Why does the efficient markets hypothesis matter ?
Reading the discussion of earlier posts about the efficient markets hypothesis, it seems that the significance of the issue is still under-appreciated. In this post, Daniel Davies pointed out the importance of EMH as a source of pressure on less-developed countries to liberalise capital flows, which contributed to a series of crises from the mid-1990s onwards, with huge human costs. This is also an issue for developed countries, as I’ll observe, though the consequences are nowhere near as severe. The discussion also raised the California energy farce, which, as I’ll argue is also largely attributable to excessive faith in EMH. Finally, and coming a bit closer to the stock market, I’ll look at the equity premium puzzle and its implications for the mixed economy.
To recap, the efficient markets hypothesis says that the prices generated by capital markets represent the best possible estimate of the values of the underlying assets. So, for example, the price of a share in Microsoft is the best possible estimate of the present value of Microsoft’s future earnings, appropriately discounted for time and risk.
The EMH comes in three forms. The weak version (which stands up pretty well, though not perfectly, to empirical testing) says that it’s impossible to predict future movements in asset prices on the basis of past movements, in the manner supposedly done by sharemarket chartists, Elliot wave theorists and so forth. The strong version, which almost no-one believes says that asset prices represent the best possible estimate taking account of all information both public and private. For policy purposes, the important issue is the “semi-strong” version which says that asset prices are at least as good as any estimate that can be made on the basis of publicly available information. There’s a heap of evidence to show that the semi-strong EMH is false, but the most dramatic, as I pointed out recently, is that of the dotcom boom, when obviously hopeless businesses like Pets.com were valued at billions of dollars. I’ve discussed this a bit more here
An apparent, but not real, moderation of the semi-strong EMH is the formulation offered by James Surowiecki of a number of commentators, summarised by Daniel, that “whether or not markets are perfectly efficient, they’re better than any other capital allocation method that you can think of.” It’s clear that, if capital markets are always the best possible way of allocating capital then they must produce the best possible estimates of asset values, given available information, and this is just a restatement of semi-strong EMH.
A much more defensible position is that, even if capital markets are not perfect, neither is any alternative, and it is therefore an empirical question whether unregulated capital markets or some alternative, such as regulation or public investment, will yield better outcomes in any particular case. This formulation leads straight to the basic economic framework of social democracy, the mixed economy, leaving, of course, plenty of room to argue about the optimal mix.
Now let’s look at some specific cases where the EMH gives a simple and misleading answer, and where a more careful analysis leads to mixed-economy conclusions.
First, there’s the question of foreign exchange markets. As Daniel observed, the EMH implies that the optimal policy is to allow exchange rates to be determined by unregulated capital markets, in what is called a ‘clean float’. The experience of the Asian crises, and also of Chile and China, suggests that developing countries may do better with controls that limit short-term capital flows. But even among developed countries, belief in the desirability of a clean float has faded. The volatility of exchange rates since the collapse of the fixed exchange rate regime in the mid-1970s has been much greater than expected, and much more than can be explained by any sensible model of rational markets. Most central banks have come to adopt a policy of ‘leaning against the wind’, that is, buying their own currency when the exchange rate is well below the long-run average (sometimes called Purchasing Power Parity, though this isnt quite accurate) and selling when it is well-above. On average, this has been a profitable long-term strategy. A lot of economists, starting with the late James Tobin, would go further and tax foreign exchange transactions in the hope of reducing volatility. The lesson here is that neither rigidly fixed exchange rates nor a perfectly free float is likely to be optimal. You can read some more on this here
Second, there’s energy, and particularly electricity. Until the 1990s, electricity was supplied either by public enterprises (most places except the US) or regulated monopolies. These did not do a perfect job in making investment decisions. Roughly speaking, when engineers were in charge, there was “gold-plating”, notably in the form of excessive reserve capacity. By contrast when accountants or Treasury departments were in charge, capital was rationed tightly, and investment decisions were determined largely on the basis of attempts to get around the resulting artificial constraints.
The EMH implied that private capital markets would do a far better job. This in turn led to the creation of spot markets for electricity which otherwise made little sense, since, in the absence of highly sophisticated metering, most users could not respond to market signals in the required fashion (the big consumers, who could respond, typically had contracts specifying if and how much their supply could be cut in the event of a shortage). In practice, this produced huge reallocations of wealth while failing to produce either sensible signals to consumers or rational investment. Instead, the pattern in investment was one of boom and bust. As a result, there’s been a steady movement away from spot markets and towards the reassertion of more co-ordination and planning of investment. I’ve had a bit more to say about this here
Finally, there’s the equity premium puzzle, that is, the fact that average returns to equity are far higher than can be accounted for by any sensible model incorporating both reasonable risk attitudes and the efficient markets hypothesis. If you reject the efficient markets hypothesis, it’s natural to conclude that the true social cost of capital is close to the real bond rate. This in turn allows us to evaluate privatisation by comparing the profits forgone with the interest saving that arises when sale proceeds are used to repay debt. For Australia and the UK, such an analysis produces the conclusion that most privatisations have reduced welfare. Some of the evidence is in this paper (PDF file), and there’s a more rigorous analysis here (PDF). Responding to this argument, Brad DeLong observes, with customary acuteness that my argument implies that “the natural solution to all this is the S-Word:Socialism:public ownership of the means of production”.
Actually, however, while this aspect of the argument, taken in isolation, would imply that public ownership always outperforms private, there are plenty of factors going in the other direction, such as the principal-agent problems associated with the absence of an owner-manager or a threat of takeover. So, the correct conclusion, as foreshadowed above, is that the optimal arrangement will be a mixed economy.
Determining the optimal mix is a difficult task, requiring lots of case-by-case analysis, but I’ll offer the view that the optimal public share of production and consumption is unlikely to be below 25 per cent, and is typically close to 50 per cent. For a contrasting view, I’ll point to my distinguished predecessor at the University of Queensland, Colin Clark, who thought 25 per cent was an upper bound. I’ll need another big post to spell out my claims here, and this will take some work. In the meantime, feel free to pitch in with your own views.