Longevity Risk and Annuities in Singapore

Transcription

1 1 Longeviy Risk and Annuiies in Singapore Joelle H.Y. Fong, Olivia S. Michell, and Benedic S. K. Koh* Absrac This paper explores he curren annuiy marke in Singapore and discusses he pros and cons of a proposal o mandae annuiizaion under he Singaporean Cenral Providen Fund (CPF). We evaluae he pricing of various annuiy policies in order o assess wheher plan paricipans migh benefi from higher annuiy reurns per dollar premium and lower adverse selecion coss under he new annuiizaion mandae. Our resuls indicae ha privae annuiy providers currenly offer good value-for-money annuiies, wih money s worh values in line wih hose found for oher developed counries. This has implicaions for proposals o mandae annuiizaion. * Fong is a docoral suden and S.S. Huebner Foundaion Fellow a he Wharon School, Universiy of Pennsylvania; Michell is Inernaional Foundaion of Employee Benefi Plans Professor of Insurance and Risk Managemen, and Execuive Direcor of he Pension Research Council/Boener Cener a he Wharon School of he Universiy of Pennsylvania; and Koh is Associae Professor of Finance a he Singapore Managemen Universiy. All opinions are solely hose of he auhors who acknowledge research suppor from by he Russell Ackoff Docoral Suden Fellowship and he Wharon Risk Managemen and Decision Processes Cener, and he Pension Research Council/Boener Cener a The Wharon School of he Universiy of Pennsylvania. Wihou implicaing hem, we acknowledge helpful assisance and advice from Desmond Chew, Jean Lemaire, Libby Sang, and Lee Ee Jia Fong, Michell, and Koh.

2 2 Longeviy Risk and Annuiies in Singapore Joelle H.Y. Fong, Olivia S. Michell, and Benedic S. K. Koh A cenral concern in he debae over pension reform in defined conribuion (DC) sysems is how plan paricipans should draw down heir accumulaed asse balances during reiremen. Annuiizaion is ofen recommended as a means o help plan paricipans manage heir longeviy risk, since oherwise hey may oulive heir asses in reiremen. Some form of annuiizaion in he payou phase helps ensure ha plan paricipans have a dependable flow of income beyond he reiremen dae all he way o deah. For insance, in he Unied Kingdom, reirees have been required o use a leas par of he lump sum available a reiremen o purchase an annuiy (Finkelsein and Poerba 2002, 2004); in Chile he DC reiremen sysems give plan members he choice of aking scheduled wihdrawals or buying life annuiies upon reiremen (Michell and Ruiz 2009). This paper reviews he naure of longeviy risk and annuiies in Singapore, in order o draw some implicaions abou he prospecs for fuure annuiizaion under one of he world s larges defined conribuion schemes, he Cenral Providen Fund (CPF) of Singapore. In paricular, we examine how he curren life annuiy marke appears o be operaing and assess he likely araciveness of compulsory annuiizaion under proposed reforms. In wha follows, we firs describe he way in which he reiremen sysem works in Singapore. Nex we assess he value-for-money of exising annuiy producs. We conclude wih a brief discussion of he issues ha arise when discussing he opions for a mandaory annuiy model such as hose recenly suggesed in he Singaporean conex.

3 3 The Reiremen Framework in Singapore Singapore s Cenral Providen Fund is one of Asia s oldes reiremen programs, as i was esablished in Buil around individual accouns, he scheme is mandaory and employmenlinked. The curren conribuion rae under he defined conribuion plan is divided beween employers and employees; currenly he oal sands a 34.5 percen (hough i has been as high as 40 percen in he pas). 1 Since incepion, paricipans have been able o leave heir conribuions wih he CPF o earn a guaraneed risk-free ineres rae (currenly a leas 2.5 percen [CPF 2009a]). The sysem has also been reformed several imes wih he goal of enhancing he sysem s asse accumulaion by simulaing more saving for reiremen, housing, and healhcare needs. The CPF Invesmen Scheme, inroduced in 1986 and broadened in 1993, allowed pension conribuions o be invesed in muual funds and alernaive asse classes including gold. In 1993 and again in 1996, CPF members were permied o buy shares of Singapore Telecom a a discoun wih heir CPF conribuions. The CPF Board, responsible for managing he sysem, has underaken a se of reforms focused on he asse decumulaion process, responding in par o he rapid aging of he CPF membership base. In he las wo decades, he proporion of members age 55+ experienced a four-fold increase from 5.5 percen in 1985 o 22.9 percen in 2005; a he same ime, he proporion of hose younger han age 24 fell from 25.1 percen o 9.2 percen (CPF 2007a). This rend occurred, in par, because Singapore has one of he world s lowes feriliy raes (1.29 per 1 For workers age 50 and younger, he Governmen has se he long-erm arge CPF conribuion raes a percen, wih an employee conribuion of 20 percen and an employer conribuion varying beween percen. For hose older han age 50 and up o age 55, he long-erm arge range is se a percen, wih he employee conribuion a 18 percen and he employer conribuion varying beween 6 12 percen (SPMO 2003).

4 4 female) and longes life expecancies (80.6 years a birh). 2 These facs combined imply ha reiremen expenses are projeced o rise, as people live longer and have fewer children on which o rely. The CPF regulaes how reirees can access heir money via he so-called Minimum Sum (MS) Scheme. This includes hree main componens: he value of he Minimum Sum, he age a which draw-down can sar, and he form of payous. This scheme was inroduced in 1987 o ensure ha CPF members could anicipae a leas a basic sandard of living in reiremen. A presen, sysem paricipans a age 55 mus se aside he MS in heir Reiremen Accoun from heir oal accumulaions; 3 official draw-down age. In July 2007, for insance, he required MS was se a $99,600 and he official draw-down age a 62. The MS is no a hreshold easily me; for insance, only 36.4 percen of acive members could se aside he required MS in 2006 (CPF 2007b). In response o he rapid aging of he CPF membership base and he need o save more for reiremen, he componens have been fine-uned. Table 1 summarizes he evoluion of he MS Scheme and projeced changes o Here we see ha he sipulaed MS rises progressively unil i reaches $120,000 by his amoun is hen preserved and may be paid ou only as of he 4 Concurrenly, he draw-down age is rising gradually from 62 o 65 by Table 1 here Reirees can currenly ake heir payous in he form of phased wihdrawals, hough as of 2013, he governmen has announced here will be a ransiion o a compulsory deferred 2 Figures are from year end 2007 and obained from he Singapore Deparmen of Saisics (SDOS 2008a). 3 If a member s oal balance is higher han he Minimum Sum, any remaining balance can be wihdrawn as a lump sum. If he oal balance is less han he Minimum Sum, he following wihdrawal rules currenly apply for members who reach 55 beween 1 Jan 2009 and 30 June 2009: oal balance $5,000 (wihdraw everyhing), $5,000 < oal balance $12,500 (wihdraw $5,000 and se aside remainder in Reiremen Accoun), and $12,500 < oal balance $176,667 (wihdraw 40 percen of oal balance and se aside remainder in Reiremen Accoun; see CPF 2009b). 4 This change was announced in The oher wo changes on he draw-down age and payou srucure were announced in 2007 in Prime Miniser Lee s Naional Day Rally speech (SPMO 2007).

5 5 annuiizaion forma, abou which we discuss below. Under he presen rules, mos reirees ake drawdowns from heir MS over abou 20 years, or unil he balance is exhaused. An alernaive o his phased wihdrawal approach is a life annuiy sold by privae insurers. In 2007, nine Minimum Sum Scheme annuiies were offered; hese involve a life annuiy where he lump-sum premium is he sipulaed MS. 5 These privae insurers end o be well-known players in he Singapore and inernaional insurance markes including AIA, Prudenial, and HSBC Insurance, as well as some local insurance providers. Table 2 summarizes aribues of he nine qualifying MS annuiies offered in 2007 by privae insurers. There are some differences in provisions, bu he producs offered are similar in ha hey are all single premium, deferred, life annuiies. The lump-sum premium is he enire MS amoun of $99,600 o be paid a age 55, in exchange for annuiy paymens beginning a exacly age 62. All have fixed (level nominal) payous, wih wo annuiies boosing an addiional paricipaing bonus feaure; hese laer are no guaraneed and depend on he profis of he insurer each year. Non-paricipaing annuiies payous average $526 per monh for males and $478 for females, while paricipaing payous (minus he bonus) average $499 for males and $474 for females. Paricipaing payous are slighly lower since he consumer may receive bonus payous on op of he specified base rae. Overall, Table 2 shows ha women receive lower payous for he same premium given heir longer life expecancies; i is also noeworhy ha he annuiy payous are sex-specific, resuling in females obaining a lower annual payou han males for a given premium (by conras, he annual payous under phased wihdrawal are sex-neural). I is also worh noing ha all MS annuiies on offer include a guaraneed amoun feaure. Thus a he annuian s deah, his beneficiary receives a leas he premium paid (a leas a premium of 5 I mus be noed ha he CPF Board does no endorse any specific life annuiy produc offered under he MS Scheme nor does i screen privae insurers (alhough any life insurer exhibiing poor conduc or unaccepable behavior may have is conrac suspended).

6 6 $99,600 6 ) less annuiy payous already made. In effec, his means ha he proecion has an elemen of capial-proecion on he premium remaining. Table 2 here Despie he assormen of annuiies available on he marke, mos CPF reirees o dae have eleced phased wihdrawal raher han life annuiies. Using esimaes from CPF annual repors, we find ha only four o five percen of he reiring cohors in recen years have volunarily purchased MS annuiies. Ye his small percenage mus be evaluaed in he proper perspecive. For one hing, abou a quarer of he reiring cohor was exemped due o medical grounds, small balances, or oher reasons. For anoher, almos half (48 percen) of he reirees were ineligible o buy his annuiy because hey had no se aside he full MS in cash; his group is, by defaul, channeled ino he phased wihdrawal payou opion. In oher words, of he remaining 27 percen who had a choice beween phased wihdrawals versus lifeime annuiy payous, a relaively high proporion one ou of six oped for annuiizaion. 7 This relaively high annuiizaion rae among he eligibles compares o much lower annuiizaion raes in oher counries, where i has been suggesed ha people may fail o annuiize because of crowd-ou from public defined benefi pensions, a desire o leave bequess, he need for liquidiy, and adverse selecion, among oher reasons (Michell e al. 1999). In he Singaporean conex, however, we can rule ou he crowd-ou by a public defined benefi pension as here is none. The beques moive is unlikely o be a deerren since exising rules permi bequess via refunds o beneficiaries. More plausible is a desire for liquidiy, since he phased wihdrawal approach yields monhly payous of $790, compared o he average annuiy payou of 6 In some producs, he guaraneed amoun is he premium plus annual ineres accrued from age 55 when he annuiy is purchased up o age 62 when payous sar; see Table 2. The resuls in Table 4 accoun for he differen specificaions of guaraneed amoun for each produc. 7 This is consisen wih Doyle, Michell, and Piggo (2004).

7 7 abou $ I is plausible ha myopic paricipans as well as hose expecing o live a shorer period will op for phased wihdrawal. Anoher facor may be ineria: pension plan paricipans are ofen found o accep whaever is he defaul opion, which in his case is a phased wihdrawal (Koh e al. 2008). The possibiliy of adverse selecion may also be a consideraion in he Singaporean annuiy marke, hough he issue has no ye been fully evaluaed. We urn o an examinaion of his issue nex. Money s Worh Valuaion for Singaporean Annuiies To examine he exen of adverse selecion in he Singaporean volunary annuiy marke, i is necessary o compare he money s worh of he life annuiy benefi using populaion survival versus annuian ables. Specifically, we noe ha a life annuiy is a conrac ha pays he buyer a benefi as long as he lives, which insures he annuian agains he risk of ouliving accumulaed resources, in exchange for a premium. By so doing, he purchaser ransfers his longeviy risk o he insurer, who pools he survival experience of muliple buyers. Following Michell e al. (1999), he money s worh raio (MWR) is herefore he raio of he expeced presen discouned value (EPDV) of annuiy payous divided by he iniial premium (K): MWR = [EPDV(benefis)]/K (1) In he Singaporean conex, K reflecs he $99,600 lump-sum premium (he Minimum Sum in 2007). The age of enry is age 55 when he annuiy is purchased. The general expression for he EPDV is: pa A EPDV = = + ( i ) 1 1 a (2) 8 This is for he case of a member who has se aside he full Minimum Sum of $99,600 a age 55 as of Under phased wihdrawal, he can draw down his amoun plus ineres via monhly payous of $790; of course his will las for only abou 20 years a which poin he balance is likely o be exhaused.

8 8 where a is he age a which he annuiy is purchased, represens he number of monhs beyond annuiy saring dae, A a refers o he (level or fixed) monhly nominal annuiy payou for he individual purchasing annuiy a age a, i is he nominal ineres rae a monh, and p a is he probabiliy of an individual of age a sill surviving afer monhs. The expression runs over he maximum lifespan in a given moraliy able; for a deferred annuiy, paymens A a are zero during he deferred period. Assuming no commercial coss (loads), acuarial fairness requires ha he discouned value of he annuiy sream will equal he premium paid; accordingly, he MWR for an acuarially fair annuiy is uniy. In pracice, wo facors make annuiies acuarially unfair for he average person. Firs, insurers mus charge enough o cover adminisraive coss and earn a profi. Second, hose who buy annuiies end o live longer han hose who do no. Accordingly, in a marke where people buy annuiies volunarily, his adverse selecion raises prices for hose who buy. As demonsraed by Michell e al. (1999), one can separaely value hese wo sources of acuarial unfairness by comparing he difference in he MWR using he populaion versus he annuian survival ables. An appeal of he MWR concep is ha i is readily quanifiable and faciliaes comparisons across producs and counries. Neverheless, hese calculaions are necessarily sensiive o underlying moraliy and ineres rae assumpions. Prior Sudies on he Singaporean Annuiy Marke Two sudies have previously evaluaed he money s worh of MS Scheme annuiies offered under he Singaporean sysem; boh use daa available in he year Fong (2002) 9 A oal of 11 fla-rae (level) non-paricipaing and paricipaing annuiies were offered in July 2000 (excludes wo annuiies wih increasing payous). Doyle, Michell, and Piggo (2004) sampled five non-paricipaing annuiies. Fong (2002) included all eigh non-paricipaing annuiies plus one paricipaing annuiy bu did no incorporae poenial bonus payous from he paricipaion feaure.

9 9 invesigaes nine annuiy producs using a fixed ineres rae he proxies wih he 10-year Governmen bond yield. He repors a mean MWR of $0.997 for he male populaion so his implied cos of adverse selecion is abou $ Doyle, Michell, and Piggo (2004) use five fla-rae annuiies and employ a erm srucure of ineres raes which more accuraely discouns fuure cashflows. Tha analysis generaes a slighly lower MWR of $0.947 for he male populaion and a much lower cos of adverse selecion, $ Boh of hese sudies aribue he small degree of adverse selecion deeced o he fac ha he lack of a public defined benefi pension sysem makes he CPF annuiizaion scheme close o a capive marke. Ye here is reason o worry ha hese money s worh values could be oversaed and he degree of adverse selecion biased down, due o he lack of good moraliy daa. For insance Doyle, Michell, and Piggo (2004) uses abridged life ables from he World Healh Organizaion 10 and Fong (2002) exrapolaes moraliy paerns esimaed from 1960 period life ables. In addiion, boh sudies assume a consan force of moraliy for fracional ages wihin a year wihou jusifying why his migh be appropriae in he Singapore conex. 11 Anoher daa limiaion in pas sudies is ha heir moraliy ables have differen limiing ages for he populaion and he annuian group. For insance, Fong (2002) assumed a maximum lifespan of 99 years for he populaion bu 109 years for he annuian group; his naurally leads o higher MWRs for he annuians. In wha follows, we seek o improve on hese shorcomings. Two oher drawbacks of prior sudies are worh noing. Firs, hey use low ineres raes (long duraion T-bonds were no available a he ime) which may oversae he MWR resuls. The emergence of longer-duraion bonds offers us he opporuniy o improve on his issue. And 10 The World Healh Organizaion (various years) offers abridged life ables ha provide moraliy esimaes in 5- year age inervals; hey are less deailed han complee life ables providing moraliy raes for every individual age. 11 Three acuarial assumpions could be used for fracional ages wihin a year, namely, a uniform disribuion of deahs assumpion, a consan force of moraliy assumpion, and a hyperbolic assumpion.

10 10 second, boh sudies do no model he specific characerisics of he MSS annuiies currenly offered. Specifically, hey ignore he guaranee effecive during he seven-year deferral period and he lump-sum naure of he guaranee paymens. Tha is, hey assume ha CPF life annuiies have wo erms, where he firs erm applies o he 15-year guaranee period, and he second erm o he life payou period hereafer: EPDV = A p A a a + = 1 1+ ( 1+ i ) = 181 ( i ) a (3) Bu Equaion (3) does no capure he value of he refund if deah occurs prior o age 62, 12 so i will undersae he MWR. I also does no correcly capure he fac ha if deah occurs during he 15-year guaranee period, he refund is a lump-sum paymen o he beneficiary minus payous. Our MWR equaion exends he approach used in he pas in four key ways. Firs, we mach he limiing age of he populaion group wih ha of he annuian group. Second, we include all he annuiies currenly on offer under he CPF scheme, as opposed o selecing a subsample; we also incorporae expeced bonus payous for paricipaing annuiies using hisorical raes o reflec he paricipaion upside on such producs. Third, we accoun for he guaraneed amoun inheren in he 2007 MSS annuiies when underaking he valuaion analysis. Fourh, we apply a uniform disribuion of deahs assumpion o beer reflec he paern of moraliy in Singapore. 13 Adaping he valuaion model for he join-and-coningen annuiy, and using acuarial echniques o incorporae produc-specific characerisics, he following formula is hen suiable for valuing a MS nominal annuiy wih guaraneed amoun refund: 12 In resuls no repored here, we find ha 3-5 percen of EPDV is aribuable o refunds o he beneficiary upon he annuian s deah (hence failure o capure his undersaes he MWR by 3-5 percen). 13 The uniform disribuion of deahs assumpion for fracional ages wihin a year is appropriae given he lack of variaion in Singapore's weaher (so deah raes are unlikely o vary seasonally).

11 11 EPDV = 83 = 1 ( 1) p a q a+ ( 1) G + a A ( 1) q a+ ( 1) ( 1+ i ) = 84 ( 1+ i ) p a + p a max[0, G 84 s= 0 A a, s ] (4) Here a,, A a, i, and p a are defined as before, G is he guaraneed amoun (premium plus accrued ineres) a ime, s is a couner for he number of annuiy paymens made o annuian before deah, (-1) p a is he probabiliy of an annuian age a being alive afer (-1) monhs, q a+(-1) is he probabiliy of he annuian age a+(-1) monhs dying wihin he following one monh. Taken ogeher, (-1) p a. q a+(-1) is he probabiliy of an annuian aged a surviving o (-1) monhs and hen dying beween monh (-1) and monh. Thus his model exends Fong (2002) and Doyle, Michell, and Piggo (2004) by explicily including he refund upon deah before age 62 o represen expeced benefis due o he annuian and his beneficiaries. 14 In implemening his valuaion, we are also forunae o have access o new populaion moraliy ables recenly published by Singapore Saisics (SDOS 2008b) wih a limiing age of 100. Building on his base, we hen mus cohorize he populaion ables, as cohor moraliy ables are no available in Singapore o dae. To derive birh cohor ables using period life ables using he year 2007 period life able, we compue: x q x ( α ) qˆ ( ) = (2007) 1 (4) x where q x (2007) is he annual moraliy rae for age x in year 2007, qˆ x ( ) is he esimaed 14 This formula is appropriae for money s worh values when he annuiy has a guaraneed amoun or capialproecion feaure. Previous sudies using US daa have focused on simpler producs, mainly single-premium, immediae, nominal annuiies, and hey differeniae beween he single-life versus join-life annuiies (see Michell e al. 1999, Brown e al. 2001). Sudies on he UK compulsory and volunary annuiy markes (Finkelsein and Poerba 2002) have compared he money s worh of nominal, real, and escalaing annuiies, some wih guaranee periods of 0, 5, and 10 years; hese repor ha MWRs rise wih he lengh of he guaranee period. Thorburn, Rocha, and Morales (2005) repor ha he MWRs of guaraneed annuiies in Chile are smaller han hose of non-guaraneed annuiies, possibly due o he fac ha long periods of guaranee end o increase duraion, hus reinvesmen risk, forcing premiums up for a given value of benefis.

12 12 annual moraliy rae for age x in year ( ), and α x represens he esimaed annual moraliy improvemens for an individual aged x exrapolaed from moraliy changes beween 1990 and As in previous sudies, moraliy improvemen raes are projeced from he abridged period populaion ables for Singapore published by he World Healh Organizaion. In addiion, we mach he limiing age of he populaion group wih ha of he annuian group by exrapolaing populaion moraliy esimaes o he common maximum age of 117 o properly capure he longeviy ail risk in he populaion group; his is paricularly imporan for females. Lile informaion is currenly publicly available on he annuian moraliy experience in Singapore. Sandard insurance indusry pracice and previous research (Fong 2002) adoped he UK annuian moraliy experience wih adjusmens for local condiions, similar o wha is done in Ausralia. Moreover, he Moneary Auhoriy of Singapore (MAS), in is capaciy as insurance regulaor, requires firms o employ he UK a(1990) Ulimae Tables raed down five years for reserves and liabiliy valuaions peraining o annuiies sold (MAS 2008). 15 Accordingly, we use he a(1990) ables wih a five-year seback o esimae he annuian experience for our valuaion year, and hen we cohorize he resuling annuian ables. 16 We compue cumulaive survival probabiliies from he cohor ables as follows: 1 j= 0 ( 1 q ) p = (5) a a+ j 15 The Sixh Schedule of he Insurance Regulaions 2004 sipulaes ha insurers may employ he raes in he UK a(90) ables wih a five-year seback o value heir annuiy liabiliies. Previously, he Insurance Regulaions 1992 regulaions required insurers o employ he a(1990) ables wih a wo-year seback. These a(1990) ables are consruced based on UK annuians moraliy experience from wih moraliy improvemens projeced o By applying he 5-year seback, we effecively age he ables o Year 2007 and hen cohorize i for he MWRs. 16 As a robusness check, we verify ha our calculaions yield a lower moraliy for annuian cohor han he populaion cohor; for insance, a 65-year-old male in he general populaion has a moraliy of compared o for an annuian, which seems reasonable.

13 13 where p a is he cumulaive probabiliy of a person aged a surviving for years, and q a+j is he probabiliy of a person age (a+j) dying wihin he year. These cumulaive survival probabiliies are sex-specific and calculaed on a monhly basis o mach he frequency of he annuiy payous. Figure 1 plos our esimaes for he cumulaive survival probabiliies for 55-year old males and females in Singapore, for he general populaion and also for annuians. The key akeaway from he figure is ha he annuian survival curves lie above ha of he general populaion, confirming ha cumulaive survival probabiliies for annuians are generally higher han hose of he general populaion (or conversely, moraliy for annuians is lower since hey live longer). Figure 1 here Annuiy Quoes and Ineres Raes In 2007, eigh privae insurers offered life annuiies under he MS Scheme o CPF members; hey provided a oal of nine qualifying annuiies for he MS premium of $99,600 (CPF 2008a). All annuiies paid level benefis; wo were also paricipaing annuiies (see Table 2). The NTUC Income (2009) paricipaing annuiy offers an annual projeced bonus rae of abou wo percen; incorporaing his rae for boh paricipaing annuiies means he average payou across all 2007 MSS annuiies averages abou $535 per monh for a male paricipan and $492 per monh for a female paricipan. Table 2 here Daa on ineres raes are drawn from marke informaion. As in Michell e al. (1999), we use a erm srucure of ineres raes o discoun he sream of annuiy paymens o he presen. We judge he Singaporean Treasury bond raes as appropriae here since he MSS annuiies are

14 14 viewed as capial-proeced and hus riskless. Relying on he prices and yields of he Singapore Governmen Securiies Treasury bonds (MAS 2008) as a end 2007, we compue he riskless spo raes o proxy he yields on hypoheical zero coupon bonds. 17 Table 4 summarizes he key inpus and compares hem o assumpions used in prior sudies on Singaporean annuiies. Table 4 here Money s Worh Raios for MSS Annuiies We nex provide money s worh resuls using populaion moraliy ables, focusing on MS Scheme life annuiies offered by privae insurers in he volunary annuiy purchase scheme in Resuls in Table 5 show ha, on average, $1 of premium spen on a nominal MS annuiy by a 55-year-old male drawn from he general populaion would generae nearly $0.854 in expeced annuiy income (in ne presen value erms). Likewise, a female in he general populaion could anicipae receiving $ Though he monhly payous are lower for females han males, he MWR values converge for boh sexes, once life expecancy is aken in accoun. We also noe ha NTUC Income annuiy offered he highes money s worh o reirees in Singapore. The MWR of $0.944 (males) and $0.965 (females) exceeded he average MWR by almos 10.5 percen and 13.0 percen respecively. This could be explained by he fac ha NTUC Income operaes as a co-operaive company wih a mission o give back 98 percen of profis o policyholders in bonuses. Furher, is monhly annuiy payous are he highes among 17 The firs year rae is derived from he 1-year Treasury bill. Thereafer, he 2, 5, 7, 10, 15 and 20-year Treasury bond raes as of 2007 are used o esimae he riskless spo raes. Our annual spo rae ranges from 1.4 percen o 3.44 percen. Since maximum duraion available is only 20 years, we hen exrapolae he las spo rae ino he fuure, yielding a nominal riskless erm srucure of ineres raes on Singapore s Treasury bonds.

15 15 he MSS annuiies (having facored in he esimaed bonus). 18 Perhaps no surprisingly, his firm has he larges marke share of annuiies in Singapore. Table 5 here The MWR values may be used o esimae oal loadings, which average abou 14.6 percen. These esimaes are similar o hose for UK nominal annuiies of abou 14 percen (Finkelsein and Poerba 2002), and for US annuiies where he loads amoun o abou percen (Michell e al. 1999). Bu he Singaporean resuls for 2007 are much higher han hose repored in previous sudies on CPF-linked annuiies using pricing from earlier years and less precise moraliy ables. Tha is, Fong (2002) repors an average loading of only 1.4 percen for males and a negaive loading for females (MWR exceeded 1), while Doyle, Michell, and Piggo (2004) found loadings of abou 5.5 percen for boh sexes. Such small loadings make sense, given heir very high MWR figures ($ using populaion moraliy). A possible explanaion for he difference, as acknowledged by he auhors, is ha heir MWRs migh be oversaed due o he lack of long duraion Treasury bonds a he ime. Their loading figures seem implausibly low inasmuch as he producs are mainly offered by privae insurers who could no survive for long if hey paid such high benefis. Our main explanaions for he differences in he resuls are as follows: Differen producs evaluaed: We value he MS Scheme annuiies offered in 2007, which include a guaraneed amoun refund; earlier sudies valued he annuiies wih a 15-year guaraneed period. Differen moraliy assumpions: Prior sudies employ a consan force of moraliy assumpion for fracional ages wihin a year; by conras, we apply a 18 Hisorically, NTUC Income s average bonus paricipaion rae has ranged beween percen, and a 2 percen fuure bonus rae is ypically used o value is annuiy (NTUC Income 2009).

16 16 uniform disribuion of deahs assumpion o beer reflec he paern of moraliy in Singapore. Higher premium and lower annuiy payous: Annuiy quoes in he year 2000 were based on a lower premium (S$65,000) and paid ou higher average monhly benefis (abou $555 for males; Doyle, Michell, and Piggo 2004). By 2007, he premium had risen o $99,600 bu he average annuiy payous were lower ($520 for males). Nex we urn o a discussion of he cos of adverse selecion. We compue his by aking he difference beween a given annuiy s MWR using annuian moraliy, versus he same produc s MW calculaed using populaion moraliy. Table 5 shows he resuls. On average, adverse selecion coss accoun for approximaely one-fifh of he oal loadings for MS Scheme annuiies. These resuls are lower han empirical findings in he UK where adverse selecion coss amoun o abou 4.6 percenage poins (Finkelsein and Poerba 2002) and below he six percenage poins repored for Ausralian annuiies (Doyle, Michell, and Piggo 2004). They are much lower han he 10 percenage poins repored for he US (Michell a al. 1999). Overall, our findings sugges ha a reiree in Singapore s CPF having cash of a leas he MS would have been able o purchase longeviy proecion on compeiive erms, wihou much concern for adverse selecion. Toal loads migh be furher reduced, of course, if annuiizaion were mandaory, a opic o which we urn nex. Policy Proposals for Mandaory Annuiizaion Though he exising defaul phased wihdrawal wih volunary annuiizaion may have worked well in he pas in Singapore, here is now concern ha fuure reirees may be a risk of

17 17 running ou of money. In 1990, for insance, he average male and female life expecancy a birh was 73 and 77 respecively; by 2007, life expecancies a birh had risen o 78.2 and 82.9 respecively. And, of hose age 65 in 2007, wo-hirds would expec o sill be alive a age 80, and 48 percen a age 85 (CPF 2007a). This means ha abou half of all age-65 CPF members alive oday migh oulive heir CPF savings under he 20-year phased wihdrawal program (CPF 2009c). A relaed concern is ha he enire age srucure of members has aged subsanially, while membership growh has apered off. These facors are, quie sensibly, urning policymaker aenion o he role of longeviy proecion. To his end, he Singapore Governmen has recenly announced plans o implemen a mandaory annuiizaion scheme, slaed o pay ou benefis for he firs ime in This program, dubbed he CPF-LIFE program is being inegraed wih he exising CPF Minimum Sum Scheme. A he ime he plan was announced in 2007, a Naional Longeviy Insurance Commiee (NLIC) was formed o help design he program s elemens (SPMO 2007). This group released a repor in 2008 oulining preliminary deails and he design coninues o be refined. Accordingly, as of his wriing, since only he broad oulines of he new mandaory annuiizaion scheme are available, we do no offer MWR compuaions; hese mus awai more informaion of he produc and pricing srucure of producs o be offered. Neverheless he key elemens of he new proposal are usefully summarized. The new scheme will auomaically enroll members age 51 and younger from 2009 forward, who have a leas $40,000 cash saving in he Reiremen Accoun a age 55. The eniy which supervises he curren sysem, he CPF Board, is o adminiser he scheme drawing on he advice of independen acuarial consulans o deermine premium and payou levels. Members

18 18 may purchase heir annuiies eiher from a governmen-adminisered eniy, or from qualifying privae insurers. Key changes o be implemened under he LIFE scheme are laer payous and a differen defaul payou srucure. Specifically, a age 55, a paricipan mus se aside a Minimum Sum; in 2013, his is expeced o be abou $134,000 (CPF 2008b). Insead of having payous begin a age 62, as now, he MS is o be parially annuiized by defaul a age 65, so here is a 10-year deferral period from age 55 o 65. The MS will be spli ino a Term componen and an Annuiy componen, wih he spli depending on which plan he paricipan chooses. The Term invesmen amoun (T) plus ineres earned on T is inended o finance payous from age 65 o some older age Y, where Y may be eleced by he paricipan wihin some bounds (e.g. Y migh be eiher age 65, 75, or 85). The Annuiy componen (N) is inended o finance payous from age Y o deah. In any case, however, he requiremen is ha he member mus receive a fixed dollar payou every monh from age 65 o deah as long as he lives; he can also bequeah he Term amoun plus ineres minus payous. One oher decision o be made is wheher he paricipan wishes o have his remaining annuiy premium (N minus payous) provided o his heirs on his deah, which is called he refund porion. For example, he R80 plan sars he annuiy payou a age 80 and he annuiy would have a refund elemen; he NR65 plan sars he annuiy paymen a age 65 and has no deah benefi. The raionale for making he sysem compulsory is o preven adverse selecion, and i is logical o assume ha making he program mandaory will miigae his problem. On he oher hand, while he rules permi members o elec eiher he CPF-managed annuiy or a privae annuiy provider, i is unclear wheher privae firms will be able o compee. There is already some suggesion of a marke response: in 2008, for insance, only wo MSS annuiies were

19 19 offered by privae insurers, compared o nine he previous year. The shrinkage of privae annuiy offerings could indicae ha privae insurers are scrambling o re-price heir producs more compeiively, bu i could also indicae ha hey have been crowded ou more permanenly. In he fuure, privae annuiy providers migh insead refocus heir business ouside he CPF scheme for hose seeking o annuiize non-pension wealh. There is a cap o he amoun in he Reiremen Accoun ha CPF members can annuiize, namely he sipulaed Minimum Sum which will rise over ime. Very wealhy individuals seeking a higher monhly annuiy payou would sill urn o commercial annuiies. I is also worh nohing ha he plan will exemp CPF members who hold alernaive lifelong pensions or receive annuiies from he governmen-run annuiy scheme. Conclusions This chaper offers new evidence on he annuiy marke in Singapore focusing on producs offered by privae insurers, and i discusses possible impac of imposing mandaory annuiizaion hrough he naional CPF sysem. We show ha an average 55-year-old male could purchase annuiies providing a MWR value of abou 0.854, using populaion ables; his is similar o figures in oher counries. Bu he coss of adverse selecion in Singapore are much smaller han elsewhere, on he order of hree o four percen. For insance, Brown e al. (2001) find ha roughly half of he cos of purchasing a volunary annuiy in he US annuiy marke could be aribued o adverse selecion, whereas we find he proporion is less han one-fifh in Singapore. Wha his means is ha given he mos bes available daa annuiizaion coss in Singapore are more shaped by insurance company loadings han by adverse selecion.

20 20 For hese reasons, we would expec ha requiring mandaory annuiizaion in Singapore is likely o have lile impac on he money s worh valuaions of lifeime annuiy payous due o he eliminaion of adverse selecion. Insead, wha will enhance he value for money of annuiy payous is he fac ha he governmen will provide hem, presumably wihou he need o make a profi. To he exen ha aking he CPF-provided annuiy is he defaul, his will likely hold down adverising, markeing, and disribuion coss as well. Accordingly, he enry of he CPF Board ino he marke is expeced o narrow he radiional gap beween premiums and anicipaed benefis, and i will likely make he new payou producs quie aracive. Acknowledgemens The auhors acknowledge research suppor from by he Wharon-SMU Research Cener a Singapore Managemen Universiy, and he Pension Research Council/Boener Cener a The Wharon School of he Universiy of Pennsylvania. They also acknowledge helpful assisance and advice from Desmond Chew, Lee Ee Jia, Jean Lemaire, and Libby Sang.

21 21 References Brown, Jeffrey, Olivia S. Michell, James Poerba, and Mark Warshawsky (2001). The Role of Annuiy Markes in Financing Reiremen. MIT Press. Cenral Providen Fund Board (CPF) (2009a). Ineres Earned by Members. mycpf.cpf.gov.sg/members/gen-info/in-raes/in-raes.hm. Viewed 3/16/09. Cenral Providen Fund Board (CPF) (2009b). CPF Wihdrawal Calculaor. cpf.gov.sg/cpf_rans/ssl/financial_model/ wdl_assumpion.asp?prof. Viewed 3/16/09. Cenral Providen Fund Board (CPF) (2009c). Frequenly Asked Quesions on CPF LIFE. mycpf.cpf.gov.sg/members/gen-info/cpf_life/cpf_life.hm. Viewed 3/16/09. Cenral Providen Fund Board (CPF) (2008a). CPF Minimum Sum Scheme: Table of Monhly Paymen Raes for he Minimum Sum Of $99,600 placed wih a Paricipaing Insurance Company. mycpf.cpf.gov.sg/nr/rdonlyres/883dd6d3-d1ec-48dd-8b09-811c0d24ece8/0/annuiycomparison.pdf. Viewed 3/20/08. Cenral Providen Fund Board (CPF) (2008b). Repor by he Naional Longeviy Insurance Commiee (NLIC), February. mycpf.cpf.gov.sg/members/gen- Info/CPF_LIFE/NLIC.hm. Viewed 3/16/09. Cenral Providen Fund Board (CPF) (2007a). CPF Trends: Changing Age Srucure of CPF members: Membership in Brief. mycpf.cpf.gov.sg/nr/rdonlyres/1103e583-19a4-4b62- A38D-8D06B5C0E071/0/ChangingAgeSrucureofCPFMembers.pdf. Viewed 12/31/08. Cenral Providen Fund Board (CPF) (2007b). CPF Trends: Minimum Sum Scheme. mycpf.cpf.gov.sg/nr/rdonlyres/27aca8e5-7d0d-431f-b74b- D8D3A3E43463/0/CPFTrends_MSS.pdf. Viewed 12/31/08.

22 22 Doyle, Suzanne, Olivia S. Michell, and John Piggo (2004). Annuiy Values in Defined Conribuion Reiremen Sysems: Ausralia and Singapore Compared, Ausralian Economic Review, Vol. 37, No. 4, December: Finkelsein, Amy, and James M. Poerba (2004). Adverse Selecion in Insurance Markes: Policyholder Evidence from he U.K. Annuiy Marke, Journal of Poliical Economy, 112, 1, February: Finkelsein, Amy, and James M. Poerba (2002). Selecion Effecs in he Unied Kingdom Individual Annuiies Marke, The Economic Journal, Vol. 112, Issue 476, January: Fong, Wai M. (2002). On he Cos of Adverse Selecion in Individual Annuiy Markes: Evidence from Singapore, Journal of Risk and Insurance, Vol. 69, No. 2, June: Koh, Benedic S. K., Olivia S. Michell, Too Tanuwidjaja, and Joelle H.Y. Fong (2008). Invesmen paerns in Singapore's Cenral Providen Fund Sysem, Journal of Pension Economics and Finance, Vol. 7, Issue 01, March: Michell, Olivia S., James Poerba, Mark Warshawsky, and Jeffrey Brown (1999). New Evidence on he Money s Worh of Individual Annuiies, American Economic Review. December: Michell, Olivia S. and Jose Ruiz (2009). Pension Paymens in Chile: Pas, Presen, and Fuure Prospecs, This volume. Moneary Auhoriy of Singapore (MAS) (2008). ( NTUC Income (2009). Annuiy Plans. income.com.sg/insurance/annuiy. Viewed 3/16/09. Singapore Deparmen of Saisics (SDOS) (2008a). Saisics: Demography (Themes). singsa.gov.sg/sas/hemes/people/demo.hml. Viewed 3/16/09.

23 23 Singapore Deparmen of Saisics (SDOS) (2008b). Complee life ables for for he Singapore Residen Populaion, Informaion Paper, Singapore, May. Singapore Prime Miniser s Office (SPMO) (2007). Transcrip of Prime Miniser Lee Hsien Loong s Naional Day Rally English Speech: Ciy of Possibiliies; Home for All, Singapore, 19 Augus. Singapore Prime Miniser s Office (SPMO) (2003). Miniserial Saemen by Prime Miniser Goh Chok Tong in Parliamen: Reuning he CPF, Singapore, 28 Augus. Thorburn, Craig, Robero Rocha, and Marco Morales (2005). An analysis of Money s Worh Raios in Chile, Journal of Pension Economics and Finance, Vol. 6, Issue 03. November: World Healh Organizaion. (Various years.) Life Tables for WHO Member Saes. who.in/whosis/daabase/life_ables/life_ables.cfm. Viewed 3/16/09.

24 24 Table 1: The Minimum Sum Scheme (MSS) Schedule, Period Required Required Minimum Sum Minimum Sum (2003 Dollars ) (Real Dollars) Drawdown Age MSS Payou Srucure 2003 $80,000 $80, Defaul: Phased wihdrawal 2004 $84,000 $84, (adminisered by CPF) $88,000 $90, Alernaive: Volunary purchase 2006 $92,000 $94, of annuiies $96,000 $99, {Choose from MSS annuiies 2008 $100,000 $106, offered by privae insurers $104, Saring from 2009, paricipans 2010 $108, may also choose from LIFE 2011 $112, annuiies} $116, $120, Defaul: Compulsory annuiizaion. {Choose from MSS annuiies offered by privae insurers or LIFE annuiies offered by CPF}. Alernaive: nil. Source: Auhors compilaions; see ex. Noes: As of 2003, he Minimum Sum will be raised from S$80,000 o $120,000 by 2013 (in 2003 dollars) and adjused for inflaion. The Draw-Down Age refers o he official age a which he member may sar drawing down he Minimum Sum plus ineres. For example, a member who urns 55 as a 1 Jul 2007 mus se aside $99,600 as he Minimum Sum in his Reiremen Accoun a age 55 and may sar draw down only when he reaches age 64. If he dies before drawdown sars, he balance in he Reiremen Accoun will be fully refunded o his beneficiary. From , CPF is operaing a volunary annuiy purchase scheme; life annuiies may be purchased wih he Minimum Sum volunarily from qualifying annuiies offered by privae insurers. Saring from 2009, paricipans may also choose from LIFE annuiies offered by CPF. From 2013, CPF will operae a mandaory annuiizaion scheme wih compulsory purchase of life annuiies using he Minimum Sum; annuiies will be offered by CPF-LIFE and privae insurers.

25 25 Table 2: Monhly Nominal Payous for Life Annuiies purchased a he Minimum Sum of $99,600 (2007; S$ per monh) Company & Produc Monhly Annuiy payou for enry age of 55 Male (S$) Female (S$) Guaraneed amoun upon deah Non-paricipaing Annuiies Asia Life Assurance Premium - oal annuiy paymens. Prudenial Assurance Premium - oal annuiy paymens. American Inernaional Premium - oal annuiy paymens. Assurance (AIA) Grea Easern Life (GE Life I) Premium + ineres accumulaed a 0.75% p.a. o age 62 - oal annuiy paymens Overseas Assurance Corporaion (OAC) Premium + ineres accumulaed a 0.75% p.a. o age 62 - oal annuiy paymens. Aviva Premium + accrued ineres compounded a 1% p.a. o annuiy commencemen dae - oal annuiy paymens. Grea Easern Life (GE Life II) [Noe: This produc includes long erm care benefi.] Sub-average Premium + ineres accumulaed a 0.5% p.a. o age 62 - oal annuiy paymens. Paricipaing Annuiies NTUC Income Co-op ( # ) HSBC Insurance ( # ) ( # ) ( # ) Sub-average Premium + ineres accumulaed a 2.5% p.a. and bonuses o age 62 - oal annuiy paymens. Premium + ineres accumulaed a 2% p.a. o age 62 - oal annuiy paymens. Average wihou bonus adjusmen Average wih bonus adjusmen Source: Auhors compuaions; derived from CPF (2008a). Noes: Monhly payous for a nominal deferred annuiy purchased a age 55 wih paymens saring a age 62. The lump-sum premium is he Minimum Sum of S$99,600 for members age 55 (July 07-June 08). Annuiies under he MS Scheme currenly guaraneed for a given amoun in he even of annuian s deah; he posiive difference of he guaraneed amoun less annuiy

26 paymens made would be paid o nominaed beneficiaries. Previously (in 2000) mos MS annuiies were guaraneed for a cerain period so if deah occurred during he guaraneed period, remaining annuiy paymens would be convered ino a lump sum paid o beneficiaries. # Bonus raes depend on company performance; NTUC Income s annual bonus raes were percen hisorically; a 2 percen bonus is used in NTUC Income benefi illusraions. Original payous wihou bonus expressed wihou brackes; figures in brackes incorporae bonus componen assuming an annual projeced bonus rae of 2 percen and a projeced annual invesmen rae of reurn of 5 percen. 26

27 27 Table 3: Money s Worh Raios Compued for MSS Life Annuiies Sudy Fong W.M. (2002) Doyle, Michell and Piggo (2004) Presen sudy (2009) Valuaion Dae & Sample chosen 2000; subse of 8 nonparicipaing annuiies & 1 paricipaing annuiy. 2000; subse of 5 nonparicipaing annuiies wih a 15-yr guaranee period or similar. 2007; all MSS annuiies: 7 nonparicipaing & 2 paricipaing annuiies. Moraliy Assumpion Ineres Rae Assumpion MWR for 55- year-old Male Adverse selecion Annuian Populaion Ann. Pop. (%) a(90) wih 2- year seback. Limiing age used is 109. a(90) wih 2- year seback. Limiing age used is 109. a(90) wih 5- year seback. Limiing age used is 117. Derived from 1960 and 1990 Ordinary Male and Female Lives Tables (Singsa). Limiing age of 99. Abridged life ables for Singapore (World Healh Organizaion). Limiing age of 100. Complee life ables for Singapore residen populaion 2007 (Singsa), plus exrapolae from age 100 o 117. Limiing age of 117. Fla ineres rae (proxy by he 10-year Governmen bond yield of 4.6%). Term srucure (yield curve wih long-erm rae assumpion of 4.76%). Term srucure (yield curve wih long-erm rae assumpion of 3.44%) Source: Auhors compuaions; see ex. Noes: A oal of 13 MSS life annuiies were offered in July 2000 of which 9 were fla-rae annuiies, 2 were paricipaing annuiies and 2 were increasing annuiies. The increasing annuiies offered by AIA were dropped afer ha year (Source: Personal communicaion from CPF Board). The a(90) able refers o he UK a(1990) period life able for annuians. I is based on UK annuians experience from , wih moraliy improvemens projeced o Because of lack of annuian experience in Singapore, previous sudies used he a(90) and wih a wo-year seback o accoun for lower moraliy among annuians. A wo-year seback means ha a 65-year-old is reaed as having he same moraliy rae as a 63-year-old has in he iniial able.

28 28 Table 4: Money s Worh Raios and Adverse Selecion Cos of MSS Annuiies (nominal life annuiies offered by privae insurers under CPF plan; 2007) Company & Produc Populaion MWR Male Annuian MWR Adverse Selecion (%) Populaion MWR Female Annuian MWR Adverse Selecion (%) Non-paricipaing Annuiy Asia Life Assurance Prudenial Assurance AIA GE Life I OAC Aviva GE Life II Paricipaing Annuiy NTUC Income Co-op HSBC Insurance Mean Source: Auhor s compuaions, see ex. Noes: Money s worh raios are in decimals. Adverse selecion coss are in percenage poins. Compuaions perain o a CPF paricipan who purchases he MSS annuiy a enry age 55 for a premium of $99,600, and sars receiving payous a age 62.

29 29 Figure 1: Cumulaive Cohor Survival Probabiliy: General Populaion and Annuian groups (condiional on aaining age 55 and limiing age of 117; 2007) A. Singaporean Males 1.0 Cumulaive survival probabiliy Age Populaion Annuian B. Singaporean Females 1.0 Cumulaive survival probabiliy Age Populaion Annuian Source: Auhors compuaions; see ex.