The US Federal Reserve is strongly considering a proposal to publish the implied losses for sample portfolios under its stress-testing models – a move that could help banks align their capital models more closely with the regulator’s expectations.
Risk.net understands the idea of disclosing model-implied losses for a set of hypothetical portfolios – which was first floated by former Fed governor Daniel Tarullo in a speech in April – is gaining traction with bank supervisors. Such transparency would allow banks to make a fairly accurate inference of the expected stressed losses on any given set of assets.
“Showing the world indicative loss rates on these hypothetical portfolios would provide even more information on how different risk characteristics impact our model,” Timothy Clark, deputy director of supervision and regulation at the Federal Reserve Board, said in a recent interview with Risk.net.
Bank risk managers say such disclosures would take some of the guesswork out of the Fed’s annual stress test, the Comprehensive Capital Analysis and Review (CCAR). “This will be very helpful to move banks towards a more level playing field,” says Jimmy Yang, global head of credit and operational risk analytics at BMO Financial Group in Chicago.
More data on the Fed’s model outputs might help banks identify discrepancies in their own CCAR models. “Hypothetical portfolios from the Fed with the associated impacts would be a helpful data point,” says a stress-testing expert at a large US bank. “There are still quite a few variances between the Fed’s results and those of participating banks, and this may help bridge the gap around key assumptions and model methodologies.”
Lourenco Miranda, a stress-testing expert at Societe Generale, says the Fed’s model-implied losses could be a useful benchmark for banks, similar to Basel III’s standardised approaches for calculating minimum capital requirements for market, credit and operational risk. “I like the idea of having the same view as Basel, with standardised portfolios and methodologies for comparison purposes,” he says.
The model-implied losses for the Fed’s hypothetical portfolio could be used to create a ‘look-up table’ of estimated stressed capital requirements for specific assets. For instance, the data may allow banks to infer the required capital and expected losses for a four-year, BBB-rated unsecured commercial loan to a manufacturing company. “It could be helpful for all banks to use the same look-up table to assign capital and stressed losses,” says Yang at BMO.
The difficulty of the mapping depends on how close the characteristics of the bank’s portfolio are to the hypothetical portfolio
Michael Alix, PwC
Banks have long complained that the secrecy around the Fed’s models has made it difficult to meet the requirements of CCAR. However, Tarullo warned in his speech in April that revealing the inner workings of the Fed’s models would repeat an error made in the 1990s, when Congress established revised capital standards and stress tests for Fannie Mae and Freddie Mac. The stress-testing models were made public and any changes were subject to the standard notice and comment process.
“With the model in the hands of the government-sponsored enterprises, even a scenario of the severity of the 2006 to 2008 experience produced only mild losses for them,” said Tarullo. “This result stands in stark contrast to the actual losses, which were sufficient to drive them into conservatorship in September 2008.”
Randal Quarles, who succeeds Tarullo as the Fed’s vice-chair for supervision, has backed calls for greater transparency, however. “The benefits of transparency outweigh any theoretical costs. If you are clear about what it is you expect, you will get more compliance,” Quarles said at his Senate confirmation hearing on July 27.
If the Fed goes ahead with the proposal, much will depend on the granularity of the disclosures and the extent to which the data aligns with the Fed’s models. “If, for example, the Fed segments a credit card portfolio into eight Fico bands when it models stressed losses, but for the hypothetical portfolio segments the portfolio into only two less granular Fico bands, banks will be unable to reverse-engineer the Fed’s models,” says Arthur Angulo, a managing director at Promontory Financial Group, and a former senior supervisor at the Federal Reserve Bank of New York.
Even if the Fed disclosures are highly granular, the process of mapping the hypothetical portfolios to an individual bank’s holdings will be difficult, says Societe Generale’s Miranda. “The complexity and idiosyncrasies of the different CCAR banks is so wide that hypothetical portfolios would not be able to cover this diversity,” he says. “The idea of hypothetical portfolios makes sense, but my scepticism is related to the execution.”
Michael Alix, leader of the risk advisory practice at PwC and a former senior vice-president at the New York Fed, agrees: “The difficulty of the mapping depends on how close the characteristics of the bank’s portfolio are to the hypothetical portfolio.”
He says there is little chance of banks using the implied losses for hypothetical portfolios to reverse-engineer the Fed’s models, no matter how granular the detail the Fed provides. “More granularity provides greater ability to roughly determine what the Fed models look like, but I don’t think they’re going to get to the point where they could determine every line of code,” says Alix. “There’s no way they’ll be able to do that.”
The week on Risk.net, October 6-12, 2017Receive this by email