Net Pension Liabilities, Intergener Title and Pension Reforms.

Transcription

1 Net Penson Labltes, Intergener Ttle and Penson Reforms Author(s) Osho, Takash Ctaton Issue Date Type Techncal Report Text Verson publsher URL Rght Htotsubash Unversty Repostory

2 Net Penson Labltes, Intergeneratonal Equty, and Penson Reforms December 2002 Takash Osho Tokyo Gakuge Unversty Abstract Ths paper examnes an deal penson system that would prevent an ncrease n net penson labltes and redress ntergeneratonal nequaltes. The drecton of penson system reform that ths paper proposes s to lmt the amount of penson benefts to the sum of premums, state contrbutons, and yelds on nvestment of reserves, and also to mnmze the burden on the nsured of premum payments. Ths reform would have roughly the same effect as a swtchover to a funded system. It wll be, however, no easy task to carry out ths knd of reform, because the reform requres a poltcally unacceptable step: to request those generatons already recevng pensons and those who have pad penson premums to accept reductons n the amount of penson that the government has pledged to pay. One possble means to deal wth ths problem s to wthdraw the reserve funds, whch are now over 00 trllon yen. However, wthdrawal of the reserves would reduce the opportunty of future generatons to use the funds and yelds on nvestment of the funds. Any attempt to reform the penson system wll cause conflcts of nterests between generatons. Ths s why the government should postvely dsclose the outlook for ts penson fnance, manly net penson labltes, and nformaton about ntergeneratonal dspartes to provde the publc wth data for objectve dscussons on a more desrable form of publc penson. JEL Classfcaton: H55, H3. Key words: net penson labltes, ntergeneratonal equty.

3 . Introducton Publc pensons n Japan have a number of problems, and the followng three are the most mportant. Frst, net penson labltes have reached a rather hgh level. Penson labltes are the amount of penson benefts pledged to be pad to the nsured of a penson system or to the recpents of pensons accordng to ther past premum payment performance (benefts cost correspondng to the past perod). Net penson labltes are defned as penson labltes less those supported by avalable penson funds. It s often sad that publc pensons barely exst by shftng burdens onto future generatons, and these burdens are net penson labltes. Accordng to the "Government Balance Sheet" for 2000 publshed by the Mnstry of Fnance, the net penson labltes of the Employees' Penson Insurance reached 552 trllon yen (about 0% of GDP) at the end of If these labltes contnue ncreasng, fears about the sustanablty of the penson system wll be ntensfed. Second, ntergeneratonal nequaltes are ncreasng, whch s related to the frst problem referred to above. Even after the 2000 reform, there reman great dspartes between generatons as to the benefts and the burdens of the Employees' Penson Insurance. For example, an average couple of the generaton born n the 990s would pay about % of ther lfetme wages, whch s more than the amount of penson benefts they would receve (based on the premums pad by the couple and ther employers; state contrbutons excluded). On the other hand, those of the generaton born n the 950s would pay only about 6% of ther lfetme wages and would receve benefts that are greater than the premums they would pay. The greater the net penson labltes are, the heaver are the burdens future generatons would have to bear. Fnally, the future of the penson systems s uncertan. Penson reforms n the past were repettons of rasng premum rate and makng penson payment condtons strcter as a result of past brth rate estmates beng found to be too optmstc. Ths means that the government contnued to change ts promse to the people, so t wll be natural that ther worres about the future of the penson system wll not be allevated however the government may explan ts polces to them. In fact, the new populaton estmate, 2

4 publshed n late January 2002 by the Natonal Insttute of Populaton and Socal Securty Research, revsed down ts total fertlty rate forecast for 2050 from.6 to.39. If ths forecast s correct, a substantal ncrease n the premum rate wll be nevtable, whch wll further deepen publc anxetes concernng penson systems. Wth the three questons above n mnd, ths paper dscusses the desrable drecton of penson reforms. In other words, t examnes the knd of publc pensons that could attan the three objectves: () avod a further ncrease n net penson labltes; (2) correct ntergeneratonal nequaltes; and, (3) enhance the transparency of the penson system. Secton 2 summarzes the concepts of net penson labltes, and Secton 3 dscusses three basc methods for avodng an ncrease n labltes (fundng, full pay-as-you-go, and balanced ncome and expendture methods). Secton 4 compares the theoretcal characterstcs of the three methods, and Secton 5 evaluates the exstng systems, especally from the vewpont of the dvergence of net penson labltes. Fnally, Secton 6 proposes several reform plans on the bass of the dscussons n prevous sectons, and Secton 7 gves a summary of all of the arguments. 2. Concepts of net penson labltes 2. Formulaton of net penson labltes Ths secton summarzes problems related to the handlng of net penson labltes usng very smplfed overlappng-generatons models for two generatons and two perods. Let us suppose that generaton lves n two perods, the workng and retrement perods, and generaton + s stll n the workng perod when generaton retres. Generaton pays premums n the amount of P, and receves penson benefts n the amount of B on retrement. The perod when generaton s workng s defned as perod. Also, suppose the government owns reserves amountng to A 0 at the end of perod 0, the perod when generaton 0 s workng. If the government pledged to pay generaton 0 penson benefts amountng to B 0 startng from perod, ts net penson labltes at the end of perod 0, N 0, can be expressed by 3

5 N B0 = 0, () + r 0 A where r s the nterest rate, whch s supposed to be fxed for each perod below to smplfy the dscussons. Although the government promsed generaton 0, who had pad premums, to pay pensons amountng to B 0, the government now has reserves n hand only, and has to procure the amount of the shortfall by some means. Smlarly, the government's net penson labltes at the end of perod, expressed as N and estmated by the dscounted value at the end of perod 0, are gven by N B A =. (2) 2 + r ( + r) The relaton between the reserves at the end of perod 0 and those at the end of perod s expressed as A ( + r) A + P. 0 B0 = (3) In addton, by substtutng () and (2) for (3) we get N B0 = + + r B ( + r) P + r A 2 0 That s, the government pays generaton 0 the pledged amount of penson B 0 n perod and at the same tme promses generaton anew to provde benefts n the amount of B, whereas the penson funds avalable for such payments are the premums collected from generaton n perod plus the reserves already exstng at the end of perod 0. By the same token, the value of the net penson labltes at the end of perod s gven by N t t ( + r) t= ( + r). Bt Pt = A0. (4) + t= 0 As noted, the porton of the sum of penson benefts the government pays by the end of perod and penson benefts t pledges to pay n perod + (frst term, rght sde), whch exceeds the sum of total premums collected by the government by the end of the same perod (second term) and the ntal reserves (thrd term), represents the net penson labltes the government has n perod. 4

6 2.2 Net penson labltes and generatonal accounts Now let us take a look at the relaton between net penson labltes and generatonal accounts. The net penson labltes at the end of perod should be jontly shouldered by generaton +, whch s n the workng perod n perod + and by generaton + 2 and subsequent generatons. Therefore, the assumed generatonal account equaton can be gven by N = P + B + + = + t t+ ( + r) ( + r) t ( + r) ( + r) P B + t+ + t+ 2 where the frst term at the rght sde shows the net burdens of generaton + and the second term, those of generaton + 2 and after. If we call generaton + the "actve generaton" at the end of perod, and generaton + 2 and after "future generatons," the above equaton suggests that the actve and future generatons must share the cost of net penson labltes. In ths sense, the economc nterest of the actve and future generatons has a trade-off relaton. The net penson labltes at the end of perod + are shown by B. 2 2 ( ) ( ) + P + P + t+ B + t+ N + = N + = t+ + t+ + r + r t= ( + r) ( + r) Ths equaton shows that the porton of the net penson labltes that have not been borne by the actve generaton, that s, the porton whose lqudaton has been shfted to the responsblty of future generatons, becomes the net penson labltes of the next perod. In other words, as long as the government contnues to promse each generaton the provson of penson benefts n an amount that s greater than the premum payments of the generaton, net penson labltes wll contnue ncreasng. The above dscusson shows that an ncrease n net penson labltes s closely related to an ncrease n ntergeneratonal nequaltes n publc pensons. Avodng the dvergence of net penson labltes s bascally the same as preventng an ncrease n ntergeneratonal dspartes n generatonal accounts., 3. Methods of avodng the dvergence of net penson labltes 5

7 One of the problems the government should solve s how to avod the dvergence of net penson labltes, whch s expressed by (4). As s evdent from ths equaton, net penson labltes dverge because the exstng penson system has, n the long run, a structure by whch the government contnues to pay nsured amounts of benefts that are greater than the premums t collects from them (plus the porton of ts ntal reserve funds wthdrawn). Therefore, what s needed s to change the exstng structure. Several methods are avalable, but the followng three would be the most basc strateges: fundng, full pay-as-you-go, and balanced ncome and expendture methods. 3. Fundng methods The frst opton s to ntroduce a fundng method to the penson system or shft to a funded scheme -- for generaton and after 2. In ths case, the premums and the penson benefts have the same dscounted present value for each generaton, that s, B = P, =, 2, L. + r In addton, (4) can be changed nto N = N 0 + t= ( + ) t r Bt Pt + r and the [ ] n the second term of the rght sde of ths equaton s zero n the fundng method. Hence, n ths case, net penson labltes are fxed at the value at the end of perod 0 n terms of dscounted present value, that s, N = N 0. Also, the dscounted present value of reserves at the end of perod (assessed at the end of perod 0), A, s expressed as A A P B = = N = N. (5) + 0 ( + r) ( + r) ( + r) The reserves the government holds at the end of perod are equal to the amount obtaned by deductng the ntal net penson labltes from the premums collected from generaton as funds for penson benefts to be provded to the generaton after retrement. In the next perod, the government has to spend these premums as benefts for retred generaton, 0 6

8 whle t receves premums from the actve generaton, whch wll provde penson funds for the generaton after the next. And, unless the amount of premums (or amount of penson benefts) ncreases at a faster pace than nterest rates, the reserves wll gradually decrease n terms of dscounted present value. 3.2 Full pay-as-you-go method The second opton s to change the penson system nto a full pay-as-you-go one n perod and after. The term 'full' s used here because the current pay-as-you-go method may have promsed to provde benefts that are greater than those whch can be supported by premums (plus state contrbutons), nstead of adoptng a textbook-lke system where penson ncome and expendture are balanced at a pont n tme 3. The pay-as-you-go method, n whch premums and benefts agree wth each other n each perod, can be expressed as: B = P, =, 2, L Snce (4) can be rewrtten as N = B A + t Pt, + 0 t ( + r) t= ( + r) B the thrd term of the rght sde becomes zero. Therefore, the net penson labltes n the full pay-as-you-go method are shown by: N B = A. (6) + ( + r) 0 Ths equaton contrasts wth (5) for the fundng method. In the full pay-as-you-go method, all the government must pay attenton to s the amount of benefts t pledges to provde n the next perod, because premums and benefts agree wth each other n each perod. The government needs to procure the amount of these benefts less ts reserves, but can collect premums equal to benefts from the actve generaton n the next perod and does not have to wthdraw ts reserve funds. Instead, t has to worry about the benefts t wll provde n the perod after the next. Thus, unless the benefts ncrease at a faster pace than nterest rates, the dscounted present value of net penson labltes wll not dverge and wll rather declne lttle by lttle. 7

9 In case publc pensons are managed by the full pay-as-you-go method, net penson labltes can be calculated, but are no longer mportant n the government polcy. In fact, there seem to be no cases where other countres record penson labltes as government debts when preparng ther government balance sheets. In addton, f the government swtches over to a penson system that would automatcally adjust the amount of benefts wthn the lmts of premum ncome 4, t wll dstrbute only the premum ncome to the elderly as benefts (but the premum payments of the respectve recpents wll be reflected on the allotment of benefts), so the concept of penson labltes wll tself become meanngless. Fnally, let us pay attenton to the fact that because the amount of premums and that of benefts agree wth each other n each perod, the dscounted present value of reserves, A, s fxed as A = A Balanced ncome and expendture method The thrd opton s to always make the amount of benefts agree wth the sum of premums and yelds on nvestment of reserves. Ths opton s referred to as the "balanced ncome and expendture method" below. In ths method, the followng equaton apples to each perod: B, = P + ra =, 2, L. Here, because the reserves are fxed at ther ntal value 5, we get A0 A = A0, A =, =, ( + r) 2, L whch means that the dscounted present value of the reserves gradually decreases. And snce the net penson labltes are expressed as N B = + A 0 ( ) ( ), + r + r net penson labltes wll not dverge unless the amount of benefts ncreases at a faster pace than nterest rates. Ths method wll also obvate the need to pay attenton to net penson labltes n the polcy-makng process. 8

10 Ths balanced ncome and expendture method may be regarded as a method for wthdrawng reserve funds to apply them to part of the penson funds. In other words, f only ra - /( + r) of the reserves avalable at the begnnng of perod s wthdrawn and operated, and the prncpal thus obtaned s used as penson funds, ths wll be the same as the balanced ncome and expendture method. In ths method, too, f there s a mechansm for automatcally adjustng benefts wthn the lmts of premums and yelds on the nvestment of reserves, the concept of penson labltes wll become substantally meanngless. 4. Evaluaton of the three methods In ths secton, let us compare the three methods explaned n the prevous secton from four standponts: () redempton of net penson labltes, (2) possblty of "clam renouncement," (3) rate of return of pensons, and (4) change over tme of net penson labltes and reserves. 4. Redempton of net penson labltes The frst comparator s the redempton of net penson labltes, whch poses the most mportant problem n the case of a swtchover to the fundng method. In ths case, the porton of penson benefts promsed to generaton 0 that cannot be covered by reserves wll manfest tself as net penson labltes to be redeemed. Ths s often referred to as the problem of so-called "double burdens." As means to redeem these net penson labltes, two methods have been studed so far: (a) to make generaton and subsequent generatons to shoulder the labltes as addtonal premums and/or taxes, and (b) to cover the labltes by ssung government bonds, redeemng those bonds and payng nterest on them n subsequent years. However, both methods cannot avod layng addtonal burdens on generaton and subsequent generatons, and the sze of the burdens agrees wth the amount of the already exstng net penson labltes 6. By contrast, wth the full pay-as-you-go and balanced ncome and expendture 9

11 methods, there s lttle need to pay any partcular attenton to net penson labltes. Ths s because the benefts pledged to each generaton wll be fully provded when they retre. In case the amount of benefts s automatcally adjusted wthn the lmts of premums and state contrbutons, t s possble to consder that no penson labltes arse themselves. Two thngs should be mentoned here. Frst, even f the government swtches over to the fundng method, t does not always need to redeem ts net penson labltes mmedately. It may nstead make only avodng an ncrease of labltes ts polcy target for the tme beng. Ths resembles the strategy of attanng the prmary balance frst and avodng rses n the dscounted present value of government debts, nstead of amng at reducng such debts at once. Second, net penson labltes are now off-balance sheet labltes not lsted on the government balance sheet, and there s a problem of whch of two methods should be adopted: explctly redeem the labltes as on-balance sheet ones by recordng them on the balance sheet, or mplctly redeem them as off-balance sheet ones. In case the penson system s changed to the fundng method, there wll arse the need to turn exstng net penson labltes nto on-balance sheet ones. In such a case, the on-balance sheet labltes should be lsted as natonal debt, whch wll pose another problem: whether the natonal debt should be made marketable or not. In the case of the full pay-as-you-go and balanced ncome and expendture methods, there wll be no need to make net penson labltes on-balance sheet ones. 4.2 Possblty of "clam renuncaton" In carryng out a penson reform program, an mportant problem s whether the government should fully provde the penson benefts t promsed to generaton 0. We can consder the method by whch the government cancels part of ts pledge and asks generaton 0 to gve up some porton of the benefts t has promsed to provde to them (ths means partal "clam renuncaton" to the generaton). If generaton 0 s gven favorable treatment n terms of penson rate of return under the current system, ths strategy mght be acceptable at least from the vewpont of ntergeneratonal equty. In the case of the fundng method, however, clam renuncaton by generaton 0 s 0

12 very dffcult. In case the fundng method s ntroduced n perod and after, f generaton and subsequent generatons pay premums only to secure ther own penson benefts, funds for benefts for generaton 0 must be rased from sources other than premums. Conversely, f the government wants to lmt clam renuncaton by generaton 0 to a small scale, t has to ask generaton and subsequent generatons to pay addtonal premums. In the case of the full pay-as-you-go and balanced ncome and expendture methods, however, partal clam renuncaton by generaton 0 becomes more realstc. Wth these methods, t s possble to adopt the management strategy of automatcally adjustng the amount of benefts by premum ncome from the actve generaton (plus yelds on the nvestment of reserves n the case of the balanced ncome and expendture method) n perod and after. In ths case, we can employ the method of determnng the amount of benefts for generaton 0 by ths strategy, too, askng the generaton to gve up only the defcences. In case adequate premum ncome from the actve generaton can stll be expected to some extent or n case there s a substantal yeld on nvestments of reserves, clam renuncaton, whch ths strategy forces generaton 0 to accept, can be controlled to a smaller amount. How then can the sze of clam renuncaton be expressed? Let us suppose penson management where penson benefts at each perod are automatcally adjusted to an amount wthn the lmts of premum ncome for the full pay-as-you-go method, and wthn the lmts of premum ncome plus yelds on nvestments of reserves for the balanced ncome and expendture method. In each case, the sze of the clam renuncaton by generaton 0 can be shown as follows: B0 P The full pay-as-you-go method: + r + r B0 P ra 0 The balanced ncome and expendture method: +. + r + r + r Needless to say, to carry out clam renuncaton, there s also the need to solve a problem related to property rghts, that s, partal cutback on already determned pensons. Compared to a shft to the fundng method, however, clam renuncaton wll be easer wth a changeover to these two methods.

13 If no clam renuncaton s approved for generaton 0, that porton of benefts wll be added to net penson labltes. If generaton and subsequent generatons pay addtonal premums to make up for the porton added, the ncrease n net penson labltes wll be checked. Also, n case generaton and subsequent generatons do not make up for the porton and the government ssues bonds and contnues to refund exstng bonds, an ncrease n net penson labltes can be avoded n terms of dscounted present value. Another possble alternatve s to wthdraw reserves. In ths case, however, future generatons wll be unable to use yelds on the nvestment of the wthdrawn part of the reserves. 4.3 Rate of return of pensons The thrd comparator s the rate of return of pensons. To make comparson easer, suppose that premums are collected n proporton to wages (the premum rate s fxed). In ths case, the ncrease rate of premum ncome s equal to that of wage ncome (sum of per capta wage hke and populaton growth rates; shown by ). Frst, what s the rate of return at the completon stage of the system? Wth the fundng method, t s evdent that the rate of return equals nterest rate, r, regardless of the ncrease rate of wage ncome. Wth the full pay-as-you-go method, the rate of return equals the ncrease rate of wage ncome. In a stuaton of low brth rates and agng populaton, ths method s lkely to be nferor to the fundng method n terms of rate of return. The rate of return of the balanced ncome and expendture method s not constant, because t s affected by the ncome and expendture of penson fnance n each perod. Suppose the rate of return for generaton s, then we get µ B P + + ra0 ra0 = = = ρ +. P P P Thus, t s evdent that ths method has a hgher rate of return than that of the full pay-as-you-go method. Also, snce r µ r = ρ + ( A0 P ), P the possblty cannot be theoretcally dened that ths method has a hgher rate of return 2

14 than the fundng method, too, especally n the case where the ntal reserves are large enough relatve to premums n each perod, and also the ncrease rate of wage ncome does not show a large negatve value. Ths means that a shft to the balanced ncome and expendture method should preferably be made whle reserves are abundantly avalable. We should note that f the premum rate s lowered and the scale of publc pensons s reduced, the full pay-as-you-go and balanced ncome and expendture methods wll have substantally the same effects as a change to the fundng method. Ths s because people wll be able to operate, at the current nterest rate, part of ther ncome that wll become avalable as a result of the reducton n the scale of publc pensons. Next, what wll be the rate of return durng the transton perod of the penson system? In the case of a shft to the fundng method, the rate wll dffer greatly accordng to how we deal wth net penson labltes. The more generaton and subsequent generatons are caused to bear the labltes, the lower the rate for these generatons wll be. In the case of a changeover to the full pay-as-you-go method or to the balanced ncome and expendture method, f the amount of benefts s automatcally adjusted, the rate of return wll be determned by ether (for the full pay-as-you-go method) or (for the balanced ncome and expendture method) for all generatons. But, ths does not apply f the government has already promsed generaton 0 to provde an amount of benefts n whch the rate of return would exceed and provdes benefts accordng to ts promse. 4.4 Change over tme of net penson labltes and reserves The fnal comparator s how net penson labltes and reserves change over tme. If premums and benefts do not ncrease faster than nterest rates, the labltes and reserves could both avod dverson and wll fnally converge as expressed as Fundng method: N = N, lm A = N, lm ( N + A ) = 0, Full pay-as-you-go method: 0 0 lm N = A, 0 A = A, lm 0 ( N + A ) = 0, Balanced ncome and expendture method: 3

15 lm N = 0, lm A = 0, lm ( N + A ) = 0. In all cases, the labltes and the reserves wll offset each other n the long run, but the process wll be dfferent. Wth the fundng method, net penson labltes are fxed at ther ntal amount and reserves gradually decrease and converge to the value wth a sgn that s opposte to the sgn of the labltes. The labltes are the amount obtaned by deductng the premums receved so far by the government and reserves from the benefts pledged by the government; f benefts and premums come to be roughly balanced n the long run n terms of dscounted present value, net penson labltes and reserves begn to offset each other. In the case of the balanced ncome and expendture method, both net penson labltes and reserves wll ultmately converge on zero. 5. Evaluaton of the current penson systems 5. Outlook for ncome and expendture The "Government Balance Sheet" for 2000 (draft proposal) publshed by the Mnstry of Fnance n September 2002 provdes nformaton on the net penson labltes of the Employees' Penson Insurance (EPI) at the end of Accordng to ths publcaton, the "benefts cost correspondng to the past perod" of the EPI amounts to 695 trllon yen. Of ths amount, 43 trllon yen s reserves, that s, the porton supported by penson funds, whle the remanng 552 trllon yen s composed of EPI s net penson labltes. These labltes (552 trllon yen) can also be dvded nto () state contrbutons (97 trllon yen), whch to be borne by people n the future as taxes, and (2) addtonal premum ncome (455 trllon yen), whch to be covered by future rases n the premum rate. Here state contrbutons are supposed to cover one-thrd of the Basc penson benefts. The problem s whether or not the labltes wll contnue to ncrease and place burdens on future generatons. It s, however, no easy task to forecast the future of net penson labltes, because t requres the calculaton of premums to be pad by each generaton at each perod and of the penson labltes that would arse accordng to the premums to be actually pad. 4

16 Accordng to the "Actuaral Revaluaton" publshed by the Mnstry of Health, Labor and Welfare n 999, the EPI's ncome and expendture s expected to be bascally n the black for 50 to 60 years (see Fgure below). If ths outlook s correct, an ncrease n net penson labltes wll be avoded. Intergeneratonal nequaltes seen from generaton accounts wll not ncrease further. However, the "Actuaral Revaluaton" s based on the former medan populaton projecton publshed n 997, whch showed that the total fertlty rate would recover to.6 by 2050 (from the.33 n 200). The new medan populaton projecton announced n 2002 states that the fertlty rate would rse to only.39 by 2050, whch means that to mantan penson benefts at the current level, substantal rases n the premum rate wll be needed 7. Conversely, ths ndcates that the benefts level establshed by the 2000 reform s unfeasble, consderng the dynamcs of populaton and other factors. Ths fact can also be confrmed by replacng the assumptons of penson fnance used n the 2000 reform wth more realstc ones, and by examnng changes n penson ncome and expendture. Here, the followng three factors should be taken nto consderaton. The frst and most mportant factor s populaton estmates, as mentoned above. Here, we use the new medan estmate of populaton publshed n The second factor s the estmate of yelds on nvestment, whch was set at 4% n the 2000 reform. Ths fgure seems to be a lttle unrealstc consderng the low nterest rates n recent years. Thus, we lower the yelds on nvestment to 2.5%. However, f t s supposed that nterest rates fall due to dsnflaton, the rates of wage and prce hkes wll also drop to % and 0%, respectvely. Thus, lower yelds on nvestment are unlkely to have very adverse effects on penson fnance. The thrd factor s the ntal value of reserves. The reserves recorded n the actuaral revaluaton nclude the "substtutonal part" (Dako Bubun) of the Penson Fund Assocaton, and were 77.2 trllon yen at the end of 999. However, because of the slower growth n premum ncome and the worsenng fund operaton envronment due to recessons, the actual amount of reserves at that tme was only 63 trllon yen (of whch 28.2 trllon yen was substtutonal). The lower the ntal value s, the lower s the expected yeld on nvestment. 5

17 Fgures and 2 shows how the estmate n the 2000 reform are affected f the above three assumptons are changed smultaneously, n terms of the fscal postons and reserves. In the 2000 reform, the fscal balance s expected to reman generally n surplus, although t wll temporarly suffer defcts around 2050 (Fgure ). However, when more realstc assumptons are used, the balance s n the black around the frst peak years of populaton agng, but s n defct before and after these years, and wll consstently reman n the red after the surplus perod. The 2000 reform expected that reserve ratos would reach stablty at about three tmes around 2060, but a more realstc estmate shows that the ratos wll become about zero around that year, reflectng worsenng flow-base ncome and expendture. As stated above, f the assumptons used n the actuaral revaluaton are replaced wth more realstc ones, the reserves wll run out by 2060, whch means that no provson for the government's lablty to provde benefts wll be avalable n that year (Fgure 2) 8. Although no drect estmate of future net penson labltes s made here, the above dscusson clearly suggests that the 2000 reform has very serous problems from the standpont of controllng any ncrease n labltes. 5.2 Level of penson benefts that can be mantaned by the 2000 reform Now, let us change our pont of vew and roughly estmate the level of penson benefts that can be mantaned wthout dffculty by the 2000 reform. The followng two methods are avalable for ths purpose. Frst, we can suppose that benefts for each year are provded wthn the lmts of the premum ncome obtaned by the planned rases n the premum rate, whch are assumed n the 2000 reform, plus state contrbutons (whch are supposed to cover one-thrd of the Basc Penson benefts), and examne to what extent the amount of benefts should be reduced (Case I). Ths means that we assess to what extent problems wll arse f the full pay-as-you-go method, whch supposes the automatc adjustment of benefts, s ntroduced. Second, we can suppose that benefts for each year are provded by yelds on nvestment of reserves n addton to premum ncome and state contrbutons, whle adoptng the plan to ncrease the premum rate scheduled n the 2000 reform (Case II). In 6

18 ths case, we examne to what extent benefts should be adjusted when the balanced ncome and expendture method, whch supposes automatc benefts adjustment, s adopted. Fgure 3 shows the cutback n benefts needed to mantan the exstng penson system n Cases I and II. In Case I, where benefts provson reles only on premum ncome and state contrbutons, the necessary reducton rate of benefts wll be about 0% untl about 2020, and wll decrease a lttle around 2025, when populaton agng s expected to reach the frst peak, but wll rse agan to 20% or so by around As s evdent from ths, f benefts are provded only from premum ncome and state contrbutons as assumed by the exstng system, t s expected at present that a substantal cutback n benefts wll be needed for younger generatons. In Case II, where yelds on nvestment of reserves are also used for the provson of benefts, the reducton n benefts requred s consderably smaller. In partcular, around 2025, t wll be possble to ncrease benefts by about 5%. However, as n Case I, there wll arse the need to slash benefts after that. Cases I and II both suggest that the government wll become unable to pay the amount of benefts t has pledged to the people, manly because the 2000 reform overestmated future brth rates. Ths shows the possblty that net penson labltes wll contnue ncreasng n the long run, whch s consstent wth the dscusson n the prevous secton. If benefts are pad accordng to the assumptons of the 2000 reform and the premum rate s not rased, the burdens of net penson labltes wll be passed onto future generatons. Moreover, f the amount of benefts s cut back, as n Cases I and II, n an effort to avod an ncrease n net penson labltes, ntergeneratonal dspartes are lkely to ncrease further. The 2000 reform supposes that the premum rate wll be rased from 3.58% at present to 2.6% on a total remuneraton bass. If benefts have to be substantally reduced at the same tme, the lfetme penson rate of return wll be substantally lower. 6. Penson reform plans 7

19 6. Basc prncples What knd of penson reform plans then can we consder to avod a further ncrease n net penson labltes and to correct ntergeneratonal nequaltes? Three theoretcal methods are avalable. The frst method s a changeover to the fundng method, or the total or partal abolton of publc pensons, whch would have smlar effects to such a changeover. If the current system s replaced by a funded one, no new net penson labltes wll arse after that. However, the problem s how to deal wth the exstng labltes. For example, f the government plans to swtch over to a full fundng method, t wll have to take mmedate measures to prepare penson funds amountng to about 30 trllon yen a year. It may cope wth ths by wthdrawng ts reserves for some tme, but the amount avalable s far from meetng the needs. One possble strategy s to ask the retred generaton to accept a partal cutback n benefts, but the cutback wll have no meanng unless t s large, because no penson funds are avalable, and t wll be mpossble to make such a large cut n benefts. The second method s a shft to the full pay-as-you-go method, n whch the amount of benefts s automatcally adjusted wthn the lmts of premum ncome and state contrbutons, wthout determnng the amount of benefts n advance. In ths case, a benefts oblgaton arses formally, but can be controlled well by the government because t can always collect premums n the same amount as that of the oblgaton. However, the past premum payment performance of the nsured wll not be fully reflected on the amount of benefts they receve. In ths sense, ths method asks the people to make a partal clam renuncaton. But, as noted n the prevous secton, the adjustment n ths method wll be less than n a changeover to the fundng method. Because the sum of premums and state contrbutons always agrees wth the amount of benefts, reserves wll contnue ncreasng n nomnal value, but wll reman at the ntal level n terms of dscounted present value. The thrd method s a changeover to the balanced ncome and expendture method. As n the second method, the amount of benefts s automatcally adjusted so that t may become equal to the total revenues, that s, the sum of premum ncome and state contrbutons plus yelds on nvestment of reserves. Here, the reducton n benefts s controlled by the amount of yelds on nvestment. In ths case, too, the benefts oblgaton 8

20 wll be fulflled successvely, and net penson labltes wll not ncrease. Reserves wll be fxed at the ntal value n nomnal terms and wll declne n dscounted present value. All three methods have the effect of checkng the dverson of net penson labltes, but also have advantages and dsadvantages. From the vewpont of transtonal measures, the second and thrd methods ask the retred generaton to accept cuts n benefts, but wll be easer to ntroduce than the frst method, whch wll requre the mmedate redempton of huge labltes or the ssuance of government bonds. In terms of penson rate of return, however, younger generatons (who do not face the "double burdens" problem) should prefer the frst method. To make the most of these methods and mnmze ther defects as much as possble, we have no alternatve but to fnd a compromse plan. More specfcally, we can consder the strategy by whch the system s based on ether the full pay-as-you-go method or the balanced ncome and expendture method, where the amount of benefts s automatcally adjusted so that t may equal total revenues, that s, the sum of premums and state contrbutons plus, where approprate, yelds on nvestment of reserves, and the scale of publc pensons s reduced by lowerng the premum rate as much as possble. 6.2 Eght reform plans Now let us examne eght plans to reform the penson system (see Table ). Case A adopts the full pay-as-you-go method wth automatc benefts adjustment, and rases the premum rate gradually up to 20% (3.58% at present on an annual remuneraton bass, whch s equvalent to 7.35% on a monthly remuneraton bass; total of employers' and employees' shares), wth the rate fxed at 20% thereafter. The schedule supposed n the 2000 reform s used for these rases. The premum rate reaches 20% at around 2020, and s fxed at ths level after that. Case B s the same as Case A as to automatc benefts adjustment and the 20% celng on premum rate ncreases, but adopts the balanced ncome and expendture method, whch also uses yelds on nvestment of reserves, nstead of the full pay-as-you-go method. Compared to Case A, the cuts n benefts wll be much smaller. Case C adopts an automatc benefts adjustment and the full pay-as-you-go 9

21 method as n Case A, but fxes the premum rate at the current level of 3.58%. Case D resembles Case C n terms of automatc benefts adjustment and the fxed premum rate of 3.58%, but adopts the balanced ncome and expendture method. All four cases do not provde benefts above the level assumed by the exstng system. In other words, should revenues exceed such a level, the surplus s carred forward to the next year (t s added to the reserves). In addton, all of the cases nclude controls on the premum rate, and so reduce the scale of publc pensons to that extent. In ths sense, these cases have the common feature of shftng to the fundng method, and ths wll be more obvous n Cases C and D than n Cases A and B. The four cases mentoned above do not wthdraw any reserves. In Cases A and B, reserves contnue to yeld nterest, but none of ths nterest s utlzed for benefts provson. Cases C and D do use the nterest on reserves for provdng benefts, but am at permanently holdng reserves (at a fxed level n nomnal terms). Here, let us adopt a mechansm for wthdrawng reserves n each case. More specfcally, take out 3% of the reserves at the end of the prevous year and apply the amount to the provson of benefts. We name these cases correspondng to the above four cases Case A', Case B', Case C', and Case D', respectvely. All of these cases only offer the amount of benefts assumed by the current system and surpluses, f any, are carred forward to the followng year. 6.3 Effects of reform (): reducton rate of benefts and change n reserves Now let us examne the effects of each case. The frst factor for examnaton s cutbacks n benefts (see Fgure 4). Let us begn wth comparng Cases A to D, where no reserves are wthdrawn (see Fgure 4, ()). In Case A, where the celng of premum rate s fxed at 20%, the reducton n benefts s the same as that n Case I untl around 2020, but t grows larger after that and ultmately reaches about 25%. Because the amount of premums s controlled, the benefts to be provded are naturally restraned, too. In Case B, because yelds on nvestments of reserves are avalable, a cutback n benefts s checked to some extent, and fnally rses to about 20%. As already dscussed for Cases I and II, the adjustments untl around 2020 pose a problem. In Case A, the amount of benefts should be reduced from about 0% to nearly 20

22 5%. In Case B, where yelds on nvestment can be utlzed, the reducton rate can be controlled to about 0-5%. Compared to Case II, however, an ncrease n benefts, whch s possble n Case II, s almost unfeasble n Case B, partly because the latter supposes no rases n premums around 2020 and after. What s the stuaton n Cases C and D where the premum rate s fxed at the current rate of 3.58%? Naturally, the amount of benefts supposed by the current system must be cut back substantally. In Case C, whch depends on premum ncome and state contrbutons only for penson funds, the ultmate reducton rate of benefts s about 45%. In Case D, whch also uses yelds on nvestment, the rate ncreases to nearly 40%. For future generatons, tght controls both on premums and benefts mean a shrnkage n the scale of publc pensons. Next let us have a look at Cases A' to D', where reserves are wthdrawn by 3% each year (see Fgure 4, (2)). In Cases A' and B', whch fx the celng of premum rates at 20%, there s no need to cut back benefts untl around Ths shows up the effect of takng out reserves. Those who have already receved or wll soon receve benefts would welcome ths. The need to slash benefts begns to arse around 2040, but the ultmate reducton rate s 20% or so both n Cases A' and B'. In Cases C' and D', where the premum rate s fxed at 3.58%, the present level of benefts can be kept by takng out reserves only untl around 200. Ths suggests that the procurement of penson funds wll become dffcult as a result of restrants on premum ncome. Ths wll be a more serous problem wth a shft to the fundng method. The second factor for consderaton s the outlook for reserves (n dscounted present values). Fgure 5 shows that the outlook dffers greatly accordng to the reform plan. As shown n Fgure 3 above, the 2000 reform wll dran reserves by around 2060 f the assumptons are replaced wth more realstc ones. However, n Cases A and C, whch ntroduce the full pay-as-you-go method, ncome and expendture always equal yelds on nvestment of reserves because the amount of benefts always agrees wth the sum of premums and state contrbutons. Reserves contnue ncreasng n nomnal terms, but are fxed at the ntal level n terms of dscounted present value. In Cases B and D, where the balanced ncome and expendture method s adopted, yelds on nvestment are added to 2

23 benefts provson, so reserves are fxed at the ntal value n nomnal terms, but gradually decrease n dscounted present value. As of 2060, there wll reman about 40 trllon yen of reserves (n dscounted present value), or about 23% of the 999 level. By contrast, n Cases A' to D' where reserves are taken out, reserves naturally decrease more quckly. However, as n Cases B and D, the pace of declne s stable n contrast wth the non-convergent decrease expected under the current system as shown n Fgure Effects of reform (2): effects by generaton Next, let us examne what effects the eght reform plans would have on each generaton. Here, we defne the amount obtaned by deductng lfetme premums pad from lfetme benefts receved, whch s expressed as a rato to lfetme wage ncome, as "net penson benefts rato." And, we nvestgate the level of the net penson benefts rato and changes n the rato due to reform n the case of average marred couples, where the husband works accordng to the average wage profle and the wfe, who s two years younger, s a full-tme homemaker, and n the case of unmarred sngle workers who smlarly work accordng to the average wage profle 9. The estmate results are summarzed n Table 2 and Fgure 6. What we can pont out from these data frst s that even n the 2000 reform, there are great ntergeneratonal nequaltes. The net penson benefts rato s 32.9% for couples of the generaton born n 930, but s mnus 4.3% for those of the generaton born n 200. The ratos for unmarred sngle workers are lower than those for couples, because they cannot receve the beneft of non-workng dependent spouses, who are not oblged to pay any premums n the current system. Both for couples and unmarred workers, however, we can realze that the current system has serous problems regardng ntergeneratonal equty. What are the results of Cases A and B? In both cases, no sgnfcant reducton n dspartes can be observed. In Case A, the net penson benefts rato s lower both for elderly and future generatons. Whle cutbacks n benefts affect the former generaton, the fact that the premum rates are not substantally lower relatve to the present level has an adverse effect on the latter generaton. In Case B, where the use of yelds on the 22

24 nvestment of reserves can also be expected, the results are not very favorable, although they are not so bad as n Case A, except for some ncreases n the rato for those of the generaton born n the 200s. Cases C and D, where the premum rate s set at the current level of 3.58%, ask the elderly generaton, who enjoy advantages under the current system, to bear addtonal burdens, but substantally reduce the burdens on future generatons. In partcular, Case D demands a slghtly smaller sacrfce by the elderly generaton, and also reduces the burdens on future generaton more than n Case C. Ths s because n Case D, as shown n Fgure 4 above, the reducton n benefts s smaller than n Case C, because yelds on nvestment of reserves are added to penson funds. What are the results of Cases A' to D', where reserves are wthdrawn? In all four cases, no great dfferences are observed for young generatons compared to the cases where no reserves are taken out. In Cases B' and C', where account s taken of the wthdrawals of reserves, young generatons are less fortunate to the extent that they cannot expect yelds on nvestments of reserves compared to Cases B and C, whch assume the use of such yelds. On the other hand, addtonal burdens of the reform on the elderly generaton are lessened thanks to wthdrawals of reserves. Ths trend s especally remarkable n Cases A' and C', whch adopt the full pay-as-you-go method and do not rely on yelds on nvestments of reserves. Ths s because takng out reserves controls any ncreases n the cutbacks n benefts resultng from fxed premum rates. 7. Concluson The current publc penson system promses the people to provde benefts that are more than the burdens they can bear wthout dffculty. As a result, t shfts the burdens onto future generatons. Ths means that net penson labltes, the porton of penson labltes supported by no cash funds, contnue ncreasng, and that ntergeneratonal nequaltes contnue to ncrease. Ths paper examnes the penson systems that could prevent an ncrease n net 23

25 penson labltes and correct ntergeneratonal nequaltes, and would also be easer for people to understand. The drecton of penson reforms proposed by ths paper s to keep the amount of benefts wthn premums and state contrbutons plus yelds on nvestment of reserves, where approprate, and to mnmze the level of premum rates. Ths knd of reform would have substantally the same mpact as a changeover to the fundng method. Needless to say, t wll be no easy task to carry out ths type of reform. Ths s manly because the government has to ask the generaton that s already recevng benefts and the generaton that has pad premums to accept partal cutbacks n the benefts t has pledged to offer, whch s a poltcally unacceptable measure. One possble method for copng wth ths problem s to wthdraw reserves, whch now exceed 00 trllon yen. However, takng out the reserves means a dmnshng the possblty of future generatons usng the reserves and ther yelds on nvestments. Any nsttutonal reform that could acheve a smultaneous ncrease n economc benefts for all generatons s a theoretcal mpossblty. Unless we can overcome such ntergeneratonal conflcts of nterest, we wll be unable to go ahead wth penson system reform. A poltcally acceptable reform does not attempt to solve the mportant problem of these conflcts of nterest and mantan benefts at the current level, thereby passng the burdens of dealng wth nsuffcent penson funds onto future generatons. However, ths sort of reform wll only make the problems n the exstng penson system more serous. The government-managed penson system also plays an mportant role n ntergeneratonal support, and should not be evaluated only from a generaton account standpont. However, we cannot approve of the current penson system because of ths. Snce the system rases a serous queston about ntergeneratonal conflcts of nterest, the government should postvely dsclose ts outlook for penson fnance, manly net penson labltes, as well as nformaton about ntergeneratonal nequaltes to provde us wth data for objectve dscussons on better publc pensons. 24

26 Notes:. The dscusson here s based on Osho (2002). 2. For problems related to a shft to the fundng method, see the detaled analyss of Hatta and Oguch (2000). 3. In fact, we can also consder the modfed fundng method, by whch more premums than benefts are provded ntally, and are collected and reserved to lessen the burdens on future generatons. Ths s an ntermedate method between the fundng method and the full pay-as-you-go method referred to above. The publc pensons n Japan are regarded as belongng to the modfed fundng method, but t s doubtful that premum ncome (and state contrbutons), ncludng reserves, s at a level that can cover benefts adequately. Ths problem s dscussed n Secton Ths knd of automatc benefts adjustment mechansm has been adopted by the penson system n Sweden. For an economc evaluaton of the Swedsh method, see Takayama (2002), the Cabnet Offce (2002), etc. 5. Note that yelds on nvestments of reserves are also fxed at ra 0 f the nterest rate s constant. 6. Ths stuaton s often referred to as the "equvalence proposton." 7. Accordng to the "Effects of the New Populaton Estmate on the Fnance of the Employees' Penson and Natonal Penson" publshed by the Mnstry of Health, Labor and Welfare n May 2002, the ultmate premum rate of the Employees' Penson Insurance should be rased to 24.8% based on the new populaton estmate, compared to 2.6% based on the former populaton estmate (both cases assume state contrbutons for the basc penson are one-thrd of benefts). 8. However, a change to a lower-level estmate also works to reduce the benefts labltes of future generatons. Ths effect s not analyzed here. 9. The assumptons about average couples are as follows: the wfe was an offce worker for sx years before her marrage, and the husband has been an offce worker snce he was 20 years of age. He marred at 28 and reached the age lmt at 60. The wfe was born two years after her husband, and was nsured under Employees' Penson Insurance durng the sx years she worked before marrage. After her marrage, the wfe became a full-tme homemaker or 25