The 2009 Card Act required lenders to "consider" customers' "ability to pay" before granting credit on cards.

A senior regulator recently told me that banks reacted strongly to this—meaning they complained a lot. I responded that banks have reacted thoughtlessly and dysfunctionally, rather than strongly.

Consequences of this seemingly common-sense requirement have included banks incurring hundreds of millions of dollars of wasted marketing costs and depriving tens of millions of people of sound credit needed for our economic recovery. It's an object lesson in an obtuse response to misconstrued legislation, like charging a fee for debit cards.

What's the purpose of a mandate to consider ability to pay before extending credit? It'a fundamental safety-and-soundness requirement that every prudential regulator must enforce anyway, without any Congressional nudge.

Evidently Congress had something further in mind — because in the Great Recession credit cards performed well and didn't cause banks to fail.

What apparently provoked this enactment was the accusation that banks made money by lending to high-risk customers who suffered (while the banks somehow profited) from the resulting defaults. There may have been some truth in that for mortgages—where banks could hope to shift the losses to distant investors and the borrowers might eventually suffer foreclosure. None of that applies to cards.

But one bank's card program has been profitable despite losses up to 40% in the customer's first year. Maybe Congress was saying "It's unfair, even if profitable, to lend with a 40% chance of default." The most logical reading of the law is that card credit should not be extended to those who very likely can't pay. On typical sub-prime cards, annual losses can be 15%. That's very high relative to almost all other bank loans.

So, the new requirement could target just the few issuers (only two large ones) who were able to profit while incurring very high credit losses. The only way "ability to pay" would have caused any rational difference in underwriting would be if regulators had interpreted it as requiring a crack down on cards with especially high loss rates.

Now consider the far more numerous banks that issue only "prime" or "super prime" cards. Fewer than 3% of these are charged off every year. It follows that the overwhelming majority of these customers must have had "ability to pay." The proof of the pudding is the eating. They pay, they keep paying!

How do we achieve this low frequency of defaults? Primarily by using models and criteria based on credit bureau attributes. Namely, the best measure of propensity and hence ability to pay is the customer's actual pattern of past payments. These models have been continually improved and updated for decades. They do far more than just project that a customer who's paying a lot every month, could pay still more. But whatever is in the secret formulas, they work--even during deep recessions.

Use of checking account data can substantially improve the models, particularly in line management for current customers. However, "income" numbers play only a modest role in the models, because any arbitrary stab at "income," such as stated income, has been much less predictive of card payments than is payment history. That makes intuitive sense.

In response to the new requirement, however, lawyers, compliance experts and model builders, with excessive help from vendors with formulas and data to sell, came up with new credit screens based on estimates of income. A customer who qualified under earlier models will get the credit for which he is qualified only if he passes an additional screening based on his supposed income.

The result has been an increase of as much as 30% in declined applications and a decrease of as much as 50% in dollars of credit granted per 1,000 applications or per marketing dollar. Much more cost and much less credit.

If two people have equal chances of paying on time but one of them has lower income, then offering less or no credit to the one with lower income — which is exactly what most banks are now doing — will fail the Reg. B "effects test" by discriminating against protected classes such as women. The ability-to-pay requirement does not establish a business necessity for this, because compliance does not require discrimination.

This deflection of lending is bad for consumers, terrible for banks, and also harms prospects for economic recovery — in addition to its likely illegality. Only banks, not regulators, can fix it.

Andrew Kahr is a principal in Credit Builders LLC, a financial product development company, and was the founding chief executive of First Deposit, later known as Providian.