Grant-funded Project Nr. 465/2004/A-EK/CERGE Final Report
Project title:
The alternative model of portfolio management
Research leader:
Mgr. Martin Vojtek, 2001
Co-researcher:
Doc. Ing. Evžen Kočenda, PhD.
Period of project:
2004-2004
Overall grant:
77 000 CZK
Project Results
The objective of this work is to propose a new, alternative model of portfolio management. IT is often interesting task to determine a portfolio of bonds or assets which cash-flow replicates the cash flow of liability stream. This is important task for example for the insurance companies that often face a liability stream reaching several years into the future. This stream is usually stochastic as well as interest rates. Therefore the usual way of solving is to use the stochastic programming approach. The problem with this approach is that it is hard to solve numerically, as it is often large-scale program with hundreds of thousands of constraints and decision variables. Similarly, the probability distribution of scenarios has to be known in advance what is often problematic. Therefore a new model is proposed which deals with these drawbacks.
The process is as follows: The manager chooses a scenario which he deemed to be the most probable to occur and he solve the task of finding the best portfolio as deterministic program with respect to this scenario. He also chooses several others scenarios and he requires that the yield of portfolio is bounded from below for these scenarios. It is a way of hedging against the movement in interest rates etc. The reality behind this is the practice of reassurance of insurance companies. This practice allows to spread losses in the case of unexpected event among more companies, lessening the impact of claims on one company. It is a way of bounding the expected loss from below.
The model was constructed and the basic properties of optimal solutions were proved as well as the sensitivity analysis with respect to chose of scenario etc was performed. Preliminary analysis shows that it may not be the best model, as it does not use all information in the market as stochastic programming does but it is much more easy to compute. it also has some drawback as the optimal solution may not exist if the manager chooses too strict boundaries on the yield under alternative scenarios.
The preliminary version of this paper was presented in the Stochastic Finance 2004 International Conference in the poster session. It is also being published in the Discussion Paper Series at CERGE-EI, No. 2005-141