The test got harder, yet the pupils scored better.

The Federal Reserve Board concluded Thursday that the nation's 33 biggest bank holding companies are collectively better prepared to withstand a severe recession than they were roughly a year ago. Against a dark macroeconomic scenario hypothesized as part of the Fed's stress testing process, the firms were projected to see a key capital ratio fall from 12.3% to 8.4%.

The industry's losses in that scenario would be substantial – the Fed projected that loan losses would total $385 billion over the course of more than two years. Still, the average ratio of common equity Tier 1 capital to risk-weighted assets would be slightly higher than the 8.3% found in last year's tests.

The Fed also found that the banks, which hold more than 80% of U.S. banking assets, would lose a smaller percentage of their capital cushions in this year's Dodd-Frank Act Stress Test than they would have lost under last year's test.

Those improvements are in spite of the fact that the severely adverse scenario this year is harsher than last year's. The most negative scenario envisioned as part of the 2016 stress tests includes a severe global recession, a 5-percentage-point rise in the U.S. unemployment rate and negative yields for short-term Treasury securities.

Given the Fed's tougher grading, banking industry representatives bragged about their report card on Thursday.

"The tremendous effort banks have made to build capital and liquidity has allowed them to perform strongly even under scenarios that are unrealistically severe," Rob Nichols, president of the American Bankers Association, said in a statement.

Earlier in the day, the head of another banking trade group told a Senate committee that the stress tests and other post-financial-crisis measures have made the banking system safer. Greg Baer, president of The Clearing House Association, urged policymakers to start focusing more on the possibility that the next crisis could emerge from the so-called shadow banking system.

"We do not know what geopolitical shock or asset bubble will cause such a crisis," he told the Senate Banking Committee. "But the chances of its first victims being banks – with three times the capital they held before the last crisis, and with much of their assets held in cash or Treasuries – appear extremely low."

The Fed said Thursday that each of the nearly three dozen banks that were stress tested emerged from the severely adverse scenario with a ratio of common equity Tier 1 capital to risk-weighted assets above the regulatory minimum of 4.5%. Last year, all 31 banks that were tested under DFAST also met the regulatory minimums.

Among the top performers this year were Discover Financial Services, American Express and Bank of New York Mellon; each of those firms was projected to weather the severely adverse scenario with a capital ratio of at least 11%.

Toward the lower end of the results table were Huntington Bancshares, BMO Financial, KeyCorp, Ally Financial and BBVA Compass; each of those firms was projected to have a capital ratio of 6.5% or less at the end of the Fed's nine-quarter scenario.

Among the nation's largest banks, Citigroup was projected to see its capital ratio fall from 15.3% in the fourth quarter of 2015 to 9.2% under the severely adverse scenario. A senior Fed official said that the biggest banks were expected to experience larger declines in their capital ratios than midsized peers that were also tested against the severely adverse scenario, reflecting larger proportional losses.

But the same Fed official also said that the 33 firms that were tested have been building up their capital, and have made improvements in their ability to assess the risks that they are taking.

"Since 2009, these firms have added more than $700 billion in common equity capital," the Fed said in a press release.

In the 15 months between the Fed's last two stress tests, the U.S. economy has improved, and so has the credit quality of banks' loan portfolio, said David Wright, a former San Francisco Fed official who is now a managing director at Deloitte & Touche.

"I'd say that the report is hopeful, in that it's showing progress by banks in terms of the quality of their portfolios and other positions, and that the improvements in the economy since the last stress tests are helping firms weather these storms better," he said.

But banks are hardly out of the woods.

The Fed is scheduled on Wednesday to release the results of the Comprehensive Capital Analysis and Review, a second round of stress testing that is generally considered more important than those that were published Thursday.

One key difference between the two stress tests is that DFAST is based on a standard formula of capital payouts, while CCAR takes into account each bank's specific plan for returning capital to shareholders.

As a result, DFAST better enables comparisons between banks than the CCAR does. But because DFAST does not take into account each bank's actual capital allocation plan, its real-world implications are more limited.

Another key difference between the two tests is that DFAST is a purely quantitative exercise, while CCAR also includes a qualitative element.

"Meeting the quantitative benchmark is just the first step," Anna Krayn, senior director at Moody's Analytics, said in an email Thursday. "We are expecting to see greater divergence in next week's results, once the focus turns to qualitative aspects and proposed capital distributions."

There is no penalty in DFAST for falling below the regulatory minimum standards, whereas failing to pass CCAR carries the possibility of significant consequences.

Any bank that falls short on CCAR's quantitative test will have the opportunity prior to Wednesday to take a so-called mulligan – that is, to make revisions to their capital distribution plan in order to meet the minimum regulatory standards.

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.