Sunday, October 5, 2008

Some thoughts on an interesting article on bailouts in The Economist's Voice

Note: Like most of my posts, I edited and improved this one several times after I initially placed it. I will not hesitate to do this, and without cross-outs, as my main goal is to teach clearly and well, and to help with discovery and good idea creation, not to leave a historically accurate evolution of my writing. In fact, if you think the writing of a post of mine has clunky spots or mistakes, you might want to try looking at it hours or days later. These things may be fixed, plus valuable new material may have been added.

Jonathan Carmel, a finance professor from my Alma matter, the University of Michigan, makes some important and good points in his article in The Economist's Voice , "Pitfalls of the Paulson Plan".

First, he notes that in auctioning the packages of toxic sub-prime mortgage securities, or just selling them in a market, there is a potentially very serious Lemons problem, because in spite of how they were originally marketed, they are very heterogeneous, very different in quality, yet it is hard to estimate the quality well without rare and specialized expertise, and very time consuming and expensive evaluation. Thus, a classic lemons problem. As a result of the lemons problem, and on top of it, there is a liquidity problem, and an associated liquidity discount. So asymmetric information is key. Still, Carmel argues:
After a bit of reflection, I think it becomes clear that asymmetric information should be a relatively minor issue [at this stage] in these markets. Most subprime has been securitized and is no longer held by the originator or the servicer of the subprime mortgages. The originator may have had some extra information about the characteristics of the mortgage borrowers in a certain pool. But given the originator no longer holds any stakes in the pool, this is not relevant. Both current holders of subprime mortgages and potential buyers have the same access to the original information released by the originator concerning the mortgage pool characteristics. So this cannot be a source of asymmetric information.

What about default rates? Perhaps there is asymmetric information regarding the payment history of loans in the pool. Perhaps this is the source of asymmetric information. But mortgage servicers keep track of the cash flows paid by the mortgages it services. This payment history is available to all potential buyers. Current mortgage holders have no additional information about payment rates beyond that known by the mortgage servicer. So this cannot be a source of asymmetric information either. The only potential asymmetric information issue involves default modeling. There was a time when the proprietary default models of some investment banks and “quant” desks were thought to give them an edge in using this common data set to predict default. Perhaps there is some truth to this. But the groups with the most well-regarded default models tended to be the ones with the largest subprime holdings and some of the largest losses. The current default experience has been sufficiently unlike any past time period that even the best of these proprietary default models appears to provide little edge in valuing these mortgages.
Carmel does note, "The originator may have had some extra information about the characteristics of the mortgage borrowers in a certain pool.", but then he quickly moves on with, "But given the originator no longer holds any stakes in the pool, this is not relevant." Maybe it's not relevant for the specific point Carmel wants to make there, but it is potentially important in addressing this crisis and preventing another one, because this is potentially one group that does have much better – and asymmetric – information about sub-prime and other mortgages, these local local banks and other providers that originate the mortgage.

Local banks and providers know the local real estate market, local economy, the particular states laws and regulations, and even personal details about the borrower that can be very relevant and aren't in the applications. Also, when the issuer of the mortgage is also the holder of it, it eliminates a principle-agent problem, which as we've seen can be very serious (liar-loans). This is a very important part of the problem that I haven't been hearing about. The principles -- buyers and re-buyers of these packages are highly disconnected from the agents -- the issuers. This allowed the agents to pull all sorts of very profitable (for them) shenanigans ans fraud, which made the principles think these packages of mortgages were worth much more than they actually were, leading to great inefficiency and a crisis.

The local bank or provider can have a big advantage – or asymmetry – in information over a large national finance house. As a personal finance expert, I have studied extensively mortgages from the perspective of an individual or family, how they can get the best deal, and often it is through a local provider, precisely because of their greater local knowledge. It's something we should really think about when deciding if we should keep moving away from (or move at least to some extent back towards) the old model of local banks and S&Ls, or other issuers, providing the bulk of mortgages, at least for the trickier and more unique (heterogeneous) non-premium types.

Such local knowledge also gives local providers an advantage in servicing and salvaging the mortgage. It's really analagous to the finding in finance that for very unique corporate lending situations, the corporation's bank can usually do the financing more efficiently. It knows the corporation, and often similar corporations, well, and it's close relationship makes it easier to get information and cooperation.

With regard to the lemons problem, local banks and other providers may be like mechanics specializing in that type of car. There is far less of a lemons problem for them, or essentially no lemons problem. They can evaluate the car pretty well, and at a reasonable evaluation cost.

Carmel ends acting as though doing any government bailout would endanger the good deregulation, and not be worth the cost, even if it avoided a recession (or a worse recession than we already have). Some of the de-regulation was, in fact, good, but much of it wasn't, and, despite what Carmel seems to imply, it's certainly possible to devise a highly effective bailout plan – for example like what the University of Chicago's John Cochrane describes – and still keep the good de-regulation, while adding only new regulation that's good – and very needed. Whether such a plan can be proposed and passed politically, either before or after the election, however, is another question.