Friday, August 29, 2014

After Dodd-Frank

(Youtube link) A talk given at the Mercatus Center / CATO conference "After Dodd-Frank: The Future of Financial Markets." (The link has videos of the whole conference.) The talk is taken from the paper "Towards a Run-Free Financial System," which answers many objections you may have to claims in the talk. (Yes, I have plugged it before on the blog and will likely do again.)

The more I read about it, the more I think it's important to define what is not a problem, and can be left alone. If we have to solve housing subsidies, Fannie and Freddie, global imbalances, Wall Street greed, compensation, inequality, savings gluts, predatory lending, financial utilities, bankruptcy law, behavioral biases of equity managers, living wills, stress tests, capital ratios, Basel regulation, macroprudential bubble-detection and pricking, complexity of derivatives, exchange vs. otc trading, and so on and so on just to save ourselves from the next crisis, we might as well give up now.


  1. My biggest objection is to the claim that $18 trillion is enough to meet "any conceivable" demand for money. Why would we need a hierarchy of money to ensure elastic supply, if this were true? Runs affect institutions that bridge between two levels of the hierarchy, such as banks that trade base money for deposits, or central banks that try to fix an exchange rate. We could indeed avoid runs if we could flatten this hierarchy, and dollarize the whole world. But that would require so many dollars that the value of the dollar would not be stable.

    If we want to prevent runs on a system that deals in run-prone liabilities, we need to develop insurance that stabilizes the balance sheets of institutions that produce various kinds of money. Isolated runs on individual banks and other types of dealers should not lead to a crisis throughout the entire system.

    1. Insurance for small banks is OK: hence FDIC. But with large ones, the “insurance company” would need an absolutely huge stock of assets to sell given a crisis like the recent one. Selling those assets would cause the price of the relevant assets to collapse, I’d guess. So in effect, there is only one entity that can stand behind large banks with runnable liabilities, and that’s the state.

      So we could have the state act as insurance company with large banks paying an annual insurance premium. But what’s the point? All that does is to enable commercial banks to issue runnable liabilities, i.e. cash or money, and the state can perfectly well issue any amount of money anytime. Indeed, the state ALREADY HAS TO issue varying amounts of money (e.g. via QE) to make up for the erratic money / credit creation activities of commercial banks.

      So I vote for just forbidding commercial banks from issuing any form of money or runnable liability, as advocated by Milton Friedman, Merton Miller and two or three other Nobel laureates.

    2. Hans Gersbach has written about "Banking with Contingent Contracts" where the contract between the bank and depositors is such that depositors insure the bank against macroeconomic risk. He shows that the "risk allocation is efficient, provided that banks are not bailed out". If we all agree that Dodd-Frank is a dead end, then contractual innovation should be considered as a possible way forward. If we insist on using the types of contracts we have today, we are limited to options like the Chicago plan.

    3. Anwer,

      The paper you are referring to can be found here:;h=repec:eth:wpswif:08-93

  2. Bingo! The Dodd Frank act is a mess because it does not define what is a systemic risky institution. It looks for something but it doesn't know exactly what is it that it is looking for.

    "Definitions are the guardians of rationality, the first line of defense against the chaos of mental disintegration" (Ayn Rand)

    1. Worse, it does not define what is NOT a systemically risky institution. If the FSOC so "determines," car washes can be designated as systemically important. There is no safe harbor.

  3. "we might as well give up now"

    In the long run we're all dead. It is still worthwhile looking both ways before we cross the street.

    We can't solve all problems. We cannot eliminate all risk. All we can do is seek some economic optimum in how we regulate banks. We know that no regulation is not the optimum and that there can be economic gains from well designed regulation.

    For me the priorities would be: accurate disclosure by banks of their total liabilities; economically efficient capital rations (probably somewhere between 5% and 20%); and a lot more transparency and regulation in the derivatives markets. The fact that the derivative markets dwarf the real economy ten to one tells us there are huge, world threatening, risks lurking in that market and may suggest that there is something profoundly amiss.

  4. The Volker rule and other penumbras around Dodd-Frank are making the big banks get out of proprietary trading. In the next recession, there will be a lot fewer agents willing to 'provide liquidity' via some value-trading, and assets will drop as never before. Add to that the post-hoc penalties applied to banks for firms that were acquired during the crisis, so no big institution will take over another's obligations merely because its equity price is low--the legal risk is too great.

    So, we haven't seen anything yet.

  5. two concerns:

    1. imagine i want to start a business in maturity transformations. and i put up a big neon board outside my door which says, "NON FDIC INSURED". would you feel that the government has any right to stop me and my depositors from conducting our business even though i might be issuing demand liabilities.

    2. (i am less sure about this) but are you concerned that narrow banking might be a form of financial repression making it easier/cheaper for the government to raise capital and ultimately increasing the size of government.

  6. John,

    Please address the fungibility of money issue -

    I sell $5 million in short term debt and I sell $5 million in equity. I make $5 million in long term loans and I buy a new building for $5 million. Have I borrowed short to lend long?

    1. That's why I like Pigouvian taxes better than capital ratios. In your case, you issued $5 million short term debt, so send the US Treasury (say) $100k. No arguing about capital ratios, stress tests, systemic importance, etc.

    2. Okay,

      So I send the US Treasury $100k. How is this an improvement over what we have now? I just roll that cost into the interest hat I charge on long term loans.

      You haven't answered my question - with money being fungible and with a variety of financing means and investment objectives, how do you identify which money is being directed to what objective?

    3. John,

      So I send the U. S. Treasury $100k. And I can't pass that cost onto my long term borrowers?

      It is my belief that Pigouvian taxes would be ineffective because they would increase long term borrowing costs as much as they increase short term borrowing costs.

    4. Incentives. Margins. If you're paying a 2% tax on debt you'll think really hard about just how bad is it going to be to fund the project with equity. Economics is about margins, not about levels.

    5. The Pigouvian tax seems like an insurance premium that covers government assistance in bad states of the economy. Hans Gersbach would have depositors bear that risk, instead of taxpayers. What I like about the Gersbach approach is that the wholesale shift out of risk-free deposits toward contingent contracts would raise the equilibrium rate of interest, and people are looking for ways to do that.

    6. John,

      If I am paying a 2% tax on short term debt I pass that tax onto my long term borrowers. How does a Pigouvian tax preclude me from borrowing short and lending long? How does it stop a Lehman Brothers / Bear Stearns from leveraging up 30 to 1?

      So what if Lehman would have "thought really hard" about the right debt / equity mix with a Pigouvian tax. The ownership of Lehman may have still have concluded that cashing in stock options before the company imploded was more important than the survival of the company.

      Greenspan made mention of this in his mia culpa / I was wrong. He was under the 1950-ish "company man" assumption that the ownership of a firm would always do the logical thing to ensure the survival of the firm. Times have changed.

      Golden parachutes and the revolving door between Wall Street and government make running a company into the ground a sound business strategy.

    7. Anwer,

      "The Pigouvian tax seems like an insurance premium that covers government assistance in bad states of the economy."

      I thought the whole point of the Pigouvian tax was to dissuade people from engaging in an activity (in this case - short term borrowing). A successful Pigouvian tax would collect no tax revenue.

      If instead (as you say) it is an insurance premium, then am I (the short term borrower) given government protection (insurance) in case I become over leveraged? With a Pigouvian tax as insurance policy paid for by the short term borrower, who is the insured party?

      The government sells me home fire insurance when I have the tendency to construct camp fires around my house?

    8. Frank, I realized after posting my comment that I had forgotten the aim of Pigouvian taxes. We want people to engage in activities at efficient levels, and externalities do cause distortions. The current equilibrium with razor-thin equity levels at banks is clearly a corner case, and Cochrane is explaining it as Anat Admati does: low capital ratios are an inefficient result of the externality that leads to implicit promises of bank bailouts.

      We get bailouts of banks in bad states of the economy, which is why I'm describing them as macroeconomic insurance, though the bailouts don't cover only realized macroeconomic risk but also bad bets made due to moral hazard.

      Cochrane's proposal internalizes the costs of that insurance/bailout with a price of 2% for coverage of run-prone funding. Because equilibrium interest rates are so low (perhaps negative), we might see little practical change in outcomes - and remain in this corner case - except that bailouts are funded in advance, and not by taxpayers. Or if that 2% is sufficiently high then we might see a higher reliance on equity funding, less moral hazard, and fewer crises - within any country where this rule was applied. Perhaps we will also have fewer global financial crises.

      Gersbach doesn't dwell on externalities, but his proposal does stipulate that the government should be committed to a policy of no bailouts, which suggests that he has eliminated the same problem that Cochrane addresses with his own plan. Indeed Gersbach demonstrates, with a formal model, that his proposal leads to an efficient distribution of risk. I think it is also more likely to raise equilibrium interest rates than the 2% tax proposed here, which would make it more helpful to the global financial system.

    9. Anwer,

      "Cochrane's proposal internalizes the costs of that insurance/bailout with a price of 2% for coverage of run-prone funding."

      And I would interject that a bank would simply pass that cost along. What Cochrane does not demonstrate is how a Pigouvian tax would shrink the term premium between short term borrowing by a bank and long term lending by a bank. The only way to do that (with central bank acting as lender of last resort) is for the central bank to deliberately flatten the yield curve. Meaning whatever long rate (30 year in the case of the U. S.) the market demands, the central bank lends overnight at that rate.

      As long as it is profitable to borrow short and lend long, someone is going to find some way to do it. A Pigouvian tax on short term debt will not in my opinion change that.

      I read through Gersbach's paper and the same questions keep popping in my head.

      From the paper (page 7):

      "If macroeconomic risk is present, we allow for contracts to be conditioned on the realization of qt or on the resulting GDP in period t − 1. In such cases, state contingent deposit or loan contracts can be written."

      Gersbach is referring to both types of contracts (debt, deposits) being conditioned to the state of the economy. But he does not clarify how that conditioning is / should be applied.

      Should debt and deposit contracts be written where their values are pro cyclical (values rise and fall with GDP growth) or counter cyclical (values rise with falling GDP growth and fall with rising GDP growth)? Or should one of the two (debt / deposits) be pro cyclical and the other be counter cyclical?

    10. If we know our objective, I'm sure that there are mathematical methods to determine the optimal form of indexation. I remember long ago in grad school they were showing us how to use the calculus of variations to find rules that optimize a problem. I don't know if that could be done here, but I also remember John Taylor writing about the search for good Taylor rules was done by testing several candidates using formal models, to see which worked best. Here the proposal is for banks to implement a rule, instead of the monetary authority.

      I've mentioned borrowing and lending Trills in previous comments, because they seem like a natural candidate for the indexed debt I'm talking about. If we abstract from default risk, then what should I pay to divert resources from an economy growing at g? If I pay a rate of g for that loan, then we have an equilibrium between real investment in the economy and the rate charged by financial investors, with banks acting as a very smooth conduit for risk. In the present system, risk accumulates at banks.

    11. Anwer,

      "If we abstract from default risk, then what should I pay to divert resources from an economy growing at g? If I pay a rate of g for that loan, then we have an equilibrium between real investment in the economy and the rate charged by financial investors, with banks acting as a very smooth conduit for risk. In the present system, risk accumulates at banks."

      Suppose all debt was of the floating rate type and the interest rate on that debt was positively correlated with g. A higher growth rate g causes the floating interest rate on that loan to rise and a lower growth rate g causes the floating interest rate on the loan to fall.

      Problem - A negative growth rate (recession) would reverse the flow of interest payments from "borrower to lender" to "lender to borrower". This creates a really perverse incentive to borrow money, bury it in the ground, and wait for the inevitable recession.

      Possible answer #1: Limit downside risk to lenders to 0% return on lending.
      Possible answer #2: Make the value of both deposits and loans positively correlated with the growth rate g. Sure the value of loans falls during a recession, but so does the value of borrowed money buried in the ground.

      Problem with possible answer #2: What prevents the value of loans and deposits to fall to zero?

    12. Frank I think the key here is to distinguish between ex ante and ex post. You are describing adjustable interest rate contracts, which are already in use e.g. ARM mortgages. Gersbach considers contingent contracts (like insurance) where the payment due for a period is determined at the end of it, after risks are realized.

      Atif Mian and Amir Sufi are also advocating contingent contracts when they talk about indexed debt. They want borrowers to be properly hedged against aggregate risks. That is not how things are now. We have businesses, homebuyers, banks etc. all borrowing dollars, without proper hedging. Banks especially have significant moral hazard and externalized costs, and they expect bailouts in bad states of the economy. Some people think that this can be managed with complex regulation, but there are very good criticisms of that approach, such as the lecture above by Prof. Cochrane. I believe that it's more useful to support the "macro markets" project Robert Shiller started over 20 years ago. Let's address the missing markets for contingent contracts that hedge macroeconomic risks.

      Here is a simple way to answer your question about the proper way to index debt - pro-cyclical vs. countercyclical: instead of borrowing dollars, you borrow an index that is a suitable benchmark for your performance. For example, it could be a measure of regional income, or an index of local home prices (which Shiller helped establish on the Chicago exchange). The payments due on your debt are determined by this index, which hedges you against aggregate risks. If you expect to beat the relevant benchmark, you should seek funding, and not if you are likely to under-perform. This arrangement should support an efficient distribution of risk.

    13. Anwer,

      "Frank I think the key here is to distinguish between ex ante and ex post."

      I am not sure why that is relevant. Ex ante / ex post analysis of single time period contracts makes sense. Most debt contracts extend across multiple time periods (12 months, 60 months, 360 months).

      Whether periodic adjustments to the contract terms are made starting at the beginning of every month (prior to that month's economic news) or starting at the end of every month (after that month's economic news) make little difference except for very short term debt.

      Banking that relies on short term funding can be subject to undo volatility in short term interest rates caused by the central bank. But the last time the U. S. Fed raised short term interest rates, they went out of their way to limit that volatility - raising aways in 25 basis point steps.

      "You are describing adjustable interest rate contracts, which are already in use e.g. ARM mortgages."

      Is this not a form of contingent contract? Sure ARM mortgage rates tend to track the short term interest rate set by the central bank rather than an index, but I would still consider them contingent rather than fixed.

      "Instead of borrowing dollars, you borrow an index that is a suitable benchmark for your performance....The payments due on your debt are determined by this index, which hedges you against aggregate risks."

      Let me try my hand at a simple example - Debt and deposits are pro cyclical

      I borrow 100 Trills. The value of a trill is 1 trillionth of the GDP level. I enter into a continent debt contract that says I will make monthly payments on those 100 Trills over 5 years equal to principle plus an "interest rate" equal to the monthly growth rate of GDP over the same 5 years.

      First, whether changes in the monthly interest rate that I am paying are made ex post or ex ante to measured monthly changes in GDP should not affect my net payments a whole lot unless: I am borrowing for a very short period of time or the volatility of measured GDP is significant.

      Second, the problem, as I mentioned before is one where the value of both debt and deposits begin a death spiral. There is nothing in this solution (pro cyclical debt and deposits) to stop the value of both debt and deposits from declining to zero.

      So we address the missing markets by creating a mechanism where two markets (debt and deposits) can completely evaporate?

    14. You're right, Frank - banks do use contingent contracts, including ARMs. I realized that the issue was more complex after posting my comment. Gersbach wants to make deposits more risky, so that banks can pass the risk on their balance sheets on to depositors. Cochrane and others want the same thing when they call for money funds to report net asset value (NAV). I'm looking for a way to do that while retaining more of the money-like aspects of deposits, namely information insensitivity, because Gary Gorton explains the latest financial crisis as a breakdown in this property. Indexed debt is risky, but it is still debt and still alleviates the information problems that motivate the use of debt.

      I think you are also commenting on excess financial elasticity and the possibility of multiple equilibria, which I've seen as a topic of discussion right now amongst economists. I think that this problem is linked to writing too many contracts using dollars. We don't know how much wealth there will be in the future, and it's easy in good times to become overly optimistic. But we do know that it is impossible to own more than 100% of the economy, and also that there is no way to make the real economy shrink to zero. Trills are contracts written in real terms, and greater use of them might reduce the excess elasticity that contributes to crises.

    15. Anwer,

      "Gersbach wants to make deposits more risky, so that banks can pass the risk on their balance sheets on to depositors....I'm looking for a way to do that while retaining more of the money-like aspects of deposits"

      Rather than focusing on adding riskiness to deposits, I think your best bet is looking at means of adding more equity (a risk bearing asset distinguishable from both deposits and debt) into an economic system. In that case (additional equity), those that want to take risk are free to do so, and those that don't want to take risk are free to do so as well. Adding riskiness to deposits seems heavy handed to me.

      I don't believe Gersbach mentions equity financing in his paper, though Cochrane has been a proponent of banks utilizing equity to fund lending (rather than short term debt).

      I don't believe that John's Pigouvian tax on short term borrowing is an effective means of accomplishing the goal of making an economic system more reliant on equity and less reliant on debt.

      "I think that this problem is linked to writing too many contracts using dollars."

      I think that this problem is linked to writing too many guaranteed contracts (debt) payable in the medium of exchange.

  7. Hi John,

    I enjoyed your presentation and this is my first time on your blog. I agree entirely with your focus on bank liabilities and the concept that "runs defined the events as a crisis". A question though, why 100% equity? Granted that equity is perpetual funding and so (over)matches illiquid and long-dated assets, but is there no room for matching long-dated debt that does not share the loss-absorbing property of equity? Tech stocks were naturally 100% equity funded due to high or extreme riskiness - but bank assets as you say are on a different scale and so could accommodate some (approximate) matching leverage.

  8. An intriguing feature of a banking system financed by equity rather than debt would be its compliance with sharia law. Profit sharing through investment in risk taking enterprises is acceptable to Islam, so interest bearing demand deposits based on purchase and sale of floating price ETF units, should be quite kosher.

  9. John - the Fed seems to be interested in implementing your idea. From today's WSJ:

    "The Fed also will penalize U.S. banks that rely heavily on volatile forms of short-term funding, such as overnight loans, in determining the size of a new capital surcharge."


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.