By Pete Schroeder
WASHINGTON (Reuters) -The U.S. Federal Reserve is poised to unveil an overhaul of its annual "stress tests" of large banks, handing the industry a long-sought win over an exercise they have long complained is opaque and onerous.
The Fed is set to meet on Friday to vote on a proposal that is expected to dramatically change the annual exam, shifting it from a closely-held exercise where banks are told little about the process to one in which the U.S. central bank opens up its models
and the scenarios it devises for each year's exam.
Banks have complained for years that the test should be more transparent and predictable, and that message is being echoed by President Donald Trump's appointees, like Fed Vice Chair for Supervision Michelle Bowman, who is leading the overhaul.
The examination of large banks' resilience became an annual exercise following the 2008 global financial crisis, as the Fed sought to determine how banks would perform during a hypothetical severe economic downturn. The results dictate how much capital the big banks must hold in the coming year, with greater losses in the test requiring a larger "stress capital buffer" against potential losses for each firm.
But proposed changes to make the test more transparent are likely also to make it more predictable, allowing banks to more precisely set aside capital and deploy excess funds via more lending, dividends or share buybacks, according to analysts.
"The Federal Reserve's disclosures should make the annual stress test less volatile. That should allow banks to reduce the amount of extra capital they hold to guard against this volatility. It also should help banks better optimize their use of capital," Jaret Seiberg, an analyst with TD Cowen, said in a note.
The feud over stress tests came to a head late last year, when banking industry trade groups sued the Fed over the exercise. The lawsuit was postponed after the central bank said it was already considering several of the changes sought by the industry, including making its testing models and scenarios public and responsive to feedback.
(Reporting by Pete Schroeder; Editing by Paul Simao)
