Curren Research Quesions APRA Research Mehodology for Analysis of Superannuaion Funds Wha are he deerminans of he cross-secional variaion in superannuaion reurns? Asse allocaion, manager skill, expenses/axes % 11.0 10.0 Karina Ellis Based on work by Alan Tobin, Belinda Tracey, Wilson Sy, Chris Inman and Karina Ellis 9.0 8.0 7.0 6.0 5.0 Fund A Fund B Fund C 1 Ne reurn Expenses Benchmark reurn 2 Previous research by APRA Moivaion for new APRA research The impac of agency coss on he invesmen performance of Ausralian pension funds Coleman, Esho and Wong (2006) Celebraing 10 years of superannuaion daa collecion 1996-2006 APRA Insigh Special Ediion (2007) Reques by he Council of Financial Regulaors APRA was able o use sauory powers o collec addiional daa for prudenial purposes in he form of a governance survey and a invesmen performance survey To gain a beer undersanding of daa problems from fund perspecive Curren research projec: How can we improve he mehodology and daa o suppor or refue prior APRA resuls? 3 4 1
Previous research on superannuaion reurns Previous research on superannuaion coss Early sudies: Allen, Brailsford, Bird and Faff (2002) survey Gallagher (2001) Drew, Sanford, Veeraraghavan (2003); Drew, Sanford (2003) Langford, Faff, Marisey (2006) Holmes and Faff (2008) Leaving aside how performance differs across funds, explici informaion on various fees provides anoher daa source o be examined Problem: no all fees are capured in adminisraion coss and direc invesmen coss Higher fees may indicae ha invesors are geing higher benefis no only in reurns, bu in oher producs and services 5 6 Criicisms of previous APRA research New research objecives Comparing reurns a he fund level may no be a direc comparison since: Some funds may have annuiies, capial guaranee producs, ec. in holdings DB and DC funds included Produce survey daa ha will ry o answer he following quesions: A he invesmen opion level (raher han whole fund level) wha are he deerminans of superannuaion reurns? Holding accoun balance consan, wha are superannuaion reurns? Some funds may have differen clienele base and hus asse allocaions would be differen and migh affec resuls Some funds may have lower average accoun balances resuling in more adminisraive coss ha erode reurns 7 8 2
The survey sudy Invesmen Performance Survey Daa In lae 2006 surveys were disribued o 197 funds wih more han $200 million in asses under managemen as of 30 June 2005 Invesmen performance survey focus is on DC funds and DB/DC funds were asked o supply informaion regarding DC componen Examinaion of he governance survey was compleed firs (APRA Insigh 2008) followed by he invesmen performance survey daa (o be released publicly in several weeks) General Informaion Asse Allocaion of he Defaul Invesmen Opion Asse Allocaion of he Toal Fund Financial Performance of he Defaul Invesmen Opion Financial Performance of he Toal Fund 9 10 Invesmen Performance Survey Daa con. Represenaive Invesor A: Individual who invess in he Defaul Invesmen Opion Mehodology: Self-Benchmark Use he asse allocaion weighs for he defaul invesmen opion Muliply weighs by reurns on asse class indices Represenaive Invesor B: Individual who invess in he Defaul Invesmen Opion via an Employer Maser Trus Fees for Represenaive Invesor Uni Prices / Crediing Raes Asse Class Cash Ausralian Fixed Income Inl Fixed Income (unhedged) Inl Fixed Income (hedged) Ausralian Equiy Inl Equiy (unhedged) Inl Equiy (hedged) Lised Propery Benchmark Index UBS Bank Bill Index UBS Composie Bond Index All Mauriies JP Morgan World ex-aus $A JP Morgan World ex-aus Hedged (raded) S&P/ASX 200 Merged Accumulaion Index MSCI TR Ne World Ex-Ausralia Local MSCI TR Ne World Ex-Ausralia USD S&P/ASX 200 Propery Merged Accumulaion Index Unlised Propery Ausralian Mercer Unlised Propery Funds Index Pre-Tax 11 Oher UBS Bank Bill Index 12 3
Mehodology: Self-Benchmark 10 = j= 1 j j Benchmark Reurn weigh index reurn The self-benchmark provides he reurn ha is driven by he asse allocaion of he defaul invesmen opion How imporan are asse allocaion differences for explaining he cross-secional differences in reurns? -es of benchmark reurns: each fund ype versus he enire sample If asse allocaion differences do explain he under-performance of reail funds versus oher fund ypes we would expec o see a significan -es resul Gross Reurns: Acive Manager Skill ( ) Gross Reurn = Gross Reurn Benchmark Reurn +Benchmark Reurn = (Acive Skill Embedded Coss ) + Benchmark Reurn = Invesmen Value Added + Benchmark Reurn Gross Earnings ha are repored are no free from embedded coss We only have an imperfec measure of acive manager skill, ne of embedded coss How imporan are acive manager skills? -es of Invesmen Value Added: each fund ype versus he enire sample 13 14 Ne Reurns: Expenses and Taxes Ne Reurn = Gross Reurn - Expenses - Taxes = Invesmen Value Added + Benchmark Reurn- Expenses - Taxes = Ne Invesmen Value Added + Benchmark Reurn Wha is he Ne Invesmen Value Added, including acive skill, embedded coss and axes plus explici coss and axes -es of Ne Invesmen Value Added: each fund ype versus he enire sample If Benchmark Reurns are comparable in he cross-secion hen we can also direcly compare ne reurns of all funds Improvemens in he survey sudy Only focus on DC funds Only focus on one represenaive invesmen opion Conrol for asse allocaion differences Examine he componens of reurns: acive skill, asse allocaion, expenses and axes Remaining poenial problem: expenses of an invesmen opion may be driven by he average accoun balance in he fund Wih Represenaive Invesor we also conrol for accoun balance and compare he performance of he same dollar invesmen in he defaul invesmen opion 15 16 4
Improvemens in he survey sudy con. If benchmark reurns are no saisically differen, hen asse allocaion differences canno explain APRA s prior resuls If he represenaive invesor shows he same cross-secional resuls, hen average accoun balances canno explain APRA s prior resuls If he same resuls are found wih only DC invesmens, hen he DC/DB mix canno explain APRA s prior resuls Remaining problems: shor hisory of daa (5 years) and infrequen (annual) daa 17 5