Wednesday, March 03, 2004

The post just below got a link from Brad
and so was visited by over 100 people (a lot for this blog). In case any come back for more, I will explain a little.

First I agree with Brad that the idea that the treasury can make a killing playing the stock market is separate from the issue of saving social security. Social security shortfalls have a present value in the trillions, so making a killing would help the social security administration, but one could as well
present the multi trillion dollar almost sure thing as a plan to fund socialized medicine or the expansion of the US empire or whatever.

The question is can the treasury make a lot of money without hurting private agents by selling a few more hundreds of billions of bonds and buying stock ?

Before getting (again) to the problems, I want to explain why this might be a good idea, or, more exactly, would have been a good idea if done in 1945. First there is the fact that in the postwar period, at least until the exuberant 90s, stock was a great buy for a huge deeper than deep pockets investor like the US treasury. The average return on stock was much higher than the average return on treasury bills. This is (or was) presumably a risk premium. Oddly if one looked only at aggregate data on market returns and consumption, it is (was) almost impossible to explain this risk premium (as pointed out by Mehra and Prescott). They worked a while ago when the Prescott et al approach to macroeconomics was to do microeconomics with a representative consumer.

The idea of a representative consumer is to assume that we are all equal so each has wealth income and consumption etc equal to average wealth income consumption etc.. Then assume we are rational and fiugre out what would happen. Such models are also used to guide policy by people who don't care at all about income distribution (including Prescott).

In a rational representative consumer model, assets should earn a risk premium if their return is positively correlated with average consumption. The idea is that a high average return isn't so great if it is especially high when we consume a lot and are not hungry and low when we consume little and are hungry. Oh the market return is the return on an index fund. It should be the return on an index fund of all listed stocks.

Oddly the market return is almost perfectly uncorrelated with consumption ?!? This is very odd and implies that, unless the representative consumer is incredibly risk averse, stocks were (are) under-priced. That the average person would be better off buying more stock. This is also true for someone with a long time horizon. The return on stock over *all* long periods was better than the return on treasury bonds.

The reader (if any) will have noticed many parentheses. The reason is that, while it seems clear that the market was under-priced from 1945 to say 1980 at least, it is not at all clear that it is still under-priced. In fact, part of the 90s bull market might have been a correction of this mistake.

OK so why is (was) the stock market undervalued ?
1) One possible reason is that people systematically underestimate(d) the returns on stock, that is, that they are (were) systematically irrationally pessimistic.
2) Another possibility is that people don't (didn't) know to invest. The risk on a portfolio of only one or a few stocks is much greater than the risk on a diversified market portfolio. Indeed with time people began investing in index funds and the aggregate price earnings ratio rose.
3) the representative stock owner is different from the representative consumer. In the period studied by Mehra and Prescott almost all of the stock market risk was born by very few people. about 50% of individually owned shares were owned by 1% of people. More importantly almost all pensions were defined benefit so the worker didn't need to care how the pension fund managers did. The correlation of food consumption with stock returns was higher for people who actually owned stock as shown by Campbell and Mankiw (yes the head of the CEA).
d) There might have been a "peso problem" a low risk of a catastrophic collapse of stock prices. This scared rational people but, in the event, it didn't happen. this means that the average return over the period would be a very bad estimate of the expected (ex ante average) return.


Imagine that we are back in say 1955 (predictions are hard especially about the future). For a utilitarian, it would have been a good idea for the treasury to sell more bonds and buy stock. This is true for any of the theories of the stock premium listed above.

1) if people are irrational they are helped by removing (part) of their consumer sovereignty. If people would serve their own interests by buying stock, forcing them to hold stock as citizens is in their interest. One key question is whether people would eliminate the public program by buying less stock personally. I think this is silly. I think it is clear that people don't keep track of what the government is doing on their behalf. Also brokerage fees would keep many people at 0 personal stock holding.
2) the plan would force people (as citizens) to hold a diversified portfolio.
3) the plan would make all people (as citizens) shareholders spreading the risk. The risk was strangely concentrated. Again the strange concentration of share ownership could have been due to brokerage fees etc.
4) A catastrophic lasting collapse of share prices (a in 1929) would happen exactly when the govt should run a huge deficit. If there was such a risk, the plan would serve as an automatic stabilizer.

I really think that the government could have changed the path of consumption in a way which increased expected utility of almost everyone by selling more bonds and buying stock. Thus I think it could have spent more on say health care and left the path of private consumption one that gave higher utility too.

The problems are.
a) First it is easy to predict the past. Maybe the great opportunity has passed now that price earnings ratios have increased. Maybe the stock premium is now what it would be if there were a rational representative agent.
b) if the treasury is required to buy all stocks proportional to their market capitalization there is a chance for fraud (see the 2 cows joke below)
c) if some bureaucrat is allowed to pick stocks there is a huge risk (certainty?) of corruption.
d) if individuals are allowed to pick stocks (as in the Bush admin proposal) and to chose to not pick stocks at all the whole thing doesn't work at all.

I think a is the key thing. I have to admit that, even though I just argued that stocks are a great buy, I personally don't own any and have a defined benefit pension.

No comments:

Post a Comment