The role of risk measures’ choice in ranking real estate funds: evidence from the Italian market Claudio Giannotti, University LUM Casamassima, Bari Gianluca Mattarocci, University of Rome “Tor Vergata” Milano – June 23 td -26 th, 2010

 Introduction  Literature review  Empirical analysis:  Sample  Methodology  Results  Conclusions Index

Introduction (1/2)  In the asset management industry, the Risk Adjusted Performance (hereinafter RAP) measures are the more well known instruments used in order to give advice about the quality of an investment (Cucurachi, 1999). The more widespread measures assumes the hypothesis of normality of returns and provide a judgment of the quality of the investment as a ratio between a return and a risk index.  Empirical analysis proposed in literature about the real estate investment vehicle performance demonstrate that the return distribution is asymmetric (Hutson and Stevenson, 2008) and is significantly skewned (Lizieri et al. 2007).

Introduction (2/2) Research questions -Does the normality hypothesis fit for the Italian real estate funds’? -Is there any difference in the ranking constructed using RAP measures that assume the normality of returns and those that do not consider this simplified assumption? -Is there any relationship between leverage or volume and the fitness of the RAP measures?

Index  Introduction  Literature review  Empirical analysis:  Sample  Methodology  Results  Conclusions

Literature review (1/2)  The analysis of the performance achieved by listed real estate property companies and REITS demonstrate a lack of normality in the return distribution (Lizieri and Ward, 2000) and shows dynamics for returns achieved that are not always coherent with those achieved by other financial instruments (Hutson and Stevenson, 2008).  Real estate investment vehicles show frequently a returns’ distribution with higher skewness and kurtosis respect to other financial instruments (Myer and Weeb, 1993).

Literature review (2/2)  The non normality of results is explained on the basis of the liability structure that could defined in order to ensure to the lender a fixed minimum return and a premium in some market scenarios (Ward and French, 1997).  The performance dynamics of real estate vehicles could be also explained on the basis of the lack of liquidity that characterized the markets in which they are traded (Li et al., 2009). There are some evidence for more developed markets (like US) of an increasing number of transactions and a lowering level of transaction costs (Jirasakuldech and Knight, 2005) but these results could be not generalized to the overall world industry.

Index  Introduction  Literature review  Empirical analysis:  Sample  Methodology  Results  Conclusions

Empirical analysis: Sample Fund nameListing date Asset Under Management December 31 th, 2009 Alpha immobiliare July 04 th, ,833,183 € Atlantic 1 June 7 th, ,495,349 € Atlantic 2 – Berenice July 19 th, ,476,570 € Beta immobiliare October 24 th, ,287,272 € BNL portfolio immobiliare January 2 nd, ,315,443 € CAAM RE Europa November 17 th, ,227,779 € CAAM RE Italia June 03 rd, ,583,711 € Caravaggio May 16 th, ,375,253 € Delta Immobiliare March 11 th, ,204,487 € Estense Grande Distribuzione August 3 rd, ,789,091 € Europa Immobiliare 1 December 04 th, ,237,566 € Immobilium 2001 October 29 th, ,979,669 € Invest Real Security January 01 st, ,286,908 € Investietico November 01 st, ,844,486 € Obelisco June 14 th, ,707,118 € Olinda December 09 th, ,305,787 € Piramide Globale November 26 th, ,430,399 € Polis April 20 th, ,633,481 € Risparmio immobiliare uno June 04 th, ,088,699 € Securfondo October 02 nd, ,575,750 € Tecla fondo uffici March 4 th, ,515,749 € Unicredit Immobiliare uno June 4 st, ,349,929 € Valore Immobiliare globale November 29 th, ,644,612 € N° of Italian real estate funds (listed and unlisted)154 AUM of the overall Italian Market (listed and unlisted)38,316,900,000 € Sample representativeness n° funds = % of the number of Italian real estate funds AUM > 38 billions euros 21.87% of the AUM of the overall Italian real estate funds market

 Performance achieved is computed using the following formula: Empirical analysis: methodology (1/2) Where P t is the closing price a time t, D t is the dividend eventually paid at time t and ln is the natural logarithm. Normality testShapiro & Wilk We select to test the usefulness of new RAP measure corrected for the non-normality looking only at those that are constructed starting from the excess return respect to a risk free rate

Empirical analysis: methodology (2/2) Omega risk measure VaR risk measuresMDD risk measures Lower partial moments risk measures Value of each measure Ranking correlation Ranking peristence

Empirical analysis: results (1/6) Shapiro – Wilk test of normality Alpha immobiliare *** *** *** *** *** *** *** *** Atlantic *** *** *** *** Atlantinc 2 - Berenice *** *** *** *** *** Beta immobiliare *** 8.97 *** *** *** *** BNL portfolio immobiliare *** *** *** *** *** *** *** *** CAAM RE Europa *** *** *** *** *** *** *** CAAM RE Italia *** *** *** *** *** *** *** 6.36 *** Caravaggio *** *** *** *** *** Delta Immobiliare *** Estense Grande Distribuzione *** *** *** *** *** *** Europa Immobiliare ** *** *** *** Immobilium *** *** *** *** *** *** *** Invest Real Security *** *** *** *** 9.08 *** Investietico ** *** *** *** *** *** Obelisco *** *** *** 8.49 *** Olinda *** *** *** *** *** Piramide Globale *** *** *** *** *** *** *** *** Polis2.130 ** *** *** *** 8.72 *** *** *** *** *** Risparmio immobiliare uno *** Securfondo4.645 *** *** *** *** *** *** *** *** *** Tecla fondo uffici *** *** *** *** *** *** Unicredit Immobiliare uno2.564 *** *** *** *** *** *** *** *** *** Valore Immobiliare globale3.291 *** *** *** *** *** *** *** *** *** Notes: *** test significant at 99% level ** test significant at 95% level * test significant at 90% level

Empirical analysis: results (2/6) Correlation among rankings SharpeROPSROASSortinoKappa (n=3)Kappa (n=4)CalmarSterlingBurkeVaR RatioCVaR ratioMVaR ratioSharpe Omega Sharpe Mean Max Min ROPS Mean Max Min ROAS Mean Max Min Sortino Mean Max Min Kappa (n=3) Mean Max Min Kappa (n=4) Mean Max Min Calmar Mean Max Min Sterling Mean Max Min Burke Mean Max Min VaR ratio Mean Max Min CVaR ratio Mean Max Min MVaR ratio Mean Max Min Sharpe Omega Mean Max Min Max range of variation 60% Mean correlation with Sharpe index: 60%

Empirical analysis: results (3/6) Correlation among rankings (breakdown by leverage) SharpeROPSROASSortino Kappa (n=3) Kappa (n=4) CalmarSterlingBurke VaR Ratio CVaR ratio MVaR ratio Sharpe Omega Sharpe H L ROPS H L ROAS H L Sortino H L Kappa (n=3) H L Kappa (n=4) H L Calmar H L Sterling H L Burke H L VaR ratio H L CVaR ratio H L MVaR ratio H L Sharpe Omega H L Note: H = Funds with leverage at least equal to the mean value L = funds with leverage lower than the mean value Correlation of rankings for highly leveraged funds lower in the 77% of cases

Empirical analysis: results (4/6) Correlation among rankings (breakdown by volume) SharpeROPSROASSortino Kappa (n=3) Kappa (n=4) CalmarSterlingBurke VaR Ratio CVaR ratio MVaR ratio Sharpe Omega Sharpe H L ROPS H L ROAS H L Sortino H L Kappa (n=3) H L Kappa (n=4) H L Calmar H L Sterling H L Burke H L VaR ratio H L CVaR ratio H L MVaR ratio H L Sharpe Omega H L Note: H = Funds with volume at least equal to the mean value L = funds with volume lower than the mean value Funds less traded are frequently characterized by less coherence of rankings

Empirical analysis: results (5/6) Persistence analysis Time horizon Sharpe Overall HL LV ROPS Overall HL LV ROAS Overall HL LV Sortino Overall HL LV Kappa (n=3) Overall HL LV Kappa (n=4) Overall HL LV Notes: Funds are classified on the basis of the leverage and the volume identifying as high leverage the funds with leverage at least equal to the mean value and as low volume the funds with a volume of trade lower respect to the mean

Empirical analysis: results (6/6) Persistence analysis Time horizon Calmar Overall HL LV Sterling Overall HL LV Burke Overall HL LV VaR Ratio Overall HL LV CVaR ratio Overall HL LV MVaR ratio Overall HL LV Sharpe Omega Overall HL LV Notes: Funds are classified on the basis of the leverage and the volume identifying as high leverage the funds with leverage at least equal to the mean value and as low volume the funds with a volume of trade lower respect to the mean

Index  Introduction  Literature review  Empirical analysis:  Sample  Methodology  Results  Conclusions

The choice of risk measures more complete respect to the standard deviation affects not only the yearly ranking position of each fund but also the variability of rankings over time. Measures constructed on distribution of losses, on the maximum drawdown and looking also at the asymmetry of returns allow to achieve the highest level of raking persistence over time. Especially for less traded funds and/or highly leveraged ones, approaches normally adopted for analyzing the asset management industry had so to be revised in order to consider the specific characteristics of the real estate investment that do not allow to simplify the performance analysis assuming the normality of returns distribution. Conclusions

Claudio Giannotti University of LUM Casamassima Gianluca Mattarocci University of Rome Tor Vergata Contact information