A MARKOVIAN APPROXIMATED SOLUTION TO A PORTFOLIO MANAGEMENT PROBLEM



Approximated Realisations. Average Utility = 748

x 104

ip

0          1          2         3         4         5         6         7         8         9


tions ∈ [159,162]).10 Averaging the utility over
more realisations and diminishing the simulation
step could help to improve the utility variance
estimate. However, the improvement would not be
substantial, as the portfolio pcrfomancc is highly
“volatile” whether optimal (Figure 4) or not (Fig-
ure 13).

□ 0.5-


5. PORTFOLIO MODEL MODIFICATIONS

0---------------------1---------------------1---------------------1---------------------1---------------------1---------------------1---------------------1---------------------1---------------------L

0          1          2         3         4         5         6         7         8         9

Fig. 13. Approximated wealth, investment and
consumption rate realisations.


Fig. 14. Utility realisation distribution.


5.1 Time Varying Parameters

σ(t) = σ ɑ — 0.09 cos ŋ


(36)


Suppose now that the agent- expects the volatility
coefficient-
σ to vary as follows
where σ = .4. This means that the volatil-
ity
σ used in Sections 3 and 4 was an average
value. Now, the volatility will rise from 36.4 %
to 43.6 % in the middle of T, and then drop. The
agent- would like to know whether this information
should change their investment- strategy or not-.

Figure 15 reveals the modified strategy obtained
as a solution to the discrct-iscd portfolio problem
with ∂' = .05 and
h = 500. As expected11, the
agent- will invest- more (i.c., above .75) when the
volatility is low. The solution is “exact” in that we
can read how much one has to invest- at each time.
Interestingly, the optimal consumption strategy
remains unchanged.

The wealth and strategy sample paths for the
cosine volatility problem arc shown in Figure 16.
The discounted total utility is 664 (std=158).

In summary, the agent- should modify their in-
vestment- strategy once the information about- a
volatility scenario becomes available. In a similar
way, a portfolio problem with a time dependent-
interest- rate (or other parameters) can be solved.

as judged by index J for other approximating
rules (c.g., ∂' = .05, .02 and
h = 500, 100) was
comparable (J ∈ [701, 720] with standard devia-
10Howovor, using ∂' = .5, h = 10000 that is clearly a “bad4
policy, resulted in
J = 124.13 with the standard deviation
equal 100.

11 Soo footnote 19.

11




More intriguing information

1. Climate change, mitigation and adaptation: the case of the Murray–Darling Basin in Australia
2. The name is absent
3. The name is absent
4. Long-Term Capital Movements
5. Modellgestützte Politikberatung im Naturschutz: Zur „optimalen“ Flächennutzung in der Agrarlandschaft des Biosphärenreservates „Mittlere Elbe“
6. The name is absent
7. Parent child interaction in Nigerian families: conversation analysis, context and culture
8. Une Classe de Concepts
9. Conservation Payments, Liquidity Constraints and Off-Farm Labor: Impact of the Grain for Green Program on Rural Households in China
10. LABOR POLICY AND THE OVER-ALL ECONOMY
11. The name is absent
12. Are combination forecasts of S&P 500 volatility statistically superior?
13. Fiscal Insurance and Debt Management in OECD Economies
14. Imperfect competition and congestion in the City
15. Recognizability of Individual Creative Style Within and Across Domains: Preliminary Studies
16. Benefits of travel time savings for freight transportation : beyond the costs
17. The name is absent
18. Feature type effects in semantic memory: An event related potentials study
19. FUTURE TRADE RESEARCH AREAS THAT MATTER TO DEVELOPING COUNTRY POLICYMAKERS
20. The name is absent