Friday, October 03, 2008

CR confusing about rationality.

Calculated Risk attempts economic theory without the assumption of rational expectations. It is not his fault that the result is confusing.

The fact is that no one seriously believes that economic agents have rational expectations (one economist once told me he did but I think he was joking). Economists also know that violations of rational expectations which aren't gross and obvious can make a huge difference (A game theorist told me this in a lecture in 1978).

However, economists just can't do without the assumption. Otherwise anyone can argue anything.

Here CR totally demolishes an argument by Glenn Hubbard and Chris Mayer that What is to be done is to lower interest rates to drive up home prices. The excerpted argument is

We are in a vicious cycle: falling housing values cause losses on securities, which reduce bank capital, thereby tightening lending and causing house prices to fall further.

I will discuss the argument of Glenn Hubbard and Chris Mayer as excerpted by CR (GHCMAECR).

In response CR first asserts that agents sure aren't rational "First, house prices are falling because prices are too high when compared with fundamentals like incomes and rents." , "analysts are finally getting realistic on their house price projections" and "it is important for a healthy housing market to allow prices to return to more fundamental levels (and that means further price declines and/or increases in household incomes)" CR is sure the prices were wrong 2 years ago and are still too high.

On the other hand, when discusses the price of mortgage based securities CR assumes that agents are rational "Second - and this is important to understand - the value of the securities is based on projections of future house prices, not on current house prices."

The argument is First that if financial distress is causing a temporary decline in house prices, it would be irrational to extrapolate the temporary decline to low house prices in the future. The value of mortgage based securities depends mostly on future house prices because house prices matter when mortgage initiators foreclose or when debtors try to refinance. Most foreclosures of houses whose mortgages are currently securitized will take place years from now (even in expected present value and even if the foreclosure rate will be extraordinarily high in the near future). The second part of the argument is that financial operators are rational and, therefore, if something is causing a temporary decline in house prices it won't cause them to incorrectly predict a permanent decline.

Basically the argument is First we know that financial operators are totally irrational Second if we assume that they are rational we conclude.

One could as well claim that they were rational before and are now irrationally panicking. I absolutely agree with CR, but the argument beginning "second" is only convincing if one assumes that financial operators are rational. If they irrationally extrapolate recent trends, then events which would, if people were rational cause only a temporary downturn in house prices can cause a vicious circle.

Now as to the solution Again we get First agents are Irrational and Second If we assume agents are rational then.

"First, it is important for a healthy housing market to allow prices to return to more fundamental levels" Agents aren't rational so prices can differ from fundamental values. Indeed house prices are above their fundamental value.


Second, this shows a misunderstanding of the role of interest rates with regards to house prices. This gets complicated, but if the interest rate is artificially low today, the buyer can expect rates to rise - and therefore that the home price will not be as high in the future (all else being equal). The buyer should discount this lower house price back to the present, and we discover that interest rate changes only play small role in house prices.

That is Glenn Hubbard doesn't understand how interest rates should affect house prices, but it is reasonable to assume home buyers and sellers understand perfectly.
Ordinary people understand that if interest rates are low now they will rise (including the people who didn't understand that their mortgage would reset).

This is standard practice for economic theory. Economists are perfectly willing to say that other economists do not understand the implications of economic theory. However, the standard assumption is that everyone else understands perfectly. Thus the standard assumption among economists is that we are the least able to understand the economy.

The problem is very serious. If we assume that people have rational expectations we must conclude that the price is right now and it was before and ... well it's so crazy that no one would stick to the assumption when talking about policy during a crisis (there are no fresh water economists in a depression).

If we allow ourselves to hypothesize this or that kind of irrationality which sounds semi plausible, we can justify any policy, by claiming that the problem is one of irrational pessimism which will be eliminated by the proper burnt offerings or whatever so long as people believe that the burnt offering will solve the problem.

The situation is so desperate that it has driven Paul Krugman himself to become an accidental theorist. When linking to CR he presents a graph. The implicit theory is that price to rent ratios have a normal level and will decline if they are above that normal level. This is a hypothesist, but it is presented without any formal model or testing of competing hypotheses.

Now I agree with PK and CR, but my argument is "it's obvious".


Actually, now that I think of it, I'm not even sure that the complicated argument is valid as a statement in economic theory. If I have a 30 year fixed rate mortgage and am sure that I (and my heirs) will never move or want to sell my house, I really shouldn't care about future interest rates. If home buyers were of this type, future interest rates wouldn't matter. They matter if home buyers have some kind of floating rate mortgage, or if there is a significant probability (in expected discounted value) that they will sell the house. Standard simple asset pricing models assume that people can make infinitesimal changes to their holdings (sell one brick of the house) and that there are no transactions costs. These assumptions make no sense for house prices.

Now I think the asset pricing models have realistic implications (including that there can be bubbles if people are irrational in a way that isn't grossly obvious) because many many homebuyers actually made the crazy assumptions typical in simple asset pricing models, that is, they acted as if they could sell part of their house by taking out a home equity loan and they decided that the transactions costs were negligible. Then they assumed that house prices would go up forever and interest rates would stay low forever (neither grossly obviously irrational as predictions are hard especially about the future).

No comments: