A TRANSITION MODEL FOR PORTFOLIO REVISION*

Abstract

PORTFOLIO MANAGEMENT is a decision process having to do with choosing and manipulating, from a large population of possible investments, that subset which best satisfies certain constraints and achieves certain specified objectives. Within the field of finance, portfolio management has become an area of growing importance. Investment companies, corporate pension funds, and the variety of trust funds managed by commercial banks are all important examples of portfolio management in practice. The literature of finance contains several distinct subject areas which, directly or indirectly, relate to the problem of portfolio management. Included are techniques of security analysis, utility theory and investment behavior, stock market movements, empirical measurement of security performance, and portfolio selection methodology. Unfortunately, there is little published work on the portfolio manager's most important problem-how to revise or alter a portfolio of investments over some finite period of time. The programming application of Naslund and Whinston' and Chambers and Charnes2 both consider portfolios over time, but only at the beginning of the investment horizon. That is, there is no adjustment mechanism provided which is dependent upon what happens after the initial portfolios are purchased. The purpose of this paper is to extend an existing methodology for portfolio selection to an intertemporal basis. The suggested technique is an adaptive type of mechanism which is performed at finite intervals. The technique is illustrated over the 1957-1964 period using a population of 150 common stocks. Portfolio yields which result from the revision procedure are compared with similar performance measures from unrevised portfolios. Section I is a brief review of the Markowitz/Sharpe diagonal model which furnishes a basis for selecting and revising portfolios. A transition model which is used as a framework for portfolio revision is presented in Section II. Although the data inputs into the Markowitz/Sharpe model are generated by historical extrapolation, two important adjustments are necessary-these are discussed in Section III. Section IV includes a summary of the research design used to test the suggested transition model. Results of the empirical test of the transition model are presented in Section V. The final section is a summary of the findings of the study and some implications for subsequent research.