Updated: Aug 25, 2020
HER editor-in-chief Bonnie Liu conducted an interview with Jeremy C. Stein, Harvard Moise Y. Safra Professor of Economics and chair of Harvard Department of Economics, on the Federal Reserve’s response to the COVID-19 pandemic, in particular financial regulation and the Special Purpose Vehicles. The following features the interview, synthesized with Prof. Stein’s presentations at the Brookings Institution and the Princeton Bendheim Center for Finance.
The COVID-19 pandemic presents a massive shock to the economy and contracts both aggregate supply and aggregate demand, due to severed supply chains, staggering unemployment, and lost output. What do you identify as the primary economic threats, and what are the main policy objectives in response?
The overarching goals of economic policy should be to smooth consumption, allocate losses fairly and efficiently, and contain forces that can amplify the initial economic shock. The high-level policy goals can be split into two types: micro-efficiency and macro amplification. On the micro level, the priority should be preventing destruction of socially valuable business capital, both physical and organizational. The goal is not to shield debt/equity investors from losses: in an idealized world, you could put losses on creditors, but leave assets in their first best use. The process of restructuring the financial side leads to imperfect allocation on real side, because the technology for allocating losses – the bankruptcy system – is imperfect and risks excessive destruction of productive capital, for two main reasons. First, for smaller firms that are hard hit, there is more bias towards indiscriminate liquidation. Second, to the extent that the bankruptcy system works tolerably well normally for larger firms, it has real capacity constraints. As Ben Iverson finds, the less time bankruptcy courts have to handle corporate bankruptcies, the less efficient the outcomes, and the magnitudes are significant: in a recession, bankruptcy caseload increases by 50%, which implies an increase in losses to debtholders of 32%, against a base-rate loss of 16% – essentially doubling the losses. Even if many firms ultimately need to be restructured, there are benefits to “flattening the curve” in the bankruptcy procedure.
On the macro front, the policy objective is to contain the macroeconomic amplifications. There are a host of amplifiers, including financial accelerators like firm, household, and bank balance sheets, fire sales and credit crunches in the credit markets, aggregate demand externalities and Keynesian multiplier effects, and congestion externalities in bankruptcy courts.
Given these high-level economic priorities, what are the overarching principles that should guide the design of the Fed’s emergency facilities, especially the Special Purpose Vehicles? In the spirit of these principles, how can regulator modify existing measures?
In our paper and presentation, we focus on longer-term credit provision as opposed to shorter-term liquidity program. Specially, we examine the bond-buying programs in which the Fed and Treasury buy corporate bonds for the largest investment grade facilities, in the Primary and Secondary Market Corporate Credit Facilities. Government buys bonds directly issued by corporations, for the largest investment grade facilities. The goal is not to completely reimagine the suite of programs from scratch, but to suggest modest tweaks. Given the enormous uncertainty, desirable program features should include: 1) Wide access: not too stringent on ex ante credit quality. 2) Relatively junior claims and ability to defer interest. 3) Staged financing: control exposure via quantity allocated to preserve optionality.
First, the government should preserve optionality with a “venture capitalist of last resort” mindset. The first order concern is uncertainty: don’t think of the government as a “senior lender of last resort” that’s only extending credit to firms that can pay back. Rather, it should make a collection of risky investment that on the flip side may have a lot of social value.
Second, the government should make aid widely available. Bagehot’s “solvency and good collateral” criteria is not very appropriate here. Don’t impose excessively stringent ex ante credit standards when deciding which firms to admit. For instance, open the door a little bit on the Primary Market Corporate Credit Facility, which currently does not admit any non-investment grade firms. Dial that back a little, admit the highest levels of junk (BB and B-rated firms). There will be more risk of loss, but the government can shore up its position in an expected value sense with warrants, as is done in risky finance to compensate for limited upside claim.
Third, they should provide aid with less senior claims, ideally with preferred stock plus warrants rather than senior debt. Preferred stock preserves firm balance sheets and reduces future cashflow problems and mitigate corporate debt overhang that would otherwise impede recovery. Adding warrants helps to shore up the government’s position and to align incentives because firms will no longer participate solely to gain the upside and stick the government with the downside; they’ll have to share the upside under this alternative arrangement.
Fourth, staged finance is a helpful approach. Let’s suppose the Fed has the same budget authority, and we don’t want to put the Fed at meaningful risk, so if the Fed makes riskier loans, by definition it can make fewer loans. This is exactly the trade we’d like to sign up for. We’re willing to forego the “lend without limit” aspect of Bagehot to gain the optionality. In our preferred design, the Fed would cap loan size by firm, which protects taxpayers by limiting the amount firm can borrow, not by excluding bad firms. The strategy would be, instead of giving firms money upfront for year, give them enough to survive a quarter to cover fixed obligations, and work on a quarter-by-quarter basis, not on a year-by-year basis. It’s relatively easy to estimate recurring fixed obligations – rent, interest payments, lease payments – from 2019 tax returns, and the Fed might allow firms to borrow up to 1/4 of this amount each quarter. Then the Fed can dynamically adjust the program as public health conditions change. If the situation improves, the Fed can lend more for longer. If the situation deteriorates, the Fed can realize that they actually need to let firms restructure or go bankrupt. In response, it can gradually reduce the amount firms can borrow in subsequent tranches; this would allow waves of bankruptcies to take place in an orderly fashion to avoid congestion costs in the bankruptcy process.
What’s a better conceptual framework for risk this time as compared with the Global Financial Crisis? The distinction between Lender of Last Resort purchasing investment grade bonds and Venture Capitalist of Last Resort providing credit in more default-prone scenarios seems to boil down to how much risk the Fed is willing to undertake. Along similar lines, should the Fed also be an Investor of Last Resort and buy equities? If the difference between debt/bonds and equity/stocks a matter of repayment risk and amount, can and should the Fed purchase equity ETFs beyond existing corporate bond ETFs?
The classic Lender-of-Last Resort Logic goes back to Walter Bagehot. Bagehot’s dictum states “lend freely to solvent firms, against good collateral, at a penalty rate.” Every word here is important. This essentially means central banks should lend to firms that are illiquid but fundamentally solvent. The underlying theory of LOLR is based on a very particular type of crisis. Think about the standard Diamond-Dybvig bank run model, there’s a good equilibrium and a bad equilibrium; the logic goes “if everyone runs, you should run too to get your money back.” In the no-run equilibrium, the bank is fundamentally solvent, and everyone can ultimately pay back their loans. In the bad equilibrium, central bank lending stems the run, and the economy is fundamentally healthy. This is the classic case where central bank should lend and expect its money back. In hindsight, the LOLR approach is a decent superficial characterization of 2008-2009. The TARP (Troubled Asset Relief Program) funds were almost entirely repaid, and the Fed didn’t lose a nickel. Ex ante, this might not have been a LOLR situation, but ex post, it seems to have been in significant part a liquidity crisis. This is not the case today. No matter what the Fed and Treasury do, the present scenario is more than a liquidity problem; it’s also a solvency problem.
As you remarked, the Fed has been decisive in rolling out its emergency program and aggressive in providing liquidity and credit. Would the broad scope of the programs be seen as overstepping the Fed’s legal purview as outlined in Section 13(3) and 14 of the Federal Reserve Act and thereby jeopardize central bank independence? Would “fiscal” quantitative easing rather than purely monetary QE blur the line between Treasury and Fed?
By design, the Special Purpose Vehicles involve both the Fed and the Treasury in partnership. The Treasury, as the fiscal authority, acts as the junior partner, as it has congressional authorization and $450 billion of congressionally approved fiscal support. The Fed, as the monetary authority, acts as the more senior lender. The idea is to expose the Treasury and insulate the Fed: The Treasury holds the equity tranche and absorbs the first losses, while the Fed holds more senior claims and benefits from the liquidity cushion.
The Special Purpose Vehicles work as follows: a shell company is created; the treasury puts in equity, and the Fed lends to the SPV. So even if that company buys risky bonds, or in principle even equity, as long as the company has enough backing from the Treasury, the Fed’s loans can be secured. Granted, the Fed losing a lot of money adds a fiscal flavor and threatens its independence, but can it lend to a vehicle that has been set up with the express intention that some of the money may get lost? I think the answer is yes as long as it’s adequately backed by the Treasury.
Do you identify any potential shortcomings or loopholes in the Fed’s current programs, and if so, what solutions would you propose to address them?
The Main Street programs have been open for business, but it’s done almost no business. It’s a little hard to disentangle what exactly the reasons are, and we’re hard-pressed to parse through the several possible alternatives. The banks are unwilling to participate, but I don’t know if it’s the simple economics or something broader. The Fed instituted a similar program called public private investment partnerships in 2009, and there was a pervasive stigma associated with being embedded within government. Do I as a bank want to partner with the government? Maybe they thought the optics were bad and they might be accused of taking advantage.
There are several concerns with the Main Street Program design. First, the bank risk retention programs require banks to act as gatekeepers and take 5-15% of the loan. This is designed to ensure government is making loans on “commercial” terms seen as positive-NPV by banks. The banks’ incentive alignment is contaminated by pre-existing positions. Banks are more likely to participate when the new loan bails out an existing troubled position, which results in allocative distortion. Even if a loan has a socially positive net present value, the banks probably won’t be inclined to participate unless they have an existing stake in the company. Given all the externalities at play, this is not the right social criterion.
A second concern is the accelerated and aggressive repayment schedule. In the new loan program 1/3 of principal has to be paid back in each of years 2-4. This is likely to create cashflow problems for borrowing firms with a high operating leverage, because a year of lost revenue is many years of profits.
With the Fed currently in the “do whatever it takes” phase as described by Powell, short-term solutions to ensure survival and prevent damage appear tangible and feasible, while long-term policies aimed at restoring the economy back to pre-crisis equilibrium seem more nebulous. In your opinion, what is an appropriate exit strategy for the government when the economy prepares to reemerge following the discovery of a vaccine?
There is an exit strategy issue with the Main street program, which lends directly to medium-sized companies. The hardness of debt claims presents a particular worry. The combination of fast repayment plus senior claims is likely to put many Main Street borrowers in distress when economy is still fragile. The natural next question is: Who will manage the workout if loans go bad? The Fed hopes that the Treasury as the residual claimant will assume this role. The Treasury may want to act leniently by recontracting, deferring interest, reducing principal, but given participation of banks, it’s hard for the Treasury to be soft. Senior bank lenders with strong collateral positions may be keen about liquidating companies that default, relative to a socially optimal resolution. Meanwhile, the Treasury doesn’t want to make politically fraught decisions of going company-by-company and deciding which firms live and which firms die, so they may hire a third-party agent to perform the workout. If the workout is delegated to private entities on private-market terms, there is reduced scope for ex-post socially desirable solutions. Hardness of claims, combined with partnership with banks, and political imperative to avoid overly-interventionist ex post actions can be problematic.
With regard to the workout, is there an alternative solution that more efficaciously preserves social welfare than the third party contracting you just described?
We want to avoid a scenario whereby when the economy reemerges, companies are too burdened with debt to be economically viable. The venture capital analogy is helpful here, and the question is “what is the right way to provide financing to firms in an environment of high uncertainty?” I argue for financing with preferred stock over hard debt. With preferred claims, interest payments can be deferred without forcing default. If companies can’t pay dividends on preferred stock, they get deferred without risking bankruptcy. This is what the government did when investing in the banks in 2008-09: the government needed to fund the banks, but it didn’t want to give debt because banks were overly indebted – that was the whole problem. Ideally, the government would give banks equity, but it didn’t want to risk the specter of government nationalizing and controlling the banks and create the impression of “owning the banks.” As a compromise, preferred stock is somewhere in between – less junior than equity, and more junior than debt. This would be a reasonable compromise for the current crisis as well. It’s very soft and friendly: “You’re supposed to pay, and you can pay. But if you don’t pay, then you can pay tomorrow.” The more junior status helps companies attract future rounds of financing and lessens debt overhang. If the Fed is concerned about junior position, it can strengthen its overall claim by adding warrants to align banks’ incentives.
Following the Great Recession, there has been abundant research illuminating the Fed’s conventional and unconventional tools, via interest rate and balance sheet policy, etc. How do recent events shed new light on previous research about monetary policy at the zero lower bound? What are the implications of the Fed’s large balance sheet due to large scale asset purchases? Does this corroborate your views in Greenwood, Hanson, and Stein (2016)?
The Fed had been shrinking their balance sheet for a while post 2008, and they reached a point where further shrinkage was creating tension. Our 2016 paper argued for the Fed to keep a large balance sheet to provide supply of short-term liquid assets and maintain financial stability. Now the Fed will have a large balance sheet purely as a byproduct of the crisis, those issues have been subsumed.
The only worry with the humongous balance sheet concerns the leverage ratio for banks, a non-risk-based capital requirement that obligates banks to maintain equity equal to some fraction of all assets, independent of how risky they are. Thus, you have to hold equity even against safe assets like treasuries and reserves. Normal time regulation and the Fed’s emergency response are at cross purposes: The fed is forcing banks to hold trillions of reserves, while regulation is penalizing them by mandating them to put aside costly equity capital for reserves held. In response, the Fed instituted a temporary fix by tweaking the rule to exempt treasuries and reserves from the regulatory requirement. In the longer run, the Fed should continue to exclude reserves at the least – maybe not Treasuries – from the requirement, which would allow for a system of abundant liquid assets.
The pandemic’s economic repercussions have reignited debate about unconventional monetary policy tools. Former chairmen Ben Bernanke and Janet Yellen and current governor Lael Brainard proposed yield curve control to directly target long-term rates by committing the Fed to buy or sell Treasury securities at the predetermined price. Similarly, negative rates allow the Fed to push short-term policy rates below zero. Do you think these are worth experimenting with?
Yield curve control and negative interest rates are interesting new gadgets and fun to talk about, but I don’t think they are the first order worry right now. Yield curve control could be used to lower interest rates 2-5 years out a little, but short term and long-term rates are already well controlled with forward guidance and verbal commitment to keep rates lower for longer.
Instead of safe treasury rates, the Fed should be thinking about rates on risky debt, which fluctuated dramatically since March. The real issue is not the safe rate of interest, but the magnitude of the credit spread added to the safe rate. The Fed should prioritize thinking about “what is our reaction function if high yield bonds start to default?” and it is rightly focusing on the Main Street programs, the Primary and Secondary Corporate Credit Facilities.
When the market wobbled from the enormous hit, the Fed’s announcement in March caused an unbelievably powerful rally in the bond market. On Bloomberg today, a company set the all-time record for the lowest junk bond rate ever paid. Here we are, in the midst of a historical recession, yet firms are borrowing at the cheapest levels even for junk bonds, solely because of some words the Fed said. The market pricing is hard to understand if you don’t believe the Fed is going to act extremely aggressively.
What will happen if things get worse? The market seems to believe that the Fed has their back, but the Fed cannot buy defaulted bonds according to its guidelines. It’s as if the market has heard Powell say “Don’t worry about a thing, we got you,” when in fact the Fed has said “We’ll buy some investment grade bonds.” On the one hand, the Fed’s LSAP announcement is an unbelievably successful policy. It’s magic: markets rallied; there’s an enormous amount of issuance; companies are borrowing a lot, which strengthens their balance sheet with reserves. However, there is a potential flip side: by stimulating the market that much, I worry that the Fed run the risk of later disappointing the market. It’s a very interesting market dynamic, though it feels to me a little overdone.
In your recent Brookings paper with Blank, Hanson, and Sunderam, you propose a two-pronged conceptual framework for bank capital regulation, including the idea of “dynamic resilience.” Could you elaborate on it?
We are addressing the question “How should regulators respond to the pandemic?” To give you the short answer up front, our view is “more forcefully than they have done so far.” To put this in context, the Fed’s overall response has been generally outstanding; they’ve been appropriately aggressive, imaginative, willing to push the envelope, except strikingly in bank regulations, where it feels like they’ve been somewhat passive.
In our paper, we argue that bank regulation should serve two logically distinct goals. First, regulators should use risk-based capital ratio requirements to reduce the probability of bank failures in normal times. Since banks do not fully internalize the social externalities of bank failure, they will under-finance themselves with equity and take excessive risk in absence of regulation. The second aim is to ensure an early recapitalization of banks, in the spirit of what we call “dynamic resilience” as opposed to the current strategy of “watchful waiting.” The Fed should ensure that banks remain well-capitalized so they can continue to provide credit to the economy. This framework implies that regulators should simultaneously relax marginal bank capital ratio requirements and encourage banks to raise new dollars of equity.
With the goal of encouraging bank lending in mind, there are two ways to achieve this. If you think about bank lending in terms of a fraction – the numerator being loans and the denominator being equity – the arithmetic is straightforward. To expand the amount of loans, you can either have a lower capital requirement – which means allowing more dollars of loans per dollar of equity -- or you can have more equity. On the one hand, you can lower the marginal capital requirement, that means mechanically for a dollar of equity, the banks could lend more. In an ideal world, if you have a very high starting capital requirement, you’d cut it to stimulate lending, while not jeopardizing things because it’d still be pretty high. If you don’t have room to do that, the alternative is more dollars of equity. This strategy is analogous to optimal taxation, which combines a reduction in marginal tax rate (decreasing marginal capital requirements) with an expansion in lump-sum taxes (encouraging new equity capital). Regulators have actively relaxed existing capital requirements, but have inadequately emphasized conserving or raising new equity capital. Most of the official regulatory actions fall into the first category; only the announcement that the largest banks would temporarily suspend share repurchases falls into the second category. Bank regulators in many countries have curtailed or suspended bank payouts to common shareholders, and the Bank of England instructed banks to halt cash bonus payments to senior staff since March.
Looking beyond the current crisis, what long-term recommendations do you have for the Fed and other regulators post-recovery?
Instituting a countercyclical capital buffer could cushion banks against future shocks. The capital requirement must start off high, because following an adverse shock, even if the regulators are willing to cut the ratio, the market may be unwilling to lend. Regulatory requirements may not be the binding constraint for banks, as credit market investors may be more stringent and refuse to extend short-term wholesale funding to banks that are not well capitalized, as was the case in 2008-09. In good times, the countercyclical capital buffer should always be turned on by default, so that policymakers have room to cut this element of the overall capital requirement in a downturn. Other countries have countercyclical capital buffers that mandate bank to hold more capital with the expressed idea of bringing it back down in bad times. The U.S. never took the step of raising it in the first place, so it has nothing to lower. This leaves bank regulators with less ability to encourage lending in the current crisis. Given this policy error, the only possible adjustment is to raise equity capital.
Moreover, it is difficult to influence the lending to capital ratio in practice. Even if regulators relax it temporarily, the market players look at banks and say, “yeah but you’ll have to bring it back up eventually.” The banks know that the present requirement will eventually revert back to its equilibrium value, so their lending incentive may not change.
In your paper, you additionally recommended that the Fed more directly incorporate bank stock prices information into stress tests, because forward-looking market prices provide more timely information than backward-looking accounting metrics. What would this entail?
The stress tests, as they are currently done, don’t make references to market values. They ask about accounting values, or the book value of equity: “Will your capital be high enough as a buffer in adverse scenarios?” Accounting values work tolerably well in normal times, but when things change rapidly, they are slow to reflect the evolving circumstances. Both stock prices and stress tests suggest meaningful risk of large losses on bank assets, but the banks haven’t recognized them yet. In contrast, the stock market is forward looking and contains an element of judgement: we saw bank stocks fall by 40% more than the rest of the market. Bank stocks were down 40% from their January peak as of May 25, compared to the overall stock market, which is down just 13% from its peak. Beyond an overall forecast, the market can clearly identify which banks are more vulnerable: During the Global Financial Crisis, banks with the biggest stock market losses were the ones that later recognized the largest loan losses. This time around, stocks of consumer-focused banks seems to be hardest hit, which is reasonable as these are the banks that engage in the most consumer lending and are therefore most affected by high unemployment. Clearly, the market has an early warning property.
On the other hand, the market is of course very noisy and prone to overreact and mis-react, so it’s undesirable to use market information as purely mechanical, judgement-free input and build regulatory framework that says “you have to have this ratio of market value to total lending.”
There ought to be some reconciliation in the stress tests. It should explain “how can we have relatively optimistic outcomes when the market disagrees?” The market isn’t always right, but it’s objective and less subject to lobbying and accounting manipulation. There shouldn’t be a simple mechanical relationship between, say, bank stock prices and recapitalization requirements, but the market has useful information. Regulators have to use it cautiously.