Stock market performance and pension fund investment policy: Rebalancing, free float, or market timing?

Transcription

1 Pelmnay no o be quoed Sock make pefomance and penson fund nvesmen polcy: Rebalancng fee floa o make mng? XAS MFA 2008 Sepembe Absac hs pape s he fs ha examnes he mpac of sock make pefomance on he nvesmen polcy of penson funds. We fnd ha sock make pces nfluence he asse allocaon of Duch penson funds n wo ways. In he sho em oupefomance of eques ove bonds and ohe nvesmen caegoes would auomacally esul n hghe equy pofolo allocaons (and vce vesa) as penson funds do no connuously ebalance he asse allocaon. ach quae penson funds ebalance on aveage aound 39 pecen of excess equy euns leavng 6 pecen fo fee floang. In he medum em oupefomance of eques would nduce penson funds o ncease he saegc equy allocaon (and vce vesa). hese fndngs sugges ha he nvesmen polcy of penson funds s paally dven by he cyclcal pefomance of he sock make. Penson funds espond asymmecally o sock make shocks: ebalancng s much songe afe negave equy euns. On aveage boh he make mng saegy poposed by he penson fund admnsaon and he make mng execued by nvesmen manages led o negave excess euns ove he peod unde consdeaon. Invesmen polces of lage funds devae fom ha of small funds: hey have moe equy and he equy allocaon s much songe affeced by acual equy euns eflecng less ebalancng. he lages funds eac vey asymmecal o excess equy euns and adjus he pofolo wh sgnfcanly moe ha 00% eflecng oveshoong of fee floang o negave ebalancng. Appaenly manages of lage funds apply a de you pofs cu you losses saegy. JL Classfcaon Codes: G G23 Keywods: Penson funds euns pofolo choce excess euns saegc equy allocaon sze effecs asymmecal behavo;

2 2 Inoducon he opmal equy allocaon of penson funds s subjec o consdeable debae. A hgh pecenage of asses nvesed n eques esuls n sgnfcan exposue of penson wealh o flucuaons n sock make pces. Nomnal defned-benef penson lables ae bes esembled by bond euns whle consdeable equy holdngs may be opmal when ndexaon of benefs s conngen on he fundng ao of he penson fund. he sks of equy holdngs sufaced dung he nnees when abundan equy euns led o conbuon holdays and subsequenly n he couse of he fall of sock make pces dung whch esuled n lage penson funds losses. hs epsode ased a debae on nvesmen saeges of Duch penson funds and paculaly on he exposue o equy makes. he nvesmen saegy of Duch penson funds s of key mpoance o socey as nvolves moe han 600 bllon n asses o ove pe nhaban. he way n whch hese asses ae nvesed may sgnfcanly affec he level of equed pemums o fnal benefs. A one pecen lowe annual eun ove he lfecycle of a ypcal woke anslaes no 27 pecen lowe accumulaed penson asses. 2 Consequenly one of he mos mpoan esponsbles of penson fund s usees s o maxmze he expeced eun a an accepable level of sk e.g. measued n ems of he pobably of undefundng. hs sudy nvesgaes whehe sock make pefomance ha devaes fom ha of he ohe asse caegoes nfluences penson funds nvesmen polcy. In pacula we examne wo ways n whch sock make pefomance mpac he equy allocaon of penson funds: () n he sho em as a esul of make mng o mpefec ebalancng and () n he medum em as a esul of adjusmens o he saegc asse allocaon. Consecuvely we examne he value added of acve make mng decsons. hs s paculaly elevan fo compulsoy ndusy wde penson funds. If he pefomance of hese funds measued as a Nomnal defned benef penson lables ae bes esembled by nomnal govenmen bonds. Insead defned benef penson lables ha ae fully ndexed o pces ae bes esembled by Inflaon Lnked Bonds. In many Duch defned benef penson deals ndexaon s conngen on he fundng ao of he penson fund. he make value of hs conngen ndexaon can be deved usng opon pcng heoy. In hs case mgh be opmal o have consdeable equy exposue see e.g. Boedes (2006). 2 he hee majo componens deemnng he coss of penson povsonng ae he qualy of he penson scheme he ae of eun on nvesmens and admnsave and nvesmen coss (see also Bkke and de Deu 2007).

3 3 fve yea movng aveage dops below a cean heshold he afflaed companes can call fo dspensaon wh jonng ha seco fund. 3 able pesens he allocaon of he asses of Duch penson funds ove he followng fve boad classes: ques Bonds Real sae Cash and Ohe Asses. Penson fund nvesmen polcy ncludes he saegc asse allocaon decson whch efes o he popoonal combnaon of asses based on expeced aes of eun and (co)vaance chaacesc fo each asse class. Of he afoemenoned asse classes eques have he hghes expeced eun bu also have he hghes volaly. Moeove s he lages asse caegoy. Consequenly he equy allocaon s one of he key polcy vaables deemnng he sk-eun pofle of a penson fund. able : Penson fund saegc and acual asse allocaon 999:I 2006:IV (n %) Asse classes Aveage saegc asse allocaon Sandad devaon Aveage acual asse allocaon Sandad devaon ques Bonds Real sae Cash 0 Ohe 8 3 oal Noes: he asse shaes ae aveages ove de Duch penson funds and weghed by oal nvesmens. Souce: De Nedelandsche Bank. Penson funds geneally deemne he saegc asse allocaon usng asse lably managemen sudes n whch hey consde long-em expeced euns eun vaances and covaances of boad asse classes gven he sze and chaacescs of he penson lables see e.g. Campbell and Vcea (2002). 4 he saegc asse allocaon s ypcally long em n naue usually wh a hee o fve yea hozon. Fo many penson funds he saegc asse allocaon ncludes bandwdhs fo he acual asse allocaon o df. hese bandwdhs ae chosen n such way ha he maxmum ex ane ackng eo does no exceed a gven heshold. hs ackng eo () s usually defned as = w' Σw whee w s he veco of acual pofolo weghs mnus he veco of saegc pofolo allocaons and Σ s he vaance-covaance max. As nvesmen oppounes change ove me devaons n expeced euns fom he long-em aveages may waan changes n he nvesmen mx. 5 Choosng acual pofolo weghs ha devae 3 hs s known as he Sa-es whch s based on he so-called z-scoe. he heshold s based upon a efeence pofolo usually he saegc asse allocaon of he penson fund. 4 Shefn and Saman (997) use behavoual fnance heoy o explan he asse allocaon of penson funds. hey ague ha nvesos buld pofolos as pyamds of asses laye by laye. In conas o mean-vaance heoy covaance beween asse classes ae geneally gnoed esulng n subopmal pofolos. 5 Pedcably n expeced asse euns may affec he opmal pofolo choce of nvesos wh long hozons (see e.g. Babes 2000; Paso and Sambaugh 200).

4 4 fom he saegc asse allocaon s known as accal asse allocaon o make mng. Make mng mples he akng of sho-em (nfomed) bes on he elave euns of asse classes. Make mng can be mplemened hough acually buyng and sellng he undelyng secues alhough n pacce devaves ae also commonly used as an effcen manne o change he asse allocaon. Howeve he poenal exa eun hough make mng s lmed as also ndcaed by he so-called fundamenal law of acve managemen see Gnold and Kahn (999). 6. hs sudy examnes he mpac of hghe o lowe socks euns compaed o he ohe asse caegoes on he equy allocaon of penson funds. o he bes of ou knowledge hs s he fs sudy ha examnes hs elaonshp. Fgue shows he vaous facos ha nfluence he equy allocaon of penson funds. Ove he long em he equy allocaon s deemned by he saegc asse allocaon. Howeve seveal facos nfluence asse allocaon n he sho o medum em. We dsngush he followng hee equy nvesmen saeges ha penson funds may use o espond o posve o negave sock make euns: ebalancng fee floa and make mng. Fgue : he mpac of sock make pefomance on equy allocaon Sock Make Pefomance Make mng/ Fee floang Rebalancng Saegc asse allocaon quy Allocaon 6 hs law saes ha he nfomaon ao equals he nfomaon coeffcen mes he squae oo of he numbe of ndependen nvesmen decsons. he nfomaon ao s he sk-adjused excess eun ove a passve nvesmen saegy. An nfomaon ao of 0.5 whch s consdeed as hgh eques ha he asse manages ean a 50 bass pons excess eun ( alpha ) pe pecen of esdual sk on a yealy bass. he nfomaon coeffcen measues he skll of he asse manage and s defned as he coelaon beween hs foecass on nvesmen euns and he acual oucomes. he numbe of ndependen nvesmen decsons s fou f he penson fund makes quaely make mng decsons. o geneae a make mng nfomaon ao of 0.5 eques n ha case an nfomaon coeffcen of 0.25 whch s consdeed as exemely hgh. I would eque he asse manage o foecas he decon of he sock make coecly 63 ou of 00 mes. heefoe he poenal added value of make mng s lmed.

5 5 Rebalancng efes o he nvesmen pocess ha ensues ha he acual equy allocaon connuously equals he saegc equy allocaon whch mples sellng eques afe elave hgh sock make euns and buyng afe elave low equy euns. Insead we use fee floa o ndcae a passve nvesmen saegy n whch penson funds allow he equy allocaons o df wh make developmens. 7 Fnally as menoned above make mng efes o he hghe o lowe weghng of eques (o ohe asse classes) elave o he saegc asse allocaon polcy of he penson fund. Noe ha n he case of absence of any equy ades s dffcul o dsngush beween fee floa (passve managemen) and make mng (acve managemen) as allowng he asse allocaon o df could be seen as a knd of acve nvesmen decson. A numbe of sudes show ha saegc asse allocaon domnaes pofolo pefomance. In pacula hs saegc asse allocaon s shown o explan moe han 90 pecen of he vaably n penson fund euns ove me whle he addonal vaaon explaned by make mng s less han 5 pecen (Blake Lehmann and mmemann 999; Bnson e al ; and Ibboson and Kaplan 2000). 8 Moeove n lne wh he effcen make heoy evdence shows ha penson funds ae no successful n explong make mng o geneae excess euns. In pacula make mng s shown o cause an aveage loss of bass pons pe yea (Blake e al. 999; Bnson e al and Danel e al. 997). Whle a numbe of empcal sudes examne he mpac of nvesmen polcy on euns 9 vey few papes nvesgae he mpac of make developmens on nvesmen polcy. Blake e al. (999) and Kakes (2006) epo a negave coelaon beween asse class euns and ne cash flows o he coespondng asse class whch pons o ebalancng. Howeve Blake e al. (999) fnd also ha he asse allocaon fo UK penson funds dfs owad asse classes ha pefomed elavely well n lne wh a fee-floa saegy. Appaenly UK penson funds only paly ebalanced he nvesmens n 7 Penson funds can ebalance connuously; heeby ensung ha he asse allocaon always equals he saegc asse allocaon. Howeve penson funds ae also known o use ebalancng saeges whch have some fee floa componen. xamples nclude calenda ebalancng wheeby penson funds ebalance he pofolo back o s saegc weghs a egula nevals and band ebalancng wheeby penson funds ceae bands aound each asse class and ebalance he pofolo as soon as one asse class beaches s band. 8 A shocomng of hese sudes s ha he (aveage) dffeence beween acual and saegc asse allocaon s no epoed. hs dffeence beween acual nvesmens and nvesmen polcy s lkely o be elaed o he pecenage of eun vaably ha can be abued o make mng. 9 he leaue nvesgang he effecveness of sock pckng and make mng n mpovng nvesmen pefomance s exensve. Mos sudes focus on US muual funds and fnd ha fund manages ae no able o explo selecvy and mng o geneae excess euns (e.g. Fama (972) Henksson and Meon (98) Kon and Jen (979) and Kon (983)). he semnal pape by Lakonshok e al. (992) examnes he US money managemen ndusy and shows ha nvesmens by penson funds have conssenly undepefomed he make. he auhos abue hs fndng o exa layes of agency poblems n he penson fund ndusy. Howeve hs fndng s conesed n a ecen pape by Baue e al. (2007) who show ha penson funds pefom close o he benchmaks whle muual funds songly undepefom. he lae auhos ague ha penson funds ae less exposed o hdden agency coss han muual funds snce hey have bee negoang powe and monong capacy and moe effcenly pool he funds.

6 6 esponse o dffeen euns acoss asse caegoes. Hence he degee of ebalancng vesus fee floa n penson fund asse allocaon emans an open queson. Agnew e al. (2003) epo ha equy allocaons of pacpans n 40(k) plans ae posvely elaed o he pevous day s equy eun (feedback adng). Howeve no sgnfcan coelaon s found beween changes n equy allocaons and euns ove he followng hee days suggesng he absence of make-mng ables. hs pape uses quaely daa fom he majo Duch penson funds ove he 999:I 2006:IV peod. Fgue 2 gves a pevew of he empcal esuls depcng he saegc and he acual equy allocaon fo Duch penson funds as well as he MSCI Wold Index. hee paens sand ou fom hs fgue. Fs he acual equy allocaon ends o have a smla paen as he MSCI Wold Index bu wh some eveson o he saegc equy allocaon. Geneally he acual equy allocaon nceases when he sock make goes up and vce vesa. he man explanaon fo hs paen s ha penson funds only paly ebalance he asse allocaon n esponse o changes n he value of he equy pofolo. Fgue 2: Sock make euns and equy nvesmens (999:I 2006:IV) :Q 999:Q3 2000:Q 2000:Q3 200:Q 200:Q3 2002:Q 2002:Q3 2003:Q 2003:Q3 2004:Q 2004:Q3 2005:Q Pofolo nvesmens n equy (%) 2005:Q3 2006:Q 2006:Q MSCI Wold ndex quy nvesmen polcy (%) quy nvesmens (%) MSCI Wold ndex Second Fgue 2 pons o neacon beween sock make pefomance and saegc asse allocaon. he saegc equy allocaon appeas o follow he pefomance of he equy make alhough wh a me lag. Followng he sock make boom n he second half of he 990s he saegc equy allocaon nceased unl he end of 200 bu deceased fom 2002 o 2003 n

7 7 esponse o he sock make pces declne sang n Geneally he saegc equy allocaon s adjused gadually and wh a subsanal delay n esponse o sock make movemens. hd he fgue suggess ha penson funds may have los money fom make mng ove he busness cycle. hey seem o have gadually nceased he equy allocaon unl he downun of he sock make was well unde way so ha hey encouneed elavely lage losses. Convesely penson funds have no sgnfcanly nceased he equy allocaon pofolo nvesmens o ake full pof fom he subsequen upwad end of he sock make. he sucue of hs pape s as follows. Secon 2 pesens he daa used n he analyses. Secon 3 nvesgaes he nfluence of make movemens on asse allocaon wheeas ebalancng s moe closely examned n Secon 4. he nex secon analyses he elaonshp beween sock make euns and saegc asse allocaon. he queson whehe penson funds ae successful n adjusng equy allocaons n ancpaon of expeced make movemens s exploed n Secon 6. Fnally he las secon summazes and concludes. 2 Descpon of he daa We use a dealed daase wh quaely nfomaon on 748 Duch penson funds fo he 999:I 2006:VI peod. he daa s fom De Nedelandsche Bank whch s esponsble fo he pudenal supevson of penson funds and he egulaoy complance. Fo each penson fund daa s avalable on he saegc asse allocaon asse sales and puchases he make value of nvesmens n dffeen asse classes and he me-weghed euns. 0 he sample s an unbalanced panel as no all penson funds epo daa fo he ene sample peod due o new enans meges emnaons and epong falues. Snce ou am s o sudy asse allocaon ove me we exclude penson funds wh daa fom less han wo yeas. Fnally we exclude nconssen obsevaons and obsevaons wh clea epong eos. Ou fnal sample ncludes 748 penson funds epesenng aound 85% pecen of oal penson fund asses n he Nehelands. able 2 pesens summay sascs on he penson funds nvesmens of ou sample ove 999:I 2006:IV. he penson funds n he sample dvege enomously: he smalles penson fund has asses woh less han mllon whle he lages fund has asses of moe han 0 We calculae he aveage euns n hee seps. Fo each penson fund we fs calculae he aveage of s log euns ove me. Nex we conve hese aveages o smple euns. Fnally we calculae he aveage of hese smple euns acoss penson funds. hs pocedue avods a dsoon n calculaed aveage euns esulng fom he fac ha he log of an aveage s no equal o he aveage of logs. We also an egessons fo a balanced sample of only 382 penson funds ha epoed a leas seven yeas of daa. he egesson esuls wee smla as hose epoed n ables 5-9 suggesng ha he so-called suvvoshp bas s no a sgnfcan poblem.

8 8 200 bllon. he aveage and medum sze of penson fund asses equal 799 mllon and 53 mllon especvely. We dsngush beween sze classes and ypes of penson funds and beween ypes of penson plans. Small funds end o have elavely less nvesmens n equy compaed o lage funds and moe n bonds eflecng lowe sk appee. Alhough lage n numbe small funds have only a mno shae of all penson fund nvesmens. able 2: Invesmens acoss sze classes and penson fund ypes (999:I 2006:IV) Numbe Aveage Aveage Aveage Max mn Max mn Sze classes based on of oal bond equy equy saegc equy oal nvesmens penson nvesmens nvesmens nvesmens nvesmens allocaon ove Invesmen (mln euo) funds (mln euo) (%) (%) ove me (%) me (%) gap (%)* Column () (2) (3) (4) (5) (6) (7) 0-00 (Small) (Medum) >000 (Lage) Aveage / oal ype of penson fund Indusy (all) Compulsoy Non-Compulsoy Company Pofessonal goup Plan ype DB DC Noes: All sascs ae aveages weghed by oal nvesmens excep fo he fs and las column (Aveage oal nvesmens and Numbe of penson funds). * Invesmen gap s he dffeence beween he saegc equy allocaon and he acual pecenage of equy pofolo nvesmens. Ou sample ncludes 528 company funds 94 ndusy-wde funds and 0 pofessonal goup funds. 2 Compulsoy ndusy funds ae lages n ems of nvesmens. Company funds and pofessonal goup funds have moe bonds han ohe ypes of funds and a faly equal shae of equy eflecng hghe sk aveson. he dffeence s manly due o eal esae nvesmens. On aveage defned benef (DB) funds have hghe equy and lowe bonds nvesmen han defned conbuon (DC) funds possbly eflecng ha he fome may ake moe sk as hey can benef fom negeneaonal sk shang. Columns 5 and 6 ndcaes how especvely he acual and saegc equy allocaon vay ove me. Fo he aveage penson fund he ange of acual and saegc equy allocaons s 6% and 3% especvely. hus boh he acual and saegc equy allocaon move sgnfcanly ove me. he 2 Company funds povde penson plans o he employees of he sponso company. hey ae sepaae legal enes bu ae un by he sponso company and employee epesenaves. Indusy funds povde penson plans fo employees wokng n an ndusy. Such penson plans ae based on a collecve labo ageemen beween an ndusy s companes and he labo unons epesenng he employees n hs ndusy. Fnally pofessonal goup funds offe penson schemes o specfc pofessonal goups (e.g. geneal pacones publc noaes).

9 9 las column shows ha he dffeence beween he saegc and acual equy allocaon s on aveage 0.8%. able 3 shows ha saegc and acual equy allocaons dffe sgnfcanly acoss penson fund obsevaons. A small majoy of funds nves pecen of he asses n eques. A quae of he funds nves moe han 40 pecen n eques whle aound one-ffh of he funds nves less han 20 pecen n eques. able 3: Dsbuon of equy allocaons acoss penson funds (999:I 2006:IV; n %) Range* Saegc equy allocaon Acual equy allocaon oal Noe: * he allocaons n each class ae calculaed as a pecenage of he oal obsevaons. We use wo benchmaks o assess he mpac of sock make euns on acual and saegc equy allocaon and o evaluae he pefomance of make mng saeges. he MSCI Wold oal eun ndex denomnaed n euos s ou pncpal benchmak eun fo equy pofolo nvesmens. 3 Penson funds use devaves fo hedgng puposes and also o effcenly adjus he accal asse allocaon avodng buyng o sellng (coss) of he undelyng asses. able 4 pesens he use of devaves by penson funds of vaous sze classes. he use of devaves nceases wh he sze of penson funds bu even he lages penson funds (.e. wh oal nvesmens lage han 0 bllon) hold only 2 pecen of he equy posons n devaves. Fo small penson funds (below 0 mln oal nvesmens) he noonal value of quaely puchased fuues and fowads (Column (5)) s small compaed o he ne puchase of eques (Column (4)). Howeve fo he lages penson funds he ne noonal value of ne puchased fuues and fowads s hgh and even exceeds he ne puchase of eques. In conas he ne pemum pad o eceved fo opons and waans s elavely small fo all funds alhough agan nceasng wh fund sze. Oveall we conclude ha only he lages penson funds make sgnfcan use of devaves. 3 Alenavely n some cases we use he AX ndex as a benchmak fo sock make euns n he Nehelands alhough should be noed ha penson funds geneally have boade and moe nenaonal equy pofolo allocaons han capued by he lae ndex.

10 0 able 4: he use of devaes acoss penson funds (999:II 2006:IV) Numbe of Aveage equy of whch Ne puchase (n mln euo) of oal nvesmens (mln euo) penson funds nvesmens (mln euo) usng devaves (%)ques Fuues and fowads Opons and waans Column () (2) (3) (4) (5) (6) 0-0 (Small) (Medum) (Lage) Aveage / oal Noes: All sascs ae smple aveages. Fo ne puchases of fuues and fowads we epo noonal amouns. Fo opons and waans he pemum s epoed. 3 Relave sock-make euns and sho-em changes n equy allocaon o sa ou empcal analyss hs secon examnes he sho-em mpac of sock make pefomance on equy allocaons. Ove me he acual equy allocaon may change n wo ways: () due o excess euns on eques compaed o ohe asse classes (fee floang) and () due o ne puchases o ne sales of eques (ebalancng and make mng). o nvesgae he mpac of elave sock make euns on penson funds equy allocaons we esmae he followng equaon: ( j j ) γ Polcy j δsze k w = α Σ j 0β j ε = () he dependen vaable w s he acual pecenage of he pofolo nvesed n eques of penson fund ( = N) a quae ( = ). he vaable ( ) s used o measue excess sock make euns compaed o ohe nvesmen caegoes on a quaely bass. Fo he sock make eun ( ) we use he eun on he MSCI Wold ndex and fo he eun on he oal penson fund pofolo ( ) we mulply he saegc asse allocaon of fou key asse classes by epesenave boad make ndexes. 4 We consde wo vaans on quaon (). he base model s whou lagged sock make euns (k = 0) wheeas alenavely we nclude excess sock make euns wh me lags (k = 5) o nvesgae whehe pas euns nfluence penson funds nvesmens wh some delay. he saegc equy allocaon (Polcy) also expessed as a pecenage s ncluded o conol fo penson fund nvesmen polcy. Sze whch s measued as he logahm of he oal nvesmen pofolo conols fo he endency of lage funds o nves elavely moe n eques. he saegc equy allocaon (Polcy) and Sze ae ncluded wh one lag o avod endogeney poblems and snce may ake some 4 We consde fve nvesmen caegoes: eques bonds eal esae money make nsumens and ohe asses. Fo bonds we use he JP Mogan MU bond ndex fo eal esae we use he FS PRA Nehelands eal esae ndex and fo money-make nvesmens we use he 3-monh ubo nees ae. We assume ha he ffh caegoy ohe asses ae popoonally nvesed n he pevous fou nvesmen caegoes (o ohewse have a smla eun). We calculae excess euns as follows: excess eun = eun MSCI [(eun on bonds * bond nvesmens eun on eal esae * eal esae nvesmens 3-monhs ubo * money make nvesmens) / (bond nvesmens eal esae nvesmens money make nvesmens)].

11 me befoe changes n hese vaables lead o changes n he equy pofolo nvesmen. he panel s unbalanced whch mples ha he numbe of obsevaons may vay acoss penson funds. 3. mpcal esuls of effec on acual equy allocaon able 5 pesens he mpac of sock make euns on he pecenage of equy pofolo nvesmens. A pecen elave oupefomance of he MSCI leads o an ncease n equy allocaon of 0.6 pecenage pon n he subsequen quae (see fs column). Noe ha 0.6 should be nepeed agans he aveage equy allocaon of 40 pecen. he second column shows ha acual euns also have a sgnfcan mpac on equy allocaons one up o fve quaes lae 0.53 pecenage pons n oal. able 5: smaes of he penson funds equy nvesmens model (999:II 2006:IV) All funds Small funds Medum szed funds Lage funds Column () (2) (3) (4) (5) xcess equy euns 0.59 *** 0.63*** 0.44*** 0.96*** *** Idem lagged quae 0.4*** 0.03*** 0.34*** 0.39 *** Idem lagged 2 quaes 0.083*** 0.07*** 0.0*** 0.38 *** Idem lagged 3 quae s 0.060*** 0.05*** 0.079*** *** Idem lagged 4 quaes 0.058*** 0.053*** 0.065*** 0. *** Idem lagged 5 quaes 0.047*** 0.044*** 0.056*** *** oal effec Invesmen polcy (-) *** 0.98*** 0.933*** 0.892*** *** Sze (-) *** 0.002*** 0.003*** *** Inecep *** *** Numbe of obsevaons R 2 adjused Noes: *** ** and * denoe sgnfcance a he % 5% and 0% levels especvely. he sandad eos have been coeced fo possble heeoskedascy o lack of nomaly usng he Hube-Whe sandwch esmaos. Hence sock make oupefomance nduces nceases n he equy allocaon n he nex peods alhough he mpac decays ove me. able 5 eveals also ha a pecenage pon ncease n he saegc equy allocaon causes a sgnfcan se of aound 0.93 pecenage pon n acual equy pofolo nvesmens n he nex peod. Appaenly penson fund nvesmen manages adjus he equy pofolo nvesmens almos fully n esponse o changes n he saegc equy allocaon. Fnally he posve sgn fo he sze of nvesmens affms ha lage funds nves elavely moe n eques. 5 A possble explanaon s ha lage penson funds end o be less sk avese han small penson funds. If we consde he nvesmen behavo acoss sze classes (las hee columns) whee sze classes ae defned as n able 4 we obseve ha he equy allocaon s songe affeced by acual equy 5 able also shows ha lage penson funds have elavely hghe equy nvesmens han smalle ones.

12 2 euns he lage he penson fund s boh mmedaely and n he long un. Appaenly lage funds show songe fee floang wheeas smalle funds ebalance moe. In lne wh hs lage funds eac less o changes n he nvesmen polcy compaed o smalle funds. 3.2 Relave sock make euns and ebalancng he posve mpac of excess equy euns on equy allocaons n he pevous secon may be (paly) due o mpefec ebalancng by penson funds. xcess equy pefomance wll lead o changes n he equy allocaon f penson funds do no fully ebalance he nvesmen pofolos mmedaely and connuously. hs secon pesens an empcal ebalancng model whch esmaes o wha exen penson funds ebalance ha s adjus he asse allocaon n esponse o excess equy euns. 6 hs model s deved as follows sang fom he defnon of he acual equy allocaon: A w / = (2) whee epesens he equy nvesmens of penson fund a me and A sands fo oal asses. akng fs dffeences of quaon (2) we oban: = A A w w ) ( ) ( = A A ) ( ) ( ) ( ) ( w w = ) ( ) ( w = (3) whee s sho fo new nvesmens as a facon of oal nvesmens fo new equy nvesmens also as a facon of equy nvesmens fo he eun on eques ove he las quae and fo he eun on oal asses (all fo fund and quae ). Dvdng boh sdes by w - esuls n: w w w = (4) hs equaon explans he pecenage change n equy allocaon by: () excess equy euns and () ne cash flows o eques whee boh vaables ae scaled by he change n he oal pofolo sze. he fs gh-hand em s exogenous snce excess euns ae deemned by make developmens and 6 An alenave appoach o measue ebalancng based on penson funds equy sales and puchases s pesened n he appendx.

13 3 ne cash flows o he penson fund ae based on decsons by employes and employees. Gven he small sze of penson fund nvesmens elave o oal sock make capalzaon we can safely assume ha changes n equy allocaons do no affec sock make euns. he second gh-hand em howeve s endogenous. Whle ne cash flows o equy nvesmens decly nfluence he equy allocaon of penson funds he evese can also be ue: changes n he equy allocaon may sway penson funds o adjus he ne cash flows o equy nvesmens. hus hee s evese causaly beween changes n equy allocaons and ne cash flows o equy nvesmens. o esmae he mpac of excess equy euns on he equy allocaon we apply he decomposon above excludng he endogenous second gh hand em. hs esuls n he followng empcal egesson model: w w w Polcy α 2 β = 2 γ 2 ε Polcy 2 (5) he gowh n he saegc equy allocaon (Polcy) s ncluded o conol fo changes n nvesmen polcy ncluded wh a lag of one quae snce may ake some me befoe hs vaable adjus he equy pofolo nvesmens. In quaon (5) β 2 esmaes he degee of fee floa and equvalenly - β 2 assesses he pecenage of ebalancng. As an alenave model we spl he excess equy eun vaable fuhe no posve and negave equy euns n ode o be able o obseve possble asymmec effecs n he esponse on sock make shocks. 3.3 mpcal esuls of ebalancng able 6 pesens he esmaed mpac of excess equy euns on he equy allocaon. he esuls show ha penson funds ebalance on aveage aound 39 pecen of excess equy euns leavng 6 pecen fo fee floang. he lae ndcaes ha 6 pecen of excess equy euns nceases he equy allocaon n he nex peod. Column (2) shows ha penson funds ebalance dffeenly n esponse o posve and negave equy euns. Only 2 pecen of posve equy euns ae ebalanced whle 49 pecen of negave equy euns esul n ebalancng. Appaenly penson funds do no auomacally sell eques n bull makes bu ae moe wllng o buy addonal eques n bea makes. In lne wh expecaons changes n polcy affec he acual allocaon posve (sgnfcan a he % level) wh a lag of one quae.

14 4 able 6: smaes of he equy allocaons model: ebalancng vesus fee floa (999:II 2006:IV) All funds Small funds Medum szed funds Lage funds Column () (2) (3) (4) (5) (6) (7) (8) xcess equy euns Posve excess equy euns Negave excess equy euns Change n polcy (-) Inecep Numbe of obsevaons R 2 adjused Noes: All symmec and asymmec excess equy euns effecs ae sgnfcan a he % level. he sandad eos have been coeced fo possble heeoskedascy o lack of nomaly usng he Hube-Whe sandwch esmaos. he las sx columns pesen he model esmaes fo he vaous sze classes. In lne wh he esuls of Secon 3. we obseve ha n he symmec model vaan he lage funds ebalance a 30 pecen less han he small and medum-szed funds (wh aound 40 pecen) leavng 70 pecen fo fee floang. Changes n he one quae lagged saegc equy allocaon (Polcy) affec he acual allocaon sgnfcanly (a he % level) fo he small funds only. If we un o he asymmec effecs on excess equy euns we obseve ha he posve effecs ncease sgnfcanly wh penson fund sze whle he negave effecs ae smla acoss he sze classes. he coeffcen of posve euns of he lages funds s a.2 even sgnfcanly above eflecng oveshoong of fee floang o negave ebalancng. hs s n lne wh esuls n able 5 whch ndcae ha lage funds espond songe on excess equy euns han small ones. A possble explanaon s ha manages of lage funds have moe feedom o espond o make developmens and apply a de you pofs cu you losses saegy. Fgue 3 pesens he asymmec elaon beween excess equy euns and ebalancng dscussed above. 7 xcess equy euns would fully lead o popoonae nceases n he equy allocaon f penson funds would use a fee floa saegy and no ebalance a all. hs s epesened by he dagonal lne. Insead n case of full ebalancng excess equy euns would have no mpac on equy allocaons: he x-axs. he cuvaue dvdng he fee floa and ebalancng aeas eflecs he acual ebalancng behavo by he Duch penson funds. Skngly ebalancng by penson funds depends on boh he sgn and sze of excess equy euns. Small posve equy euns (of aound 0-5%) ae no ebalanced a all bu he degee of ebalancng nceases wh he sze of excess equy euns. Insead small negave euns (of aound 0 o -0%) ae ebalanced fo he lages pa bu he degee of ebalancng deceases wh he sze of he negave excess euns. 7 o esmae hs fgue we adjused quaon 5 by addng hee addonal ems: squaed excess equy euns and excess and squaed equy euns mulpled wh 0- dummes ndcang posve and negave euns.

15 5 Fgue 3: smaed behavou wh espec o ebalancng and fee floa Rebalancng Change n equy allocaon (%) 0 Fee floa Fee floa -0 Rebalancng xcess equy eun (%) 4 Relave sock make euns and medum-em changes n saegc equy allocaon he pevous secon descbed he effecs of excess equy euns on he acual equy allocaon. hs secon nvesgaes he mpac of sock make pefomance on an annual bass (a - a ) on he penson funds saegc equy allocaon (Polcy). heefoe we esmae he followng equaon: ( a a ) γ 3Polcy δ 3Sze Polcy = α 3 β 3 ε (6) he sock make pefomance has been aken on an annual bass ndcaed by (a a ) whee a efes o annual. 8 We assume ha he penson fund admnsaon bases s polcy on longe em measues of pefomance as also efleced by he empcal esuls. 9 As abovesze conols fo he endency of lage funds o nves elavely moe n equy pofolos. We nclude also a lag of he dependen vaable Polcy as we expec ha polcy changes gadually ove me only. Hence he equaon descbes he quaely adjusmens n polcy. We also esmae an alenave model whch explans fs dffeences n Polcy n sead of gadual adjusmen. 8 xcess equy eun compaes he MSCI Wold ndex o ohe nvesmen caegoes as above. 9 Annual euns povde bee esuls han quaely euns.

16 6 4. mpcal esuls of he effec on saegc equy allocaon able 7 povdes he mpac of sock make euns on he saegc equy allocaon. he nvesmen polcy s adjused sgnfcanly o changes n equy euns especve of whehe hey ae measued by he MSCI o he acual nvesmen euns eaned by penson funds. hs shows ha he nvesmen polcy s no consan ove me bu o some exend follows make developmens. he coeffcen of he lagged dependen vaable of 0.97 ndcaes how slowly he saegc equy allocaon eacs o changes n he quaely euns. On aveage 97 pecen of he equy nvesmen polcy s deemned by he nvesmen polcy n he las quae wheeas make developmens ae followed fo he emanng 3 pecen. hese make developmens he yealy excess eun ae small bu vey sgnfcan boh based on he MSCI and on he equy eun penson fund. he fnal mpac on equy nvesmen polcy ove me s 0.28 (= /( )). he sze effec s also small bu sgnfcan. 20 Whee hs equaon shows how nvesmen polcy s nfluenced by make developmens does no povde a model of he undelyng deemnaon of nvesmen polcy whch s geneally based on asse lably managemen sudes. able 7: smaes of he saegc equy allocaon model (999:II 2006:IV) All funds Medum szed Small funds funds Lage funds Column () (2) (3) (4) (5) quy nvesmen polcy (-) *** 0.976*** *** *** *** Yealy excess eun MSCI *** *** *** 0.00** Yealy excess equy eun penson fund *** Sze (-) *** ** ** Inecep *** Numbe of obsevaons R 2 adjused Noes: *** ** and * denoe sgnfcance a he % 5% and 0% levels especvely. he sandad eos have been coeced fo heeoskedascy usng he Hube-Whe sandwch esmaos. he esuls acoss he penson fnd sze classes seem smla. Neveheless he slgh nceases n he coeffcens of he lagged dependen vaable and he yealy excess eun fo he lage funds mply an ncease n he long-un effec of he yealy excess eun fom 27% (small funds) o 47% (lage funds). 20 We esmae also an alenave specfcaon of quaon (6) wh as dependen vaable he fs dffeence of equy nvesmen nsead of s level so ha we dop he lagged equy nvesmen polcy em. In ha specfcaon he vaous yealy excess eun coeffcens ae all a facon hghe and equally sgnfcan. he vaable sze s no longe sgnfcan. Whee equy levels ae hghe fo lage penson funds compaed o small funds he adjusmen of he saegc equy allocaon does no devae sgnfcanly beween lage and small funds.

17 7 5 he mpac of make mng on penson fund euns Boh Secons 3 and 4 show ha he acual asse allocaon of penson funds as well as he saegc asse allocaon s nfluenced by he elave pefomance of equy makes. Penson fund manages may pof fom make mng povded hey have some ably n foecasng sock make ends. o ean hghe sk-adjused pofolo euns sklled nvesos can ceae a posve nfomaon ao hough nceasng equy allocaons n bull makes and convesely deceasng hem n bea makes. hs secon examnes whehe penson funds ndeed have pofed fom make mng dung he sample peod. We use he followng equaon adaped fom Gnbla and man (993) o esmae he pa of excess euns ha can be abued o make mng (R M ): 2 R M ( / = = w w ) (7) he vaable equals quaely euns epoed by penson fund a me and w s (agan) he equy wegh n he asse pofolo. In addon we eplace ndvdual euns by euns of he MSCI Wold ndex o he AX ndex heeby smulang he eun ha would have been obaned fom usng he make mng saegy o nves n a boad sock make ndex. Unde he null hypohess ha he pofolo manage has no ably n foecasng expeced sock make euns he pofolo euns ae uncoelaed o ove- o undeweghng of equy allocaons elave o he mean. able 8: he aveage mpac of make mng on euns (999:II 2006:IV; n %) Saegc make mng by Aveage sandad Aveage equy devaon of equy Aveage excess MSCI penson fund boad: wegh weghs (ove me) equy euns ndex AX ndex a) All funds b) Small funds c) Medum-szed funds d) Lage funds Acual make mng by pofolo manages: a) All funds b) Small funds c) Medum-szed funds d) Lage funds Noes: All sascs ae aveages weghed by oal nvesmens. able 8 pesens he esmaon esuls of quaon (7). On aveage boh he make mng saegy poposed by he penson fund admnsaon and he make mng execued by nvesmen manages 2 he pefomance measue poposed by Gnbla and man (993) s dffeen fom ous n wo ways. Fs hey compae cuen pofolo weghs o pas nsead of aveage pofolo weghs. Second hey do no specfcally focus on make mng bu nsead sudy whehe acve sock pckng geneaed posve skadjused euns.

18 8 led o negave excess euns alhough no sgnfcanly dffeen fom zeo. 22 Acual aveage negave excess euns equal 20 bass pons pe euo nvesed. able 8 also epos he excess euns fom make mng when penson funds would have nvesed he money n a pofolo ha mmcs a boad make ndex. he aveage negave excess euns fom usng acual make mng by penson funds o nves n he MSCI Wold and AX ndex equal especvely 23 and 20 bass pons pe euo nvesed bu agan hese esmaes ae no sgnfcanly dffeen fom zeo. I s mpoan o emphasze ha he coss of make mng ae no fully nenalzed no he fgues pesened. If coss of pesonnel and ansacons would have been ncluded euns fom make mng wee even moe negave. Oveall hese esuls show ha fo he aveage penson fund make mng led o negave non-sgnfcan excess euns dung he sample peod. unng o he sze class esmaes we fnd agan ha small funds have less equy. Fo small funds he make mng execued by nvesmen manages led o smalle negave excess euns han fo lage funds bu he dffeences ae small and no sgnfcan. 7 Conclusons hs pape s he fs ha examnes he mpac of sock make pefomance on he nvesmen polcy of penson funds. We fnd ha sock make pefomance nfluences he nvesmen polcy of penson funds n vaous ways. In he sho em he elave oupefomance of eques ove bonds and ohe nvesmen caegoes auomacally esuls n hghe equy allocaons (and vce vesa) as penson funds do no connuously ebalance he asse allocaon. ach quae penson funds ebalance on aveage aound 38 pecen of excess equy euns leavng 62 pecen fo fee floang. In he medum em he elave oupefomance of eques leads penson funds o ncease he saegc equy allocaon (and vce vesa). Oveall ou esmaes sugges ha he nvesmen polcy of penson funds s paally dven by he (cyclcal) pefomance of he sock make. We fnd ha penson funds equy eallocaon s sgnfcanly dffeen n esponse o posve equy euns compaed o negave sock make yelds. Only 6 pecen of posve equy euns ae ebalanced whle 48 pecen of negave shocks esul n ebalancng. We obseve ha he acual equy allocaon of penson funds vaes sgnfcanly ove me. Howeve penson funds ae no ewaded fo akng addonal sk elave o he saegc asse allocaon. On aveage boh he make mng saegy poposed by he penson fund admnsaon and he make mng execued by nvesmen manages led o negave excess euns ove he peod unde consdeaon alhough no sgnfcanly dffeen fom zeo. In lne wh mos empcal evdence we 22 Sgnfcance s based on he annualzed sandad devaon of he calculaed excess euns whch s aound 0.9% well above he aveage excess euns n able 8.

19 9 fnd ha he make mng saegy of Duch penson funds does no geneae excess euns suggesng ha also Duch penson fund manage ae no able o pedc make movemens. Oveall hs pape suggess ha ove he sample peod he value added of make mng s lmed and consequenly penson funds would do bee by elyng on ebalancng. Lage funds demonsae dffeen nvesmen behavou compaed o small funds n he sense ha hey have moe equy and ha he equy allocaon s much songe affeced by acual equy euns. he lae mples ha lage funds ebalance less eflecng less sk aveson han small funds. We fnd asymmec effecs on excess equy euns whee he posve effecs ncease sgnfcanly wh penson fund sze. he coeffcen of posve euns of he lages funds s even sgnfcanly above eflecng oveshoong of fee floang o negave ebalancng. A possble explanaon s ha manages of lage funds have moe feedom o espond o make developmens and apply a de you pofs cu you losses saegy. Refeences Agnew Jule Pelug Balduzz and Annka Sundén 2003 Pofolo Choce and adng n a Lage 40(k) Plan Amecan conomc Revew 93 (): Babes Ncholas 2000 Invesng fo he long un when euns ae pedcable Jounal of Fnance 55 (): Baue Rob Rk Fehen Hube Lum and Rogé Oen 2007 he pefomance of US penson funds. New nsghs no he agency coss debae mmeo. Bkke Jacob A. and Jan de Deu 2007 Opeang coss of penson funds: he mpac of scale govenance and plan desgn Jounal of Penson conomcs and Fnance fohcomng. Blake Davd Buce N. Lehmann and Allan mmemann 999 Asse allocaon dynamcs and penson fund pefomance Jounal of Busness 72 (4): Bnson Gay P. Randolph Hood and Glbe L. Beebowe 986 Deemnans of pofolo pefomance Fnancal Analyss Jounal 42 (4): Bnson Gay P. Ban D. Snge and Glbe L. Beebowe 99 Deemnans of pofolo pefomance II: An updae Fnancal Analyss Jounal 47 (3): Boedes Dk W.G.A. he valuaon of condonal penson lables and guaanees unde sponso vulneably DNB Wokng Pape no. 82 De Nedelandsche Bank Amsedam. Campbell John Y. and Lus M. Vcea (2002) Saegc Asse Allocaon: Pofolo Choce fo Long- em Invesos Oxfod Unvesy Pess. Danel Ken Mak Gnbla Shedan man and Russ Wemes 997 Measung muual fund pefomance wh chaacesc-based benchmaks Jounal of Fnance 52 (3): Fama ugene F. 972 Componens of nvesmen pefomance Jounal of Fnance 27 (3):

20 20 Gnbla Mak and Shedan man 993 Pefomance measuemen whou benchmaks: An xamnaon of Muual Fund Reuns Jounal of Busness 66 (): Henksson Roy D. and Robe C. Meon 98 On make mng and nvesmen pefomance. II. Sascal pocedues fo evaluang foecasng sklls Jounal of Busness 54 (4): Ibboson Roge G. and Paul D. Kaplan 2000 Does asse allocaon polcy explan o 00 pecen of pefomance? Fnancal Analyss Jounal 56: Kakes Jan 2006 Fnancal behavou of Duch penson funds: a dsaggegaed appoach DNB Wokng Pape n. 08 De Nedelandsche Bank Amsedam. Kon Sanley J. 983 he Make-mng pefomance of muual fund manages Jounal of Busness 56 (3): Kon Sanley J. and Fank C. Jen 979 he nvesmen pefomance of muual funds: An empcal nvesgaon of mng selecvy and make effcency Jounal of Busness 52 (2): Lakonshok Josef Ande Schlefe and Robe W. Vshny 992 he sucue and pefomance of he money managemen ndusy Bookngs Papes on conomc Acvy: Mcoeconomcs Páso Luboš and Robe F. Sambaugh 200 he equy pemum and sucual beaks Jounal of Fnance 56 (4): Shefn Hesh and Me Saman 2000 Behavoal Pofolo heoy he Jounal of Fnancal and Quanave Analyss 35 (2): 27-5.

21 2 APPNDIX: PNSION FUNDS QUIY PURCHASS AND SALS An alenave pocedue o esmae ebalancng s o use he puchases and sales of equy nvesmens. quaon (8) esmaes he mpac of excess equy euns on changes n equy holdngs. quy sales = 4 β4 ) γ 4 Polcy δ4( Polcy w ) α ( ε (8) quy sales s defned as equy puchases o sales of fund a quae as a pecenage of oal equy. he explanaoy vaables ae he same as befoe: ( ) measues excess sock make euns compaed o ohe nvesmen caegoes w s he pecenage of penson fund equy nvesmens whle Polcy s he saegc equy allocaon. Addonally we consde asymmec effecs of excess equy euns on equy ansacons. We conol fo changes n he saegc equy allocaon and fo dffeences beween he saegc equy allocaon and acual equy nvesmens (he nvesmen gap ) boh lagged one quae. able 9 pesens evdence on ebalancng as he pecenage of equy pofolo sales and puchases ae sgnfcanly affeced by excess equy euns. Negave equy euns ae he man foce behnd hs phenomenon (see Column (3)). he nvesmen gap s also a sgnfcan dve of equy pofolo sales and puchases. Alhough he ebalancng and nvesmen gap effecs ae sgnfcan only a ny poon of he vaaon n equy pofolo sales and puchases can be explaned by ou model (see adjused R 2 ) able 9: smaes of he equy puchases and sales model (999:II 2006:IV) All funds Small funds Medumszed funds Lage funds Column () (2) (3) (4) (5) (6) xcess equy euns -0.27*** -0.27*** -0.27*** -0.28*** -0.22*** Posve excess equy euns 0.03 Negave excess equy euns -0.4*** Change n saegc equy allocaon (-) ** 0.0** -0.0 Invesmen gap (-) 0.37*** 0.35 *** 0.44*** 0.33*** 0.26*** Inecep 0.02*** 0.02*** 0.0 *** 0.03*** 0.02*** 0.0*** Numbe of obsevaons R 2 adjused Noes: *** ** and * denoe sgnfcance a he % 5% and 0% levels especvely. he sandad eos have been coeced fo heeoskedascy usng he Hube-Whe sandwch esmaos. unng o he sze class esmaes we fnd less ebalancng behavou of he lage funds compaed o small funds. hs s fuhe emphaszed by he obsevaon ha lage funds also eac less han small funds on changes n polcy and he nvesmen gap.