Stratégie poistenia portfolia s viacnásobnými investičnými horizontmi

Abstract

The thesis is focused on definition of a multi-horizon portfolio choice model. The goal of the thesis is to invent and analyze multi-horizon models that utilize portfolio insurance strategies. Several models are created with different parametric and functional complexity. The analysis of presented models is dedicated to sensitivity analysis and comparison with classic one-horizon models through an adjusted historical simulation. The thesis is divided into two parts, theoretical-methodological and application part. In the theoretical part of the thesis, classical portfolio choice models are firstly described and sequentially supplemented by models of current research in financial economics. It is shown that both rational expectations models as well as models based on behavioral economics provide theoretical background to the portfolio insurance models. Portfolio insurance model is defined as a portfolio choice model that limits realizations of negatively anticipated outcomes of an investment. The portfolio insurance models’ theory is followed by definition and quantification of the impact of overall portfolio horizon and frequency of parameters change mismatch. Presentation of multi-horizon models assumes prudent solving of time propagation of utilized estimates, what is devoted the second part of the thesis to. Advanced dynamic models together with structural models are built on the demonstration of basic model. The third part presents and quantitatively defines multi-horizon portfolio insurance models. In the application part applicability and economical rationality requirements of invested models are verified. The sensitivity to models’ parameters is analyzed. Special emphasis is devoted to the comparison with regular one-horizon models. Presented multi-horizon models are back-tested based on various assumptions regarding time propagation of utilized estimates. Models are subsequently compared to each other based on qualitative criterions with respect to the flexibility and complexity. Multi-horizon portfolio insurance models can be used as portfolio choice engine in the mutual or pension funds management where invest investors that are homogeneous in terms of their risk/reward preferences but heterogeneous in terms of their invest horizons. Introduced models decrease probability of the negative outcome of investment even in multi-investor setting. Multi-horizon models can be simultaneously utilized to eliminate problem of allocation jumps in other dynamic portfolio models.