Abstract

This thesis examines three major issues dealing with the risk of Canadian stocks. The first issue is what are the differences in various measures of idiosyncratic volatility (IV) and is this risk priced. To this end, various measures of realized, conditional and idiosyncratic volatility are examined for Canadian stocks for the 1975-2003 period. As for other markets, smaller firms exhibit higher total and idiosyncratic risks than their larger counterparts, and IV accounts for almost three-quarters of total volatility for the six studied samples. Unlike other markets, Canadian IT firms exhibit considerably lower volatilities. The relationship between returns and IV is examined using various approaches with(out) the presence of control variables for liquidity and firm-specific information embedded in stock prices. The conditional relation between returns and asymmetric IV is highly significant, robust and as expected (i.e., positive and negative for correspondingly signed excess returns). The second issue is whether a minimum portfolio size (PS) should be prescribed to achieve a naively but sufficiently well-diversified portfolio for investment opportunity sets (un)differentiated by cross-listing status and market capitalization. To this end, various (un)conditional metrics are used to measure diversification benefits for stocks listed on the TSX for 1975-2003. The minimum PS is found to depend upon the chosen investment opportunity set, the metric(s) for measuring diversification benefits, and the criterion for determining when the portfolio is sufficiently well diversified. The third (final) issue is to re-examine volatility transmission for stocks cross-listed in synchronous markets. To this end, four bi-variate GARCH models are used to examine contemporaneous co-movement, asymmetry and volatility transmission effects for equal (value-)weighted daily returns for Canadian stocks cross-listed on the TSX and U.S. markets for 1975-2003. Contemporaneous and asymmetric comovements decrease during the 1990s and increase thereafter, mostly due to the entrance of new (and smaller) stocks into both national markets. Conclusions about the directional change of asymmetric volatility spillovers depend upon the choice of GARCH model. Thus, a researcher should use more than one multivariate GARCH model in order to draw robust inferences on cross-market volatility dynamics