Style anomalies on the Toronto Stock Exchange : a univariate, multivariate, style timing and portfolio sorting analysis

Master Thesis


Permanent link to this Item
Journal Title
Link to Journal
Journal ISSN
Volume Title

University of Cape Town

A growing body of empirical evidence has found inconsistencies in the Capital Asset-pricing Model (CAPM) of Sharpe (1964), Lintner (1965), and Black (1972) and Ross's (1976) Arbitrage Pricing Theory (APT). Numerous attempts to explore the validity of these theories of modern finance have led to the identification of various firm specific attributes that explain the cross-sectional variation of returns. These attributes have appropriately been termed 'style anomalies '.This thesis investigates the existence and exploitability of style anomalies for the shares comprising the Toronto Stock Exchange (TSX) for the period 31 January 1989 to 31 July 2005. The investigation is divided into four areas of research. First, a methodology similar to Fama and Macbeth (1973) is used to explore the cross-sectional relationships between some 904 firm-specific attributes and the unadjusted and risk adjusted monthly returns of equities constituting the S&P TSX Composite Index. A myriad of uncorrelated style anomalies are found to persist before and after controlling for systematic risk, and are categorized as either size, growth, momentum, value, liquidity and bankruptcy (risk) effects. The most significant attributes from each respective style group include: Price, eighteen month change in net tangible asset value, price change over twelve months, twelve month change in price to net tangible asset value, three month change in the absolute volume ratio and interest cover before tax. Multivariate testing confirms the ability of anomalies to explain excess returns. In and out sample cross sectional tests show inconsistent anomaly persistence, raising the question of whether they are perhaps perennial in nature. Second, the predictability of style payoffs is examined through the analysis of autocorrelation and six style timing models. Strong positive autocorrelation at lower orders for the majority of style payoffs suggests that the ability to time payoffs is possible. The six month moving average timing model shows the best forecasting skill, followed by twelve month and eighteen month moving average models. Third, the presence of firm specific attributes among three classified sectors namely: Basic materials, Cyclicals and Non-Cyclicals are compared. Risk, value and liquidity based anomalies dominate the Basic Materials shares. Liquidity effects stand out within the Cyclicals group, and the Non-Cyclicals sectors exhibit value and size effects. The ability to exploit all style-based anomalies after accounting for transaction costs is evaluated using a portfolio sorting methodology. The tests illustrate that increased exposure to the anomalies has delivered substantially higher returns with lower volatility than a buy and hold approach using an equally weighted all share benchmark. These abnormal returns are confirmed after adjusting for systematic risk. Further testing shows that the attributes, rather than loading on those attributes, are better at explaining share returns. Finally, the seasonal nature of Canadian equity returns is investigated. A six month strategy of "Selling in June and going away till December" provides the most optimal returns. The calendar month tests find January, February and December to be the strongest months of the year. Attribute payoffs seem to show vague seasonal tendencies.

Includes bibliographical references.