Roll A Critique Of The Asset Pricing Th
Journal of Financial Economics 4 (1977) 129-176. (0 North-Holland
A CRITIQUE
OF THE ASSET PRICING
Publishing Company
THEORY’S
TESTS
Part I: On Past and Potential Testability of the Theory*
Richard ROLL*
University
of California,
Los Angeles,
l
CA 90024,
U.S.A.
Received June 1976, revised version received October 1976
Testing the two-parameter asset pricing theory is difficult (and currently infeasible). Due to a mathematical equivalence between the individual return/beta’ linearity relation and the market portfolio’s mean-variance efficiency, any valid test presupposes complete knowledge of the true market portfolio’s composition. This implies, inter alia, that every individual asset must be included in a correct test. Errors of inference inducible by incomplete tests are discussed and some ambiguities in published tests are explained.
If the horn honks and the mechanic concludes that the whole electrical system is working, he is in deep trouble. . .
Pirsig (1974)
1. Introduction and summary
The two-parameter asset pricing theory is testable in principle; but arguments are given here that: (a) No correct and unambiguous test of the theory has appeared in the literature, and (b) there is practically no possibility that such a
*This is Part I of a three-part study. Parts 11 and III are summarized in the introduction here. but will appear in later issues. A copy of the complete paper can be obtained by writing the author at: Graduate School of Management, University of California, Los Angeles,
CA 90024. USA.
**This paper was written while the author was at the Centrc d’Enseignement Sup&ieur des
Afl’aires. France. Eugene Fama, Michael C. Jensen, John B. Long, Jr., Stephen Ross and
Bruno H. Solnik provided many useful comments and Patricia Porter provided excellent secretarial service. While the paper was being written, Fama pointed out that his new book
(1976) contains some of the same analysis and conclusions. New papers by Stephen Ross
(forthcoming)
A CRITIQUE
OF THE ASSET PRICING
Publishing Company
THEORY’S
TESTS
Part I: On Past and Potential Testability of the Theory*
Richard ROLL*
University
of California,
Los Angeles,
l
CA 90024,
U.S.A.
Received June 1976, revised version received October 1976
Testing the two-parameter asset pricing theory is difficult (and currently infeasible). Due to a mathematical equivalence between the individual return/beta’ linearity relation and the market portfolio’s mean-variance efficiency, any valid test presupposes complete knowledge of the true market portfolio’s composition. This implies, inter alia, that every individual asset must be included in a correct test. Errors of inference inducible by incomplete tests are discussed and some ambiguities in published tests are explained.
If the horn honks and the mechanic concludes that the whole electrical system is working, he is in deep trouble. . .
Pirsig (1974)
1. Introduction and summary
The two-parameter asset pricing theory is testable in principle; but arguments are given here that: (a) No correct and unambiguous test of the theory has appeared in the literature, and (b) there is practically no possibility that such a
*This is Part I of a three-part study. Parts 11 and III are summarized in the introduction here. but will appear in later issues. A copy of the complete paper can be obtained by writing the author at: Graduate School of Management, University of California, Los Angeles,
CA 90024. USA.
**This paper was written while the author was at the Centrc d’Enseignement Sup&ieur des
Afl’aires. France. Eugene Fama, Michael C. Jensen, John B. Long, Jr., Stephen Ross and
Bruno H. Solnik provided many useful comments and Patricia Porter provided excellent secretarial service. While the paper was being written, Fama pointed out that his new book
(1976) contains some of the same analysis and conclusions. New papers by Stephen Ross
(forthcoming)