Multifactor Expected-Returns Models and the Performance of Superstock Portfolios in the UK Equity Market

Abstract:     
Two empirically testable behavioural finance hypotheses are that (1) 'superstock' portfolios derived from multifactor expected-returns models will have higher than average returns and lower than average risk in terms of statistical and economic significance; and, (2) the expected-returns factor models will demonstrate greater predictive and explanatory power than the risk- and expected-returns-risk factor models of modern finance. Using a dataset comprised of the entire universe of fully-listed stocks in the UK market for the period 1987 to 2002 a multifactor expected-returns model is constructed to estimate security payoffs to factors related to various characteristics such as risk, price level, liquidity, growth potential and previous performance. These payoffs are then used to estimate out-of-sample expected returns and to construct a 'superstock' portfolio. Our analysis suggests that compared to the existing risk-factor models, an 'expected-returns' factor model exhibits increased predictive power of expected returns and has consistently higher than average realised returns with lower than average risk. These results are significant in both statistical and economic terms and corroborate the above behavioural finance hypotheses.

Number of Pages in PDF File: 33

Keywords: Behavioural Finance, Multifactor Models, Superstocks, Expected-returns models

JEL Classification: G11, G12, G14

working papers series