Risk and Regulation

I thought I’d elaborate a bit on my post from yesterday about Basel III. I’m not sure I got across just how fundamentally problematic I think the Basel approach is.

The overall trend in the financial industry for a couple of decades now has been towards a “scientific” approach to risk-management. The idea is, rather than relying on old-fashioned rules of thumb, you actually try to measure the risk one is taking by holding a certain portfolio. Then, once you know what your risk is, you know how far wrong things can plausibly go, and you know how much you can borrow (or someone else can lend) against your position.

Of course, measuring risk is a tricky thing, because you’re trying to predict future events and all you have is historical performance data. And, as the old saying goes, past performance is no guarantee of future results. Nonetheless, the risk-managers have tried to do their best with the tools they have: historic price data, default rates, and correlations. And they’ve developed a collection of tools like value-at-risk and stress-tests designed to probe both likely outcomes and 2- or 3-sigma possible events.

All of this data does give real information. I don’t want to belittle the utility of knowing the historic correlations between the risks you’re taking, or how much money you would lose if there’s a two-sigma negative move across all asset classes. But precisely because you now feel like you know what your risks are, you’re going to be inclined to hedge them, until you get your risk position down to something de minimis. Indeed, if you’re a bank, you’ll start to think of yourself as a financial intermediary whose job isn’t so much to take risk as to serve as a kind of risk-middleman. And if you’re a conventional asset manager, you’ll start to think of your job as compiling an asset pool with a risk-profile designed to match some benchmark. And so forth.

Everybody still has to make money, though. And since taking any additional measurable risk is now stigmatized, the game becomes how to increase returns without increasing measurable risk. And that, predictably enough, means that more and more money piles into products with risks that either cannot yet be measured or, even worse, that the financial world is not aware exist.

Developments in banking regulation in the last decade, meanwhile, have turbocharged this process, and I’m increasingly convinced contributed mightily to the financial crisis. At the heart of the financial crisis, after all, was banks investing in highly-rated debt backed by lousy mortgages. But why did they hold so much of this debt? In part because they could plausibly argue that it was risk-free or nearly risk-free. Double- and triple-A-rated debt had very low risk-weighting. Similarly, anything hedged with a double- or triple-A-rated financial counterparty had a very low risk-weighting. Use of an internal ratings-based approach (encouraged under Basel II) could reduce capital requirements even below the standard approach carried over from Basel I. If the exposure was classified as market risk rather than credit risk (which it might be, depending on the bank’s internal organization and the structure in question), again the Basel II framework was based on Value-at-Risk, which is based on historic performance data – which in the case of highly-rated tranches on mortgage deals showed very low volatility (because there hadn’t been a crisis since the product took off).

The big-picture point to take home is: the regulatory framework recommended by Basel II assumes that banks are in the best position to measure their own risks, and that a regulatory framework that aligns regulatory capital requirements with the risk being taken is to be desired. In other words, the regulatory framework was pushing banks hard in the direction they were already going: towards avoiding measurable risks and hence (since you still have to make money) into risks you can’t easily measure (or don’t know exist).

Basel III retains this basic framework, but increases (somewhat) the amount of capital that banks need to hold generally. Therefore, it should further increase the incentive to avoid measurable risks and to hold positions whose risks are not well-measured.

The boys in Basel are undoubtedly correct that regulators are not going to be better at measuring risk than the banks themselves. So the question isn’t whether there is a better way to measure risk – the question is whether we want to incentivize banks to hold portfolios that have little apparent risk (but may have substantial unmeasured risks) or whether we want to incentivize banks to hold portfolios with meaningful measured risks, and to hold appropriate capital against those risks, so that when the unmeasured risks do rear their ugly heads they don’t represent a multi-sigma event on the bank’s portfolio. One way to achieve that would simply be to increase the minimum risk-weight. This would discourage banks from holding large portfolios of government debt (which may be one reason that Basel didn’t make that recommendation) and other highly-rated, apparently low-risk securities. Instead, they’d have to get back into the business of making risky loans – a business that, in the last decade, was increasingly ceded to the unregulated hedge fund community.

I don’t want to make it sound like this is an easy call to make. But it is striking that after a crisis driven by huge bank exposures to a largely unmeasured risks, Basel III doesn’t appear to have questioned the assumption that measured risks are all we have to worry about.