How do you measure the value of a relationship? One way is forbearance—what the involved parties would give up in order to stay together. That’s the principle used by Stephen Karolyi, associate professor of finance at the Costello College of Business at George Mason University, in his co-authored research on relationship lending in the banking industry.
“There are many reasons why banks like to use relationship lending, which in practice involves loan officers developing relationships with executives, entrepreneurs and other borrowers,” Karolyi says. “An alternative would be arm’s-length lending where each transaction carries its own narrow cost-benefit analysis, without consideration for other services that are being provided or could be provided in the future.”
Prior research has found that investing in long-term relationships helps lenders gather information on borrowers, which they can use to cross-sell and reduce uncertainty. But just how much that information might be worth to lenders has been a scholarly blind spot.
To resolve this gap, Karolyi identified a moment where lender forbearance becomes a primary issue. When borrowers end up breaching contractual terms—e.g., by exceeding mandated debt-to-earnings ratios or falling below profitability thresholds—lenders can choose to either extend leniency or crack down by imposing fees or renegotiating the loan. When lenders choose the latter, breaching borrowers often take steps to reduce default risk—an additional benefit lenders receive for non-forbearance.
“The lender has the upper hand in this negotiation because the alternative outcome is that the loan is recalled,” Karolyi says.
His paper in Journal of Financial Intermediation closely examines how lenders weigh the benefits of covenant breach enforcement against its chief drawback—which is that it might alienate the borrower, thus severing the relationship. The paper was co-authored by Andrew Bird of Chapman University, Michael Hertzel of Arizona State University, and Thomas G. Ruchti of Virginia Tech.
The researchers assembled a data-set comprising loan packages initiated between 1990 and 2016. They cross-referenced contractual terms with financial data for the borrowers to calculate “covenant slack”—a quarter-by-quarter measurement of the likelihood that a given borrower was in breach. They also quantified changes in the cost of default for borrowers found to be in breach, as well as the probability that those borrowers would switch to a different lender for their next loan. Additionally, the research team mined borrower 8-K filings to collect data on enforcement actions, i.e. actual fees imposed by lenders.
A key aspect of the methodology was comparing outcomes for borrowers that were just barely in breach to those that were very close, but not quite over the line. “When we model two borrowers at similar distance to the threshold, we can be safer in assuming that changes in their outcomes have less to do with differences in borrower quality, and more to do with the covenant violation itself,” says Karolyi. “On one side, the lender has to make a choice whether or not to enforce. On the other side, they don’t have that choice.”
As a result, the researchers could estimate how big the rewards had to be, as compared to the risk of losing the relationship, for banks to discipline borrowers in breach. The sample-wide estimate was 11.6 percent of the loan amount. This figure represented the limit of the lender’s forbearance—how much they were willing to give up for the sake of preserving the future relationship with the borrower.
As the researchers expected, the relationship premium was higher for borrowers that were more mysterious to the wider marketplace (thus placing a higher value on the info-gathering dimension of the relationship), as well as those with fewer financing options.
When it comes to drawing conclusions based on his estimates, Karolyi urges caution. “Whether you’d consider the premium large or small depends upon the size of the loan compared to the bank’s overall portfolio.”
Further, Karolyi warns that the paper’s findings might not be applicable to all loans. “These are all loans that end up in a situation where a borrower is at least close to breaching a covenant threshold, which is not uncommon but certainly not always the case.”
Nevertheless, these results suggest that despite the ascendancy of financial algorithms, non-transactional approaches still add value—and in many cases that value can be quantified. “We were trying to measure the intangible value that is generated for banks through these activities that we know they pursue. What we’re showing is that these lending relationships do behave like information-based intangible assets,” Karolyi says.