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.

w*t = glW,;Z,-1,...,Z0]<br />

= gT_i[W>] if the Z's are independently<br />

distributed.<br />

Our problem is now solved for every case<br />

but the important case of infinite-time horizon.<br />

For well-behaved cases, one can simply let<br />

T -» oo in the above formulas. Or, as often<br />

happens, the infinite case may be the easiest of<br />

all to solve, since for it C*t = j{Wt), w*t =<br />

g(W,), independently of time and both these<br />

unknown functions can be deduced as solutions<br />

to the following functional equations:<br />

0= U'[f(W)] - (l+p)- 1<br />

/•<br />

>[(W - j(W)) {(1+r)<br />

-g(W)(Z- l-r)}][(l-t-r)<br />

-g(W)(Z-\-r))dP(Z) (17')<br />

0= f"v'[{W-j(W)}<br />

{l+r-g(W)(Z- 1-r)}]<br />

[Z-l-r]. (17")<br />

Equation (17'), by itself with g(W) pretended<br />

to be known, would be equivalent to equation<br />

(13) of Levhari and Srinivasan [4, p. f]. In<br />

deriving (17')-(17"), I have utilized the envelope<br />

relation of my (13), which is equivalent to<br />

Levhari and Srinivasan's equation (12) [4,<br />

P- 5].<br />

Bernoulli and Isoelastic Cases<br />

To apply our results, let us consider the interesting<br />

Bernoulli case where U = log C. This<br />

does not have the bounded utility that Arrow<br />

[1] and many writers have convinced themselves<br />

is desirable for an axiom system. Since<br />

I do not believe that Karl Menger paradoxes<br />

of the generalized St. Petersburg type hold any<br />

terrors for the economist, I have no particular<br />

interest in boundedness of utility and consider<br />

log C to be interesting and admissible. For this<br />

case, we have, from (12),<br />

Ji(W) = Max logC<br />

{CM<br />

+ E(l+p)-*\og[(W-C)<br />

{(l-Bi)(l+r)+i»Z}]<br />

= MaxlogC+ (l-HO-Hogr.W-C]<br />

{C}<br />

LIFETIME PORTFOLIO SELECTION 243<br />

+ Max /"log[(l-si)(l+r)<br />

•'o<br />

M<br />

+ wZ)dP(Z). (18)<br />

Hence, equations (12) and (16')—(16") split<br />

into two independent parts and the Ramsey-<br />

Phelps saving problem becomes quite independent<br />

of the lifetime portfolio selection problem.<br />

Now we have<br />

0= (l/C) - (l+P)- 1 (W r -C)- 1 or<br />

CT-i = (l+p){2+P)- 1 Wr-.1<br />

(19')<br />

• /<br />

(Z-l-r)[(l-u.)(l+r)<br />

+ wZ]~ l dP(Z) or<br />

wT~i = w* independently of WT-i- (19")<br />

These independence results, of the CT_i and<br />

wT_l decisions and of the dependence of wT-i<br />

on WT_lt hold for all U functions with isoelastic<br />

marginal utility. I.e., (16') and (16")<br />

become decomposable conditions for all<br />

V(C) = l/yCv, y

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

Saved successfully!

Ooh no, something went wrong!