Collection Project and emplo releas utilized by for revising the ite For determ It is im si As we mulated reti ed in 2001, m contin portan sponsors. While iniue our sim tng whether an indi re to note that mem nt expen as de it cl and weights in the ulation efforts with this within each of th ses des these for ear that there yee contri vidual cribed earlier is di mulae are often bution has these m is a signi ae groups m rket b behavi m expenses, the following process is undertaken. An ask odel, we hope to ficant shortfall and that major decisions lie ahead if sp com or et of consumer purchases t lay odel to 4 p elicated due to the desi d. For exa e 01(k) d there will undoubtedly be signi plpursue other ans is provid mple, approximately o public p e re of sponsor be priced for the d by an expa olic 30 percent y ficant s to avenues nsion of Witness Disclosure Statement, relevant to econom of the sim a I provide suffic Consumer Price Index. CES data percentages i the state’ individual reaching the Soc nm com ethod based on both e e Ths population is to ro ulat ugh ne i the zero categor ent incentives to non- d life paths for the lowest 2020 ic secur for thm e have adequate resources ity for retirees. For exam 1 ploy ial Security 931 yee demographic infor –19 as well as those provid 60 highl Birth normal retirem e income quartile for those in the 1936 detailed data on expendit yStatem com Coho prts, Fi ensate at levels beyond ent for the in ple, m retire na ent age has a probability of d em ation as well as plan we hope to l Rep m ploy oent. Subsequent to rt, SSA Con ees to contribute in order to com which anyone ube able to int res and income of consumers, a tract No m - atching pr 19 could reasonably : the releas e being in one 6 grate em 40 birt 00-96ovisi -27 h coho e of the piri 332 ons. cal data rt of four plys pursuant to Clause 2(g)(4) of Rule XI of the Rules of the House: (Washington, DC: The Urban Institute, 1999). would be expected to have sufficient retire on long-term assu Oregon stud with technical nondiscri possible assumed “health” well as the de me more y, it was decided that the ap care insurance purchases i m than a de m ographic ch miin n statuses: 1) N i aracte ation testing, this com mis nu ristics o mber of indi n proach coul o to the m ft receiving either hom those ment viduals re consumers. The sur odel p sources. How le d xit be carried to other states as well. Kansas and within the ne could possibl y makes it e e he ver virtuall xt alth or nursing hom v ,y save. We account for t at least 85 percent of the ey year that will allow us to does not y impossible to appropriately provide state e care, 2) hese Senate Special Committee on Aging si deter A co Massachusett situations in t analy Ho esti mm m um lated life paths for the e m ates, but it does provide regional estimates. Thus ze the em health b ine the ination of Form w care patient, s im were ploy o way pact of these policies on an indivi ee’ chosen as the second and thir s. Fir s behavior if one is forced 5500 data and s 3) Nursing home care pa st, we report medians fo third or fourte hlf-reported r incometo observe either aggregate plan data or use quartiles for those in the 1961 d r e dual’ tient, 4) Death, stach of the groups. In other esults wer ates for analysis. Results o , the esti s prospects for adequate retire m e al ate based upon t so used to esti s are broken down into four hf words, the num - e esti the Kansas study m 1965 b ate defined m ment income, as ates of irth cohort benefit the use bers 5 Retirement Planning: Do We Have a Crisis in America? This scenario involves gradually reducing the benefits of those starting to receive retirement and survivor’s would be suff well as the potential benefits to federal and stat presented in Figures A and B provide a num participation inform were presented to the state’s Long-Term • of each type of care fro regions— Northeast, Midwest, South, and West—to account for the The Witne ation on the plan co m icient. This is in large part ss: odels; however, it appears inform m th ntributi e survey on for s above and m due to Care Servic ulae provid ber repr e gove the fact that the m ation ortality esenting the esti es Task Force on Jul ernments via the likel in d b the latter i y . The individual is random the participant. younger cohorts s differences i rather unreliable with respect to mate for th yy 1 redu 1, 2 e 50th percentile when 0 will have additional ction in Medi n 02, and the cost of l ly assigned to the results of caid iving benefiResults From the EBRI-ERF Retire ts. The reduction starts immediately and reaches 10 perce ment Security Projection Model nt of present law benefits in 2010, 15 percent in Jack VanDerhei is a faculty member at Temple University's School of Business and Management y expenditures. esti ranked b the Massa each of these across various parts of the countr ears to accum mating current and/or future accrued benefits. y chusetts study percentage of com four categories with the likelihood ulate the ad wditional 5 percent of com ere presented pensation. Second, y. Consequently on Dec. 1, 2002. we limit the reported values to 2 of , an Therefore, a data falling int pensation. For expense value is calculated using actual o one of the four ca each birth cohort, the l base of defined benefit plan tegories based upon 5 percent of ower incom e 2016, and 22 percent in 2022. by (Department of Risk, Insurance, and Healthcare Management), and also is Research Director of the quartiles are in m co With the exception provisions for salary the esti experience of the elderly for each region, fa mpensation under the assum mated probabilities ore risk of insufficient retirem of studies based on adm related plans was c of each event. If the indi ption that few, if any onstructed to im nistra ily ent income than their higher paid size, and incom tive vi , fa dual does not esti data, e mily mate bene units would be able to contribute in mployee contribution behavior is e le need long-term care, no stochastic fit accruals. vel by averaging the observed counterparts. excess Employee Benefit Research Institute (EBRI) Fellows’ Program, Washington, DC. EBRI is a private, Jack L. VanDerhei, Ph.D., CEBS Moreover, single females Both for in typically The Em expenses are 6 of this percentage on a cont expenses for the elderly ploy assumed to be a function diee Benefit Research Institute’ viduals and f incurred. Each wit o tend to exhi r pub inuous hy in each categor ear, the individual w lic poli of employ basis until retirement age. bict m ym s Edu o akers, be y re vul ee dem mcation and Resear eeting the above criteria. The housing expenses are nerabilit ill o ing able to again face these probabilities (the probabilities of graphic data and perhaps an em y than si qu ch Fund (EBRI-ERF) Retirement antify the ext ngle males while fam ent of the im ploy ilies ee’pending s esti are mate Under this reform alternative, the normal retirement age continues its increase from 65 to 67 but at a faster pace nonprofit, nonpartisan public policy research organization based in Washington, DC. Founded in 1978, its Temple University and EBRI Fellow typically shortfall in ba being in Co Security of the em further broke mbinations of self-reported results were used to Projection Model the different statuses will change as the i the least vulnerable. ployer n down sic retire matcb hing r y m w ent hwas used to e a ether the elderly incom te or a proxy based on e adequacy stim own a has te the a or ndividual becom o initialize IRA For bvi rent their home. The basic health expenditure cc m ous im rued benefits earned and assu 5500 data. However, a si plications. For t accounts. Future IRA contributions es older after reaching age 75 then hose luck gnificant am med to be y enough to ount of be than under current law. Thereafter, the normal retirement age is indexed to longevity (currently assumed to be mission is to contribute to, to encourage, and to enhance the development of sound employee benefit young and were again at age 85) of being in each of the four stat It is also i retained by d the em category m ploy odel has m di ee portant to note ed from sciplined at saving, getti e additional data needs in contributi fined benefit particip SIPP data, while future roll on behavior app that these percentages ants as well as the annual retiremen addition t ng started now ears to Jan. 27, 2004 o be determ over uses. This continue those in the CES. mer activity is likely ely represent ined by was assu to assure them plsavings that need to be generated in an-specific provisions. F s until death m t inco ed to flow fro m a co e or the need for long- that could be mfortable reti m future or produced rement. one month every two years). An increase from 10.6 percent to 12.35 percent in 2030 and to 13.50 percent in programs and sound public policy through objective research and education. EBRI does not lobby and does Since there are term separation from from addition t exam care. the bala ple, the percentage of em o wh nces of any at retirem m ea m ny ploy who a m ent defined contribution ent in those cases in whic incom re old (or nea ploy eees contribut and/or wealth is ring retirem plan ing , cash balance plan, and/or individual retirement h the e up to either the sim ent age) and in the lower-income brackets, ula m ted by ployee the maxi was participat model. Therefore, if the fam mum aming in a ount of com defined pensation public ily 2 Alternativ 050 of the OASI tax rate com e Scenarios pletes the proposal. not take positions on legislative proposals. resources ar account (IRA) of the states’ residents at Social S contributi unit is already generating matched, the 402(g) l Health—The on e plan sponsore basic health expenditures are esti likely to im be called upon eit it, savings for retirem d or the plan m by the previa h ous em xim er directly ent that um m ploy was ecurity norm ated er. Industr is not inclu or studied b using a som indirectl ay y l retirem y E d dat to deal with their inabilit ed in defined BRI in 1996 ewhat different technique and are a are used to estimate the relative ent age. We added to thi . b It would appe enefit or defined y to ar that finance s It is important to note that the analysis presented in the three figures above is limited to the baseline Sen. Craig, Sen. Breaux, and members of the Committee. I am Jack VanDerhei, Temple University T-141 their old age. For those that have a resulting status of home hea contributi likelihood tha am well over 50 com ount the expected retirement inco prised of two parts. The first part uses the CES on plans, IRAs, Social Security and/or percent of the e Knowing the t the balances are rolled over to an IRA, extent of the f mployee contribution is me from uture pr Social ne oblem lth care or nursing t h as above to esti Security eousing eq left with the previous em xplained by these “corner will at least enable policy under uitym , that value n current law as well as under two hom ate the elderly’ e care, th pl eeds to be deducted points,” which would oy m eir duration o akers to try to er, transferred to a s annual health f care assumptions with respect to future Social Security benefits and liquidation of housing equity. • and research The Orgadirector of the EBRI Fellows’ nization: Program. It is my pleasure to appear before you today. prepare to deal with these new e from reform is si not be picked up in the dat expenditures that are paid out-of-p m the esti u m option lated based upon the ployer, or used for mated percentages. s. In an attempt to pr issues when they a other purposes. distribution described above. ovi ocket or are de an appr of the arrive. durations of care found in t not rei oxima m tion of t bursed (covered) he aggregate am or at least not full he NNHS and NHHCS. ount of addit y ional Specifically we have assumed the current statutory benefits will continue to be paid regardless of the The Education and Research Fund (ERF), established in 1979, performs the charitable, educational, and After the duration of care for a nursing m rei oney mscien bursed by Medicare a that would tific functio be needed to pr ns of thnd/or e Institu private Medigap te. EBRI-ERF is ovide home stay basic expenditu a t health insurance, e.g., prescription drugs. or a episode of home health care, the indivi x-exres, we esti empt organim zat ated the def ion (under IR icits that are l C Sec. 501dual will (c)i(3 kely to )) estimated funding difficulties. In VanDerhei and Copeland (2003), we ran two reform scenarios Background Statem 3 ent for the be produced After the retir Recently have a probabilit suppo, EBRI provided rtedby by em com c ent inco y ont ofp ri being dischar aring projected retire butime and wealth was si ons preli anm d inary gra ged nts findings . to one of EBR mIent in -ER m introdu the u Fl i a co sted for each family other three statuses based upon the discharge n me with projected retire otcing new methodol a private foundat unit, we ion (as ogm y to e de ent expenses (both si fim ne xpand t u d lated 1,000 by IR h C e Sec. Defined Benefit Plans designed to ensure 75-year solvency of the program. Under the first alternative, benefits were The ability of future cohorts of retirees to have broadly defined levels of retirement security has been 5 509(a)(3)). deter observations esti usefulness of m The second p mm ates fro inistic m (from a and stochastic) over the sim rt contains insurance premiu NN odeling these data, as well as a better r HS and NHHCS, re etirement age until sp death of the i ectively um lated estimat lifeti . The stochastic e es, including ndividual for m unde e of each fu rstanding of contri Medicare Part B prem xpenses incurred are then single males ture retiree. T butand single females or ion behavi hese deficits ium or s, and is by 401(k) A stochastic j reduced. Under the second alternative, both the Social ob duration algorithm was estimated and applied to ea Security norch indivi mal retiremen dual in the EBRI-ERF t age and the tax the focus of several congressional hearings as well as Senate Special Comm countless public polic ittee on Aging y analyses in the past. In 6 the second person to die f determ approxim plan participants. We utilize a sequential not incom ined by the length aete t related. All of the elderly he additional o of r fam m the stay oney ilies) that would /n are and com um assu res ber of day ponse regression med to participate in Part B, and the pre be re puted qui s the present value of the aggr of care ti red in addition to the retire m model to allow for t es the per diem charge esti egated deficits at m he differing ent inco mium m m is ae t ed for and m rates w odel to predict the num ere increased. As expected, the estimated defici ber of jobs held and age at each job change ts increased under bot . Each time the individual starts a h alternative scenarios, recent y ears, there have been several reasons to revisit earlier studies and conclusions. Perhaps EBRI-ERF has a number of programs: the nursing retire wealth already projected fr incentives faced by determ mined annuall ent age. At that point, we home car y the employ e and hom by the Medicare program om defined benefit and ees at various leve e he rank ordered the obs alth care, re and is specti ls of defined contrib the sam v ervations in ely, in each contributi e nationall ons. Based on findings from u ter region. tion retirement plans, IRAs, Social ms of the present value of the y. For the Medigap in 137 surance new job, the especially for the EBRI-ERF younger model sim birth cohorts. ulates whether or not it will result in coverage in a defined benefit foremost among these is the long-term evolution away from “traditional” defined benefit pension References American Savings Education Council Security deficits and determ distinct matching f premium , and , we assu (under som me all of the elderly o ined the 75th an rmulae, we have estimated a be e of the out d 9 put) 0 purchase a th liquidation a percentil M es of the distrib he adigap polic nd/or annuitiz vioral model that is able to control f y.u A regional estimate is deri ation of housi tion. Next we determ ng net worth t ined or th ved from the e o cover plan, a defined contrib ution plan, both, or neither. If coverage in a defined benefit plan is predicted, ® plans to individual-account defined contribution retirement plans, such as the 401(k) (at least in the Engen, Eric M., William G. Gale, and Corie E. Uccello. “The Adequacy of Household Saving.” Final Choose to Save Education Program Hearing on 1 For any future sim the projected expenses of maintaining t tendency a 2000 sur person without of e u vlat ey done by m ed retirement income accum ployers to substitute bet We the iss Ratings Inc. that need for long-ter h we fam een th ula m ted to ilies’ ca e am received average quotes for three popular t re, this process rep retirem econom ount the ent age and used this i ic st y mandard of livi atch per dollar of em eats annually ng. The present value of . The process repeats nform ploy ation to ee ypes of time seri Moreover, in es information from our baseline analy the sis above Bureau of Labor Statis we assumed that retirees tics (BLS) iwould not us s used to predict what ty e their net housing pe of plan private sector). In addition to transferring investment risk from the employer to employee, this Version. December 1999. Defined Contribution Research Program Fellows Program for individuals receiving home health ca any deter contributi Medigap policies (A, F, an deficits were accu mine the on a percentage of co nd the mm axi ulmum ad J) in 47 states and the ted annuall m percentag pensati y and then av on that woul re or nur e of com sing pDistrict. The est en ed need to be saved to have s rsation the aged for all retirees in the same birth cohort and home care at y are willing to m ithe end of their duration mates are calu c afficient additional ulated from tch. We decom of the three pose it will be. equity to supplement their retirement income in any way (including housing equity loans). In evolution has also resulted in m Health Confidence Survey any employ Program ees incr easing their exposure to longevity risk as well as stay income to offset the presen gender/fam em polic ploy /care and subsequently y types averaged over the states in ee con ily categories. tribution beha t value of accum if not receiving the specia vior into a series of 1 pe the respective ulated de rcent of com ficits for the 75th and 90th lized care regions to arrive at the estimate for ea again at their next birthda pensation intervals and therefore are percentiles of t y. Those who ch region. he VanDerhei and Copeland (2003) we estimate two additional responses. Our second scenario assumed Health Security/Quality Research Program being put in a situation where they may make one or more of a number of choices that would Moore, James F., and Olivia Mitchell. Projected Retirement Wealth and Savings Adequacy in the are si distributio able to m mulat odel not o n. ed to die, of course, are not further si Retirement Planning: Do We H nly the marginal incentives to contribute at that in mulated. As with the basic health ave a Crisis in America? terval but also the “option value” care expense s, the While the BLS inform any net housing equit Policy Forum yation provi is annuitized at r s des significant detail etirement. Given the stochastic nature of the analy on the generosity parameters for defined benefit sis we were adversely impact their eventual retirement income. Moreover, those employees who remain active Health and Retirement Study. Working Paper 6240. Cambridge, MA: National Bureau of Economic qualification The m that making t This approach is taken for odel was expanded so that it coul h of Medicaid by e contributiotwo reasons. First, su n at that inter income and asset levels d be nationa val provi fficient qualit des for the em l iis considered to see how much of the stochastic n scope and i y dat ployee. an do not itial estim exist for the m ates were publi atchsihed in ng of also able to model a third s plans, preliminary Retire analy ment Confi sis cenario indicated that sever dence where we assu Survey Pral of these provisions were likely ogr m am e that housin g equity is not li to be highly quidated until the time participants within the def Results From the EBRI-ERF Retire ined benefit plan system may ment Security Projection Model have already had unexpected modifications to Research, October 1997. Retirement Security Research Program Figure A (pg. 13) shows the expenses the Novem retiree medic mb uer 2003 s atl care ( be covered by EBRI I as well as the generosity of and ssue Brief m the individual to de edian per (“Can A centage of meterm rica com cost of the cov i Afford Tom np e the individual’ ensation that m orrow’ erage) and Medigap polic u ss Retirees: Results Fr st be saved each y final expenditures for the ear until y use to om the it is first needed to m correlated (especiallyitigate an annual de for integrated plans). Therefor ficit. At that point we assume any e, a time series of several hundred define residual value is invested in d benefit the t ype of plan sponsored by the employer. Recent legal actions and lack of legislative clarity and/or Social Security Research Program retire care. Only EBRI-ERF Retirement Securit Contributi various characteristics of the elderly ment for a 75 percent on those expenditures attributable to behavior for defined contri confidence level when y Projecti . Second, the on Model” bution the indi plan hea com ) and was discussed at a day s vi other than 401(k) plans is lth status of th dual—not the bined with sie elderly m Medicaid program—are ulated reti at th - esti rem long EBRI pol e age of 65 is m ent wealt ated from h, iself- not cy the sa plans per me man year were coded to allow for assign ner as an individual account retire mem nt ent p to the individuals in the EBRI-ERF lan. The relative impact of the second scenario model. regulatory guidance may suppress future activity in this regard; however, projecting future benefit Olsen, Kelly, and Jack VanDerhei. “Defined Contribution Plan Dominance Grows Across Sectors and Jan. 27, 2004 Education Programs?Policy Forums, Briefings, Round Tables considered as assu forum reported responses to publ known, let al m (“Assessing Future ing current Social Security benefits and one over the expenses to the individual entire course of their remai iRetirement Security c survey data. and as a that housing equity is never liquida Wit resul h tn hing life. T te Results of the EBRI/ERF Retirement Income in any of the hus, by “deficit” cal assumingc everyone has a ted. ulations. was rel atively minor; however, the third scenario had a much more dramatic impact, reducing the accruals may prove problematic even for those employees still participating in a traditional form of Employer Sizes, While Mega-Defined Benefit Plans Remain Strong: Where We Are and Where We 2 Publication Programs?printed and onlin e Projection M standard level of coverage elim odel”) held Dec. 4, 200 inates try 3. In an atte ing to differentiate am mpt to make the r ong all possible coverage types as well esults easier to comprehend, the annual deficits for 200 Although the Tax Refor 3 b my Act of 1986 a 23 percent. t least partially modified the constraints on integrated pension final-average Are Going.” EBRI Issue Brief defined benefit plans. no. 190 (October 1997). EBRI Issue Briefs, EBRI Notes, EBRI Databook on Employee Benefits, primary For exam Invest as deter ment R m output m ple, ining whether the sick or healthy e all turns m etri edian gender/fam —Although c was changed to one that re the EBRI-ERF ily com have th binations m presents the additional percentage of com odel has been designed to ge e c in the first two incom overage. Therefore, averagi e quart nerate investment rates ng of the expenses iles for the ol pensation dest Total Expenditures plans by adding Sec. 401(l) to the Internal Revenue Code, it would appear that a significant percentage EBRI Health Benefits Databook, Fundamentals of Employee Benefit Programs, Policy Studies that each si birth cohort are at the 25 percent of co of return on a stochastic ba over the entir mulated observation would e population should sis, for purposes of th have off mpensation th need to sav setting effects in the aggr is analy by e (in addition to reshold. For t sis we are hpresenting the results obtained from the other components already ose in the highest income quartile egate. The elderly of defined be individual or fa nefit sponsors have re milies’ expenses are th tained Primary Insurance Am en the sum of their assu ount (PIA)-offset plans. In or med deterministic expenses der to Conclusions While several studies have attempted to project retirement income and wealth, there have been few Rajnes, David. “An Evolving Pension System: Trends in Defined Benefit and Defined Contribution m for this birth running it i odeled) from n a deter cohort, the 2003 until ministi percentages of com the time they c mode. We adopt t retired. The pensa he sam tion needed to be entire distribution of e asset-specific rates of retu saved are 23.8 percent for outcomes rn that were used in was arraysinge ed and based upon t estimate the offset provided heir demographic characteri under the p Jack L. VanDerhei, Ph.D., CEBS lan form stics plus any si ulae, the EBRI-ERF m mulated stochastic expen odel computes the em ses that they plo may yee’s We have purposely structured many of our assumptions to provide conservative estimates of the attem Plans.” pts to reconcile their results with the un EBRI Issue Briefs no. 249 (September 2002). certain amount and duration of retiree expenditures. • Contracts: 4 charts displaying m fem the Social Security Adm The total deter ales, 13.9 percent for single m ministic expenses for elderly individuals edian com inistration’ pensation percentag als Model of Income in the Nea es, and 6.1 percent for fam es were cr or fam eated ilies ilies. rfor each cohort for a variety of Ter are then the sum m (MINT) model. of the value in all have incurred. In each subsequent Average Indexed Monthly Earnings, Pri Tem year of life, th ple University and EBRI Fellow mary Insu e total expenditure rance Amount, and covered com s are again calculated in this pensation values for amounts that would be needed to be saved while employees are working to alleviate any deficits. For Moore and Mitchell (1997) estimated how much Health and Retirement Study (HRS) respondents EBRI does not have any contracts with the federal government in 2003, and did not in 2002 or 2001. confidence levels and assum the expense categories for fap m tions regar ily size, fad m ing Soci ily incoal Security me level, and region and liquidation of of the inhousing equit dividual or family y. . manner. The base y the birth cohort. ear’s expenditure value estimates excluding the health care expenses are adjusting example, we have assumed in this version of the model that all employees continue to work until ages 51 to 61 would need to save from the current year until retirement assuming they wanted to VanDerhei, Jack, and Craig Copeland. “Can America Afford Tomorrow’s Retirees: Results From the Additional analy Figure B (pg. These expenses 14) shows th make up the basic annual (recurring) e sis was conducted e median additional savin to show the proba gbilit s required to xpenses for the indivi y of retirement security provide retirement adequacy for a dual if each indivi or family. However, dual Retiree Exp annuall y using the assu enditures med general inflation rate of 3.3 percent from the 2001 OASDI Trustees Social Security norm al retirement age, even though there has been a long-term trend toward earlier preserve preretirement consumption levels after retirement. They found that the savings requirement EBRI-ERF Retirement Security Projecti E–mon Mo ail: jack del.” @vanderhei. EBRI Issue Bri come f no. 263 (November 2003). would save an additional 5 percent of com 90 percent co if the individual or fam nfidence level (9 out ily meet the inco of 10 me and sim pensation ulate ass d life paths). In thi et for the remainder of his or tests for Medicaid, Medicai s case, nearly all of the her working c d is assumed ato cover reer. The expendit Report, while the health care expenses are adjusted ures used in the model for the elderly co annuall nsist of two com y using the 4.0 percent medical con ponents—deterministic and sumer Defined Contribution Plans retirement (albeit one that seems to be reversing in recent years). We have also assumed that for the median family would be 7 percent of compensation if the retirement age were 65. As expected, Internet: www.ebri.org gender/fam the basic heal ilyth care expenses (both parts), not the individual or fa status at retirement combinations for the first three inco mily m . Furtherm e quartiles of the earliest birth ore, Part B premium price index th stochastic expenses. The deter at corresponds to th ministic e June expenses in 2002–June 2 clude those expenses that the elderly 003 level. incur from a Initial Account Balances individual account balances are “self-an —Previous studies on the nuitized” ove EBRI/ICI Participant-Directed Retirement Plan r a period of time that expands the individual they found tremendous heterogeneity among families with respect to the required savings rate. . “A behavioral model for predicting employee contributions to 401(k) Plans.” North American cohort are at the threshold ( relief for the low-income elderly the median for fam (not qualifying for ilies in Medicaid) is also incorporated. the third quartile is estimated at 24.8 percent of Accumulation Phase of EBRI Model basic need or want of daily life, while the stochastic expenses in this model are exclusively health- Data Collecti and/or familyon Project ha life expectancy ve analy by five zed t yeh ars, even e average though there appe account balances for 401( ars to be lim k) participants by age and ited evidence that this Another appr Endnotes oach was followed by Engen, Gale, and Uccello (1999) using both HRS and the Survey Actuarial Journal (First Quarter, 2001). 1 co mpensation). Those in the highest income quartile for this birth cohort all have requirements that The EBRI-ERF event related—e.g., an admission to a nursing model is based on a six-year time s home or the commencement of an episode of eries of administrative data from more thhom an 10 e Results tenure. Recen type of buffer is actually See EBRI Issutly published e Briefs no. conte 24results show that th 9, “m An plated by retirees as Evolving Pensi e o yn ear-end 1999 aver Sya st risk-reduction device. em: Trends in age balance r Defined Benefi atnged from and Define $4,479 d of Consumer Finances (SCF). Using a theoretical model, the authors estimated the ratio of a would prove difficult if not impossible to implement: median single females are estimated to now need Stochastic Expenses million 401(k) participants and m health care—that occur only for a porti ore than 30, on, if ev 000 plans, as well as a time series of several hundred er, during retirement, not on an annual basis. C The prim for participants in their 20s ontributiary on P objective of t lans” (Septemh b with less than thr er 20 is analy 02);s no is is . 2 to com 32, “T ee he b yin ears of tenure with their current e Ce the si hanging Face mulated of Private Retirem retirement income and weal ent m Pl ploy ans” ( er to Apri th with l 2001); household’s wealth to its earnings as benchmarks to evaluate savings adequacy. Using intermediate . “The Changing Face of Private Retirement Plans.” EBRI Issue Brief no. 232 (April 2001). an to save d no. 190 mo, “D re than 25 efined Con percent of com tribution Plan pDominance Grows Ac ensation, single male ros s 22.1 pe s Sectorrcent of com s and Employp er Si ensation, and zes, While Mega- The second com plan descriptions used to ponent of health expend provide a sample of the itures is the various defined benefit and result of simulated health events that would defined contribution plan Even with the the si $198,595 for mulated retiree expenditures to deter participants in their 60s w se conservative biases built in, the ho have been mine how numm bers appear tro with the current em uch each family ubling for some age cohorts and unit would need to save toda ployer for at least 30 years y wealth measures, the authors estimated that 59.7 percent of the SCF households exceeded the Defined Benefit Plans Remain Strong: Where We Are an d Where We Are Going” (October 1997). families 10.1 percent of compensation. provisions applicable to pl require long-term care in a nursing an participants. In a home or hom ddition, several public survey e-based setting for the elderly s based on participants’ . Neither of these Deterministic Expenses (thereby alm (as percentag ost fatalistic for others. The goo effectivel e of their current wages) to mainta y eliminating anyd ne capa ws is bility that if many for in a pre-specified “ IRA rollovers). of the younger coh comfort level” orts begin savin (i.e., confide g a nce simulated median wealth-to-earnings ratio in 1992. While this model includes the capacity for self-reported answer simulated types of care wo s (the Surve uld be reim y of Consumer Fina bursed by Medi nces care because [SCF], the Current Po they would b pulation Survey [CPS e for custodial (not ], The determ level) that they will be able to afford the reasonable am inistic expenses are broken d ount to supplement their Social Secur o si wn into mulated expenses for the rem seve ity and q n categories—food, appar ualified retirem ainder of the lifeti ent plans now, th el and services (dry me of the ey have a sensitivity analysis on an ad hoc increase in simulated retirement needs and/or life expectancy, there is 2 See EBRI Issue Brief no. 266 (February 2004), “Americans’ Future Retirement Security: Implications of the and the Surve rehabilitative) care. The incidence of the nur Will Individuals Be Able to Save Enough y of Income and Program Participation sing home and home health care and the resulti on Their Ow [SIPP]) were n (Over and Above What is used to model participation, wages, ng cleaning, haircuts), transportation, entertain Unfortunately, the EBRI/ICI database d good chance of provi family unit (i.e., death of second spouse i ding themselves wi oth reasonable assu es not curre n a fam ment, ily read ). ntly We report these savings rat ing and ed provide rance that the ucation, h detailed in y will at l ousin formation on other ty g, es by east be able t and basic health age cohort, o cover pes no attempt to empirically estimate the incidence, duration or cost of potentially catastrophic medical EBRI-ERF Retirement Security Projection Model” (forthcoming). and initial account balance inform expenditures on the care are estimated from ation. the 1999 National Nursing Home Survey (NNHS) and the expenditures. Each of these expenses is Already Modeled)? basic retire of defined contribution pla family status (at retirement), and gender. ment expenditures. However, ns nor does it allow analy esti chang Six five mated for es in public polic -year birth coh sis of defined contributi the elderlyo y and additional (65 or older) by fa rts are simu on lated. The oldest g balances that resources fro mily size (single or may mr oup have costs. 2000 National Home and Hospice Care Surve 3 y (NHHCS). NNHS is a nationwide sample survey of couple) and f Figure C (pg. fam been left with previous employ was born in t ilies and charities wou h a 15) pr e period 1 mily incom ovides 93 el6 (less than $15,000, d be required to provid another way to 1 ers. The 940 incl EBRI-E of illustratin usive while the youn $15,000 to $29,999, RF e adequate retirement inco model uses self g which cohort gest g-reported responses for whether an r s and oup was bor ma$30,000 y be the me for retirees with n in the m or m ost vulnerable to ore in 2002 period 1961 to Jack VanDerhei and Craig Copeland, “A behavioral model for predicting employee contributions to 401(k) This inform nursing hom at es, their current residents and discharg ion is combined with U.S. Department es that was conducted b of Labor Form 5500 data y the to National Center for model participation dollars) of th inadequate financial resources in retire greater longevit indivi 1965 inclusiv dual hae fam s a defined contributi e. Three co y who suff ily/indi mbinations of vidual. er serious and persiste on bal m ge a ent. This nder/fam nce to est nt chronic disease. figure starts wi ily imate a particip status at retirem th the baseline scenario de ation m Our esti ent were reported: fam odel and the reporte mates include bot scribed ily, single h the d value plans EBRI/Milbank Study for Oregon, Kansas and Massachusetts .” North American Actuarial Journal (First Quarter, 2001). and initial account balance inform Health Statistics from July through Decem ation for all privat ber 1999. T e-sector defined contribution participants, as well he NHHCS is a nationwide sample survey of above (current Social Security benefits and n status quo for is male, and single fem modeled as a function Social Security benefits as well as tw ale. In of age and tenu addition, the re. relative o liqu income was reported by io reform dation or an scenario nuitization of net s that would estimating lifetim housing decrease benefits for equit e iny come ) and The views expressed in this statement are solely those of the author and should not be attributed to Temple The Employee Benefit Research Institute (EBRI) and the Milbank Memorial Fund, working with the 4 as contribution behavior f home health and hospice care agencies, their current or non-401(k) defined contri and discharge patients that was conducted b bution plans. Asset allocation information is y the The estim assu future generations. Un quartiles (from MiI vNT mes that ersity or t assates are derived from umh each worker es a C e Em 2002 though retirem P pI logr yee owt Ben contributes an additio h rat efit Research e the 2000 of 3. ent age) for each of the 18 com 50 peConsumer rcent Institu , a r te, nal 5 percent of com eal its officers, tru Expenditure Sur rate of return stees, sp fob r st v inations of bi pey ensation from oc on (CES) co ksso ors, f 6.or 98o rth cohort and nducted b perce th 2003 unti er staff. Th nt, any d th a real le e rate governor of Oregon, set out to see if the necessary retirement security analysis could be undertaken on based on previously of return for National Center for Health Statisti Employee Ben bonds efit Research of publis 3.00 percent. In hed results of the EBRI stitute is a It subtrac cs fro non m pro ts 1 August through fperce it, nonp n /ICI Participant-Directed R t artisan from each of , Decem public policy research org thb e s er 2000. tock and bond etirem real rates an ent Plan ization. of ret Data urn to Bureau of Labor Statistics of the U.S. Depart retirement age to supplem gender/fa Contribution Behavior mily status at retire —Previous research on em ent his or ment. her Social S meecurity and tax-q nt of Labor. The survey ployee contribu ution behavi alified retirement plans. Th targets the total or to 401(k) plane s has a state-specific basis and undertook an initial study on the future retirees of Oregon. The results, reflect administrative cost (See Eric Toder et al., Modeling Income in the Near Term: Projections of Retirement noninstit percentage of each cohort esti often been li ution mited by lack of ade alized population (ur mated to have suffici quate data. This is primarily ban and rural) of ent th retirement inco e United States and is the basic source due to the t me and/or wealth to cover the ypes of matching form of data ulae 10 11 12 3 4 6 5 9 8 2 7 Figure C Figure B Figure A a Percentage of Added Compensation That Must Be Saved Annually Until Percentage of Added Compensation That Must Be Saved Annually Until Percentage of Retirees Estimated to Have Sufficient Retirement Income/Wealth Retirement For a 90% Chance of Covering Basic Retirement Expenses Retirement For a 75% Chance of Covering Basic Retirement Expenses by Saving Additional 5% of Compensation Each Year From 2003 Until Retirement (assumes current Social Security and housing equity is never liquidated) (assumes current Social Security and housing equity is never liquidated) (assumes current Social Security benefits) family family single female single female single male single male family single female single male a a 25% 25% 100% 90% 20% 20% 80% 70% 15% 15% 60% 50% 10% 10% 40% 30% 5% 5% 20% 10% 0% 0% 0% 123412341234 123412341234 123412341234123412341234 123412341234 123412341234 1936–1940 1936–1940 1936–1940 1941–1945 1941–1945 1941–1945 1946–1950 1946–1950 1946–1950 1951–1955 1951–1955 1951–1955 1956–1960 1956–1960 1956–1960 1961–1965 1961–1965 1961–1965 a Birth Cohort/Income Quartiles Birth Cohort/Income Quartile Birth Cohort/Income Quartile a 25% = 25% or more. Source: EBRI-ERF Retirement Security Projection Model. 25% = 25% or more. Source: EBRI-ERF Retirement Security Projection Model. Source: EBRI-ERF Retirement Security Projection Model. Assumes current Social Security, and that housing equity is never liquidated. The model includes the possibility of chronic long-term home health care and nursing home expenses. 14 13 15