Student Number : 9005994G - MSc research report - School of Construction Economics and Management - Faculty of Engineering and the Built Environment / The objectives of this research are to assess the returns, risks and correlation of a mixed-asset portfolio, establish the role of real estate in a mixed-asset portfolio and suggest an appropriate real estate allocation in the South African pension fund industry. The issue of low real estate allocation has been a subject of interest to practitioners and academics, both locally and internationally, despite the diversification benefit that real estate provides in a mixed-asset portfolio. A statistical approach was considered most the appropriate tool for analyzing returns. Solver optimizer in the excel spreadsheet package was used to generate efficient frontiers and the associated values of portfolios. Real estate provided a lower return of 1.25% and a lower standard deviation of 4.90% compared to equities with a return of 1.39% and the highest standard deviation of 6.23%, whilst bonds provided the best risk-return trade off, with a return of 1.42% and the lowest standard deviation of 2.64%. An equally -weighted portfolio consisting of bonds and stocks and a portfolio consisting bonds, stocks and real estate was simulated. The equally -weighted portfolio of bonds and stocks provides a return of 1.41% and a standard deviation of 3.76%. The minimum variance with bias to bonds provides a higher return of 1.42% at a lower level of risk of 2.62%. The equally -weighted portfolio consisting of bonds, stocks and real estate provides a return of 1.35%, with a lower risk of 3.49%. The minimum variance with bias to bonds provides almost the same return of 1.40% at a lower level of risk of 2.54% compared to the bond and share portfolio. The Chi-Square statistical tool was used to test the diversification benefit of real estate. It can be concluded that the standard deviation of the portfolio with property is close enough to the standard deviation without property of 3.76% and cannot statistically say that it is different given the 5% level of significance. The Sharpe ratio was used to test the favourable risk adjusted returns offered by real estate. It concluded that property provides favourable risk adjusted returns and diversification benefits, as illustrated with the increasing portfolio return from 7.44% to 8.66% on Sharpe ratio basis and standard deviation of the portfolio decreasing from 3.76% to 2.54%. The literature review generally supported the view that real estate has a role in a mixed-asset portfolio. Research topics such as securitized versus unsecuritized real estate, real estate allocation and diversification, returns and risk, inflation hedging, modern portfolio theory and the efficient -frontier were analysed and related to the research report. The empirical analysis supports the hypothesis that real estate provides diversification benefits. The property cycle is positive and it is supported by positive property fundamentals like (vacancies are at lowest levels, capitalisation rates are strengthening, the property cycle is turning positive and a stable interest rate environment). The positive property fundamentals will lead to earnings growth. An allocation of between 10% based on the lower end of the minimum variance and 15% based on the lower end of the risk/return ratio is recommended for a mixed-asset portfolio.

The weight of real estate in a multi-asset portfolio is a highly discussed matter and the main purpose for every investor is to reach an optimal diversification. The aim of the thesis is to apply a new allocation model, which considers market imperfections characterized by real estate. The most known and used method today is the mean-variance approach, founded in the modern portfolio theory. Modern portfolio theory is based on several assumptions, where one of these is the assumption of an efficient market. However, real estate is not considered the be a part of the efficient market due to several market imperfections, such as illiquidity, transaction cost etc. Market imperfection generate risk, which naturally should decrease the optimal weight of real estate in the portfolio. In order to assess optimal real estate allocation in a multi-asset portfolio when accounting for market imperfections an extended approach of the mean-variance model is applied. The extended model accounts for risk-aversion, transaction cost and time-on-market. The model is divided in two approaches, the benchmark- and normative approach, based on two different papers. The model is tested on the Swedish market with current market conditions in order to assess the models applicability. The result from the benchmark approach suggested an optimal real estate allocation of 0,72 –  5,84 %. The normative approach has been dismissed as inconclusive and unreliable due to abnormal weight of real estate. However, the value of risk-aversion is identified as the strongest determinant in both the models.  The allocation from the benchmark approach is, as expected, lower than results of the standard mean-variance approach. The extended model is a useful and valid tool in the consideration of market imperfections characterized by real estate. / Vikten av fastigheter i en portfölj med fler tillgångsslag är en omdiskuterad fråga där det huvudsakliga målet för en investerare är att uppnå en optimal diversifiering. Målet med uppsatsen är att testa en ny allokerings modell som tar hänsyn till fastigheters speciella attribut, marknadsimperfektioner. Den vanligast förekommande metoden för att beräkna allokeringen mot fastigheter härstammar från modern portfolio teori (MPT). MPT baserar på ett antal antaganden, varav ett är en effektiv marknad. Problematiken ligger i att fastigheter inte kvalificerar sig inom ramen för den effektiva marknaden då de delvis räknas som en illikvid tillgång, förenade med höga transaktionskostnader. Marknadsimperfektioner resulterar i risk vilket logiskt leder till en lägre vikt av fastigheter i portföljen. För att utvärdera den optimala allokeringen mot fastigheter i en portfölj med flera tillgångar används en modifierad version av den klassiska medelvariationsmodellen som tar hänsyn till vissa av marknadsimperfektioner. Den modifierade versionen tar hänsyn till risk aversion, transaktionskostnad och försäljningstid. Modellen är uppdelad i två tillvägagångsätt, den ena benämnd som benchmark- och den andra som normativa metoden, baserat på varsin vetenskaplig artikel. Modellen testas på den svenska marknaden under nuvarande marknadsförutsättningar för att bedöma dess tillämplighet. Resultatet från benchmarkmetoden visar på en optimal fastighetsallokering om 0,72 – 5,84 %. Resultaten för den normativa modellen är förkastade som ofullständiga och opålitliga. I vilket fall identifieras riskaversionen som den mest avgörande faktorn i båda modellerna. Benchmarkmetoden ger en lägre optimal allokering gentemot fastigheter, än den som räknas fram med den klassiska mean-variance metoden. Den utvidgade modellen anses vara en användbar och giltig metod som tar hänsyn till de marknadsimperfektioner som fastigheter karaktäriseras av.