plan assets (pplau) Compustat,If fiscal year >= 1998, Pension plan assets (pplao) Form 5500Prepackaged Dummy = 1 if a bankruptcy is prepackaged or prenegotiated BRDRecovery Rate Unsecured creditors’ recovery rates Capital IQRefile Dummy = 1 if the bankrupt firm refiles for bankruptcy, and 0 otherwise BRDReturn on Assets Operating income before depreciation (oibdp) / Adjusted Assets CompustatTerminate Dummy = 1 if the firm terminates at least one defined benefit pension plan in bankruptcy, and 0 otherwise PBGC, Form 5500Union Dummy = 1 if at least one of the firm’s defined benefit pension plans is represented by a union, and 0 otherwise Form 5500, 8K/10K97Table 3.2: Sample SelectionThis table reports statistics for the sample of 236 bankrupt defined benefit sponsors from1987 to 2012. Panel A outlines the steps taken to construct the sample and the remainingnumber of firms in the sample after each step is applied. Pension and financial data are ob-tained from Compustat. Chapter 11 filings are identified using Lynn LoPucki’s BankruptcyResearch Database (BRD). Panel B reports the industry distribution of sample firms basedon the Fama and French 48-industry classification. The panel lists the industry name, thenumber of firms in that industry, and the percent of firms in the industry relative to the236 firms in the sample. Industries are ranked from highest representation in the sample tolowest.Panel A: Sample SelectionSampleNumber of firms from Compustat Pension Annual data from 1987 to 2012 8874with data on DB assets and liabilities 5695with all control variables on Compustat 2005filed for Chapter 11 244with data at most two years prior to Chapter 11 236Panel B: Industry AffiliationIndustry N % Industry N %Retail 37 16% Food Products 5 2%Transportation 18 8% Entertainment 5 2%Steel Works Etc 16 7% Fabricated Products 5 2%Automobiles and Trucks 16 7% Electrical Equipment 5 2%Textiles 14 6% Computers 3 1%Consumer Goods 11 5% Electronic Equipment 3 1%Chemicals 11 5% Coal 2 1%Machinery 11 5% Petroleum and Natural Gas 2 1%Apparel 9 4% Personal Services 2 1%Rubber and Plastic Products 9 4% Other 2 1%Business Services 9 4% Recreation 1 0%Business Supplies 9 4% Healthcare 1 0%Printing and Publishing 7 3% Non-Metallic Mining 1 0%Construction Materials 7 3% Measuring Equipment 1 0%Wholesale 7 3% Shipping Containers 1 0%Construction 6 3% All 236 100%98Table 3.3: Summary StatisticsThis table reports summary statistics for the main sample of 236 bankrupt defined benefit sponsors from1987 to 2012 as well as for several subsamples used in this paper. Panel A reports firm characteristics,defined benefit pension plan characteristics, and bankruptcy characteristics for the main sample of 236bankrupt defined benefit sponsors. Panel B reports summary statistics for the subsample of 107 definedbenefit sponsors for which information on the number of claim classes in bankruptcy was obtained. PanelC presents summary statistics at the plan level for the subsample of 83 defined benefit sponsors (176 plan-years) with data on Form 5500. Panel D provides summary statistics for the subsample of 32 defined benefitsponsors for which data on unsecured creditors’ recovery rates was available on Capital IQ. All variablesare defined in Table 3.1.Panel A: Main SampleN Mean StDev Min 25th Median 75th MaxFirm CharacteristicsAssets 236 1,701 4620 161.8 273.5 500.4 1,545 55,002Adjusted Assets 236 2,139 6,431 174.8 311.1 589.2 1639 78,625Leverage 236 0.446 0.342 0 0.247 0.388 0.590 3.532Additional Leverage 236 0.573 0.230 0.058 0.417 0.544 0.727 1.814Cash 236 0.035 0.041 0 0.007 0.02 0.047 0.228Cash Flow Volatility 236 0.018 0.014 0.002 0.009 0.014 0.023 0.085Asset Tangibility 236 0.252 0.156 0.015 0.132 0.232 0.348 0.808Return on Assets 236 0.040 0.061 -0.164 0.001 0.041 0.074 0.485Firm Age 236 25.35 16.51 3 10 20 41 62Employees 230 13.04 26.64 0.031 2.500 5.225 14.70 252.0Union 176 0.602 0.491 0 0 1 1 1Pension Plan CharacteristicsDB Assets/Adj. Assets 236 0.109 0.118 0 0.023 0.077 0.147 0.569DB Liab./Adj. Assets 236 0.131 0.141 0.002 0.027 0.086 0.164 0.775DB Liab./Adj. Liabilities 236 0.128 0.13 0.003 0.03 0.085 0.169 0.590(DB Liab.-DB Assets)/Adj. Liabilities 236 0.020 0.053 -0.215 0 0.008 0.028 0.474Bankruptcy CharacteristicsTerminate 172 0.581 0.495 0 0 1 1 1Prepackaged 236 0.246 0.431 0 0 0 0 1Delaware 236 0.390 0.489 0 0 0 1 1Duration 236 20.03 17.17 0.667 8.133 16.85 25.82 131.8Ln(Duration) 236 2.617 0.975 -0.405 2.096 2.824 3.251 4.882Emerge 236 0.725 0.448 0 0 1 1 1Refile 236 0.186 0.390 0 0 0 0 1Claim Classes 107 9.953 6.371 3 7 9 11 6299Panel B: Claim Classes SubsampleN Mean StDev Min 25th Perc. Median 75th Perc. MaxAdjusted Assets 107 2,754 7,996 174.8 378.4 758.1 2,360 78,625Leverage 107 0.480 0.440 0.014 0.246 0.393 0.592 3.532Additional Leverage 107 0.598 0.206 0.145 0.441 0.566 0.748 1.191Cash 107 0.035 0.040 0 0.007 0.019 0.05 0.194Cash Flow Volatility 107 0.017 0.013 0.002 0.009 0.013 0.021 0.071Asset Tangibility 107 0.253 0.140 0.015 0.146 0.239 0.355 0.729Return on Assets 107 0.045 0.069 -0.102 0.005 0.040 0.074 0.485Union 92 0.587 0.495 0 0 1 1 1DB Liabilities / Adj. Liabilities 107 0.139 0.131 0.005 0.037 0.094 0.214 0.590(DB Liab.-DB Assets)/Adj. Liabilities 107 0.023 0.048 -0.215 0.002 0.01 0.045 0.136Terminate 78 0.385 0.490 0 0 0 1 1Prepackaged 107 0.308 0.464 0 0 0 1 1Delaware 107 0.467 0.501 0 0 0 1 1Duration 107 18.97 18.95 0.667 7.300 16.07 22.53 131.8Emerge 107 0.860 0.349 0 1 1 1 1Refile 107 0.196 0.399 0 0 0 0 1Panel C: Plan Level SubsampleN Mean StDev Min 25th Perc. Median 75th Perc. MaxAverage # of Plans 90 1.911 2.882 1 1 1 2 26Plan Age 165 28.77 17.37 2 15 27 42 78Ln(DB Assets) 176 2.871 1.985 -2.398 1.607 2.559 3.973 8.164DB Assets / Adj. Assets 176 0.050 0.090 0 0.003 0.016 0.051 0.656DB Liabilities / Adj. Assets 176 0.060 0.107 0 0.004 0.019 0.060 0.740(DB Assets-DB Liab.)/DB Liab. 176 -0.068 0.327 -0.668 -0.245 -0.127 0.045 2.025DB Liabilities / Adj. Liabilities 176 0.056 0.094 0 0.004 0.018 0.060 0.502(DB Liab.-DB Assets)/Adj. Liabilities 176 0.008 0.025 -0.085 0 0.001 0.008 0.169Union 176 0.432 0.497 0 0 0 1 1Employees 176 5.780 19.70 0 0.28 1.094 4.394 216.4Active Share of Employees 176 0.324 0.294 0 0 0.302 0.512 1Firm Age 176 32.33 18.195 5 13 33 51 56Terminate 176 0.438 0.497 0 0 0 1 1Claim Classes 103 8.670 3.014 4 8 8 10 20Panel D: Recovery Rates SubsampleN Mean StDev Min 25th Perc. Median 75th Perc. MaxRecovery Rate 32 0.404 0.303 0.002 0.064 0.380 0.670 1Adjusted Assets 32 3,053 4,264 188.0 469.9 1,264 3,676 19,464Leverage 32 0.571 0.578 0.118 0.31 0.429 0.701 3.532Additional Leverage 32 0.527 0.220 0.058 0.411 0.519 0.647 1.066Asset Tangibility 32 0.267 0.167 0.015 0.126 0.282 0.395 0.631Return on Assets 32 0.057 0.102 -0.164 0.010 0.060 0.084 0.485Employees 32 14.27 21.29 0.038 2.140 4.850 16.500 80.11DB Liabilities / Adj. Liabilities 32 0.133 0.129 0.007 0.033 0.101 0.196 0.590(DB Liab.-DB Assets)/Adj. Liabilities 32 0.030 0.044 -0.046 0.003 0.014 0.046 0.127Equity 32 0.091 0.147 0.003 0.014 0.046 0.099 0.751Default Barrier 32 0.441 0.599 0.065 0.197 0.297 0.412 3.532LTD Issuance 32 0.564 0.439 0 0.021 0.811 0.988 1100Panel E: Pairwise CorrelationThis panel reports pairwise correlation coefficients among the variables used in the tests for the sample of 236 bankrupt defined benefit sponsors from 1987 to 2012. All variables are defined in Table 3.1. p-values arereported in brackets.(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17)(1) Ln(Adj. Assets) 1.00(2) Leverage -0.17 1.00(0.01)(3) Additional Leverage 0.20 -0.37 1.00(0.00) (0.00)(4) Cash 0.18 0.14 0.18 1.00(0.01) (0.03) (0.01)(5) Cash Flow Volatility -0.20 0.03 0.05 0.06 1.00(0.00) (0.68) (0.48) (0.37)(6) Asset Tangibility -0.03 0.14 -0.33 -0.10 -0.15 1.00(0.67) (0.03) (0.00) (0.12) (0.02)(7) Return on Assets 0.04 0.52 -0.22 -0.00 -0.30 0.14 1.00(0.50) (0.00) (0.00) (0.99) (0.00) (0.03)(8) Prepackaged -0.15 0.30 -0.20 0.09 -0.09 -0.03 0.21 1.00(0.02) (0.00) (0.00) (0.18) (0.18) (0.64) (0.00)(9) Delaware -0.01 0.03 -0.10 -0.04 -0.05 -0.01 0.10 0.09 1.00(0.84) (0.69) (0.14) (0.51) (0.45) (0.87) (0.13) (0.17)(10) DB Liabilities / Adj. Liabilities 0.17 -0.30 0.46 -0.04 -0.14 -0.09 -0.15 -0.15 -0.13 1.00(0.01) (0.00) (0.00) (0.50) (0.03) (0.19) (0.02) (0.02) (0.05)(11) (DB Liab.-DB Assets)/Adj. Liabilities 0.01 -0.07 0.30 -0.01 -0.08 0.07 -0.09 -0.02 -0.16 0.41 1.00(0.84) (0.27) (0.00) (0.88) (0.23) (0.30) (0.17) (0.81) (0.01) (0.00)(12) Union 0.18 0.06 -0.05 -0.05 -0.01 0.17 0.05 -0.01 -0.08 0.03 0.19 1.00(0.02) (0.43) (0.55) (0.53) (0.85) (0.03) (0.53) (0.92) (0.26) (0.67) (0.01)(13) Claim Classes -0.00 -0.02 -0.10 -0.00 0.03 0.25 -0.06 -0.10 0.06 -0.10 -0.05 -0.14 1.00(0.99) (0.83) (0.30) (0.99) (0.77) (0.01) (0.51) (0.30) (0.54) (0.33) (0.62) (0.19)(14) Duration 0.20 -0.26 0.15 -0.04 0.03 -0.01 -0.01 -0.46 -0.03 0.12 -0.12 -0.05 0.09 1.00(0.00) (0.00) (0.02) (0.57) (0.70) (0.86) (0.87) (0.00) (0.61) (0.08) (0.06) (0.52) (0.35)(15) Emerge 0.11 0.25 -0.09 0.07 -0.10 0.11 0.13 0.31 0.01 0.00 0.12 0.17 0.15 -0.15 1.00(0.09) (0.00) (0.19) (0.26) (0.11) (0.08) (0.05) (0.00) (0.92) (0.98) (0.07) (0.02) (0.12) (0.02)(16) Terminate -0.24 -0.29 0.11 -0.04 0.22 -0.06 -0.23 -0.22 -0.10 0.05 0.05 -0.16 -0.19 0.09 -0.61 1.00(0.00) (0.00) (0.15) (0.65) (0.00) (0.47) (0.00) (0.00) (0.21) (0.48) (0.51) (0.06) (0.10) (0.23) (0.00)(17) Refile -0.11 0.06 -0.01 -0.05 -0.03 0.05 0.04 0.18 -0.03 -0.13 -0.05 0.04 -0.01 -0.09 0.27 -0.26 1.00(0.09) (0.36) (0.83) (0.41) (0.61) (0.48) (0.54) (0.01) (0.69) (0.06) (0.49) (0.56) (0.94) (0.16) (0.00) (0.00)101Table 3.4: Changes in Firm Variables- Before vs. After BankruptcyThis table presents univariate comparisons of firm and plan characteristics betweenthe last fiscal prior to the bankruptcy filing and the first fiscal year after bankruptcyemergence. The results are based on 64 defined benefit sponsors with informationin both years. All univariate comparisons are based on median values. Reported p-values for significance of differences are based on Wilcoxon two-sample tests. Allvariables are defined in Table 3.1.T-1 T+1 z-stat p-valueFirm CharacteristicsLn(Assets) 7.388 6.845 1.78 0.0759Leverage 0.481 0.261 5.21 0.0000Additional Leverage 0.505 0.554 0.46 0.6490Cash 0.024 0.044 2.44 0.0148Cash Flow Volatility 0.012 0.012 0.04 0.9715Asset Tangibility 0.252 0.211 1.48 0.1402Return on Assets 0.063 0.063 1.45 0.1467Plan CharacteristicsDB Liabilities 119.2 128.0 0.14 0.8920DB Liabilities/Adj. Assets 0.104 0.143 1.58 0.1153DB Assets 90.56 101.8 0.42 0.6732DB Assets/Adj. Assets 0.094 0.108 0.99 0.3204DB Liabilities/Adj. Liab. 0.101 0.163 3.22 0.0013(DB Liab.-DB Assets)/Adj. Liab. 0.013 0.029 2.92 0.0035102Table 3.5: Determinants of Bankruptcy DurationThis table reports results from cross-sectional regressions of bankruptcy duration with industry and year fixed effects.The sample includes 236 defined benefit sponsors from 1987 to 2012, as well as two subsamples of sponsors with unionand claim classification data. All variables are defined in Table 3.1. t-statistics are reported in parentheses and arestatistically significant at the 1%(***), 5%(**), and 10%(*) levels. Standard errors are robust and are clustered at theindustry level.Dependent Variable: Duration (1) (2) (3) (4) (5) (6) (7)DB Liabilities/Adj. Liabilities 0.50 -0.06 0.17(0.63) (-0.05) (0.19)(DB Liab.-DB Assets)/Adj.Liabilites 1.23 3.92 -2.62(1.06) (1.63) (-0.89)Ln(Adj. Assets) 0.08* 0.07 0.15** 0.10 0.08* 0.15** 0.11(1.91) (1.59) (2.04) (0.76) (1.79) (2.23) (0.80)Leverage -0.67*** -0.63*** -0.89*** -0.89* -0.67*** -0.95*** -0.83*(-3.31) (-3.02) (-3.04) (-1.79) (-3.24) (-3.52) (-1.96)Additional Leverage -0.13 -0.23 -0.09 0.45 -0.24 -0.48 0.83(-0.40) (-0.66) (-0.16) (0.49) (-0.70) (-0.86) (0.70)Cash 0.73 0.92 1.03 -0.79 0.84 1.32 -1.38(0.59) (0.73) (0.63) (-0.39) (0.65) (0.85) (-0.60)Cash Flow Volatility 6.07** 6.12** 10.48* -1.57 6.19** 12.55** -5.41(2.34) (2.36) (1.86) (-0.12) (2.35) (2.06) (-0.35)Asset Tangibility 0.34 0.34 1.00 0.12 0.27 0.74 0.27(0.66) (0.68) (1.31) (0.15) (0.55) (1.03) (0.30)Return on Assets 2.42** 2.40** 3.07*** 4.16 2.44** 3.44*** 3.69(2.45) (2.49) (2.82) (1.35) (2.51) (3.52) (1.34)Prepackaged -1.45*** -1.44*** -1.28*** -1.50*** -1.47*** -1.36*** -1.40***(-7.33) (-7.30) (-5.37) (-5.21) (-7.34) (-5.63) (-4.55)Delaware -0.30*** -0.29*** -0.36** -0.18 -0.29** -0.32** -0.22(-2.84) (-2.72) (-2.45) (-0.72) (-2.72) (-2.15) (-1.09)Union -0.23** -0.27**(-2.04) (-2.42)Claim Classes -0.01 -0.01(-0.66) (-0.51)Intercept 2.72*** 2.74*** 2.16*** -0.27 2.82*** 2.34*** -0.43(4.51) (4.47) (2.75) (-0.20) (4.38) (3.18) (-0.33)Industry FE Yes Yes Yes Yes Yes Yes YesYear FE Yes Yes Yes Yes Yes Yes YesModel LPM LPM LPM LPM LPM LPM LPMObservations 236 236 176 107 236 176 107R-squared 0.68 0.68 0.71 0.82 0.68 0.72 0.83103Table 3.6: Likelihood to Emerge from BankruptcyThis table reports results from cross-sectional regressions of the likelihood of emerging from bankruptcy with in-dustry and year fixed effects. The sample includes 236 defined benefit sponsors from 1987 to 2012, as well as twosubsamples of sponsors with union and claim classification data. All variables are defined in Table 3.1. t-statisticsare reported in parentheses and are statistically significant at the 1%(***), 5%(**), and 10%(*) levels. Standarderrors are robust and are clustered at the industry level.Dependent Variable: Emerge (1) (2) (3) (4) (5) (6) (7)DB Liabilities/Adj. Liabilities -0.02 0.09 0.22(-0.04) (0.17) (0.39)(DB Liab.-DB Assets)/Adj. Liabilities 0.38 0.33 0.07(0.59) (0.23) (0.06)Ln(Adj. Assets) 0.09*** 0.09*** 0.08** 0.09* 0.09*** 0.09** 0.09*(2.82) (3.07) (2.33) (1.76) (2.79) (2.14) (1.76)Leverage 0.31** 0.31* 0.23 0.17 0.31** 0.22 0.17(2.22) (1.92) (1.23) (1.25) (2.24) (1.32) (1.16)Additional Leverage 0.09 0.10 0.03 0.01 0.06 0.02 0.00(0.61) (0.56) (0.15) (0.05) (0.32) (0.09) (0.01)Cash -0.84 -0.85 -1.21 -0.66 -0.81 -1.23 -0.65(-0.74) (-0.78) (-0.81) (-0.66) (-0.69) (-0.75) (-0.55)Cash Flow Volatility -1.10 -1.10 -0.01 4.50 -1.06 0.16 4.60(-0.52) (-0.52) (-0.00) (0.69) (-0.50) (0.05) (0.65)Asset Tangibility 0.16 0.16 -0.17 0.35 0.14 -0.19 0.35(0.55) (0.54) (-0.57) (0.52) (0.46) (-0.59) (0.53)Return on Assets -0.84 -0.84 0.01 1.04 -0.83 0.06 1.05(-0.99) (-0.98) (0.01) (0.71) (-0.99) (0.06) (0.69)Prepackaged 0.31*** 0.31*** 0.30*** 0.26 0.30*** 0.30*** 0.26(3.60) (3.58) (3.26) (1.60) (3.30) (2.81) (1.66)Delaware 0.04 0.04 0.06 0.17*** 0.05 0.06 0.18**(0.65) (0.63) (0.88) (3.07) (0.71) (0.92) (2.73)Union 0.14* 0.14*(1.80) (1.69)Claim Classes 0.01 0.01(1.11) (1.13)Intercept 0.22 0.22 0.29 0.12 0.25 0.30 0.13(0.66) (0.67) (0.92) (0.16) (0.70) (0.80) (0.17)Industry FE Yes Yes Yes Yes Yes Yes YesYear FE Yes Yes Yes Yes Yes Yes YesModel LPM LPM LPM LPM LPM LPM LPMObservations 236 236 176 107 236 176 107R-squared 0.41 0.41 0.46 0.57 0.41 0.46 0.57104Table 3.7: Distressed Pension Plan TerminationsThis table reports results from cross-sectional regressions of the likelihood to terminate a defined benefit pension planin bankruptcy with year fixed effects. All variables are defined in Table 3.1. t-statistics are reported in parentheses andare statistically significant at the 1%(***), 5%(**), and 10%(*) levels. Standard errors are robust and are clustered atthe industry level.Panel A: Likelihood of Defined Benefit Pension Plan Termination at the Plan LevelThis panel reports results for the likelihood of defined benefit plan termination in bankruptcy using plan-level data fromForm 5500. The sample includes 176 plan-year observations for 83 defined benefit sponsors from 1990 to 2007, as wellas two subsamples of sponsors with union and claim classification data.Dependent Variable: Terminate (1) (2) (3) (4) (5) (6) (7)DB Liabilities/Adj. Liabilities 1.04** 1.05** 0.89**(2.18) (2.19) (2.31)(DB Liab.-DB Assets)/Adj. Liabilities 2.35 2.17 1.67(1.40) (1.23) (0.73)(DB Liab.-DB Assets)/DB Liab. -0.01 -0.02 0.00 -0.20 0.05 0.07 -0.13(-0.07) (-0.21) (0.04) (-1.36) (0.49) (0.62) (-1.20)Active share of participants -0.01 0.03 0.02 0.01 0.01 -0.00 -0.01(-0.06) (0.28) (0.16) (0.12) (0.09) (-0.01) (-0.07)DB Assets 0.00 0.00 0.00 0.00* 0.00 0.00 0.00(1.13) (1.62) (1.45) (1.68) (1.07) (0.93) (1.13)Ln(DB Assets) 0.00 -0.03 -0.03 -0.04* -0.00 -0.01 -0.02(0.09) (-1.30) (-1.50) (-1.74) (-0.23) (-0.48) (-1.07)Leverage -0.53*** -0.50*** -0.46*** -0.02 -0.57*** -0.52*** -0.09(-3.23) (-3.10) (-2.73) (-0.18) (-3.50) (-3.10) (-0.85)Additional Leverage 0.07 -0.00 0.06 0.86*** -0.04 0.03 0.86***(0.28) (-0.00) (0.24) (3.65) (-0.14) (0.13) (3.28)Union 0.13 0.10(1.24) (0.99)Claim Classes -0.01 -0.01(-0.81) (-0.76)Intercept 0.33* 0.26 0.09 0.89*** 0.36* 0.22 0.95***(1.75) (1.38) (0.43) (6.36) (1.85) (0.95) (5.68)Year FE Yes Yes Yes Yes Yes Yes YesModel LPM LPM LPM LPM LPM LPM LPMObservations 176 176 172 103 176 172 103Firm-years 91 91 88 45 91 88 48R-squared 0.51 0.53 0.55 0.77 0.52 0.54 0.75105Panel B: Likelihood of Defined Benefit Pension Plan Termination at the Firm LevelThis panel presents results for the likelihood of defined benefit plan termination in bankruptcy using firm-level datafrom Compustat. The sample consists of 171 firm-level observations for defined benefit sponsors from 1987 to 2012for which information on plan termination was obtained, as well as two subsamples of sponsors with union and claimclassification data.Dependent Variable: Terminate (1) (2) (3) (4) (5) (6) (7)DB Liabilities/Adj. Liabilities 1.27*** 1.24** 0.33(3.43) (2.47) (0.47)(DB Liab.-DB Assets)/Adj. Liabilities 1.34* 2.33 0.55(1.73) (1.44) (0.30)(DB Liab.-DB Assets)/DB Liab. -0.06 0.06 -0.04 0.03 0.10 0.09 0.07(-0.44) (0.40) (-0.18) (0.11) (0.56) (0.29) (0.24)DB Assets 0.00 0.00 0.00* 0.00 0.00 0.00 0.00(1.30) (1.60) (1.76) (1.55) (1.25) (1.40) (1.49)Ln(DB Assets) -0.11*** -0.17*** -0.16*** -0.13*** -0.12*** -0.11*** -0.12***(-4.04) (-5.22) (-3.78) (-2.76) (-4.14) (-2.96) (-3.06)Leverage -0.71*** -0.60*** -0.68*** -0.37** -0.71*** -0.80*** -0.39**(-4.76) (-4.07) (-3.93) (-2.17) (-4.95) (-4.84) (-2.14)Additional Leverage 0.38* 0.24 0.21 0.99*** 0.25 0.14 0.99***(1.70) (1.08) (0.80) (3.27) (1.06) (0.44) (2.87)Union -0.04 -0.07(-0.44) (-0.75)Claim Classes -0.01 -0.01(-0.87) (-0.83)Intercept 0.83*** 1.05*** 1.05*** 1.34*** 0.92*** 0.97*** 1.34***(4.78) (5.65) (4.38) (6.37) (4.89) (3.68) (5.72)Year FE Yes Yes Yes Yes Yes Yes YesModel LPM LPM LPM LPM LPM LPM LPMObservations 171 171 139 78 171 139 78R-squared 0.35 0.41 0.42 0.57 0.36 0.39 0.56106Table 3.8: Likelihood to Refile for Chapter 11This table reports results from cross-sectional regressions of the likelihood to refile for bankruptcy. The sample includes 64 defined benefit sponsorsfrom 1987 to 2012 with information in the year prior to bankruptcy and in the year after emergence, as well as a subsample of 50 sponsors withunionization data. All variables are defined in Table 3.1. t-statistics are reported in parentheses and are statistically significant at the 1%(***),5%(**), and 10%(*) levels. Standard errors are robust and are clustered at the industry level.Dependent Variable: Re f ile (1) (2) (3) (4) (5) (6) (7) (8) (9)DB Liabilities/Adj. Liabilities -1.46*** -1.47*** -1.57*** -1.52***(-3.00) (-3.69) (-3.07) (-3.54)Change in DB Liab. / Adj. Assets -1.06 -0.58(-0.94) (-0.41)(DB Liab.-DB Assets) / Adj. Liabilities -3.70*** -4.28** -5.12*** -5.67***(-3.09) (-2.68) (-4.49) (-2.98)Change in (DB Liab.-DB Assets)/Adj. Assets -3.90** -4.42**(-2.58) (-2.60)Ln(Adj. Assets) -0.08 -0.05 -0.07 -0.06 -0.07 -0.06 -0.07 -0.04 -0.02(-1.69) (-1.15) (-1.03) (-1.40) (-1.11) (-1.10) (-0.99) (-0.69) (-0.30)Change in Ln(Adj. Assets) -0.14 -0.16 -0.40 -0.13 -0.37 -0.11 -0.29 -0.20 -0.45(-0.64) (-0.73) (-1.34) (-0.62) (-1.19) (-0.51) (-0.88) (-1.00) (-1.48)Leverage -0.64 -0.64 0.07 -0.73 -0.00 -0.57 0.09 -0.42 0.35(-1.12) (-1.16) (0.10) (-1.25) (-0.00) (-1.14) (0.14) (-0.92) (0.58)Change in Leverage -0.44 -0.45 0.36 -0.50 0.30 -0.36 0.40 -0.23 0.57(-0.77) (-0.81) (0.46) (-0.86) (0.35) (-0.71) (0.56) (-0.52) (0.91)Additional Leverage 0.50 0.79** 1.39** 0.79** 1.38** 0.69 1.31** 0.81* 1.45**(1.12) (2.27) (2.61) (2.15) (2.50) (1.45) (2.25) (1.79) (2.49)Change in Additional Leverage 0.78* 1.02** 1.07 1.25** 1.21* 0.76* 0.84 1.12** 1.43**(1.89) (2.06) (1.66) (2.28) (1.79) (1.83) (1.43) (2.55) (2.42)Cash -0.35 -0.55 -1.51 -0.42 -1.38 -0.49 -1.28 -0.69 -1.77(-0.36) (-0.64) (-1.40) (-0.48) (-1.21) (-0.51) (-1.21) (-0.71) (-1.47)Cash Flow Volatility -6.10** -6.90** -16.80** -7.16** -17.22** -6.30** -16.77** -7.51** -18.83***(-2.22) (-2.17) (-2.58) (-2.16) (-2.66) (-2.29) (-2.85) (-2.69) (-3.36)Asset Tangibility 0.27 0.25 -0.12 0.25 -0.10 0.31 -0.09 0.32 -0.18(0.68) (0.67) (-0.25) (0.66) (-0.20) (0.72) (-0.16) (0.70) (-0.32)Return on Assets 2.09 1.62 2.02 1.80 2.12 1.64 1.82 1.70 1.74(1.47) (1.28) (1.24) (1.37) (1.26) (1.18) (0.95) (1.13) (0.82)Duration -0.00 0.00 -0.00 0.00 -0.00 -0.00 -0.01 -0.00 -0.00(-0.15) (0.17) (-0.78) (0.36) (-0.68) (-0.45) (-1.32) (-0.31) (-1.00)Union 0.13 0.14 0.20 0.17(0.76) (0.78) (1.08) (0.88)Intercept 0.74 0.61 0.41 0.72* 0.49 0.60 0.37 0.40 0.00(1.56) (1.49) (0.83) (1.75) (0.92) (1.08) (0.57) (0.78) (0.01)Model LPM LPM LPM LPM LPM LPM LPM LPM LPMObservations 64 64 50 64 50 64 50 64 50R-squared 0.14 0.22 0.29 0.23 0.29 0.19 0.27 0.25 0.35107Table 3.9: Determinants of Unsecured Creditors’ Recovery RatesThis table reports results from cross-sectional regressions of unsecuredcreditors’ recovery rates. The sample includes 32 defined benefit spon-sors from 2006 to 2012 with data on creditor recovery rates from Cap-ital IQ. All variables are defined in Table 3.1. t-statistics are reportedin parentheses and are statistically significant at the 1%(***), 5%(**),and 10%(*) levels. Standard errors are robust and are clustered at theindustry level.Dependent Variable: Recovery (1) (2) (3)DB Liabilities/Adj. Liabilities 0.48**(2.01)(DB Liab.-DB Assets)/Adj. Liabilities 1.24(1.07)Equity -0.78** -0.70** -0.69*(-2.64) (-2.42) (-1.91)Default barrier -0.24 -0.18 -0.20(-1.70) (-1.27) (-1.18)LTD Issuance -0.08 -0.04 -0.04(-0.65) (-0.35) (-0.26)Asset Tangibility 0.01 0.11 0.01(0.07) (0.53) (0.07)Profitability 0.20 0.02 0.06(0.29) (0.03) (0.09)Ln(Adj. Assets) 0.08 0.07 0.08(1.23) (1.18) (1.25)Ln(Employees) -0.04 -0.05 -0.05(-0.82) (-0.94) (-0.94)Constant 0.10 0.05 0.01(0.21) (0.12) (0.02)Model LPM LPM LPMObservations 32 32 32R-squared 0.28 0.31 0.30108Table 3.10: Changes in Pension Obligations in BankruptcyThis table reports results from cross-sectional regressions of the determinants of changes in definedbenefit pension obligations. Changes in pension obligations are measured as the difference betweenpension obligations at bankruptcy emergence and pension obligations prior to filing for bankruptcy,scaled by adjusted assets in the year prior to bankruptcy. The sample includes 62 defined benefitsponsors from 1987 to 2012 with information in the year prior to bankruptcy and in the year afteremergence, as well as a subsamples of sponsors with data on plan terminations. All variables aredefined in Table 3.1. t-statistics are reported in parentheses and are statistically significant at the1%(***), 5%(**), and 10%(*) levels. Standard errors are robust and are clustered at the industrylevel.Dependent Variable: ∆PBO (1) (2) (3) (4) (5) (6)DB Liabilities/Adj. Liabilities -0.10 -0.12(-1.01) (-0.98)(DB Liab.-DB Assets)/Adj. Liabilities -0.68*** -0.75***(-3.30) (-2.75)Ln(Adj. Assets) -0.02 -0.02 -0.01 -0.02 -0.02 -0.01(-1.09) (-1.02) (-0.90) (-0.75) (-0.65) (-0.50)Leverage 0.04 0.04 0.04* 0.06 0.06 0.04(1.37) (1.38) (1.74) (0.93) (0.89) (0.71)Change in Leverage 0.05 0.05 0.06* 0.07 0.08 0.08(1.28) (1.31) (1.99) (0.87) (0.91) (1.23)Employees 0.00 0.00 0.00 0.00 0.00 0.00(1.49) (1.45) (1.69) (1.53) (1.49) (1.65)Change in Employees 0.00 0.00 0.00* 0.00 0.00 0.00*(1.56) (1.61) (1.91) (1.55) (1.72) (1.96)Terminate -0.05 -0.05 -0.05(-1.01) (-1.03) (-1.11)Intercept 0.13 0.12 0.11 0.12 0.13 0.11(1.08) (1.07) (1.04) (0.71) (0.69) (0.65)Model LPM LPM LPM LPM LPM LPMObservations 62 62 62 39 39 39R-squared 0.11 0.12 0.19 0.17 0.19 0.25109Chapter 4ConclusionIn this thesis, I examine the role of defined benefit claimants in Chapter 11bankruptcy at two time periods: prior to bankruptcy and in Chapter 11 reorga-nization. In the first essay, I find a negative relationship between corporate definedbenefit pension claimants and the likelihood of bankruptcy. In the second essay, Idocument a role for defined benefit claimants beyond that of traditional creditorsin various aspects of the bankruptcy restructuring process.Both essays in this thesis are motivated by the similarities and differences be-tween defined benefit claimants and other firm lenders. Defined benefit obligationsresemble financial liabilities because firms are liable for both types of debt and thepromised payouts for either obligation do not depend on the financial performanceof the underlying assets. Moreover, pension liabilities require regular contribu-tions akin to interest payments which are tax-deductible and can trigger bankruptcyif missed. At the same time, defined benefit obligations differ from other corpo-rate liabilities because they are reported off the balance sheet, they are guaranteedby the government and they are held by claimants who are less diversified thantraditional creditors. These characteristics give pension claimants an incentive toact differently from traditional firm creditors, especially in critical times such asdefault. This thesis is aimed at understanding the link between defined benefitclaimants as firm lenders and Chapter 11 bankruptcy.Prior to bankruptcy, I document that defined benefit claimants are an importantdeterminant of the firm’s decision to file for Chapter 11. Since firms’ financialhealth is one of the key drivers of bankruptcy filings, I control for overall firm in-debtedness by accounting for all balance sheet obligations and the two largest off-balance sheet liabilities, defined benefit obligations and operating leases. I proxy110for the role of defined benefit claimants with the ratio of defined benefit obligationsto overall liabilities. Holding the level of debt constant, I find that defined benefitclaimants are negatively related to the likelihood of bankruptcy. This result is eco-nomically important. Given the sample average likelihood of bankruptcy of 50%,an increase in defined benefit liabilities relative to other liabilities is associatedwith a 6% decrease in the probability of bankruptcy. Moreover, the results extendto different samples and are robust to alternative measures of pension obligationsand the role of defined benefit claimants.The first essay’s findings are significant as they highlight a need to account fordefined benefit claimants in studies of negotiations between the firm and its credi-tors. Pension claimants are firm lenders with different wealth, diversification, andcharacteristics from financial lenders and little is known about how defined benefitclaimants negotiate with the firm and how they influence bargaining with the firm’sother lenders. While there have been important advances in the literature in deter-mining the role of pension claimants in other corporate decisions and firm charac-teristics, defined benefit claimants’ impact on negotiations prior to bankruptcy hasremained largely unexplored. The first essay in this thesis attempts to address thisquestion. The sample in this study is designed with identification in mind so thedecision to sponsor a defined benefit plan is taken for granted and some of the mainbankruptcy determinants are accounted for through matching bankrupt sponsors tosimilar solvent sponsors. While I find that defined benefit claimants are related to alower likelihood of bankruptcy, the study remains silent on the private negotiationsthat ensue between pension claimants and the firm. Thus, further insight could begained from studying the private negotiations with defined benefit claimants priorto Chapter 11 to better understand the exact mechanisms through which pensionclaimants impact the firm’s decision to file for Chapter 11.In bankruptcy reorganization, I find that defined benefit claimants play a rolebeyond that of the firm’s other lenders. Therefore, the differences between de-fined benefit claimants and the firm’s traditional creditors are relevant during therestructuring process as well. I provide evidence that defined benefit claimantsare associated with a higher likelihood of pension plan terminations in bankruptcy.Such a relationship points to plan terminations as one loss that defined benefitclaimants incur in bankruptcy. In expectation of the losses they will have to in-111cur in bankruptcy, defined benefit claimants have high incentives to negotiate withthe firm prior to Chapter 11, as argued in the first essay. Furthermore, I find thatdefined benefit claimants are associated with higher recovery rates for unsecuredcreditors and a higher probability of firm survival post-reorganization. The factthat defined benefit claimants are less likely to refile for Chapter 11 after emergingfrom bankruptcy provides further support for the first essay’s findings of a negativerelationship between pension claimants and the probability of bankruptcy. Thus,the results from the second essay reinforce the first paper’s claims. In the secondessay, I also identify defined benefit claimants’ willingness to accept pension ben-efit cuts as one potential channel through which pension claimants influence thebankruptcy reorganization process. After controlling for the expected reduction inliabilities for all creditors due to bankruptcy, I show that the relative role of definedbenefit claimants explains the majority of the variation in these benefit cuts. Thismechanism is only at work for unfunded pension obligations, which suggests thatdefined benefit claimants give up the most when they stand to lose the most.Altogether, the second essay highlights a role for defined benefit claimants inChapter 11 reorganization beyond that of traditional creditors. These findings arerelevant because defined benefit claimants are firm creditors who differ from tradi-tional lenders and may act differently from other creditors. Prior research has doc-umented such a role for banks, private equity funds and hedge funds but definedbenefit claimants differ from these lenders as well because of the human capitalinvestment in the firm and the lack of diversification unique to pension claimants.As members of a creditors’ committee in bankruptcy, defined benefit claimants canvote on the firm’s reorganization plan and may thus be important players in Chapter11. The second essay documents several specific bankruptcy outcomes that definedbenefit claimants influence. Furthermore, Benmelech et al. (2012) show that airlinemanagers use the threat of pension plan termination to extract wage concessions innegotiations with labor. The third chapter in this thesis confirms that defined bene-fit plan termination is a threat that pension claimants face and a loss that they mayhave to agree to in bankruptcy. However, my sample includes numerous industriesand hence indicates that the threat of plan termination is an important considera-tion for pension claimants across a wide range of industries and that negotiationsbetween pension claimants and the firm in all sectors of the economy influence112the Chapter 11 reorganization process. While my study expands the applicabilityof Benmelech et al. (2012)’s results, the level of detail that the authors are ableto obtain for the airline industry is not available for all industries in this essay’ssample.Overall, the results in this thesis uncover a role for defined benefit claimants asfirm lenders both prior to and in bankruptcy. While my essays provide empiricalsupport for the influence of defined benefit claimants in bankruptcy, a theoreticalmodel that captures defined benefit claimants as a unique lender in bankruptcy willbenefit our understanding of the bargaining among creditors and the role of de-fined benefit claimants in negotiations with the firm and other lenders. Moreover,this thesis identifies benefit concessions as one potential channel through whichdefined benefit claimants influence the bankruptcy restructuring. However, thismechanism only holds for unfunded pension obligations. Given the results for planterminations and recovery rates in which the total pension obligation matters whilethe unfunded portion does not, additional channels through which defined benefitclaimants influence the restructuring process can be investigated. A comprehensivestudy of these additional channels represents an interesting direction for further in-vestigation. Last, the second essay documents a role for labor unions in explainingsome of the outcomes of Chapter 11 bankruptcy. Even further, the results show thatunions provide explanatory power for outcomes which defined benefit claimants donot influence, while pension claimants explain bankruptcy outcomes that unionsdo not appear to impact. Thus, the interplay between unions and defined benefitclaimants and their respective negotiations with the firm provide a fruitful area forfuture research.113BibliographyE. I. Altman. Financial ratios, discriminant analysis and the prediction of corporatebankruptcy. The Journal of Finance, 23(4):589–609, 1968.R. J. Arnott and M. Gersovitz. Corporate financial structure and the funding ofprivate pension plans. Journal of Public Economics, 13(2):231–247, 1980.K. Ayotte and D. A. Skeel. Why do distressed companies choose delaware? anempirical analysis of venue choice in bankruptcy. Working Paper, University ofPennsylvania Law School, 2004.J. Begley, S. Chamberlain, S. Yang, and J. Zhang. Ceo incentives and the health ofdefined benefit pension plans. Review of Accounting Studies forthcoming, 2015.E. Benmelech, N. Bergman, and R. Enriquez. Negotiating with labor under finan-cial distress. Review of Corporate Finance Studies, 1(1):28–67, 2012.D. Bergstresser, M. Desai, and J. Rauh. Earnings manipulation, pension assump-tions, and managerial investment decisions. The Quarterly Journal of Eco-nomics, 121(1):157–195, 2006.S. T. Bharath, V. Panchapegesan, and I. Werner. The changing nature of chapter11. Working Paper, Arizona State University, 2009.F. Black. The tax consequences of long-run pension policy. Financial AnalystsJournal, 36(4):21–28, 1980.D. M. Bryan, S. L. Tiras, and C. M. Wheatley. The interaction of solvency withliquidity and its association with bankruptcy emergence. Journal of BusinessFinance & Accounting, 29(7-8):935–965, 2002.J. I. Bulow and J. B. Shoven. The bankruptcy decision. The Bell Journal of Eco-nomics, 9(2):437–456, 1978.S. V. Campbell. Predicting bankruptcy reorganization for closely held firms. Ac-counting Horizons, 10:12–25, 1996.T. J. Carroll and G. Niehaus. Pension plan funding and corporate debt ratings.Journal of Risk and Insurance, 65(3):427–441, 1998.114X. Chang, J. Kang, and W. Zhang. Corporate pension funding status and the marketfor corporate control: The disciplinary role of pension deficits in mergers andacquisitions. Working Paper, Nanyang Technological University, 2011.S. Chava and R. A. Jarrow. Bankruptcy prediction with industry effects. Review ofFinance, 8(4):537–569, 2004.H. Chen, M. Kacperczyk, and H. Ortiz-Molina. Labor unions, operating flexibility,and the cost of equity. Journal of Financial and Quantitative Analysis, 46(1):25–58, 2011.J. F. Cocco and P. F. Volpin. Corporate pension plans as takeover deterrents. Work-ing Paper, London Business School, 2012.P. Colla, F. Ippolito, and K. Li. Debt specialization. The Journal of Finance, 68(5):2117–2141, 2013.S. Dahiya, K. John, M. Puri, and G. Ramirez. Debtor-in-possession financing andbankruptcy resolution: Empirical evidence. Journal of Financial Economics, 69(1):259–280, 2003.D. K. Denis and K. J. Rodgers. Duration, outcome, and post-reorganization perfor-mance. Journal of Financial and Quantitative Analysis, 42(1):101–118, 2007.Y. Duan, E. S. Hotchkiss, and Y. Jiao. Corporate pensions and financial distress.Working Paper, University of Alberta, 2013.A. L. Eisfeldt and A. A. Rampini. Leasing, ability to repossess, and debt capacity.Review of Financial Studies, 22(4):1621–1657, 2009.F. Franzoni and J. Marin. Pension plan funding and stock market efficiency. TheJournal of Finance, 61(2):921–956, 2006.S. C. Gilson, K. John, and L. H. Lang. Troubled debt restructurings: An empir-ical study of private reorganization of firms in default. Journal of FinancialEconomics, 27(2):315–353, 1990.J. R. Graham, H. Kim, S. Li, and J. Qiu. Human capital loss in corporatebankruptcy. Working Paper, Duke University, 2013.E. Hotchkiss. Investment Decisions under Chapter 11 Bankruptcy. PhD thesis,New York University, 1993.E. Hotchkiss, D. Smith, and P. Stro¨mberg. Private equity and the resolution offinancial distress. In NBER Conference on Systemic Financial Risk, 2012.115E. S. Hotchkiss and R. M. Mooradian. Vulture investors and the market for controlof distressed firms. Journal of Financial Economics, 43(3):401–432, 1997.R. A. Ippolito. The economic function of underfunded pension plans. Journal ofLaw and Economics, 28(3):611–651, 1985a.R. A. Ippolito. The labor contract and true economic pension liabilities. TheAmerican Economic Review, 75(5):1031–1043, 1985b.R. A. Ippolito. Bankruptcy and workers: Risks, compensation and pension con-tracts. Washington University Law Quarterly, 82(4):1251–1304, 2004.V. Ivashina, B. C. Iverson, and D. C. Smith. The ownership and trading of debtclaims in chapter 11 restructurings. Working Paper, University of Virginia, 2013.C. James. When do banks take equity in debt restructurings? Review of FinancialStudies, 8(4):1209–1234, 1995.R. Jankowitsch, F. Nagler, and M. G. Subrahmanyam. The determinants of recov-ery rates in the us corporate bond market. Journal of Financial Economics, 114(1):155–177, 2014.W. Jiang, K. Li, and W. Wang. Hedge funds and chapter 11. The Journal ofFinance, 67(2):513–560, 2012.L. Jin, R. Merton, and Z. Bodie. Do a firm’s equity returns reflect the risk of itspension plan? Journal of Financial Economics, 81(1):1–26, 2006.D. R. Korobkin. Employee interests in bankruptcy. American Bankruptcy InstituteLaw Review, 4(5):5–34, 1996.Y. Lin, S. Liu, and J. Yu. Pension aasset and liabilities, cost of debt, and debtmaturity. Working Paper, University of Nebraska-Lincoln, 2014.A. W. Lo. Logit versus discriminant analysis: A specification test and applicationto corporate bankruptcies. Journal of Econometrics, 31(2):151–178, 1986.O. S. Mitchell and S. P. Utkus. The role of company stock in defined contributionplans. Working Paper, National Bureau of Economic Research, 2002.H. F. Mittelstaedt and P. R. Regier. The market response to pension plan termina-tions. Accounting Review, 68(1):1–27, 1993.W. N. Moulton and H. Thomas. Bankruptcy as a deliberate strategy: Theoreticalconsiderations and empirical evidence. Strategic Management Journal, 14(2):125–135, 1993.116J. Neyman and E. L. Scott. Consistent estimates based on partially consistentobservations. Econometrica: Journal of the Econometric Society, 16(1):1–32,1948.C. A. Olson. Do workers accept lower wages in exchange for health benefits?Journal of Labor Economics, 20(S2):S91–S114, 2002.M. A. Petersen. Pension reversions and worker-stockholder wealth transfers. TheQuarterly Journal of Economics, 107(3):1033–1056, 1992.J. M. Poterba. Employer stock and 401 (k) plans. American Economic Review, 93(2):398–404, 2003.A. A. Rampini and S. Viswanathan. Collateral and capital structure. Journal ofFinancial Economics, 109(2):466–492, 2013.J. Rauh. Investment and financing constraints: Evidence from the funding of cor-porate pension plans. The Journal of Finance, 61(1):33–71, 2006.J. Rauh. Risk shifting versus risk management: Investment policy in corporatepension plans. Review of Financial Studies, 22(7):2687–2733, 2009.J. D. Rauh and A. Sufi. Explaining corporate capital structure: Product markets,leases, and asset similarity. Review of Finance, 16(1):115–155, 2012.J. D. Rauh, I. Stefanescu, and S. P. Zeldes. Cost shifting and the freezing of corpo-rate pension plans. Working Paper, Stanford University, 2012.A. Shivdasani and I. Stefanescu. How do pensions affect corporate capital structuredecisions? Review of Financial Studies, 23(3):1287–1323, 2010.T. Shumway. Forecasting bankruptcy more accurately: A simple hazard model.The Journal of Business, 74(1):101–124, 2001.M. Stone. A financing explanation for overfunded pension plan terminations. Jour-nal of Accounting Research, 25(2):317–326, 1987.I. Tepper. Taxation and corporate pension policy. The Journal of Finance, 36(1):1–13, 1981.I. Tepper and A. Affleck. Pension plan liabilities and corporate financial strategies.The Journal of Finance, 29(5):1549–1564, 1974.J. K. Thomas. Why do firms terminate their overfunded pension plans? Journal ofAccounting and Economics, 11(4):361–398, 1989.117J. L. Treynor. The principles of corporate pension finance. The Journal of Finance,32(2):627–638, 1977.I. Welch. Two common problems in capital structure research: The financial-debt-to-asset ratio and issuing activity versus leverage changes. International Reviewof Finance, 11(1):1–17, 2011.118AppendicesAppendix A. Accounting Standards for Defined BenefitPension PlansThe way firms account for defined benefit plans on financial statements has changedover the past decades. In this section, I briefly discuss the accounting principles thatgovern the reporting of defined benefit pension variables on financial statements,set by the Financial Accounting Standards Board (FASB).Prior to 1987, there were no standardized accounting rules that governed de-fined benefit pension accounting. As a result, defined benefit pension variableswere not comparable across firms or over time, and significant pension-related as-sets and obligations were not recognized on firms’ financial statements.In December 1985, the FASB issued the Statement of Financial AccountingStandards No. 87 (SFAS 87), effective for fiscal years beginning after December15, 1986, which largely governed defined benefit pension accounting until 2006.SFAS 87 introduced a standardized method to measure pension costs and requiredactuarial estimation of pension liabilities and fair value measurement of pensionassets. As a result, employers had to report on the balance sheet a liability (accruedpension cost) if the net periodic pension expense exceeded the employer’s cashpayments into the plan, or an asset (prepaid pension cost) if the net periodic pen-sion expense was lower than the employer’s cash payments. The prepaid/accruedpension cost reported on the balance sheet was obtained by netting several off-balance sheet items: the projected benefit obligation, the fair value of plan assets,and deferred items such as actuarial gains/losses, prior service costs and transitionobligations, among others. Nonetheless, the reported pension variable under SFAS87 represented only a portion of the actual pension assets and liabilities. Moreover,employers had to immediately recognize a minimum liability whenever the accu-119mulated benefit obligation exceeded the fair value of plan assets. Finally, firmswere reporting defined benefit pension variables separately for underfunded andoverfunded plans.In February 1998, the FASB issued the Statement of Financial AccountingStandards No. 132 (SFAS 132), effective for fiscal years beginning after December15, 1997, which was amended by the Statement of Financial Accounting StandardsNo. 132 (Revised) (SFAS 132(R)) in 2003. While SFAS 132 and SFAS 132(R)significantly revised the defined benefit disclosure requirements, neither standardamended the measurement or recognition of these plans. Along with the detailedreconciliations of the pension liability and plan assets and a detailed computation ofthe pension expense, the disclosure requirements after SFAS 132 and SFAS 132(R)included information about expected future benefit payments, cash contributions,and information on the composition of pension assets and the plan’s investmentpolicy and strategy. Following SFAS 132, pension data was consolidated across allof the firm’s plans, regardless of their funding status.In September 2006, the FASB issued the Statement of Financial AccountingStandards No. 158 (SFAS 158), effective for fiscal years ending after December15, 2006. SFAS 158 requires employers to report either a net liability (when theprojected benefit obligation is greater than the fair value of pension assets) or a netasset (when the projected benefit obligation is less than the fair value of pensionassets) on the balance sheet. Under SFAS 158, employers no longer report theaccrued/prepaid pension cost or a minimum liability but instead record the plan’sunderfunded or overfunded status on the balance sheet. In addition, firms must rec-ognize the financial effects of certain plan events in other comprehensive income,and they must include previously deferred items, such as prior service cost and netgain, among others, in accumulated other comprehensive income on the balancesheet.While over the years different pension variables were accounted for on firms’financial statements, the full defined benefit assets and obligations are only reportedin the financial statements’ footnotes. Defined benefit pension obligations are mea-sured by the projected benefit obligation (PBO), which is the discounted value ofexpected future payments that have been earned to date. Future payments are fore-casted based on assumptions about mortality rates, employee turnover, retirement120dates, and future salary levels. Defined benefit pension assets are measured as thefair value of the pension plan assets. The value of pension assets depends on boththe contributions to the pension plan and the market return on plan assets.121Appendix B. Employees’ Legal Rights in BankruptcyThis section examines the treatment of employee interests in bankruptcy pro-ceedings under Chapter 11 of the Bankruptcy Code. First, I discuss several fea-tures of firm employees, such as if workers can hold equity stakes in a firm and therelationship between defined benefit pensions and wages. Next, I discuss whetherlabor can trigger bankruptcy and the role of workers in Chapter 11, such as thepriority status of their claims and the mechanisms through which employees caninfluence the reorganization process. The description of the legal framework in thissection has greatly benefited from the thorough description of the bankruptcy lawprovided by Korobkin (1996).Can Labor Hold EquityEmployees can hold firm equity through employee stock ownership plans(ESOPs). In 2014, 13.5 million employees in 7,000 companies in the U.S. werecovered by an ESOP. In an ESOP, a company sets up a trust fund and either thefirm contributes new shares of its own stock or cash to buy existing shares or thetrust borrows money to buy shares and the firm contributes cash to the plan to repaythe loan. Shares in the trust are allocated to all full-time employees’ individual ac-counts. Allocations are made either on the basis of relative pay or some predefinedformula. When employees leave the company, they receive their stock, which thecompany must buy back from them at its fair market value. ESOPs are used for dif-ferent purposes, including to buy shares of a departing owner, to borrow at a lowercost, or to make contributions to employee pension plans. For instance, rather thanmatching employee contributions to a defined contribution pension plan with cash,the firm will match them with stock from an ESOP. An example of an ESOP isthe plan created by United Airlines in 1994 when the company realized it could nolonger compete in the deregulated U.S. airline industry without substantial wagereductions. After negotiating with its unions, the firm obtained five-year pay cutsfrom its employees which were enough to secure a $5 billion loan package throughwhich employees acquired 55% ownership of United Airlines.122Defined Benefit Pensions’ Impact on WagesAccording to the implicit contract theory discussed by Treynor (1977), workerswho anticipate a career with a firm will consider the package of wage and pensionbenefits they expect to collect over their life cycle. In a model of defined benefitliabilities, Ippolito (1985a) assumes that firms do not provide pensions for free andworkers sacrifice current wages for future pension promises. In the model, workersforego an amount which is precisely equal to the present value of the expected pen-sion payments. Thus, employees sacrifice a portion of their compensation through-out their career in exchange for a pension at retirement50. In a subsequent study,Ippolito (2004) notes that the amounts workers are willing to pay for the pensionpromise have declined over time due to the higher incidence of defined benefit planterminations.Can Labor Trigger BankruptcyFirm employees are able to trigger bankruptcy by not accepting concessions inprivate negotiations with the firm or by seeking the help of the PBGC. Firms oftencite failed negotiations with labor as a reason for filing for Chapter 11 reorgani-zation. For example, in its 2012 bankruptcy filing, Hostess Brands indicated itsinability to reach an agreement with its labor representatives on wage and benefitcuts as the main reason for bankruptcy. By not accepting reductions in their salariesand in their pensions, employees are able to bring about a bankruptcy filing.In addition, labor can trigger bankruptcy with the help of the PBGC. If thePBGC determines that continuing a defined benefit plan will be more expensiveto the PBGC than terminating it immediately, the PBGC can trigger bankruptcy toterminate the plan. An example of the PBGC forcing a firm in bankruptcy is thecase of LTV Corp in 1987. LTV sponsored a defined benefit pension plan whichwas seriously underfunded. As the size of the firm’s unfunded pension obliga-tions grew, the PBGC increased its monitoring efforts. Eventually the companystopped contributing to its defined benefit pension plan and the PBGC forced LTV50In a related paper, Olson (2002) documents that workers receiving more generous health bene-fits earn lower wages than comparable workers who prefer fewer fringe benefits.123in bankruptcy in 1987 and took over its pension plan.Priority Status of Employees’ Claims in BankruptcyEmployees’ prepetition claims are generally unsecured claims, i.e. they havethe same status as most other creditors’ claims. Each employee claim is entitledto receive a pro rata share of the firm’s assets. Nonetheless, the Bankruptcy Codegives priority treatment to certain types of employee claims, such as parts of directcompensation, damage claims and promised benefits.At the time of a Chapter 11 filing, employees may have wages or commissionsthat have been earned but not yet paid. Under section 507(a)(3), these types ofclaims have priority limited to $4,000 per individual, earned in the 90 days prior tothe bankruptcy filing51. These claims for direct compensation have “third priority”status and come after administrative expenses. In addition, employees’ claims forwages and benefits earned postpetition have administrative expense priority52.Moreover, some damage claims which arise from the firm’s breaching its obli-gations under employment contracts may be given priority in bankruptcy. Thestatus of these claims depends on whether the debtor decides to assume or rejectthe underlying contract53. Under the Bankruptcy Code, some of the claims arisingfrom the contractual breach, such as wages and certain benefits, may qualify forpriority treatment. Beyond these limited priorities, damage claims arising from therejection of a prepetition contract typically constitute general unsecured claims54.In bankruptcy, workers’ prepetition claims for benefits or for unpaid plan con-tributions may receive priority. According to the Bankruptcy Code, certain welfarebenefit claims, such as vacation, severance, and sick pay leave, have third partypriority55. However, the total claim, including both direct compensation and fringebenefits, is limited to $4,000 for each employee, and must be earned in the 9051All claims for direct compensation that fall outside the language of section 507(a)(3) are generalunsecured claims.5211 U.S.C. §503(b)(1)(A).53If the firm assumes a prepetition contract, it must pay any damages arising from the breach ofcontract, but if the firm rejects a contract, no damages are owed. 11 U.S.C. §365(a).54Id. §502(b)(7).55Id. §507(a)(3).124days before the bankruptcy filing. In addition, part of the prepetition contributionsrequired by a benefit plan receive priority treatment. The claim must arise from ser-vices rendered to the company within 180 days before the filing of the bankruptcypetition and may not exceed $4,000 for each employee enrolled in a plan56. Evenfurther, if the PBGC files a lien prior to the firm’s bankruptcy, the PBGC’s claimbecomes senior and must be paid in full before unsecured creditors can be paid off.How Employees Influence Chapter 11In bankruptcy, there are at least three mechanisms that give employees a voicein the reorganization process: direct representation, labor unions and participationin the creditors’ committees.Under the Bankruptcy Code, any “party in interest, including the debtor, thetrustee, a creditors’ committee, an equity security holders’ committee, a creditor,an equity security holder, or any indenture trustee, may raise and may appear andbe heard on any issue” in a Chapter 11 case57. If an employee is also an equityholder, as a member of an ESOP or a defined contribution plan, or the employee isa creditor, as a beneficiary of an underfunded defined benefit pension plan, the em-ployee would presumably have the right to speak on any issue. Therefore, one wayin which employees’ interests can be voiced in bankruptcy would be through directaddress of the court. However, this is likely the least common channel throughwhich employees may influence the bankruptcy process due to employees’ lack oftime or money to take an active stance on their own behalf.Furthermore, Bankruptcy Rule 2018(d) specifically states that “a labor unionor employees’ association, representative of employees of the debtor, shall havethe right to be heard on the economic soundness of a plan affecting the interestsof the employees,” but “shall not be entitled to appeal any judgment, order, or de-cree relating to the plan, unless otherwise permitted by law.” A union may have theorganizational and financial resources to take an active role in the case on behalfof its members. In addition, under section 1114, the Bankruptcy Code specifically5611 U.S.C. §§507(a)(4)(A), 507(a)(4)(B).57Id. §365(a).125provides for the appointment of a committee of retirees58. In contrast, nonunionemployees often lack organization, and the Bankruptcy Code does not require theappointment of a formal committee to represent their interests. Therefore, an-other channel through which employees can influence the reorganization processis through their participation in a labor union.A third channel through which employees’ interests can be heard in bankruptcyrestructuring is through representation on a creditors’ committee. Soon after thestart of a Chapter 11 case, the United States trustee must appoint creditors’ com-mittees “consist[ing] of the persons, willing to serve, that hold the seven largestclaims against the debtor of the kinds represented on such committee.” Membersof the creditors’ committee can consult with the debtor on the case’s administrationand investigate the financial condition and operation of the reorganizing debtor59.The committee may hire attorneys, accountants, and other professionals, whoseexpenses and fees constitute administrative expenses charged to the estate60. Ifrequested, employee creditors should gain a seat on the creditors’ committee. Em-ployees’ claims differ from other creditors’ claims and thus would seem to be oneof the “kinds” that the committee must represent61. Additionally, courts gener-ally permit a union representative to serve on the creditors’ committee. For bothunion and nonunion workers, membership on the creditors’ committee provides animportant channel to have their interests considered in the Chapter 11 proceedings.58Id. §1114(b)-(d).5911 U.S.C. §1103(c).60Id. §§330(a), 530(b)(2), 1103(a).61Id. §1102(b)(1).126Appendix C. Similarities and Differences Between DefinedBenefit Obligations and Financial LiabilitiesDefined benefit liabilities resemble financial liabilities in numerous ways, in-cluding the firm obligation to pay off both liabilities, the comparability betweenpension contributions and debt interest payments, and the treatment of both typesof lenders as creditors in bankruptcy. At the same time, defined benefit pensionliabilities differ from traditional firm obligations along several dimensions, includ-ing pension liabilities’ government insurance, reporting discretion and off-balancesheet treatment, and the flexibility in contributions. I compare and contrast definedbenefit liabilities and traditional firm liabilities, such as bank loans or public bonds,in more detail here. The discussion in this section has largely benefited from thework of Shivdasani and Stefanescu (2010).Defined benefit pension obligations resemble the firm’s other liabilities in var-ious respects. For instance, defined benefit pensions create an ongoing liability forthe firm which does not disappear if an employee leaves the firm or if the pensionplan is terminated. While companies in distress may preserve cash flows by layingoff employees and thus saving on salaries and defined contribution payments, theobligation to pay the defined benefit promise remains. Therefore, firms are liablefor their defined benefit pension promises in the same way they are liable for otherobligations.Under the Employee Retirement Income Security Act of 1974 (ERISA), firmsare required to make regular minimum contributions to their defined benefit pen-sion plans, equal to the normal cost of the plan plus the level of underfundingamortized over thirty years62. Like interest payments on debt, the minimum con-tributions to defined benefit plans are tax deductible63. Prior studies have indicatedthat the tax deductibility of contributions to defined benefit plans as one of the62The normal cost of a defined benefit plan equals the benefits that active participants earn un-der the plan in a given year. In minimum contribution estimations, underfunding is defined as thedifference between the present value of future benefits and the fair value of plan assets.63While there is no cap on the contributions a firm can make to its defined benefit plans, tax ruleslimit the amount of contributions that is deductible in a given year. All contributions beyond thislimit are taxed with higher rates on an incremental basis, thus reducing firms’ incentive to contributeto the pension plan.127main reasons why these pension schemes survived over time (Black, 1980; Tepper,1981; Tepper and Affleck, 1974).If the Pension Benefit Guarantee Corporation (PBGC), a government entityestablished under ERISA to protect corporate defined benefit plans, determinesthat the plan has not met the minimum funding requirements, if the plan cannotpay current benefits when due, if a lump sum payment is made to a owner of thecompany, or if the loss to the PBGC is expected to increase unreasonably if theplan is not terminated, the PBGC may trigger bankruptcy in order to terminatethe pension plan (Shivdasani and Stefanescu, 2010)64. An example of the PBGCforcing a firm in bankruptcy is the case of LTV Corp in 1987. LTV sponsoreda defined benefit pension plan which was seriously underfunded. As the size ofthe firm’s unfunded pension obligations grew, the PBGC increased its monitoringefforts. Eventually the company stopped contributing to its defined benefit pensionplan. As a result, the PBGC forced LTV in bankruptcy in 1987 and took over itspension plan.Upon a bankruptcy filing, underfunded defined benefit obligations are usuallyconsidered unsecured claims and are entitled to receive a pro rata share of the firm’sassets (Korobkin, 1996). However, there are some exceptions to the general case.If the PBGC files a lien prior to the firm’s bankruptcy, the PBGC’s claim becomessenior and must be paid in full before unsecured creditors can be paid off. If thebankruptcy is already in effect, contributions attributable to service during the 180days prior to the Chapter 11 filing receive priority treatment under section 507(a)(5)of the Bankruptcy Code65. Altogether, pension claimants generally become a partof the unsecured creditors’ committee in bankruptcy and are never junior to thefirm’s other debt obligations.While defined benefit obligation resemble other firm liabilities, pension claimsalso differ from the firm’s other obligations. For example, under ERISA, corporatedefined benefit pension liabilities are guaranteed by the PBGC up to a certain level.In 2014, the maximum benefit guaranteed by the PBGC for a 65-year-old retiree is64In addition, if the defined benefit plan assets are insufficient to pay pension benefits currentlydue, the PBGC must terminate the pension plan.65Even further, some contributions attributed to post-petition services may also receive adminis-trative priority status.128approximately $4,940 per month. Due to these government guarantees, firms maynot bear the full costs of undertaking higher risk in their pension plan decisions andinvestments. Therefore, pension liabilities may encourage firms to take more riskthan other of the firm’s liabilities (Shivdasani and Stefanescu, 2010). Rauh (2009)documents that firms in financial trouble allocate plan assets to safer securities,such as government debt and cash and Duan et al. (2013) show that defined benefitpension plans reduce their exposures to company stock prior to debt default. Theseresults provide some evidence that the government guarantees may not necessarilyencourage higher risk taking by defined benefit sponsors.Defined benefit liabilities also differ from other firm obligations because firmsare allowed some leeway in the valuation assumptions concerning their pensionplan assets and liabilities. For instance, firms can choose the discount rate they useto discount the future value of the pension obligations. According to regulationsset by the Securities and Exchange Commission, the discount rate for defined ben-efit liabilities should be based on Moody’s Aa interest rate index but firms havediscretion as to the actual rate they use. Bergstresser, Desai, and Rauh (2006) showthat managers use the flexibility in valuation assumptions to influence reportedearnings. Even further, managers are more aggressive whenever changes to pen-sion assumptions have a greater impact on earnings. In that way, firms can avoidcovenant violations tied to earnings performance which differentiates pension obli-gations from other firm liabilities.Unlike financial liabilities, defined benefit liabilities are largely reported in thefootnotes to firms’ financial statements. While some pension plans characteristicsare reported on the balance sheet, defined benefit assets liabilities are presentedonly in the footnotes to firms’ financial statements. The accounting rules whichgovern pension plan reporting have changed over time and different pension planamounts have been reported on financial statements over time, but the total definedbenefit asset and liability are only reported in the footnotes66. Such off-balancesheet treatment distinguishes pension liabilities from other types of debt. Never-theless, prior studies have documented that defined benefit liabilities are reflectedin firms’ market valuation (Franzoni and Marin, 2006), equity beta (Jin et al., 2006)66For a more detailed description of the accounting rules and the variables reported on financialstatements over time, see Section 4 in the Appendix.129and debt ratings (Carroll and Niehaus, 1998). Therefore, there is some evidencethat the off-balance sheet treatment does not prevent investors from accounting fordefined benefit liabilities similarly to other types of debt.One further disparity between defined benefit obligations and other firm liabil-ities is managers’ ability to time the contributions to the pension plan. In general,firms have to make regular minimum contributions to the plan which are tax deduc-tive, as discussed above. However, firms with underfunded defined benefit pensionplans can make additional contributions to their plans beyond the specified mini-mum. In this way, firms can contribute more to their plans in high marginal taxstates to maximize their tax benefits. Firms do not experience the same flexibilitywith their other obligations.Last, defined benefit liabilities differ from traditional liabilities in that theclaimants of pension promises are firm employees who are not diversified (Ip-polito, 1985a,b, 2004). Unlike a bank with a diversified portfolio, defined bene-fit claimants’ wages, employment, and retirement income are vested in their em-ployer. Since their pension wealth is invested in the firm, defined benefit claimantscan and may be plausibly expected to diversify this risk away in their own portfo-lios. While this essay is not focused on pension claimants’ investment decisions,previous studies shed some light on how employees invest in their 401(k) accounts.For example, Mitchell and Utkus (2002) estimate that 5.3 million workers, or aboutone out of eight 401(k) participants, hold more than 60% of their account in owncompany stock. Moreover, the authors show that about 2.3 million workers holdbetween 41 and 60% in own company stock, and another 3 million workers holdbetween 21 and 40% in company stock. Poterba (2003)studies the cost of poor di-versification and finds that the high concentration of own stock ownership leads toa substantial reduction in expected utility upon retirement. In light of the evidencethat when they can, pension claimants do not diversify their portfolio away fromown company stock, it is possible that these claimants may not diversify the risksinherent in defined benefit pension schemes in their own portfolios.Altogether, defined benefit liabilities resemble financial liabilities and also dif-fer from traditional liabilities along various dimensions. The similarities betweenpension obligations and traditional debt give pension claimants incentives to act insync with the firm’s other lenders and to influence corporate decisions similarly to130the impact that the rest of the firm lenders have. Yet the differences between pen-sion liabilities and the firm’s other obligations may lead defined benefit claimantsto influence firm decisions differently from other lenders. Whether defined benefitclaimants act similarly to or differently from other firm lenders is a priori unclear.131

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0132736/manifest

Comment

Related Items