By Samuel Antill (Harvard Business School) and Aymeric Bellon (UNC Kenan-Flagler Business School)
Many firms file for Chapter 11 bankruptcy in a court located outside of the state in which they are headquartered. Roughly half of these “forum-shopped” bankruptcies are filed in the Delaware bankruptcy court. For decades, policy makers and academics have debated whether this Delaware forum shopping is a positive or negative feature of bankruptcy. Critics argue that forum shopping reflects corrupt court incentives to attract cases. Six congressional bill proposals have sought to limit forum shopping. Several of these bills are predicated on the idea that forum shopping is harmful for employees, who cannot represent themselves in a distant court. On the other side of the debate, proponents argue that forum shopping is beneficial because Delaware judges have expertise in handling difficult bankruptcies.
In a separate research article titled “The Real Effects of Bankruptcy Forum Shopping,” which we now summarize, we provide new empirical evidence to inform this forum shopping debate. We combine a novel dataset with a natural experiment to estimate the causal effect of Delaware forum shopping on firm outcomes.
Our dataset covers a nearly comprehensive set of Chapter 11 bankruptcies filed by public and private firms of all sizes. We merge detailed bankruptcy records with establishment-level data from the U.S. Census, allowing us to track establishment-level outcomes for bankrupt firms before and after their bankruptcy filings. In most of our tests, we focus on roughly 16,000 Chapter 11 bankruptcies filed by firms headquartered in three Delaware-neighboring states: Pennsylvania, Maryland, and New Jersey.
Our natural experiment exploits the fact that firms are more likely to file in a court that is physically close to their headquarters; indeed, in a Government Accountability Office survey, 33% of bankruptcy professionals stated that they consider physical proximity as a factor in choosing courts. We confirm this empirically: within a Delaware neighboring state, we show that firms headquartered closer to the Delaware court are more likely to file in Delaware. Conditional on extensive control variables, we show that firms near the Delaware court are comparable to those further from the Delaware court. Most notably, in a placebo test examining firm outcomes before they file for bankruptcy, we show that firms near the Delaware court have similar pre-bankruptcy employment growth trajectories to firms further from the Delaware court. Thus, exploiting within-state variation in the distance to Delaware allows us to “quasi-randomly” assign firms to forum shop in Delaware.
Formalizing this in two-stage least squares regressions to estimate causal treatment effects, we show that forum shopping to Delaware dramatically improves bankruptcy outcomes. Relative to a counterfactual of filing for bankruptcy outside of Delaware, filing for bankruptcy in Delaware substantially reduces the likelihood that a Chapter 11 bankruptcy is converted to Chapter 7. Similarly, filing in Delaware drastically reduces the likelihood of establishment closure, increasing post-bankruptcy employment growth as a result. Suggestive evidence indicates that avoiding liquidation through forum shopping benefits creditors: filing in Delaware increases the likelihood that unsecured creditors receive positive recovery.
Finally, we argue that these positive effects of forum shopping are driven by the pool of judges in Delaware. To show this, we construct a judge-level measure of how “Delaware-like” a judge’s outcomes appear. Constructing this measure for judges in each bankruptcy court, we show that an outside-Delaware bankruptcy with a Delaware-like judge tends to have a Delaware-like outcome. Formally, the treatment effect of forum shopping to Delaware shrinks to zero once we control for the assignment of a Delaware-like judge. This suggests, consistent with practitioner interviews, that the distribution of judge talent in Delaware drives our results.
Click here to read the full article.