Empirically Bankrupt EMPIRICALLY BANKRUPT Robert K. Rasmussen* I. Introduction ................................................................ 180 II. Warren and Westbrook and Bankruptcy Choice ...... 184 A. Warren and Westbrook's Intervention of Bankruptcy Choice .............................................. 186 B. The Need to Look in the Right Place .................. 191 C. Selecting the Population ..................................... 192 D. From Population to Sample ................................ 196 E. The Need to Ask the Right Questions ................ 201 F. The Need to Draw the Right Inferences ............. 205 III. Steve Lubben and Equity Receiverships ................. 209 A. Lubben on the Effectiveness of Railroad Receiverships ....................................................... 210 B. Selecting the Appropriate Baseline .................... 213 IV. Lynn LoPucki and Bankruptcy Court Competition. 217 A. Competition and Corruption ............................... 218 B. The Need to Look in the Right Place .................. 222 C. The Need to Ask the Right Questions ................ 230 1. Traditional Cases ........................................... 231 2. Prenegotiated and Prepackaged Cases .......... 237 V . C onclusion .................................................................. 244 * Milton Underwood Professor of Law, Vanderbilt University Law School. The thoughts in this essay grow out of my long, fruitful, and continuing collaboration with my colleague Douglas Baird. This piece also benefited from the comments of Ken Ayotte, Bernie Black, Rebecca Brown, John Goldberg, Larry Helfer, Lynn LoPucki, Stephen Lubben, Ed Morrison, Eric Rasmussen, Alan Schwartz, David Skeel, Randall Thomas, and Todd Zywicki. Chris Champion provided excellent research assistance. Prior versions of this article were presented at the Columbia University School of Law, the Vanderbilt Law School, and the Annual Meeting of the American Law and Economics Association. I. INTRODUCTION Empirical analysis seeks to transform the terrain of legal scholarship. In the spirit of the natural scientist, the new cadre of empiricists attempt to use data they have gathered to end debate over a wide array of legal issues confronting scholars and law makers. A case in point can be found in the law of corporate reorganizations. Three recent empirical efforts in this area, taken together, conclude that the realities uncovered require that much prior theoretical work be rejected outright and that future scholarly efforts be directed to other areas. Elizabeth Warren and Jay Westbrook, writing in the Harvard Law Review, conclude their data demonstrate that academic calls to allow corporations to select their insolvency rules would result in inefficient redistributions and would create transaction costs that would swamp whatever benefits the new system would bring.' Stephen Lubben, in the Cornell Law Review, uses empirical data to rebut the notion that equity receiverships, which, similar to current bankruptcy practice, were dominated by senior lenders, were effective at resolving financial distress.2 Finally, Lynn LoPucki, in a book published by the University of Michigan Press, says that his data demonstrate that bankruptcy courts have become corrupted, and that this corruption is destroying companies that could otherwise be saved.' The promise of empirical legal scholarship is demonstrated by the claims these works put forward; each author asserts that he or she has uncovered facts that put an end to the central theoretical debates that have dominated the literature. The data show that contract bankruptcy is out, and bankruptcy needs to remain a mandatory rule. It ' See Elizabeth Warren & Jay Lawrence Westbrook, Contracting Out of Bankruptcy: An Empirical Intervention, 118 HARv. L. REV. 1197 (2005) [hereinafter Contracting Out]. 2 See Stephen J. Lubben, Railroad Receiverships and Modern Bankruptcy Theory, 89 CORNELL L. REV. 1420 (2004). LYNN M. LoPuCKI, COURTING FAILURE: How COMPETITION FOR BIG CASES IS CORRUPTING THE BANKRUPTCY COURTS (Michigan 2005). COL UMBIA B USINESS LA W RE VIE W [Vol. 2007 EMPIRICALL Y BANKRUPT demonstrates that equity receiverships were ineffective, and hence recent changes in today's bankruptcy practice that seem to take us back towards the dominance that investment bankers played in those reorganizations should be resisted. Finally, the data suggest that corporations need to have their venue choices severely limited in an effort to eliminate the competition among bankruptcy courts. It is, in the language of one such effort, "time to move on"4 from the debates of the past. Yet, precisely because empirical work seeks to end debate, law professors must be on guard against over- reading data and jumping to conclusions. This is not one theory battling another; such debates produce iteration after iteration, and each new effort confronts the theories that have come before. Each of the three pieces asserts that it has established certain facts that are not subject to dispute. The other side is not pressed to come up with a better defense of its theory; rather, each argues that the opposing theory should not be heeded because it runs aground on facts. This is not Burke versus Paine; it is Copernicus versus Ptolemy. The seductive finality offered by empirical legal scholars requires an evolution of current scholarly discourse. By and large, law review staffs often do not have the expertise necessary to assess empirical claims critically. Indeed, the penchant of law reviews to publish strong claims creates incentives for authors to relax the necessary caution that pervades rigorous empirical work in other fields. Law reviews also disfavor work that evaluates other claims rather than putting forth their own affirmative cases. This lack of rigor in the publication decision combined with reluctance to expose flaws may lead to the publication and immunization of works that contain erroneous assertions. Flawed claims become part of the discourse and increasingly difficult to root out. Simply put, were the adage, "It takes a theory to beat a theory," applied to empirical claims (i.e., "It 4 Warren & Westbrook, Contracting Out, supra note 1, at 1254. No. 1: 179] COLUMBIA BUSINESS LAW REVIEW takes data to beat data"), spurious claims would infect the field. The works assessed in this essay illustrate that risk. The claims of the empiricists are very strong. Taken as a whole, they would transform our understanding of the law of corporate reorganizations. Yet their assertions do not hold up upon examination. The fundamental problem is that the data does not support the claim.5 All three works fail to explain why the data that they gathered supports the conclusions that they reached. Data is trotted out as a trump, banishing the theoretical claims made by other scholars. When the data is taken on its own terms, however, it falls far short of supporting the ambitious conclusions the authors reach. Indeed, when all of the claims are examined closely, it becomes clear that it is the critic's own theoretical assumptions, not empirical evaluation, that is doing the heavy lifting. At a minimum, empirical work should look in the right place. For example, Warren and Westbrook investigate corporate reorganization proposals aimed primarily at large corporations. Such entities, however, cannot be found in their dataset. Half of their sample is comprised of individual debtors, and the other half is effectively devoid of public companies. By not creating a representative sample, Warren and Westbrook's study gets no traction on the issue they wish to explore. Conversely, LoPucki is only able to generate statistically significant results by combining full-blown Chapter 11 cases, which tend to last for months or even years, with prepackaged cases that last a few weeks. He ' The issue on which this essay focuses is not the accuracy of the data per se. LoPucki freely makes his data base available and corrects any errors that are drawn to his attention. See Lynn LoPucki, Web Bankruptcy Research Database, http://lopucki.law.ucla.edu (last visited Jan. 17, 2007). Lubben seems to have included most if not all of his data in his article. See Lubben, supra note 2. Warren and Westbrook have described their database in a prior work, but they have not made the database itself publicly available. See Elizabeth Warren & Jay Westbrook, Financial Characteristics of Businesses in Bankruptcy, 73 AM. BANKR. L.J. 499 (1999) [hereinafter Financial Characteristics]. [Vol. 2007 EMPIRICALLY BANKRUPT does not offer a theory as to why the cases should be lumped together in this manner. Empirical work also must ask the right question. Lubben is able to conclude that equity receiverships were ineffective only because he establishes a standard for reorganization law that no law to date has met. Compared to what we know about the effectiveness of Chapter 11, Lubben's data actually shows that equity receiverships performed surprisingly well. Warren and Westbrook make assertions about the ways in which bankruptcy choice proposals could redistribute money away from certain creditors that they believe bankruptcy law should protect, but they present no data on what these creditors actually received. Data collected by others suggests that the creditors Warren and Westbrook worry about by and large receive no distributions under current law. The changes that Warren and Westbrook condemn could only improve the lot of these creditors. Finally, empirical work should draw the correct inferences. Data does not speak for itself. Warren and Westbrook's study counts the number of unsecured creditors in each case. From this number, Warren and Westbrook infer that the costs of allowing debtors to commit to insolvency rules would outweigh the benefits. However, they have no information about what parties would spend in a world of bankruptcy choice or what benefits would be gained. The number of creditors standing alone cannot support an inference regarding the net welfare effects of a bankruptcy choice regime. LoPucki assumes that there were cases filed in Delaware in which the company would have reorganized successfully had the case been located elsewhere. An examination of the cases themselves, though, reveals that little could have been done to change the ultimate outcomes for these enterprises. LoPucki cannot rule out the possibility that it is a selection effect rather than a treatment effect that explains the pattern that he observes. This essay presents a cautionary tale. The growing use of empirical methods in legal scholarship is among the most No. 1: 179] noteworthy scholarly trends of the last ten years.' It is beyond cavil that this work has deepened our understanding of a wide range of legal topics. Warren and Westbrook, Lubben, and LoPucki have all made valuable contributions over the years to our understanding of reorganization practices. That said, one must always keep in mind that facts do not speak for themselves. Data must inevitably be interwoven with explanatory theory. The lesson of the studies considered here is that we must remain vigilant in this era of empiricism to avoid reflexively crediting arguments that advance tendentious theoretical claims as if they were "just the facts." We do not all need to become empirical scholars. Theory and doctrine remain honorable callings. This empirical turn, however, requires that legal discourse broaden to include work that assesses empirical claims on their own terms to ensure that their contributions are sound and that their value is properly assessed. II. WARREN AND WESTBROOK AND BANKRUPTCY CHOICE In a recent issue of the Harvard Law Review, Elizabeth Warren and Jay Westbrook report one of the first results from their project of over a decade of collecting data on business bankruptcies in this country.' This project includes hundreds of thousands of data points culled from thousands of business bankruptcies across the United States. Warren and Westbrook combed through this data in an attempt to resolve the central academic debate in corporate 6 Indeed, Empirical Scholarship was the theme of the 2006 Annual Meeting of the Association of American Law Schools. As evidence of the increasing role of empirical legal scholarship, Tracey George has complied a ranking of law schools based on their output of such work. See Tracey E. George, An Empirical Study of Empirical Legal Scholarship: The Top Law Schools, 81 IND. L.J. 141 (2006). ' Warren and Westbrook presented a summary of their data in Financial Characteristics, supra note 5. Warren, with Bob Lawless, used the database in Robert M. Lawless & Elizabeth Warren, The Myth of the Disappearing Business Bankruptcy, 93 CAL. L. REV. 743 (2005). COL UAMIA B USINESS LA W RE VIE W [Vol. 2007 reorganization law: whether bankruptcy should remain a mandatory rule.' In Contracting Out of Bankruptcy: An Empirical Intervention,' they assert that, contrary to proposals put forth by various scholars, the data leaves no doubt that the federal government and not corporations should select insolvency rules. It has been, they claim, "an entertaining debate, but it is time to move on."10 8 See Barry E. Adler, Financial and Political Theories of American Corporate Bankruptcy, 45 STAN. L. REV. 311, 313-15 (1993) (in the absence of bankruptcy law, creditors would contract to forgo individual collection rights, thus ensuring there would be no common pool problem); Robert K. Rasmussen, Debtor's Choice: A Menu Approach to Corporate Bankruptcy, 71 TEx. L. REV. 51, 55-68 (1992) (corporations should be allowed to commit to insolvency rules in advance of financial distress); Alan Schwartz, A Contract Theory Approach to Business Bankruptcy, 107 YALE L.J. 1807, 1820-39 (1998) (formal model of how bankruptcy contracts can increase social welfare); cf Lynn M. LoPucki, Contract Bankruptcy: A Reply to Alan Schwartz, 109 YALE L.J. 317, 317 (1999) ("Since the publication of Professor Robert Rasmussen's landmark article in 1992, the central focus of bankruptcy scholarship has been to discover a practical method of contracting for bankruptcy procedure."). ' See Warren & Westbrook, Contracting Out, supra note 1. 10 Id. at 1254. Warren and Westbrook have posted a reply to some of the arguments that I raise in this essay. See Elizabeth Warren & Jay Lawrence Westbrook, The Dialogue Between Theoretical and Empirical Scholarship, U. OF TEXAS LAW AND ECON. RESEARCH PAPER No. 88, http://www.ssrn.com/abstract=945155. Their arguments are in many ways misguided. To avoid assertions that I have changed my arguments in order to evade Warren and Westbrook's critique, I have left the substance of the essay as it was posted on SSRN, making only stylistic changes. I respond to Warren and Westbrook's new arguments in the footnotes. That said, it appears that Warren and Westbrook's response focuses on an early draft that I sent to them and not the draft that is currently posted publicly on SSRN and was submitted for publication to the COLUMBIA BUSINESS LAW REVIEW. Of the four "claims" that they attribute to this essay, two- one concerning the difference in debt levels between their overall sample and their sub-sample and the other exploring the differences between the sub-sample and a limited sample of cases from 2002-I removed in the process of focusing this essay on the more egregious flaws in Contracting Out prior to submitting the essay for publication While I still believe these points raise legitimate questions about the validity of Warren and No. 1:1791 EMPIRICALL Y BANKRUPT A. Warren and Westbrook's Intervention of Bankruptcy Choice In the 1990s, law-and-economics scholars put forward several "bankruptcy choice" proposals. The gist of these proposals was that social welfare will increase if corporations are allowed to commit to one set of insolvency rules well in advance of financial distress. These companies could, either at or before the time they borrowed, bind themselves to selling the corporation at auction, liquidating the company, canceling the interests of junior investors, or going through something akin to Chapter 11 should the company run into financial difficulties. 1 The intuition was that the managers of companies, seeking to reduce their overall borrowing costs, would select the set of rules that would maximize the assets of the enterprise upon financial distress. By promising creditors more return in the event of financial disaster, the Westbrook's conclusion, they are not part of the argument that I present in this piece. " There is no shortage of proposed alternatives to Chapter 11. Douglas Baird has suggested a mandatory auction. See Douglas G. Baird, The Uneasy Case for Corporate Reorganizations, 15 J. LEGAL STUD. 127 (1986). Baird and Randy Picker considered a regime where the senior creditor would not be stayed, but other creditors would. See Douglas G. Baird & Randal C. Picker, A Simple Noncooperative Bargaining Model of Corporate Reorganizations, 20 J. LEGAL STUD. 311, 348 (1991). Lucian Bebchuk suggested issuing investors of the bankrupt company a series of options that reflected their contractual priority. See Lucian A. Bebchuk, A New Approach to Corporate Reorganizations, 101 HARv. L. REv. 775 (1988). Barry Adler proposed a regime of "chameleon equity" under which default on any debt results in elimination of equity and a conversion of the lowest priority debt into equity. See Adler, supra note 8. Philippe Aghion, Oliver Hart and John Moore argued for an auction regime that allows for both cash and non-cash bids, effectively combining the Baird and Bebchuk proposals. See Philippe Aghion et al., The Economics of Bankruptcy Reform, 8 J.L. ECON. & ORG. 523 (1992). Barry Adler and Ian Ayers have suggested a new approach to sell interests in the bankrupt corporation. See Barry E. Adler & Ian Ayers, A Dilution Mechanism for Valuing Corporations in Bankruptcy, 111 YALE L.J. 83, 140-49 (2001). COL UMIA B USINESS LA W RE VIE W [Vol. 2007 EMPIRICALLY BANKRUPT debtor would reduce what it had to pay for credit.12 (Creditors price their loans to ensure a market rate of return. The more they receive should things go awry, the less interest they will have to charge to cover this risk.) Bankruptcy choice proponents have always been sensitive to the transaction costs their proposals would entail. Few companies would have the incentive to craft their own set of insolvency rules if the cost of drafting and disseminating such rules could well exceed the benefit that such rules would bring. Also, there may be benefits from using a set of rules used by others as well. Hence, bankruptcy choice proposals suggest that the state supply a menu of terms from which the corporations would select.13 These proposals did not meet with unanimous acceptance in the academy. A vocal group of critics asserted that the proposals were ripe for mischief because sophisticated creditors and debtors would select regimes that would systematically reduce the recoveries of creditors who lacked the ability to protect themselves through adjusting the prices that they charged for credit.14 Those opposed to bankruptcy 12 See Rasmussen, supra note 8, at 56-57; Schwartz, supra note 8, at 1826-32; Alan Schwartz, A Normative Theory of Business Bankruptcy, 91 VA. L. REV. 1199, 1207-11 (2005). 13 Warren and Westbrook believe that the state would only supply the terms for selection under one variant of bankruptcy choice. See Warren & Westbrook, Contracting Out, supra note 1, at 1242 (pointing out that Rasmussen's menu-approach provides a set of choices). In this they are mistaken. See Schwartz, supra note 8, at 1850 (state should supply choices). Moreover, they assert that having a set of standard forms lessens the gains from bankruptcy choice. See Warren & Westbrook, Contracting Out, supra note 1, at 1242. Such standardization, however, is a benefit of the system, not a cost. See Rasmussen, supra note 8, at 66; Schwartz, supra note 8, at 1843. For a general explanation of the efficiency enhancing properties of standardized terms, see Charles J. Goetz & Robert E. Scott, The Limits of Expanded Choice: An Analysis of the Interaction Between Express and Implied Contract Terms, 73 CAL. L. REV. 261 (1985). 14 See, e.g., LoPucki, supra note 8, at 339 (Bankruptcy choice has "the potential to redistribute wealth from noncontracting parties to contracting parties."); Susan Block-Lieb, The Logic and Limits of Contract Bankruptcy, 2001 U. ILL. L. REV. 503, 516 (opponents of bankruptcy choice "contend No. 1: 179] COLUMBIA BUSINESS LAW REVIEW choice also complained that the inevitable transaction costs attendant on these schemes would consume the benefits that choice would bring. 15 Those in favor of bankruptcy choice responded that concerns about redistribution were unfounded, and that transaction costs would not loom large. The theoretical debate over the basic structure of reorganization law thus devolved into competing factual assumptions. Attempting to end this debate, 6 Warren and Westbrook turned to the database they had developed as part of their business bankruptcy project. 7 The Business Bankruptcy database is comprised of 3,201 business bankruptcy cases filed in 1994.18 Warren and Westbrook generated this that [bankruptcy choice] would impose distributive costs on involuntary creditors, such as tort victims, unsophisticated creditors, and creditors whose small claims would not justify the cost of these complicated contractual remedies"). Warren and Westbrook make no effort to engage these arguments. 1 Indeed, concerns over the extent to which creditors can adjust their interest rates and transactions costs have been staples of the bankruptcy choice literature since its inception. See, e.g., Barry E. Adler, Finance's Theoretical Divide and the Proper Role of Insolvency Rules, 67 S. CAL. L. REV. 1107, 1131 (1994) (describing contracting costs as "trivial"); Donald R. Korobkin, The Unwarranted Case Against Corporate Reorganization: A Reply to Bradley and Rosenzweig, 78 IOWA L. REV. 669, 720 (1993) ("[Tlhe administrative and disruption costs of coordinating the negotiation of a full network of default-contingent contracts are sure to be substantial."); Block-Lieb, supra note 14, at 517 ("Despite claims about the cost-saving effect of [bankruptcy choice], commentators are dubious that contractual substitutes will be less costly than the current bankruptcy process." (citations omitted)). 16 Warren and Westbrook complain that I believe "that empiricists mean to stamp out all traditional theoretical work." Warren & Westbrook, supra note 10, at 2. This mischaracterizes my concern, which is merely with empirical work that claims to resolve a theoretical debate even though the data offered do not support such a conclusion. Warren and Westbrook asserted that they had put an end to the debate over bankruptcy choice; this essay demonstrates that they have not. 17 See Warren & Westbrook, Financial Characteristics, supra note 5, at 503-17. " See Warren & Westbrook, Contracting Out, supra note 1, at 1209 & n.43. [Vol. 2007 sample by identifying the judicial districts within each court of appeals (other than the Federal Circuits and court of appeals for the District of Columbia) that had the greatest and the smallest number of bankruptcy cases filed. They then added an additional district from the Ninth Circuit to account for the fact that the bankruptcy courts in that circuit handle roughly one-third of all bankruptcies in the country.19 Two of the least active districts were swapped out for other districts in the relevant circuit due to lack of Chapter 11 activity.2" After selecting these twenty-three districts, Warren and Westbrook then selected fifty business cases filed under each Chapter of the Bankruptcy Code during 1994.21 To ensure that these cases were dispersed throughout the calendar year, they picked the first twelve or thirteen cases of each type filed at the start of each calendar quarter.22 Warren and Westbrook did not use the entire database for testing bankruptcy choice. Instead, they constructed a 19 Warren & Westbrook, Financial Characteristics, supra note 5, at 510. 20 See Warren & Westbrook, Financial Characteristics, supra note 5, at 510 (Eastern District of North Carolina selected from Fourth Circuit even though it was the third lowest in terms of cases filed), 511 (District of Connecticut selected from the Second Circuit even though it was the second-lowest). This selecting from the tails of the distribution may well be problematic. Standard protocol provides for the selection of samples such that "each element in the total population has a known (and preferably the same) probability of being selected." Lee Epstein & Gary King, The Rules of Inference, 69 U. CHI. L. REV. 1, 108 (2002). These districts, however, were chosen because they were the outliers in their respective judicial circuits. While it is unclear whether selecting from the tails in this fashion biased the overall results, the study does not include any reassurances to allay fears of some systematic difference between the sample and business bankruptcies generally. Given the other problems with the ability of this data to shed any light on bankruptcy choice proposals, there is no need to pursue the matter any further. 21 Warren and Westbrook collected cases from Chapter 13, which is only available to individuals, as well as Chapters 7 and 11. In Contracting Out, they do not rely on any data regarding these Chapter 13 filings. See Warren & Westbrook, Contracting Out, supra note 1, at 1211. 22 See Warren & Westbrook, Financial Characteristics, supra note 5, at 511. No. 1: 179] EMPIRICALL Y BANKRUPT COLUMBIA BUSINESS LAW REVIEW sub-sample of this larger pool by taking every fifth Chapter 7 and Chapter 11 case from the main source. This provided them with 386 cases (the "sub-sample"). For each debtor within this sub-sample, Warren and Westbrook catalogued the number of creditors each debtor had, the nature of each creditor, and the amount of each creditor's claim.23 It is this data on which their empirical claims rest. Warren and Westbrook examined the characteristics of each creditor because they were attempting to discover how many creditors lacked the ability to alter their interest rates in response to a debtor's bankruptcy selection.24 They argued that tort claimants, taxing authorities, utility companies, employees, and individuals generally lack the ability to price credit based on the circumstances of the debtor.2 5 In addition, Warren and Westbrook posited that creditors with small claims would not make the effort to adjust their rates.26 Warren and Westbrook focused on both of these groups of creditors because they were at risk should the law change to one of bankruptcy choice. Warren and Westbrook worried that debtors would systematically choose bankruptcy rules so as to redistribute money from these creditors to others and that these redistributions would be inefficient. 2 1 In addition to concerns about redistribution, Warren and Westbrook sought to identify the expenses that would be associated with bankruptcy choice. To do this, they counted the total number of creditors each debtor had. They assumed that the greater the number of creditors, the more costly it would be for the creditors to learn about which set of rules would govern a particular debtor. They also assumed that the more creditors a debtor had, the more contentious 23 See Warren & Westbrook, Contracting Out, supra note 1, at 1210- 11. 24 See id. at 1214-15. 25 See id. at 1216. 26 See id. at 1214. 27 See id. [Vol. 2007 EMPIRICALLY BANKRUPT and costly would be the negotiations over the bankruptcy selection." Warren and Westbrook asserted that their data on these points resolved the bankruptcy choice debate. Creditors who could not adjust interest rates were abundant. Warren and Westbrook found that roughly a quarter of the creditors in their sample had claims held by entities who did not adjust their interest rate on a debtor-by-debtor basis." Added to this, roughly six percent of the remaining debt was held in small chunks. ° From these numbers, Warren and Westbrook infer that bankruptcy choice would cause redistribution among creditors and that these redistributions would cause "substantial inefficiencies."3 Warren and Westbrook also reported information on the number of creditors each debtor had. They found in their sub-sample that the mean for each debtor was nineteen and that the maximum was 255.32 The existence of this number of creditors implied to them that bankruptcy choice "would produce substantial transaction costs that would likely overwhelm any claimed gains."3 For these reasons, they concluded that those who had argued for allowing debtors to choose bankruptcy rules in advance may have provided an interesting theoretical case, but that the data demanded that the proposal be rejected. B. The Need to Look in the Right Place The first task of any empirical project is to select the appropriate population to study. Here, the population is the set of entities to which the relevant legal reform would 28 See id. at 1249-50. 29 See id. at 1236. 30 See id. at 1248 n.148 (reporting that small claims were roughly seven percent of all unsecured claims and that twelve percent of these claims were held by creditors who could not adjust their interest rates regardless of claim size). 3' Id. at 1248. 32 See id. at 1250-51. 3" Id. at 1253. No. 1: 179] COLUMBIA BUSINESS LAW REVIEW apply. 4 Gathering information about those whom a legal change would not affect pollutes the data. Conversely, leaving out those whom a change would affect impoverishes the analysis. When a researcher reports results drawn from too broad a population, the readers of the analysis do not know whether the information comes from those to whom the legal change would apply or from those for whom the change would be a non-event. Alternatively, when the researcher draws from too narrow a pool, the reader can only speculate about how those not studied might affect the result. Thus, failure to specify the correct population limits the value of an empirical project before even the first piece of data is collected. C. Selecting the Population What, then, is the appropriate population for studying the impact of rules permitting choice of bankruptcy rules? The literature advocating bankruptcy choice has made it clear that, based on the selection of titles and the substance of the arguments, bankruptcy choice proposals would apply only in the corporate setting. The first article in this spirit was called Debtor's Choice: A Menu Approach to Corporate Bankruptcy.35 Other works in this genre describe rules that a corporation may select for parceling out its equity interests in an attempt to resolve financial distress. None of the proposals extend to individuals running sole proprietorships. This is not a trivial or esoteric point. Bankruptcy law distinguishes between corporations and individuals. While both corporations and individuals can file for relief under Chapter 7, only individuals can receive a discharge of their debts. Corporations are not similarly entitled to such relief.36 In Chapter 11, an individual's post-bankruptcy earnings are not subject to creditor claims; all the earnings of a corporation, in contrast, go to its creditors. The law thus 34 See DAVID COPE, FUNDAMENTALS OF STATISTICAL ANALYSIS 102 (Foundation Press 2005). Rasmussen, supra note 8 (emphasis added). See 11 U.S.C. § 727(a)(1) (2004) (limiting discharge to "individuals"). [Vol. 2007 draws a clear line between the financial distress of individuals and that of legal entities.3 v Bankruptcy scholarship and commentary explain the reason behind this distinction. Corporate bankruptcy law seeks to put assets to their highest valued use.38 It decides whether the assets should be sold to new investors, either piecemeal or in bulk, or kept in their current configuration. If the latter route is chosen, Chapter 11 provides a mechanism by which the capital structure of the business is readjusted. 9 Out-of-the-money interests are eliminated, and the remaining investors receive new rights against the business. Just as corporate law exists to maximize the value of the corporation,4" so does corporate bankruptcy law.4 Individual bankruptcy law, in contrast, seeks to discharge debts so as to provide the individual with a fresh start in life.42 The future earnings of the debtor go to the individual, " Warren and Westbrook "are puzzled by the claim that natural- person debtors do not matter in a study of business bankruptcy or a critique of contractualism." Warren & Westbrook, supra note 10, at 4. In neither their original article nor their response do they either disagree with the proposition that the proposals they are investigating are expressly limited to legal entities or set forth a theory as to why they think this distinction should be ignored. 38 See Rasmussen, supra note 8, at 62; Robert K. Rasmussen, An Essay on Optimal Bankruptcy Rules and Social Justice, 1994 U. ILL. L. REV. 1, 2 (1994); Schwartz, supra note 8, at 1807. 3' There is no confusion on this point in the literature. For scholars making the point explicitly, see Ralph Brubaker, Taking Exception to the New Corporate Discharge Exceptions, 13 AM. BANKR. INST. L. REV. 757, 759-61 (2005); DOUGLAS G. BAIRD ET AL., BANKRUPTCY: CASES, PROBLEMS, AND MATERIALS 561 (Foundation Press rev. 3d ed. 2001); Schwartz, supra note 12, at 1220. 40 See FRANK H. EASTERBROOK & DANIEL R. FISCHEL, THE ECONOMIC STRUCTURE OF CORPORATE LAw 34-35 (Harvard 1991); REINIER R. KRAAKMAN ET AL., THE ANATOMY OF CORPORATE LAw 17-19 (Oxford 2004). 41 Warren and Westbrook, while expressing some discomfort with this limitation of the goals of bankruptcy, accept it for the purposes of their piece. See Warren & Westbrook, Contracting Out, supra note 1, at 1203- 04. 42 See Barry E. Adler, Bankruptcy Primitives, 12 AM. BANKR. INST. L. REV. 219 (2004); THOMAS H. JACKSON, THE LOGIC AND LIMITS OF BANKRUPTCY LAW 225 (Harvard 1986). No. 1: 1791 EMPIRICALL Y BANKRUPT COLUMBIA BUSINESS LA W REVIEW not to his or her pre-bankruptcy creditors. The individual can enjoy the fruits of his or her human capital unburdened by the claims pre-bankruptcy creditors. The discharge of past debts is the defining feature of individual bankruptcy law but is absent when discussing corporate reorganizations. Giving a fresh start to the travel agent running a sole proprietorship out of her house is a fundamentally different endeavor from sorting out the affairs of United Airlines. Any empiricist setting out to test the bankruptcy choice proposals, therefore, needs to begin with the population of corporations that may file for bankruptcy. Contracting Out did not so limit its sample population.43 Bankruptcy courts receive cases from both individuals and corporations. Both individuals and corporations can file under Chapter 7 or Chapter 11. When Warren and Westbrook sampled cases from their chosen districts, they included individuals who indicated that they were engaged in business. 44 Later, in constructing the sub-sample for Contracting Out, they once again did not limit their debtors to corporations. The only accommodation they made was to not include individuals who filed under Chapter 13. The result of the decision to include individuals in the sub-sample is that roughly half of the cases that Warren and Westbrook examined were cases filed by individuals rather than corporations. 45 Thus, the population was, presumably, fifty percent comprised of hairdressers, limo drivers, travel agents and others who were not doing business in the corporate form. Bankruptcy law gives these individuals a right to enjoy the income from their human capital. Current ' Warren and Westbrook's reply to this essay implies that I did not raise this (and other) criticisms with them privately. See Warren & Westbrook, supra note 10, at 3. In fact, I wrote them a letter on April 14, 2004 raising this issue. I also contacted them via email to respond to the arguments in their reply well in advance of publishing this essay. " See Warren & Westbrook, Financial Characteristics, supra note 5, at 512. 45 See id. at 532 (noting one-quarter of their Chapter 11 cases and three-quarters of their Chapter 7 cases were filed by individuals rather than legal entities). [Vol. 2007 law does not allow these individuals to waive their right to discharge, and no bankruptcy choice proposal would change this. The proposals Warren and Westbrook claim to test thus do not encompass half of the debtors on which they rely for their conclusion. The decision to include subjects who were not in the relevant population undermines the study's conclusions. To see how, recall that Warren and Westbrook devote much effort in Contracting Out to identifying certain types of creditors whom they believe could be harmed by bankruptcy choice. These were creditors who could not adjust their interest rates to take account of the governing insolvency rules, and they consisted of tort creditors, tax authorities, utility companies, employees, individuals and other creditors with claims of less than $5,000.46 But these classes of creditors are not distributed uniformly between corporate and individual debtors. Rather, one would expect that individual debtors, as compared with corporate debtors, would be more likely to have inadequate insurance, inadequate systems in place to pay taxes, utilities and employees, and indebtedness to individuals as well as having smaller debts. In other words, individuals are more likely than corporations to have the very type of creditors whom Warren and Westbrook worry would be ill-equipped to confront bankruptcy choice. By including individuals in their sample, Warren and Westbrook thus increase the incidence of the phenomena that forms the basis of their claim. We have no way of knowing what their figures would have looked like had they limited their sample to the domain of debtors relevant to bankruptcy choice. We can say, however, that the skewed data cannot support the broad conclusions that the authors draw.47 46 See Warren & Westbrook, Contracting Out, supra note 1, at 1227- 44. 41 Warren and Westbrook's response asserts that "we can strip out the individuals in business from the data, and the results of the analysis do not change in any meaningful way." Warren & Westbrook, supra note 10, at 4. It is difficult to understand the assertion being made in the following sense. Warren and Westbrook report that the overall number of creditors No. 1: 179] EMPIRICALL Y BANKRUPT COLUMBIA BUSINESS LA W REVIEW D. From Population to Sample Identifying the appropriate target population is only the first step in the analysis. Resource constraints mean that researchers often cannot collect information about every member of the population. 48 Instead, they have to create a sample of the population.4 9 The ultimate goal is to be able to assess how the proposed law reform will affect the entire population. By looking at a carefully selected subset of the population, the researcher endeavors to glean information that can be generalized to the relevant population in its entirety. In order to support such a generalization, the sample must share the attributes of the population that are relevant to the question being probed. Contracting Out does not construct its sample according to this principle and actually excludes from its sample the debtors most relevant to the issue of bankruptcy choice.5" goes up when one limits the data to corporations. This is not a surprise. I never asserted that corporations have fewer creditors than do individual debtors. As to maladjusting and small creditors, Warren and Westbrook do not offer any sense as to what the data they presented in their original piece would look like if individual debtors were removed. They say that the number of debtors having such creditors remains roughly the same, but they give no evidence as to the prevalence of these types of claimants in the corporate cases. Recall that roughly thirty percent of the debt in their undifferentiated sample was held by small or maladjusting creditors. They provide no statement or evidence as to how this number changes when individual debtors are excluded. Obviously, a population study is better than a sample. For example, Lynn LoPucki's well known database on large, publicly held corporations that file for bankruptcy is a census of that population rather than a sample. See Lynn M. LoPucki's Bankruptcy Research Database, http://lopucki.law.ucla.edu/contents-of the webbrd.htm (last visited Nov. 20, 2006). 41 See Epstein & King, supra note 20, at 28-30. " Warren and Westbrook's sample comes from cases filed in 1994. This presents two problems. The first is that bankruptcy practice today is fundamentally differently from practice at that time. See Douglas G. Baird & Robert K. Rasmussen, Chapter 11 at Twilight, 56 STAN. L. REV. 673 (2003); Harvey R. Miller & Shai Y. Waisman, Is Chapter 11 Bankrupt?, 47 B.C. L. REv. 129 (2005). The second is that 1994 was a quiet year for corporate reorganizations, especially those of public [Vol. 2007 EMPIRICALLY BANKRUPT For years, commentators, including Warren and Westbrook, have recognized that the dynamics of corporate bankruptcies differ based on the size of the corporation.51 As Warren has observed, "the experience of large, publicly traded companies in bankruptcy differs sharply from that of smaller, private companies." 2 The financial distress of small corporations presents different problems than the distress of large companies. The problems of the limo service doing business as a limited liability company are different from the problems of General Motors.53 Whether bankruptcy choice would provide gains to small corporations is a different inquiry as to whether it would provide gains to the likes of Kmart. Large corporations tend to have a separation between ownership and control.54 The owners invest none of their human capital in the business. Investors tend to hold small interests and lack the ability to coordinate among companies. See THE 2006 BANKRUPTCY YEARBOOK & ALMANAC 31 (New Generations Research 2006) (listing number of Chapter 11 filings by public companies for each year since 1980 along with the total and average assets). 51 See ELIZABETH WARREN & JAY WESTBROOK, THE LAW OF DEBTORS AND CREDITORS 678 (5th ed. 2006) ("But the huge difference between business and consumer cases can obscure another difference-a staggering variety within each of the two classifications. Business cases include Tina's Tax Preparation & Tanning Salon, a Tupperware party planner, and a lawn service man who has lost his mower (all companies in our Business Bankruptcy Project) along with some-what better known companies such as Enron, Worldcom, Alephia[sic], and perhaps every major airline carrier in the country. These tiny little businesses and great big businesses face many of the same formal provisions when they try to reorganize in Chapter 11, but the practical realities facing these businesses may be very different."). 52 Elizabeth Warren, The Untenable Case for Repeal of Chapter 11, 102 YALE L.J. 437, 443 (1992). See Douglas G. Baird & Edward R. Morrison, Serial Entrepreneurs and Small Business Bankruptcies, 105 COLUM. L. REV. 2310, 2311 (2005) (Small businesses have "few assets beyond the entrepreneur's human capital, and these rarely have more value inside the business than outside."). 5 See ADOLF A. BERLE & GARDINER C. MEANS, THE MODERN CORPORATION AND PRIVATE PROPERTY (Transaction Publishers 1933). No. 1: 179] COLUMBIA BUSINESS LA WREVIEW themselves. Professional managers run the enterprise. The business is perfectly capable of functioning with a new set of investors and a new set of managers.55 It is here that bankruptcy choice proposals have focused much of their attention. There are many ways in which the financial distress of these companies can be addressed: the business could be sold as a going-concern, it could be liquidated piecemeal, or it could be reorganized with a new capital structure.56 One can quickly identify a range of other possible alternatives in the literature, few of which would make any sense for a struggling restaurant. 57 Bankruptcy choice rests on the proposition that the investors in the business are better able to make this selection than the government. Small corporations, in contrast, are basically human- capital firms.5 " The major asset of the business is the talent of the owner. The other assets are primarily standard goods that have no value above what they could fetch in a sale. The problem that bankruptcy law needs to focus on in these cases is whether the owner should stay with this corporate entity or move to another. The problems of large, publicly held companies are ones of corporate finance; those of small corporations are of labor economics. To explore the effects of bankruptcy choice across corporations, one needs to divide the population of corporations appropriately. One can offer a number of plausible classifications to accomplish this.59 It is not "5 There is an active market of managers whose sole task is to address a corporation's financial distress. See Douglas G. Baird & Robert K. Rasmussen, Private Debt and the Missing Lever of Corporate Governance, 154 U. PA. L. REV. 1209 (2006). '8 See Baird, supra note 11. 5' See, e.g., Adler, supra note 8 (proposing "chameleon equity" financial structure); Bebchuk, supra note 11 (proposing exchanging existing debt and equity claims for financial options); Aghion et al., supra note 11 (proposing an auction system that allows for non-cash bids). See Baird & Morrison, supra note 53, at 2330-32. 5 See, e.g., Arturo Bris et al., The Costs of Bankruptcy: Chapter 7 Liquidation vs. Chapter 11 Reorganization, 61 J. FIN. 1253, 1278 (2006) (breaking down sample according to less than $100,000 in assets; $100,000 [Vol. 2007 difficult to find laws that depend on the size of the corporation being regulated. 0 Perhaps the most basic and easiest demarcation would be to distinguish between private and public companies, a distinction common in both the bankruptcy and corporate law literatures.6' Despite the central importance of large corporations to the bankruptcy choice debate, Warren and Westbrook's study appears to have omitted all or nearly all public corporations. Warren and Westbrook provide little descriptive information about the cases in their sub-sample, but from what they do provide it seems that, at most, there may be one relatively small public corporation in the sub- sample.62 Thus, out of 386 debtors in their sub-sample, 385 are either individuals or privately-held companies. This failure to include public companies in the data they examine means that the effects of bankruptcy choice on such companies are simply beyond the reach of the study. The creditors on whom Warren and Westbrook focus for much of their article, including tort claimants, unpaid utilities, tax to $1 million, $1 million to $10 million, and greater than $10 million); Douglas Baird et al, The Dynamics of Large and Small Chapter 11 Cases: An Empirical Study, working paper (using less than $100,000, $100,000 to $200,000, $200,000 to $500,000, $500,000 to $1 million, $1 million to $2 million, $2 million to $5 million, and above $5 million to break down the sample); Warren & Westbrook, Financial Characteristics, supra note 5, at 521 (using less than $100,000, $100,000 to $500,000, $500,000 to $1 million, $1 million to $5 million, and above $5 million to break down the sample). 60 For example, the Family and Medical Leave Act only covers private employers with fifty or more employees. See 29 C.F.R. § 825.104(a) (1995). The American with Disabilities Act and Title VII both cover private employers with at least fifteen employees. See 42 U.S.C. § 12111(5)(A) (ADA) (1991); 42 U.S.C. § 2000e (Title VII) (1991). 61 See, e.g., Warren, supra note 52, at 443 ("the data suggest a critical difference between the bankruptcy experiences of private and public corporations"). 6 In their larger database, Warren and Westbrook have six of the seventy public companies that filed for bankruptcy in 1994. See Warren & Westbrook, Financial Characteristics, supra note 5, at 548-49. In their response to this essay, Warren and Westbrook do not take issue with my conclusion that public companies are not represented in their sub-sample. EMPIRICALL Y BANKRUPTNo. 1:1791 authorities, employees owed money and individuals, are not uniformly distributed with respect to all debtors. Large businesses have systems in place to ensure that taxes, utilities, payroll and the like are paid. They carry insurance that covers all but the most catastrophic claims. They borrow money from large commercial lenders, not friends and family. It may well be that large, publicly held companies, on the whole, lack the type of creditors on which Warren and Westbrook's conclusions rest. The absence of public corporations from the sample also makes it impossible to assess the aggregate welfare effects of bankruptcy choice. To be sure, these companies do not dominate bankruptcy court in terms of their numbers. Even in the most active year, bankruptcies of publicly held companies number less than three hundred." Every year they are less than one percent of business bankruptcies.' But the assets that these companies own ensure that they have effects that extend well beyond their small numbers. In six of the last seven years, the assets of public companies filing for bankruptcy exceeded $50 billion. Three of these years saw assets of over $100 billion in bankruptcy, and the largest single year witnessed almost $400 billion in assets entering into Chapter 11 as part of the filings of public companies. 5 Even though 1994 was the nadir in terms of public companies filing for bankruptcy, there were still seventy public companies that filed for bankruptcy,66 and the total assets for these companies exceeded $8 billion.6 In 1994 itself there were six filing companies, each of which reported assets that exceeded the debts of Warren and 6 See THE 2006 BANKRUpTcY YEARBOOK & ALMANAC, supra note 50, at 28 (listing the number of publicly held companies filing for bankruptcy for each year since 1980). 6 Compare id. (listing number of publicly held companies filing for bankruptcy each year) with id. at 6 (listing the number of business bankruptcies filed each year). Id. at 28. 66 This is the fewest number of corporate bankruptcies for any year since 1980. See id. " Again, this figure is low by historical standards. See id. COL UAMIA B USINESS LA W RE VIE W [Vol. 2007 EMPIRICALLY BANKRUPT Westbrook's entire sub-sample. The entire debt for their sub-sample, secured and unsecured, is $376 million. Memorex, which filed for bankruptcy in 1994, reported assets of over three times this amount." Five other companies, Resorts International, House of Fabrics, Kash N Karry Food Stores, Merry-Go-Round Enterprises, and F & M Distributors, each reported more assets than the cumulative debt in Warren and Westbrook's sub-sample. 9 Public companies play a crucial role in ascertaining the merits of bankruptcy choice, yet they are not included in the Contracting Out study. The end result is that Warren and Westbrook have a dataset half of which is comprised of debtors who would never be subject to a bankruptcy choice regime, and the other half of which omits the type of debtor that motivated bankruptcy choice scholarship in the first instance. The study is unable, therefore, to make any contribution to the debate about bankruptcy choice. E. The Need to Ask the Right Questions There is additional reason to doubt the conclusions of the Warren and Westbrook study. The paper devotes much of its discussion to the inefficiencies that might arise if debtors could redistribute wealth from the various "maladjusting creditors" that Warren and Westbrook catalogue.7 ° The concern is that if these creditors are receiving substantial recoveries today, they will lose something in a system of bankruptcy choice. Such redistribution might be inefficient, " Lynn LoPucki, Web Bankruptcy Research Database, http://lopucki. law.ucla.edu (last visited Jan. 17, 2007). 69 Three more cases: Westmoreland Coal, Crystal Brands, O'Brien Environmental Engineering each reported more than $300 million in assets. The data on the asset value of the corporations comes from LoPucki's web-version of his database. See Lynn LoPucki, Web Bankruptcy Research Database, http://lopucki.law.ucla.edu (last visited Jan. 17, 2007). " See Warren & Westbrook, Contracting Out, supra note 1, at 1219-53 (thirty pages discussing maladjusting creditors as compared with five discussing transaction costs). No. 1: 179] and hence would weigh in the balance against bankruptcy choice. 71 To quantify the risk of such redistribution, it is essential to have information about the types of creditors a debtor has and how much these creditors receive under current bankruptcy law. The type of creditor is relevant insofar as it suggests whether the creditor could adjust its interest rate in response to changes in insolvency rules. Creditors who adjust their prices cannot be systematically disadvantaged. The amount the creditors receive is equally important to the overall efficiency inquiry. The concern over redistribution assumes a change in position for these creditors, from distributions that they are receiving to something less. Nevertheless, Warren and Westbrook report no infor- mation on creditor recoveries.72 They do report that almost a third of the dollar amount of unsecured claims in their sub- sample is held by the types of entities that will not respond for one reason or another to changes in insolvency rules. But they offer no data on what these entities recovered under the existing bankruptcy regime. The dollar amount of debts owed to unsecured creditors provides no information as to these creditors' prospects for actual recovery, in light of the limited assets in the bankruptcy estate and the priority accorded to secured creditors and administrative expenses. Instead of providing data on this point from their sub- sample, Warren and Westbrook cite to Lynn LoPucki's work on large, public corporations suggesting that, in those cases, general creditors often receive a substantial return on their " See generally id. Warren and Westbrook equate redistribution with efficiency, but they fail to specify the relationship between the two. Generally, a redistribution is inefficient only to the extent that the redistribution itself consumes resources. It is not the amount of the resources actually transferred. Since the transfer under bankruptcy choice would come through the selection of an insolvency regime, the redistribution argument is to a large extent parasitic on the transaction costs argument. 72 It appears that they actually do have such data. Another scholar to whom they have granted access to their database does report the distributions that were made in the sub-sample he studied. See Stephen J. Lubben, The Other Liquidation Decision (working paper). COL UMIA B USINESS LA W RE VIE W [Vol. 2007 claims.73 Yet again, commentators for years have cautioned that one cannot reflexively transfer results from large cases to small ones. 4 It thus is not obvious that the distribution to the bondholders in the Kmart bankruptcy provides information as to the distribution to trade creditors in the bankruptcy of the local drugstore. Data from other sources provides reason to conclude that the large majority of unsecured creditors in Warren and Westbrook's sub-sample received nothing. 5 Half of the cases in Warren and Westbrook's study were filed under Chapter 7, and other studies have shown that in virtually every completed Chapter 7 case general creditors receive no distribution.76 For the other half of debtors, those that filed under Chapter 11, the general creditors probably did not fare much better. We know that approximately two-thirds of corporate cases that start out in Chapter 11 are either eventually dismissed so that creditors can exercise their 71 See Warren & Westbrook, Contracting Out, supra note 1, at 1218 n.75 (citing prepublication version of Lynn M. LoPucki, The Myth of the Residual Owner, 82 WASH. U.L.Q. 1341 (2004)). 74 See, e.g., Warren, supra note 52. 71 Warren and Westbrook's response to this assertion misinterprets the point being made in this article. They write that I claim that "unsecured creditors.. . already collect nothing in bankruptcy." Warren & Westbrook, supra note 10, at 4. This misreads my argument. Of course some unsecured creditors recover money. The point is that these recoveries tend to be in the larger cases, cases that are not included in the sub-sample Warren and Westbrook are relying on. As explained in this article, Warren and Westbrook have created a sub-sample that is dominated by the type of debtor one would expect to have a greater percentage of its debt held by small and maladjusted creditors. The unsecured creditors of these debtors are unlikely to receive any distributions. When we get to cases where we find unsecured creditors receiving substantial recoveries, it is unlikely that these debtors have a substantial portion of their unsecured debt held by small and maladjusting creditors. 76 See Bris et al., supra note 59, at 1289 ("We find that unsecured creditors receive nothing in 95% of our Chapter 7 cases. The mean recovery rate is 1%, all driven by one case."). No. 1:179] EMPIRICALL Y BANKRUPT COLUMBIA BUSINESS LA W REVIEW state law remedies or are converted to Chapter 7.77 As to the cases that the bankruptcy court dismisses, conventional wisdom is that a judge should dismiss a case when the senior lender is owed more than all of the assets of the business are worth.78 Based on this, it is fair to infer most of the unsecured creditors in the dismissals receive nothing, just as the unsecured creditors do in the completed Chapter 7 cases. If Warren and Westbrook's Chapter 11 cases share the two- thirds conversion/dismissal rate found for similar cases in other data sets, then in roughly two-thirds of their Chapter 11 cases there were no distributions to unsecured creditors. Combining the cases filed in Chapter 7 with the Chapter 11 cases that were either dismissed or converted, it would appear likely that over eighty percent of the cases in the Warren and Westbrook sub-sample resulted in no distribution to unsecured creditors. The creditors in the remaining twenty percent of the cases in Warren and Westbrook's sample likely did not fare much better. Even in the remaining one-third of Chapter 11 cases, general unsecured creditors often are left out of the distribution. In these cases, the pattern tends to be that the smaller the case is, the less likely it is that the unsecured creditors will see a return. For example, in completed Chapter 11 cases with assets of less than $200,000, general creditors usually receive nothing. Indeed, only when one starts looking at businesses with more than $5 million in assets does one find recoveries to the unsecured creditors 71 See id. at 1271 (reporting two-thirds dismissal or conversion rate); see also Edward R. Morrison, Bankruptcy Decisionmaking: An Empirical Study of Continuation Bias in Small Business Bankruptcies, 50 J.L. & ECON. (forthcoming 2007) (manuscript at 11-16) (showing that over sixty percent of Chapter 11 filings were dismissed or converted to Chapter 7). It appears that Warren and Westbrook's dataset includes information on whether a case that was filed under Chapter 11 was ultimately converted to Chapter 7 or dismissed. See Lubben, supra note 72. Warren and Westbrook, however, do not report this information. 78 See Morrison, supra note 77, at 12-13. [Vol. 2007 EMPIRICALLY BANKRUPT which approximate those reported by LoPucki. 7s There are very few such debtors in Warren and Westbrook's dataset s It is likely that the creditors on whom Warren and Westbrook rely did not receive any satisfaction of their claims; therefore, it is difficult to credit a concern that forecasts a loss to this group. These creditors received nothing under current law. Thus, adopting bankruptcy choice could only improve their lot. Because this study failed to ask the right question, it cannot claim to have found the right answer. F. The Need to Draw the Right Inferences Warren and Westbrook report that for the businesses in their data set, the average business had nineteen creditors, and the largest had 255.1 The inference that they draw from these data is that bankruptcy choice would create substantial transaction costs. s2 This inference, from data " See Baird et al., supra note 59. Warren and Westbrook's response takes me to task for not citing this work even though I provided comments on it. Again, Warren and Westbrook appear to be referencing an old version of this essay. I have included references to the Baird et al. piece since it was circulated. The data in Baird et al. confirm that the creditors on which Warren and Westbrook's case against bankruptcy choice rests- small and maladjusting creditors in the cases they examine-most likely recover nothing under the present system. "0 Debtors of this size comprise 8.5% of the Chapter 11 debtors in their larger sample. See Warren & Westbrook, Financial Characteristics, supra note 5, at Fig. 3. There is no information which allows one to calculate how many of this size debtor is in the Contracting Out sub-sample, though given the information provided in terms of total debts for the entire sub- sample, the figure is likely to be less than 5%. " See Warren & Westbrook, Contracting Out, supra note 1, at 1250- 51. These numbers again reflect the truncated nature of Warren and Westbrook's sample. When Baird, Bris and Zhu examined all corporate bankruptcies filed between 1995 and 2001 in the Southern District of New York and the District of Arizona, looking for cases that ended in a confirmed Chapter 11 plan, they found, on average, over 160 unsecured creditors in each case. See Baird et al., supra note 59, at tbl. 1. 82 See Warren & Westbrook, Contracting Out, supra note 1, at 1251- No. 1: 179] COLUMBIA BUSINESS LA W REVIEW (the number of creditors) to conclusion (substantial transaction costs), lacks a solid theoretical foundation. The inference reflects a concern that, in a regime of bankruptcy choice, each creditor will have to gather information about the bankruptcy rules that the debtor has selected and then negotiate with the debtor over them.83 This is a concern about transaction costs and, as such, it contains assumptions about how bankruptcy choice would operate in practice. The first assumption is that creditors would need to ascertain what rules the debtor has chosen. This assumption is undoubtedly correct, but trivial. Bankruptcy choice proposals all recommend that the state supply a set of bankruptcy rules from which the debtor could select.84 A creditor would simply need to learn which set of rules had been selected and price its terms accordingly. This effort is the kind of inquiry that creditors make on a regular basis. Creditors need to assess various aspects of the debtor's business in order to determine the likelihood of repayment. Indeed, many creditors rely on ratings agencies for information about the likelihood of repayment. Given the amount of information that ratings agencies already collect on the companies that they monitor, the added cost of each company's bankruptcy choice selection would be negligible. Warren and Westbrook fail to explain why asking about the debtor's chosen bankruptcy regime would pose any challenge to creditors different from what they routinely do before making a loan. Warren and Westbrook assert that there would be additional transaction costs arising from the need to negotiate over the insolvency rules selected. While they are not explicit on the point, this seems to be where they believe the bulk of the costs lie.85 Bankruptcy choice proposals, '3 See id. at 1249. 4 See supra note 15. 85 See Warren & Westbrook, Contracting Out, supra note 1, at 1249 ("If, however, the reality of business bankruptcy is that most debtors have many claimants and that any negotiations will have to take place in a rented hall, then the efficiency gains from contract bankruptcy quickly [Vol. 2007 EMPIRICALLY BANKRUPT however, have been sensitive to (and have expressly addressed) this concern. Bankruptcy choice, in all of its permutations, operates by the debtor selecting the appropriate insolvency rule, and the creditor pricing the extension of credit accordingly. (What they differ over is whether that selection is made in the corporate charter or in the lending documents themselves.) In competitive credit markets, one cannot systematically exploit creditors who can adjust their lending behavior. A term that lessens their recovery in bankruptcy will cause them to raise their interest rate. In expectation, they receive a competitive rate of return regardless of which choice a debtor makes. 6 There is no advantage for creditors to negotiate extensively here. They simply need to ascertain the governing bankruptcy choice and price their credit appropriately. Creditors have nothing to gain by hammering out with the debtor which regime should be chosen. The number of creditors, therefore, does not necessarily translate into transaction costs that loom large. The contrary inference that Warren and Westbrook draw depends on a caricature of bankruptcy choice. On the other half of the cost/benefit equation, Warren and Westbrook do not evaluate any possible benefits of bankruptcy choice before concluding that the transaction costs outweigh them. They offer no evidence on this point whatsoever. Had they produced data on the transaction cost issue, they would be in a position to at least set the parameters for the benefits that bankruptcy choice must generate to counsel its adoption. To be sure, measuring efficiency gains presents a challenge, but one does not have to look all that hard to find reasons to suspect that gains can be had. Bankruptcy choice rests on the assumption that it would provide a better sorting system for companies in financial distress than current law does. Some corporations would find themselves fade, overwhelmed by the negotiating costs of dealing with many creditors."). 8" That creditors are in competitive markets is a standard assumption in the literature, and Warren and Westbrook do not take issue with it. No. 1: 179] in regimes that handle financial distress more efficiently than Chapter 11. Proponents of bankruptcy choice suggest that alternative regimes will spare some of the deadweight costs seen under the existing Chapter 11. Indeed, eliminating even some of these costs would bring substantial benefits. Consider extreme cases like Eastern Airlines, where upwards of a billion dollars appears to have been wasted.17 To give another example, Merry-Go-Round is a case in Warren and Westbrook's initial dataset where $100 million dollars in cash went out the door during Chapter 11.88 A substantial number of cases such as these are initially filed in Chapter 11 but ultimately are either converted to Chapter 7 or dismissed. This suggests that these debtors are being put into the wrong type of insolvency proceeding. Were these cases to start out under a more appropriate set of rules, they would probably be resolved more quickly, and creditors would see increased returns. Quantifying the gains to be achieved by a counter-factual state of proposed law reform is a challenge, and no one can fault this study for failing to do so. But that does not mean that its authors are entitled to assume, without data or theory, that these gains are less than the costs that would arise in a world of bankruptcy choice. One cannot locate in their work the basis for this assumption. In sum, Warren and Westbrook in Contracting Out make sweeping claims. They claim to have proven that a regime of bankruptcy choice would decrease overall welfare and that scholars should abandon work on such proposals. But the proof is elusive. Their claims rest on a series of fundamental flaws embedded in their project, each one fatal. Rather than looking in the right place for data, they look at a sample comprised half of individuals and virtually no public companies. Rather than asking the right questions, they 17 See Lawrence A. Weiss & Karen H. Wruck, Information Problems, Conflicts of Interest, and Asset Stripping: Chapter 11's Failure in the Case of Eastern Airlines, 48 J. FIN. ECON. 55 (1998). ' See Elizabeth MacDonald & Scott J. Paltrow, Merry-Go-Round: Ernst & Young Advised the Client, but Not About Everything, WALL ST. J., Aug. 10, 1999, at Al. [Vol. 2007COL UMBIA B USINESS L4AWRE VIE W simply assume that creditors in their sub-sample received substantial recovery when other empirical evidence strongly suggests the opposite. Rather than drawing the right inferences, they attribute large negotiation costs to bankruptcy choice without support in either theory or data, and they infer that these costs would exceed the benefits of the new law without making any effort to ascertain what benefits would flow from permitting companies to select their own insolvency rules. Like all research and analysis, empirical research and analysis must cross a minimum threshold to generate insight and value. In part, this is accomplished through authors self-consciously policing their own assumptions, choices and inferences. Equally important is external critique. Flaws in empirical arguments need to be included in the discourse. It is incumbent on scholars like Warren and Westbrook to be more careful and circumspect than they have been in their use of empirical data. It is incumbent on the rest of us to subject work such as this to rigorous review. The idea of "data" putting an end to complex theoretical and policy disputes is alluring-too alluring in this case. Contracting Out may be an entertaining read for those predisposed to accept its conclusions, but the data it offers bring us no closer to resolving the debate. III. STEVE LUBBEN AND EQUITY RECEIVERSHIPS Whereas Warren and Westbrook attempt to use empirical analysis to evaluate a proposed change in governing law, Lubben turns to data to scrutinize the past. Bankruptcy scholars have recently shown renewed interest in the equity receiverships that were created to handle the financial distress of the nation's railroads. The accepted wisdom since the New Deal is that investment bankers dominated these railroad organizations in order to benefit themselves and their clients at the expense of the general public.89 As one " See Douglas G. Baird & Robert K. Rasmussen, Boyd's Legacy and Blackstone's Ghost, 1999 SuP. CT. REV. 393, 408-17 (describing New Deal hostility to equity receiverships). EMP1RICALL Y BANKRUPTNo. 1: 1791 COLUMBIA BUSINESS LAW REVIEW critic colorfully put it in 1938, "Railroad reorganization was a racket many years before the word 'racket' was coined; and thousands of investors have paid tribute to it with the loss of their fortunes."" Recent work, however, suggests that these receiverships performed about as well as could be expected. No one has asserted that railroad reorganizations were perfect. The problems raised by insolvent railroads were novel, and it would be shocking if those responsible for developing this mechanism had hit upon the ideal solution right out of the box.9 Rather, the new scholarship attempts to highlight the ingenuity behind the process. Receiverships were by no means perfect, but neither were they the corrupt mechanism for fleecing the unsophisticated as the New Deal reformers alleged.92 A. Lubben on the Effectiveness of Railroad Receiverships Against the backdrop of these divergent views about equity receiverships, Stephen Lubben investigated the extent to which railroad reorganizations were "effective."" 90 HAROLD PALMER, INVESTMENT SALVAGE IN RAILROAD REORGANIZATIONS 1 (Beard Books Inc. 1938). 91 Lubben concludes that "receiverships were lengthy and perhaps quite expensive by modern standards." Lubben, supra note 2, at 1452. While making this observation, he never defends the proposition that one can glean any insights by comparing the length and cost of receiverships with those of today's Chapter 11. Improvements in information technology, the thickening of capital markets, and the learning that comes with decades of experience render comparisons such as the ones Lubben makes meaningless. 92 See DAVID A. SKEEL, JR., DEBT'S DOMINION 56-69 (Princeton 2001); Douglas G. Baird & Robert K. Rasmussen, Control Rights, Priority Rights, and the Conceptual Foundations of Corporate Reorganizations, 87 VA. L. REV. 921, 931-32 (2001) [hereinafter Control Rights]; Baird & Rasmussen, supra note 89, at 403-06; David A. Skeel, Jr., The Past, Present and Future of Debtor-in-Possession Financing, 25 CARDozo L. REV. 1905, 1908-13 (2004). 9 See Lubben, supra note 2, at 1423 ("railroad receivership offers a poor example of effective corporate reorganization"), 1452 ("were they [Vol. 2007 EMPIRICALLY BANKRUPT His study concludes that the data he has gathered shows that they were not.94 Although the data is sound, the conclusions are suspect. One simply cannot get from the interesting data Lubben collected to the conclusions he puts forth. If anything, the data point against his argument condemning railroad receiverships. Lubben collected data on fifty-three railroads. Equity receiverships were developed to handle the financial distress of large railroads, so Lubben's study limits the criteria to railroads of more than five hundred miles in length. Also, since Lubben wanted to assess the effects of equity receiverships that took place between 1890 and 1937, he defines his population as all railroads that had more than five hundred miles of track in 1900 and that still existed as separate legal entities in 1937.95 Lubben divides his sample into those railroads that were reorganized in an equity receivership at least once between 1890 and 1917 and those that were not. Lubben gathers data on two fronts. He first compares the capital structure of those railroads that had been through an equity receivership during this time period with those that had not. He next examines whether a railroad that went through a receivership in the first period (1890 to 1917) was more likely to file for receivership in the second period (1921 to 1937) than was a railroad that did not undergo a receivership in the first period. effective?"), 1464 ("The Effectiveness of Receiverships-Regression Analysis"). 94 Id. at 1473. " Fifteen railroads that had 500 or more miles of track in 1900 disappeared by 1937. Id. at 1453-54. One tantalizing fact which Lubben does not explore is that of the fifteen railroads which drop out of his sample because they were acquired by other railroads, only one had gone through a receivership in the pre-war period. See id. In other words, reorganized railroads were not attractive acquisitions. There was something about going though a receivership that made these railroads toxic. As discussed below, it may be that the railroads that went through receivership early had less desirable routes, which would explain both why they encountered financial distress earlier and why other railroads did not find them attractive takeover candidates in a contracting market. No. 1: 1791 As to the capital structure of the reorganized railroads, Lubben finds that receiverships reduced a railroad's fixed charges (its obligations on its securities) by more than twenty-five percent. 6 Lubben compares the capital structure of those railroads that had undergone a receivership with those that had not for the period between World War I and World War II. He finds that the fixed charges of both sets of railroads, when compared to each railroad's total income, are virtually identical. 8 Receiverships thus returned the capital structure of the distressed railroad to the industry average. Despite having similar capital structures, the railroads that had undergone a receivership prior to the Great War were more likely to experience a receivership between the wars than were those railroads that had not been in a receivership before 1917. Lubben finds that railroads that had gone through receivership were roughly two and a half times as likely to go through receivership in the interwar period as railroads that had never done so before.9 Lubben concludes that this recidivism rate "throws into question the efficacy of the receiverships that occurred between 1890 and the United States' entry into World War I. " 10 The problem, according to Lubben, is that receiverships did not trim sufficient debt. "[RIeceiverships were not designed to provide railroads with optimal capital structures, but rather with typical capital structures such as those that might be found in a non-bankrupt railroad."10' Hence, receiverships were not effective. 9 See id. at 1462. There is thus a risk that Lubben has not captured fully the dynamics of the receivership process. A railroad's capital structure may have changed between the time it left the receivership and the interwar period that Lubben measures. 98 Lubben, supra note 2, at 1462-63. " The difference is significant at the 90% confidence level but not at the 95% confidence level. See id. at 1465, tbl. 11. 100 Id. at 1466. 101 Id. at 1462. COL UMBIA B USINESS LA W RE VIE W [Vol. 2007 B. Selecting the Appropriate Baseline Lubben's project is to evaluate the past. He wants to assess the effectiveness of equity receiverships. Unlike Warren and Westbrook, Lubben looks in the right place. He has compiled a census of railroads over five hundred miles in length. While equity receiverships eventually came to be used for different types of businesses, they arose to meet the financial distress of the railroad industry. Railroads were thus the primary, though not exclusive, clientele.12 Lubben, in part, has the right focus: he looks at the capital structure of the railroad after reorganization. The problem presented by railroads is that they cost more to build than they were worth. 0 3 Once they were built, however, the assets were best used in their current configuration. Thus, the primary challenge of the receivership was to create a capital structure that better reflected the operation of the business." Lubben also explores the fate of the reorganized entities, again a relevant question. Those who devised equity receiverships sought to bring the railroad's obligations in line with its revenues. To see whether these reorganizers succeeded, Lubben assesses the performance of the railroads after reorganization and determines whether a reorganized railroad filed for reorganization a second time. Lubben goes astray, however, in evaluating the performance of receiverships against a standard that no bankruptcy system has yet to meet. By asking whether receiverships were "effective," he both asks the wrong question and draws the wrong inference. To be "effective" for Lubben, the railroads reorganized between 1890 and 1917 would have to seek reorganization in the second period at the same rate as those which had no prior receivership. An initial problem is that the lag between the first 102 On the expansion from railroads to other industries, see SKEEL, DEBT'S DOMINION, supra note 92, at 104-05; Baird & Rasmussen, supra note 89, at 408-10. 103 See Baird & Rasmussen, Control Rights, supra note 92, at 925-27. '04 Id. at 927-33. No. 1: 179] EMPIRICALL Y BANKRUPT reorganization and the second was quite long. On average, it was sixteen and a half years.0 5 This extended period of time raises questions as to the causal relationship that Lubben posits. It would take some foresight to spot problems that would crop up a decade and a half after the first reorganization. Indeed, the railroads that underwent a subsequent receivership only did so after the onset of the Great Depression. It borders on the incredible to suggest that those conducting receiverships prior to the end of World War I could have designed a capital structure that would have insulated the reorganizing railroad from the ravages of the Depression." 6 More problematic, however, is the standard for "effectiveness" itself. Lubben posits that an effective bankruptcy law would leave reorganized companies with the same chance of reorganizing a second time as all companies have of needing reorganization in the first instance. Current law is clearly not effective against this metric. Regardless of the time period selected, when one looks at companies that reorganize under Chapter 11, they are more likely to file a second case than is another company likely to file a first. '07 This pattern is not surprising. Companies file a Chapter 11 case because they are in financial distress. All things being equal, one would expect that a group of companies that 105 Lubben, supra note 2, at 1466, n.210. '0 The start of Lubben's second period coincides with the general decline of the entire railroad industry. See JAMES W. ELY, JR., RAILROADS & AMERICAN LAW 265 (Kansas 2001) ("After World War I, the rail industry entered a prolonged period of contraction and stagnation."); JOHN F. STOVER, AMERICAN RAILROADS 192 (2nd ed., Chicago 1997) ("Ever since the First World War, American railroads have experienced a general decline. The year 1916 in several ways marked the end of the golden age of railroads."). 107 The probability of any public company filing for Chapter 11 is less than one percent, and the odds of a company that at one time had gone through Chapter 11 filing again is much higher. See Lynn M. LoPucki & Sara Kalin, The Failure of Public Company Bankruptcies in Delaware and New York: Empirical Evidence of a "Race to the Bottom," 54 VAND. L. REV. 232, 242 (2001) (Background rate of 0.84%), 245 (refiling rate after first bankruptcy of more than three-and-a-half times background rate). COLUAMA B USINESS LA W RE VIE W [Vol. 2007 EMPIRICALLY BANKRUPT have experienced financial distress to be more likely to encounter a second bout than a group of companies that had no such prior problems. It is a bit like going to the doctor for surgery. We would not be surprised to learn that people who have had surgery are more likely to have a second operation than people who never had surgery are to have a first one. There is an obvious selection effect at work. The control sample and the treatment sample are created through the use of a variable that we expect to be highly correlated with the effect being investigated. Simply put, it is doubtful whether any bankruptcy system could ever ensure that a reorganized company would have the same probability of filing for bankruptcy as a company that had never had financial difficulties. This is especially true in the railroad context. One cannot redeploy a railroad's assets nimbly. The route that the railroad takes is essentially fixed. One can prune operations by selling some track here and there, but the basic contour of the line is fixed. To the extent that the necessity for a receivership prior to 1917 suggests a less desirable route structure and hence a greater likelihood of financial distress, there is little that can be done to alter this aspect of the enterprise. The future is uncertain, and these railroads may have been able to prosper had the demand for railroads increased. However, when demand fell, one would expect that these railroads would be more vulnerable." 8 Asking the wrong question leads Lubben to draw the wrong inferences. Consider the inference that he draws from the data which shows that reorganization resulted in the railroad having a capital structure similar to those that prevailed in the industry. Lubben concludes that this data suggests that the reorganizations were not effective, but one 108 Lubben argues differently. He asserts that "[if receiverships effectively resolved a railroad's financial problems, we would expect that [the railroads that underwent a receivership before 1917] would encounter financial distress as often (or even less often) as the railroads [that did not]." Lubben, supra note 2, at 1464. Yet, he never says why this is so. He never articulates a way in which the railroads that did need a second reorganization could have in fact been saved. No. 1: 179] can draw this inference only by asking the wrong question. Those negotiating the new capital structure for a railroad in receivership had information about the current operating revenues and some prediction about future revenues as well. They had to decide how much debt the reorganized railroad should carry. As Lubben demonstrates, they seem to have chosen to give the reorganizing railroad a debt load similar to the debt carried by healthy railroads. One would be hard pressed to gainsay this decision. The receiverships developed because the railroads could not service their debts. The investment bankers and lawyers transformed the device of equity receivership into one that could be used to alter the railroad's capital structure. In deciding how much debt the railroads could handle, mimicking the capital structure of successful railroads seems to be a sensible place to start. If other railroads could avoid financial distress with such a capital structure, perhaps the reorganizing railroads were wise to follow suit. Lubben, however, suggests that those orchestrating the reorganization should have taken a different route. He believes that the process should have put an "optimal capital structure" in place. Rather than following the example set by other railroads, the reorganizers should have aimed for a capital structure that better fit the railroad they were reorganizing. Yet, precisely how were they to know what the "optimal capital structure" was? Lubben finds fault with the capital structure that the reorganizers selected because these railroads needed a subsequent reorganization at a higher rate than other railroads. This information, of course, was not available to those drawing up the capital structure for the reorganizing railroads. There is an even deeper problem. Even today economists do not know what constitutes an optimal capital structure. Modigliani and Miller in 1958 famously demonstrated that in a frictionless world, the mix of debt and equity does not affect the value of the corporation. 9 Ever since, economists 109 See Franco Modigliani & Merton H. Miller, The Cost of Capital, Corporate Finance and the Theory of Investment, 48 AM. ECON. REv. 261 (1958). COL UMBIA B USINESS LA W RE VIE W [Vol. 2007 EMPIRICALLY BANKRUPT have offered reasons as to why a corporation may choose debt or equity, such as bankruptcy costs, tax benefits and ease of financing.11 ° Robust debates about the determinants of capital structures continue to this day. Lubben certainly cannot point to any knowledge available at the time of the equity receiverships that would have told the participants what would constitute an optimal capital structure for their railroad. It is a bit like faulting Newton for not anticipating the work of Einstein. There is much to learn from Lubben's data. It provides us with more information on a crucial aspect of railroad reorganization, namely, the extent to which they altered the capital structure of insolvent railroads. We cannot conclude, however, that this data convicts the equity receiverships on the charge of being ineffective. If anything, his data points to the opposite conclusion. IV. LYNN LOPUCKI AND BANKRUPTCY COURT COMPETITION Whereas Warren and Westbrook look to the future and Lubben to the past, LoPucki sets his sights on the present. His empirical claims relate to the present state of American corporate reorganization practice. Starting in the late 1990s, bankruptcy scholars began to focus on the competition that seemed to be taking place among bankruptcy courts for large Chapter 11 cases.' Some took the position that this 0 See, e.g., Franco Modigliani & Merton H. Miller, Taxes and the Cost of Capital: A Correction, 53 AM. ECON. REV. 433 (1963); Merton H. Hiller, Debt and Taxes, 32 J. FIN. 261 (1977); Henry DeAngelo & Ronald W. Masulis, Optimal Capital Structure Under Corporate and Personal Taxation, 8 J. FIN. ECON. 3 (1980); Michael Bradley, G. Jarrell & E.H. Kim, On the Existence of an Optimal Capital Structure: Theory and Evidence, 39 J. FIN. 857 (1984); Stewart Myers, Determinants of Corporate Borrowing, 5 J. FIN. ECON. 147 (1977); Eugene F. Fama & Kenneth R. French, Testing Trade-off and Pecking Order Predictions About Dividends and Debt, 15 REV. FIN. STUD. 1 (2002). .. See Theodore Eisenberg & Lynn M. LoPucki, Shopping for Judges: An Empirical Analysis of Venue Choice in Large Chapter 11 Reorganizations, 84 CORNELL L. REV. 967 (1999); Robert K. Rasmussen & Randall S. Thomas, Timing Matters: Promoting Forum Shopping by No. 1: 1791 COLUMBIA BUSINESS LAW REVIEW competition would be destructive; others that it would be beneficial; and still others that it would be something of a mixed bag."' LoPucki claims that his data has put an end to this theoretical uncertainty. A. Competition and Corruption LoPucki asserts that his data leaves no doubt that this competition has led to a bankruptcy system that is both "corrupt" and failing. "Corruption" is a serious charge, and it is a term that LoPucki uses deliberately."' He does not mean that bankruptcy judges are on the take, at least not directly. But it is close. LoPucki asserts that "[t]heir actions are 'corrupt' in that they are dictated not by an attempt to apply the law to the facts of the case but by the need to remain competitive."14 They reflexively heed the requests of the "debtor's executives, professionals, and DIP lenders"11 5 rather than applying the dictates of the Bankruptcy Code. When these actors, who are in control of the debtor, make requests, they are granted without a second thought. The reason is that it is these groups who decide in which venue the debtor will file, and judges need to placate them in order to attract large cases. In other words, the judges are not making a good faith attempt to apply the law. Not following the law is bad enough, but the results are even worse. Perhaps adopting a strained reading of the Insolvent Corporations, 94 Nw. U. L. REV. 1357 (2000); David A. Skeel, Jr., Bankruptcy Judges and Bankruptcy Venue: Some Thoughts on Delaware, 1 DEL. L. REV. 1 (1998). 112 See Eisenberg & LoPucki, supra note 111 (skeptical about competition); Rasmussen & Thomas, supra note 111 (arguing that competition may be beneficial for prepackaged cases but deleterious for traditional ones); Skeel, supra note 111 (arguing that the competitive pressures would improve the bankruptcy process). .. See LoPucKi, supra note 3. "' Id. at 137; see also Lynn M. LoPucki, "Corruption is the Right Word," BANK". CT. DEC., July 19, 2005, at A7. 115 LoPUCKI, supra note 3, at 138. [Vol. 2007 Bankruptcy Code could be justified on pragmatic grounds."' The bankruptcy judges, drawing on their expertise, would adopt the interpretation most likely to facilitate a successful reorganization. Yet, this has not occurred. The effects of modern reorganization practice are, according to LoPucki, disastrous. Kowtowing to those who decide where large reorganizations will be filed has "destroyed companies that could otherwise have been saved.""7 Moreover, it has spawned a number of practices which, in LoPucki's view, will make matters even worse. Things are bad, and they promise to get worse; there is no hope for a turnaround."8 Few seriously question that at least some bankruptcy courts compete for cases.119 The debate has been over the effect of this competition. The linchpin in LoPucki's argument is his claim that the Bankruptcy Court for the District of Delaware, the leader in the competition for large cases, handles cases in such a way as to destroy what could have been valuable companies. LoPucki considers companies that completed their stay in Chapter 11 from 1991 through 1996.120 This is the period during which the Delaware bankruptcy court established itself as the venue of choice for large corporations. Prior to this period, companies had 116 On using pragmatic concerns in statutory construction, see RICHARD A. POSNER, THE PROBLEMS OF JURISPRUDENCE 299-309 (Harvard 1990); William N. Eskridge and Philip P. Frickey, Statutory Interpretation as Practice Reasoning, 42 STAN. L. REv. 321 (1990). 117 LoPucKi , supra note 3, at 258. 118 Indeed, somewhat apocalyptically, LoPucki argues that the destructive competition among bankruptcy courts is going to engulf the world. See LOPuciK, supra note 3, at ch. 8 ("Global and Out of Control?"). 19 See Rasmussen & Thomas, supra note 111, at 1369. 120 It is clear why LoPucki begins his study with the 1991 calendar year. That is the year in which the Delaware bankruptcy court began receiving a large number of Chapter 11 cases where jurisdiction was based solely on the fact that one member of the corporate group was incorporated in Delaware. The reason that LoPucki chose 1996 as the cutoff date is that he first began to look at recidivism rates in 2001, and he explored whether the corporation refiled within five years of emerging from bankruptcy. See Lynn M. LoPucki & Sara D. Kalin, The Failure of Public Company Bankruptcies in Delaware and New York: Empirical Evidence of a "Race to the Bottom," 54 VAND. L. REV. 231, 250 (2001). No. 1: 179] EMPIRI CALL Y BANKRUPT COLUMBIA BUSINESS LA W REVIEW preferred filing in the Southern District of New York. The metric by which LoPucki assesses the performance of a bankruptcy court is whether, if a company emerges from a Chapter 11 proceeding, that company files a second bankruptcy case within the ensuing five years. According to LoPucki's data,'21 we see the following result: DE SDNY Other Total Cases 32 27 99 158 completed Cases with 28 19 77 124 emerging company Cases with 11 (39%) 3 (16%) 4 (5%) 18 (11%) emerging company that refiled in 5 years The problem is obvious. Companies that reorganized in Delaware need a second reorganization at rates substantially higher than companies that reorganized elsewhere. This difference is statistically significant. "From the data it appears that if the Delaware-reorganized companies had filed in other courts, many more of them would have survived."22 Yet, the problem is even worse according to LoPucki. It is not only that Delaware was sick, but that it was contagious. By the mid-1990s, bankruptcy professionals had realized that large, publicly held corporations were demonstrating a 121 In his writings LoPucki reports data based at times on "public" companies that emerged from Chapter 11 and at other times on "companies" that emerge, which includes both public and private companies. Compare LoPuc I , supra note 3, at 113 tbl. 6 (using public companies emerging), 100 (using companies emerging). The numbers in the text are for all companies because there is no reason to think that the problems that LoPucki sees with Delaware are limited to cases where a public, rather than a private, company emerges. 122 LoPucKi, supra note 3, at 118. [Vol. 2007 marked preference for Delaware. Some bankruptcy judges outside of Delaware, at times prodded by the local bankruptcy attorneys, responded by mimicking the Delaware practices. The hope was to attract large cases to their venue. Local attorneys can expect significant additional work when a large case ends up in their backyard. At the time, the failure rate of Delaware reorganizations was only beginning to manifest itself, and no one recognized the looming problem. When other courts adopted Delaware practices, "they reproduced Delaware's failure."123 In the four years between the beginning of 1997 and the end of 2000, Delaware's performance as measured by refiling rate remained substantially unchanged, but the failure rates of the other courts caught up with that of Delaware: DE SDNY Other Total Cases 63 14 30 117 completed Cases with 38 10 20 68 emerging company Cases with 16 (42%) 4 (40%) 7 (35%) 27 (40%) emerging company that refiled in 5 years Thus, Delaware was a failure. Other courts emulated the Delaware practice, and they became failures as well. This failure drives the remainder of LoPucki's analysis. It reveals the ills of competition. Hence, any subsequent developments in reorganization practice are the result of competition and these results are presumptively objectionable. 123 Id. at 122. No. 1: 1791 EMPIRICALL Y BANKRUPT B. The Need to Look in the Right Place LoPucki has defined his population so that he could conduct a census rather than take a sample. He is only interested in the bankruptcy of large, publicly held companies."" He defines "large" as a corporation that files for Chapter 11 and lists at least $100 million in assets, as measured in 1980 dollars. He defines "public" as any corporation that has publicly traded securities, be they stocks or bonds. Focusing on large companies is a justifiable restriction on LoPucki's project. As discussed above, the economic effects of Chapter 11 are concentrated in the large cases. Thus, to the extent that LoPucki is interested in the economic impact of Chapter 11, he will pick up a large portion of that impact by looking at large cases. While $100 million in assets in 1980 dollars is at some level arbitrary, it is a reasonable place to divide the cases. It is fair to assume that not much is lost by omitting publicly held companies with fewer assets.125 Limiting the study to public companies is understandable as well. Public companies, by definition, have reporting requirements. Being a public company means having to file various disclosures with the SEC. These disclosures provide useful information on the financial condition of the business. 124 LoPucki has long been a leader in exploring the dynamics of large company reorganizations. See, e.g., Lynn M. LoPucki & William C. Whitford, Bargaining Over Equity's Share in the Bankruptcy Reorganization of Large, Publicly Held Companies, 139 U. PA. L. REV. 125 (1990); Lynn M. LoPucki & William C. Whitford, Patterns in the Bankruptcy Reorganization of Large, Publicly Held Companies, 78 CORNELL L. REV. 597 (1993). 125 For example, in 2004, ninety-two public companies holding a combined $47 billion in assets filed for Chapter 11. See THE 2005 BANKRuPTcY YEARBOOK & ALMANAC 28 (New Generations Research 2005). LoPucki's database for that year contains thirty cases with roughly $32 billion in assets. [Vol. 2007COL UMBIA B USINESS LA W RE VIE W To be sure, there are some large privately held businesses that file for bankruptcy, but these are few and far between.12 Things become more contestable when one explores the types of cases LoPucki considers. LoPucki's empirical results depend entirely on his decision to combine different types of bankruptcy cases. When academics, lawyers and the popular press consider large Chapter 11 cases, they focus on cases where a corporation files for bankruptcy and, while in bankruptcy, a decision is made about the future of the business. The headline grabbers of the past, such as Johns- Manville, Eastern Airlines, LTV, Texaco, WorldCom, and Enron, have been involved in cases of this type. For want of a better term, these cases have been called "traditional" cases.'27 When LoPucki catalogues the ills that he finds rife in current bankruptcy practice, he focuses primarily on these traditional cases. 28 When we look at all cases in LoPucki's dataset, of the over seven hundred cases it contains, over seventy-five percent are traditional. The remaining twenty-five percent of Chapter 11 cases in LoPucki's dataset consist of so-called prepackaged cases and pre-negotiated cases. In a prepackaged case, the debtor works with its major creditors and crafts a plan of reorganization before a bankruptcy petition is filed. The debtor also solicits sufficient acceptances to ensure that the 126 For example, New Generations Research reports seven private cases in 2002 with more than $100 million in nominal assets. See THE 2003 BANKRUPTCY YEARBOOK & ALMANAC 78-80 (New Generations Research 2003). LoPucki's database, in contrast, reports that there were eighty-one public companies that filed for bankruptcy that had more than $100 million in 1980 dollars. 127 "Traditional" is something of a misnomer. As Baird and I have explained elsewhere, corporate reorganization practice bears little semblance to the practice of decades ago. See Douglas G. Baird & Robert K. Rasmussen, The End of Bankruptcy, 55 STAN. L. REV. 751 (2002); Baird & Rasmussen, supra note 50, at 675-85. Still, I would rather use the term "traditional" than burden the reader with "nonprepackaged, nonnegotiated." 128 For example, he begins his book with a discussion of the Enron bankruptcy. See LoPuCiu, supra note 3, at 9-16. Of the practices that LoPucki labels as "corrupt," only one applies in prepackaged cases; the remaining six center on traditional ones. See id. at 139-81. No. 1:179] EMPIRICAL LY BANKRUPT plan will be approved. These plans tend to reduce the company's debt and make little or no change to its operations.'29 Indeed, the parties may first attempt an out- of-court workout, but turn to bankruptcy if the debtor cannot convince a sufficient number of debt holders to compromise their claims.13 ° To avoid a dissent that would impede quick implantation of the plan, prepackaged plans by and large pay trade creditors in full. Indeed, trade debt in these cases tends to be "unimpaired," which means that this debt is paid in the ordinary course of business. 3' In LoPucki's database, there are sixty-one of these cases, which comprise a little less than nine percent of the overall sample. The frequency of pre-packaged cases has fluctuated over time. In 2003, such cases had declined to less than five percent of cases involving companies with publicly traded securities, though for the period that LoPucki studies (1991-1996) they made up over twenty percent of the sample."2 The remaining cases, roughly fifteen percent of LoPucki's database (but only about six percent of the cases between 1991 and 1996), are pre-negotiated cases. In these cases, the debtor reaches agreement with its major creditors about what will happen during the Chapter 11 proceeding. The ends to which pre-negotiated plans are put have expanded over time. Originally, and continuing throughout the 1991 to 1996 period that forms the heart of LoPucki's empirical claims, pre-negotiated cases were similar to prepackaged ones in that few operational issues were addressed as part of the bankruptcy process, and the primary purpose of the proposed plan of reorganization was to reduce the company's debt level. The company would reach an agreement with its '29 See WEIL, GOTSHAL & MANGES, LLP, REORGANIZING FAILING BUSINESSES, at 12-20 (ABA 1998) ("This technique is practical only in those situations in which the debtor's financial distress primarily is traceable to burdensome debt levels and the company does not need a comprehensive rehabilitation of its business operations."). 130 See id. at 12-16. 131 See 11 U.S.C. § 1124 (2005). 132 See THE 2004 BANKRUpTCY YEARBOOK & ALMANAC 163 (New Generations Research 2004). [Vol. 2007COLUAMA B USINESS LA W RE VIE W EMPIRICALLY BANKRUPT major creditors, but would not obtain enough acceptances prior to bankruptcy to ensure that the plan would be approved. Some of the details of the financial reorganization still had to be ironed out.' Today, however, pre-negotiated cases may include situations where the debtor has decided to sell itself, and the bulk of the proceeding is concerned with dividing up the proceeds."' Thus, in LoPucki's dataset, three-fourths of the cases are traditional cases and one-fourth is prearranged. LoPucki's claims about competition and corruption necessarily are about traditional cases. Judges wield much more power here than in prepackaged and pre-negotiated cases. Judges have the power to keep control of the case in the hands of the debtor. The Bankruptcy Code at the time granted to the debtor the exclusive right to file a plan of reorganization for the first six months of the case, but the bankruptcy judge has the power to extend this period of exclusivity.'35 Bankruptcy judges also rule on contested motions, such as whether the debtor's bankruptcy financing package should be approved, whether disgruntled shareholders can oust the board of directors and whether pension plans can be jettisoned.13 Bankruptcy judges also entertain first day 133 For example, in the Grand Union case that was filed in Delaware in late January of 1996, the debtor's corporate structure was that there was a holding company whose sole asset was the stock of the operating company. The holding company had issued debt, as had the operating company. Roughly two months before the bankruptcy filing, Grand Union entered into negotiations with its bond holders. At the time it filed for bankruptcy, it had reached an agreement with those who held bonds issued by the operating company but had yet to come to terms with those who held bonds issued by the parent company. An agreement with these bondholders was reached shortly after the company filed for bankruptcy, and the reorganization plan was confirmed in late May of 1996. 131 See Baird & Rasmussen, supra note 50, at 675-85 (describing the dynamics of modem Chapter 11 practice). 135 See 11 U.S.C. § 1121 (1994). Amendments in late 2005 limited the ability of the bankruptcy court to extend exclusivity more than eighteen months after the case was filed. See 11 U.S.C. §1121(d)(2) (2006). 136 See 11 U.S.C. § 364 (2006) (court approval for financing); 11 U.S.C. § 363 (2006) (court approval for transactions outside the ordinary course of No. 1: 179] COLUMBIA BUSINESS LAW REVIEW orders which often establish the ground rules for the case.137 Traditional reorganization cases present the complex challenges that judges are suspected of seeking. LoPucki (and others) have presented cogent arguments as to why courts would compete for traditional cases; he does not, however, offer any reason to suspect that judges would compete for prearranged cases. Traditional cases also offer more interest to lawyers. Not only are there more of them, but also each case generates more fees. 3 They are longer and have more disputed matters and thus require more effort on the part of lawyers. More importantly, local attorneys benefit more from a prolonged Chapter 11 taking place in their hometown than they do from a brief prepackaged reorganization. There are many more opportunities to serve as local counsel on a contested Chapter 11 case than there are in prearranged cases that go smoothly through the system. To the extent that lawyers are looking for big paydays, traditional cases are the mother lode. Managers should also be more sensitive to traditional cases than to prepackaged or pre-negotiated ones. Much has been written about managerial turnover during reorganization cases."' Yet the threat to managers of losing their jobs during a reorganization case looms much larger in a traditional case than in a prearranged one. Prearranged cases tend to be relatively quick affairs that focus on revamping a corporation's balance sheet. Given this, one would expect to see more turnover in traditional cases. business); 11 U.S.C. § 1114 (2006) (court approval for termination of pension plans). ... See DEBRA GRASSGREEN, FIRST-DAY MOTIONS MANUAL (AM. BANKR. INST. 2003). 138 See John J. McConnell et al., Prepacks as a Mechanism for Resolving Financial Distress: The Evidence, 8 J. APP. CORP. FIN. 99, 101 (1996). ' See Stuart C. Gilson, Management Turnover and Financial Distress, 25 J. FIN. ECON. 241 (1989); Lynn M. LoPucki & William C. Whitford, Corporate Governance in the Bankruptcy Reorganization of Large, Publicly Held Companies, 141 U. PA. L. REV. 669 (1993). [Vol. 2007 A second problem involves LoPucki's choice of how to present the data. LoPucki breaks out the results for Delaware and the Southern District of New York and treats the rest of the courts the same. However, when one looks at the Chapter 11 activity in the rest of the courts during the 1991 to 1996 period, it turns out that the Central District of California received traditional cases roughly in line with that of the "Big Two." These three courts dominate traditional reorganization activity during the relevant time period. When we separate out prepackaged cases, pre-negotiated cases and traditional cases, and include the Central District of California as a separate category, we get the following: Traditional Cases DE SDNY CDCA Others Total Cases 12 23 12 67 114 ending Cases 9 16 9 48 82 ending with emerging company Cases 3 (33%) 3 (19%) 2 (22%) 2 (4%) 10 (12%) with refiling No. 1: 179] EMPIRICALL Y BANKRUPT Pre-negotiated Cases DE SDNY CDCA Others Total Cases 5 1 0 5 11 ending Cases 5 0 0 5 5 ending with emerging company Cases 1 0 0 0 1 with refiling Pre-pack ged Cases DE SDNY CDCA Others Total Cases 15 3 2 13 33 ending Cases 14 3 2 13 32 ending with emerging company Cases 7 (50%) 0 (0%) 0 (0%) 0 (0%) 7 (22%) with refiling These tables reveal that the Delaware refiling effect that forms the heart of LoPucki's normative conclusion is really a prepackaged refiling effect. Indeed, Delaware has the same number of traditional cases as does the Central District of California. The Southern District of New York has almost as many of such cases as its two closest competitors combined. To the extent that one focuses on which jurisdictions attracted large cases for traditional reorganizations, this remains a New York story during this period, though Delaware did increase its share of traditional cases in later [Vol. 2007COL UMBIA B USINESS LA W RE VIE W EMPIRICALLY BANKRUPT years, only to see the Southern District reassert its dominance. 140 For traditional cases, the refiling rates are relatively the same across these three districts and noticeably lower elsewhere. The difference between Delaware and the Central District is not statistically significant, and neither is the difference between Delaware and the Southern District. Now, when one adds up the "Big Three" and compares them to the other courts, one can get a statistically significant difference. One may have a story as to why the three busiest districts have refiling rates higher than the rest of the country, but it is not the story that LoPucki is telling. Delaware's refiling rate is statistically indistinguishable from that of the other busy courts. When looking at prearranged cases, one can see that there is a noticeable Delaware effect. Delaware dominates to the extent that it attracts these cases. If anything, the chart above does not reveal the extent of Delaware's dominance in this area. Delaware had no prepackaged cases in 1991, but the Southern District of New York and the other courts had a total of three. Delaware received its first prepackaged case in 1992. During that year, there were nine prepackaged cases nationwide, of which Delaware received three. From 1993 to the end of 1996, however, Delaware received twelve of twenty cases. From 1997 to 2004, Delaware increased its dominance in this area, garnering nineteen of twenty-four cases. Prearranged cases thus drive LoPucki's results. Had the bankruptcy court for the Central District of California been the home to the prearranged cases, we would be talking about the Los Angeles effect rather than the Delaware effect. Locating the driving force behind LoPucki's empirical claims in only a subset of the cases renders his empirical 140 On the rise of Delaware, see Rasmussen & Thomas, supra note 111, at 1372-76. On the current division of cases, see THE 2005 BANKRupTcY YEARBOOK & ALMANAC, supra note 125, at 63 (reporting that in 2004 the Southern District of New York received about sixteen percent of the cases filed by public corporations, and the District of Delaware received about eleven percent of such cases). No. 1: 179] assertions unreliable. For traditional cases, the cases on which bankruptcy law focuses, there is no Delaware effect in terms of refiling rates. C. The Need to Ask the Right Questions It is tempting to quote percentages in this context, but it is also misleading. The number of cases that are examined is small. Delaware has only three traditional cases in the six-year period that end up as repeat filers. To be sure, this is a large percentage of the companies that emerged, but there is still the fact that the overall sample is low. In such a situation, it is difficult to rely on statistical inferences alone. That is especially true when the instrument used, a second bankruptcy within five years of the first, is only a rough proxy of one's interests. Moreover, we know that cases were not randomly distributed. It may be a treatment effect-Delaware reorganization practices cause a subse- quent reorganization-or it may be a selection effect- Delaware receives cases that are more likely to need a second reorganization. LoPucki's data cannot distinguish between the two.' The appropriate response in such a situation is to examine the underlying cases themselves. One needs to identify the cause of the subsequent financial distress and ask whether this cause can be fairly attributed to the actions of the bankruptcy court in the first case. This need to examine these cases is reinforced once one realizes that the optimal refiling rate is above zero."' It has been known in the finance literature for years that debt 141 LoPucki has attempted to see if the companies that file in Delaware have different financial characteristics from those filing elsewhere. See Lynn M. LoPucki & Joseph W. Doherty, Why are Delaware and New York Bankruptcy Reorganizations Failing?, 55 VAND. L. REV. 1933, 1946-57 (2002). While LoPucki could not find a difference, this does not prove that no such difference exists. 142 See Robert K. Rasmussen & Randall S. Thomas, Whither the Race? A Comment on the Effects of the Delawarization of Corporate Reorganizations, 54 VAND. L. REV. 283, 295-97 (2001); Kenneth Ayotte & David A. Skeel, Jr., An Efficiency-Based Explanation for Current Reorganization Practice, 73 U. CHI. L. REV. 425 (2006). COLUMBIA BUSINESS LA WREVIEW [Vol. 2007 serves an important role in a company's capital structure. Debt serves both as a tax shield and as a disciplining device. One cost of debt, however, is that its existence creates the risk of bankruptcy. Also, in a world where the future is uncertain, the amount of debt in the capital structure determines the amount of time the managers will have to see whether the company can operate successfully after bank- ruptcy. Without knowing what the optimal level of refiling is, we cannot draw firm conclusions from statistical analysis alone. To assess Delaware's performance, we need to examine the cases themselves. When we look directly at the question of whether bankruptcy courts in Delaware were destroying cases as LoPucki asserts, one cannot find persuasive evidence on this score. 1. Traditional Cases LoPucki equates a refiling within five years with a failure of the first reorganization. Postulating a correlation between refiling and failure is not outrageous; there are undoubtedly situations where a different course of action could have prevented the second bout of bankruptcy. Bankruptcy law is designed to address the problems associated with financial distress, and a second filing suggests that those problems were not adequately addressed in the first proceeding. While we should expect some refilings even when companies leave bankruptcy with an optimal capital structure, a second filing is hardly the hallmark of a successful reorganization. But it is not necessarily the case that the second bankruptcy petition condemns the first proceeding. It may be that the subsequent failure is unrelated to the first. In other words, the five year window is a proxy. Such is often the case in empirical work; one cannot measure directly what one is interested in, so a proxy that can be measured is chosen. Here, LoPucki's proxy is an attempt to locate cases where the bankruptcy process failed to solve the problems afflicting the business. Yet, as with any proxy, it is a rough guide. If we have a sufficient number of observations, five years may be a suitable proxy. But Delaware in the relevant No. 1:179] EMPIRICALLY BANKRUPT time period had only three traditional cases that refiled in five years. Looking at the history of these three companies, the problems with Delaware are difficult to locate. Consider first the reorganization history of Harvard Industries, a company that emerged from the Delaware bankruptcy court in 1992 and returned to bankruptcy court five years later. Harvard was an automotive parts maker whose major customers were the American Big Three automakers. In 1991, some of its creditors filed an involuntary petition against Harvard in Delaware.' Harvard undoubtedly was in financial distress. Less than three years before filing it had undergone a leveraged buyout, and the net revenue from its operations could not cover the interest payments on its debts. The plan of reorganization that was eventually approved focused on the financial side of the business. It converted $200 million of subordinated debentures into preferred and common stock, and the company emerged from the Delaware Bankruptcy Court in August 1992.' In 1993, a new CEO took the reins at Harvard. The enterprise operated profitably for the next two years. Then it bought Doehler-Jarvis, a maker of cast-aluminum car parts. This acquisition significantly increased Harvard's debt level. Twenty months later, the CEO who had instigated this transaction resigned, and Harvard filed for a second bankruptcy. The automotive press attributed the bankruptcy to problems at the Doehler-Jarvis operation. 4 It strains credulity to attribute the second bankruptcy petition of Harvard to dereliction on the part of the bankruptcy court in the first case. 146 The second case was " See Harvard Industries Files for Chapter 11 Protection, WALL ST. J., May 3, 1991, at A3. 1" See Harvard and Debenture Holders Reach Revised Agreement in Principle, PR NEWSWIRE, April 9, 1992. 145 See Kris Hundley, Harvard Seeks Debt Relief, ST. PETERSBURG TIMEs, May 9, 1997, at 1E. 14 Cf. Harvey R. Miller, Chapter 11 Reorganization Cases and the Delaware Myth, 55 VAND. L. REV. 1987, 2005 (2002) ("the return to [Vol. 2007COL UMBIA B USINESS LA W RE VIE W EMPIRICALL Y BANKRUPT caused by what turned out to be a failed acquisition that was orchestrated by a CEO put in place after the conclusion of the first case. It may have been that the acquisition itself was an improvident decision. It may have been that the acquisition was a good idea at the time, but the company failed to integrate its operations well. Or it may have been that the acquisition was a good idea, but it simply turned out poorly. The market for supplying auto parts for American manufacturers is dwindling, and the past fifteen years have seen substantial consolidation in the industry. 4 v Perhaps the best strategy in such a situation is to attempt to acquire others so that, when the dust settles, you are one of the corporations still standing. Maybe Harvard missed a chance to be a survivor, or maybe it was one the many corporations that, despite best efforts, fell victim to a shrinking market. Whatever the explanation, the second bankruptcy case flowed directly from its ill-fated acquisition engineered by a new CEO and not from any defect in the first proceeding. 8 An even more attenuated relationship between first and second filings can be found in United Merchants and Manufacturers, the second of Delaware's traditional refilers. As its name implies, much of the company's operations was in manufacturing. In the summer of 1990, its then-CEO orchestrated an out-of-court workout to reduce the Chapter 11 by Continental Airlines in 1990 was not the result of a defective Chapter 11 plan in 1986"). 147 Currently a number of auto parts makers, including Delphi, Tower Automotive, Dana and Collins & Aikman are in Chapter 11. 148 Perhaps the second case was a failure. A new CEO was put in charge. He adopted a strategy of trying to move Harvard away from its reliance on the automotive sector. He shed assets and looked for acquisitions. In the end, Harvard filed for a third time in 2001. There was no effort to keep the corporation as a stand-alone entity; instead, it was put up for sale. This has been a tough time for autopart makers. Harvard was also in a vulnerable position due to its reliance on the Big Three. Its sales went down dramatically during the decade. It closed plants but was left with retiree medical expenses. Harvard simply had no going concern value. No. 1: 179] COLUMBIA BUSINESS LAW REVIEW enterprise's debt level.'49 This workout, however, was not sufficient to solve United Merchants' financial woes. Shortly thereafter, in November 1990, United Merchants filed for Chapter 11 in Delaware. By the time it had reached the bankruptcy court, the business had been in a period of decline for years. Revenue had shrunk from $1 billion in the late 1970s to $250 million. Its textile and home furnishing operations faced stiff competition. United Merchants did have one operating unit that was financially vibrant. It owned eighty percent of the stock in Victoria Creations, a manufacturer of jewelry. When United Merchants filed for bankruptcy, however, Victoria Creations did not.15° United Merchants exited the Delaware Bankruptcy Court in 1992. It had sold substantial assets while it was in bankruptcy. Upon exiting Chapter 11, it consisted of three operations-apparel textiles, home furnishings and accessories (Victoria Creations).15' By the middle of 1995, United Merchants had sold all of its assets except for its interest in Victoria Creations and some real estate assets.15 2 Thus, by this time, United Merchants was a shell corporation that only owned a single operation, an operation that had not been involved in the first Chapter 11 case. Victoria Creations, however, was not able to service its debt (some of which it acquired when it guaranteed the obligations of its parent), and its lenders forced both it and United Merchants to file for bankruptcy on February 23, 1996, four and a half years after the first reorganization had ended. One month later, potential buyers for Victoria Creations surfaced. Eventually, the company was sold to a 149 See UM&M Completes Exchange Offer, 160 WOMEN'S WEAR DAILY No. 10, July 16, 1990, at 6. 150 See UM&M, Two Units File Chapter 11, 160 WOMEN'S WEAR DAILY No. 88, Nov. 5, 1990, at 19. 151 See Sidney Rutberg, UM&M Plans to Sell its Apparel Units, 21 DAILY NEWS REC. No. 109, June 5, 1991, at 10. ' See UM&M, Victoria Creations Set Up a Refinancing Plan, WOMEN'S WEAR DALY, Aug. 7, 1995, at 19. [Vol. 2007 group led by the founder of the company.'53 United Merchants' second Chapter 11 case, like that of Harvard Industries, was filed more that four and a half years after the first and centered on a company that was not even a part of the first reorganization. The delay between the first and second petitions in Harvard Industries and United Merchants illustrates the dangers that arise when one uses proxies with a small sample. LoPucki finds a statistically significant effect for Delaware when he compares the refiling rate for Delaware- reorganized companies with companies that reorganized in courts other than the Southern District of New York. Yet this result depends crucially on the five year window. As we have seen, both Harvard Industries and United Merchants had a second reorganization more than four and a half years after the first petition. Had LoPucki used a four year window instead of five, Delaware would be statistically indistinguishable from the other courts, even if one did not include the Southern District of New York and the Central District of California. The final traditional Delaware case labeled a failure by LoPucki's definition is TWA. TWA had been taken over by Carl Ichan in 1988. In its first bankruptcy case, it shed roughly $4 billion in debt, and Ichan left as CEO."' Nevertheless, these steps were not enough to return TWA to profitability. Two years later, it filed a prepackaged case in St. Louis. The case, which lasted 30 days, pared another half-billion off of TWA's debt load.' Following TWA's second bankruptcy, it was able to remain outside of Chapter 11 until 2001, when it filed for a third time in order to consummate its sale to American Airlines. 5 6 This longer period between Chapter 11 cases, however, may well be 153 See Brian C. Jones, Victoria Creations Bid OKd, PROVIDENCE J.- BULL., June 26, 1996, at 1F. 154 See Agis Salpukas, T.W.A. Files its Plan to Leave Bankruptcy, N.Y. TIMES, Feb. 18, 1993, at D4. ... See Bankruptcy Court Clears TWA Reorganization Plan, WALL ST. J., Aug. 7, 1995, at B8. 156 See Baird & Rasmussen, supra note 50. E MPIRI CALL Y BA NKR UPTNo. 1: 1791 attributable to the economic climate of the last part of the 1990s, a time of unparalleled profitability for all airlines. In the end, though, TWA never turned a profit after it was taken over by Ichan until its demise. 57 Given the relatively short period between the first case and the second case, one can speculate that the Delaware Court should be faulted for not reducing the debt load more in the first case. Perhaps another court would have forced a more drastic reduction. Of course, such a conjecture, to be credible, would need data suggesting that other courts during this period cut debt more than did the Delaware court. In any event, it is difficult to judge any court by the way in which airlines have performed after bankruptcy. The legacy airlines have had a difficult time since the end of airline regulation in 1978. Most have filed for bankruptcy (Northwest, Delta, United, USAir, Continental) and many have gone out of business (Pan Am, Eastern, TWA, Braniff). There is not a single legacy airline of TWA's size or smaller that is still operating. Moreover, no legacy airline has yet to solve its financial problems in a single trip through the bankruptcy courts. The only legacy airline currently flying but not in financial distress is Continental. Continental filed its first bankruptcy in Houston in 1983 and emerged in 1986. It filed for bankruptcy a second time in 1990 and has not filed for bankruptcy since.15 This by all accounts successful reorganization occurred in Delaware.'59 Thus, the history of airline reorganizations questions LoPucki's assumption that, had TWA filed its first case elsewhere, it would still be in business today. Looking at the traditional bankruptcy cases both quantitatively and qualitatively, one cannot conclude that 157 See Nikhil Deogun et al., TWA Approves Chapter 11 Filing, Buyout, WALL ST. J., Jan. 10, 2001, at A3. 158 See Harvey R. Miller, Chapter 22 Reorganizations and the Delaware Myth, 55 VAND. L. REV. 1987, 2005 (2002). ' Indeed, it was this case that established Delaware as an attractive venue for Chapter 11 proceedings. See Rasmussen & Thomas, supra note 111, at 1372-73. [Vol. 2007COL UMIA B USINESS LA W RE VIE W Delaware's courts performed worse than others. That LoPucki has not produced evidence to support his assertion that Delaware reorganization practice has destroyed savable businesses does not exonerate the Delaware court. Things may have indeed been terrible there, even corrupt. But LoPucki's data provide no evidence that his theory is correct. To the extent that LoPucki wants to make an empirical claim about traditional cases, he simply has fallen short. 2. Prenegotiated and Prepackaged Cases We saw above that LoPucki's data show that prepackaged bankruptcies are more likely to require a second petition. Given Delaware's domination in the area, however, one cannot ascertain whether this is a reflection of prepackaged bankruptcies or an effect of Delaware's bankruptcy practices. Moreover, Delaware was not approving prearranged cases that others were rejecting. From the first prepackaged case until the case of Glenoit in 2000, all prepackaged plans were approved regardless of the court in which they were filed. No court rejected a prepackaged case, and not a single prepackaged proceeding involved any challenge to the way the business was to be operated. In such an environment, one would be surprised if the subsequent need for a second petition stemmed from actions taken by the Delaware bankruptcy court. Still, it is noteworthy that, of those prepackaged cases that did file a second time, all of them were in Delaware. Yet again, however, the numbers are small. Six companies comprise the entire set of Delaware failures. As we saw in the case of traditional reorganizations, LoPucki's numbers do not speak for themselves. They do, however, invite an inquiry into the cases themselves to see whether any of them support his conjecture that Delaware was destroying businesses that another jurisdiction could have rescued. Seven prepackaged cases that were filed in Delaware underwent a second Chapter 11 case within the next five years. In two of these, Westmoreland and Morrison Knudsen, the second case is traceable to factors arising after the first case was completed. Westmoreland was a coal No. 1: 179] EMPIRICALL Y BANKRUPT COLUMBIA BUSINESS LA W REVIEW company that, prior to its first bankruptcy petition, decided to dispose of its properties in the eastern United States. It filed its first case solely so it could assign a contract over a partner's objection.16 The second case was filed two years later in order to resolve a labor dispute. This case was dismissed after the parties agreed to binding arbitration. Neither the first case nor the second involved any alteration of the company's capital structure. The story of Morrison Knudsen is similar to that of Harvard Industries; both are cases where the second bankruptcy was caused by a post-bankruptcy acquisition engineered by a new CEO. Morrison Knudsen filed its first Chapter 11 petition due to decisions of its then-CEO, William Agee to move the venerable construction corporation into the mass transit business. 6' After its prepackaged bankruptcy, the business had no debt.'62 To the extent that one looks to bankruptcy law to trim debt, it is impossible to imagine a more drastic reduction than took place here. The tonic worked; Morrison Knudsen reported fourteen consecutive profitable quarters. The company then bought the construction unit of Raytheon, which turned out to be more troubled than was thought. Morrison Knudsen claimed that it had been misled by Raytheon; Raytheon disputed these allegations. It was this dispute with Raytheon that Morrison Knudsen claimed caused the second case.163 After the second case, Morrison Knudsen again had no long-term debt. As with Westmoreland, then, it strains credulity to attribute the second bankruptcy filing to any defect in the first proceeding. 160 The Bankruptcy Code grants this power to the debtor, even though the contract could not be assigned outside of bankruptcy. See 11 USC § 365(c)(1)(B) (2000). 161 Its many projects include the Hoover Dam and the Alaska Pipeline. 12 The operating company itself never filed for bankruptcy, only the parent company. The prepackaged case was designed to both merge Morrison Knudsen with the Washington Group and to eliminate all of the corporation's long-term debt. See Morrison Knudsen Files for Protection From Its Creditors, WALL ST. J., June 26, 1996, at B2. " See Court-Appointed Examiner Concludes RE&C Transaction was Direct Cause of Washington Group's Filing, BUSINESS WIRE, Aug. 28, 2001. [Vol. 2007 EMPIRICALLY BANKRUPT Putting aside the cases of Westmoreland and Morrison Knudsen, from 1991 to 1996 Delaware had five prepackaged cases where, after the completion of the first bankruptcy, a second bankruptcy petition was filed."' These five cases involved four separate companies, as Memorex Telex filed a prepackaged case in 1992, a second prepackaged case in 1994, and filed for a third and final time in 1996. Here, one can perhaps fault the Delaware court, but not for the reasons offered by LoPucki. To the extent that there was a shortcoming in the first case, it was only that the business was allowed to continue at all. The four companies share a common feature. All four of the companies had taken on substantial public debt within the prior ten years, and three of them had done so as part of a leveraged buyout.1 65 The reorganization plans in all of the cases were substantially similar. The only party affected was the public debt holders. The bank debt, which was senior to the public debt, was left intact, as was the trade debt. Typically, the public debt was reduced by half, with the bonds receiving substantially all of the equity in exchange. 1 66 '" As to the one pre-negotiated case that refiled, it was the Grand Union case, see supra note 133. It filed a second prepackaged case after it could not find its footing in the market place. Investors who had financed Grand Union initially but lost most of their investment include GE Capital, the Disney Corporation and George Soros. Eventually, Grand Union filed Chapter 11 a third time and was liquidated. 165 Indeed, when one looks at all of the prepackaged cases that ended between 1991 and 1996, the vast majority of them were situations where the corporation had taken on substantial public debt either as part of a leveraged buy-out or in connection with an acquisition. We see this pattern in Charter Medical (LBO), JPS Textile (LBO), Edgell (LBO), Gaylord Container (Acquisition Debt), USG Corp. (debt incurred in response to hostile takeover attempt); Restaurant Enterprises (LBO), Petrolane (acquisition debt), Thermadyne (LBO), Mayflower Group (LBO), Great American Communications Company (LBO). " In the second Memorex case, all of the remaining public debt was eliminated, and these bondholders received the equity in the reorganized business. In SPI Holding, the debt holders received only sixty-four percent of the equity. The remaining equity went to Marvin Davis, who had originated the LBO, in exchange for Davis's fresh contribution of $25 No. 1: 179) COLUMBIA BUSINESS LAW REVIEW There is no question that these four companies not only refiled, but eventually failed.167 Yet, LoPucki has not identified a single action that the bankruptcy court could have taken to prevent these ultimate failures. An examination of the cases themselves does not reveal any quick fixes that, even in retrospect, could have been used to preserve any of the corporations. SPI Holding's business was Spectravision, a service that provided movies to hotel guests. After its first case reduced its debt burden, it invested heavily in delivering films via satellite.1 68 This technology did not pan out, and the company filed for bankruptcy a second time. During this second proceeding, Spectravision was sold to OnCommand, which uses cable to deliver its programming. OnCommand bought Spectravision solely to acquire its access to hotels. After the acquisition, OnDemand replaced all of Spectravision's technology with its own. 69 SPI Holding, after its first case, made a sensible business decision that turned out poorly. This decision was not influenced by the first case, and there is no basis on which to conclude that the company would have adopted different technology had a court other than Delaware approved the prepackaged plan. Cherokee and Ithaca Industries were both domestic textile manufacturers that, in the end, could not compete in an outsourcing world.10 They are by no means the only causalities in the shift of American textile production to million. See SPI Holdings Inc.: Bankruptcy Judge Confirms Prepackaged Reorganization, WALL ST. J., Oct. 30, 1992. 167 Cherokee still exists, but all of its operations closed. Its only asset is its trademark, and all of its revenues come from licensing it to manufacturers overseas. '" See Jim Mitchell, Spectravision Hires Cable Exec as CEO, DALLAS MORNING NEWS, Sept. 10, 1994, at 2F. 169 See On Command Corporation Announces 1997 Fourth Quarter and Year-End Financial Results, PR NEWSWIRE, Feb. 18, 1998. 170 Ithaca closed its doors. Cherokee, in contrast, closed all of its operations, but retained the Cherokee brand. Today, the company is profitable, but its only asset is the Cherokee license. See Kathleen O'Steen, Cherokee Makes the Most of Its Trademark, L.A. TIMEs, Sept. 21, 1999, at B8. [Vol. 2007 other countries. Given the decimation of the domestic textile industry, the only plausible flaw is not that the companies filed for reorganization a second time, but rather that the operations were not closed more quickly. That leaves Memorex Telex. The company was formed by the LBO of Memorex and Telex. In the early 1990s, Memorex Telex operated at a profit. The problem was that the profits were not sufficient to service its debt obligations. Memorex Telex had hitched its star to IBM mainframes. Telex made nodes for individual users. Memorex made storage devices for IBM mainframes. In the recessionary times, computer purchasing was down, as were Memorex Telex revenues. The management team of Memorex Telex began restructuring discussions with the banks and bondholders. The two largest single bond holders were legendary vulture investors Carl Ichan and Leon Black. Eventually, the parties proposed a plan where roughly half of the bond debt would be eliminated. In exchange for reducing their debt, the bondholders would receive ninety- five percent of the equity in the reorganized company, and that stock would be publicly traded. The remaining five percent would go to the managers, the preferred share- holders, and the common shareholders. Trade creditors would be unimpaired under the plan. In other words, their bills would continue to be paid as they became due. Shortly after the parties agreed to the plan, Memorex Telex reported disappointing revenues. Still, the plan was sent out to the effected creditors and shareholders. Over eighty-five percent of each class voted for the plan. The plan was approved in then-record time."' The fortunes of Memorex Telex, however, continued to decline. By now, it was becoming apparent that having a business model tied to the market for IBM mainframes was no longer viable. The CEO who had steered Memorex Telex through its first bankruptcy retired, and he was replaced by a veteran of the computer industry. The new CEO 171 See Memorex Telex N.V., WALL ST. J., Feb. 19, 1992, at A2; LoPucKI, supra note 3, at 161. No. 1: 179] EMPIRICALL Y BANKRUPT COLUMBIA BUSINESS LA W RE VIEW undertook a review of the company's operations and began to steer away from reliance on the mainframe market. Still, the inevitable shrinking of that market continued, and the new CEO opened discussion with the debt and equity holders. The parties reached agreement on the following terms. All existing stock would be cancelled, as would all debt other than that arising from operations. In other words, all of the leverage acquired in the combination of Memorex and Telex would be eliminated. The architects of those transactions were left with no interest in the reorganized business and the equity was given to the old bondholders.'72 As with the first plan, the trade creditors would be unaffected. The new plan was approved by over ninety-five percent of the claimants. After restructuring, Leon Black and his investment company controlled roughly fifteen percent of the equity while Carl Ichan controlled another fourteen percent. 173 The CEO then launched a third business model, positioning Memorex Telex as a service provider. Rather than selling parts for a client's network, they sought to manage it.' 74 Of course, to the extent that Memorex Telex had expertise in this area, it was almost exclusively in IBM products. In the end, this strategy was unable to turn around Memorex Telex's fortunes. The CEO brought in after the first bankruptcy left in March of 1996, and Memorex Telex moved into liquidation mode. It sold its Pacific Asia operations in July of that year, and in October, it filed for bankruptcy a third time, announcing that it was searching for a buyer."' 172 The old stock holders did receive warrants that would be in the money if the stock of the reorganized corporation reached $14 a share. 173 See Cecile Gutscher, Future Uncertain for Prepackaged Memorex Telex Plan, 4 HIGH YIELD REP. No. 7, Feb. 21, 1994, at 6. 174 See Memorex Telex Outlines New Business Strategy to Target Enterprise Networking, BUSINESS WIRE, July 18, 1994. 175 See Memorex Telex N.V. Announces Intent to Sell U.S. Operation, BUSINESS WIRE, Oct. 15, 1996. [Vol. 2007 The story of Memorex Telex is not a happy one. Its employees lost their jobs; its investors lost their shirts. Yet, it is far from clear how the Delaware court's handling of the two prepackaged bankruptcies contributed to this failure. The restructuring in the first bankruptcy was based on Memorex Telex's historical earnings. Few at the time appreciated the fundamental changes that were sweeping the computing world. In the second case, all of the debt was wiped out, and the old debt holders emerged with complete control of the company. To be sure, they did not find a way to save the business, but it is doubtful that any court could have engineered a different result. It is unclear what actions would have saved any of these companies. To be sure, it may have been the case that the companies could have pared more debt. Yet, even had the debt been wiped out in its entirety in the first case, failure now seems inevitable. 7 6 SPI Holding and Memorex had technology that lost in the marketplace. Cherokee and Ithaca Industries were crushed in the outsourcing wave that decimated America's textile industry. In the end, these companies failed because they could not find their way in a competitive marketplace. It is simply a flight of fancy to suggest that had they filed their prepackaged bankruptcies in other jurisdictions, their fates would have been any different. One puzzle remains: no other court had a refiling during this period. Part of the answer to this puzzle is that, just as a second filing does not necessarily imply a failure of the first bankruptcy proceeding, the lack of a filing cannot be treated as a success. Of the twelve prepackaged cases with emerging public companies that finished between 1991 and 1996 and did not take place in Delaware, only two of those companies continue to do business as public companies today.'77 The others have been merged or acquired by other companies. Out of the Delaware cases, five companies are still in 76 No one suggests that eliminating all public debt in a prepackaged bankruptcy is the optimal strategy for every company. Debt has a disciplining effect which the investors want to keep in place. 177 They are Southland and Gaylord Container. No. 1: 179] EMPIRICALLY BANKRUPT business.17 It may be that, for prepackaged bankruptcies, simply looking at whether a second petition was filed fails to provide adequate information about the post-bankruptcy fate of the corporation. But regardless of the reason for the Delaware difference in prepackaged cases, we can clearly reject the notion that the failure of Delaware prepackaged cases arose directly from actions taken by the Delaware bankruptcy court. In the end, LoPucki has presented a theory that may well be correct. It may well be that the choice of bankruptcy venue has had a deleterious effect on corporate reorganization practice, but this theory is contestable.'79 Today, we often see creditors gaining control of the enterprise before filing. These creditors may, as LoPucki suggests, steer the companies to a venue that will serve the interests of the controlling creditors at the expense of the other investors in the business. On the other hand, these creditors may find their stakes maximized by seeking out a venue that will maximize the value of the business as a whole.' The empirical evidence that LoPucki presents does not allow for a choice between these two theories. V. CONCLUSION The strength of legal scholarship is that it borrows from various sources in attempts to answer real world problems. This strength, however, is also a weakness. Legal scholars tend to approach debates with strong views. Empirical data is often ambiguous, and the risk is high that one can invariably find what she seeks. Failing to look in the right place, to ask the right question, or to draw the right inference can confirm a bias that was brought to the project. 178 They are Charter Medical, Cherokee, Westmoreland, USG Corp. and Morrison Knudsen. " For arguments that competition may be beneficial, see David A. Skeel, Jr., What's So Bad About Delaware?, 54 VAND. L. REV. 309 (2001). 180 For an argument that controlling creditors have an incentive to maximize enterprise value in a reorganization, see Robert K. Rasmussen, The Search for Hercules: Residual Owners, Directors, and Corporate Governance in Chapter 11, 82 WASH. U. L.Q. 1445, 1460-65 (2004). COL UMBfIA B USINESS LA W RE VIE W [Vol. 2007 No. 1:179] EMPIRICALLY BANKRUPT 245 A healthy skepticism must be applied to empirical work as much as to theoretical work. In both, it is necessary to articulate the crucial assumptions and inferences on which conclusions rest.