Wednesday, November 30, 2016

A Better Choice

Roll up your shirtsleeves, financial economists. As reported by Elizabeth Dexheimer at Bloomberg, Rep. Jeb Hensarling is “interested in working on a 2.0 version,”  of his financial choice act, the blueprint for reforming Dodd-Frank. “Advice and counsel is welcome."

The core of the choice act is simple. Large banks must fund themselves with more capital and less debt. It strives for a very simple measure of capital adequacy in place of complex Basel rules, by using a simple leverage ratio. And it has a clever carrot in place of the stick. Banks with enough capital are exempt from a swath of Dodd-Frank regulation.

Market based alternatives to a leverage ratio

The most important question, I think, is how, and whether, to improve on the leverage ratio with simple, transparent  measure of capital adequacy. Keep in mind, the purpose is not to determine a minimum capital level at which a bank is resolved, closed down, bailed out, etc. The purpose is a minimal capital ratio at which a bank is so systemically safe that it can be exempt from a lot of regulation.

The "right" answer remains, in my view, the pure one: 100% equity plus long term debt to fund risky investments, and short term liabilities entirely backed by treasuries or reserves (various essays here). But, though I still think it's eminently practical, it's not on the current agenda, and our task is to come up with something better than a leverage ratio for the time being.

Here are my thoughts. This post is an invitation to critique and improve.

Market values. First, we should use the market value of equity and other assets, not the book value. Risk weights are complicated and open to games, and no asset-by-asset system captures correlations between assets. Value at risk does, but people trust the correlations in those models even less than they trust risk weights. Accounting values pretend assets are worth more than they really are, except when accounting values force marks to market that are illiquid or "temporarily impaired."

Market values solve these problems neatly. If the assets are unfairly marked to market, equity analysts know that and assign a higher value to the equity. If assets are negatively correlated so the sum is worth more than the parts, equity analysts now that and assign a higher value to the equity.

Liabilities not assets. Second, we should use the ratios of liability values,  not ratios to asset values.   Rather than measure a ratio of equity to (accounting) asset values, look at the ratio of equity to the debt that the bank issues. Here, I would divide market value of equity by the face value of debt, and especially debt under one year. We want to know, can the bank pay off its creditors or will there be a run.

In principle, the value of assets = the value of liabilities so it shouldn't matter. Accountant and regulator assets are not the same as liabilities, which raises the important question -- if you want to measure asset values rather than (much simpler) liability values, then why are your asset values not the same as my liability values?

So far, then, I think the ratio of market value of equity to (equity + face value of debt) is both better and much simpler than the leverage ratio, book value of equity to complex book value of assets.

One can do better on ratios. (Equity + 1/2 market value of long-term unsecured debt ) / market value    of short term debt is attractive, as the main danger is a run on short-term debt.

Use option prices for tails. Market value of equity / face value of debt is, I think, an improvement on leverage ratios all around. But both measures have a common problem,  and I think we can do better.

A leverage (equity/assets) ratio doesn't distinguish between the riskiness of the assets. A bank facing a leverage constraint has an incentive to take on more risk. For example, you can buy a stock which costs $100, or a call option which costs $10, each having the same risk -- when the stock market moves 1%, each gains or loses $1 of value. But at a 10% leverage the stock needs $10 of capital and the call option only $1.

The main motivation of risk-weights is to try to measure assets' risk -- not the current value, but the chance of a big loss in value -- and make sure there is enough equity around for all but the worst risks. So let's try to do this with market prices.

A simple idea: So, you're worried that the same value of equity corresponds to a riskier portfolio? Fine: use option prices to measure the banks' riskiness. If bank A has bought stock worth $100, but bank B has 10 times riskier call options worth the same $100, then bank B's option prices will be much larger -- more precisely, the implied volatility of its options will be larger.

So, bottom line: Use the implied volatility of bank options to measure the riskiness of the bank's assets. As a very simple example, suppose a bank has $10 market value of equity, $90 market value of debt, and 25% implied volatility of equity. The 25% implied volatility of equity means 2.5% implied volatility of total assets, so (very roughly) the bank is four standard deviations away from wiping out its equity. Yes, this is a simplistic example, and the refinements are pretty obvious.

(For non-finance people: An option gives you the right to buy or sell a stock at a given price. The more volatile the stock, the more valuable the option. The right to sell for $80 a stock currently going at $100 is worth more, the more likely the stock is to fall below $80, i.e. the more volatile the stock. So option prices tell you the market's best guess of the chance that stocks can take a big fall.  You can recover from option prices the "implied volatility," a measure of the standard deviation of stock returns.)

We might be able to simplify even further. As a bank issues more equity and less debt, the equity gets safer and safer, and stock volatility goes down, and the implied volatility of options goes down. Perhaps it is enough to say "the implied volatility of your at the money options shall be no more than 10%."

Here's the prettiest rule I can think of. A put option is the right to sell stock at a given price. Assemble the minimum cost of put options that give the bank the right to issue stock sufficient to cover its short-term debt. For example, if the bank has $1,000 of short-term debt, then we could look at the value of 10 put options, each giving the bank the right to sell its stock at $100. If the market value of equity is greater than the cost of this set of put options, then the bank is ok.

(It would be better still if banks actually bought these put options, so they always had sitting there the right to issue equity in bad times. But then you might complain about liquidity and counterparty risk, so let's just use this as a measurement device.)

That's probably too fancy, but one should always start with the ideal before compromising. (Back to 100% equity.... )

In summary, I think we could improve a lot on the current leverage ratio by 1) using market values of equity 2) using ratios of liabilities, not accounting asset values at all and 3) using option prices to measure risk.

I left out the use of bond yields or credit default swaps to measure risk. The greater a chance of default, the higher interest rate that markets charge for debt, so one could in principle use that measure. It has been proposed as a trigger for contingent bonds or for regulatory intervention. I'm leery of it for lots of reasons. First, we're here to measure capital adequacy, so let's measure capital. Second, credit markets don't provide good measures of whether you're three or four standard deviations from default. Third, credit markets include not just the chance of default, but also the guess about recovery in default, and thus a guess about how big the bailout will be. But there is no reason in principle not to include bond information in the general picture -- so long as we can keep to the rule simple and transparent .

Our first step is to get our regulators to trust the basics: 1) stock markets provide good measures of total value -- at least better than regulators 2) option markets provide good measures of risk -- at least better than regulators.

Why not? I think our regulators and especially banks don't trust market values. They prefer the central-planning hubris that accountants and regulators can figure out what the market value and risk are better than the actual market.

If so, let's put this on the table in the open and discuss it. If the answer is "your proposal to use market value of equity and options is perfect in theory but we trust regulators to get values right a lot more than markets," then at least we have made 90% progress, and we can start examining the central question whether regulators and accountants do, in fact, outperform market measures. The question is not perfection or clairvoyance, it's whether markets or regulatory rules do a less bad job. Markets were way ahead of regulators in the last crisis.

What if market gyrations drive down the value of a bank's stock? Well, this is an important signal that bank management and regulators should take seriously by gum! Banks should have issued a lot more equity to start with to make sure this doesn't happen; banks should have issued cocos or bought put options if they think raising equity is hard. And when a bank's equity takes a tumble that is a great time to send the regulators in to see what happened. The choice act very nicely sets the equity ratio up as the point where we exempt banks from regulation, not a cliff where they get shut down.

Let's also remember, when you read the details, the leverage ratio is not all that simple or transparent either. Here is a good summary.

And let's also remember that perfection should not be the enemy of the much better. Current Basel style capital regulations are full of distorted incentives and gaming invitations. If there are small remaining imperfections, that

Or maybe not

 Is fixing the leverage ratio all important?  What's wrong with a leverage ratio? Right now, banks have to issue capital if they take your money and hold reserves at the Fed or short term Treasury debt. That obviously doesn't make much sense as it is a completely riskless activity. More subtly, a leverage ratio forces banks to issue capital against activities that are almost as safe, such as repo lending secured by Treasuries.  Required reading on these points: Darrell Duffie Financial Regulatory Reform after the crisis: An Assessment
... the regulation known as the leverage ratio has caused a distortionary reduction in the incentives for banks to intermediate markets for safe assets, especially the government securities repo market, without apparent financial stability benefits....I will suggest adjustments to the leverage ratio rule that would improve the liquidity of government securities markets and other low-risk high-importance markets, without sacrificing financial stability.
The natural response is to start risk-weighting lite. The Bank of England recently exempted government securities from their leverage ratio.  The natural response to the response is, once we start making exceptions, the lobbyists swarm in for more. You can see in Duffie's writing that an exemption for repo lending collateralized by Treasuries will come next. Given the fraction of people who understand how that works, the case for resisting more exemptions will be weak.

The poster child for the ills of risk-weighted asset regulation: Greek sovereign debt still carries no risk weight in Europe. Basel here we come.

Interestingly, Duffie does not see banks currently shifting to riskier investing, the other major concern, though that may be because the Volker rule, Basel risk weights and other constraints also apply. So perhaps I should state the market-based measures not as alternatives to the leverage rule, but as measures to add to the leverage rule, in place of the other constraints on too much risk.

But how much damage is really done by asking capital for safe investments? Recall the Modigliani-Miller theorem after all. If a bank issues equity to fund riskfree investments, the equity is pretty darn risk free too, and carries a low cost of capital.  Yes, MM doesn't hold for banks, but that's in large part because of subsidies and guarantees for debt, and it's closer to true than to totally false -- the expected return on equity does depend on that equity's risk -- and the social MM theorem is a lot closer to holding and that's what matters for policy.

And even if funneling money to safe investments costs, say, an extra percent, does that really justify the whole Dodd-Frank mess?

In the end, it is not written in stone that large, systemic, too big to fail banks must provide intermediation to safe investments. A money market fund can take your deposits and turn them in to reserves, needing no equity at all. A bank could sponsor such a fund, run your deposits through that fund, and you'd never notice the difference until the moment the bank goes under... and your fund is intact.

Duffle again:
These resiliency reforms, particularly bank capital regulations, have caused some reduction in secondary market liquidity. While bid-ask spreads and most other standard liquidity metrics suggest that markets are about as liquid for small trades as they have been for a long time, liquidity is worse for block-sized trade demands. As a trade-off for significantly greater financial stability, this is a cost well worth bearing. Meanwhile, markets are continuing to slowly adapt to the reduction of balance sheet space being made available for market-making by bank-affiliated dealers. [my emphasis] Even more stringent minimum requirements for capital relative to risk-weighted assets would, in my view, offer additional net social benefits. 
I emphasized the important sentence here. There are many other ways to funnel risk free money to risk free lending activities. The usual mistake in financial policy is to presume that the current big banks must always remain, and must always keep the same scope of their current activities -- and that new banks, or new institutions, cannot arise when profitable businesses like intermediation open up.

So, in the worst case that a liquidity ratio makes it too expensive for banks to funnel deposits to reserves, to fund market-making or repo lending, then all of those activities can move outside of big banks.

More Choice act

The Choice act has some additional very interesting characteristics.

Most of all, it offers a carrot instead of a stick: Banks with sufficient equity are exempt from a swath of regulation.

That carrot is very clever. We don't have to repeal and replace Dodd-Frank it its entirety, and we don't have to force the big banks to utterly restructure things overnight. Want to go on hugely leveraged? The regulators will be back in Monday morning. Would you rather be free to do things as you see fit and not spend all week filling out forms? Then stop whining, issue some equity or cut dividends for a while.

More deeply, it offers a path for new financial institutions to enter and compete. Compliance costs and a compliance department are not only a drag on existing businesses, they are a huge barrier to entry. Are markets illiquid? Are there people who can't get loans? The answer, usually forgotten in policy, is not to prod existing businesses but to allow new ones to enter. A new pathway -- lots of capital in return for less asset-risk regulation -- will allow that to happen.

Both politically and economically, it is much easier to let Dodd-Frank die on the vine than to uproot and replant it.

In the department of finish sanding, I would also suggest a good deal more than 10% equity.   I also would prefer a stairstep -- 10% buys exemption from x (maybe SIFI), 20% buys you exemption from y, and so forth, until at maybe 80% equity + long term debt you're not even a "bank" any more.

Remember, the issue is runs, not failure. Banks should fail, equity wiped out, and long-term debt becomes equity. The point of regulation is not to make sure banks are "safe" and "don't fail." The point of regulation is to stop runs and crises. So ratios that emphasize short term debt are the most important ones.

Duffle (above) also comes down on the side of more capital still. The "Minneapolis plan" spearheaded by Minneapolis Fed President Neel Kashkari (Speechreport by James Pethokoukis at AEI) envisions even more capital, up to 38%.


  1. John,

    "Rather than measure a ratio of equity to (accounting) asset values, look at the ratio of equity to the debt that the bank issues. Here, I would divide market value of equity by the face value of debt, and especially debt under one year. We want to know, can the bank pay off its creditors or will there be a run."

    Possibly too technical to discuss here, but I will ask anyway:

    First, how should convertible bonds and other contingent securities that have both equity and debt aspects be treated? It would be helpful if the legislation nailed down what exactly constitutes the distinction between the two.

    Second, how should a bank be defined? Is any company that does vendor financing considered a bank under the legislation? Are intermediaries (hedge funds, holding companies, insurance companies, mutual fund companies) considered banks as well or is the definition of a bank limited to those companies that originate loans rather than any company that may purchase and hold loans and other securities?

    Third, are the quasi-government agencies (Sallie Mae, Freddie Mac, Fannie Mae, etc.) required to follow the same level of governance in terms of managing a debt / equity ratio? How do direct government purchases of equity sold by these companies come into play?

    Fourth, how does this legislation work within the international finance community? A U. S. bank based bank borrows abroad and hides those loans from the U. S. government and the bank's investors.

    Fifth, how should U. S. Treasury dealings with the primary dealers be handled? Most primary dealers borrow short term to buy debt from government auctions only to resell the debt at a later time. Trying to do that with equity issuance and re-purchases might be unsettling to equity markets when the float volatility is considered. Recognize that the primary dealers consists of domestic U. S. banks as well as banks headquartered overseas and so domestic buyers would likely be disadvantaged.

    I am cautiously optimistic that regulation of this form could have real benefit. I am concerned that Congress will ultimately get it wrong (they were the ones that instituted Dodd Frank to begin with).

    It would be much simpler (with far greater benefit) if Mr. Hensarling and newly minted Treasury Secretary Mnuchin would realize that the best angle of attack is for the U. S. government itself to sell equity claims on it's tax revenue rather than trying to perfect some form of legislation to reign in the debt excesses of the banking sector.

    It accomplishes the same purpose without relying on Congress.

  2. On another front:

    "If we have a border adjustable tax system, that can solve a lot of these trade issues that Trump is talking about, economic analyst and Trump adviser Stephen Moore said in an interview."

    “You’re going to tax what’s imported and not going to tax what’s exported. So we’re going to reduce the trade deficit and we’re going to have more companies come in here, Moore said."

    Does anyone understand reciprocity? Does anyone understand capital accounts and current accounts?

    So the U. S. raises taxes on companies that import (including retailers) and lowers taxes on companies that export. China, Japan, etc. do the same - what have you accomplished?

  3. Some more food for thought:

  4. John:

    As you point out, baking regulations are important but they ought to be simple.

    As Frank points out, the systemic global banks are a lot more complex than they might appear theoretically.

    Static risk weighted schemes such as BASEL III all invite optimization or as you say "gaming."

    Plus, the real issue has not only balance sheet measures but the required infrastructure ( data, technology, expertise, and management attention) to do timely analysis of the risk. I think practically the most straight forward approach is to require stress analysis -- similar to what is done now annually but with more rigor and transparency and less complexity. Then you could readily see the different nature of assets in each banks.

    Regulators should define static (versus dynamic over time) scenarios, banks submit their position data to regulators, and both regulators and banks would perform their analysis. The results would be shared with public. Everything transparent.

    Overtime, this approach would lead to uniform data maintenance across banks, and development of standard analytical tools for risk analysis. Things would converge.

  5. On a somewhat unrelated topic, progressive economist Larry Summers has rediscovered the merits of rule of law:

    So keep fighting the good fight prof Cochrane.

  6. Very cool stuff. And although I agree that a market based measure would be better than a book based one, many of the smaller banks are not listed so that only book equity is available. Very clever indeed!

  7. No. 100% No. Forever No. Moronic approaches will not create long term solutions to complex problems (watch how this "Trump" thing works out, ha). That said, seeking better elegant simple solutions to complex problems is worthy.

    No bank, HF, or client working with me lost money - since 2006. I know the biggest failures. I know why. Its starts and end with the silly stupidity of "mark-to-market". This was a game, setup to be played against banks. See that? That is exactly the wrong way.

    Imagine homeowners were "marked-to-market" on there mortgages? Someone would come to your door and say you needed to post 50k or 200k to stay in the home. Your equity was now too low? This is crazy absurd. Worse, it would just be tragic stupidity for everyone involved. But then isn't this exactly MTM for long-term credit and structured finance assets?

    The answer lies in the other direction. Seeing the foolhardy misplacement of MTM.

  8. I like your thoughts. I'm not qualified nor competent to comment on the detail, but my observation is that:

    1. You start with a simple concept that deals with a concern - what should an organisation do to demonstrate sufficient safety to get out of regulation. The answer to that is 100% equity.

    2. You seem to think that isn't palatable, so you start engineering ways to determine safety that are a step back from that position. But each of those is a step towards the exact regulations we would avoid

    3. You look for market solutions, but those solutions I think have political problems (believing that markets are more accurate than regulators - which may be true, but it's unlikely the regulators would agree) and seem somewhat open to gaming (for example, options on bank equity may be a thin market that you could manipulate, and really, how well did options price in the last crisis?)

    4. I wonder whether your suggestion in the body is the right one - come at it the other way. Instead of starting at perfect (100% equity) and then nibbling away at it, start articulating how much regulation we'd let you off for each thing that you do. The regulations relaxed would need to somehow correlate to the risk removed - so bank does X and we let you out of regulation Y because the thing that was chasing is now not a risk.


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.