06.06.2013 Views

STOCHASTIC

STOCHASTIC

STOCHASTIC

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

326 THE JOURNAL OF BUSINESS<br />

this opportunity are<br />

and that<br />

where fj(xj) is given by (9).<br />

It is easily verified that<br />

Thus,<br />

22 .J<br />

_ (0 with probability 1/2<br />

fty ~~ 13 with probability 1/2<br />

(9) clearly belongs to the class (7). (4) and (3) now give<br />

where<br />

fdoes not exist Xj-l < 0<br />

0 < *j_i < d .<br />

+ d) xj-i > d<br />

-I(ZJ--I) = \xj-i<br />

ll/2(^_,<br />

| = 1l/2(3s,_, + d)<br />

fj-l(.Xj-C<br />

IOJ_I(*J-_I + d) 1<br />

>' 2 + l/2d» 2<br />

/2<br />

0 < Xj_! < d<br />

xj-i > d<br />

aU J (8)<br />

fj{xj) = (xj + d)^ d>0. (9)<br />

/y(xy) = max £{/y+i[(&y - l)z2y + *y]} i = 1, . . . , 7 - 1 , (10)<br />

0£«jy£*y<br />

(11)<br />

(12)<br />

o/_, = l/2[(l/2)" 2 + 2» 2 ] . (13)<br />

We now observe that fj-i(x) is a positive linear transformation of fj(x) only for<br />

x > d; for x < d,fj^(x) < aj-ifj(x).<br />

Proceeding with the solution to (10), we obtain<br />

h(% ' > jfly(*y + « X,->bs 3 l '-'-' J l ' U * ;<br />

where ay is a positive constant, gy(xy) < O.J(XJ + d) in for 0 < Xy < 6„<br />

and<br />

J> = b, J - 1 , J- - 1 • (16)<br />

It is easily determined that hj(x,) is highly irregular except for j = / — 1.<br />

When / = 11 and d = 1,000, we obtain from (IS) that Ai = 1.023 million. Thus,<br />

when the horizon is ten periods distant, the optimal amount to invest in opportunity<br />

2 is, in this example, proportional to Xy + d only if initial wealth xt exceeds<br />

$1 million by a substantial margin. Furthermore, while/y(x) is a positive linear transformation<br />

of fj(x) for x > bj, j = 1, . . . , J — 1, it is not for x < bs, that is, for<br />

Xi < 1.023 million, x2 < 511,000, etc., in the above example. Since the constant<br />

bj > 0 depends on the distribution functions Fj, . . . , Fj-i, the short-run utility<br />

functions induced by the terminal utility function (9) are clearly not myopic. In<br />

other words, to make an optimal decision at decision point j, not only Fj but Fj+h<br />

. . . , Fj-i must be known.<br />

IV. BORROWING AND SOLVENCY<br />

The nonmyopic nature of the induced utility functions/i(xi), /2(x2), • • • ,fj-i(xj-i)<br />

in the preceding example is clearly attributable to the constraint<br />

0 < zij < XJ j = 1, .. . , J - 1 , (17)<br />

which precludes borrowing and short sales. We shall now relax this constraint.<br />

3. MYOPIC PORTFOLIO POLICIES 403

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

Saved successfully!

Ooh no, something went wrong!