In early November 2010, as the Federal Reserve began to weigh whether the nation’s biggest financial firms were healthy enough to return money to their shareholders, a top regulator bluntly warned: Don’t let them.

“We remain concerned over their ability to withstand stress in an uncertain economic environment,” wrote Sheila Bair, the head of the Federal Deposit Insurance Corp., in a previously unreported letter obtained by ProPublica.

The letter came as the Fed was launching a “stress test” to decide whether the biggest U.S. financial firms could pay out dividends and buy back their shares instead of putting aside that money as capital. It was one of the central bank’s most critical oversight decisions in the wake of the financial crisis.

“We strongly encourage” that the Fed “delay any dividends or compensation increases until they can show” that their earnings are strong and their assets sound, she wrote. Given the continued uncertainty in the markets, “we do not believe it is the right time to allow transactions that will weaken their capital and liquidity positions.”

Four months later, the Federal Reserve rejected Bair’s appeal.

In March 2011, the Federal Reserve green-lighted most of the top 19 financial institutions to deliver tens of billions of dollars to shareholders, including many of their own top executives. The 19 paid out $33 billion in the first nine months of 2011 in dividends and stock buy-backs.

That $33 billion is money that the banks don’t have to cushion themselves — and the broader financial system — should the euro crisis cause a new recession, tensions with Iran flare into war and disrupt the oil supply, or another crisis emerge.

This is the first in-depth account of the Fed’s momentous decision and the fractious battles that led to it. It is based on dozens of interviews, most with people who spoke on condition of anonymity, and on documents, some of which have never been made public. By examining the decision, this account also sheds light on the inner workings of one of the most powerful but secretive economic institutions in the world.

The Federal Reserve contends it assessed the health of the banks rigorously and made the right decisions. The central bank says the primary purpose of the stress test was to assess the banks’ ability to plan for their capital needs. The Fed allowed only the healthiest banks to return capital — and they are still not paying anything like the proportion of profits that they distributed in the boom years. And it says the stress test covered only one year. Regulators say they can revisit their decisions if the economic picture turns bleaker.

Most important, Fed officials argue that the biggest financial institutions still added $52 billion in capital to their balance sheets in 2011 despite raising dividends or buying back stock. The top 19 financial firms had a 10.1 percent capital ratio by the end of the third quarter of 2011, using the measure that regulators primarily look at, nearly double what they had in the first quarter of 2009.

But a wide range of current and former Federal Reserve officials, other banking regulators and experts either criticized the decision to allow dividend payments and stock buy-backs then, or consider it a mistake now.

Among their reasons: Allowing banks to return capital to shareholders weakened American banks’ ability to withstand a major shock. Whether they are too weak remains debated, but dividends and buy-backs matter. From 2006 through 2008, the top 19 banks paid $131 billion in dividends to shareholders, according to SNL Financial. When the financial crisis hit, the banks were weak in large part because they didn’t have those billions. Indeed, in the fall of 2008, the government invested about $160 billion in the top banks.

Today, the European economic and banking crisis, which was looming when the Fed made its decision, continues to threaten the economy. Unemployment in the U.S. remains persistently high, and the housing market fell almost 5 percent last year, according to CoreLogic, a financial information firm.

American banks are suffering metastasizing liabilities from the U.S. foreclosure crisis. A recent settlement with almost all states’ attorneys general covered only part of those costs, leaving many banks bleeding cash to cover legal costs of the robo-signing scandal and other problems related to the housing crisis.

Once banks start paying dividends, it’s difficult for a regulator to get them to stop without panicking investors. Indeed, building investor confidence was one reason the Fed allowed dividends. But by that measure, it failed: This past November, ratings agency Standard & Poor’s downgraded most of the biggest American banks, and financial stocks in the S&P 500 plummeted more than 18 percent in 2011, though they have since bounced back a little.

Many banks are trading below “book value,” meaning the value of their stock is less than what the banks say are the value of their assets. This fact is particularly sobering, because it suggests investors do not trust the banks’ accounting and are skeptical of their future profitability.

Eventually, the banks will have to raise capital to comply with new international standards, to be in place fully by 2019. The Fed’s decision leaves them further from that goal than they would be otherwise.

But the Fed’s stress-test decision was lucrative for shareholders and bank executives, who are increasingly paid in stock. Dividend payments are taxed at lower rates than ordinary income. Merely allowing the banks to pay dividends, buy back stock and pay back the government helped boost shares, albeit temporarily.

“As undercapitalized as many of these banks are, allowing them to return capital, in my opinion, is preposterous. I can’t believe a strenuous stress testing of their mortgage assets, European exposures and other questionable assets would allow them to return capital to shareholders,” says Neil Barofsky, who until March 2011 served as the special inspector general for the Troubled Asset Relief Program (TARP), better known as the bailout.

“Taxpayers should be concerned when banks pay dividends and remain thinly capitalized,” warned Anat Admati, a finance professor at Stanford in a February 2011 letter to The Financial Times signed by 15 other economists from across the political spectrum. “Taxpayers are the ones who are likely to end up covering the banks' liabilities in a crisis.”

Tarullo’s task

The Fed’s decision cannot be understood in isolation. It continued a series of actions — by the central bank and other arms of government — that were generous to the banks. When the government invested hundreds of billions in the banks through TARP, banks didn’t even have to lend out the money, and bankers could pay themselves bonuses. To keep the financial system from collapsing, the Federal Reserve provided more than $1 trillion to the banks in low-interest loans and loan programs, which were highly profitable for the recipients.

Also, the dividend decision came as the Fed was painfully reinventing its regulatory role after being blindsided by the worst economic crisis since the 1930s.

One of the world’s most powerful economic institutions, the Federal Reserve sets interest rates, controlling the supply of money to stimulate the economy and prevent it from overheating. The Fed also regulates American banks. Designed to be insulated from political tussles, the central bank makes its decisions independently of the president and Congress. Its board of governors is appointed by the president, however, with Senate approval.

Chairman Bernanke has promised the Fed will be more open and transparent, but it still conducts much of its bank regulation and supervision behind closed doors on the grounds that disclosures about individual institutions could cause bank runs and financial panics.

Before the financial crisis, bank oversight had long been a backwater at the Fed, especially under former chairman Alan Greenspan, who advocated for deregulation. The glory and promotions within the Fed lay in monetary policy — deciding what level to target for interest rates and preventing inflation or high unemployment. Indeed, before 2008, the Fed’s bank regulation and supervision had been disastrous, failing to prevent or foresee multiple financial crises, including the near-collapse of the entire financial system in 2008.

In the wake of that terrifying experience, the Fed decided it needed to determine just how strong the banks were and whether they could survive another economic shock. So, in early 2009, it carried out the first stress test, a system-wide assessment of how banks would fare under bleak economic scenarios. Following that test, in May 2009, regulators determined that the weakest 10 of the 19 banks needed to add $185 billion in capital by the end of 2010. They named those banks and announced key findings.

The second stress test, conducted from November 2010 through March 2011, is what led the Fed to allow banks to disperse capital to shareholders. And that test differed starkly from the first.

For starters, it was secretive even by the Fed’s standards and certainly by comparison to the first stress test. To this day, the Fed has disclosed little detail about how the second stress test was conducted, and virtually nothing about how it decided which banks could release capital. Unlike its actions in the first stress test, the Fed hasn’t released its estimates of banks’ revenues, post-stress capital ratios or losses for asset classes, such as real estate. It did not even announce which banks passed the test.

Instead, the Fed left it to the financial institutions to disclose the results. In March 2011, 11 of the 19 announced that they had been given the go-ahead to buy back shares, pay dividends or get out of TARP. In the first three quarters of 2011, all 19 paid some dividends on either common or preferred stock, and 10 bought back some stock — a total of $33 billion in common and preferred dividends and share buy-backs in the period, according to SNL Financial. That amounted to almost 5 percent of the most important, bedrock type of bank capital held by the 19 institutions as of the end of the third quarter.

But beyond the specifics, the political climate had also shifted by the time of the second stress test. Bankers were emerging from their defensive crouch, emboldened to push back against what they considered excessive regulation. At the same time, the sweeping 2010 financial reform act known as Dodd-Frank gave the Federal Reserve more supervision responsibility. What the Fed had done in a panic in the aftermath of the 2008 crisis was now codified as one of its central legal duties, and as it tried to adapt to its new mission, it suffered internecine battles.

Responsibility for overhauling and improving the Fed’s bank regulatory efforts rests with Daniel Tarullo, a 59-year-old former Clinton administration official and academic who became a governor in January 2009. Bankers utter his name in hushed and embittered tones, terrified of his aggressive calls for more oversight and capital. With his white hair and square jaw, he recalls a pro football linebacker from a previous era. Yet, his actions have often fallen short of his tough talk, critics say.

He oversees a sprawling, fragmented institution. Twelve regional reserve banks share bank oversight responsibility with the central board of governors, based in Washington, D.C. The reserve banks have historically taken most of the bank supervision responsibilities, while Washington has concentrated on monetary policy. The individual reserve banks are frenemies, sometimes working together but often suspicious of others’ motives and jealous of each other’s clout.

“It’s like ‘Survivor’: You make certain alliances, but that doesn’t mean you won’t cut the throat of the person the next time,” says a former Fed supervisor.

The most powerful of the regional banks, the Federal Reserve Bank of New York, rivals the central board in authority and influence. The two institutions often butt heads. The New York Fed is widely seen throughout the rest of the system as overly protective of the two biggest institutions it supervises, JPMorgan Chase and Citigroup. New York returns the view, believing the Richmond Fed to be captured by the biggest bank it oversees, Bank of America, and San Francisco by its charge, Wells Fargo.

Bank supervisors, especially the ones deployed to work physically inside the banks that they oversee, are, like all regulators, vulnerable to capture by the institutions they police. They sometimes identify with and coddle their banks rather than enforce the rules, according to multiple current and former Fed officials.

“You have to work with these people every day. You develop working relationships,” says a New York Fed official. “You have to put yourself in their shoes, but you have to make sure they are safe and sound.”

Given what its critics call its supervisory neglect, regulatory capture and bureaucratic rivalries, the Fed had a poor understanding of the sector it regulated in the lead-up to the 2008 crisis. “If a supervisor wanted to see [the Fed’s] systems, he would have flunked us miserably,” recalls a former Federal Reserve banking official. “We would joke about that a lot.”

Mandated to overhaul this system, Tarullo centralized supervision — and thus power — in Washington.

“As an academic, I think I came to have a pretty good understanding of the substance of the Fed's regulatory policies,” Tarullo tells ProPublica. “But when I got here, I was surprised that the large institution supervisory process wasn't very well coordinated across firms, and really didn't draw on all the economic expertise of the Federal Reserve.”

When he arrived, Tarullo found that the supervision and regulation division was beset by personal disagreements and turf battles. He found the unique structure of the Federal Reserve complicated bank supervision and regulation, and such problems persist.

“That's why we created the Large Institution Supervision Coordinating Committee — to make strong coordination and interdisciplinary perspectives permanent features of the supervisory function," he says.

Part of that may be due to his management style. Universally regarded as smart, Tarullo doesn’t always look people in the eye. He often leans back in his chair, tilts his head up and addresses the ceiling. The rank-and-file thought he mistrusted them and didn’t listen, according to several former Fed employees.

Tarullo and some at the Fed defend his leadership, but critics say his manner could undermine his push for tighter regulation, especially with the Fed’s old guard. “Our esteemed leader,” was how Patrick Parkinson, a Greenspan acolyte who retired at the end of 2011 as the director of the division of banking supervision and regulation, sarcastically referred to Tarullo, according to a former Fed employee. Parkinson didn’t return calls seeking comment.

Tarullo is known for his temper. He has made a few employees cry, according to people familiar with the incidents. At a meeting of the board of governors in late 2009, Tarullo blew up at Coryann Stefansson, a former supervisor who ran the first stress test in front of the entire board. As she contended that the Fed shouldn’t push Bank of America to raise more capital, Tarullo surprised people in the stuffy Fed meeting rooms, laden with heavy, dark wood furniture, shouting: “How dare you interrupt me?” a person familiar with the meeting recalls.

Yet, Tarullo gave ground eventually on BofA. In the end, the Fed pushed Bank of America to raise more capital than it initially proposed but less than the FDIC wanted.

The Fed and Tarullo declined comment on the incident.

For his part, Tarullo is cautious about his accomplishments, saying, “I don’t want to overstate how much progress has been made, but I do think that with the authority we either had already or have gained from Dodd-Frank, plus the creation of the [large institutions committee] and the requirement of regularized capital planning and oversight, there are at least the makings of something durable.”

The first stress test

At the Fed, battle lines over how rigorously to regulate banks were drawn during the first stress test. Officially known as the Supervisory Capital Assessment Program (SCAP, pronounced “Ess-Cap”), the first stress test certainly forced most banks to beef up their capital. But even coming right out of the gravest economic peril since the Great Depression, the Fed gave the banks concessions.

When it first began to test banks against bleak economic scenarios, the Fed took a conservative stance toward bank plans for the future. If a bank was going to sell a business line or other asset to raise capital, it had to complete the transaction before the Fed would count it toward fulfilling the bank’s new capital requirements.

One debate was over what are known as “deferred tax assets” — losses that can be written off against future profits. If a bank suffers losses for a prolonged period, it can lose the opportunity to take the write-off, and the asset becomes worthless. Initially, some supervisors pushed for a conservative treatment. Capital is supposed to absorb losses in a crisis. Deferred tax assets, or DTAs, can’t do that because they are little more than an accounting concept.

For the stress test, the Fed assumed that banks’ profits would suffer a prolonged hit in an economic downturn, reducing or wiping out the value of these assets. But there was an issue with Wells Fargo. Since it hadn’t lost money even at the depths of the 2008 crisis, should it be able to get some credit for its deferred tax assets?

Janet Yellen, then-president of the Federal Reserve Bank of San Francisco, supported Wells Fargo. The tough-talking Tarullo came around to her view, according to four people familiar with the negotiation. At the end of 2009, banks were haggling over the details of how to fulfill the capital-raising requirements. The Fed allowed Wells to get some credit for its remaining deferred tax assets. The softer treatment set off cascading effects: Supervisors had to scramble to give Pittsburgh-based PNC Bank similar credit because it, too, had weathered the crisis better than other banks and had a large portion of deferred tax assets on its balance sheet.

“It is common — perhaps too common — for reserve presidents to tend to believe that their firms are invariably better than average,” laments a former Fed governor.

“The treatment of Wells' DTAs was fully consistent with a rigorous” stress test, says Yellen, now vice chair of the Fed board of governors, in a statement. She argues that a key issue was that the tax write-offs were imminent, so the risk was minimal that Wells would suffer losses and not be able to use them. “Under Federal Reserve capital regulations, it is appropriate to count DTAs as capital when they are going to be realized in the very near term,” she says in her statement. “After looking at the specific characteristics of Wells’ DTAs, senior Board staff determined that they qualified as capital” for the first stress test.

PNC and Wells declined to comment.

In a speech on May 6, 2010, the one-year anniversary of the tests, Fed Chairman Bernanke called the first stress test a “watershed event,” crediting it with having helped “restore confidence in the banking system and the broader financial system, thereby contributing to the economy’s recovery.”

By the third quarter of 2011, the top 19 banks that underwent the tests had added hundreds of billions in capital. A crucial measure of their capital is known as Tier 1, and it consists mostly of common stock, reserves and “retained earnings” — income that is not paid to shareholders but instead kept by the company to invest in the business. By the end of the third quarter, the top banks’ Tier 1 capital was up to about $740 billion. Using an average weighted to account for the different sizes of the banks, that’s 10.1 percent of their assets, compared with a low of 5.4 percent at the end of 2008.

As the banks built more capital, struggles erupted among various government bodies about when and how to let banks pay back TARP money. Most of the banks wanted to pay back the government as quickly as possible, mainly because the bailout money came with intensified oversight and potential restrictions on how much executives could pay themselves. And with the bailouts deeply unpopular with the public, the Treasury also pushed for the banks to pay back the money as quickly as possible so the government could claim the program was successful and hadn’t cost the taxpayers much.

But how should the banks replace the taxpayer money? Could they borrow to pull together the money, or should they be required to raise what’s known as common equity, the basic type of stock, whose holders absorb the first losses in the event of problems? Doing the latter would force banks to do the hard work of finding investors and amassing solid capital that could cushion them against economic blows.

The Fed led the process to answer this crucial question, with contributions from the FDIC, the Treasury and another major bank regulator, the Office of the Comptroller of the Currency (OCC).

In late 2009, regulators decided that the eight financial institutions that hadn’t exited TARP immediately after the first stress test, including Bank of America, Wells Fargo and Citigroup, would be able to pay back every $2 of TARP money by issuing $1 in new common equity, according to a Sept. 29, 2011, report by the special inspector general of TARP. The banks could raise the other dollars through other ways, such as borrowing.

Yet, almost as soon as they had decided on that standard, the Federal Reserve and OCC relaxed it for some of the most troubled big banks. The FDIC “was by far the most persistent in insisting that banks raise more common stock,” the report found.

The FDIC pushed repeatedly for the banks to adhere to the guidance known as the “2-for-1” provision.

Sheila Bair’s agency was particularly frustrated when the Fed and OCC eased their conditions for Bank of America, one of the most vulnerable banks.

Bair told the TARP special inspector general that “the argument [the Fed and OCC] used against us — which frustrated me to no end — is that [Bank of America] can’t use the 2-for-1 because they are not strong enough to raise 2-for-1.” She said: “If they are not strong enough, they shouldn’t have been exiting TARP.”

The Fed decided it could ignore the FDIC’s views. On Nov. 19, 2009, an unnamed Federal Reserve governor stated that Bernanke’s position was that “we would go ahead without [FDIC] agreeing,” according to a previously unreported email from a draft version of the special inspector general’s report. And indeed, Bank of America fell more than $3 billion short of 2-for-1, raising $19.3 billion in equity to pay off the taxpayers’ $45 billion.

Since then, Bank of America has run into further troubles and been forced to sell assets and raise capital.

The Fed has taken pains to hide such tussles and compromises. The email describing Bernanke’s decision to override the FDIC, along with many others, was excised from the final draft of the special inspector general’s report. In a footnote in the final, published report, the special inspector general wrote that the Federal Reserve “strenuously objected to the inclusion of a significant amount of text” in earlier versions of the report, citing the need to keep communications with banks confidential.

Even though the special inspector general wrote that she “respectfully disagrees” with the Fed, she allowed the emails to be excised from the published report.

Stresses during the stress test

In 2010, as the Fed began the second stress test, officially known as the Comprehensive Capital Analysis and Review (CCAR, pronounced “See-Car”), bankers lobbied heavily to be allowed to return capital to shareholders. They began to feel emboldened to speak out against tightening regulations.

In late 2010, Tarullo had at least two conversations about capital planning, not previously reported, with top bank CEOs: JPMorgan’s Jamie Dimon and Citigroup’s Vikram Pandit. Dimon pressed Tarullo about the Fed’s plans for how much capital large banks should be required to have. The JPMorgan CEO, who has been an outspoken critic of the post-crisis regulatory tightening, argued to Tarullo that “it made sense to differentiate between banks,” says a person familiar with the discussion. “If we were healthy, we should be allowed to pay dividends and buy back stock.”

Bank executives such as Dimon and Pandit stood to gain personally from dividend decisions since much of their compensation comes in the form of stock.

Tarullo says he and the Fed were not unduly influenced by Dimon, Pandit and others who lobbied on behalf of the banks.

Inside the Fed, whether to pay dividends had been hotly debated for years. Less than a year after the height of the crisis, in August 2009, top officials from around the system met with Tarullo in Washington, where they mulled letting banks return capital to shareholders. Some Fed officials were shocked that the regulators were considering returning dividends so soon, according to one attendee.

Margaret “Meg” McConnell, a wiry and intense macroeconomist from the New York Fed, raised the possibility that the Fed might bar even healthy banks from paying dividends, if the regulator thought the environment was still too fragile. This is dubbed a “macroprudential” approach. Generally mild-mannered, McConnell surprised people with her emotion. She spoke “with a bit of pique,” a person at the meeting recalls.

But other Fed officials at the meeting argued that could be dangerous because it would erode investor confidence. They feared such an action might inadvertently signal that the Fed was still worried about the financial system. Preventing even ostensibly healthy banks from returning capital for a period of time might be destabilizing in and of itself. Investors could get the wrong idea and panic.

There were other debates. The European crisis, while not as acute as it would become in 2011, was clearly brewing by the time the stress test began in late 2010. Some supervisors argued that the test should include an evaluation of the banks against some European measure, such as a stock-market index, and make it public. In the end, that was rejected. The Fed worried about creating political backlash by suggesting Europe was in deep trouble, according to a person familiar with the discussion.

“It was important for us to create a regular, annual process to ensure that banks could only increase dividends or buy back shares if they could show they would remain healthy even in the face of adverse economic conditions," Tarullo says to ProPublica. “It’s not reasonable to say that a bank could never return capital to its shareholders, no matter how well-capitalized it has become, but it is reasonable to require that any such action be premised on a sound capital plan and rigorous stress testing.”

The Fed’s legal department acted as a brake on aggressive supervisors. The department, headed by the powerful general counsel, Scott Alvarez, had long served as a de facto overseer of supervision in Washington. The board of governors often asked Alvarez to report on various supervisory topics.

Tarullo appeared frustrated by this back-channel reporting and tried to curtail it, according to people familiar with the workings of the board of governors. He clashed with the legal department, viewing it as too friendly to the banks.

As supervisors conducted the second stress test, the legal department assessed whether the Fed had the authority to stop banks from raising their dividends — a move that surprised some supervisors because the banks themselves had never raised such an objection to the Fed’s power. After all, the Fed has statutory responsibility to maintain the “safety and soundness” of banks. Some staffers interpreted the legal department’s action as a sign that the Fed was looking for ways to hamstring itself, a signal that they should tread lightly when it came to restricting banks from returning capital to shareholders.

The Fed declined to comment and didn’t make Alvarez available.

Once Dodd-Frank was passed in the summer of 2010, making bank supervision a more vital part of the Fed’s mandate, the board of governors wanted to monitor Tarullo’s efforts, so the board requested regular briefings. One governor, Kevin Warsh, a George W. Bush appointee, viewed the Fed’s overall bank regulatory approach skeptically. Something of a libertarian, he thought that the Fed was overly confident in its abilities to monitor banking activities and head off crises before they became acute.

But Warsh, who oversaw financial market activities and not regulation and supervision, also worried that the banking system had too little capital. Do the banks have enough to offset losses? Do their books accurately value their assets? Do they have the proper risk management systems in place? he worried to colleagues. “There is too much confidence that these institutions won’t find themselves in the soup again,” Warsh tells ProPublica.

But the Fed had boxed itself in with its standard: The regulator had told the banks that if they hit their capital requirements under the stress-test scenarios, they could pay dividends and buy back stock.

Ultimately, the Fed did not allow every bank to increase dividends. Some banks, like Citigroup, didn’t request an increase after a signal from the Fed that it would be turned down.

In at least one instance, a signal was misinterpreted. Early in the process, the Richmond Fed left Bank of America with the impression that it would pass the stress test and be allowed to raise dividends. Encouraged, the bank asked permission.

In late 2010, Chief Executive Brian Moynihan suggested to investors that a raise would come in the second half of 2011. But in the end, the Fed nixed any dividend raise. The Fed and Bank of America declined to comment on this incident.

A triumph for Tarullo, the episode nevertheless harmed investor confidence in Bank of America and its management, becoming one of the more embarrassing in Moynihan’s tenure at the bank.

The FDIC vs. the Fed

To some regulators, the second stress test seemed like little more than a formality with officials inclined from the outset to allow most banks to return money to investors.

“Institutionally, the decision was already made” before it was completed, says a former senior regulator who was involved in the testing. A Fed spokeswoman says that was not true.

Still, when a senior regulator familiar with the FDIC’s thinking heard that the Fed was considering allowing banks to return capital to shareholders, “my first reaction was: You’ve got to be kidding me. We are still in the middle of the crisis,” the official recalls. “We believed the banks didn’t have the structure for capital distribution, for dividends and stock buy-backs.” The stress tests “continued to get better, but they were not picking up the full gamut of risks.”

The banks had portfolios of underwater mortgages, and growing legal problems stemming from their pre-crisis actions and post-crisis foreclosure practices. Given the ongoing uncertainty about the global economy and the fragility of the world’s financial system, FDIC officials repeatedly voiced concerns to their Fed counterparts.

As the Fed began the test, Bair wrote her Nov. 5, 2010, letter to Bernanke. “We would prefer to see a longer period of stability, sustained core earnings growth, and strengthening of capital buffers before dividends are considered,” she wrote.

The letter pointed out that “once the level of dividends increases, it is difficult to scale back.”

One major concern was accurately measuring legal liabilities. Banks that had assembled mortgage-backed securities often faced accusations of fraud or deception from investors in those securities. Now, the banks faced a threat that courts would force the banks to take back billions of dollars’ worth of toxic mortgages, known as “put-back” risk. With input from the legal department, the Fed had come up with a system-wide estimate for this risk that the FDIC considered too low, according to two people familiar with the process.

Worse, by the fall of 2010, banks had an emerging legal threat to worry about: foreclosure problems. In the aftermath of the housing crash, banks had abused the rights of homeowners in the process of foreclosing. The most publicized issue was “robo-signing,” in which banks had documents automatically notarized by people who weren’t reading the materials or checking for inaccuracies. As this threat emerged, banks came under investigation by state attorneys general and faced billions in new potential legal liabilities.

“The highly publicized mortgage foreclosure process flaws provide an example of how quickly material issues can arise in these institutions and how they are still exposed to the poor decisions made in the years leading up to the crisis,” Bair warned Bernanke in her letter.

Estimating these future liabilities was a task the Fed delegated mainly to the banks. In a Dec. 12, 2010, speech, Tarullo said the Fed expected “that firms will have a sound estimate of any significant risks that may not be captured by the stress testing, such as potential mortgage put-back exposures, and the capacity to absorb any consequent losses.”

But the various foreclosure problems were just emerging, so estimating those liabilities was difficult — though it was clear that they could be huge.

The FDIC was puzzled. “The direct connection between the put-back issue and the stress test never has been clear to me. They didn’t take the number and add it to the bottom line,” says the senior regulator familiar with the FDIC’s position. “They didn’t have any sizing on broader servicing liabilities.”

All the more reason, FDIC officials thought, to slow down the dividend payouts and stock buy-backs.

The Fed did have an effort parallel to the stress test to assess these liabilities, and, according to a Fed spokeswoman, eventually included the legal risks in the stress test.

(In February, banks settled robo-signing problems with state attorneys general for $25 billion but remain on the hook for other legal liabilities arising from their mortgage servicing.)

By March 2011, it was clear the FDIC had lost its argument. The stress test was winding up, and most of the big banks would get a green light. The agency made a last stand on one bank: SunTrust, a large regional bank based in Atlanta.

The Fed had determined that SunTrust passed the stress test and could exit TARP. Bair appealed to Tarullo, according to several people familiar with the matter. The objection has not previously been reported.

The FDIC didn’t think SunTrust was ready to pay back the government and leave the program. It was one of the last big banks to still have TARP money. It was loaded with high-risk assets, such as interest-only, adjustable-rate mortgages and loans to borrowers with low credit scores.

But the Fed shunted aside Bair’s appeal and allowed the bank not only to repay the government but to do so on indulgent terms that the FDIC thought left SunTrust still vulnerable. When it green-lighted SunTrust’s exit, the Fed didn’t adhere to the 2-for-1 replacement standard regulators had established for healthy banks. The bank issued only $1 billion in new common stock to repay the government’s $4.85 billion in preferred stock.

The FDIC had lost again.

Two former Fed officials who held senior posts during the crucial decision-making say they were troubled, believing that SunTrust didn’t have enough capital. “I was horrified,” says one of the officials.

Since then, investor views have mirrored those of the Fed’s critics. In 2011, SunTrust shares tumbled more than 41 percent. The bank has a market value of about half of the value of the assets on its balance sheet — a sign that investors suspect the bank is overstating the worth of its assets and exaggerating its overall health.

SunTrust and the Fed declined to comment.

Looking ahead

Some see allowing dividends and stock repurchases as a confidence-building exercise. The hope is that investors will believe that the Federal Reserve is confident in the banks and will buy bank stocks. Higher share prices mean that banks can sell stock more cheaply if necessary. The regulators’ logic seems to have been: Let the banks deplete capital to raise capital.

In the second stress test, “how much was appearance, and how much was reality? The Fed wanted the appearance of strength,” says a former senior regulator. “The Fed wanted everyone to see that the banks were profitable, back to normal and back on the road to health. ... And the banks wanted it.”

Passing the banks makes the Fed look good, too. The Fed “wanted it as a symbol of their success in mending the banks,” the senior regulator says.

Tarullo strongly defends the decision to allow some of the banks to return capital to shareholders.

“If you imagine a truly severe financial dislocation, it’s not going to matter much for the health of the U.S financial system whether banks paid out five or eight percentage points more of earnings in the preceding year,” he says. "What will matter is that the banks have been steadily building capital to much higher levels than existed before the financial crisis and that they are subject to annual stress tests to make sure they have the capital needed to withstand a quite adverse economic situation.”

Indeed, the Federal Reserve is at it again, conducting another stress test of the biggest banks. The Fed is testing the banks against much more dire scenarios than it did a year ago, which some analysts see as an implicit admission that it was too soft in the earlier test. This time, in order to comply with Dodd-Frank, the Fed must make much greater disclosure of how the tests are conducted and which banks pass.

The tests are draconian. They require the banks to plan against a scenario in which, among other drastic occurrences, the Dow Jones Total Market Index crashes to 5,668 and gross domestic product falls four quarters in a row, including one 8 percent quarterly drop, almost as much as it did in the fourth quarter of 2008, and unemployment rises to more than 13 percent. Can any of the banks truly survive such a scenario? And will the weaker ones be restricted from buying back stock or paying dividends?

The Fed seems to have put itself in a bind. Either the banks don’t pass, which could harm investor optimism and thus the fragile economic recovery. Or the banks somehow do pass, risking the Fed’s credibility in the event of another crisis.

The results will be out in mid-March.

Correction: A previous version of this story incorrectly dated the letter from Anat Admati to the Financial Times. The letter was published in February 2011. It also misstated the name of a stock index the Fed is using in its new stress test, scheduled to be completed this month. A scenario the Fed is using involves the Dow Jones Total Market Index crashing to 5,668, not the Dow Jones Industrial Average.