Earlier this summer, I wrote a column on the vital role loan-loss reserves can play in the safety and soundness of American banking.  I knew that bankers had remarkably little leeway to set their reserves — either to lower or raise them — because of strict accounting rules in desperate need of amendment.  I just didn't expect to hear from so many of those bankers directly.

For instance, last week, I met a banker who was lamenting that he sees credit as just a little bit soft right now. Pricing and terms are too loose, he said, and he'd like to bump up his allowance for loan and lease losses. But because the previously troubled credits on his books are improving, and because the softness he sees isn't "identifiable" in an accounting sense, accounting rules actually require him to reduce the allowance. 

The same banker pointed out he couldn't achieve the safety cushion he needs by raising capital. He had already brought in a capital buffer beyond minimum regulatory requirements — and now it's just sitting there, waiting for a rainy day.  But while capital may be useful in failed bank cases and can serve as a cushion in the 100-year flood, it's much less useful in the year-to-year storms, which can be every bit as deadly.  

In practice, capital is a reserve you can draw down only at your peril, in extreme cases, and even then, there is potential for collateral problems. Regulators and investors view a capital drawdown as a sign of weakness, often with both serious regulatory and market consequences. More perversely, if an institution boosts its capital above what is necessary, its return on capital declines, making the bank look weaker and diminishing its ability to raise capital.

Of course, maintaining appropriate capital levels is important. However, though capital is a powerful safety and soundness tool, it is no substitute for prudent discretionary provisioning.  Supervisors, including Comptroller of the Currency Thomas Curry, who has recently spoken out on this issue, should be commended for recognizing the importance of provisioning to safe and sound banks.

The accounting aversion to discretionary provisioning is driven by a fear that banks will use reserves to hide their true economic earnings (or losses) of from investors.  Accordingly, accounting rules make the Allowances for Loan and Lease Losses essentially formulaic and treat credit losses as either known or unknown. In that simple framework, the loan-loss reserve is there to absorb only "expected" losses, and capital is there to absorb unexpected ones. But in reality, credit risk presents more subtle distinctions.  Instead of known or unknown, credit presents varying degrees of risk over time.  Anticipating growing risk in portfolios before it can be substantiated by data is part of the art of banking. Models have their place, but so does judgment. Failure to take advantage of a banker's judgment is obviously a mistake when the banker wants to take steps that will enhance prudence, even at the expense of increasing short-term profitability.

Without giving bankers this flexibility, we move back to the dangerous scenario where past behavior determines the cushion banks are allowed to hold, and not a good-faith belief in the unrest that might yet come. In a weak economy beset by all manner of exogenous shocks, we simply can't afford accounting rules that effectively chide bankers for having a sense of caution. 

If we don't free bankers to increase their ALLL, when the time comes for a real economic speed bump, those capital reserves will be drawn down in dramatic fashion.  Loan-loss reserves aren't just a cushion for lending gone awry — they're a buffer against a capital hit, and against the market reaction that can follow quickly on its footsteps. This goes beyond fairness into systemic issues, where investor perception can make a bad situation for one bank into a worse situation for the entire sector.

The notion of decreasing the degree to which a banker can use his or her judgment to add to the ALLL not only throws the baby out with the bathwater, but it also fails to protect the investor, the very person that the accounting rules claim to be protecting.  The notion that this accounting orthodoxy keeps bankers from hiding economic reality from investors is a thin reed to lean on, if it is a reed at all.  Not only do more robust reserves at banks help investors by making banks safer and sounder; they are also completely transparent to the investor, so the investor can clearly see what the banker is doing with this account.

The debate around financial institution accounting reform is in full swing, and has only picked up since the start of the summer.  But the core issues remain the same: banks should be free to set whatever ALLL they want, as long as it's well-reasoned, consistent with regulatory requirements, and entirely transparent to the public.  Prudential supervisors, not accountants, should be in charge of this issue.

Eugene A. Ludwig is a founder and the chief executive of Promontory Financial Group LLC. He was the comptroller of the currency in the Clinton administration.