What does it mean when the head of the International Monetary Fund says that the centuries-old model of fractional banking is broken? Maybe about as much as when the Fund first pronounced that Greece would do just fine, but I don’t think so. The combined force of asymmetric regulation and warp-speed innovation is redefining finance around traditional banks, many of which have left strategic planning to languish as they wrestled with one after another costly new rule. The more banks complied with post-crisis rules, the more non-banks redefined banking in their own image, cherry-picking the traditional financial-intermediation model apart to pull out the choicest bits. Almost ten years after the close brush by the Grim Reaper in 2008, U.S. banks now face death by a thousand cuts. This wouldn’t matter to anyone but bank shareholders if macroeconomic, financial-stability, income equality, and consumer protections were secure. But, they’re not – Equifax’s debacle shows just what can happen when core banking functions go beyond the regulatory pale.
Equifax isn’t usually classified as a fintech firm. That honorific is reserved for new entrants with lots of hoodie-clad employees, not a decades-old credit bureau. But what went wrong at Equifax shows clearly how tightly linked banks are to “partners” regardless of how fintech-like the partner is when the bank delegates core functions or critical infrastructure. Cost-efficient? Quite likely, but safe, sound, and sustainable under stress?
As I discussed in two speeches earlier this week, sacrificing core functions redefines banking in one of three business models. As the Basel Committee rightly categorized them, these are a distributed, relegated, or disintermediated bank. Basel has another business-model option – the “better bank” – but one striking feature of its analysis is how pessimistic it is that banking can indeed be better.
Again, Equifax is instructive. Credit bureaus got their start by providing the data banks needed to prove they weren’t discriminating on racial or, later, gender criteria once it became illegal to do so. Instead of determining sound credit-underwriting criteria, lenders delegated first the scoring models and then the data needed to build them to third parties which couldn’t be accused of cooking the books given the “objective” nature of the credit-reporting system. Very quickly, credit decisions were made not on character – all too often subject to judgments based on what someone looked like, not what they were – in favor of simple scores. No need to know your customer – just call the credit bureau.
Who came to own the customer as credit underwriting grew ever more dependent on the credit bureaus? The bank owned the risk, but the credit bureau knew the customer and thus made the credit go/no-go decision. Although banks increasingly became automated-underwriting dispensaries, not lenders, credit-bureau dependence could make sense if credit-bureau scores were right. The 2008 crisis proved in very costly fashion that lots of loans based on credit scores can turn into lots of totally unanticipated losses for banks, borrowers, and the financial system but not the credit bureaus.
A recent NBER study we assessed found that credit growth between 2001 and 2007 was concentrated in the middle and at the top of the credit-score distribution, while borrowing by those with low credit scores was virtually constant during the boom. Even more interesting, growth in defaults during the financial crisis was concentrated in the middle of the credit score distribution. In short, those who relied on scores got it very, very wrong and a systemic crisis ensued.
Of course, banks weren’t alone in their reliance on credit scores. The GSEs also built their underwriting models then and now on credit scores. One might have thought that the hard way we learned that third-party scores are as reliable as third-party credit ratings at predicting credit risk would have chastened credit underwriting. Quite the reverse – as of this writing, industry credit underwriting is still more dictated by credit scores because – guess what – CCAR and a lot of bank regulatory-capital standards are too.
All of this market reliance on credit bureaus has made credit bureaus systemic-risk epicenters. The worse their scores, the greater the credit risk, but their operational-risk costs also translate quickly into financial risks across the U.S. financial system. In the near term, Equifax’s risk is reputational. Customers subjected to identity theft will not seek redress from the credit bureau – banks are by law responsible for making them whole. With about half the U.S. population at risk, the tab could be very, very big. If it comes in concert with an economic downturn, the earnings impact will be similarly large. That’s the first cost when banks permit a third party to make key decisions for it – the approach Basel dubs a distributed-bank model because the bank ceases to control key aspects of the customer relationship. In this case, it’s actually knowing anything about who the customer is and how risky he or she might really be. The credit score number is supposed to tell all.
Is that it for the Equifax risk and by inference all the others that come when banks distribute risk or are relegated or disintermediated by more potent competitors? I fear not. Credit-bureau ratings and the lending decisions based on them have had significant and adverse impact on first-time entrants into the U.S. credit market. As a result, established borrowers – who may well be the riskier for all their prior indebtedness – get the lower-cost loans while lower-income aspirants get higher-cost, less-sustainable loans if they get any at all. This is risky on its face, as the NBER paper shows. But it’s also a potential accelerant on the burning fire of economic equality.
As a new paper we have analyzed shows, income inequality may well be the most important predictor of financial crises. That is, the less equal America is, the riskier our financial market regardless of all the new rules designed to make big banks impregnable financial fortresses. Credit scores aren’t the only reason for U.S. economic inequality – far from it. But the more lending is based on their calls and the more these calls mistake wealth for repayment capacity, the worse economic equality gets.