Testing a mean reversion hypothesis on the Toronto stock exchange

• Main Author: Williams-Short, Rowan Peter
• Language: English
• Published: University of British Columbia 2010
• Online Access: http://hdl.handle.net/2429/27749

Monthly returns data for the 150 companies that have been listed on the Toronto Stock Exchange for at least 95% of the period from January 1963 to December 1987 are transformed into continuously compounded returns, adjusted for inflation, gathered into portfolios, and examined for the property known as mean reversion. The model used to describe this scenario comprises two components, namely a random walk and an autoregressive scheme of order one. The study's foundation is a paper by Fama and French(1988), using data from the New York Stock Excange, but in addition to all the estimation performed in that paper, other parameters are estimated. The parameter originally of prime interest is shown to behave in accordance with model predictions, in the case of nearly all 28 portfolios constructed. However, this interpretation is shown to be seriously jeopardised in view of closer scutiny of model parameters, which indicate that the model is a poor approximation to the data. === Science, Faculty of === Statistics, Department of === Graduate