The role of real estate in investment portfolios was analysed within an Asset Liability Management Framework since 2000 (see Chun et al (2000), Craft (2001, 2005), Booth (2002)). It was found that, when liabilities are taken account of, the diversification role of real estate turns out to be more limited and accounts for the reduced exposure to this asset as witnessed in the portfolio of most institutional investors._ _Brounen et al (2010) highlighted another important issue: the choice of liability stream. They demonstrated that different definitions of liability result in different allocations to real estate. They compared results obtained when the actuarial projected benefit obligations of DB Schemes as used in Chun et al (2000) to those obtained by using the market value of the projected benefit obligations. The market values were obtained by accounting for interest rate and inflation risk.

With the shift from Defined Benefit to Defined Contribution pension structures across the globe, many feel that there is no need to adopt an ALM approach in optimisation. However, the global financial crisis which occurred between 2007 and 2008 has brought to the fore the uncertainty of income that pension contributors, especially those based on defined contribution (DC) formulas. Pension legislations in some OECD countries require that DC Pension funds provide a certain minimum on pension contributions. These minimum guarantees could be absolute or relative. Relative return guarantees are set in relation to a certain benchmark synthetic investment portfolio or the average performance of pension funds in the industry. When minimum return guarantees are offered by companies that sponsor DC Plans, the plan inherently takes on DB features (OECD, 2012).

In the first part of this study, we examine the long-run relationship between the various asset classes and our chosen liability benchmarks using correlation analysis and cointegration techniques. We make use of various minimum return guarantees and other potential benchmarks as our definition of liabilities: 0% nominal return; 0% real return (inflation proxies: CPI and RPI), government bond rate, risk-free rates such as t-bill and repo rates (Sharpe, 1994). In the second part of our study, we use the dynamic Asset Liability Model of Dempster et al (2002). As in Brounen et al (2010), results of our study would demonstrate how a change in definition of liability impacts on the asset allocation decision.