The Obama Fiscal Boost: A Note
Hoisted from the Archives: Chiang Kai-Shek and the Bush Family

Robert Waldmann on the "Fundamental Values" of Risky, Long-Duration Securities

Robert writes:

Comment on DeLong: What Has Happened to Milton Friedman's Chicago School? DRAFT: What does "impaction" as in "the impaction of information" mean ? Is it like an impact or an impacted wisdom tooth ? I seek wisdom not teeth. I see a fifth cause of the decline in asset values. I think many assets were over-valued in the past, because the ratings agencies were tricked or cashing in on their late lamented excellent reputations (or both). The loss of confidence in said agencies causes an increase in estimated risk not because risk has increased or risk aversion has increased but because the old estimates are now known to be bogus. More generally, risk modelling by traders was similarly complete nonsense. I don't know to what extent the traders were tricked and to what extent they were in on the scam (nor I suspect do they).

Structured finance created a huge illusion of wealth by creating an illusion of safety. The financial engineers knew how the agencies rated (the agencies explained for a consulting fee) and how traders estimate "value at risk". They knew people assumed (or pretended to assume) that all stochastic variables are normally distributed. Thus it was profitable to sell instruments with skewed returns (fat lower tails) unless the rare negative event occured during the testing period (in which case the instruments could be re-engineered). A senior tranche of a pool of moderately risky assets has a skewed distribution. Similarly money could be made by reducing own variance for a given beta by pooling assets and issuing claims on the pool. The variance of an average is less than the average variance. The covariance of an average and the market portfolio is the average covariance. Thus a claim on a pool of BBB rated corporate bonds was rated AAA. Turning BBB to AAA is worth a lot of money except for the fact that it was a scam (investors could pool themselves -- they don't seem to have understood that pre-pooling reduces the benefit to them of their own pooling -- or they were in on the scam).

The risk of a nationwide decline in house prices was estiamted at 0 by S&P (I am not exaggerating). Now there has been such a decline and they must admit that there is the risk of such a decline in the future. The current decline would cost 1 trillion. The future possible declines also cost money. So much of the wealth was an illusion which won't come back soon. This end of systematic miss-estimation of risk is not on your list.

Also institutions took huge gigantic bets against each other (as in AIG lost writing CDSs). This increases counterparty risk. No one knows if a counterparty is solvent. That reduces the value of a huge number of instruments. The damage could have been done without involving the housing industry or the stock market if, say, investment bank CEOs played a really high stakes poker game and all claimed to have won money. Also bankruptcy is costly. Even if the CEOs had played hundred billion ante poker on camera, wealth would have been destroyed and more wealth would have been shifted from investors to lawyers. This is another item not on your list. Finally, much of the strange new finance was designed to help agents avoid prudential rules and regulations which they considered to be costly. Now they have learned two things. First that the regulations weren't so pointless so they will have to pay that cost to avoid bankruptcy. Second they will be audited by banking regulators, trustees etc and found wanting. This last point is semi redundant as it amounts to an increase in perceived risk. However, it explains why I keep speculating that this that or the other operator was in on the scam.

Here Robert Waldmann raises a whole bunch of issues that I was attempting to defer to later. I claimed (correctly) that the collapse of global financial asset values over the past year and a half was overwhelmingly do to increases in the risk and information discounts, and not to changes in expected future cash flows. Robert raises the point that perhaps the current risk and information discounts are "right," or in some sense the appropriate "fundamental values," and that the previous low values of these discounts were in some sense "wrong" or "irrationally exuberant."

To make sense of this claim we need a definition of what "fundamental values" are.

Here is mine:

The fundamental values of asset prices are the money-metric values that the costate variables associated with the commodities would have in some reasonable utilitarian central-planning social-welfare-maximization exercise under reasonable utilitarian preferences.

This entails the following, leaving what happens in Mad-Max scenarios completely out of the picture:

  1. Default discounts should be small (and in fact default discounts are small even now in the recession)--default is not a thing that should happen often to financial assets and should not be a big component of asset pricing in well-functioning asset markets that mimic some reasonable social-welfare-maximization calculation.
  2. Information discounts should be small because they reflect knowledge held by one part of the system and not by another--and well-functioning asset markets should be liquid and full enough of information that they should be able to mimic the low information discounts found in some reasonable social-welfare-maximization calculation.
  3. The proper duration discount--the safe real interest rate--in a social-welfare-maximization calculation is the utility cost of moving wealth ahead in time from one consumer to another or from early to late in a consumer's life. This should be governed by the rate of labor productivity growth divided by a reasonable utilitarian intertemporal elasticity of substitution. A reasonable utilitarian intertemporal elasticity of substitution is one. A reasonable estimate of the rate of productivity growth is 2 percent per year. The duration discount ought to be a real safe interest rate of 2 percent per year.
  4. The proper risk discount is governed by the utilitarian diminishing marginal utility of wealth and the covariance of asset returns with wealth: it is the danger that when assets lose value you find yourselves in states of the world in which your marginal utility of wealth is high. The covariance of asset returns with wealth is low. A reasonable utilitarian diminishing marginal utility of wealth corresponds to a coefficient of relative risk aversion of 1. The risk discount for an equity beta = 1 portfolio ought to be on other order of a long-run real rate of 0.3% per year. The risk discount for financial assets as a whole ought to be on the order of a long-run real rate of 0.2% per year.

Thus my view of fundamental values: financial assets ought to be priced so that safe assets yield a 2% per year real return, and so that financial assets as a whole ought to yield an expected long-run return of 2.2% per year (with equities yielding an expected return of 2.3% per year).

Right now it looks to me as if safe assets yield a real return of -1% per year, and financial assets as a whole yield an expected long-run real return of 5% per year (with equities yielding an expected long-run rate of return of 8% per year). At the peak of the housing bubble as best as I could tell safe assets yielded a real rate of return of 2% per year (about at their fundamental values) and equities yielded an expected long-run real rate of return of 4% per year (about half their fundamental value).

The most important thing about asset prices, I think, is that even in booms and bubbles they don't reach their "fundamental" values. The first-order fact about asset markets is that they currently do and historically have done a really lousy job of mobilizing the long-run risk-bearing capacity of the global economy. The risk tolerance of those who participate in financial markets is a remarkably low fraction of the true fundamental global risk tolerance, the horizon of those who participate in financial markets is a remarkably low fraction of the true social investment-planning horizon, and in addition the views of those who participate in financial markets are extraordinarily more volatile than can be generated by the variation in rational expectations of future growth rates and appropriate fundametnal discounts.