Stigma, Public Disclosure and BankruptcyFootnote 1

by Barry Scholnick

December 2011


Note: While abstracts are available in both English and French, some research papers are only available in their original language. This document is only available in English.

Abstract

The bankruptcy stigma hypothesis states that individuals with lower social costs of default (i.e., less embarrassment) will be more likely to default. We introduce and test the related hypothesis that the level of public disclosure accompanying default impacts stigma. We compare whether individual credit card holders choose to default via bankruptcy, where there is a legal requirement for public disclosure, or via credit card charge-off, where there is no legal requirement for public disclosure. We capture stigma using data on all past bankruptcies in the defaulter’s neighbourhood. We find that lower stigma, as measured by a one standard deviation increase in past bankruptcies in the defaulter’s neighbourhood, will increase the probability that the defaulter will choose (disclosed) bankruptcy rather than (undisclosed) charge-off by approximately 6 percent.

1. Introduction

The bankruptcy stigma hypothesis states that individuals will be more likely to file for bankruptcy if they face lower social costs (i.e., less embarrassment) when they default (Fay, Hurst and White, 2002; Gross and Souleles, 2002; Athreya, 2004; Sullivan, Warren and Westbrook, 2006; Dick, Lehnert and Topa, 2008; Cohen-Cole and Duygan-Bump, 2009; Livshits, MacGee and Tertilt, 2010; White, 2011). The importance of the stigma hypothesis is emphasized by Livshits, MacGee and Tertilt (2010), who argue that declining costs of bankruptcy, e.g., declining stigma, are “likely the most commonly cited explanation” (p. 166) for the recent increase in bankruptcy in the United States. The issue of stigma is also relevant to public policy. Broadly speaking, pro-creditor lobbyists (e.g., the credit card industry) argue that declining bankruptcy stigma makes bankruptcy more prevalent (supported by evidence from Fay, Hurst and White, 2002; Gross and Souleles, 2002; and Livshits, MacGee and Tertilt, 2010), thus they argue for increased legal restrictions on the ability of debtors to file for bankruptcy. On the other hand, pro-debtor lobbyists (e.g., consumer rights groups) argue that declining stigma is not an important reason for increased individual bankruptcy filings, using evidence from authors such as Sullivan, Warren and Westbrook (2006). In spite of the obvious importance of bankruptcy stigma, however, empirical evidence on how bankruptcy stigma actually operates remains limited.

The aim of this paper is to propose and empirically test the new hypothesis that individuals for whom avoiding stigma is important will be more likely to choose a mechanism for default where there will be less public disclosure about their default. Similarly, individuals who are less concerned about stigma will be less constrained about choosing a mechanism for default that may entail public disclosure. Our argument is that the level of stigma felt by an individual will be related to the possibility that others will subsequently become aware of the individual’s default, which in turn is related to the extent to which information on the default is publicly disclosed. While there is a vast literature examining the impact of public disclosure in many areas of finance and economics, our paper is the first to examine how public disclosure impacts bankruptcy stigma and the mechanism of default.

Using data from individual credit card accounts, we test our hypothesis by comparing stigma effects between (1) defaults where there is a legal requirement for information about the default to be publicly disclosed (i.e., bankruptcy), and (2) defaults where there is no legal requirement that information about the default be publicly disclosed (i.e., credit card charge-off). We explore the legal differences between bankruptcy and charge-off to identify the impact of public disclosure on stigma. This specific choice between default via bankruptcy or default without bankruptcy (e.g., via charge-off) is discussed by White (2011) in her survey of the institutional details surrounding default. She writes that “the main punishments for bankruptcy are making filers’ names public...which...stigmatize the bankruptcy filers.” On the other hand, the “punishments for debtors who default but do not file for bankruptcy, include(s) credit collectors calling them, suing them, and garnishing their wages.” (White, 2011, p. 2).

Defaulters thus face a trade-off when choosing between bankruptcy and charge-off. Bankruptcy entails greater stigma because of the increased public disclosure, but under bankruptcy all outstanding unsecured debts (e.g., credit card debt) can be written off and all recovery actions by creditors are stayed (stopped). Charge-off entails lower stigma because of reduced public disclosure, but under charge-off creditors are able to continue attempts to recover debt through wage garnishment and other actions. Bankruptcy and charge-off are the two legal mechanisms whereby the credit card contract can be terminated with unpaid balances remaining, thus both constitute formal default.

Public disclosure of every bankruptcy filing in the United States is, by design, provided through the court system, and public disclosure of every bankruptcy filing in Canada is provided on a single Government of Canada web page (from which the data in this paper are derived). This is not the case for credit card charge-offs, where there is no legal requirement that information on this kind of default be publicly disclosed. While information about both bankruptcy and credit card charge-off appear on the defaulter’s credit rating (e.g. FICO score), the distinction we explore here concerns public disclosure of the default to those without access to credit ratings, i.e., the defaulter’s broader social network, from which stigma effects are assumed to flow.

Our paper builds on, but is different from, various strands of the literature. We follow Fay, Hurst and White (2002), Gross and Souleles (2002), Dick, Lehnert and Topa (2008) and Cohen-Cole and Duygan-Bump (2009) in empirically testing for stigma effects in bankruptcy, but we are the first to specifically hypothesize that the level of stigma is related to the extent of public disclosure. We are also the first to test this hypothesis by empirically comparing the impact of stigma on publicly disclosed bankruptcy and publicly undisclosed charge-off. Our paper is also somewhat related to the literature examining the strategic interactions between defaulters and creditors during the period after default (White, 1998a; White, 1998b; Dawsey and Ausubel, 2004; Dawsey, Hynes and Ausubel, 2009; Chatterjee, 2011; Benjamin and Mateos-Planas, 2011) to the extent that we focus on comparing default with bankruptcy and default without bankruptcy (i.e., credit card charge-off). This literature does not, however, address issues of stigma, which is the central element of our paper. Furthermore, this literature addresses possible strategic interactions between defaulters and creditors during the subsequent periods after the initial default, whereas our empirical tests examine whether stigma impacts the choice of the initial mechanism for default (bankruptcy or charge-off).

To test the stigma–publicity hypothesis, we first have to measure stigma effects (i.e., identify individuals who may be more or less embarrassed about defaulting). We then have to examine the impact of these stigma effects on the choice individual defaulters make between defaulting via bankruptcy or defaulting via credit card charge-off. The methodology we use to capture stigma effects is commonly used in the literature (Fay, Hurst and White, 2002; Gross and Souleles, 2002; Dick, Lehnert and Topa, 2008; Cohen-Cole and Duygan-Bump, 2009). This approach examines the impact on individual defaults of aggregate bankruptcies in the geographic area the individual lives in. This methodology is based on the assumption that stigma effects can flow to an individual defaulter from other individuals who live in the same geographic area. Justification for such a procedure is provided by Fay, Hurst and White (2002) who argue that “if households live in a district with a higher bankruptcy filing rate, then they are more likely to hear first-hand about bankruptcy from friends or relatives because the latter are more likely to have filed.... This information will tend to make households more comfortable with the idea of bankruptcy, so the level of bankruptcy stigma falls.” (p. 710). Similarly, Gross and Souleles (2002) argue that “social stigma and information about bankruptcy might change, with the number of people in one’s community, appropriately defined, that have already filed for bankruptcy.” (p. 339).

Following the discussion of Fay, Hurst and White (2002) and Gross and Souleles (2002), we argue that there are a variety of different “channels” through which stigma effects can operate between past bankruptcies in a geographic area and the choice between defaulting via bankruptcy or charge-off. First, because bankruptcies are publicly disclosed whereas defaults without bankruptcy (e.g., charge-offs) are not, individuals in the area are more likely to hear about past bankruptcies relative to past defaults in the area. Increased knowledge of neighbours’ bankruptcies in the past could lower stigma related to bankruptcies (because “everybody else is doing it”), increasing the probability that the individual files for bankruptcy rather than default without bankruptcy. Second, it is also possible that an individual defaulter could have learned about the specific procedural process involved in bankruptcy from previous bankruptcy filers in the neighbourhood (i.e., information cascades). It can be argued that such information cascades will be more likely to emanate from previous bankruptcy filers than individuals who default via charge-offs if bankrupts are relatively less concerned about maintaining confidentiality about the procedural details of their default given that their default is already public knowledge. Third, individuals concerned about avoiding stigma could be more likely to choose non-publicized charge-off rather than publicized bankruptcy to avoid others in the area learning about their own default at some stage in the future. All of these mechanisms imply that increased bankruptcies in an area will increase the probability of an individual in that area filing for bankruptcy rather than defaulting without bankruptcy (i.e., charge-off).

To test the stigma–publicity hypothesis, we use data derived from matching two unique databases, both provided to us confidentially. First, we use individual level monthly credit card account data provided to us by an individual Canadian bank. The data contain information on a large number of individual credit card account holders, a small fraction of whom have filed for bankruptcy or have had their credit cards charged-off. These individual credit card account level bankruptcy and/or charge-off data are our dependent variable(s). The credit card data are similar in structure to data used in previous bankruptcy stigma research conducted by Gross and Souleles (2002). Our data differ from data used by Gross and Souleles (2002), however, in two important respects. First, our data flag the two separate kinds of individual default — bankruptcy and credit card charge-off. Second, the data include the Canadian six-digit postal code of each individual credit card holder, revealing the fact that these Canadian postal codes cover very small geographic areas, containing only 50 households on average, and typically extend over only a few city blocks (i.e., our definition of a “neighbourhood”).

We use these six-digit postal codes to match the credit card database with our second database, which contains counts of every past insolvency filing in every Canadian postal code (neighbourhood) in every year. This second, neighbourhood count database was provided to us by the OSB, the Canadian bankruptcy regulator, through special runs of the OSB data extraction system conducted specifically for this research project. A key advantage of our data is that the aggregate bankruptcies used to capture stigma effects are measured at a very small neighbourhood level (50 households) compared with the aggregate bankruptcies used by others in the literature that are measured over very large geographic areas (e.g., U.S. states, used by Gross and Souleles (2002), or U.S. bankruptcy court districts, used by Fay, Hurst and White (2002), of which there are 94). We argue that it is more likely that interactions between individuals, through which stigma effects flow, will occur within 50 household neighbourhoods than across U.S. states or U.S. bankruptcy court districts.

Our main test examines whether neighbourhood level stigma (the main independent variable, as measured by all past bankruptcies in the defaulter’s neighbourhood) impacts the choice of an individual defaulter to default via bankruptcy or charge-off (the dependent variable(s)). The main finding of this paper is that lagged neighbourhood level bankruptcy has a significantly positive impact on the individual’s choice to declare bankruptcy, but a significantly negative impact on the individual’s choice to allow credit card charge-off. We find, for example, that a one standard deviation increase in aggregate consumer bankruptcies in a defaulter’s neighbourhood in the five prior years will significantly increase the probability that the defaulter will file for bankruptcy rather than default via credit card charge-off by approximately 6 percent.


Footnotes

  1. 1 Financial support from the Office of the Superintendent of Bankruptcy (OSB) Canada to conduct the research on which this report is based is gratefully acknowledged. The views expressed in this report are not necessarily those of the Office of the Superintendent of Bankruptcy, Industry Canada or the Government of Canada. Funding was also provided by the Social Sciences and Humanities Research Council of Canada (SSHRC). Particular thanks to Janice Jeffs, Stephanie Cavanagh, Gord Kelly and Lynne Santerre for their help with the OSB data and to employees of the anonymous Canadian bank for their help with the bank account data. Vyacheslav Mikhed provided outstanding research assistance. Thanks to Vyacheslav Mikhed, Rasmus Fatum and Jason Allen for comments. (back to footnote reference 1)