A MARKOVIAN APPROXIMATED SOLUTION TO A PORTFOLIO MANAGEMENT PROBLEM



Fig. 6. Strategy convergence.


The vertical line t⅛=.75 shows where the utility
maximum “should” be, for it is known from Figure
2 that the optimal strategy is t⅛=.75, indepen-
dent of time. It is clear from the figure that the
discretised model strategy
rι(δ) converges to the
optimal strategy as <) → 0. Again, any reasonable
approximation requires
δ < .1.

Another point to remember is that the time step
δ should be large enough for the process to move
beyond the adjacent state i.e.,
δf. > h. On the
other hand, as shown above,
δ should be small for
high accuracy of the approximating solutions.

The convergence. There are various ways in
which the goodness of a numerical solution can
be evaluated. The most “objective” one would
perhaps be to look at the average discounted total
utility
J generated by the application of an ap-
proximated optimal numerical solution to the con-
tinuous model. However, as evident from Figure 4,
the portfolio performance is very “volatile” and
the standard deviation of utility distribution is
large. Therefore, using
J thus computed is difficult
to judge which solution is best.

We will first evaluate the convergence by compar-
ing the approximating policy profiles (Figures 7 -
12), to the optimal ones (Figure 2).9 Then, we

9 Remember that because of the model re-scaling, if l'2
was linear (Figure 2) u2 will be horizontal.
will generate a few realisation profiles to compare
them to those of Figure 4 and, eventually, we will
compute the corresponding utility distribution.

S = .2; h=10000 -:- 100

0.81-

0.7

160

0.5-

0.4-

1          2         3         4         5         6         7         8         9         10        1 1

wealth                                    x 104


0.3

0.35

0.3

025

0.2

0.15

• h=10000

0.11-------------1-------------1-------------1-------------1-------------1-------------1-------------1-------------1-------------1-------------1-------------1_|

1          2         3         4         5         6         7         8         9         10        1 1

wealth                                    χ104

Fig. 7. Approximating strategies for t = 0
(5=.2).


Examine the policy rules shown in Figures 7 -
12. The bold dotted lines correspond to opti-
mal strategies (compare Figure 2). One can see
that the strategy convergence is more difficult
to achieve for later times (t = 9) than at the
beginning of the horizon (t = 0).

Fig. 8. Approximating strategies for t = 0
(5=.1).



More intriguing information

1. What Drives the Productive Efficiency of a Firm?: The Importance of Industry, Location, R&D, and Size
2. The use of formal education in Denmark 1980-1992
3. Understanding the (relative) fall and rise of construction wages
4. Gender and aquaculture: sharing the benefits equitably
5. Large-N and Large-T Properties of Panel Data Estimators and the Hausman Test
6. The name is absent
7. Peer Reviewed, Open Access, Free
8. Industrial districts, innovation and I-district effect: territory or industrial specialization?
9. The name is absent
10. Multiple Arrhythmogenic Substrate for Tachycardia in a
11. The Values and Character Dispositions of 14-16 Year Olds in the Hodge Hill Constituency
12. Foreign Direct Investment and the Single Market
13. The name is absent
14. The Impact of Financial Openness on Economic Integration: Evidence from the Europe and the Cis
15. Consciousness, cognition, and the hierarchy of context: extending the global neuronal workspace model
16. FASTER TRAINING IN NONLINEAR ICA USING MISEP
17. The name is absent
18. ENERGY-RELATED INPUT DEMAND BY CROP PRODUCERS
19. The name is absent
20. DURABLE CONSUMPTION AS A STATUS GOOD: A STUDY OF NEOCLASSICAL CASES