Consumer credit bureaus and FICO
Each of the 3 major US consumer credit bureaus (CCBs) – Equifax, Experian, and Transunion – are an agglomeration of regional credit agencies, some dating back to the early 1800s. That a once fragmented industry consolidated into oligopoly is a natural consequence of the benefits that come from pooling data: a lender will go with a national bureau over a regional one because when they run a check on a borrower’s credit history the former is more likely to return a result and that result, coming from an aggregator that captures credit lines opened at more banks, is also more likely to reflect a complete borrower record. In accessing this database, the lender also contributes its borrowers’ credit limits, balances, delinquencies, defaults, etc., further reinforcing the national bureau’s dominant position.
In the consumer credit ecosystem, as important as data is a benchmark against which to gauge risk across borrowers and over time. In the US, that benchmark is the FICO score. Each of the 3 bureaus license FICO’s algorithms and run their data through them to produce a score that is re-sold to lenders, along with the data itself. Lenders use those scores in combination with credit data and “decisioning models”, built in-house or licensed from the bureaus or FICO, to determine whether and how much to lend to a borrower.
Both layers – credit data and FICO scores – are complementary and near impossible to dislodge. FICO is protected by a classic standards-based moat, which I described in my Moody’s post from 2017. As with a Moody’s rating, FICO’s incumbency reinforces adoption. Banks have FICO scores embedded in dozens of mission critical systems. Fannie and Freddie require a FICO score from all 3 credit bureaus on every conforming mortgage they purchase. Investors use FICO scores to evaluate and compare risk on collateral pools underlying 95% of securitizations. In other words, FICO is woven into the risk management fabric of our nation’s consumer lending apparatus, bound to risk engines, qualified over many years, and understood by investors who ultimately buy the assets that lenders originate. It wouldn’t make sense for a lender to replace FICO for an alternative score, even one that was cheaper and just as good at predicting defaults. Not only is the FICO score a self-reinforcing standard blessed by regulators, but its generation has no marginal cost to FICO, literally. They don’t even use compute resources to run the scores! The credit bureaus do that. FICO is selling math. Its variable cost is air. Operating margins in the Scores segment are closing in on 90%.
The credit bureaus, while not as competitively insulated as FICO, have a decent setup too. It’s not just that their data would be tough to acquire but also lenders’ risk models have been fine-tuned for years on countless permutations of it. Replacing one of the big 3 with an alterative vendor would entail enormous cost, effort, and risk that financial institutions, as conservative and slow-to-change as they are, have little reason to bear.
But still, there are 3 bureaus to choose from and the credit files they offer are pretty similar. For mortgages purchased by Fannie and Freddie, the Federal Housing Financing Agency (FHFA) requires that banks pull data from all 3 (this could be changing soon, more later), neutralizing the competitive instinct to cut prices. For all other loan categories, though, banks can source data from whoever they want and all 3 bureaus compete vigorously to position themselves at the top of the “waterfall”, where they typically get first dibs on ~60%-80% of credit checks (if the preferred vendor fails to return a record, the bank will turn to the second bureau and, failing that, the third), as well as pole position to cross-sell software and analytics.