(Reuters) - A few years ago, famed financial engineer Andrew Lo built a computer model that allowed one of the biggest U.S. banks to figure out which customers were most likely to fall behind on credit-card payments.
More recently, the Massachusetts Institute of Technology finance professor has been laboring for another client: the U.S. government. Working with banking regulators at the Office of the Comptroller of the Currency, an arm of the Treasury Department, he's helping build quantitative tools to find potential credit risk in the banking industry, starting with the mortgage market.
The efforts of Lo, a pioneer in his field, are part of an unprecedented push at the OCC to embrace quantitative analysis. The regulator is building models, hiring financial engineers - known on Wall Street as "quants" or "strats"- and questioning banks to a far greater degree than it ever has before.
The OCC's effort stems in part from the 2010 Dodd-Frank financial reform law, which requires the OCC and the U.S. Federal Reserve to evaluate the quantitative models that banks build and use. But the OCC is going further by building its own models from the ground up so it can check banks' results and monitor the broader financial system.
The regulator hopes that by conducting its own analyses, it might prevent a repeat of the mortgage bubble, which happened in part because underwriters, investors, rating agencies, and other market participants all assumed that U.S. housing prices would not fall.
"We need better models, more responsive models, that can identify risks more quickly," Lo said. "This is going to be a new age of financial innovation in quantitative models."
The 150-year-old regulator hasn't always found it easy to embrace 21st century quantitative analysis. For one thing, experienced quants that can earn upwards of $500,000 on Wall Street aren't necessarily willing to work for the $80,000 to $175,000 a year the OCC is offering.
For another, banks and other regulators say it's not clear that the OCC initiative to aggressively question bank models is worth the effort, particularly at a time when resources are constrained and regulators have many other, arguably more important tasks to accomplish.
Michael Sullivan, who oversees the OCC's quant staff as Deputy Comptroller for Risk Analysis, said the agency has 53 quantitative economists and a support group of seven research associates and financial analysts, compared with 31 and six, respectively, at the end of 2007. Sullivan said his team recently poached a senior quant from another agency, though he declined to specify which one. He still has a few unfilled positions and is deciding whether to expand further.
"We offer a unique job in that it combines direct involvement in supervision and interaction with bankers and examiners to figure out how banks operate and what kind of challenges they face in both research and policy," said Sullivan. "It's that combination that's attractive to the right person."
Those types of challenges are sufficiently interesting to Lo that he is advising the OCC for free. The agency's own staff is working closely with him, and he is not privy to the bank data the OCC works with. He hopes to publish the results of the analysis.
The OCC's efforts are well intentioned, but to some bank executives and even some other regulators they are also maddening.
An executive at one of the largest U.S. banks told Reuters that an argument between OCC quants and the bank's own quants over a valuation model for a particular type of bond took weeks to resolve. According to the executive, there were only minor differences in the models and barely any difference in the outcomes. The bank now runs both models concurrently to appease OCC staff, even though its internal model had already been vetted by the Fed.
A Treasury official who spoke to Reuters about the OCC's approach said the agency is wasting time challenging models that banks have already spent significant resources to develop. Often, this person added, OCC quants have less experience than bank quants and are privately ridiculed for second-guessing models that they do not fully understand.
At the same time, banks themselves are training quants who once valued exotic derivatives to work on compliance and work closely with regulators.
"Just about the only 'growth industry' in the (quant) banking world these days is regulation-based: regulatory compliance and reporting, regulatory examinations, regulatory 'optimization,'" said Leif Andersen, co-head of the quantitative group at Bank of America Corp, in a presentation he made at Aarhus University in Denmark in January. "Fortunately, for quants/strats, much new regulation is complex."
Ashwin Rao, who spent 14 years on Wall Street as a quant and now runs a business called Zlemma, which uses algorithms to match job-hunters and hirers in the science, technology and engineering fields, said quants used to spend 30 percent of their time on regulatory matters before the crisis and complain, but now spend 50 percent to 60 percent of their time on regulatory matters and hate it.
For its part, the OCC is pleased with the program so far, and believes it will have no trouble hiring good talent.
"It's one of those quiet little success stories," said Comptroller Thomas Curry in an interview with Reuters. "It's a very competitive area but one of the attractions we have for the quants that don't want to be totally academic that this is really a practical application of what they do."
Lo said working for the OCC is a career move that might appeal to some of his students. The OCC often asks him for recommendations of students they might want to hire.
"Part of what drives students is the fact that they can make a difference," Lo said. "They want to do good in the world."
Reporting by Lauren Tara LaCapra; Editing by Dan Wilchins and John Pickering