Someone has said that maternity is a matter of fact, while paternity is, at best, a matter of opinion. Concerning the field to which this issue of the Journal is devoted, however, there is little difference of opinion — Harry Markowitz is almost universally considered to be the father of portfolio analysis. The essence of the theory appeared in his 1952 article, although the full exposition was not available until several years later.

This article will concern itself with the various techniques for selecting portfolios of securities. It should be made clear at the outset that a good portfolio is not just an amalgamation of a number of “good” stocks and bonds. Rather, it is an integrated whole, each security complementing the others. Thus, the investment manager must consider both the characteristics of the individual securities and the relationships between those securities. Until recently there was no comprehensive theoretical framework for the analysis of the latter aspect of the portfolio problem. Intuitive judgment and experience were the guidelines used by investment managers.

The Markowitz analysis of efficient portfolio selection, which can be interpreted as solving the quadratic-programming problem of minimizing the variance of a normal variate subject to each prescribed mean value, easily can be generalized (in the special case of independently distributed investments) to the concave-programming problem of minimizing the “dispersion” of a stable Pareto-Lévy variate subject to each prescribed mean value. Some further generalizations involving interdependent distributions will also be presented here.

Since the publication of Markowitz's article on “Portfolio Selection,” which was subsequently expanded into a monograph, there has been a great deal of further articulation, a not inconsiderable amount of mathematical programming and sensitivity analysis, the arrival of several competing portfolio balance models, and a near revolution in the theory of money and asset preference. While few formulas for solving a practical problem can claim to have generated as much theoretical fall-out, the new approach to portfolio management apparently has not been very successful at reaching the practitioners for which it was intended.

Most of the insurance oriented literature contains several levels of abstraction regarding the financial solvency of insurance companies. The first level represents those advances in the actuarial profession which are destined to eventually solve the double problem of capacity and capitalization (hereinafter referred to as the Ruin Problem). Most of these works, however, are concerned with developing the tools necessary to determine the theoretical loss distribution in insurance. The emphasis, then, is upon the mean and variance of the total loss distribution as they can be derived from the distribution of frequency and severity.

A theoretical model capable of supporting a rigorous analysis of portfolio selection in financial intermediaries appeared only recently. In the absence of a suitable theoretical framework, the limitations of maximizing behavior as an explanation of the selection of asset and liability structures in this class of firms were obscured. Discussions bearing on this question usually focused on the structure of one or the other side of intermediary balance sheets and gave little attention to the effects of these structures on the risk associated with their equity.

Haley's line of reasoning can be reconstructed in the following way. When a borrower incurs a liability (issues a bond) he should gauge any prospective asset purchase with the proceeds against an alternative fund use, the purchase of his own bond. If the proceeds realized from the bond are B, but if the borrower would willingly pay L to be free of the obligation, L becomes a relevant variable in the asset acceptance decision. If the discounted value of any asset exceeds L, borrowing to buy it will be subjectively wealth-enhancing, whether or not the discounted value exceeds B.