Volcker and Ludwig: Relying on financial models to set loan-loss reserves is flawed

In their Wall Street Journal editorial, Paul Volcker and Eugene Ludwig praise the accounting board for adopting the idea that banks should reserve for losses expected over the life of the loan, but criticize the implementation because of its heavy reliance on financial models.

Actually, the real problem with the implementation is that it is subject to being gamed by bank management.

So long as banks are not required to provide ultra transparency and disclose on an ongoing basis their current global asset, liability and off-balance sheet exposure details, bank management can play games whether the loan loss reserves are set based on financial models or 'experience and judgement'.

The reason that the reserves can be gamed is that in the absence of ultra transparency market participants have no way of independently assessing how closely the reserves for losses expected over the life of the loan actually tracks with how the bank's loans are actually performing.

For example, what are the reserves set aside for the 'zombie' loans that are the product of regulatory forbearance and the banks engaging in 'extend and pretend'?

The good news: The board recognizes that its existing rules on the Allocation for Loan and Lease Losses may have worsened the 2008 financial crisis. These rules limited bank reserves to those that are already "incurred." This all but ensures that banks' rainy day funds will be too skinny, particularly in periods when credit markets are under stress. 

Worse yet, limiting loss estimates to events that have already occurred makes the allowance for loan and lease losses procyclical—reported earnings are too high in good times and losses hit hardest in bad times.

The FASB's draft proposal to reform these rules incorporates what is known as the "Current Expected Credit Loss Model." It is meant to expand reserves to reflect losses that are expected over the life of the loan, and it is a big improvement over the existing regime. 

But as it stands, the proposal could create risks for the financial system.

Risks that can only be removed by requiring the banks to provide ultra transparency.

In an effort to ensure that everything is "auditable," the proposal ties the loan-loss reserve to what the accounting profession will decide is an acceptable "model." While the proposal is well-intentioned and makes clear that various models can be used, this model-driven approach is dangerous. 

Modeling by its very nature is backward looking. It would push bankers to address only risks that are readily and historically quantifiable. It would discourage them from acting on forward-looking but less well-defined risks, like broader economic trends, that can be just as damaging. 

A focus on modeling also unnecessarily favors large institutions. Banks with smaller loan books and more hands-on experience have some advantages when setting their reserves.... 

While we do believe it is critical to allow bankers to use their expertise in estimating losses for reserve purposes, we also believe it is critical that they disclose to regulators and the public both the methodology they employ to set reserves and the quarter-by-quarter decisions on reserves they actually make. That way investors can follow a bank's net revenue picture before and after loan reserves are set aside, and the methods they use to establish these reserves.

Disclosing how the reserves are set without providing ultra transparency is simply an invitation for the banks to game the system.