After a longer break than I’d planned, I’m back for the second and final instalment of my series on the efficient markets hypothesis and its implications for Social Security reform and other issues. The first instalment is here.

Last time, I pointed out that, under the strong assumptions needed for the efficient markets hypothesis to hold, the diversion of social security funds to personal accounts makes no difference at all, since everyone can already choose their optimal portfolio, borrowing if necessary to finance equity investments. A more realistic version with borrowing constraints or high borrowing costs implies that either private accounts or diversification of the holdings of the Social Security Fund can be beneficial, and also that a range of other government interventions will be beneficial. (See also Matt Yglesias

In this post I want to look at the case I think is actually relevant, namely, where the efficient markets hypothesis is violated in so many ways as to be a poor guide to economic policy of any kind.

To begin with, why do I think this is the relevant case? Because the efficient markets hypothesis is way out in its predictions of the key variables it is supposed to explain: the relative prices of bonds and equity and the volatility of asset prices. Compared to the EMH, asset prices are several times too volatile, average returns to equity are several times too high and real rates of interest are much lower than they should be.

In addition, even defenders of EMH admit that the vast majority of the markets that would be required for the hypothesis too hold either don’t exist or are subject to large transactions costs. I’ve already mentioned the transactions costs of borrowing, but an equally important factor is the absence of insurance against job loss or business failure caused by recessions. In effect, defenders of the EMH look at the complexity and sophistication of corporate capital markets and assume that all the other risks and contingencies in the economy are irrelevant.

On the assumption that the difference between rates of return to equity and to government debt are largely due to market failure, the immediate implication is that the government should hold more equity, use its tax power to spread the resulting risk, and thereby achieve a massive risk arbitrage. One way to achieve this is for the Social Security Fund to invest in equity, as was proposed under Clinton. Alternatively, if the central bank holds foreign reserves for whatever reason, there’s a case for holding equity as well as debt, and it looks as if this is happening. Finally, the government can own enterprises outright, for example in the infrastructure sector.

Each of these approaches to public holdings of equity has advantages and disadvantages. Holding a diversified portfolio of small shareholdings raises the problems of ethical investment: what if some members of the public object to investments in some particular company. Ted raised this issue a while back in relation to the default portfolio for private accounts, and its equally acute in relation to diversification of the Social Security Fund.

Investments by governments in overseas equity raise a bunch of political issues in both the investing and target countries.

Finally, direct ownership raises all the issues that have been tossed about in debates over nationalisation and privatisation for decades.

All of these points imply that there are limits to the optimal public holding of equity, though there’s no good theory on this (or on the related question of the optimal level, if any, of gearing for the public sector). In any case, as public holdings of equity increase, the rate of return will fall and the rate of interest will rise, ultimately eliminating the equity premium. So we end up with the mixed economy we all know and (some of us) love.

But as long as governments can realise an average return on equity investments that exceeds their cost of debt, adjusted by the (small) cost of risk in an efficient equilibrium, there’s a case for more public investment, either direct or portfolio.

8 thoughts on “EMH&SS Part II

  1. There is a danger in treating the government as just another investor in equities or as just another entity that has to achieve an optimal degree of leverage. What is left out of this is that the government is the ONLY investor capable of providing the public goods that increase the rate of return of all other investments. This is sometimes lost in the debate on SS where one side simply assumes that all government spending is wasteful and bad, and it is also not something that the EMH can capture either – The returns to building roads, adequate schools, or whatever else you think is a public good are not readily calculated but they are no less real because of that. It would be a big mistake to put the government in the position of chasing an ever shrinking equity premium (and it would certainly shrink if it were chased with $2 trillion even if it isnt already shrunken) while ignoring the physical investments that would raise all returns in the aggregate. There is a good case for having the government first tend to the investments where it has a comparative advantage – indeed is the only player there is – rather than spend all of its time, energy and resources chasing something that may or may not be there and may or may not be eaten up by transactions cost even if it is there.

  2. John, I have been reading Benoit Mandlebroit’s, The Misbehaviour of Markets. He claims the existence of an equity premium is a myth that stems from the failure to account for the very occasional catastrophically bad outcomes in equity markets such as the 1987 crash. He says that people hold more cash and bonds than portfolio theory says they should because of their accurate perception that occasionally these extreme ‘long-tailed’ events occur. He argues that CAPM, Black-Scholes and the rest of conventional portfolio theory should be dumped and a focus reinstated on what can go extremely wrong in an equity market. Any comment?

  3. The 1987 crash is not a good example, since markets regained the lost ground within a few years.

    More generally, catastrophe-based explanations faced the problem that, historically, most stock market catastrophes (eg the Russian revolution) have also been bond market catastrophes.

    That said, I agree that long-tailed events are an important problem for CAPM.

  4. The number of shares in companies which could be considered as reasonable places to invest pension funds, is limited. Massive buying of these shares would raise the price, and thus no longer have a premium. Just how much ,in theory does the SS fund contain, compared to the stock market.

  5. US market is roughly $20 trillion. IIRC, Social Security fund is about $3 trillion, so investing half in equity would have a significant but not overwhelming impact.

  6. JQ said,

    “The 1987 crash is not a good example, since markets regained the lost ground within a few years.”

    Maybe I’m being naive here and this is already well accounted for in models or it counts as market failure. However, it seems the fact that a large proportion of investors can’t or are unwilling to sustain losses of the size of 1987 over say six months let alone a few years makes a big difference to how much premium they will demand for the risk. It is all very well to say that if we know a market will deliver higher returns over 20 years then no significant premium should be sustained, but what fraction of the total amount invested has this timeframe?

    If say 80% of the market is concentrating on pricing equities such that they get a sufficient return over the risk in that year, then they will be indifferent to the fact that outcomes such as the 1987 crash resulted in regained ground within a few years. From my experience of risk modelling in banking it is increasingly becoming the case that you don’t assume that a say a five year risk is just the product of five one year risks. Serial correlations or perhaps mean reversion exists, and not to model it is not to properly reflect risks. Thus the time frame under which you are investing matters and so long as the pricing is not determined by long term investors the fact that they may receive sustained superior returns is unsurprising so long as they are a relatively small section of the market.

  7. I did some doodling involving the composition of probability generating functions a while back. It strongly suggested that an unmanaged portfolio would have financial risk growing with twice the exponent of the expectation (unless they were shrinking).

    If this is not misleading, it suggests that active management of portfolios leads to a rebalancing within a larger notional asset class, which will have a smaller all up financial risk but the risk will still have the same larger exponential growth.

    That would certainly show up as an occasional catastrophic event. Dropping those from the model would lead to measures of how things rebalance once we ignore the house-takes-all events, the notion of quasistability in compounded PGFs.

    But I didn’t model things in any depth, just enough to satisfy my intuitions. I expect someone out there who is doing academic stuff for a living has done a great deal more, if only you could research it.

  8. John,
    I oppose this on principle, but could be convinced. The only instance I am aware of where this sort of thing has been tried was in Hong Kong after the East Asian crash in 1990. True, the reason given for it was to prop up the market (on the assumption of market failure), but I would be interested to read any studies you are aware of on the results of this.

Comments are closed.