Skip to navigationSkip to content

How to build a banking benchmark that’s not rigged

A trader works on the floor of the New York Stock Exchange.
Reuters/Brendan McDermid
Some data points are more believable than others.
  • Jason Karaian
By Jason Karaian

Global finance and economics editor

Published This article is more than 2 years old.

Does this business model sound familiar? An intermediary collects market assessments from banks, scrubs them of identifying information, and publishes an average. That average then serves as a benchmark for financial transactions worth millions, if not billions, of dollars.

With minor variations, that is how Libor, gold, foreign exchange, oil, and other key financial-market benchmarks work. And with depressing regularity, all of them have been—or are suspected to have been—manipulated in some way. Why, then, is Elly Hardwick so excited about the start-up she runs, based on a similar business model? And why did the venture capitalists at Index Ventures just sink $7 million into the company?

Hardwick, and her VC backers, seem puzzled when Quartz brings up Libor when talking about the company, Credit Benchmark. That’s because they believe they are building a new banking benchmark the right way—that is, one that is both useful and unlikely to be rigged. (Never say never.)

The loan arrangers

Credit Benchmark aggregates internal assessments from different banks of their institutional borrowers’ creditworthiness—namely, how likely the borrowers are to default, and how much of a loan the banks expect to recoup if they do. It shares the resulting averages (which require scores from at least three banks) with participating lenders. Armed with consensus estimates of credit quality for a particular borrower, a bank’s risk team can see where they stand relative to the market view, and adjust their exposure accordingly. A dozen banks in Europe and the US have signed up so far.

These assessments are more up-to-date than those from credit ratings agencies (which may change only once every few years, and then only within broad qualitative categories like AAA or BB+) and less noisy than the prices in credit-default swap markets (which fluctuate constantly), Hardwick says. They also cover a much wider universe of companies, including opaque sectors like privately-held small businesses, hedge funds, and little-known subsidiaries of larger groups. (Untangling convoluted corporate structures to figure out which legal entities can get funds from others is a key feature of the service, Hardwick notes.) Eventually, the company may expand its dataset to cover specific securities and instruments as well as borrowers.

Checks and balances

So why is Credit Benchmark so sure that the data it gets won’t be fiddled? Crucially, because it comes directly from the systems that the banks already use to calculate how much capital they hold against their various exposures—which is also shared with regulators. And since the service is being launched now, after the rate-setting processes for Libor and other benchmarks blew up, extra checks have been “baked in from the beginning,” Hardwick says. Much of the company’s new funding will be used to hire staff who specialize in data quality and data science.

The ongoing revamp of Libor is reportedly ruffling banks’ feathers because the new benchmark administrator is asking for proof of actual trade data (paywall)  to back up a bank’s submissions. It’s never easy to get banks to part with internal data, says Jan Hammer of Index Ventures, but if enough of them participate and network effects kick in, a (believable) benchmark is born. The data collected by Credit Benchmark is useful, he says, because it’s “a byproduct of what’s already produced anyway” by the banks. Imagine that.

📬 Kick off each morning with coffee and the Daily Brief (BYO coffee).

By providing your email, you agree to the Quartz Privacy Policy.