The research finds that these banks reduced the riskiness of their lending after the change (compared with a control group of other banks who didn't have a guarantee in the first place). This is what you would expect from conventional theory.
But I wonder whether their conclusion is correct:
"The results suggest that public guarantees may be associated with substantial moral hazard effects."An alternative view: perhaps banks without guarantees take less risk than is socially optimal, and public guarantees partly correct for this effect. There are several reasons why this might be the case:
- Information asymmetry - the "market for lemons" argument. Businesses (and consumers) borrowing money have much more information about their own situation than does a bank. Thus there is an incentive for non-creditworthy borrowers to borrow more than they should, and banks may respond to this incentive by imposing restrictive conditions on all borrowers.
- Herding of lenders. Partly as a result of point 1, banks are likely to identify groups of similar borrowers, with similar situations that are more easily measurable and comparable with other banks. Thus people in some types of job can borrow much more easily; businesses in specific, established sectors find it easier to borrow than those in new ones; and banks in 2005-2007 became much more likely to invest in mortgage-backed securities than other types of securitised asset. Thus, certain kinds of lending are overprovided by the market and others are underprovided, because of a higher estimated level of risk.
- Lending has positive externalities. Lending to consumers provides profits to makers of consumer goods, which are not captured by the consumer (and therefore cannot be priced by the bank). More importantly, lending to businesses enables innovation in the marketplace, new product development and increased consumer surplus - much of which cannot be captured by the businesses in increased profits. Economic theory indicates that goods with positive externalities will be underproduced - and thus, we can see that banks will under-lend.
This issue pervades all policy responses to the financial crisis. It's important not just to cut the total amount of lending or "total risk" (as if there were such a thing). What we really want to do is to stop those risks which are specifically problematic. These come in four main categories:
- risks associated with moral hazard (where I get the benefit but you pay for the losses)
- as a special case of moral hazard, those which are driven by agency problems (the employees have the upside but the shareholders - or taxpayers - have the downside)
- those risks which arose from misunderstanding of how probabilities in complex systems really work (such as the fat tails of loss distributions)
- and those which arise from, or are amplified by, bad coordination across the system (for instance, banks overinvesting in mortgage backed securities because of network effects)
But while doing this, we must consider how we can better distinguish between different risks, to make sure we don't stop people from taking risks which are socially useful, as an unwanted side-effect of preventing the bad ones.