Busy bankruptcy courts and the cost of credit

https://doi.org/10.1016/j.jfineco.2021.08.010Get rights and content

Abstract

This paper estimates the effect of bankruptcy court caseload on access to credit by exploiting firms’ plausibly exogenous exposure to the largest recorded drop in court backlog in the United States following the 2005 consumer bankruptcy reform. I show that a drop in court congestion reduces the time firms spend in bankruptcy and increases recovery values, which is priced into credit spreads and loan maturities. Consistent with a shock to credit supply, less congested courts increase firm leverage but leave default risk unchanged. A back-of-the-envelope calculation suggests that backlog in bankruptcy courts costs corporate borrowers at least $740 million per year in interest payments.

Introduction

Our judicial resources are strained. And the cost to society of an overburdened bankruptcy system ... is enormous.

–US District Judge Barbara Lynn1

Legal institutions1 play an important part in shaping the cost of financing for firms and households.2 Higher creditor and property rights, for example, are associated with the development of larger and more sophisticated financial sectors.3 Functioning bankruptcy courts in particular are often seen as playing a key role in the economy. However, there is little credible microeconomic evidence on how courts shape ex-ante outcomes outside of bankruptcy because the functioning of courts is endogenous to the state of the economy.

In this paper, I study how bankruptcy court backlog affects ex-ante contracting by exploiting the largest recorded drop in the caseload of US bankruptcy judges in the wake of the Bankruptcy Abuse Prevention and Consumer Protection Act of 2005 (BAPCPA). The BAPCPA reform fundamentally changed the bankruptcy code for individual debtors, making it considerably harder and more expensive for them to default. In the two years following its implementation, bankruptcy filings essentially halved—the largest decrease outside of war times since the beginning of records in 1899 (Federal Judicial Center, 2019). This drop in filings permanently reduced the number of cases per bankruptcy court and thus the workload of judges.

Despite its effect on individual debtors, BAPCPA left corporate bankruptcies essentially unaffected. As a result, the drop in judicial workloads after the reform was larger in bankruptcy districts with a higher share of nonbusiness filings before. This generates plausibly exogenous variation in court backlog for firms located in the bankruptcy districts that were particularly exposed to BAPCPA. Building on Iverson (2018), I exploit this variation in a difference-in-differences framework to study the impact of court backlog on publicly listed firms.

To start, I show that a one standard deviation increase in the nonbusiness share in a court’s caseload is associated with a 64-hour drop in the annual workload of judges (around two work weeks). This drop in court backlog causes a 5.5% reduction in the length of bankruptcy cases and a 6%–12% increase in recovery values for creditors. Using contract-level data on syndicated loans from DealScan, I then estimate that this improvement in the functioning of bankruptcy courts leads to a 20 basis point drop in interest rate spreads (a 10% drop) and 9% increase in loan maturities. These results suggest that lower court congestion is priced into firms’ ex-ante financing terms by increasing expected recovery values.

Consistent with the predictions of a simple theoretical framework in the spirit of Hart and Moore (1998), the drop in caseload also reduces the gap in loan terms between risky and safe borrowers. Exploiting heterogeneity in pre-reform borrower characteristics in a triple-difference approach, I find that judicial backlog matters almost exclusively for firms with a higher probability of default and a higher expected loss given default. This also allows me to absorb bankruptcy district × time fixed effects, which rules out many potentially confounding factors.

A few additional tests support a causal interpretation of these results. First, loan terms trended similarly for more and less exposed bankruptcy districts before the drop in court caseload, when uncertainty about the impact of BAPCPA lifted. This suggests that creditors interpret observable court backlog as a noisy signal about future recovery values. Initially, interest rate spreads only drop for short-term loans, for which the reduction in backlog in early 2006 is the most informative signal about expected recovery values. By late 2007, however, the effect on shorter and longer loans converges, suggesting that creditors learn over time whether changes in judicial caseload are permanent or transitory.

Second, the share of nonbusiness filings—which I use to measure exposure to BAPCPA—is uncorrelated with firm and loan characteristics before the reform. This also makes it less likely that I am capturing unobserved differences between districts. Third, some borrowers were additionally exposed to changes in credit supply because they had to roll over loans that were issued in the years before BAPCPA passed (also see Almeida et al., 2012). The estimates from using this pre-determined variation suggest that my findings are entirely driven by credit supply, not unobserved shocks to firms.

Next, I estimate the effect of court backlog on firm leverage and default risk, measured by credit ratings and ex-post bankruptcies. Consistent with an expansion of credit supply, book leverage increased by around 2% relative to the mean for a one standard deviation larger drop in court caseload. However, I find a precisely estimated zero effect on credit ratings, the probability of borrowers being rated, and the number of firm bankruptcies. This is likely because—although leverage increased—lower debt service payments decrease default risk. I also find a zero effect on the frequency of bankruptcy filings during the financial crisis. Taken together, this suggests that the drop in court backlog did not change firms’ ex-ante or realized probability of default. Rather, fewer court cases decrease the expected loss given default, which is immediately priced in by creditors.

I conduct a simple back-of-the-envelope calculation to get a sense of the costs judicial backlog imposes on borrowing firms. While it requires a set of strong assumptions, only some of which I can test empirically, changes to interest rate spreads translate naturally into a macroeconomic quantity: the approximately $400 billion per year US non-financial corporations pay to service their debt. My estimates suggest that the costs to firms arising from overburdened bankruptcy judges are around $740 million per year. While this estimate does not necessarily tell us about the welfare effects of busy bankruptcy courts, these magnitudes are large compared to the costs of creating additional judgeships that would lower court congestion. This extrapolation exercise suggests that debt enforcement plays a key role in the ex-ante design of financial contracts. From a policy perspective, it provides some evidence that bankruptcy judges may be relatively “cheap” compared to their value added, at least in terms of the savings to borrowers’ debt service payments.

I also test and reject potential alternative explanations of my findings. A host of robustness checks suggests that the housing boom in the run-up to the 2007-08 financial crisis is unlikely to play a role. Exposure to BAPCPA is not correlated with key housing variables, and excluding securitized loans or the construction and nontradable industries—or adding lender × year fixed effects—makes little difference to my point estimates. Event study plots further suggest a discontinuous and permanent effect of court caseload rather than a boom-bust pattern.

My results are also not driven by minor changes the reform made to corporate bankruptcies. I find almost equivalent estimates for firms who were clearly not affected by changes in the applicable bankruptcy framework. “Forum shopping”, that is the leeway the largest US borrowers have over where to file for bankruptcy, is also unlikely to play a role. Theory suggests that this would lead me to understate the effect of court congestion: if anything, less busy courts should increase bankruptcy filings and make judges more debtor friendly (Gennaioli and Rossi, 2010) and thus—all else equal—lead to worse contract terms for borrowers. In contrast, I find that less congested courts improve contract terms and do not change the number of firm bankruptcies. My baseline results also remain unchanged when I exclude the firms most likely to engage in forum shopping.

My paper builds on a large literature on legal frameworks and economic outcomes. First, my work builds on research that focuses on the enforcement of existing law. Djankov et al. (2008) use a representative bankruptcy case for 88 countries and show that better debt enforcement is associated with higher credit market development. Jappelli et al. (2005) show a negative correlation of judicial backlog and interest rate spreads for Italian provinces.4 However, these correlations do not constitute causal effects because court backlog is likely correlated with other factors that also matter for credit provision. For example, areas that lack public funding to hire judges likely have both more congested courts, firms at a higher risk of defaulting, and a higher cost of credit even in the absence of any causal link. My contribution is to provide, to my knowledge, a first causal estimate of how court backlog affects ex-ante contract outcomes.

I build on the insight of Iverson (2018) that BAPCPA constituted a shock to the US bankruptcy courts with the highest pre-reform share of nonbusiness bankruptcies. Iverson shows that the drop in court congestion around the reform reduced repeated bankruptcy filings, decreased the time larger firms spent in court, and lowered banks’ charge-offs for business lending. My contribution is to study whether and how such expected ex-post effects alter the ex-ante contracting environment. I show how banks adjust contracts in a forward-looking manner—usually more associated with equity markets—when news about recovery values arrive. In addition, I provide a back-of-the-envelope estimate of the aggregate costs court backlog imposes on firm borrowers. In other related work, Boehm and Oberfield (2020) study the effect of court congestion on the input choices of Indian manufacturing firms. Brown et al. (2016) study the effect of a 1953 law that imposed external state courts on Native American reservations.5

Second, I add to the literature studying how the effect of major legal reforms depends on pre-existing levels of judicial backlog. Ponticelli and Alencar (2016) and Rodano et al. (2016) show for Brazil and Italy, respectively, that financial reforms interact with legal institutions governing courts. Their work, however, estimates the value of major changes in bankruptcy regimes, not whether judicial backlog matters per se. In other words, these studies do not tell us whether courts have an effect without a contemporaneous legal reform that fundamentally alters bankruptcy law. My contribution is to estimate the effect of the caseload burden of judges on firms’ financing terms while holding the applicable bankruptcy law constant. My results suggest that court congestion also matters for the United States, which has one of the most sophisticated bankruptcy law and court systems in the world.

Third, this paper adds to the literature on legal determinants of financial contracts more broadly. Qian and Strahan (2007); Bae and Goyal (2009), and Laeven and Majnoni (2005) show that better property rights, creditor rights, and legal institutions are correlated with lower loan spreads across countries. Vig (2013) studies a legal reform in India that increased creditor rights and decreased the delay between default and liquidation. I also add to a large literature on the heterogeneous effects of bankruptcy frameworks.6

The remainder of the paper is structured as follows. In Section 2, I discuss the institutional background governing bankruptcy in the United States generally and BAPCPA in particular. Section 3 introduces the data and variable construction. Section 4 introduces the empirical strategy based on a simple conceptual framework. Section 5 discusses the main results. Section 6 concludes.

Section snippets

Courts and judges

Bankruptcy courts are units of the district courts that operate across 90 judicial districts in the mainland United States.7 Twenty-seven states only have a single bankruptcy district. The decision-making power over bankruptcy cases lies with the bankruptcy judge. As of September 2012, there were 350 bankruptcy judgeships in the United States, out of which 34

Loan contract and balance sheet information

I use loan-level data from Thomson Reuters’ DealScan. The main variables I construct from DealScan are the interest rate spread (usually over LIBOR), the natural logarithm of the loan maturity (in months), and size of the loan (in logs). I further include a dummy for whether a loan is backed by collateral.

Balance sheet data come from the annual Compustat Fundamentals, which I extend with rating data from Mergent FISD. Following standard procedure in the literature, I exclude firms in the

First stage: court backlog, case length, and recovery values

BAPCPA approximately halved the congestion of US bankruptcy courts before the onset of the Great Recession. Because BAPCPA targeted consumer bankruptcies, there is an almost linear relation between the share of nonbusiness cases in total caseload prior to the reform (Exposure) and the subsequent drop in caseload per judge. Fig. 3 visualizes this point, where the caseload drop is calculated as the difference between the averages for 2006–2007 (Post-BAPCPA) and 2004–2005 (Pre-BAPCPA). I exploit

Baseline estimates

How do forward-looking creditors price in changes to the congestion in bankruptcy courts? Column 1 of Table 4 starts to investigate this question by running Eq. (1) with interest rate spreads as the dependent variable. The estimated coefficient is around 107 and highly statistically significant. It implies that a one standard deviation increase in the exposure to BAPCPA (around 0.12) is associated with 13 basis points (bps) lower spreads (106.53×0.12113). In column 2 I allow for the

Conclusion

Exploiting exposure to BAPCPA as an exogenous shock to court caseload, I show that judicial backlog has first-order effects on the ex-ante contract environment in the corporate loan market. To the best of my knowledge, my paper provides among the first estimates of the causal effects of debt enforcement on ex-ante outcomes. I estimate that a 64-hour drop in annual caseload per judge reduces the time bankruptcy cases spend in court by 5.5% and increases recovery values for creditors by 6%-12%.

References (75)

  • V.V. Acharya et al.

    Bankruptcy codes and innovation

    Rev. Financ. Stud.

    (2009)
  • S. Albanesi et al.

    Insolvency After the 2005 Bankruptcy Reform

    NBER Working Papers

    (2018)
  • H. Almeida et al.

    Corporate debt maturity and the real effects of the 2007 credit crisis

    Crit. Finance Rev.

    (2012)
  • E.I. Altman

    Revisiting the recidivism-chapter 22 phenomenon in the US bankruptcy system

    Brook. J. Corp. Financ. Commer. Law

    (2013)
  • American Banker, 2005a. In Focus: As New Bankruptcy Era Opens, What’s Now and What’s Next. October 17, 2005. Available...
  • American Banker, 2005b. Why Some Have Yet to Detail Bankruptcy Impact. November 18, 2005. Available online at:...
  • American Banker, 2006. Bankruptcy Reform: What Has It Meant?October 17, 2006. Available online at:...
  • K.-H. Bae et al.

    Creditor rights, enforcement, and bank loans

    J. Finance

    (2009)
  • T. Berg et al.

    The total cost of corporate borrowing in the loan market: don’t ignore the fees

    J. Finance

    (2016)
  • G. Bermant et al.

    A day in the life: the federal judicial center’s 1988–1989 bankruptcy court time study

    Am. Bankruptcy Law J.

    (1991)
  • J. Boehm et al.

    Misallocation in the market for inputs: enforcement and the organization of production

    Q. J. Econ.

    (2020)
  • A. Bris et al.

    The costs of bankruptcy: chapter 7 liquidation versus chapter 11 reorganization

    J. Finance

    (2006)
  • J.R. Brown et al.

    Law and finance matter: lessons from externally imposed courts

    Rev. Financ. Stud.

    (2016)
  • G. Cerqueiro et al.

    Debtor rights, credit supply, and innovation

    Manag. Sci.

    (2017)
  • G. Cerqueiro et al.

    How does personal bankruptcy law affect startups?

    Rev. Financ. Stud.

    (2016)
  • R. Chakrabarti et al.

    Auto credit and the 2005 bankruptcy reform: the impact of eliminating cramdowns

    Rev. Financ. Stud.

    (2019)
  • CLLA, 2016. Commercial Law League of America: Bankruptcy Venue Reform. Available online at:...
  • J.D. Coval et al.

    The geography of investment: informed trading and asset prices

    J. Polit. Econ.

    (2001)
  • S.A. Davydenko et al.

    Do bankruptcy codes matter? A study of defaults in France, Germany, and the U.K

    J. Finance

    (2008)
  • S. Djankov et al.

    Debt enforcement around the world

    J. Polit. Econ.

    (2008)
  • M. Drehmann et al.

    How much income is used for debt payments? A new database for debt service ratios

    BIS Q. Rev.

    (2015)
  • K.E. Dynan et al.

    Recent changes to a measure of US household debt service

    Federal Reserve Bull.

    (2003)
  • F.R. Edwards et al.

    Derivatives and the bankruptcy code: why the special treatment?

    Yale J. Regul.

    (2005)
  • G. Favara et al.

    Debt enforcement, investment, and risk taking across countries

    J. Financ. Econ.

    (2017)
  • Federal Judicial Center, 2019. Caseloads: Bankruptcy Cases,...
  • Gayda, R., 2015. The Bankruptcy Venue Debate: A Never-Ending Story. Law360. Available online at:...
  • N. Gennaioli et al.

    Judicial discretion in corporate bankruptcy

    Rev. Financ. Stud.

    (2010)
  • Cited by (13)

    • Justice as efficiency: Courts and the allocation of electricity in China

      2023, Journal of International Financial Markets, Institutions and Money
    • Girls Just Wanna Have Funds? The effect of women-friendly legislation on female-led firms’ access to credit

      2022, International Review of Law and Economics
      Citation Excerpt :

      To the best of our knowledge, only Kelley et al. (2022) analyze the effect of gendered law on lenders’ attitudes toward female prospects in the nonmortgage fintech lending setting. In addition to legal rights, Müller (2022) shows that legal efficiency is critical; credit restrictions appear to diminish in more efficient legal systems (i.e., those that score higher on an efficiency index). As Bae and Goyal (2009) explain, strong laws cannot substitute for weak judicial enforcement.

    View all citing articles on Scopus

    I am grateful to Toni Whited (the editor) and an anonymous referee, as well as Titan Alon, Natalie Bachas, Douglas Baird, Thorsten Beck, Tony Cookson, Matteo Crosignani (discussant), Stuart Fraser, Paul Goldsmith-Pinkham (discussant), Mathias Hoffmann, Ben Iverson, Jan Keil, Paymon Khorrami, Yann Koby, Moritz Lenel, Ernest Liu, Adrien Matray, Eduardo Mattos, Sebastian Merkel, Atif Mian, James Mitchell, Ezra Oberfield, Steven Ongena, Jacopo Ponticelli (discussant), Jung Sakong, David Samuel, David Sraer, Emanuele Tarantino (discussant), James Traina, Emil Verner, Motohiro Yogo, and Eric Zwick for helpful discussions and feedback. I would also like to thank seminar participants at the SFS Calvacade 2019, WFA 2019, University of Chicago, Princeton University, EFA 2018, MFA 2019, University of Zürich, 12th Swiss Winter Conference on Financial Intermediation, University of Warwick, Christmas Meeting of German Economists Abroad, and the University of the West Indies for their comments. Most of this research was conducted while I was at Princeton University, whose hospitality is gratefully acknowledged. I was supported by a Doctoral Training Centre scholarship granted by the Economic and Social Research Council [Grant number 1500313].

    View full text