Tuesday, April 19, 2016

Chari and Kehoe on Bailouts

V. V. Chari and Pat Kehoe have a very nice article on bank reform, "A Proposal to Eliminate the Distortions Caused by Bailouts," backed up by a serious academic paper.

Their bottom line proposal is a limit on debt to equity ratios, rising with size. This is, I think, a close cousin to my view that a Pigouvian tax on debt could substitute for much of our regulation.

Banks pose a classic moral hazard problem. In a financial crisis, governments are tempted to bail out bank creditors. Knowing they will do so, bankers take too much risk and people lend to too risky banks. The riskier the bank, the stronger the governments' temptation to bail it out ex-post.

Chari and Pat write with a beautifully disciplined economic perspective: Don't argue about transfers, as rhetorically and politically effective as that might be, but identify the distortion and the resulting inefficiency. Who cares about bailouts? Well, taxpayers obviously. But economists shouldn't worry primarily about this as a transfer. The economic problem is the distortion that higher tax rates impose on the economy. Second, there is a subsidy distortion that bailed out firms and creditors expand at the expense of other, more profitable activities. Third there is a debt and size distortion. Since debt is bailed out but not equity, we get more debt, and the banks who can get bailouts become inefficiently large.
For sake of argument, I think, Chari and Pat take a benign view of orderly resolution and living wills. Their point is that even this is not enough. Though functioning resolution would solve the tax distortion and subsidy distortion, the debt-size externality remains.
The extent of regulator intervention depends on the aggregate losses due to threatened bankruptcies. Individual firms do not internalize the effect of their decisions on aggregate outcomes and, therefore, on the extent of such intervention. Just as with bailouts, individual firms have incentives to become too large relative to the sustainably efficient outcome 
Their alternative: A regulatory system that
limits the debt-equity ratio of financial firms and imposes a Pigouvian tax on the size of these firms.
The paper is not specific beyond this suggestion. It's intriguing for many reasons outside the paper.

First, they limit the ratio of debt to equity, not the ratio of debt to assets. Current bank regulation is centered on the ratio of debt to assets, but then we get in to the mess of measuring risk-weighted assets, many of them at book value.  Abandoning this whole mess is a great idea.

Thinking about some of the same issues, I came to the conclusion that a simple Pigouvian tax on debt would work better than current debt-to-asset regulations. If you borrow $1 (especially short) you pay an 5 cent tax per year.

There is an interesting question then whether this tax on debt or a regulatory debt-to-equity ratio limit will work better.

Chari and Pat don't say what the optimal debt/equity ratio should be, and how that should be enforced dynamically. If up against the limit, do they want banks to sell assets ("Fire sales" and "liquidity spirals" banks will complain), to issue equity ("agency costs", banks will complain) or what?  Chari and Pat also don't say whether they want regulators to target the ratio of debt to book value of equity or to market value of equity. I like market value, further avoiding accounting shenanigans. I suspect the regulatory community will choose book value, so inure themselves from responding to market signals.

I like announcing a price rather than a quantity -- a Pigouvian tax on debt rather than a debt-equity ratio -- as it avoids the whole argument, and the just this side vs. just that side of any cliff.   My tax could rise with size, to address their size externality as well.

But they don't analyze the idea of a tax on debt rather than their ratio, so perhaps both would work as well within their model. Their ratio of debt to equity is sufficient for their ends, but perhaps not necessary.

Chari and Pat take a benign view of debt, and the functioning of resolution authority: They
start from the perspective that because debt contracts are widespread, they must be privately valuable and, in all likelihood, also valuable to society in general.
They also posit that "orderly resolution" authority will in fact swiftly impose losses on creditors, and that by using "living wills" the offending banks can be quickly broken up.

I think they make these assumptions to focus on one issue. That's good for an academic paper. But in contemplating a larger regulatory scheme, I think we should question both assumptions.

In a modern economy, liquidity need not require fixed value, and I think we could get by with a lot less debt.  That leads me to much more capital overall. They implicitly head this way,  presuming that debt is vital, but then advocating debt equity ratio regulations that will presumably mean a lot more equity.

I suspect that resolution authorities, hearing screaming on the phone from large financial institution creditors of a troubled bank,  and with "systemic" and "contagion" in mind, will swiftly bail out creditors once again.  I think that a bank too complex to go through bankruptcy, even a reformed bankruptcy code, is hopeless for the poor Treasury secretary to carve up in a weekend. So another reason for more equity is to avoid this system that will not work, as well as to patch up its remaining limitations even if it works perfectly.

Chari and Pat also step outside the model, stating that the resolution authority
is worrisome because by giving extraordinary powers to regulators, it allows them to rewrite private contracts between borrowers and creditors...[this]... can do great harm to the well-being of their citizens. Societies prosper when citizens are confident that contracts they enter will be enforced
Their closing sentence is important
We emphasize that regulation is needed in our framework not because markets on their own lead to inefficient outcomes, but because well-meaning governments that lack commitment introduce distortions and externalities that need to be corrected.


  1. The TBTF lending subsidy channel sounds intuitive, but subsidy estimates from the GAO and the IMF find that the subsidies appear during a crisis. There doesn't seem to be an obvious distortionary effect prior to 2008: http://rooseveltinstitute.org/two-opposing-methods-tell-us-too-big-fail-subsidy-has-collapsed/

    Maybe such a Pigouvian tax should make an allowance for core deposits to maintain some of unique information management benefits. An unspecified bank debt tax would also inhibit service benefits of deposits.

    1. Let me say, I don't like a Pigouvian tax on debt for the simple reason that you are asking the federal government to conduct monetary policy.

      What is the difference between the central bank lending at 1% and the federal government charging a 5% (5 cents on the dollar) tax or the central bank lending at 6% and the federal government charging zero tax?

      If you think that the short term interest rate is too low - then say so. Don't try to hide your disdain for monetary policy stance - speak up.

  2. Here is what I worry about. Very short term bank debt (like commercial paper and CDs) functions to a degree as "money." If you crack down on bank debt, are you reducing the amount of money in the economy, with harmful consequences for the macro economy.

    Maybe this concern is wrong, but at least I would like to hear the issue discussed.

    Once again, finance economists and monetary economists are not talking to each other.

    1. I agree that if we “crack down on bank debt” that has a deflationary effect, ALL ELSE EQUAL. However, there’s no need for all else to be equal: in particular, the state can perfectly well print and spend whatever amount of new money is needed to keep the economy operating at full employment / capacity.

      Moreover, state issued money is cheaper to produce than private bank issued money: reason is that private banks have to check up on the credit-worthiness of those they lend money to, and that involves significant costs. In contrast, the state just presses buttons on a keyboard and creates new money and spends it (and/or cuts taxes). No need to check up on anyone’s credit-worthiness.

      Also, privately issued money involves horrendous risks: 2007/8 style bank collapses followed by eight year long recessions. In contrast, where lenders are funded by equity rather than debt, and they make silly loans, all that happens is that the value of those shares / equity falls in value. Insolvencies do not occur.

  3. Thanks for reading our paper and summarizing it so well John! It is much appreciated.

    1. As you say, a Pigouvian tax would be just as effective in our model. I much prefer such a tax and we should have made that clearer in the Policy piece. (We do discuss the equivalence in the academic paper.)

    2. The paragraph you cite about Orderly Resolution was meant to convey our deep unease with intrusive discretionary schemes like orderly resolution. At what point does the United States come to resemble Zimbabwe?

    3. I know you have beaten this drum a lot, and I have also raised it often: When it came to Acid Rain, economists to a man and woman said that command and control schemes like the EPA's were a terrible idea. We all screamed for a tax or for tradable permits. When such permits were set up by Congress, Acid Rain stopped being a problem. This event was, I thought, a watershed for economists' ideas. Why are so many economist now fans of Rube Goldberg regulatory schemes which dictate every kind of activity in which banks can engage? Clever economists can now spend all their time developing "systemic warning signals". I suppose it is a full employment act for economists, but I hoped for more from our profession. Why is everyone so fond of command and control in this sphere?

    Thanks again.


    1. How do Rube Goldberg regulatory schemes come about? Perhaps it is the accumulation of small sensible changes by economists who want to appear reasonable. This blog does contemplate more fundamental reforms now and then, like the narrow banking proposal, and I admire the attempt to swing for the fences.

      With respect to this specific proposal, we should also discuss tax incidence. Whether it is by imposing limits on financial ratios, or by a Pigouvian tax, either way we seem to end up with downward pressure on deposit rates. The problem of negative rates has been discussed elsewhere on this blog, and radical changes in architecture probably hold more promise than the inevitable whack-a-mole of accumulating piecemeal changes.

  4. John,

    "I like announcing a price rather than a quantity -- a Pigouvian tax on debt rather than a debt-equity ratio -- as it avoids the whole argument, and the just this side vs. just that side of any cliff. My tax could rise with size, to address their size externality as well."

    I don't, here is why. A tax on debt would be limited to private debt. It's not like the federal government is going to tax itself when it borrows. And so what you end up with is a large wedge between the cost of private versus public borrowing. This wedge already exists in the form of credit spread (AAA, AA, A, B, etc.). Your proposal exacerbates this spread which from a productivity standpoint ( Real GDP / Total Debt ) is the exact opposite of what reform should look like.

    Your inclination seems to be that borrowers would be paying an insurance premium to the federal government so that the federal government can cover potential losses in the future.

    One problem with this is that the federal government (aside from Social Security) does not act like a traditional insurer - they don't buy market securities that can be cashed in when problems arise and they don't stuff premiums into some large vault. Any tax revenue that the government collects is spent (superfluously sometimes).

    Another problem is that by collecting this insurance premium you are conceding that the government should bail out to begin with.

    Perhaps you can illuminate what government should be doing with the revenue it collects from a Pigouvian tax, and then maybe it will make a little more sense.

  5. Speaking only as a plebe, and a non-economist, although sometime regulatory lawyer, I want debt contracts. I think, checking accounts, savings accounts, and CDs are wonderful things. I think we need to encourage more people to save more. I am not sure if you economists understand how scared most people are of equity investments. If ordinary people have no ability to make fixed investments with banks, they won't buy equities, they will put money under their mattresses. That would not be a good thing.

    I agree that if the banks are going to issue debt, they need to be more heavily capitalized. But, I think the idea of a tax on debt issuance is a political non starter. The banks would claim that the tax would decrease the amount of interest that could be paid to depositors and require heavier fees on accounts and higher rates from borrowers. All of that would be very unpopular. (5 cents per dollar? What are you smoking? How can you make 5 cents on the dollar legally?).

    I think we are back to debt ratios. I would like to see BHC debt asset ratios on the order of $1 on 8 dollars of assets.

    Further, to induce ultra large institutions to right size themselves, I would add an extra capital requirement to institutions larger than $195 Billion in assets (the top 16). http://www.ffiec.gov/nicpubweb/nicweb/HCSGreaterThan10B.aspx

    The requirement would be extra percentage points equal to:

    The natural logarithm of its total assets – 26.

    So JP Morgan Chase, which has ~$2.35 Trillion in assets would have an additional capital requirement of 28.5 – 26 = 2.5% or about 15% over all. They might decide to reconfigure the business a bit.

    Further the 30 largest banks would be required to have an equal amount of their funding derive from long term subordinated debentures. Those instruments (~$350 billion for JPM-Chase) would attract a CDS market that would be a canary in the coal mine for them.

    I also think that the idea of "risk free assets" should be abandoned. It is the central conceit of the Basel "risk based" capital requirements. The one thing we have learned in the last 8 years is that all assets are risky, perhaps in inverse proportion to to their appearance of riskiness. Mortgages were supposed to be safe. Government bonds were supposed to be safe. Ha. Ha. Ha.

    Second. And libertarians should love this one. More competition. The banks best reason for accumulating assets is that building a branch network is the cheapest and least volatile form of funding. This makes tremendous size an advantage. Another business that has locations easily accessible all over the country, and also handles a lot of cash is groceries. Allow Wal*Mart, Target, Kroger, Safeway, et. al. to go into the banking business, at least to the extent of allowing them to open branches in their stores, and allowing them to provide ordinary banking services to consumers and small businesses.

    A few years ago, Wal*Mart sought to start an industrial bank ( a species of small consumer bank). A coalition of labor unions, “community activists”, and bankers lobbied furiously to stop Wal*Mart. Any idea those miscreants hate so much must be very good.

    Allowing tough competition in the consumer banking space would remove the monopoly rent possibilities and limit the incentive for existing banks to expand by acquisition.

    1. Truth. There is no such thing as a risk-free asset. I have been around US banking since the latter 1970's. There are many risks in addition to credit or default risk. In 1983, during Volcker's fight against inflation, when the prime rate soared into the teens, a US Treasury bond could have a market value of 66. And, a bank holding such could be paying say 8% to 10% on the its funding, resulting in negative net interest margins.

      I watched the movements of the Basel III capital requirement from proposals to final standards. One of my big disappointments was that they maintained the 50% risk-weight for all residential mortgage loans, when the proposal had the capital charge as high as 150% or 200% for the risky mortgage loan types that caused the subprime crisis and the Great Recession.

    2. Fat Man,

      The whole problem with mandating a debt / equity ratio ( or debt / assets ratio ) is in valuation. Suppose a bank has outstanding debt with a market value of $5 million and outstanding shares with a market value of $5 million (Debt / Equity Ratio = 1 on $10 million of capital). The government decides that the bank's debt / equity ratio should be 0.75. So, the bank sells $715,000 of equity and buys back the same amount of debt.

      Total debt is now $4.285 million market value and equity is now $5.715 million market value (Debt / Equity ratio = . Except that the market for the bank's debt rallies and the market for the bank's equity retreats pushing the market values back to $5 million for equity and $5 million for debt.

      That is why John likes using market values. He realizes that setting a leverage ratio based upon market values is impossible. To actually have a binding debt / equity ratio, you need to value the debt at par. This just makes sense.

      When equity rallies, the liability to the issuing company increases. When debt rallies, the liability to the issuing company does not increase. Meaning that a company that wants to buy back shares pays market price for those shares. When a company wants to retire debt, it pays back principle and accrued interest (and possibly a prepayment penalty).

    3. That is why the ratios are traditionally set based on book values. If the market price of the common stock is above book value, the bank can sell more common stock to increase its capital. If market is under book, then the bank must either sell assets, or use some mechanism like a dilutive rights offering. Those are popular in Europe, but almost unknown in the US.

    4. Fat Man,

      Presumably book values are established by looking at the going market price for similar assets. So what happens when the market price of similar assets fall?

      A bank trying to hit either a debt / equity ratio or a debt / book value ratio runs into the same type of valuation problem. The only way to do it is to value the debt at par.

    5. Book value is an accounting concept. It is the actual cost of acquiring the asset or the face amount of a liability. Book values are not impacted by market events.

  6. I still don't understand why a publicly held bank in which shareholders have a stake and also a board of directors, is thought to not face moral hazard. Shareholders get wiped out in a bank collapse, no?

    An interesting option would be no regulations at all on banks except that they must have 20% equity against loans and another 20% first in line of convertible bonds.

    After that equity and convertible debt is wiped out in the bank collapse, I have no problem with the Federal Reserve printing money and recapping the bank and getting it up and running ASAP.


Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.