The objective of this thesis is to find the optimal portfolio of assets for the Icelandic pension sector and compare it to actual data for evaluation of the investment performance at a macro level. Three optimal portfolios are constructed for this purpose, following a mean variance optimization model by Jorion and Khoury (1996). These portfolios provide the foundation for answering our research questions. Our period of analysis was 10.5 years from January 1998 to June 2008 and additionally we covered the turbulent month of October 2008. The basis of the actual portfolio is a report on the Icelandic pension sector´s investments, issued monthly by the Central Bank in Iceland. The optimal portfolios were built on data acquired for monthly returns from equity and bond indices. For both the actual and optimal portfolios, the investments were categorized into four broad categories: international equities, domestic equities, domestic bonds and risk free assets. Furthermore, the optimal model also included two major currencies, the USD and the EUR, in order to answer our research question about the feasibility of hedging foreign positions. An overview of the Icelandic pension sector is presented, as well as illustrating the radical change in the pension funds´ investment strategy during the period of analysis. Recent catastrophic events in the Icelandic economy made it impossible to complete the thesis without a detailed coverage of the impact that the events had on the pension funds. As the domestic currency, the Icelandic krona (ISK), plays an important role for the pension funds´ foreign investments, we devote a chapter on its analysis, as well as hedging practices of the pension funds. A theoretical framework is introduced which is the backdrop for the calculation of three constrained optimal models: one ex ante or historical portfolio and two ex post or forwardlooking portfolios. From comparing the optimal models´ investment fractions to the actual portfolios, we were able to identify indications of suboptimal investments by the Icelandic pension sector. For the historical portfolio, an “equity preference bias” is detected in the actual allocations, as the optimal model recommends substantially less allocation into equity. This suggests that the Icelandic pension funds have adopted the equity asset class faster than can be rationalized by a mean variance analysis of historical returns. The comparisons of our two forward-looking optimal portfolios to the actual loss incurred in October 2008 indicate that the pension funds were investing excessively in the local equity market. The riskiness of the domestic equity class was abruptly exposed in the events of October 2008, when the local equity market collapsed. Our conclusion is thus that there is strong evidence for that the pension sector was holding relatively risky and undiversified portfolio by making large investments in domestic equities. From mean-variance optimization perspective, a wiser choice should have been to divert the investments into high yielding domestic bonds and diversify all equity holdings internationally.
|Educations||MSc in Applied Economics and Finance, (Graduate Programme) Final Thesis|
|Number of pages||129|