About Us | Search | FAQ | Contact Us
Speculative Bubbles in Asset Prices
Hot Topic or Hot Air?
October 24th, 1929 will always be remembered as 'Black Thursday' since in one day 13 million shares of trading volume forced the Dow Jones Industrial Average to show losses of 39.6% within a week. Fifty eight years later, on October 19th, 1987, the Dow Jones Industrial Average Index showed colossal losses of 20%. At a press conference, President Reagan commented that 'the collapse of prices has nothing to do with economic fundamentals'.
The evolution of prices during these two periods is very similar as both bull markets started on the second quarter of the year (1924 and 1982 respectively) and lasted 63 months. Market prices peaked in the third quarter of the year (September 3rd, 1929 and August 25th, 1987) and 54 days elapsed between this peak and the market collapse.
There are numerous other examples throughout history of such events, where prices rise continually over a period of time and then collapse spectacularly, not just in stock markets but also in other good and asset markets. At first blush, these patterns are hard to reconcile with the notion that financial markets are efficient and rational, and it is tempting to liken investors to sheep running in a field, almost all in the same direction, but not knowing why they are running or where they are running to.
Research in behavioural finance has suggested precisely that investors exhibit irrational herd behaviour and it is this which is responsible for the protracted price increases followed by swift reversals. But proponents of speculative bubble theory suggest that this price behaviour is entirely consistent with market rationality since investors are compensated for the increased risk of a price collapse by ever increasing returns.
This article presents a broad overview of the academic research on speculative bubbles, and considers whether such research is likely to be of practical value.
and bubble tests
In the presence of speculative bubbles, positive expected bubble returns will lead to increased demand and will thus force prices to diverge from their fundamental value. If the expectation of positive excess returns remains unchanged and the investor is compensated for the increased risk of bubble collapse, then these excess or abnormal returns will be realised in an increasing fashion. Thus, even though investors observe that stocks are overvalued, they are not willing to close out their positions because the bubble component offers at least the required rate of return. Data from 1987 show that, before the October 1987 crisis, 70% of private and 85% of institutional investors knew that the market was overvalued but they did not liquidate their holdings.
There are a number of tests that have been developed for investigating the presence of speculative bubbles, and these have been classified as either indirect or direct bubble tests. Indirect tests, including those based on tests for bubble premia, tests for excess volatility, and tests for long run relationships between price and dividend series, are usually joint tests of bubble absence and of the validity of the present value model. Indirect tests are therefore usually argued to provide only 'hints' of bubble presence and not proof. Direct tests by contrast examine the presence of specific forms of speculative bubbles by identifying the presence of bubble-like behaviour in the data.
The weight of
One simplistic method is to create a "dividend index", which has a value of 100 at the start of the sample period under study, and is increased each month by a percentage equal to the dividend. The value of the FTA-All Share index is plotted in the following figure (taken from Brooks and Katsaris, 2001) for data from January 1965 to March 1999 together with the dividend index constructed in this way. Both series are measured in real terms (that is, they are adjusted for inflation).
It is evident that the stock index deviated substantially from its fundamental value for prolonged periods during late 1968, 1972-73, 1987 and again in the late 1990's as a result of the internet and technology boom. All of these were followed by sharp reversals in market fortunes. The usefulness of dividends as a measure of fundamentals may be questionable, particularly in the light of a reduction in the propensity of firms to pay dividends over the sample period, but numerous alternative measures of fundamentals paint broadly the same picture.
Although the vast majority of studies have concluded that bubbles have been present in financial markets during certain historical periods, results do appear dependent upon the choice of testing approach, sample period, and the measure of fundamental value employed. The search continues for a definitive and infallible method of bubble detection.
The future of
To summarise, evidence for the presence in asset and other markets of periodically collapsing speculative bubbles is overwhelming. However, until recently, bubble research was of academic interest but had limited practical value for financial market traders. New research from the ISMA Centre, University of Reading has shown how this research can be used, via estimation of the probability of bubble collapse, to determine when to fly from risky assets and take refuge in a safe haven.
References: Brooks, C. and Katsaris, A. (2000), "Rational Speculative Bubbles: An Investigation of the London Stock Exchange", Department of Economics, Discussion Papers in Accounting and Finance 65 Brooks, C. and Katsaris, A. (2002), "Forecasting the Collapse of Speculative Bubbles: An Empirical Investigation of the S&P 500 Composite Index", ISMA Centre Discussion Papers in Finance 2002-06.
ISMA Centre, P.O. Box 242, University of Reading, Whiteknights, Reading, RG6 6BA, UK.
|Home | About Us | Search | FAQ | Contact Us|