An empirical evaluation of the psychological barriers hypothesis of security pricing

Models of security pricing under the Efficient Market Hypothesis embody two major assumptions: first, that there exists some economic principle by which a security's future returns distribution can be discounted in competitive markets to give a measure of that security's fundamental value, and second, that investment behaviour underlying competitive trading in securities is based on rational, wealth­ maximising expectations. All models of market efficiency predict that security prices in such markets are forced by competitive, rational forces to equate to their fundamental values, and that no information remains unused by such rational forces by which future returns different from market expectations can be systematically and accurately predicted. However, there have emerged in the economic literature alternative models of security pricing which depart from the major Efficient Market assumption of rational expecations. These investor-sentiment models have different implications for the pricing of securities, allowing for both systematic variation in prices away from fundamental values and predictability of future returns on securities based on information concerning past returns and price levels. One such investor-sentiment model recently developed is the barriers hypothesis (RG. Donaldson and H.Y Kim, Journal of Financial and Quantitative Analysis, 28(3): 313-330, 1993). Like other models of this genre, it invokes investor psychology and behaviour inconsistent with rational expectations and hence with market efficiency. Using daily prices for the Dow Jones Industrial Average (DJIA) from 1952 to 1990, Donaldson and Kim provide highly significant evidence supporting the validity of their barriers hypothesis. That is, inconsistent with predictions of all Efficient Market models, psychological price barriers have been a significant property of the DJIA over approximately the past forty years. In recognizing the potential importance of these findings for our understanding the processes of security pricing, the literature review of this thesis closely examines and compares the general structure of Efficient Market models with that of alternative investor-sentiment models. The empirical so strongly supported. The findings of this thesis show major flaws in the methodology employed by Donaldson and Kim. Replication of their study revealed that all barriers tests were marred by one or more of the following problems: presence of high degrees of autocorrelation in regressions; lack of robustness of results over different subperiods; poor test construction resulting in low testing power; and generation of results, not reported by the authors, that were strongly inconsistent with the barriers hypothesis. In fact, most if not all major empirical results reported by Donaldson and Kim in support of the barriers hypothesis are shown to be artefacts arising due to poor methodology or data issues. The conclusion of this thesis, in direct contrast to that of Donaldson and Kim, is that there exists no evidence in favour of the barriers hypothesis as a model explaining daily movements in the DJIA over the past forty or so years. Furthermore, no evidence in support of the barriers hypothesis was found in tests using the Australian All Ordinaries Price Index. The null hypothesis of no barriers, which in fact can be taken as a model under the Efficient Market Hypothesis, is unambiguously accepted by the empirical tests presented in this thesis.