1 XIX Inernaional Tor Vergaa Conference on Money, Banking and Finance The role of risk measures choice in ranking real esae funds: evidence from he Ialian marke Claudio Giannoi, Universiy LUM Jean Monne Gianluca Maarocci, Universiy of ome Tor Vergaa ome December 6 h, 00
2 Index Inroducion Lieraure review Empirical analysis: Sample Mehodology esuls Conclusions
3 Inroducion (/) In he asse managemen indusry, he isk Adjused Performance (hereinafer AP) measures are he more well known insrumens used in order o give advice abou he qualiy of an invesmen (Cucurachi, 999). The more widespread measures assumes he hypohesis of normaliy of reurns and provide a judgmen of he qualiy of he invesmen as a raio beween a reurn and a risk index. Empirical analysis proposed in lieraure abou he real esae invesmen vehicle performance demonsrae ha he reurn disribuion is asymmeric (Huson and Sevenson, 008) and is significanly skewned (Lizieri e al. 007).
4 Inroducion (/) esearch quesions - Does he normaliy hypohesis fi for he Ialian real esae funds? - Is here any difference in he ranking consruced using AP measures ha assume he normaliy of reurns and hose ha do no consider his simplified assumpion? - Is here any relaionship beween leverage or volume and he ranking defined using differen AP measures?
5 Index Inroducion Lieraure review Empirical analysis: Sample Mehodology esuls Conclusions
6 Lieraure review (/) The analysis of he performance achieved by lised real esae propery companies and EITS demonsrae a lack of normaliy in he reurn disribuion (Lizieri and Ward, 000) and shows dynamics for reurns achieved ha are no always coheren wih hose achieved by oher financial insrumens (Huson and Sevenson, 008). eal esae invesmen vehicles show frequenly a reurns disribuion wih higher skewness and kurosis respec o oher financial insrumens (Myer and Weeb, 993).
7 Lieraure review (/) The non normaliy of resuls is explained on he basis of he liabiliy srucure ha could defined in order o ensure o he lender a fixed minimum reurn and a premium in some marke scenarios (Ward and French, 997). The performance dynamics of real esae vehicles could be also explained on he basis of he lack of liquidiy ha characerized he markes in which hey are raded (Li e al., 009). There are some evidence for more developed markes (like US) of an increasing number of ransacions and a lowering level of ransacion coss (Jirasakuldech and Knigh, 005) bu hese resuls could be no generalized o he overall world indusry.
8 Index Inroducion Lieraure review Empirical analysis: Sample Mehodology esuls Conclusions
9 Empirical analysis: Sample Fund name Lising dae Asse Under Managemen December 3 h, 009 Alpha immobiliare July 04 h, ,833,83 Alanic June 7 h, ,495,349 Alanic Berenice July 9 h, ,476,570 Bea immobiliare Ocober 4 h, ,87,7 BNL porfolio immobiliare January nd, ,35,443 CAAM E Europa November 7 h, 003,7,779 CAAM E Ialia June 03 rd, ,583,7 Caravaggio May 6 h, ,375,53 Dela Immobiliare March h, ,04,487 Esense Grande Disribuzione Augus 3 rd, ,789,09 Europa Immobiliare December 04 h, 004 4,37,566 Immobilium 00 Ocober 9 h, ,979,669 Inves eal Securiy January 0 s, ,86,908 Invesieico November 0 s, ,844,486 Obelisco June 4 h, ,707,8 Olinda December 09 h, ,305,787 Piramide Globale November 6 h, 00 55,430,399 Polis April 0 h, 00 36,633,48 isparmio immobiliare uno June 04 h, 00 93,088,699 Securfondo Ocober 0 nd, 00 96,575,750 Tecla fondo uffici March 4 h, ,55,749 Unicredi Immobiliare uno June 4 s, ,349,99 Valore Immobiliare globale November 9 h, ,644,6 N of Ialian real esae funds (lised and unlised) 54 AUM of he overall Ialian Marke (lised and unlised) 38,36,900,000 Sample represenaiveness n funds = % of he number of Ialian real esae funds AUM > 38 billions.87% of he AUM of he overall Ialian real esae funds marke
10 Empirical analysis: mehodology (/3) Performance achieved is compued using he following formula: ln P P D Where P is he closing price a ime, D is he dividend evenually paid a ime and ln is he naural logarihm. Normaliy es Shapiro & Wilk We selec o es he usefulness of new AP measure correced for he non-normaliy looking only a hose ha are consruced saring from he f excess reurn respec o a risk free rae
11 Empirical analysis: mehodology (/3) f Sharpe 0,0 max - f f OPS,0 max - f f OAS,0 max - f f Sorino 3 3 3,0 max - f n f Kappa 4 4 4,0 max - f n f Kappa f MDD Calmar n i f MDD n Serling n i f MDD n Burke rf Va Va aio f CVa CVa aio f MVa MVa aio,0 max( f rf f E e Omega Sharpe Omega risk measure Va risk measures MDD risk measures Lower parial momens risk measures Value of each measure (mean) anking persisence over ime () anking correlaion
12 Empirical analysis: mehodology (3/3) AP measures dynamics are sudied using he following approach AP i i LEVEAGE i EFFGES i i 3 i EFFGI i i VOLUME i 4 i HHGEO i i n i 5 i i i ConrolVariables LEVEAGE i 6 i HHSECT AGE i i 7 i i INCSG i i 8 i i MATSG VOLUME i i 9 i BPSF i i i i ln NAV i AP i LEVEAGE i VOLUME i ln NAV i EFFGES i EFFGI i HHGEO i HHSECT i where: = Value of he AP measure considered a he year for he fund I; = he raio beween overall deb a he end of he year for he fund i respec o he oal asse; = he mean volume of daily rades in he year for he fund i; = a conrol variable relaed o he size of he asse under managemen for he fund I a he end of he year i; = managemen efficiency compued as he raio beween fund overall expense and oal asse in he year for he fund i; = acive propery managemen measured as he raio beween propery expense and oal asse in he year for he fund i; = geographical concenraion of asses in he year for he fund i esimaed using he Herfindal for he propery locaion (he geographical classificaion considers 5 areas according o he Assogesioni daa; = secor concenraion of asses in he year for he fund i esimaed using he Herfindal for he propery ype (he secor classificaion considers 8 ypes of asses according o he Assogesioni daa; = years from he saring dae of he fund i a he ime ; = raio beween he managemen fees and he oal asses in he year for he fund i; = dummy variable ha assume he value if he fund i is managed by a SG ha aain o a financial group and 0 oherwise; AGE i INCSG i MATSG i BPSF i = dummy variable ha assume he value if he fund is a blind pool and 0 oherwise. Conrol variables Lee and Morri 009 andom vs fixed effec Hausman specificaion es
13 Empirical analysis: resuls (/4) Shapiro Wilk es of normaliy Alpha immobiliare *** 0.36 *** 9.04 *** *** 9.3 *** *** 6.35 *** *** Alanic *** *** *** 7.75 *** Alaninc Berenice *** 7.56 *** 0.35 *** 7.0 *** 0.8 *** Bea immobiliare 4.8 *** 8.97 *** 6.73 *** *** *** BNL porfolio immobiliare 8.09 *** 5.9 *** 5.37 *** *** 8.85 *** 8.86 *** 5.90 *** 3. *** CAAM E Europa 5.03 *** *** *** 4.59 *** 6.38 *** *** 3.79 *** CAAM E Ialia *** *** 4.64 *** *** *** *** *** 6.36 *** Caravaggio *** 8.46 *** 8.66 *** 7.89 *** *** Dela Immobiliare 9.3 *** Esense Grande Disribuzione.749 *** 5.84 *** *** *** *** 7.86 *** Europa Immobiliare.49 ** 8.36 *** *** 7.96 *** Immobilium *** 7.90 *** 8.6 *** *** 7.8 *** 8.50 *** *** Inves eal Securiy *** *** *** *** 9.08 *** Invesieico.4 ** 5.56 *** 7.07 *** 8.45 *** *** *** Obelisco *** *** *** 8.49 *** Olinda *** 9.47 *** *** *** 4.9 *** Piramide Globale.404 *** 5.84 *** 8.95 *** *** ***.380 ***.596 *** 0.33 *** Polis.30 ** 6.35 *** *** 5.87 *** 8.7 *** *** 5.0 *** *** *** isparmio immobiliare uno *** Securfondo *** 8.35 *** 7.35 *** 7.76 *** *** 9.05 *** *** *** 8.09 *** Tecla fondo uffici *** *** 9.86 *** *** *** *** Unicredi Immobiliare uno.564 *** *** 6.53 *** 6.46 *** *** 8.36 *** *** *** *** Valore Immobiliare globale 3.9 *** 7.85 *** *** *** 5.99 *** 7.5 *** *** *** *** Noes: *** es significan a 99% level ** es significan a 95% level * es significan a 90% level
14 Empirical analysis: resuls (/4) Alpha immobiliare Alanic Alanic Berenice Bea immobiliare BNL porfolio immobiliare CAAM E Europa CAAM E Ialia Caravaggio Dela Immobiliare Esense Grande Disribuzione Europa Immobiliare Immobilium 00 Inves eal Securiy Invesieico Obelisco Olinda Piramide Globale Polis isparmio immobiliare uno Securfondo Tecla fondo uffici Unicredi Immobiliare uno Valore Immobiliare globale Sharpe OPS OAS Sorino Kappa (n=3) Kappa (n=4) Calmar Serling Burke Va aio CVa raio MVa raio Sharpe Omega M M M M M M M M M M M M M M M M M M M M M M M Noes: M = Mean ranking posiion = Sandard deviaion of ranking posiions in he ime period = Sandard deviaion could no be compued because he fund is lised only for one year AP measures summary saisics Saisics on S.Dev. An example Media Max Min Sharpe OPS OAS Sorino Kappa (n=3) Kappa (n=4) Calmar Serling Burke Va aio CVa raio MVa raio Sharpe Omega Mean posiion S. Dev. Sharpe OPS OAS Sorino Kappa (n=3) Kappa (n=4) Calmar Serling Burke Var raio Cvar raio Mvar raio Sharpe Omega Legenda Same mean ranking respec o Sharpe raio Mean difference of one ranking posiion Mean difference of more han one ranking posiion
15 Empirical analysis: resuls (3/4) Correlaion among rankings Sharpe OPS OAS Sorino Kappa (n=3) Kappa (n=4) Calmar Serling Burke Va aio CVa raio MVa raio Sharpe Omega Sharpe OPS OAS Sorino Kappa (n=3) Kappa (n=4) Calmar Serling Burke Va raio CVa raio MVa raio Sharpe Omega Mean Max.0000 Min Mean Max Min Mean Max Min Mean Max Min Mean Max Min Mean Max Min Mean Max Min Mean Max Min Mean Max Min Mean Max Min Mean Mean correlaion wih Sharpe index: 60% Max range of variaion 60% Max Min Mean Max Min Mean Max Min
16 Empirical analysis: resuls (4/4) E Sharpe OPS OAS Sorino Kappa (n=3) Kappa (n=4) Calmar Serling Burke Va aio CVa raio MVa raio Sharpe Omega Leverage 9.5 ** * * 4.7** 9.69* * 0.0** 0.09* 50.3* Volume ln(nav) EFFGES * EFFGI * INCSG HHGEO.9 * HHSETT AGE * * MATSG BPFS.49 ** ** **.77** 6.** * 0.0* ** Cons Obs Groups wihin beween overall Chi Prob > Chi ho
17 Index Inroducion Lieraure review Empirical analysis: Sample Mehodology esuls Conclusions
18 Conclusions The choice of risk measures more complee respec o he sandard deviaion affecs no only he yearly ranking posiion of each fund bu also he variabiliy of rankings over ime. Measures consruced on disribuion of losses, on he maximum drawdown and looking also a he asymmery of reurns allow o achieve he highes level of raking persisence over ime. Using some conrolling variables for he characerisics of he real esae invesmen fund, he leverage has a significan impac on he ranking posiion defined using differen AP measures while volume seems o be no relevan
19 Thanks for your aenion Conacs Claudio Giannoi Universiy of LUM Jean Monne Gianluca Maarocci Universiy of ome Tor Vergaa