Linear Factor Models in Finance

Dr T. Wilding [*]
A factor analytic model has many advantages when used as a model of equity returns. However, the attribution of the raw risk statistics to familiar stock and market phenomena is not a natural by-product of the factor analytic risk model estimation process. In this chapter, we address this issue and describe several methods that can be used to translate data from factor analysis-based risk models into traditional fundamental information that can assist with portfolio management and trading. First, we develop tests to determine whether selected real world attributes indicate anything significant about risk. After we have established a useful set of attributes, we develop further methods to characterize the risk of a portfolio. Finally, we apply these methods to a variety of fundamental attributes, sectors, countries, and macroeconomic time series, in order to analyse a sample UK portfolio.
[*] Head of Research & Development, EM Applications Limited, St Martin s House, 16 St Martin s Le Grand, London EC1A 4EN,+44 (020) 73978395, tim.wilding@emapplications.com
Factor analysis based models (Stroyny (1991)) are increasingly used as models of equity returns due to their many advantages. However, results from such models can be difficult to interpret. Although factor analytic models can establish the proportion of risk due to common effects as opposed to stock-specific effects, and can rank stocks according to their marginal contribution to risk, factor analytic models are not naturally suited to characterizing the systematic bets being taken. As the factors are derived from returns data...