Summary: |
In this paper, we propose and implement the concept of Liability Driven Investment within the context of defined contribution pension funds, which do not have implicit liabilities. We adopt a transitional approach, by moving from a one period mean variance analysis through to a dynamic optimisation approach. This approach would enable us to see how different approaches could result in significant allocations to the various asset classes. We begin by constructing one period mean variance portfolios for different points on the risk aversion spectrum. We make use of the risk aversion figures provided in Hanna et al (2001). Here, we observe how the allocation to real estate is influenced by different risk tolerance levels. In general, it is believed that as people get closer to retirement, their risk aversion increases. Thus, results from this section would provide a guideline for the proportion of the various assets would have to be included at different points in time. Next, we run an Asset Liability Management (ALM) using a model similar to that used by Booth (2002). We construct the liability driven portfolios using the following liability benchmarks: (i) Long term government bonds (ii) Index-linked bonds (iii) Age profile of the UK. Further, we analyse the role that real estate plays in these portfolios by imposing regulatory constraints which directly impact the allocation to real estate. We obtain these regulations from the OECD (2013) Annual Survey of Investment Regulation of Pension funds. Our analysis is done from the perspective of a a UK defined contribution pension fund, although we use global indices to diversify globally. Also, in applying the rules, we simply apply the various rules currently in place (as at 2013) in all the countries even though these rules might not be in existence in the UK. We aim to determine, in general, what different regulations with respect to real estate would ultimately impact on pension fund outcomes. Also, further analyse role that real estate assets play in the resulting portfolios, we make use of a disaggregated IPD index instead of overall returns used in other studies. We include all the the various sectors that make up the IPD index and treat them as independent asset classes. Finally, following Dempster et al (2002), we use a dynamic ALM model which is a variant of the Computer-Aided Asset Liability Management (CALM) Model of Dempster (1993). This approach has also been used by Consigli and Dempster (1998 |