But she won't be surprised. Photographer: Andrew Harrer/Bloomberg
But she won't be surprised. Photographer: Andrew Harrer/Bloomberg

Here is a pretty amusing New York Fed blog post about bank liquidity regulation. Regulators worry a lot about bank liquidity. What that means, loosely speaking, is:

  • Banks are basically in the business of borrowing money short-term (deposits, repo) to lend it long-term (mortgages, corporate loans).
  • That's a scary business: A bank that has only short-term funding (all its money can disappear tomorrow) and only long-term assets (it can't turn any of its assets into money for years) will get into trouble very easily.
  • So you want some amount of long-term funding, and some amount of liquid (easily convertible into cash) assets, to feel safe.

Nobody disagrees with this, but there is some debate about how much of each thing is required: Long-term funding costs more than short-term funding, and liquid assets have lower returns than illiquid assets, so banks tend to argue that high liquidity requirements will cost them money and limit lending and all that bad stuff.

The New York Fed's Dong Beom Choi and Lily Zhou find some evidence against that argument:

[W]e ask if holding liquidity is costly. If so, then returns should be lower for more liquid banks. However, this does not seem to be the case. We see no correlation between realized return on equity, our measure of profitability, and LSR.

Where "LSR" is the "liquidity stress ratio," a simple measure of liquidity (higher LSR = more liquidity risk, lower LSR = safer). The chart:1

Source: Liberty Street Economics
Source: Liberty Street Economics

Hmm. So liquidity is expensive, and yet it doesn't seem to cost banks anything. Why not? Well, here are some ideas:

  • "LSR is positively correlated with Tier 1 common capital ratios after the crisis." Remember, a higher LSR means more liquidity risk, so this means that that better-capitalized banks tend to take more liquidity risk, and that banks with better liquidity tend to have less capital. "[T]he positive relationship may indicate that when banks are less vulnerable to undercapitalization, they take more liquidity risk. In other words, capital and liquidity may act as substitutes."2
  • "Next, we look at the correlation between credit risk-taking, as measured by a firm’s percentage of risk-weighted assets to total assets, and liquidity risk-taking. The LSR and credit risk are negatively correlated, suggesting that the two risks may be substitutes."

This is a blog post, not a rigorous study,3 but it has the bones of a theory: Banks are subject to stricter capital regulation, and they make up for the cost of having to have more capital by running more liquidity risk. Then they're subject to stricter liquidity regulation, and they make up for the cost of having to have more liquidity by running more credit risk. If they're subjected to stricter credit risk regulation, they'll make up for the cost of having to run less credit risk by running more ... some other risk, probably. Also various vice versas.

If you're a bank, you can make money by borrowing really short and lending really long. If you can't do that any more, you can borrow slightly longer and lend slightly shorter, but to riskier borrowers. If you can't do that, you'll find something else. Risk expands to fill the banks' appetite for risk, which seems to be exogenous to regulation.4 Regulation mostly shifts the flavor of risk from the risk regulators are paying attention to this week to some other, subtler risk.

This is a cynical counsel of despair but it's fun to see the empirical data for it laid out by the New York Fed.

Though, I mean, it's not strictly a counsel of despair.5 It's also an argument for paying attention to the interconnections between bank risks. And in fact, the Fed wants to do just that, for instance in Fed governor Daniel Tarullo's advocacy for, and Fed chairwoman Janet Yellen's apparent embrace of, linked capital and liquidity rules that would require more capital at banks that take more short-term funding risk.

Still, there remains a popular view that bank regulation should be simple and single-pronged. There is an intuitive aesthetic appeal to the idea that complicated problems should have simple solutions: Banks are big and complicated? Shrink them! Or impose a 15 percent simple leverage ratio requirement! Or bring back Glass-Steagall! Or ban bonuses! Or whatever.

I tend to be a counsel-of-despair guy on those proposals, whether alone or casually stacked on top of each other. The way to regulate a complex interconnected system is to try to understand and compensate for the interconnections. Focusing on fixing one thing at a time just pushes risks somewhere else. As the Fed itself would be happy to tell you.

1 This is just a point in time, and just for the 18 banks subject to the Fed's stress tests. But they also run it over time and it's a consistently underwhelming relationship.

2 But "this relationship is only observed after the crisis": Pre-2008, capital and liquidity didn't seem to have much relationship one way or the other.

3 And in particular, the things it measures -- LSR for liquidity risk, risk-weighted asset ratios for credit risk -- are pretty blunt instruments.

4 Though some varieties of regulation -- compensation and clawback rules, say -- are designed to target bankers' risk appetites directly.

5 Another way you could go here would be "it's an argument that bankers know better about the risk tolerances of a bank than regulators do," though that is not intuitively obvious. The bankers got it pretty wrong in 2007! So did the regulators, though, I mean.

To contact the writer of this article: Matt Levine at mlevine51@bloomberg.net.

To contact the editor responsible for this article: Zara Kessler at zkessler@bloomberg.net.