The banking system is a machine to transform risk: people put their money into a bunch of risk-free-ish-or-so-they-think banks, and those banks lend that money to risky businesses, and the banks make money on their ability to price that risk appropriately and/or on their ability to get a government bailout when they price it inappropriately. From this description you can rough out some boundary cases – a bank that loaned money only to risk-free businesses wouldn’t make very much money or serve very much purpose, while a bank that was a massively levered conduit for moving depositor money into Ponzi schemes would also not serve much social purpose – but that leaves a broad middle ground where banks need to evaluate risk carefully enough to avoid blowing up but not so carefully that they constrain promising but risky investment.
But that middle ground is where the action is so you get papers like this one, from the Bank for International Settlements, about one flavor of lending and two flavors of banks. The lending is syndicated lending, and the banks are “bailed out banks” and “not bailed out banks,” and here are some suggestive charts:
The restrained conclusion in the BIS paper is “Although the riskiness of loan signings started diminishing across the board in 2009, we do not find consistent evidence that rescued banks reduced their risk relatively more than non rescued banks during the crisis.” And in particular, in 2010, banks that had been bailed out still had more levered, higher-yield syndicated loans than banks that had avoided a bailout.
Which is I guess embarrassing for the governments that bailed them out only to allow them to continue gambling crazily? Maybe. The Fed already ran a similar exercise and got a similar result for U.S. banks, but was pretty blasé about it, pointing out:
This may reflect the conflicting influences of government ownership on bank behavior. Although TARP money was given to increase bank stability and reduce incentives to take excessive risks, it was also given with the understanding that the funds would be used to expand lending during a period of increased risk. These two objectives have an opposing influence on bank risk-taking ….
And that seems pretty applicable here: if you get a bailout from your government, that government would prefer that you become solvent, but it would also prefer that you keep lending to borrowers to keep your country’s economy humming, and there’s no a priori reason to think that one desire would dominate the other. Ponder, if you will, Ed DeMarco.
The other obvious thing to take from those charts is: before the crisis, bailed out banks were riskier than non-bailed-out banks. That’s pretty intuitive – if you make riskier loans when times are good, you end up in a worse place when they turn bad all of a sudden. But there are some fun counterintuitive footnotes. Like:
Despite being riskier*, the banks that later got bailed out traded in 2001-2006 as though they were a bit safer than the ones that later didn’t. One lazy tendentious reading of that is that markets knew which banks were too big to fail and priced their credit risk accordingly: if you’ve got a promise of government support, you can take more risks and be more levered and still be a safer credit than someone who doesn’t. If “a bank is any entity that can issue liabilities that are widely accepted as near-perfect substitutes for whatever trades as money despite being highly levered,” then the wide acceptance of the liabilities is doing more work than the quality of the levered assets.
But this was my favorite chart, though I didn’t really understand it:
The “portfolio pricing error” is the “Difference between the observed spread over Libor, and the spread predicted by a linear regression incorporating observable loan features (size, maturity, guarantees, collateral, facility purpose and type), borrower characteristics (sector, rating, first time borrower) and the state of the market (total volumes, level of interest rates).”** Negative means the loan was underpriced (too low a spread), positive that it was overpriced (too high a spread). The non-rescued banks were all over the map, but the rescued banks were always negative. Banks that later got bailed out didn’t just take more risk than those that didn’t, they also systematically mispriced that risk.
What does that mean? Two related possibilities come to mind. In one, some banks were just less competent than others, and the incompetent ones needed bailouts while the competent ones did not. That is an unobjectionable theory: few bailed-out banks choked primarily on syndicated corporate loans, but perhaps syndicated-loan incompetence was a symptom of firmwide problems.
The other theory starts from the intuition that few risks are more worthwhile to misprice than syndicated loans: extending credit at too-low rates is the price of admission to all sorts of investment banking business. If your business is, like, lending to businesses, then it behooves you to price those loans appropriately, since if you don’t then you might run into trouble. But if you lend to businesses only as a way to get in the door to conduct your real business of investment banking – and the things, like securitization and trading, that often come with it – then you have no problem underpricing your loans. Because on the one hand those loans aren’t your profit center: the riskier investment banking businesses are. And on the other hand, if you do get blown up – on syndicated loans or, more likely, otherwise – and you’re a big and universal and interconnected enough bank, well, then you can confidently expect a bailout.
* One one metric, the syndicated loan portfolio. This is a suggestive toy model, not a real one that can prove stuff.
** So, whatever. Pricing loans is an art not a regression etc. etc. But the aggregate data is interesting.