Capital regulation: it’s complicated!

Many people criticize the way in which bank capital regulation is done. They know that banks can and do game complicated regulatory rules, a form of regulatory arbitrage. One focus of their criticism is risk weighting – the idea that banks should hold capital commensurate with the riskiness of their assets. The more risky the loans and securities a bank holds, the bigger the capital buffers should be to ensure that banks and the banking system are robust.

On its face, risk-weighted capital requirements sound pretty sensible. But like everything in life, the devil is in the details. Prime among them is how the riskiness of individual assets, or portfolios, is assessed. The key issue is whether the estimated riskiness accurately reflects the asset’s true risk.

For globally active institutions, the answer has been to use the estimates that come from the bank’s own internal models – the models that are presumably also used for internal firm risk management. Yet, whenever markets assess an asset’s risk differently from the bank’s model or from the measure embedded in the regulatory rules, the incentive for regulatory arbitrage emerges. Such incentives are virtually certain to arise over time, and can be difficult for regulators to identify and counteract before they lead to systemic threats.

Critics view the experience with risk-weighting as an unvarnished failure. For example, Citigroup had risk-weighted capital in the 10% range at the end of 2007, but by many accounts was insolvent less than a year later. Even more damning is the recent work by the Basel Committee on Banking Supervision, which shows banks’ risk-weighted calculations of the same hypothetical portfolios differ by up to 50%! That is, one bank’s model might tell it that it only needed two-thirds the capital to support the same exact assets as some other bank’s model. This is an unacceptable amount of variation.

Since risk-weighting is so fraught, why not just throw it out in favor of a simple leverage ratio? That is, simply monitor the ratio of capital to total assets, rather than to risk-weighted assets. And, importantly, set the regulatory capital minimum at a high enough level that it provides a sufficient buffer even in the event that banks take on more risk – something they would be expected to do in the absence of the sort of penalty that risk-weighting is supposed to create.

Unfortunately, leverage ratios are not so simple. First there are the accounting standards. For global banks that engage in large amounts of derivatives trading, “simple” leverage ratios can vary by a factor of two or more depending on whether the accounting used follows US GAAP (Generally Accepted Accounting Standards) or IFRS (International Financial Reporting Standards). For example, in 2009, Deutsche Bank reported leverage of 50 under IFRS and 25 under US GAAP. (Note: their Tier 1 regulatory capital ratio at the time was 11%!) The primary reason for these differences is numbingly complex, having to do with something called derivatives collateral netting. But the point is that depending on how you measure things, in the case of what is one of the five biggest banks in the world, capital was either 2% or 4% of assets.

But that’s just one reason leverage ratios are not simple. What about lines of credit that are guaranteed but not drawn? Many firms have these arrangements with banks: they are the corporate version of overdraft protection or of a credit card that hasn’t reached its limit. Economically, such lines of credit are analogous to checking accounts: they provide the customer with funds on demand. For a bank, they are contingent assets. Should unused lines of credit be counted when measuring “simple leverage” because they could be drawn at any moment? Should they be discounted by some factor until they are drawn? If so, how should that discount be set? Should it change with economic and financial conditions? What about backup lines of credit for commercial paper? What about trade credit guarantees?

The list of off-balance-sheet activities of banks goes on and on. In computing risk-weighted assets, the Basel Committee has set standards, and the national regulators have made (or will make) rules. Setting leverage ratios will require many of the same judgments. The point is that they are not so straightforward. [If you want to see the current state of the standards for the leverage ratio, including the approach to assessing off-balance-sheet exposures, have a look here.]

What to do? The Basel Committee’s solution is to use both risk-weighting and a leverage ratio. The idea is that combining these mechanisms reduces the risk of catastrophe for two reasons. First, there will be times when markets assess risk far higher than the risk weights prescribed by regulators. Here, the leverage ratio will serve as a backstop. Second, even when the market and regulatory assessments are aligned, things that looked safe ex ante can turn out to be risky ex post.

This approach assumes that bankers and regulators will ensure the reliability of internal models used to assess risk. One possibility is to use hypothetical portfolio comparisons across banks to ensure consistency of capital buffers.

Of course, the most important question remains unanswered: How much capital is needed by either standard to contain systemic risk? That issue is unlikely to be settled soon, in theory or in practice. Yet, the experience of the 2007-2009 crisis and recession makes us sympathetic in practice to the calls for higher capital requirements on both measures.

It would be nice to have something simple to replace risk weights. Unfortunately, leverage ratios are not simple because banks are complicated. And, since banks are complicated, so is capital regulation.