MENU
April 29, 2013 | Research Feature

A Better Standard for Credit Risk

A new metric uses publicly disclosed bank information to better predict credit losses from loans.


Financial crises across the world’s banking system are nothing new, having occurred regularly — if not always predictably — throughout history. Banks, investors, and regulators have sought ways to report and analyze credit risk and profitability as a means to head off such crises, but just how to measure these objectively and fairly remains controversial.

For most banks, lending is the primary activity and source of value creation. Loan yield provides a good estimate of expected interest income. Banks can estimate just how much value they are creating by subtracting expected credit losses from lending against expected interest income from lending, which gives one measure of profitability.

But established credit metrics and disclosures, based on the currently required incurred loss model, don’t predict credit losses very well. The prediction challenge is perhaps best reflected in how banks currently report allowance for loan and lease losses (ALLL) and the provision for loan and lease losses (PLLL), two of the main credit-related disclosures.

ALLL represents management’s best estimate of the total loans and leases (held for investment) that the bank believes it will be unable to collect, based on information and events as of the date of the financial statement. That’s an important detail: in estimating for ALLL the regulatory rule is that banks can only use information and events as of the date of the financial statements. Net loans reported on the bank’s balance sheet are estimated by deducting the ALLL from gross loans. PLLL is the amount charged against revenue that reflects credit risk from actual write-off of loans (net charge-offs) and the change in the ALLL during the reporting period. PLLL usually does not incorporate the actual credit risk taken on in the period itself.

In the past, regulators offered more flexibility in calculating the PLLL — managers were allowed to include some expected future losses in the provision. But over time, regulators put increasingly strict guidelines in place — the incurred loss model. As with ALLL, a bank can provide (i.e., charge) for losses only if it can document that the loss is probable and can be reasonably estimated. Regulators believe this is a more objective and reliable approach than allowing managers discretion about the uncertain future. They also argue that in the absence of such rules managers might manage earnings to make performance look more consistent or less volatile, and manage regulatory capital levels as well. “Under the current rules, a certain amount of ALLL can be counted as regulatory capital,” Professor Urooj Khan says. “Managers of banks whose pre-managed regulatory capital is low have incentives to use ALLL to report higher regulatory capital.” Indeed, research shows that earnings and capital get managed despite the rules.

Currently, accounting regulators are rethinking their approach. But are there currently available measures that can help to assess future credit losses? Khan and Professors Trevor Harris and Doron Nissim developed a timely, unbiased measure of expected credit losses that can be used to better assess the risks and profitability of lending. The researchers’ model combines publicly available credit-related measures disclosed by banks between 1996 and 2012, garnered from consolidated financial statements (FY9Cs) that bank holding companies are required to file with the Federal Reserve.

They developed their new metric, Expected-RCL, or expected rate of credit losses, using reported credit losses, nonperforming loans, average loan yield, and duration and composition of the loan portfolio. When they analyzed financial statement data using Expected-RCL, they found its predictive power offered a significant improvement over current metrics.

“We believe our measure is a better — more precise — predictor of expected credit losses than any of the existing credit risk-related metrics publicly disclosed in financial statements,” Khan says. “It can be used to better evaluate value creation and lending as well as bank risk and performance to a large extent.” While there is no easy way to quantify the improvements that Expected-RCL offers over other measures, Khan notes, it performs substantially better than net charge-offs, realized credit losses, or the fair value of loans in predicting credit losses.

The researchers’ work comes at a time when standard setters, including the Financial Accounting Standards Board (FASB) and International Accounting Standards Board (IASB), are reconsidering the incurred loss model in recognition of the challenges it presents.

“The research offers these standard setters and regulators material to contemplate as they consider giving managers more discretion in reporting expected credit losses,” says Khan. “It also offers academics a more informative measure to use in research, and it offers investors and banks a new measure to assess profitability and risk.”

You can listen to Urooj Khan talk about his related research in this Columbia Ideas at Work podcast:

 

Trevor Harris is the Arthur J. Samberg Professor of Professional Practice and director of the Center for Excellence in Accounting and Security Analysis at Columbia Business School.

Urooj Khan is assistant professor of accounting at Columbia Business School.

Doron Nissim is the Ernst & Young Professor of Accounting & Finance in the Accounting Division at Columbia Business School.

Read the Research

Trevor Harris, Urooj Khan, Doron Nissim

"The expected rate of credit losses on banks' loan portfolios"

View abstract/citation  Download PDF   View Research  

Have You Read Columbia Business?

Columbia Business School’s alumni magazine connects alumni with each other and the School; celebrates alumni milestones and accomplishments; and chronicles the impact of Columbia Business School alumni, faculty members, and students on the global business landscape.

Read Columbia Business>