Ben Bernanke gave another Augustinian give-us-QEn-but-not-yet* speech at Jackson Hole today and you could go read it but honestly why would you, you know what it says, which is "everything is bad, but not as bad as it could be, and we want to make it a bit better, but only once it's gotten a bit worse." Moving right along.
To Andrew Haldane's speech, which is a treat! It is here and its title is "The dog and the frisbee," so obviously he had Dealbreaker on his side right there. Haldane, the Bank of England's financial-stability guy, basically argues that while the financial system is complex, it should be regulated simply - "As you do not fight fire with fire, you do not fight complexity with complexity" - just as a dog uses only elementary trigonometry and differential calculus to solve the complex and multivariate problem of catching a frisbee.**
Haldane's main example of overcomplexity in regulation is risk-based capital regulation, in which the Basel accords have moved from simple leverage tests - common equity divided by total assets - to complicated tests where the numerator is made up of different tiers of capital and the denominator uses risk-weights that are largely driven by the bank's own models of riskiness. One thing you could do is compare the performance of those measures in the recent crisis, so he did. Here is how Basel risk-based capital did:
That looks bad and also is bad, with no statistically significant difference between banks that blew up and banks that did not. This is just boring leverage:
This looks better, and Haldane adds "The pre-crisis leverage ratio of failing banks was statistically significantly lower than surviving banks at the 1% significance level, by on average 1.2 percentage points." Dumb leverage is much better at predicting bank failure than smart leverage.
He points out some other complexifications that have had mixed results - VaR being a big one*** - and sums up:
Over the past 30 years or so, the regulatory direction of travel has been towards pricing risk in the financial system, rather than prohibiting or restricting it. ... [R]egulators have pursued price over quantity-based regulation. That makes sense when optimising in a risky world.
It may make less sense when optimising in an uncertain world. Quantity-based restrictions may be more robust to mis-calibration. Simple, quantity-based restrictions are the equivalent of a regulatory commandment: “Thou shalt not”. These are likely to be less fallible than: “Thou shalt provided the internal model is correct”. That is one reason why Glass-Steagall lasted for 60 years longer than Basel II.
The speech is very good but this sort of thing always leaves me a bit unmoved. That Basel leverage vs. regular leverage test is biased by the fact that banks were regulated to Basel leverage, meaning:
- everyone had to have roooughly the same amount of Basel capital,**** and
- your incentive was to maximize the actual risk you took with each dollar of regulatory risk.
Thus higher total leverage in that regulatory regime is less about "who was more levered" and more about "who was more aggressive in gaming capital requirements to maximize risks and returns"; there's no reason to think that in a world regulated to total leverage the aggressive ones wouldn't be gaming that. (By, for instance, using all of their allowed leverage to buy risky assets.) Perhaps related is this interesting paper from Robert Jarrow claiming that capital-adequacy regulation based on VaR can increase systemic risk by encouraging banks to sell a lot of tail-risk insurance:
Under a VaR capital adequacy rule, firms can easily adjust their assets and liabilities to increase the probability of catastrophic losses, while improving or keeping VaR unchanged. Given limited liability and for banks (deposit insurance), management and shareholders would have an incentive to do so. The balance sheet modi fication is conceptually very simple. The firm needs to sell put options on the risks in the "far" left-tail of its loss distribution. These increased liabilities are not registered by VaR. But selling these put options brings in cash, which can be used to purchase additional assets. The newly purchased assets decrease VaR. This is an example of regulatory arbitrage.
That passage is model-based but grounded in history - basically it corresponds to "write a lot of puts on AAA assets" - and pretty much describes how you get low levels of risk-weighted leverage but high levels of actual leverage.
Jarrow's solution is to add another capital-adequacy metric, based on conditional expected loss given exceeding VaR, to make that arbitrage impossible; Haldane's would presumably be to subtract a metric: "don't worry about VaR, just don't write too many puts on any assets." Either would probably work to shut down this particular game.
But this is just one game. And Haldane's other simplifiers aren't that encouraging:
Quantity-based regulatory solutions have gained currency during the course of the crisis. In the US, the Volcker rule is a quantity-based regulatory commandment: “Thou shalt not engage in proprietary trading”. In the UK, the Independent (“Vickers”) Commission on Banking has also proposed structural, quantity-based reforms: “Thou shalt not co-mingle retail deposit-taking and investment banking”.
Yet even these notionally simple, structural proposals run some risk of backdoor complexity. For example, the consultation document accompanying Volcker already runs to 298 pages.
Yeah yeah yeah the Volcker Rule is too complicated, it should be simpler, why not just say "don't prop trade"? But of course then no one could tell you what prop trading is. The Vickers/Glass-Steagall rule is simple, but a rule like "don't use retail deposits to fund investment banking" leads to money market funds that invest in repos on securities dealers' inventories, which are very different words that mean "using retail deposits to fund investment banking."
I feel like the model underlying these simplify-your-regulations arguments is:
- risk and reward are correlated,
- bankers have asymmetric upside and so want more risk (as more risk = more upside, and their downside is floored at zero),
- regulators can price/limit/ban/whatever any finite number of risks, with the cost in regulator heartache increasing in the number of regulated risks,
- but the number of possible risks is infinite and bankers only need to think of one that the regulators didn't think of to make a lot of money and maybe blow up the world, so
- regulators should stop trying to think of lots of things and try thinking of fewer things.
The first four of those points seem more or less inarguable. The fifth is a very natural thing to think in response, and solves the important (for regulators) optimization problem of reducing regulator heartache. I just don't see why it would work.
The dog and the frisbee - paper by Andrew Haldane [BoE, and press release]
Jackson Hole Paper 2 – Destroying the Tower of Basel [Money Supply]
Capital Adequacy Rules, Catastrophic Firm Failure, and Systemic Risk [Harvard Law and SSRN]
** Haldane actually says "For studies have shown that the frisbee-catching dog follows the simplest of rules of thumb: run at a speed so that the angle of gaze to the frisbee remains roughly constant," and I cannot stop myself imagining the dog reasoning that out with a few simple diagrams and an HP-12C.
*** This is a good thing to ponder:
To give some sense of scale, consider model-based estimates of portfolio Value at Risk (VaR), a commonly-used technique for measuring risk and regulatory capital in the trading book. A large firm would typically have several thousand risk factors in its VaR model. Estimating the covariance matrix for all of the risk factors means estimating several million individual risk parameters. Multiple pricing models are then typically used to map from these risk factors to the valuation of individual instruments, each with several estimated pricing parameters.
Taking all of this together, the parameter space of a large bank’s banking and trading books could easily run to several millions. These parameters are typically estimated from limited past samples. For example, a typical credit risk model might comprise 20-30 years of sample data – barely a crisis cycle. A market risk model might comprise less than five years of data – far less than a crisis cycle.
**** I mean, not really, but everyone had to have adequate Basel capital, and wanted to minimize capital subject to that requirement. Note that that chart has a narrower range than the total leverage chart. So you shouldn't really expect differences in adequate Basel capital levels to distinguish failed and non-failed banks.