Abstract
-
The theoretical assumptions and practical limitations of the CAPM have been subjected to
early scrutiny, however, causal relationships embedded in these data, and the index
management practices directly affecting the characteristics of models, have been
ignored in the empirical literature.
Stock market indexes are effectively "closed systems" comprising relatively few ex post
variables (i.e. security size and performance). Without invoking the CAPMs theoretical
framework, its conjectures about security-market relationships can be scrutinised
axiomatically. Two immediate conceptual queries arise. First, since stock market indexes
capture the market contribution of individual securities, why is it necessary to rely upon
statistical processes to estimate the market risk contribution of index constituents? Second,
since index performance ex post derives from the constituents and a controlled experiment is
conducted within a closed market model excluding "exogenous" factors, how can non-market risk arise?
This paper analyses index constituent changes occurring within the S&P/ASX50 Index - a
leading institutional equity index of Australia's largest and most liquid stocks between 1994
and 2002. This identifies index turnover as an important source of statistical anomalies not
previously documented in the literature. In addition, the efficacy of beta values is scrutinised
using a scaled market model not afflicted by constituent turnover: this permits direct
evaluation of security-market relationships in accordance with Markowitz original portfolio
analysis approach. It is argued that the CAPM's theoretical specification of systematic risk is
mistaken and that security betas are unreliable measures for their stated purpose.