International Monetary Fund Washington, D.C.

Transcription

1 2007 Inernaional Moneary Fund February 2007 IMF Counry epor No. 07/7 Cyprus: Seleced Issues his Seleced Issues paper for Cyprus was prepared by a saff eam of he Inernaional Moneary Fund as background documenaion for he periodic consulaion wih he member counry. I is based on he informaion available a he ime i was compleed on January 2, he views expressed in his documen are hose of he saff eam and do no necessarily reflec he views of he governmen of Cyprus or he Execuive Board of he IMF. he policy of publicaion of saff repors and oher documens by he IMF allows for he deleion of marke-sensiive informaion. o assis he IMF in evaluaing he publicaion policy, reader commens are invied and may be sen by o publicaionpolicy@imf.org. Copies of his repor are available o he public from Inernaional Moneary Fund Publicaion Services 700 9h Sree, N.W. Washingon, D.C elephone: (202) elefax: (202) publicaions@imf.org Inerne: hp:// Price: $8.00 a copy Inernaional Moneary Fund Washingon, D.C.

2

3 INENAIONAL MONEAY FUND CYPUS Seleced Issues Prepared by Alexander W. Hoffmaiser (EU), Mario Caalan and Jaime Guajardo (INS), and Andrei Levchenko (ES) Approved by he European Deparmen January 2, 2007 As background o he 2006 Aricle IV Consulaion, his Seleced Issues paper explores he economic consequences of aging and is impac on long-run fiscal susainabiliy. he sudy analyzes he poenial macroeconomic impac of differen approaches o deal wih he fiscal coss of aging. he sudy goes beyond a simple quanificaion of he fiscal impac by explicily examining he rade-offs of alernaive policies wihin he conex of a general equilibrium overlapping generaion framework. his approach capures an individual s responses o changing incenives ha follow pension reform. I concludes ha addressing he fiscal consequences of aging will require increasing he reiremen age o 65 years, followed by furher increases o keep up wih demographic rends. Moreover, he simulaions verify he harmful macroeconomic effecs of relying on increases in labor axes insead of consumpion axes. he consequen reducion in welfare reflecs he fac ha he burden of labor axes falls on he shoulders of a dwindling working force. Also, swiching he indexaion of pension benefis o prices can help moderae age-relaed spending and, by seing he sage for lower axes han oherwise possible, proec ne-of-ax pension benefis. ebalancing he generous public pension benefis wih hose of he privae secor furher improves he macroeconomic resuls. he auhors are hankful o Zenon Konolemis for invaluable help in undersanding he Cyprio pension sysem, and he paricipans of he informal seminar held in Nicosia on Ocober 30, 2006.

4 2 Conens Page Pension eform: Addressing he Consequences of Aging...3 A. Inroducion...3 B. he Cyprio Pension Sysem...8 C. he Cyprio Sysem in Conex and eforms in Seleced Counries... D. he Model and Simulaion...5 E. Summary of esuls and Conclusions...22 ables. Key Demographic rends Affecing he Pension Sysem, Pension Expendiures Pension eform Scenarios...26 Figures. Populaion and Employmen, Baseline Scenario wih Consan and Variable Ineres aes Increasing he eiremen Age and Payroll axes (Consan Ineres aes) Increasing he eiremen Age and Payroll axes (Variable Ineres aes) Addiional eforms Scenario (Consan Ineres aes) Addiional eforms Scenario (Variable Ineres aes)...40 Appendixes I. he Model...43 II. Value Funcion a eiremen...52 eferences...6

5 3 PENSION EFOM: ADDESSING HE CONSEQUENCES OF AGING A. Inroducion. In he coming 50 years, he populaion of Cyprus is expeced o age suanially. While people over 65 years of age currenly make up 2 percen of he populaion, ha raio is se o double o over 25 percen by Aging will be even more pronounced for he very elderly. he aolue number of people over he age of 80 is se o quadruple beween 2004 and 2050, more han ripling heir share in he oal populaion, from abou 2½ percen o over 8 percen. Mirroring his, he share of he working-age populaion (age 5 64) is se o decrease from abou 68 percen o 6 percen over he same period (ex figure, and Figure ). 2 hese aging rends are in line wih, hough slighly less severe han, hose in he EU (ex able). able 2. Cyprus and EU: Populaion by Age, (Percen of oal populaion) Working age (5-64) Cyprus EU Elderly (65) Cyprus EU Very elderly (80) Cyprus EU Source: European Commission, Aging Working Group. Age Age Pyramid for Cyprus, 2004 and Populaion (in housands) Source: European Commission, Aging Working Group. 2. Jus as in many oher counries, declining feriliy raes and increasing life expecancy are underlying hese aging rends (able ). In paricular, while as lae as 980 he Cyprio feriliy rae was 2.5 birhs per woman, by 2005 i had dropped o.5, and is projeced o remain a ha level in he medium erm. Also, life expecancy a birh increased from 70 years for men and 73 years for women in 970 o, respecively, 77 years and 8 years in Moreover, furher gains in life expecancy of abou five years are expeced by I is noeworhy ha hese populaion projecions assume a suanial amoun of immigraion. Cumulaive ne arrivals are projeced o exceed a quarer of a million people in he course of he firs half of his cenury (ex able). Wih no immigraion, he populaion would decrease by 3 percen beween now and 2050, exacerbaing he aging problem 2 he populaion forecass used in his sudy correspond o hose of Eurosa s baseline projecions.

6 4 correspondingly. Insead, in he baseline projecion, he populaion of Cyprus grows by 33.5 percen by oal Populaion Immigran populaion / Share of immigrans (In percen) Sources: Eurosa and Saff Projecions. Cyprus: Populaion Projecions and Immigraion (In housands, unless oherwise indicaed) / eflecs projeced immigraion flows saring in 2005 and is esimaed as he difference beween he baseline populaion projecion and he zero ne migraion scenario. 4. hese demographic rends challenge he long-run susainabiliy of he pension sysem in Cyprus. ecen esimaes by he EC s Aging Working Group (AWG) for he payas-you-go sysem (PAYG) sugges ha, wihou reforms, pension expendiures in Cyprus will rise by 2.9 percenage poins of GDP by 2050, compared wih an average increase of less han 3 percenage poins in he EU (ex able). Alhough hese esimaes accoun for all of he elemens of he sysem including he general and he civil servan regimes he projeced increase may be oversaed as i does no reflec he recen increase in he reiremen age of civil servans. Sill, even afer accouning for his reform as in he baseline simulaion discussed below old-age expendiure will increase suanially. In large par, expendiure will rise because he sysem has been essenially unchanged since i was inroduced in 980, when he life expecancy was 75 years 5 years less han i is currenly and he feriliy rae was wice as high. hus, by 2050, expendiure is poised o leap o among he highes in he EU from is curren low posiion (ex figure). Cyprus: he Demographic Shock (In percen of GDP, unless oherwise noed) Pension Expendiure (in percen of GDP) Old-Age Dependency aio / Public Pension Expendiures Peak Pension Expendiures Change o Change o Peak Year Value Cyprus EU EU EU Source: Eurosa, and European Commission, Aging Working Group. / aio of dependens o oal populaion (in percen), where dependens are hose above 64 years of age Esonia Lavia Mala Poland Lihuania UK Slovakia Ireland Neherlands Sweden Ausria Denmark EU-25 Germany Finland Czech epublic Ialy France Belgium Spain Hungary Luxembourg Slovenia Cyprus Porugal Source: European Commission, Aging Working Group

7 5 5. Healh expendiures are also se o increase. he problem in Cyprus is less severe han in he res of he EU. Cyprus s curren public healh expendiure is 2.9 percen of GDP, less han half he EU-25 average (ex figure), alhough healh expendiures have roughly doubled as a share of GDP in he pas 20 years. Expendiures are expeced o increase by. percenage poins of GDP by 2050, agains he EU average increase of.6 percenage poins Healh Expendiure (in percen of GDP) Cyprus Lihuania Poland Mala Lavia Luxembourg Slovakia Esonia Hungary Greece Ausria Finland Ialy Germany Porugal Ireland Neherlands Belgium Sweden Denmark Slovenia Spain Czech epublic UK France Source: European Commission, Aging Working Group Agains his background, in 2005 he governmen announced increasing he reiremen age of civil servans from 60 o 63 by Alhough his reform places civil servans reiremen age in line wih he average reiremen age in Cyprus, i sill me wih siff opposiion from rade unions. he auhoriies are considering exending such increases o educaional services and semi governmen organizaions. In addiion, he Convergence Program considers wo waves of parameric reforms for he Cyprio pension sysem. hese reform would include ighening he crieria for pension eligibiliy and early reiremen hereby increasing he effecive reiremen age o 65 years and increasing social securiy conribuions. he ime line for hese reforms has no been spelled ou. 7. his sudy analyzes he poenial macroeconomic impac of pension reform for Cyprus using an overlapping-generaions general equilibrium model. Alhough ohers have measured he fiscal implicaions of aging in Europe and in Cyprus (noably he EC s AWG and he auhoriies Acuarial Noes) hese exercises have largely involved exrapolaing pas macroeconomic rends in ligh of demographic profiles. hose resuls are useful in illusraing he implausible adjusmens in conribuions or benefis ha, in he aence of broader reforms, would be required o ensure he susainabiliy of he pension sysem. Bu hey provide a parial view of he implicaions of reforms, as an individual s behavior is aken as independen of he implici incenives change. 8. he model used in his sudy in he radiion of Auerbach-Kolikoff endogenizes an individual s lifeime labor-leisure and consumpion-savings decisions. In each period, and hroughou a person s work life, an individual maximizes his or her uiliy by deciding how much ime o devoe o work and how much o save. An individual s produciviy, moreover, changes as he or she gains work experience. Also, he model used in his sudy explicily capures key insiuional feaures of he Cyprio pension sysem, including he differen regimes for privae and public employees. he model assumes ha Cyprus is a small, open economy and hus faces an exogenously deermined world ineres rae. As discussed below, each scenario is simulaed under wo alernaive world ineres rae pahs. he firs is he unlikely case of unchanged ineres raes. he second mimics he pah of world ineres raes, reflecing worldwide aging paerns on capial flows (Börsch-Supan,

8 6 Ludwig, and Winer, 2005); a conservaive pah is considered in his case o illusrae he sensiiviy of he simulaion resuls o world ineres rae developmens. Also, he model includes populaion growh capuring he ransiional effecs of he demographic shock and micro- and macro-economic feedback effecs. his sudy is hus beer able o explore he macroeconomic effecs of pension reform in Cyprus he model also accouns for he effec of aging on healh care spending. Specifically, healh care expendiures in he model reflec he j-shaped profile of healh care spending over a household s life (ex figure). In oher words, healh care expendiure is ypically a bi higher for young children han for young aduls bu rises sharply laer in life. For Cyprus, he simulaions use he healh care profile oerved for large European counries, scaled o mach healh care expendiure given Cyprus s demographic profile. 4 his approach, however, is incomplee. I does no accoun for he impac of medical advances underlying he developmen of new medical reamens and demand for new capial-inensive reamens (Heller, 2003), nor does i reflec he Healh Care Expendiure by Age Group (Share of GDP per capia) France poenial beneficial impac on an individual s produciviy semming from advances in healh he model is used o analyze he macroeconomic implicaions of hree policy scenarios: Baseline scenario. his provides a benchmark o conras he macroeconomic effecs of he pension reform. Individuals ener he workforce a age 23 and, having worked Germany Spain Ialy Source: OECD working paper, Spending on healh and long-erm care: projecions o 2050 revisied. 3 In he model, he labor force enry (inclusive of immigraion flows) was calibraed so ha, ogeher wih gains in life expecancy, he model mached he dependency raio pah in Eurosa s populaion projecions. 4 Specifically, he profile used is ha of Spain, which is roughly an average of he profiles of France, Germany, and Ialy. Over ime, healh care coss for individuals increase wih produciviy gains, and hus he healh care expendiure profile (as a share of GDP) remains consan. 5 Also, he analysis does no accoun for he inroducion of he Naional Healh Insurance Scheme in 2008, which will phase in much-needed compeiion in he provision of healh care. Public hospials will remain shelered from compeiion for he firs hree years.

9 7 for 40 years, reire a age 63. Afer reiremen, privae and public secor employees receive heir corresponding pension benefis hrough he remainder of heir lifeime. A he ouse of he simulaions, he reiremen period is 8 years, corresponding o a life expecancy of 80 years. In he simulaions, he reiremen period increases wih life expecancy projecions (one year per decade) unil i reaches 28 years in 00 years, ha is, in 208. he economy benefis from (labor augmening) echnological progress 2½ percen per year in line wih hisorical rends. he scenario assumes ha consumpion axes are adjused as needed o keep public deb unchanged in percen of GDP; also governmen consumpion remains consan as a share of GDP. 6 he Convergence Program (CP). his scenario is designed o capure he key feaures of he parameric reforms envisaged in he CP. Specifically, i considers he effecs of increasing he effecive reiremen age from 63 years o 65 years by 208 and adjusing social securiy conribuions by 5 percenage poins. hese reforms are aken o be announced in 2008, and, effecive immediaely hereafer, social securiy conribuion raes would be increased. Saring in 203 (208) individuals would work one (wo) more year(s) before reiring. Alhough he model does no accoun for addiional once-and-for-all improvemen associaed wih a ighening of eligibiliy requiremens, 7 savings in old-age pensions are suanial. In large par, his is because he implemenaion of hese reforms assumes limied grandfahering. Addiional parameric reforms. As improvemens in life expecancy are envisaged for he nex several decades, his scenario proposes o furher increase he reiremen age gradually o ensure ha he pension sysem keeps up wih demographic rends. Specifically, afer he reiremen age reaches 65 years, for he nex 00 years i would increase one year per decade unil i reaches 73 in o furher limi age-relaed expendiure increases, and, in conras wih he CP, his scenario also includes swiching he indexaion so ha all pension benefis are indexed o prices. In addiion, he scenario addresses he mismach in generosiy beween public and privae pension benefis by gradually reducing he lump-sum paymen awarded o public secor employees a reiremen from 28 monhs o 2 monhs of pay.. he simulaions sugges ha he severiy of he macroeconomic consequences of aging can be miigaed by implemening parameric pension reforms. No reforming he social securiy sysem would suanially reduce oupu growh and economic welfare for 6 Public deb can be lowered by prefunding he demographic shock, ha is by increasing ax raes ahead of ime, and/or by reducing public consumpion. 7 he CP proposes ighening he eligibiliy requiremens for old-age pensions. Currenly, individuals who have worked a leas 0 years and conribued o social securiy for 3 years are eligible for old-age pension benefis a age 65. Eligibiliy for a basic pension requires he average annual income and conribuion period o be a leas 25 percen of he basic earnings and 3 years, respecively. Inroducing his effec in he model requires heerogeneiy in individuals, he complexiy of which lies beyond he scope of his analysis.

10 8 decades. Alhough he reforms envisaged in he CP lower age-relaed expendiure, hey would no suffice, as fuure gains in life expecancy will offse hese savings. Moreover, he perverse incenives o work associaed wih social securiy rae increases, end o lower household s welfare. As discussed below, addiional reforms are needed o keep up wih demographic rends, limi welfare losses and align he generosiy of public and privae pension schemes. B. he Cyprio Pension Sysem 2. he pension sysem comprises he General Social Insurance Scheme (GSIS), he Governmen Employees Pension Scheme (GEPS) and a few smaller schemes. In 2005, oal pension expendiure in Cyprus was abou 8.5 percen of GDP (able 2). Expendiures in he GSIS will grow from 6.3 percen of GDP in 2005 o 2.6 percen in 2050 (epublic of Cyprus, 2005). Alhough he PAYG GSIS has an elemen of prefunding as i currenly runs a surplus and accumulaes reserves he auhoriies esimae ha social conribuion raes (payroll axes) will have o increase by abou 5 percenage poins o 22.9 percen by Expendiures in he GEPS are se o grow from 2.2 percen of GDP in 2005 o 4. percen of GDP in 2050 (epublic of Cyprus, 2005). he GEPS is financed by (reduced) conribuions of public secor employees and ax revenues. he curren pension sysem was esablished in Sepember 980, 8 and is componens are briefly discussed in urn. he General Social Insurance Scheme (GSIS) 3. he GSIS is a universal, compulsory pension scheme and by far he larges componen of naional pension expendiures. I accouns for abou hree-fourhs of oal old-age pension expendiure (able 2). I is funded by riparie conribuions oaling 6.6 percen of earned income (ex able). Conribuions, however, are capped, as income in excess of C 2,077 per monh is no subjec o compulsory conribuions. Pas favorable demographic rends have generaed large surpluses; reserves have reached he equivalen o 37 percen of GDP, or abou eigh imes he annual pension oulays. Almos he enire reserve is invesed in cenral bank bonds. Conribuion aes o Social Securiy (In percen of earnings) General social insurance scheme / 6.6 Employee 6.3 Employer 6.3 Governmen 4.0 Governmen employee pension scheme 2/ 3.2 Employee 3.2 Source: he Minisry of Labor and Social Insurance. / Income in excess of CYP 2,077 monhly is no subjec o compulsory conribuion. 2/ he governmen does no conribue a fixed share of earnings. Benefi paymens exceeding employees' conribuions are financed hrough he budge. 8 he previous sysem based on fla-rae conribuions and benefis had been in place since he lae 950s. I esablished reiremen a 65 years of age (lowered o 63 years in 99) and provided a suisence-level benefi, which, in essence, was preserved in he form of he basic pension in he curren sysem.

11 9 4. he old-age pension benefi under he GSIS comprises an earnings-relaed basic and supplemenary pension benefis as follows: Pension = Basic Supplemenary, GSIS i i i where and i denoe he reiremen year and he number of years since reiremen. hese componens reflec an individual s work life and annual income earnings, Basic Ps = α BasicEarnings InsurableLife LowerBand i B i, and where CPI i Supplemenary i = α S PsUpperBand BasicEarnings, CPI α denoes he replacemen rae and he sucrip denoes he benefi componen (0.6 for he basic componen and 0.05 for he supplemenary componen); 9 BasicEarnings corresponds o he sauory earnings se each year and expressed as an annual amoun (in 2005, i was se a C 79.9 weekly, or C 4,55 annually); Ps, represens cumulaive poins earned oward each pension componen: Ps = ps. Poins per year are calculaed as ps = wages / BasicEarnings, wih = he firs poin accruing o he lower band (basic pension) and remaining poins accruing o he upper band (supplemenary pension); InsurableLife is he number of years in he workforce since an individual s 6 h birhday (or since Ocober 5, 964, whichever is less); 0 and CPI is he consumer price index. he basic pension benefi is compued each year based on he corresponding basic earnings. hus, i is indexed o wages because basic earnings are. In conras, he supplemenary pension is compued once, given he basic earnings a he ime of reiremen, and suequenly adjused by inflaion. 9 α B increases by 0. for each dependen child and by 0.2 if an individual is married. 0 If an individual s wages exceed BasicEarnings every year, he basic componen of he pension would be Basic = α BasicEarnings. B i i

12 0 5. Currenly, he basic pension is abou 27 percen of he average insurable earnings (up o C 24,924 per year). his is abou 20 percen of he average income earnings. he supplemenary benefi virually doubles his amoun, and hus he afer-ax replacemen rae is close o 60 percen. Individuals who have worked a leas 0 years and conribued o social securiy for 3 years are eligible for old-age pension benefis a age 65. Sill, individuals who have conribued o social securiy for 28½ years can reire a 63, and many do so: he average reiremen age is close o 63. o be eligible for he basic pension, Basic i, average annual income during an individual s work life mus be a leas 25 percen of he basic earnings, and he insurable life mus be a leas hree years. Also, noe ha here is an upper limi on his annual pensionable income ( C 24,924 in 2005). Beyond ha level, he social pension conribuions sop, and insurable poins are no earned. he Governmen Employees Pension Scheme (GEPS) 6. he GEPS is a compulsory pension scheme for public secor employees and he second-larges social insurance expendiure. In 2005, i accouned for abou 25 percen of he oal social insurance expendiure and covered abou 30,000 people. I is funded by employee conribuions 0.75 percen of earning up o he maximum insurable earnings and.75 percen above ha level and shorfalls in conribuions are financed by he public purse. he governmen employee scheme has a reiremen age of 60, which is se o increase o 63 by For cerain careers, such as police, he reiremen age is 55. Furhermore, civil servans are allowed o reire five years early wihou an early reiremen penaly. he average age of reiremen of governmen workers is he reiremen pension benefi under he GEPS comprises he basic pension, a monhly pension and a oneime lump-sum paymen a reiremen. Besides he basic pension, which is compued in he same manner as for privae households, public secor employees receive an addiional monhly pension benefi ha is a fracion α of he final salary, Salary,: Pension = Basic Public GPES i i i, Basic Public PsLowerBand = α BasicEarnings InsurableLife Wages i = α Salary. Wages i B i i he replacemen rae α is equal o he number of monhs of service, up o a maximum of 400 monhs, divided by 800. hus, he pension is a mos 50 percen of a governmen employee s, Provided ha he average salary is a leas equal o he basic earning.

13 final salary. Suequenly, he GEPS benefi is indexed o wages. Besides he basic and monhly pension benefi, a reiremen governmen employees receive a lump-sum paymen Lump Sum = λ Salary, equal o 28 (λ) imes heir final monhly salary if hey have served a leas 400 monhs; a proraed amoun is paid o hose wih shorer service. 8. Employees of he semigovernmen insiuions have heir own pension schemes. he semigovernmen employees pension schemes, which cover employees of public uiliies, local governmens, and similar eniies, are funded by he respecive employers. In oher words, shorfalls in hese pension schemes will also be refleced in he governmen s budge. However, hese represen a much smaller porion of old-age spending. Oher componens 9. hese componens comprise he Special Allowance o Pensioners, he Social Pension Scheme and Volunary Providen Funds. hese are considerably smaller as a share of oal pension expendiure, and are as follows: he Special Allowance (beween C per year) graned o anyone whose pension income is below C6,500 (ranging abou wo o hree imes he basic pension). his allowance is financed by ax revenues and accouns for 6 percen of oal pension expendiures by he governmen. he Social Pension scheme designed o provide a minimum sandard of living o hose who for whaever reason, did no ake par in he GSIS; i provides a minimumsuisence sandard of living. he amoun of pension is se a 8 percen of he basic social insurance pension, which iself is abou 25 (20) percen of he median (mean) labor income. he pension is also financed by ax revenues and amouns o 3 percen of oal pension expendiures by he governmen. Volunary Providen Funds are employer- or profession-based defined-conribuion savings schemes ha provide a lump-sum paymen upon reiremen. hese are no widely available: for he majoriy of paricipans in he GSIS, he universal pension was heir only pension income. Less han 43 percen of pensioners benefi from addiional pension income. C. he Cyprio Sysem in Conex and eforms in Seleced Counries 20. Several elemens of he Cyprio pension sysem are similar o hose in advanced counries. Like Cyprus, mos OECD counries have a defined-benefi scheme. he sauory reiremen age in he privae secor in Cyprus is roughly in line wih he OECD averages. In mos of he advanced counries, pension amouns are calculaed based on average lifeime earnings, a mehod ha is similar o he Cyprio poin sysem (discussed above). In abou half

14 2 of he OECD counries, pensions are indexed o wages, while in he oher half hey are indexed o prices. As a resul, he replacemen raes in Cyprus are similar o OECD averages. Because many of he advanced counries also face demographic pressures, many pension sysems have undergone reforms. his secion briefly reviews he pension reform experiences of hree counries. 2 Greece illusraes a highly complex sysem wih a number of differen pension schemes. Ialy presens he case of a sysem wih wage indexaion and differenial reamen of public and privae secor employees and reforms ha swiched indexaion o prices and unified he public and privae secor pension sysems. Sweden offers an alernaive sysem wih innovaive elemens. Greece 2. he Greek pension sysem comprises of a large number of self-governed social insurance funds, which arose in par from prior occupaional schemes. As a resul, rules for conribuions, invesing, and payous vary suanially. Effors have been under way for some ime o reduce hese differences and consolidae he funds, including by merging hem wih he larges (IKA). Sill, in 2005, 73 social insurance funds remained, of which 24 were primary funds which provide he main pension, and he res were supplemenary funds. All are backed by he governmen, and, herefore. all are included in he general governmen fiscal posiion. 22. he Greek pension sysem as a whole already has a significan defici, and oulays are projeced o rise sharply in he years ahead. According o Gagales and oehler (2006), under he baseline scenario he curren pension sysem is in a dire sae, wih he public deb se o reach some 400 percen of GDP by 2050 in he aence of policy adjusmen. 23. he governmen began a series of reform aemps in he lae 980s. However, i was no unil ha a deep crisis in Greece s public finances, combined wih he iniiaion of EMU negoiaions promped a firs wave of reforms: he Sioufas Law of 990 inroduced some spending cus and inroduced gradual increases in conribuions bu fell well shor of correcing he srucural deficiencies of he sysem. he Manos Law of 992 was a suanially waered-down version of an iniial comprehensive reform draf ha had proved poliically infeasible. he reform was largely parameric: reiremen age and conribuions were increased, bu he fragmenaion of he sysem and he inequiies of provision remained. Sill, is susainabiliy was ensured hrough For more deailed reamen and furher counry cases, see OECD (2005).

15 3 he Spraos epor of 996, he Yannisis epor of 200, and oher reform iniiaives were hwared by fierce poliical opposiion. he eppas package of 2002, aking such opposiion ino accoun, provided for differenial reamen of older and younger workers, and was included along wih a large (.3 billion) up-fron injecion of governmen funds, ino he larges social securiy fund. he sum oal of measures was designed o ensure pension fund susainabiliy o 2030, bu a he cos of annual ransfers equivalen o percen of GDP from he sae budge o he pension funds. Ialy 24. he Ialian pension sysem is based on noional accouns. his amouns o a modified pay-as-you-go public pension sysem. Individual conribuions earn a (noional) rae of reurn ha is relaed o real GDP growh. he resuling pension benefis are a funcion of accumulaed noional capial and an acuarial facor; he laer reflecs average life expecancy a reiremen. his sysem applies o labor marke enrans from 996 onward. 25. Pension expendiures increased rapidly in Ialy during he 970s and 980s, as he sysem was becoming more inclusive. By 992, pension expendiure was 4.9 percen of GDP, considerably higher han he EU average. While Ialy has he highes pension expendiure as a share of GDP in he EU, his share, according o he laes EU esimaes, is projeced o increase minimally beween now and hese projecions, however, may be based on overly opimisic assumpions abou growh and produciviy, as well as full implemenaion of he above reforms. Alernaive projecions by IMF saff as well as hose by some academics sugges much higher increases if labor produciviy fails o suanially exceed he rae achieved in Ialy over he las decade. 26. hree successive rounds of reforms sough o conain rising spending and ensure long-run susainabiliy: he 992 Amao reform raised he reiremen age by 5 years; lenghened he reference period for calculaing pensions from 5 o 0 years, and o he enire working life for younger workers; raised he minimum number of years of conribuions o be eligible for a pension from 5 o 20 years; changed he indexaion of pensions from wages o prices; and harmonized pension rules across public and privae secor workers. he reform was widely seen as a success, reducing he ne presen value of pension liabiliies by more han 00 percen of GDP. Furhermore, i se he one for an ambiious pension reform agenda. he 995 Dini reform had as is primary objecive reducing he disorions in he labor marke and improving fairness. In effec, i marked a srucural change in he Ialian pension sysem, beginning he process of a gradual ransiion from a defined-

16 4 benefi o a defined conribuion scheme, which will be compleed by he reform insiued noional individual social securiy accouns while making labor marke decisions more flexible. For insance, while he reform direced pension amouns o be calculaed based on he conribuion over he enire working life, i also reduced he reiremen age and he minimum number of years in he scheme needed o be eligible o draw a pension. While he reform fundamenally ransformed he srucure of he pension sysem in Ialy, i had lile immediae impac on pension spending, as he replacemen raes changed only marginally for older workers. However, hese raes will drop significanly for hose reiring afer 202. he Maroni-remoni reform adoped in 2004 furher raised he reiremen age for cerain ypes of pensions, in a sequence of seps envisioned o begin in In addiion, i provided for he creaion of a supplemenary privae secor pillar, also due o come ino effec in Oher hings equal, he increase in he reiremen age would conribue significanly o fiscal susainabiliy by lowering pension spending by up o ¾ of percenage poin of GDP in a he ime of he expeced hump of age-relaed spending in Ialy. However, proposals have recenly been made o modify some aspecs of his reform before i comes ino effec ha could weaken is fiscal impac. Sweden 27. he prereform Swedish pension sysem was a defined-benefi one. However, wih he impac of fuure demographic rends, a slowdown in growh during he 980s, and he recession in he early 990s, he sysem was recognized as being unsusainable in he long run. Legislaion for a new pension sysem was enaced in he key objecives in designing he new pension sysem were o ensure is longrun susainabiliy and o mainain social jusice, even in he face of demographic shocks. As a resul, a so-called noional defined-conribuion scheme was inroduced. I has four main feaures. Firs, he pension is calculaed based on conribuions over he enire lifeime. Second, pensions are indexed o wages. hird, annuiy paymens are adjused o changes in life expecancy, so ha an increase in life expecancy implies eiher a decrease in pension amouns or a laer reiremen age. As a resul of hese changes, he sysem is also more flexible on reiremen age, lowering i from 65 o 6, and allowing reired persons o accumulae pensionable income if hey coninue working pas reiremen. And fourh, he reform inroduced a supplemenal pension o op up he earnings-based pension for low income individuals. 29. According o he laes EU projecions, pension spending in Sweden will no rise much as a fracion of GDP by However, i is imporan o noe ha susainabiliy in 3 Currenly, however, he pension sysem coninues o be mosly defined benefis.

17 5 his case is ensured by auomaic adjusmens o pension benefis, and hus a risk remains ha hese benefis may decrease beyond o wha is poliically accepable. D. he Model and Simulaions he model is designed o capure he main feaures of he Cyprio sysem and examine he macroeconomic effecs of pension reform. As discussed below, he model akes as given world ineres rae and demographic developmens. Model overview and demographic ransiion 30. he macroeconomic effecs of he demographic shock in Cyprus are examined using an overlapping-generaions model in he Auerbach-Kolikoff radiion. 4 In his ype of model, he economy is populaed by overlapping generaions of households, aomisic firms, and he governmen. Households consume and accumulae asses during heir lifeime, work during heir youh and middle age, and reire when old. Firms produce a single good using labor and capial, and he governmen collecs income, consumpion, and payroll axes o finance governmen expendiures and pension benefis, and redeem he iniial governmen deb. Alhough he general equilibrium srucure is sandard, he model incorporaes specific feaures of he Cyprio pension sysem. Specifically, as discussed above, i includes separae old-age pension regimes for privae and public secor employees. he former reflecs he poins based sysem and indexaion mechanisms of he basic and supplemenary pension benefis; he laer accouns for he pension benefi based on he las salary and he oneime lump-sum paymen a reiremen. he model also includes labor-augmening produciviy growh. Deails of he formal model and is calibraion can be found in Appendix I. 3. he demographic shock and he pah of he world ineres rae are criical exogenous elemens in he simulaions. he ime line in he model corresponds o 80% Cyprus: Dependency aio in Model (In percen) a 300-year period, wih he middle cenury 70% ( ) covering he demographic 60% ransiion from a high o a low feriliy 50% rae. In he firs cenury, he growh rae of 40% new enrans o he labor force was se a 30% 0.85 percen, which is he average 20% populaion growh during % During he demographic ransiion, Source: Saff calculaions. however, he labor force growh rae varies o replicae he dependency raio defined as he raio of he populaion aged o 4 See Auerbach and Kolikoff (987). For a more recen survey of his lieraure, see Kolikoff (2000); numerical mehods o solve hese models are described in Heer and Maußner (2004) and Judd (999).

18 6 he populaion over 63 in Eurosa s baseline populaion projecions for he period (ex figure). In oher words, he dependency raio in he model peaks a over wice is curren rae oward he end of he middle cenury (2050). he final cenury sees labor force enrans growing a a rae of 0.5 percen per annum, and he dependency raio falling back bu remaining higher han i is currenly. As noed above, wo ineres rae pahs are considered in he simulaions discussed below. he firs keeps he real ineres rae unchanged. In he conex of global aging, however, his scenario is unlikely. As he world ages a differen raes and saring from differing saring poins global savings are envisaged o increase. 5 Increased savings will depress real ineres raes for he nex few decades. Börsch-Supan, Ludwig and Winer (2005) esimae ha under alernaive regional pension reform scenarios real ineres raes will decline by abou 00 basis poins by In his sudy, a more conservaive 50-basis-poin reducion in real ineres raes is considered. 32. For he discussion of he resuls below, i is imporan o undersand household s behavior in his framework: wo ses of condiions solve he household s objecive maximizing lifeime uiliy under sandard dynamic opimizaion echniques. he firs se refers o household s consumpion-leisure decision in a specific year (inraemporal firs-order condiions). In each period, he household equaes he marginal uiliy of consumpion (scaled by wages) made possible by increasing ime devoed o work o he marginal uiliy of leisure. he second se governs he household s consumpion-savings decisions over ime (ineremporal firs-order condiions, or Euler equaion). 6 In his case, households equae he marginal uiliy of curren consumpion o he marginal uiliy of fuure consumpion (scaled by he rae of reurn on savings). Each se of equaions reflecs wheher a household is in he labor force or no, and he peculiariies of he pension rule. he pension rule inroduces wo subperiods in a household s life: working years and reiremen. Privae secor households, in he firs subperiod, comply wih he inraemporal (consumpion-savings decision) and ineremporal (labor-leisure decision) condiions. he inraemporal condiion reflecs he fac ha wage earnings provide addiional uiliy during reiremen because of heir effec on he pension benefi. When he household reires, here is no labor supply choice, by definiion, and only he consumpion-savings decision remains. For public secor households, however, here is one difference: in he firs subperiod, he inraemporal (consumpion-savings) condiion does no reflec he fac ha wage 5 he increase in savings follows even hough, a differen sages of aging, a counry or geographical region will iniially expor capial and accumulae foreign asses as i saves for reiremen; suequenly i will draw down hese asses in reiremen. 6 When households reire, hey face only an ineremporal condiion, as hey no longer supply labor.

19 7 Simulaions earnings accrued in his subperiod provide addiional uiliy during reiremen. 7 his is because he pension benefi is based on he salary in he year before reiremen. 33. As noed above, hree scenarios (able 3) are used o assess he medium- and long-run consequences of aging on he economy. Noe ha he deerioraion in public finances caused by demographic ransiion sems from wo sources: an increase in pension expendiure as he dependency raio increases, and he impac of aging on healh care expendiures. 8 he simulaion resuls are as follows. he baseline 34. he macroeconomic consequences of an unreformed sysem are severe (Figure 2). he simulaions sugges ha pension expendiures will increase by beween 8 and 0 percen of GDP by 2050, regardless of he real ineres rae scenario. he economy suffers as his increase in expendiures would require sharp increases in consumpion axes so ha public deb remains susainable, ha is, a consan deb (as a percen of GDP) and compliance wih he ineremporal budge consrain. Moreover, when ineres raes decline he adverse macroeconomic oucome is worsened by a negaive wealh effec associaed wih he lower reurn on savings. 35. Consider firs he simulaion resuls assuming a consan real ineres rae: A is peak (abou 2050), he consumpion ax rae would increase by abou 0 percenage poins o finance old-age expendiures. Unil 2050, oupu growh and GDP per capia would suffer, even hough consumpion per capia would remain robus. As he share of he populaion in he work force declines, aggregae hours of work per capia decline sharply by Since he capial-labor raio is pinned down by he (consan) ineres rae, he capial sock will fall. As forward-looking households live longer, hey will save more o finance a longer reiremen. his improves he curren accoun balance (CAB) during he demographic ransiion; 9 large surpluses will urn Cyprus ino a ne credior counry. 7 Formally, he fac ha he pension benefi for public employees is based on he final wage implies ha labor effor should jump sharply he year before reiremen. However, he model resrics labor effor o be independen of is impac on pension benefis, as public employees are no compensaed by increasing heir work hours. 8 Noe ha a hird source of sress emerges in he variable ineres rae scenarios: changes in facor prices. 9 his mirrors he accumulaion of asses in counries wih aging populaions (Börsch-Supan, Ludwig, and Winer, 2005).

20 8 Privae secor employees see heir pension benefi increase because he decline in he basic pension benefi is more han offse by he increase in he supplemenary pension. he former declines as households exer less work effor during he iniial years in he workforce and hus acquire fewer basic poins. Sill, hey work harder laer in heir work life as heir work skills improve and receive greaer rewards for heir effors (wages); hus hey acquire more supplemenary poins. Public secor households work effor increases hroughou heir work life o save more for a longer reiremen, bu hese households are consrained no o increase heir work effors a he end of heir work life (ex figures). 20 he presen value of public-versus-privae pension benefis (relaive generosiy) remains roughly unchanged, favoring he former by abou 80 percen. Individual Labor Allocaions in Iniial and Final Seady Saes 0.45 Individuals in he Privae Secor 0.38 Individuals in he Public Secor Iniial Final Iniial Final Source: Saff esimaes. 36. Consider he resuls assuming a variable ineres rae: Alhough macroeconomic developmens are qualiaively similar o hose discussed above, heir effecs are more severe in his case. Forward-looking households will need o save more han before o finance longer reiremens because he reurn on savings falls wih he world ineres rae. his negaive wealh effec would lower aggregae consumpion and savings. he CAB worsens, and Cyprus would remain a ne debor counry. As he ax base declines and old-age pension and healh expendiure remain roughly unchanged, higher ax raes will be needed o finance hese expendiures As noed above, an unconsrained public secor household would maximize uiliy by reducing work effor in he firs 39 years on he job and sharply increasing is work effor in he final year, when wage earnings have he exra benefi of increasing pension benefis for he remainder of is life. 2 his does no depend on he assumpion ha consumpion axes will be adjused o finance old-age expendiure. Alernaive assumpions of adjusing income or payroll axes would also imply higher raes, as heir corresponding ax bases income and work hours also decline. Moreover, using eiher of hese axes creaes more perverse macroeconomic effecs. his is because he burden of hese axes falls disproporionaely on working-age generaions.

21 9 he relaive generosiy of public-versus-privae pension ils more oward he public secor as he addiional work effor of public secor employees booss he monhly benefi and he lump sum paymen. he Convergence Program 37. he Convergence Program (CP) envisages, over he period , a gradual increase in he normal pension age o 65 years of age from 63 years. I also envisages increasing social conribuions (payroll axes) o finance increased age-relaed spending. Compared wih he baseline, he reform package generaes a reducion of abou 3 percenage poins in old-age spending a he peak of he demographic shock. In oher words, he simulaions suggess ha pension expendiures will increase beween 5 and 7 percen of GDP by 2050; as above, he real ineres rae scenario does no affec old-age expendiures. Noe ha he higher labor axes in his reform package ranslae ino lower consumpion ax raes hrough Consider he CP simulaion resuls wih a consan real ineres rae assumpion (Figure 3). Alhough he macroeconomic resuls under he CP are beer han hose under he baseline scenario, quaniaively he differences are no suanial. By 2050, GDP, hours worked, and capial sock (all in per capia erms) are abou 5 percen higher. Beween now and 2030, GDP growh and GDP per capia would suffer, bu less so han before: consumpion per capia would increase due o an easing of he ax burden. Aferwards, consumpion per capia declines due o lower ne wages. he CAB and exernal deb would remain broadly as in he baseline scenario; Cyprus becomes a ne credior counry. I is ineresing o noe he pah of he consumpion ax raes. On impac, hese decline because he ax collecion from payroll axes increases. Also, because of he increases in payroll ax rae, consumpion ax raes would only increase, a heir peak (abou 2050), by roughly 2 percenage poins (abou 8 percenage poins less han in he baseline). In he long run, consumpion ax raes would increase by abou 2 percenage poins while payroll axes would remain 5 percenage poins higher. he reform package has a differen impac on privae and public secor labor supply. Privae secor households face wo opposing effecs. he increase in he reiremen age means ha hey need o save less o finance a shorer reiremen period, and hus households lower heir work effor. Noneheless, he increase in he work life adds wo more years of wages o be considered under he poins sysem, implying ha he average poins under he basic pension rises. Moreover, he increase in labor axes adversely affecs he incenive o supply labor. hus, when he reform is implemened, he basic pension benefi increases bu he supplemenary pension benefi decreases. In conras, public secor workers do no see any increase in heir basic pension, and hus heir labor effor declines, resuling in lower savings, pension benefis and lump-sum paymens.

22 20 Similarly, hese reforms have a differenial impac on he presen value of pension benefis for privae and public secor employees. he presen value of he pension income (a he ime of reiremen) for privae households declines as he reiremen period shrinks. However, he presen value of pension income of privae secor households increases afer 2050 because, in conras wih public secor households, hey adjus heir labor effor and hus receive a higher basic pension. hus, he relaive generosiy of public-versus-privae pension benefis declines by roughly 5 percenage poins. he reform reduces welfare for boh privae and public secor employees. his is because of he higher labor axes needed o finance pension benefis and he reducion in pension benefis. All fuure generaions will be hur unambiguously, bu he welfare of some younger generaions will remain unchanged. 39. Consider he CP resuls assuming a variable ineres rae (Figure 4): he macroeconomic developmens under his assumpion are qualiaively similar o hose discussed above, bu he effecs, while slighly worse han above, are sill less severe han in he baseline scenario. he lower ineres rae would reduce households wealh, and, hence, consumpion and savings decline. As a consequence, he ax base narrows and higher consupion ax raes are needed o finance he old-age pension and healh expendiures. Moreover, ax raes remain higher in he long run. he CAB worsens, and Cyprus remains a ne debor counry. Oher resuls broadly reproduce hose of he consan ineres rae case, albei wih more severe effecs. Addiional parameric reforms 40. Furher reforms are needed o caegorically address he demographic shock and offse he increasing relaive generosiy of public-versus-privae pensions. In paricular, his scenario envisages: (i) following he increase in he reiremen age in he CP, furher increases in he reiremen age: one year per decade beginning in 2028 unil he reiremen age reaches 73 in 2098; (ii) swiching he indexaion so ha all pension benefis o prices; and (iii) reducing he lump-sum paymen by 4 monhs every five years unil i declines o 2 monhs in he increases in he reiremen age are designed o offse he increase in he share of he reiremen period in life expecancy ha emerges in he CP reform scenario, and o gradually lower he reiremen period (as a share of life expecancy) o ha prevailing in 2005 (ex able). he oher wo reforms are o reverse he rend increase in he relaive generosiy of public-versus-privae pensions. Noe ha payroll axes are held consan.

23 2 Cyprus: Life Expecancy and eiremen Period (In years, unless oherwise indicaed) Life expecancy eiremen age / eiremen Period 2/ Memo iem (In percen): eiremen as a share of life Source: World Bank, World Developmen Indicaors ; and Minisry of Finance. / he reiremen age was lowered o 63 years in 99. 2/ Compued as he difference beween he life expecancy and he reiremen age plus one. 4. he addiional parameric reforms safeguard he sysem from declining world ineres raes and enhance he fairness of he pension benefis. Alhough he increase in overall pension spending by 2050 is abou 2 percenage poins of GDP, pension spending will decline below curren levels in he long run (Figures 5 and 6). he use of less disorionary consumpion ax increases and he more limied need o increase hese raes, underlies he suanially beer macroeconomic resuls: per capia hours worked, and oupu, consumpion, and capial sock are abou 0 percen higher. In he long run, he resuling increase in he ax base leads o a consumpion ax rae ha is abou 3 percenage poins lower han in he CP, and hus abou 5 percenage poins lower han i is currenly. eflecing he higher levels of consumpion, he CAB is slighly higher han in he CP. 42. In conras o he CP, hese addiional parameric reforms increase labor supply for privae and public households. he furher increases in he reiremen age imply ha privae households need o save less o finance a shorer reiremen period, and hus hey lower heir work effor. However, labor supply increases in he aggregae as he number of cohors in he labor marke rises. As in he CP, he increase in he work life adds wages ha are considered in he poins sysem, implying ha he average poins for he basic pension rise. For public secor employees, he addiional reforms reduce he labor effor in each period and hus lower heir pension income: boh he monhly benefi and he lump-sum paymen fall. 43. Alhough wih addiional parameric reforms pensions become less generous for boh privae and public secors, he mismach beween he wo sysems narrows. he mismach beween privae and public pension generosiy declines as he reiremen age increases, bu he decline in public pension generosiy is greaer because of he swich o price indexaion and he fall in he lump-sum paymen. he raio of public o privae generosiy falls o abou. (.3) in he consan (variable) ineres rae scenario. In any even, welfare for boh households will improve because he ax rae needed o keep public deb consan will decline. Privae households experience higher welfare in he long run and he losses o younger generaions are far smaller han in he CP. Public households would also experience higher welfare in he long run, bu younger generaions would experience some losses as hey shoulder a large share of he reform burden.