RDP 2000-01: The Efficient Market Hypothesis: A Survey 4. Discussion and Conclusion

The introduction of the efficient market hypothesis thirty years ago was a major intellectual advance. The hypothesis provided a powerful analytical framework for understanding asset prices, and has been responsible for an explosion of research into their behaviour.^[12]

Within a decade, the efficient market hypothesis was so well established that Jensen (1978) was prompted to write that he believed there to be ‘no other proposition in economics which has more solid empirical evidence supporting it’.

Such confidence portends a reversal, and the subsequent twenty years of research and asset-market experience have rendered the efficient market hypothesis a much more controversial proposition.

On some issues, the evidence continues to suggest that the hypothesis gives the right answers, at least to a close approximation. Asset price movements over short horizons are close to a random walk, new information is rapidly incorporated into asset prices (at least most of it is), and fund managers rarely outperform the stockmarket on a consistent basis.

Nevertheless, despite these successes, other features of asset-market behaviour seem much harder to reconcile with the efficient market hypothesis. Some stockmarket anomalies have been shown to be quite robust, including surviving extension to alternative sample periods. In this category, for example, is post-earnings-announcement drift. In the foreign exchange market, the bias of the forward exchange rate as a predictor of the future spot exchange rate has resisted explanations based on economic fundamentals for over a decade. Instead, the evidence from surveys suggests participants in the foreign exchange market do not have rational expectations on average, in violation of one of the building blocks of the efficient market hypothesis.

Supporters of the efficient market hypothesis can argue that many seeming violations of the hypothesis are instead examples of the ‘bad model’ problem. Under this interpretation, predictable excess returns represent compensation for risk, which is incorrectly measured by the asset-pricing model being used. While this is a logical possibility, it presumably applies with progressively less force the longer the violations remain unexplained using models based on the efficient market hypothesis.

Longer-run asset price misalignments almost certainly represent the most serious manifestation of the failure of the efficient market hypothesis. Most tests of the hypothesis do not provide evidence, one way or another, about the possibility of such misalignments. Other types of evidence, however, strongly suggest that such misalignments exist, at least at times.

In the stockmarket, the pricing of closed-end funds is hard to understand as the outcome of an efficient market. The 1987 stockmarket crash, and the unprecedented run-up in US stock prices over the 1990s are both hard to understand except in terms of markets which have moved some distance away from levels consistent with fundamentals.

The inability of models based on economic fundamentals to explain more than a small fraction of the year-to-year movements in floating exchange rates has undermined confidence in the capacity of the efficient market hypothesis to provide a convincing description of this market. This confidence has been further eroded by the anomalous behaviour of the US dollar in the 1980s and the Yen in the 1990s.

The efficient market hypothesis is almost certainly the right place to start when thinking about asset price formation. Both academic research and asset market experience, however, suggest that it does not explain some important and worrying features of asset market behaviour.

Footnote

Ball (1990) provides an extended discussion on the contribution made by Fama et al (1969) in the paper that introduced the term ‘efficient market’. [12]