ESSAYS ON VOLATILITY OF STOCK PRICES : OF BUBBLES AND HETEROGENEOUS EXPECTATIONS

Abstract

Recent literature on stock price volatility have suggested that asset price volatility is too great than can be justified by traditional asset pricing model. Trying to explain this phenomena poses a challenging question in financial and economic research. This Academic Exercise consists of three individual and separate essays pertaining to this rich and exciting area of research. It is hoped that their contributions, both individually and collectively, will either complement or enhance the pace of research carried out thus far. The first essay investigates the possibility of incorporating the concept of heterogeneous expectations into the conventional security valuation model to explain the extreme volatility exhibited by real stock prices. A model is created based on differing expectations held by market participants and market prices are subsequently determined by constant interaction of demand and supply at each period. By varying the degree of divergence in opinions of the rational agents, we were able to determine the volatility of prices over the sample period. The results obtained provide strong support for the viability of the incorporation of the heterogeneity input into standard asset pricing models to account for the observed dispersion of real stock prices from their intrinsic values. The second essay, following the research done by Diba and Grossman (1988), examines the presence of rational bubbles based on the non-stationarity properties. While most empirical studies in face of evidence that real stock prices and real dividends being integrated processes are carried out on difference data, an approach which is suitable for tests to be done on levels is introduced. The Canonical Cointegration Regression (CCR) developed by Park (1988) transform the data at levels so that valid statistical inference can be made with the use of conventional tests. It has been shown that statistical tests employed on data at their levels will yield more efficient estimates than those that obtained with difference data. By adopting CCR on the American Stock Market data and with the use of an appropriate F-test, the results obtained provide evidence for the presence of explosive rational bubbles.