One of the key metrics of how likely a set of investments is to go bust is the amount of leverage involved. If you buy a house with a 20 percent down payment, then even a largish decline in the value of the home will leave you with more assets than debt. If you buy a house with a no money down loan, by contrast, even a tiny decline in the value of the house will leave you underwater.
And so it is with banks. The more leverage they use to make their investments—be those investments simple loans or complicated trades—the more likely a bad bet is to bring the house down. So a natural question to ask about a bank is how leveraged it is. Unfortunately, the way banks like to answer this question is to obfuscate in terms of core tier one equity divided by risk-weighted assets. As Brooke Masters writes for the FT, when you strip this layer of abstraction away most major banks look a good deal riskier:
The disclosed leverage ratios strip out the effect of risk modelling, which has the effect of more than doubling the size of every bank’s balance sheet and more than tripling that of Barclays. These numbers partly reflect the size of low-risk, high volume businesses. But they will be a boon to critics who think some banks have been tweaking their models to cut their capital requirements.
The reported leverage ratio also uses the tighter definitions of capital that will be required in 2018 under the Basel III reform package, despite hopes by some banks that they will be allowed to boost their ratios by counting some old capital that doesn’t quite fit the new definitions. This unvarnished look at each bank’s borrowing produces far different results. As of December 31, Standard Chartered had the strongest ratio at 4.5 per cent, meaning that it has assets equal to 22 times its capital. Barclays has the weakest, with 2.8 per cent. That means Barclays had assets worth more than 35 times its capital base.
So which way is right? I think the way that strips out the risk modeling is clearly the correct way to measure bank capital. The reason is that with any given investment the sensible thing to do is ask two separate questions about risk. One is simply "how risky is this investment or portfolio of investments?" That's where risk modeling comes in. The second is "how risky is the capital structure?" In other words, how much leverage has been used. These are both important questions but they're separate questions. The use of risk-weighting mixes them up. The relevance of the leverage ratio is that a highly leveraged bank can be forced into insolvency by even a small loss on its investment portfolio. Risk modeling is great, but not only can the system be rigged obviously risk models can simply be mistaken. The proper role of regulating leverage is to ensure that bank failures are very unlikely even if the investments go bad.
Here's me endorsing a call for much tougher leverage regulations than are required under Basel III. Unfortunately, as Masters writes at the moment the big political momentum in most countries is around banks trying to gut the already-too-weak Basel III rules.