The purpose of this research is to analyze the behavior of stock market prices in three phases: (1) An extensive and exhaustive documentation of the statistical properties of stock returns for all the commonly used data sets, with particular attention to how the forecastability of returns differs across sample periods, sampling frequencies, and data bases: (2) An analysis of the statistical side-effects caused by the institutional features of the stock market, such as transaction costs and the existence of bid and offer prices instead of a single price; (3) An examination of the economic implications from the finding of the first two phases. Novel aspects of this research include the construction and analysis of a new database consisting of time-stamped transaction prices for all securities traded on the Toronto Stock Exchange from 1979 to 1987 which included continuous bid-ask quotes; the specification and estimation of statistical models that are capable of fitting stock returns data over many periods; the derivation and empirical implementation of dynamic portfolio strategies that optimally exploit the predictability in returns, so as to gauge the economic importance of forecastable components in stock price changes. This project is important because it will shed light on the possible effects of transaction taxes or other types or forms of governmental regulation on the behavior of the stock market.
