Stress Test

by Timothy F. Geithner

  The deterioration of the U.S. fiscal position before the crisis, the product of the Bush administration’s deficit-financed tax cuts and spending increases, helped limit the political appetite for a more aggressive and sustained fiscal response to the crisis. We could have afforded to do more than we did for a longer period of time, and the world would have financed it at low interest rates, but with the deficit already projected to be over $1 trillion before President Obama took office, there was significant opposition in Congress and among the public to additional doses of stimulus. Profligacy in good times can painfully constrain a government’s ability to respond to crises, substantively and politically. By contrast, the Fed came into the crisis with a strong reputation for keeping inflation low, which helped Chairman Bernanke pursue such aggressive monetary policy actions without a damaging increase in inflation expectations, and helped make those actions more effective than they would have been without market credibility.

  But no matter how responsibly governments and central banks act before a crisis, they need emergency authority that can be deployed quickly on a massive scale when a crisis hits, just as heads of state have the power to respond to a military attack. This authority should be reserved for the most dangerous situations, and there should be some uncertainty about its deployment to reduce the risk of moral hazard, but it should come with a lot of discretion and a lot of force.

  What should be in the toolbox? The vital tools are: an ability to extend the lender-of-last-resort authority to provide liquidity where it’s needed in the financial system; resolution authority to allow the orderly wind-down of failing financial firms; and, along with deposit insurance to prevent It’s a Wonderful Life–style bank runs, broader emergency authority to guarantee other financial liabilities.

  We went into our crisis with a toolbox that wasn’t exactly empty, but also wasn’t remotely adequate for our complicated and sprawling modern financial system. At times it felt like we were fighting World War III with General Washington’s army. The Fed did use its authorities to provide liquidity throughout the shadow banking system, while supporting markets like tri-party repo and commercial paper. The FDIC had an excellent emergency room for banks of modest size, but the government did not have resolution authority for complex giants like Bear or Lehman or AIG, so we had no way to wind them down through an orderly process. The U.S. government as a whole fell too far behind the curve of the panic, lurching from emergency to emergency, averting disasters with duct tape and string and ad hoc authorities, failing to persuade the markets we had the ability to stop the run.

  It was only after Hank Paulson and Ben Bernanke persuaded a reluctant and angry Congress to give us the authority to inject capital into troubled firms, and the FDIC agreed to guarantee bank debts, that we were able to break the momentum of the panic. We eventually put out the fire, and Congress ultimately provided resolution authority in Dodd-Frank, but we would have put out the fire a lot faster if we had had all those tools from the start, and that would have limited the damage suffered by millions of Americans.

  While Dodd-Frank’s resolution authority will help firefighters in the future, its elimination of the FDIC guarantee authority will hinder the response to the next crisis. And the final legislation took away the Fed’s ability to intervene with specific firms, as it did with Bear, AIG, Citi, and Bank of America; that power wouldn’t be as necessary if the FDIC still had broad guarantee authority, but losing both could be disastrous.

  These firefighting authorities do create some moral hazard, just as the existence of smoke alarms and firehouses make smoking in bed feel somewhat less risky. But it’s a lot better to have the power to put out the fire. The potential benefits of avoiding a depression far outweigh the potential costs of saving people who don’t deserve to be saved. And when you have the ability to provide guarantees and other emergency assistance that can prevent a financial fire from spreading, you can afford to let the fire burn for a while.

  In fact, once the time for prevention and preparation is over, letting the fire burn for a while is the right opening move.

  IN THE early stages of a financial fire, there’s no way to be sure if it’s systemic, if the problems of a few firms reflect pervasive problems that will lead to a broader panic. Substantial increases in private credit, financial leverage, and short-term financing can all indicate a system’s vulnerability to a severe crisis, but there are no clear thresholds that determine the boundary between the modest and catastrophic shock, the five-year and hundred-year flood. You have to feel your way.

  This fog of diagnosis is among the toughest problems for a crisis manager. If you treat every shock as a catastrophe and ride to the rescue too quickly, you can create too much moral hazard and plant the seeds of a future catastrophe. Protecting private investors from the consequences of their risk-taking encourages risk-taking in the same way generous flood insurance can encourage excessive development in floodplains. But if you assume every shock will be benign and escalate too slowly, you risk letting a panic gain too much momentum. Once you fall behind the curve of a crisis, it can be hard to catch up.

  There’s nothing wrong with starting out tentative, as long as you’re prepared to get aggressive if circumstances change. Letting the fire burn can help determine the severity of the crisis. It can also wipe out the weakest firms and the riskiest financing mechanisms, laying the groundwork for a restructuring of the financial system the way a natural wildfire can clear out the underbrush and improve the health of a forest.

  But you don’t want to let it burn too long or too hot. You don’t want the fire to gain enough force to create a general panic that can consume relatively strong firms along with the hopelessly insolvent. And the line between containable and uncontainable is not always easy to discern. You can afford to start slow only if your government and central bank have the authority and the will to accelerate quickly into something approaching overwhelming force. Ideally, you want to provide just enough liquidity and other support for the system to prevent it from falling apart, but not so much that you sustain unsalvageable firms or unsustainable asset prices. The goal is to let the system start to adjust and deleverage without tipping into panic and collapse, which sounds a lot easier in theory than it is in practice.

  In any case, once the fire burns too hot, once the mania turns to panic, once the diagnosis is complete and dire, the time for tentative is over.

  THE INCONVENIENT truth of financial-crisis response is that the actions that feel right are often wrong. The natural instinct is to wait as long as possible before intervening, to escalate as gradually as possible, to minimize taxpayer exposure to losses, to impose stringent conditions on assistance, to teach the arsonists a lesson, to address the root causes of the crisis. Let failing firms fail. Let the creditors who financed their binges pay the price. But that is a recipe for making a systemic crisis worse. The public will want Old Testament justice, punishment for the venal. The moral hazard fundamentalists will want to send a message that irresponsible behavior will not be rewarded. If policymakers listen, they will court disaster.

  The principles of effective crisis response are mostly counterintuitive. The more you commit to do, the less you’ll have to do. If you take the extreme risk out of the market, you’ll assume less risk of extreme losses, and you’ll attract more private capital to provide stability that would otherwise require government capital. You should err on the side of doing too much rather than too little; you’ll make mistakes no matter what, but you should try to make the mistakes that are cheaper to correct. It’s easier to arrest a financial panic than to clean up after an economic disaster.

  Eventually, you’ll want to address the root causes of the crisis, to reform the financial system, to rein in excessive leverage. But as Saint Augustine said, not yet. In an emergency, you need to lean against the forces of panic, to restore confidence, to reduce uncertainty, to make the system investable again. That means no messy failures of systemic firms and no haircuts that woul
d encourage runs. The goal is to make it irrational to run. There will be intense pressure to let major firms fail, avoid moral hazard, minimize government intervention. But that’s a formula for a larger crisis that will ultimately require greater government intervention and create more moral hazard.

  When the government does intervene, it will be tempting to try to maximize the immediate losses absorbed by the private sector, as the FDIC tried to do when it haircut WaMu’s senior creditors. But in a severe crisis following a boom, the private sector is already by definition overextended and in retreat. It doesn’t have the financial capacity to overwhelm the forces of panic, and shifting more risk its way will only amplify the panic, the way the WaMu haircuts prompted Wachovia’s creditors to run. In general, the more the government commits to do up front, the less it will end up doing. The more risk taxpayers take up front, the less they end up paying.

  The original Powell Doctrine emphasizes that military policymakers should go to war only as a last resort, and that is true of financial policymakers as well. But it’s also true for financial policymakers that once war is unavoidable, you need to commit to overwhelming force—a combination of fiscal policy, monetary policy, and financial firefighting. None of those instruments will be powerful enough alone, and weakness in any of them will undermine the effectiveness of the others.

  You should also plan for a long war. Long credit booms can produce a lot of damage that can require years of sustained government support for the economy to heal. The deleveraging process makes post-crisis recoveries slow and fragile. Governments tend to step on the brakes too early, weakening the recovery, adding to the economic costs as well as the fiscal costs of the war.

  The playbook for monetary and fiscal policy is fairly simple. You want to be as expansive as possible, providing substantial stimulus for the economy for as long as necessary. After a major shock that depresses demand and creates a risk of deflation, central bankers should ease monetary policy, aggressively lowering interest rates. Once the overnight rate approaches zero, they should find new ways to stay on the accelerator, as Ben did through quantitative easing. They need to signal that they’ll eventually hit the brakes, and that they’ll remain vigilant about inflation going forward, but the threat of future inflation is much less worrisome than the threat of imminent deflation and depression. Loose monetary policy can have limited power in a crisis, because low interest rates don’t help that much when borrowers don’t want to borrow and lenders don’t want to lend, but as the central bankers of the 1930s demonstrated, tight monetary policy can be disastrous.

  On the fiscal side, the government should do as much as it can to reduce taxes and increase spending in order to offset the loss of wealth, the tightening of credit, and the collapse in private demand. It’s important to make credible commitments to unwind this emergency stimulus down the road, especially if you go into the crisis with large deficits or unsustainable debt, but an emergency is no time to be preoccupied with immediate spending restraint or a quick path to balanced budgets. The composition of fiscal stimulus is important, too, and should ideally combine speed and power. Tax cuts tend to be fast but weak; infrastructure investments tend to be powerful but slow; aid to low-income families is fast as well as powerful. But in a crisis, you can’t worry too much about optimal design. Politics will ultimately determine what gets done, and you can’t afford to get bogged down in a protracted fight.

  The financial rescue protocol is more complex. You want to conduct a form of triage that distinguishes between the insolvent and the merely illiquid, between firms that will never be viable and firms that just need protection from being caught in the stampede. You have to decide where to set the boundaries of your firewall, your defense against the run. If you set these too narrowly and leave too much undefended, you can get overwhelmed and allow too much damage, but if you set them too broadly, to try to save everyone, you can end up spending trillions of dollars, propping up zombie banks, and laying the groundwork for the next crisis.

  There is no automatic formula for success. The inherent complexity and unpredictability of markets and human behavior require flexibility. Policymakers need authority, they need discretion to use their authority as the situation demands, and they should be accountable for having used that authority effectively and responsibly. In using that discretion, they should consider a few basic guidelines.

  THE FIRST step in crisis management is classic Bagehot. A lender of last resort should make funding available to the banking system, starting with short-term loans backed by safe collateral, going longer and broader if needed. A central bank should lend freely, at a penalty rate, and the “freely” part is important. Ideally, you want to make the loans expensive, so that banks will replace them with private funds when the panic recedes, but you don’t want them so expensive that borrowing becomes a signal of desperation. If the rate is too high, the stigma associated with borrowing will cause banks to sell assets and reduce lending instead, adding to the destructive pressures on the financial system and the economy. In general, crisis managers have to be careful about tying too many strings to emergency assistance, to avoid undermining its effectiveness.

  If the financial system relies heavily on nonbank financing, you might need to backstop those markets and institutions as well, to protect them against runs. We extended Fed lending to investment banks before we brought the largest of them under the constraints of the banking system. We backstopped Fannie and Freddie by placing them into conservatorship. We also supported commercial paper and other highly-rated securities backed by mortgages, auto loans, credit cards, and student loans. If you allow runs outside the traditional bank safety net, they can cause a lot of damage in a hurry.

  These funding facilities can be very powerful, but there is only so much risk that central banks can take. In a severe crisis, governments ultimately will need to take catastrophic risk off the table by guaranteeing a broader range of financial liabilities. They should charge a fee for the guarantees to help reduce reliance on them as conditions improve. They should cover catastrophic risk, not all risk. But there’s no way to break a true panic without guarantees. While they’re often attacked on moral hazard grounds, because they’re seen as indulgence for the weak, they can actually reduce moral hazard, because you can let the weak fall only when you have the power to protect everyone else. And there’s no way to attract private capital to a financial system in crisis without credible guarantees.

  In our crisis, we expanded FDIC deposit insurance and guaranteed business transaction accounts at banks. The Treasury guaranteed money market funds, which were behaving like banks and experienced runs reminiscent of banks before there was deposit insurance. And once Sheila Bair agreed to use the FDIC’s authority to guarantee new bank debt, we were able to remove some of the uncertainty that had hovered over creditors since the Lehman collapse and the WaMu haircuts. Unfortunately, our successors will have to go back to Congress and ask for the restoration of that authority when they need to put out a future financial fire.

  Central bank lending and government guarantees are both essential. But crisis managers should also insist on forcing capital into the financial system quickly, to make sure it can eventually stand on its own and provide credit to the broader economy without support from the government. The financial community will always prefer naked guarantees—the more widely available and generously structured, the better—in part because they don’t want shareholders to be diluted. But while guarantees are a vital short-term palliative, more capital is a long-term cure, preparing the financial system to support an economic recovery and absorb losses in the future. Naked guarantees can preserve the status quo, helping weak firms limp along in a dependent state, but that’s not the goal of crisis management.

  The goal is to stabilize and restructure the financial system so that it can lend again to families and productive businesses. Insolvent firms should be allowed to fail, but not to tear down the entire system in the process. Viable firms sho
uld be protected from the hysteria of the moment, but also required to thicken their capital buffers in order to survive a long and severe downturn. This kind of triage would be simple if policymakers had a perfect way to distinguish merely illiquid banks from insolvent ones during a crisis, along with unlimited resources they could use to provide guarantees to stop runs and inject capital to repair banks. But the real world doesn’t work like that, so triage in a crisis is exquisitely hard. And if a country’s financial system has been allowed to grow very large relative to its economy—as was the case in Ireland, Iceland, Switzerland, the United Kingdom, and other parts of Europe—it dramatically narrows the options available in extremis and makes triage even harder.

  We conducted a lot of triage during our crisis, not all of it intentional. We allowed a lot of failure, and we pushed a lot of doomed firms into the arms of healthier competitors—not just out of deference to the Darwinian imperative, but also because we didn’t have the authority to guarantee liabilities or inject capital into the system’s weaker links. We got that authority after the horrific fall of Lehman, when Congress passed TARP and Sheila Bair agreed to put the FDIC’s guarantee authority behind many bank liabilities. That quelled the panic for a few months. But the TARP capital and the FDIC guarantees were not unlimited. And as the broader economy deteriorated, ratcheting up the expected losses facing financial firms, markets once again lost confidence that we had the capacity to protect bank creditors and repair the financial system.

  We designed the stress test to solve these problems for us, letting the private sector conduct our triage while committing government resources to avoid chaotic failures of systemic firms. By disclosing the losses the firms could face in a horrible downturn and forcing them to raise enough capital to survive that scenario, we let the market decide which ones could survive with private capital and how much government help the others would need. The objective of the stress test was to maximize the likelihood that private investors would recapitalize the system, minimize the eventual burden to the taxpayers, and ensure that the worst-capitalized firms would face the worst dilution. We hoped to limit the government’s long-term involvement in the financial system and accelerate the restructuring necessary for the financial system to support growth.