06.06.2013 Views

STOCHASTIC

STOCHASTIC

STOCHASTIC

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

utility functions dependent on mean and variance parameters when the returns<br />

have normal or two equally likely point distributions.<br />

One may compute the mean-variance efficient surface in several ways; for<br />

example, by solving a parametric qudratic program as discussed in Exercise<br />

CR-13. Paine (1966) illustrates this calculation when the returns follow the<br />

Markowitz-Sharpe (see Sharpe, 1970) diagonal model. When nonnegativity<br />

constraints are not present, one can, as in Exercise CR-14, explicitly solve for<br />

the optimal portfolio allocations. The efficient surface is then an explicit<br />

quadratic function of the stipulated mean portfolio return. See also Hart and<br />

JafFee (1974) for a novel application of mean-variance analysis to the problem<br />

of financial intermediation. Their paper also develops some properties of the<br />

mean-variance efficient set when positive and negative holdings are allowed.<br />

When there is a risk-free asset the efficient surface is a ray in mean-standard<br />

deviation space, and one may find the optimal proportions by solving the<br />

fractional program or linear complementary problem developed in Lintner's<br />

paper. Exercise CR-3 discusses some aspects of the fractional program (see<br />

also Exercises II-CR-19 and 20 and II-ME-18). The continuity properties of<br />

the efficient surface in this case are explored in Exercise ME-3. The number of<br />

efficient portfolios in a given portfolio problem is determined largely by the<br />

constraints on the investment allocations. Exercise CR-11 illustrates how a<br />

useful constraint relaxation increases the number of efficient portfolios. The<br />

analysis is carried out for a numerical problem with one safe and two risky<br />

assets.<br />

As indicated above, if the investment returns have normal distributions the<br />

mean-variance approach and the expected utility approach imply the same<br />

optimal portfolio behavior. That is, all optimal solutions of the expected<br />

utility problem are mean-variance (or equivalently mean-standard deviation)<br />

efficient. In their paper, Pyle and Turnovsky investigate whether or not an<br />

investor who operates with a safety-first criterion also has the same implied<br />

behavior. The basic notion in a safety-first criterion is that the investor wishes<br />

to minimize the chance of obtaining a large loss. There are several ways to<br />

formulate such a notion and Pyle and Turnovsky consider the following three<br />

alternative versions. Choose the investment allocations to: (a) minimize the<br />

probability of obtaining a return below a certain stipulated (aspiration) level;<br />

(2) provide the largest (fractile) return level that is achieved with at least a<br />

given probability level; and (3) provide the largest mean return such that a<br />

stipulated aspiration level is achieved with at least a given probability level.<br />

All of these criteria lead to linear indifference surfaces in mean-standard<br />

deviation space: hence if the mean-standard deviation efficient surface is<br />

convex but not linear there is a unique indifference surface that is tangent to<br />

the efficient surface. Hence there is a unique safety-first investor who will<br />

choose the same portfolio as that chosen by an investor with a given concave<br />

INTRODUCTION 205

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!