I hardly seem to open a newspaper or read influential blogs without seeing the crisis being blamed on the failure of regulation – more regulation is supposedly the only way to prevent future crisis.
Does this really make sense? After all, it was the most regulated part of the financial system that was worst affected by the crisis, the banks.
Do we really know how to regulate banks?
Is the problem a lack of regulation? Or is the problem that we really don't know how to regulate financial institutions properly? Depending on the answer, the approach to future financial regulations should be very different. Recall, the last time financial regulation was revamped – Basel II – the designers got it wrong.
There is plenty of time – the next serious banking crisis is not due for quite a number of years. There is time to get it right.
The prudent way forward is to first identify how financial regulations failed and define how to best regulate banks before regulating everything in sight.
With the crisis (often rightly) being blamed on inadequate regulations, pro-regulation views are not surprising. Indeed, the pro-regulation crowd is taking advantage of the crisis to push a pre-existing agenda.
Unfortunately, there is a tendency to view the efficacy of regulations with rose-tinted glasses. We are surrounded in our daily lives by a pretty effective regulatory system - traffic. Regulations - traffic laws - are robust and transparent. Supervision - the traffic cops - is pretty effective in enforcement.
The state-of-the-art in banking regulations, Basel II, is not that effective, in particular its reliance on risk measurements is pretty disastrous.
Value-at-Risk and friends
The founding philosophy of most risk systems, including those at the heart of Basel II, is the measurement of risk. Risk engines are a sort of a black box, in one end goes data and out the other comes Value-at-Risk (VaR).
VaR is funded on the notion that we can represent the financial system by a sequence of mathematical equations, and if we chance upon the right ones, we can measure risk. This is a phenomenon I called the myth of the riskometer in a previous Vox column.
VaR gets risk wrong. Frank Knight, in 1921, studying the concept of randomness, proposed a classification system differentiating risk from uncertainty. With risk we can assign mathematical probabilities to outcomes, while with uncertainty we cannot do that.
We will arrive at very different approaches to financial regulation depending on how we view randomness in the financial system – it’s either risk or uncertainty.
Let's be clear about one thing. VaR assumes randomness is risk. That is why VaR is unreliable. Indeed, I think most individuals who actually work with risk models would agree with me in their heart of hearts.
Trying to go to even higher order risk measures like tail VaR and CoVaR is even less reliable.
Risk models are least reliable when we need them the most
Risk models are least reliable when we need them the most, because human beings are intelligent. If we see a measure of risk, we react to it. In most cases, different people will make different decisions. However, in times of crisis we tend to do the same thing. The fact that we measure risk makes us react more strongly than we otherwise would.
VaR was a big step backwards. It replaces a flawed subjective judgment with even a more flawed quantitative judgment. The problem is it's a lot harder to argue with a computer.
Onto Basel II
Basel I was originally designed to prevent banks from competing away their capital. This was a real danger in the 1980s. Japanese banks were taking over the world simply because they had 4% capital, while the banks in the West had at least 8%. The result was that banks were competing capital away.
Basel I was successful in stopping that situation. It, for all its flaws, had one redeeming quality – it was not based on the notion of risk sensitivity. It does not really depend on measuring risk. Basel II, by contrast, is founded on risk measurements – both pillar 2 with its emphasis on internal risk management and pillar 1 with its focus on capital.
The calculation of bank capitalisation is a surprisingly convoluted affair. Do we focus on a broad measure like tier 2, only tier 1, even narrower measures such as core tier 1, or even tangible equity? Is the denominator composed of risk-weighted assets (RWA) or total assets (TA)?
In looking at tables showing bank capitalisations and rankings, very different pictures emerge depending on whether we look at something like tier 1/RWA as in Europe under Basel II, or the US leverage ratio - tier 1/TA. Many financial institutions are well capitalised according to the former but really poorly capitalised under the latter.
The financial engineering premium
The reason is what may be called the financial engineering premium. Sophisticated banks can make risk-weighted assets really low by a judicious measurement of risk, perhaps even if they hold a sizable chunk of toxic waste. They can not do that with total assets.
The problem with risk-weighted capital is that it is only as good as the quality of the risk measurements. If the problem of measuring risk is much harder than it is claimed, then immediately risk-weighted capital becomes suspect. Perhaps then we don’t trust banks when they tell us they are well capitalised. Perhaps then they become stigmatised.
Perverse impacts of risk sensitive capital - Banks won’t lend
With risk-sensitive capital, banks are supposed to reduce risky activities when the models tell them risk is increasing.
This has a particular perverse impact, making banks behaviour similar to behaviour during a crisis. Banks in general will have to sell the same risky assets and buy the same assets. That by itself makes the prices of the risky assets fall, with further increases in risk which erodes capital. This is modelled formally in Danielsson and Zigrand (2008).
This creates a vicious feedback cycle - the so-called fire-sale externalities.
The perversity of risk-sensitive capital makes banks withdraw from risky activities at exactly the moment when we want them to do the opposite.
In a financial crisis, where financial institutions are required by regulations to hold minimum capital, just that fact is destabilising. If asset prices are falling, the financial institutions need to sell high-risk assets which depress the price and therefore by itself erode their capital. This is why risk-sensitive capital increases systemic risk and forces banks to withdraw lending.
Indeed, the banks are now doing what they are supposed to do. They are being prudent. It is a bit disingenuous of regulators and politicians demanding that the banks increase lending when the banks are just following the regulations designed by the very same regulators and approved by the very same politicians.
The “regulate everything” crowd
The crisis is bringing out pro-regulation views that, mercifully, have been hidden until now.
I think the crisis is allowing a lot of people to dust off a financial regulation agenda that they could not get passed in a typical year. One example comes from the EU, where Poul Nyrup Rasmussen, MEP and President of the Party of European Socialists, has declared that hedge funds have contributed to the crisis and that EU legislative action is proven.
I had thought that the one sector of the financial system that was fairly innocent in the crisis was hedge funds. Indeed, on the contrary, hedge funds have been beneficial, adding to liquidity.
If it wasn't for the hedge funds, who would buy the toxic assets from the banks? Far from being the villains, if it wasn't for the hedge funds the banks might even be in a worse situation then before. There is no evidence that the unregulated status of hedge funds and private equity firms in any way contributed to the crisis.
Indeed, trying to regulate them would be quite tricky. I pity the poor supervisor who gets full disclosure from a sophisticated hedge fund. It would be like drinking from a fire hose.
What is the supervisor to do with this information? Presumably the supervisor would measure systemic risk. When looking at systemic risk it is not enough to look at an individual financial institution, instead it’s necessary to see how their actions in aggregate contribute to financial instability.
The supervisor therefore gets position-level information from at least the largest hedge funds around. Suppose each hedge fund operates at the edge of the technology. The poor supervisor therefore has to go beyond the edge.
Ill-conceived disclosure regimes provide little or no information about financial stability, but have the downside of transferring responsibility or moral hazard to the supervisor.
The folly of narrow banking
Generally, policy makers and commentators have done a good job of avoiding the mistakes of the Great Depression, but when discussing the scope of banking many seem hell-bent on repeating them.
In the Great Depression, countries with narrow banking, such as the US, saw significant parts of their banking systems collapse, suffering from all the costs narrow banking entails. Canada, just to name one counter-example, experienced no banking failures. Its banks were comfortably universal and have remained so to this day.
Narrow banks, which are supposed to be safer, provide regular banking services - utility banking - whilst the casino banks take risks. Such distinctions are arbitrary and losses can occur everywhere. Narrow banks are inevitably less diversified, less stable, and less resistant to a crisis. Splitting banks up along business lines would be a mistake.
Regulations and the crisis
The crisis has raised many challenges in the area of financial regulations. It is, of course, clear that regulations failed. However, the crisis did not happen because of a lack of regulations.
It happened because financial institutions and the whole economy used seemingly infinite amounts of cheap credit to create an asset price bubble. The banks played their part by creating all these complex structured products that are causing the difficulties. The root causes of the crisis are typical.
Banks lend to increasingly marginal credits, asset values are increasingly out of touch with the underlying economy, and it takes increasingly little to burst the bubble. When that happens everything reverses but at much higher-speed.
This is how most financial crisis has played out throughout history. We can try to prevent the same exact things from happening in the future, but surely the next crisis will take different forms, something completely unforeseen. You cannot regulate against unforeseen events.
However, there is a real cost to regulating. The only way to completely prevent financial crisis is to live in places without financial systems – places like North Korea or Cuba.
A better course of action would be to study what went wrong and then in a few years carefully change regulations at a time when we know more.
There is no hurry – we still haven't solved this crisis, and the next one will not come immediately after the end of this one. The costs of inappropriate regulations are high and we have time to wait.
Unfortunately, the “regulate everything that moves” crowd surely will get their way and we will all be more than sorry for it.
Danielsson, J. (2009). The myth of the riskometer. VoxEU.org.
Danielsson, J. and Zigrand, J.-P. (2008). Equilibrium asset pricing with systemic risk. Economic Theory, 35:293–319.
Knight, Frank, 1921, Risk, Uncertainty and Profit.