Upload
karachibreeze
View
216
Download
0
Embed Size (px)
Citation preview
7/30/2019 The Success of the U.S. Retirement System
1/64
The Success of the U.S. Retirement System
7/30/2019 The Success of the U.S. Retirement System
2/64
Copyright 2012 by the Investment Company Institute. All r ights reserved.
Suggested citation: Brady, Peter, Kimberly Burham, and Sarah Holden. 2012. The Success of the U.S. Retirement
System (December). Washington, DC: Investment Company Institute.
7/30/2019 The Success of the U.S. Retirement System
3/64
Pet er Brady, ICI Senior Economist ; K imberly Burham, ICI Economist ; a nd Sarah Holden, ICI Senio r Director
of Retirement and Investor Research, prepared this report.
Key Findings ............................................................................ ...................................................................... 3
Introduction ..................................................................................................................................................4
Households Savings Goals ......................................................................................................... .................. 5
Focus o Saving Changes over the Lie Cycle ....................................................................... ................... 5
Focus o Saving Varies with Household Income .....................................................................................6
Using Focus on Retirement Savings to Understand Statistics on Retirement Plan Coverage .................8
The U.S. Retirement System: A Retirement Resource Pyramid ....................................................................9
Evidence o the Success o the U.S. Retirement System .......................................................................... .... 10
Assets Earmarked or Reti rement over Ti me .......... ...... ..... ...... ..... ...... ..... ...... ...... ..... ...... ..... ...... ..... ...... 10
Wealth and Retirement Accumulations o Successive Generations ....................................................... 12
Poverty Among Older Individuals over Time ..................................................................... .................. 14
Changes in Households Standard o Living at Retirement .................................................................. 14
Changes in Household Wealth in Retirement ..................................................................... .................. 17
Components o the U.S. Retirement Resource Pyramid ............................................................................. 17
Social Security Beneits ........................................................................... .............................................. 17
Homeownership..................... ............................................................................... ................................ 22
Employer-Sponsored Retirement Plans and IRAs ............................................................................ ..... 26
Other Assets ................................................................... ....................................................................... 32
TheSuccessoftheU.S.RetirementSystem
7/30/2019 The Success of the U.S. Retirement System
4/64
2 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
Estimating the Components o the Retirement Resource Pyramid ............................................................ 34
What Do Current Trends Suggest About Retirement in the Future? .......................................................... 36
Historical Trends in Reti ree Income rom Employer-Sponsored Retirement Plans and IRAs .............. 36
Projecting the Future o Retiree Income rom Employer-Sponsored Retirement Plans and IRAs ....... 37
What Are the Primary Areas o Concern Going Forward? ......................................................................... 39
Conclusions ......................................................................... ........................................................................ 42
Appendi x: Additional Deta il on Surveys, Housing-Related Data, and Near-Retiree Households ...... ..... ..... 43
Notes ..................................................................... ............................................................................... ....... 53
Reerences .......................................................................... ......................................................................... 57
7/30/2019 The Success of the U.S. Retirement System
5/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 3
Key Findings
The U.S. retirement system h s successfully provided dequte retirement resources to genertions
of Americns. Studies that examine spending, i ncome, and wealth conclude that households, on
average, maintain their standard o living in retirement. By some measures, retirees appear to be
better o than other segments o the population: in 2011, a lower percentage o the population
aged 65 or older lived in poverty (9 percent) than the percentage aged 18 to 64 (14 percent) or the
percentage younger than 18 (22 percent).
To dte, successive genertions of retirees hve been better off thn previous genertions.Analysis
shows that, on average, more-recent generations o households have higher levels o resources to
draw on in retirement than previous generations. Other measures also indicate improvements in
retiree well-being. For example, the poverty rate among people aged 65 or older has declined rom
nearly 30 percent in 1966 to 9 percent in 2011.
The shift in privte-sector retirement plns from predominntly defined benefit (DB) plns to
predominntly defined contribution (DC) plns is unlikely to reduce retirement prepredness. The
extent to which previous generations o retired households relied on income generated by private-
sector DB plans is oten exaggerated. Since 1975, the prevalence o income generated by private-
sector retirement plans o all types (measured by both the share o retirees with the income and the
amount o income) has increased substantially. In act, because they are better suited to the mobile
U.S. workorce, several studies conclude that the shit rom DB plans to DC plans in the private
sector will increase retirement resources or most households.
The focus of household sving chnges over the life cycle. In 2010, only 14 percent o households
younger than 35 reported that retirement was their primar y sav ings goal, compared with nearly
hal o households aged 50 to 64. Younger households typically are ocused on other goals:
32 percent o households younger than 35 reported that saving or education, homes, or other
large purchases was their primary saving goal. Because households may choose to save orretirement when older, it is di icult to assess retirement preparedness or households that are
not in or near reti rement.
Rther thn the trditionl three-legged stool n logy, pyrmid is more ccurte depiction of the
resources Americns rely on in retirement. The retirement resource pyramid has ive components:
Social Security; homeownership; employer-sponsored retirement plans (DB and DC); IRAs; and
other assets. Households do not rely on each resource equally; the composition o the retirement
resource pyramid var ies across households.
Socil Security benefits provide brod bse of resources for ne rly ll retirees. Social Security has
evolved into a system that provides substantial retirement resources throughout the income andwealth distribution, and provides the primary retirement resource or workers with low lietime
earnings. For workers born in the 1940s, Social Security is projected to replace 70 percent o
average lietime earnings or the bottom 20 percent o earners and 29 percent o average lietime
earnings or the top 20 percent o earners.
7/30/2019 The Success of the U.S. Retirement System
6/64
4 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
For mny households, the homes they live in represent the second most importnt retirement resource
fter Socil Security. Older households are more likely to own their homes; more likely to own
their homes without mortgage debt; and, i they sti ll have mortgages, are more likely to have small
mortgages relative to the value o their homes. Retired households typically access this resource
simply by living in their homes and not paying rent.
Employer-sponsored retirement plns nd IRAs ply complementry role to Socil Security benefits,
incresing in importnce for households for whom Socil Security replces smller shre of ernings.
Nevertheless, employer-sponsored plans and IR As are an important resource or households
regardless o income or wealth. In 2010, about 80 percent o near-retiree households had accrued
beneits in retirement plans or IRAs. Nearly hal o near-retiree households with income less t han
$30,000; 71 percent o near-retiree households with income o $30,000 to $54,999; and 94 percent
o near-retiree households with income o $55,000 or more had retirement accumulations.
Introduction
This study examines the empirical evidence on the eectiveness o the U.S. retirement system. The
empirical evidence demonstrates that the U.S. retirement system is successul. On average, households areable to maintain their standard o living in retirement. To the extent that there has been a trend in retiree
well-being, measures such as income, wealth, and poverty rates show that successive generations o retired
households have become better onot worse oover time. The U.S. retirement system will ace many
challenges andas has always been the casethe uture is uncertai n. However, changes to pr ivate-sector
retirement savingin particular, the growing importance o employer-sponsored deined contribution
(DC) retirement plans and individual reti rement accounts (IRAs)do not represent a major challenge or
the system. To date, the shit to a more account-based system has not been associated with a reduction in
the income o retired households, and there is reason to believe that many households will beneit rom
this shit.
The U.S. retirement system has ive key components. For retirees as a group, Social Security beneits
represent the largest component o retirement resources. For some retirees, homeownership represents
the second most important resource. By owning rather than renting the house that they live in, these
households do not need to generate as much monthly income in retirement. Employer-sponsored
retirement plansincluding both deined beneit (DB) plans and DC plansand IRAs provide a
supplement to Social Security bene its. In act, or many households, retirement plans and IRAs are more
valuable than either their Social Security bene its or the equity they have in their homes. Although less
important on average, retirees also rely on other assets in retirementassets such as inancial holdings
outside o retirement plans and IRAs, investment real estate, and business investments. The importance
o these ive components in providing retirement resources diers rom household to household. In their
entirety, these ive components have allowed recent generations o retirees, on average, to maintain t heir
standard o living in retirement.
7/30/2019 The Success of the U.S. Retirement System
7/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 5
The irst section o this paper illustrates that saving or retirement is one o many savings goals o U.S.
households; moreover, the ocus o household saving changes over the lie cycle and varies with household
income. The second section introduces a new way to think about retirement planning in the United
States: the retirement resource pyramid. In the thi rd section, the success o the U.S. retirement system is
evaluated using various measures. The sections that ollow describe the ive key building blocks o the U.S.
retirement system, the roles they play, and how those roles have evolved. Estimates o the components
o the retirement resource pyramid are provided or households approaching reti rement to analyze howthe pyramid d iers across households nearing retirement. Some insights into what current trends suggest
about the uture o retirement are explored, including discussions o the likely impact o the shi t among
private-sector employers rom DB retirement plans to DC retirement plans, and the major risks aced by
uture reti rees. A summary o the key results concludes the paper.
Households Svings Gols
Households have many savings goals and the goals vary across the lie cycle and across income groups.
In addition to saving or retirement, households save to und education expenses, to purchase homes, to
pre-und other large purchases, or to have cash on hand or emergencies or unexpected needs ( liquidity).
The Survey o Consumer Finances (SCF) ask s households their most important motivation or saving.1
In the 2010 SCF, 35 percent o U.S. households listed saving or liquidity as their most important reason
or saving (Figure 1). In act, regardless o age or income, liquidity is an important motivation or saving
or a substantial portion o households. Retirement was the next most common reason, with 30 percent
o households listing reti rement as the primary reason they save.2 Another 23 percent o U.S. households
listed education, buying homes, or saving or other large purchases as their primary savings goal.
Retirement is not the most important reason to save or all households, and it is oten not the most
important reason to save or any given household in every stage o lie. For this reason, it is di icult to
assess the adequacy o retirement resources or households other than those near or in retirement.
Focus of Sving Chnges over the Life Cycle
As households approach retirement age, they become more ocused on sav ing or retirement. For
example, in 2010, only 14 percent o households with a household head younger than 35 saved primarily
or retirement; 32 percent o these households cited education, homes, or large purchases as the primary
reason they saved (Figure 1). Those numbers are reversed or older households aged 50 to 64: nearly hal
o these households reported that retirement was their primary savings goal and only 15 percent reported
that saving or education, homes, or large purchases was their pri mary savings goal.
The household survey data showing that older households are more ocused on retirement saving are
consistent with economic models o lie-cycle consumption, which predict that households rationally delay
saving or retirement until later in their working careers when they typically have higher earnings. It is
not that younger households do not save, but rather that they t ypically save or other reasons. Oten the
irst savings pr iority or a household is to build up a rai ny-day und. In addition, younger, newly ormed
households oten invest in their uture in ways that do not result in the accumulation o inancial assets or
ormal retirement savings. Examples include purchasing a home; purchasing consumer durables such as
vehicles, appliances, and urniture; unding education; and raising children. These households may choose
to ocus on saving or retirement when older.
7/30/2019 The Success of the U.S. Retirement System
8/64
6 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
Focus of Sving Vries with Household Income
Households at dierent income levels also have dierent primary reasons or saving. Having cash on hand
in the event o emergency or unexpected need (l iquidity) was an important reason to save or all i ncome
and age groups (Figure 2). However, regardless o their age, households with lower incomes were more
likely to report liquidity as their primar y reason or saving. Among households aged 35 to 49 in 2010,
34 percent o households with income less than $25,000 indicated that liquidity was their primary savings
goal, compared with 28 percent o households with income o $100,000 or more (Figure 2, top panel).
A similar pattern by income is seen among households aged 50 to 64 (Figure 2, lower panel).
Saving or education, homes, or large purchases is more important to households with lower incomes than
saving or retirement. For example, among households aged 35 to 49 in 2010, 36 percent o households with
income less than $25,000 reported that saving or education, homes, or large purchases was t heir primary
reason or saving, compared with 23 percent o households with income o $100,000 or more (Figure 2,
top panel). At all levels o income, older households were less likely to save primarily or these reasons.
Among households aged 50 to 64 in 2010, 26 percent o households with income less than $25,000 cited
saving or education, homes, or large purchases as their primar y reason to save, compared with 10 percent
o households with income o $100,000 or more (Figure 2, lower panel).
FIgURe
Households Primary Reasons for Saving Vary by Age
Percentge of U.S. households by ge, 2010
Primry reson for sving
All
households
Age of hed of household
Younger
thn 35 35 to 49 50 to 64 65 or older
Retirement 30 14 29 48 24
Liquidity (cash on hand, emergencies,
unexpected needs)35 39 35 30 39
Education, home, or large purchases 23 32 27 15 19
education 8 11 16 4 2
Buy own om 3 9 3 1 (*)
Lar purcass 12 12 9 10 17
Other 8 14 6 4 11
Invstmnts 1 2 1 1 1
For t family 6 11 5 3 6
No particular rason 1 1 1 1 4
Cannot or do not save 4 2 3 3 7
Total 100 100 100 100 100
(*)=lsstanprcnt
Not:Componntsmaynotaddtottotalbcausofroundin
Sourc:InvstmntCompanyInstituttabulationsoftSurvyofConsumrFinancs
7/30/2019 The Success of the U.S. Retirement System
9/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 7
FIgURe
Households Primary Reasons for Saving Vary with Household Income
Householdincome*
Householdincome*
Primaryreasonforsaving
Cannotordonotsave
Other(investmentsforthefamilynoparticularreason)
Liquidity(cashonhandemergenciesunexpectedneeds)Educationhomeorlargepurchases
Retirement
All
ormore
to
to
to
to
Lessthan
All
ormore
to
to
to
to
Lessthan
Percentageofhouseholdswithhouseholdheadagedtobyhouseholdincome*
Percentageofhouseholdswithhouseholdheadagedtobyhouseholdincome*
*Totalisousoldincombfortaxsin
Not:Componntsmaynotaddtoprcntbcausofroundin
Sourc:InvstmntCompanyInstituttabulationsoftSurvyofConsumrFinancs
7/30/2019 The Success of the U.S. Retirement System
10/64
8 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
In contrast to other reasons to save, the ocus on saving or retirement increases with household income.
For example, among households aged 35 to 49 in 2010, 15 percent o households with income less than
$25,000 said retirement was their pr imary reason or saving, compared with 44 percent o households with
income o $100,000 or more (Figure 2, top panel). Among households aged 50 to 64 in 2010, 27 percent
o households with income less than $25,000 indicated retirement was their primar y reason or saving,
compared with 65 percent o households with income o $100,000 or more (Figure 2, lower panel).
One reason lower-income households may be less ocused on saving or retirement is that, with limited
resources, they prioritize saving or liquidity. Another reason is that the Social Security beneit ormula is
highly progressive and resu lts in beneits that replace a much higher percentage o earnings or workers
with lower lietime income.3 The result is thatat any given agelower-income households tend to be
less ocused on saving to supplement Social Security in reti rement.
Using Focus on Retirement Svings to Understnd Sttistics on Retirement Pln Coverge
Understanding di erences between households in their motivation or saving can provide insight into
statistics on employer-sponsored retirement plan coverage. Using Current Population Survey (CPS)
data or 2011,4 Brady and Bogdan (2012a) ind that 50 percent o private-sector wage and salary workers
were employed by irms that sponsored retirement plans (including both DB and DC plans). However,
access to retirement plans is not random. Workers who are part o groups who tend to be more ocused
on saving or reti rement also are much more likely to work or employers that oer plans. For example,
limiting the analysis to ull-time, ul l-year workers aged 30 to 64, access to retirement plans increases to
60 percent. I the analysis is narrowed urther to the groups o workers most likely to be ocused on saving
or retirementworkers aged 30 or older with at least moderate levels o earnings and all but the lowest-
earning workers aged 45 or olderthen 69 percent work or employers that sponsor plans. In add ition,
some in this group without access to plans at their own employers have access to plans through their
spouses employers. Taking into account access through spouses, 74 percent o workers who are likely to
be ocused on saving or retirement have access to employer-prov ided retirement plans, and 93 percent
participate in the plans oered.
Looking at the percentage o all workers who have access to retirement plans at their employers at any
single point in time understates the share o the population who will reach retirement with work-related
retirement beneits. Many young workers, low-wage workers, or part-time workers are more concerned
with saving or a ra iny day, to purchase homes, or to und education than they are with saving or
retirement. However, young workers do not remain young throughout their working careers, and many
low-wage and part-time workers do not remain low-wage and part-time throughout their careers. Many
workers who do not have access to employer-sponsored retirement plans today will have access to a plan
either through their own employers or their spouses employersprior to reti rement.
7/30/2019 The Success of the U.S. Retirement System
11/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 9
The U.S. Retirement System: A Retirement Resource Pyrmid
Households rely on many types o resources in retirement and the role each type plays has changed over
time and varies across households. The traditional analogy i s that retirement resources are like a three-
legged stool. This analogy implies that everyone should have resources divided equally among Social
Security, employer-sponsored pension plans, and private savings. This is not currently, nor has it ever been,
an accurate picture o Americans retirement resources. A pyramid is a better representation o retirement
resources.
The retirement resource pyramid has ive basic components, which draw rom government programs,
deerral o compensation until reti rement, and other savings. The ive components o the retirement
resource pyramid are Social Secur ity; homeownership; employer-sponsored retirement plans (both
private-sector employer and government employer plans, as well as both DB and DC plans); IR As
(including rollovers); and other assets.
At the base o the ret irement resource pyramid is Socia l Security (Fig ure 3). Social Securit y covers
households across a ll levels o earnings; however, it replaces the largest port ion o average lietime
earnings or households with low lietime earnings.
A resource avai lable to the vast majorit y o retired households is the home in which they l ive (Figure 3).
Homeownership increases with age and is high across all i ncome groups among near-retiree households.
Households who own homes oten have no or low mortgage debt by the time they reach retirement age.
Households do not have to sell their homes to beneit rom them in retirement; they simply have to live
in them. Homeownership is like having an annuity that provides rent, as the home provides a place to live
that otherwise would have to be rented.5
The next two layers o the reti rement resource pyramid consist o accumulations in employer-sponsored
retirement plans (both private-sector employer and government employer plans, as well as both
DB and DC plans) and IRAs (both contributory and those resulting rom rollovers rom employer-
sponsored plans) (Figure 3). In 2010, the SCF data show that accrued beneits and asset accumulations
in employer-sponsored retirement plans and IR As constituted a resource or about 80 percent o near-
retiree households. Near-retiree households across all income groups have these retirement beneits, but
employer-sponsored retirement plans and IR As ty pically provide a larger share o resources or higher-
income households.
Finally, at the top o the retirement resource pyramid are other assets that a household may own
(Figure 3). These assets can be inancial assetsincluding bank deposits and stocks, bonds, and mutual
unds owned outside o employer-sponsored retirement plans and IR As; and noninancial assets
including business equity, nonresidential propert y, second homes, vehicles, and consumer durables
(long-lived goods such as household appliances and urniture). Assets in this category tend to be ownedmore requently by higher-income households.
7/30/2019 The Success of the U.S. Retirement System
12/64
10 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
Evidence of the Success of the U.S. Retirement System
Generations o retirees have beneitted rom the resources provided to them by the U.S. retirement system,
and the empirical evidence demonstrates that, on average, retirement outcomes have improved over
time. The amount o assets earmarked or use during retirement has increased over time and successive
generations o households have reached retirement with higher levels o wealth, on average, than prior
generations. Furthermore, poverty rates or people aged 65 or older have allen over time. In addition,
expenditure and income data generally indicate that households are able to maintain their standard oliving when they retire. Final ly, research inds that retirees, on average, maintain su icient wealth to
generate as much income as they could when irst reti red.
Assets Ermrked for Retirement over Time
The amount o assets earmarked or retirement has grown, even when adjusted or in lation and growth
in the number o U.S. households.
Assets earmarked or retirement include IRAs, DC plans, private-sector DB plans, state and local
government pension plans, ederal pension plans, and annuities. In mid-2012, total retirement assets stood
at $18.5 tri llion (Figure 4, top panel) and accounted or 36 percent o U.S. households inancial assets.6
DC plans and IRAs accounted or $9.8 tril lion, or more than hal, o total retirement assets. Fueled in
signi icant part by rollovers rom employer-sponsored retirement plans (both DB and DC; both private-
sector employer and government employer plans), IRAs represented more than one-quarter o the total
U.S. retirement assets in mid-2012, compared with about 10 percent in 1985. DC plans, which include
401(k) plans, 403(b) plans, and 457 plans, also have risen in share over the past two decades, largely the
result o private-sector adoption o 401(k) plans. Despite the decline in private-sector DB plans, such plans
had assets o $2.4 tril lion in mid-2012, and they continue to pay beneits to retirees.7 Federal, state, and
local government plans had $4.6 tr illion in mid-2012, predominantly in DB plans.8
FIgURe
Retirement Resource Pyramid
Other assets
IRAs
(including rollovers)
Employer-sponsored retirement plans
Homeownership
Social Security
Sourc:InvstmntCompanyInstitut
7/30/2019 The Success of the U.S. Retirement System
13/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 11
FIgURe
Retirement Assets Have Grown over Time
Annuities
Federalpensionplans
Stateandlocalgovernmentpensionplans
PrivateDBplansDCplans
IRAs
UStotalretirementmarketassetstrillionsof(nominal)dollarsend-of-periodselecteddates
AverageretirementassetsperUShouseholdconstantdollarsroundedtonearestend-of-periodselecteddates
Q
Q
e e e
AnnuitisincludallfixdandvariablannuityrsrvsatlifinsuranccompanislssannuitisldbyIRAs(b)plansplansand
privatpnsionfundsSomoftsannuityrsrvsrprsntasstsofindividualsnotspcificallyforrtirmntowvrinformation
tosparatoutsucrsrvsisnotavailablBcausannuitisldinIRAsplansand(b)plansarnttdfromtflowoffunds
accountsannuitis(lifinsurancpnsionfundrsrvs)fiurandrportdintirrspctivcatorisbyICIICIrportsalowrannuitis
totaltantflowoffundsaccounts(sUSFdralRsrvBoarda)FdralpnsionplansincludUSTrasuryscurityoldinsoftcivilsrvicrtirmntanddisabilityfundtmilitaryrtirmntfund
tjudicialrtirmntfundstRailroadRtirmntBoardandtforinsrvicrtirmntanddisabilityfundTsplansalsoinclud
scuritisldintNationalRailroadRtirmntInvstmntTrustandFdralemploysRtirmntSystm(FeRS)TriftSavinsPlan(TSP)AltoutTSPissimilartoa(k)planasstsldbytTSParincluddinfdralovrnmntrtirmntasstsandarnot
includdintDCplantotalAtyar-ndtTSPld$billioninntasstsFormordtailsrardintTSPsCliftonLarson
AllnLLP Tiscatoryincluds(b)plansplansandprivatmployr-sponsordDCplans(includin(k)plans)Dataarstimatd Not:Componntsmaynotaddtottotalbcausofroundin Sourcs:InvstmntCompanyInstitutUSFdralRsrvBoardNationalAssociationofgovrnmntDfindContributionAdministrators
AmricanCouncilofLifInsurrsIntrnalRvnuSrvicStatisticsofIncomDivisionUSDpartmntofCommrcUSCnsusBurau
USDpartmntofLaborBurauofLaborStatisticsandUSDpartmntofLaboremployBnfitScurityAdministration
7/30/2019 The Success of the U.S. Retirement System
14/64
12 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
Even when adjusted or both in lation and growth in the number o U.S. households, assets speci ically
earmarked or reti rement have increased signi icantly over time. As o June 30, 2012, the average amount
o retirement assets per U.S. household, adjusted or in lation, was 2.7 times higher than in 1985, and
5.6 times higher than in 1975 (Figure 4, lower panel). In constant 2012 dollars, average retirement assets
per U.S. household were approximately $153,100 in mid-2012, compared with about $56,100 in 1985 and
about $27,300 in 1975.
Welth nd Retirement Accumultions of Successive Genertions
Data show that more recent cohorts o retirees tend to enter retirement wealthier than previous cohorts.
Haveman et al. (2007) use data rom two surveysthe New Beneiciary Survey (NBS) and the Health
and Retirement Study (HRS) to construct a comprehensive measure o wealth that includes the present
value o Social Security beneits and DB pensions.9 The authors ind that new reti rees in the mid-1990s had
higher levels o wealth than new reti rees in the early-1980s. Compared with new retirees i n the early 1980s,
wealth or new retirees in the mid-1990s was 60 percent higher or married couples and about 35 percent
higher or single men and women. A more recent study suggests that this trend has continued. Using HRS
data, Gustman, Steinmeier, and Tabatabai (2009) construct a similarly comprehensive measure o wealth
and compare three groups o households approaching retirement: those aged 51 to 56 in 1992; those aged51 to 56 in 1998; and those aged 51 to 56 in 2004. The authors ind that each successive cohort approaching
retirement was wealthier, with the 2004 cohort 7 percent wealthier than the 1998 cohort and 12 percent
wealthier than the 1992 cohort.
The global inancial cri sis o 2008 and the great recession (December 2007 through June 2009) negatively
impacted the wealth o households across the board.10 Coming ater the bear market and recession earlier
in the decade, the inancial cr isis and great recession represented the second time household balance
sheets had taken a substantial hit in less than 10 years. Nevertheless, over the past decade, wealth and
inancial assets held up better or older households than younger households. Bricker et a l. (2012) report
that between 2001 and 2010 average net worth ell or most age groups, but it ell less or households aged
55 to 64 than younger age groups and actually rose or households aged 65 or older.
Further, although retirement assets ell between 2007 and 2010, assets in retirement accounts ell less
than other assets and grew as a portion o household inancial asset holdings. Looking over a longer
period and ocusing on near-retiree households, the SCF data show their retirement assets have increased
substantially, despite the recent drop in value. For example, since 2001, about seven in 10 near-retiree
households had DC plan assets, IRAs, or both (Figure 5). In addition, the median amount o retirement
assets was $101,350 in 2010, compared with $63,719 in 2001 (in 2010 dollars).
7/30/2019 The Success of the U.S. Retirement System
15/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 13
Ner-Retiree Households Anlyzed in This Pper
To provide insight into what Americans have accumulated to prepare or retirement, this paper uses
the Survey o Consumer Finances (SCF) to analyze households headed by a working individual aged
55 to 64, which are identiied as near-retiree households. In the SCF, a head o household is the male ina mixed-sex couple or the older person in a same-sex couple. The SCF collects data on household total
income beore taxes or the calendar year preceding the survey. For the analysis i n this paper, a near-
retiree household is de ined as a household in the 2010 SCF that is headed by a working individual
aged 55 to 64, excluding the top and bottom 1 percent o the income distribution.
Note: For a brie description o the SCF, see Figure A.1 in the appendix. For a summary o indings
rom the 2010 SCF, see Bricker et al. (2012).
FIgURe
Real Median Balance and Percentage of Near-Retiree Households with Retirement Assets
Near-retiree households, assets expressed in constant 2010 dollars, 19892010
Medianretirementassets
amongsavers
(leftscale)
Sharewithretirementassets
(rightscale)
Not:RtirmntasstsincludasstsinIRAsandDCplanaccountsNar-rtirousoldsarousoldswitaworkinadadto
xcludinttopandbottomprcntoftincomdistribution
Sourc:InvstmntCompanyInstituttabulationsoftSurvyofConsumrFinancs
7/30/2019 The Success of the U.S. Retirement System
16/64
14 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
Poverty Among Older Individuls over Time
One way to measure the success o the U.S. retirement system is to consider poverty rates among the
elderly. The poverty rate among individuals aged 65 or older has declined substantial ly since 1966
(Figure 6). In 1966, nearly 30 percent o people aged 65 or older had income below the poverty line. In
2011, the poverty rate or thi s group stood at 9 percent. Even during the great recession, the poverty
rates among the elderly declined slightly. In act, poverty rates or people aged 65 or older are lower than
poverty rates or both the working-age population and chi ldren. In 2011, only 9 percent o people aged65 or older had income below the poverty line, compared with 14 percent o people aged 18 to 64, and
22 percent o people younger than 18.11
Chnges in Households Stndrd of Living t Retirement
It is di icult to measure reti ree well-being directly. One approach to determine i households can maintaintheir standard o liv ing in reti rement is to compare a households actual wealth to a prediction o optimal
wealth accumulation. Another approach is to analyze actors that impact standard o livingsuch as
consumption and income. The data illustrate that the U.S. retirement system provides most households
with su icient resources to maintain their standard o living in retirement.
FIgURe
Poverty Rates Among People Aged 65 or Older Have Fallen over Time
Poverty rtes by ge, percentge of individuls in ge group, 19 662011
Youngerthanto
orolder
Individualsage
Sourc:USCnsusBurauCurrntPopulationSurvy:toAnnualSocialandeconomicSupplmnts
7/30/2019 The Success of the U.S. Retirement System
17/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 15
WealthAccumulationsatRetirementandOptimalWealth
Research suggests that most U.S. households nearing reti rement have accumulated wealth that is equal to
or greater than the optimal amount o wealth they should have accumulated, as predicted by a lie-cycle
model o consumption. Lie-cycle models assume that individuals maximize lietime well-being (or utility)
by optimal ly choosing consumpt ion (and thus savings) over the course o their li eti me.12 Using HRS data
combined with administrative data on lietime earnings, Scholz, Seshadri, and Khitatrakun (2006) use a
lie-cycle model to solve or the optimal amount o wealth or each household and compare that to eachhouseholds actual wealth accumulations. They ind that, among households aged 51 to 61 in 1992, ewer
than 20 percent o households had wealth less than the target amount, and that, among these households,
projected shortal ls were smallon average about $5,000.13
RetireeConsumption
Research on retirees consumption patterns suggests that reti red households generally maintain their
liestyles when they transition into retirement. Hurd and Rohwedder (2008) analyze HRS data and ind
that, other than those who retired because o poor health, spending declines by only a small amount
when a person retires.14 The small declines can be explai ned by reductions in ood and work-related
expenditures. In other research, Hurst (2008) analyzes Consumer Expenditure Survey (CEX) data and alsoinds that most o the declines in spendi ng near retirement are due to reductions in ood and work-related
expenditures. When Hurst (2008) uses Universal Product Code (UPC) data tracking speci ic purchases to
examine ood expenditures more closely, he concludes that reti red households pay lower prices or their
grocery bil ls than younger households. In addition, urther analysis o HRS-Consumption and Activities
Mail Survey (CAMS) data by Hurd and Rohwedder (2006) inds that retirees eat at home more oten.15
Overall, the data suggest that retired households reduce expenditures by substituting money to purchase
goods in the market or time producing goods at home, such as in ood preparation or shopping research.
RetireeIncome
Anot her way to gain insight into retiree well-being is to analyze retirees income. Analysis oadminist rative tax data shows that individuals, on average, maintain their in lation-adjusted, net-o-tax
income near the time o their irst claiming o Social Security beneits. Brady and Pierce (2011) use data
on individuals aged 55 to 61 who did not receive Social Security beneits and who iled a tax return in 1999
to examine the transition into retirement.16 The data allow the authors to ollow this group o individuals
through 2008 and to measure their employment-related income, which consists o wages, Social Security
bene its, and distr ibutions rom employer-sponsored plans (both private-sector employer and government
employer plans, as well as both DB and DC plans) and IRAs. Because the data include inormation
reported to the IRS through inormation returns (such as Form W-2, Form 1099-SSA, and Form 1099-R),
employment-related income can be measured even i individuals do not ile a tax return i n uture years.
To allow su icient data or comparison, the study ocuses on those individuals who irst received SocialSecurity retirement beneits at some point between 2000 and 2005 and compares the employment-related
income o each individual in later years to the individuals employment-related income in t he year prior to
irst claiming Social Security beneits.
7/30/2019 The Success of the U.S. Retirement System
18/64
16 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
The study inds that employment-related income, net o taxes and adjusted or in lation, remains ai rly
stable, on average, in the year individuals irst c laim Social Security beneits, and or three years thereater
(Figure 7). In the irst year Social Security beneits were claimed, the median individual had inlation-
adjusted ater-tax employment-related income that was 107 percent o their real net employment-related
income in the prior year. By the third year ater claiming Social Security beneits, the median individual
had real net employment-related income o 100 percent o their real net employment-related income in the
year pr ior to cla iming Social Security beneits.
FIgURe
Employment-Related Income Before and After Claiming Social Security Benefits
Per capita1 work-related income,2 net of taxes and adjusted for inflation, expressed as a percentage of same measure
one year prior to first receipt of Social Security retirement benefits; working individuals ag ed 55 to 61 in 199 9 who
had no Social Security benefits, filed a tax return, first received benefits between 2000 and 2005, and were alive
three years after the year be nefits were first received
10 and 90
25 and 75
Median
Percentile of the distribution
8779 76
71
187
209199 195
5549 46 42
132140 136 134
107 107 104 100
Year of first receipt One year after first receipt Two years after first receipt Three years after first receipt
Formarridindividualsincomistsumofwork-rlatdincomfrombotspoussdividdbytwoemploymnt-rlatdincomisdfindastsumoffdrallytaxablwasandtipsrportdonFormW-pnsionandIRAdistributions
rportdonForm-RandSocialScuritybnfitsrportdonForm-SSAlsstsumofIRAcontributionsrportdonForm
andRotcontributionstoanmployrplanrportdonFormW-lsstsumoffdralincomtaxs(iffilinataxrturn)andpayroll
taxs(ifworkin) Sourc:BradyandPircanalysisoftaxrturndataproviddbytStatisticsofIncomDivisionoftIntrnalRvnuSrvic
7/30/2019 The Success of the U.S. Retirement System
19/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 17
Chnges in Household Welth in Retirement
Households may begin reti rement with su icient resources, but this does not necessarily mean that they
are able to maintain those resources through many years o retirement. However, studies that examine
households later in retirement ind that reti rees, on average, maintain su icient wealth to generate as
much income as they could early in reti rement.17 Haveman et al. (2005) analyze NBS data and report that
10 years ater reti rement, retirees could generate as much annual income rom their wealth as they could
when irst reti red. Love, Palumbo, and Smith (2008) use HRS data and show that the income stream thatcould be generated by wealth actually rose as households moved through retirement. Poterba, Venti, and
Wise (2012) use HRS data to st udy wealth and income at the end o lie. They ind that, although many
households had low inancial assets at the end o lie, income at the end o lie was similar to income when
the households were in their it ies and sixties.
Components of the U.S. Retirement Resource Pyrmid
This sect ion examines the components o the U.S. retirement system, or the retirement resource pyramid,
and how they have evolved. The resources oten relied on by U.S. households in retirement are Social
Security beneits; homeownership; work-related retirement accumulationsrom both pr ivate-sectoremployer and government employer plans, as well as both DB and DC plans; and IRAs (including
contributory and rollover). In addition, households may have other assets.
Socil Security Benefits
For most households, one o the most valuable resources is their Social Security retirement beneits.
However, this resource typical ly is not included in measures o household wealth. Social Secur ity is
designed to be progressive; that is, it provides a h igher proportion o retirement beneits relative to
pre-retirement earnings or workers with low earnings than or workers with high earnings. Thus, it
comprises a higher share o lower-earning households retirement income, and i it were to be counted as
an asset, would comprise a higher share o assets in such an augmented balance sheet or those households.
SocialSecurityBenefitChangesoverTime
When Social Security was signed into law in 1935, it was intended to replace a modest portion o income.
Changes to the system since its inceptionin particular, two periods o expansion, irst in the 1950s
and then again in the 1970s increased beneits substantially, especially or those with low lietime
earnings.18 Described as a cornerstone or U.S. retirement security at its beginning,19 Social Security has
transormed into a comprehensive government-provided pension or workers with lower li etime earnings
and a strong oundation or retirement security or those with higher lietime earnings.
7/30/2019 The Success of the U.S. Retirement System
20/64
18 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
One way to see the evolution o the Social Security system is to look at estimates o replacement rates
calculated by the Social Security Administration (SSA) or hypothetical workers. Each year the SSA
calculates an average wage index (AWI) that roughly corresponds to the average annual earnings or
workers covered by the Social Security system.20 The SSA uses the AWI to create three hypothetical
workers. These hypothetical workers are low earners, with career average earnings approximately
45 percent o the AWI; medium earners, about 100 percent o the AWI; and high earners, about 160
percent o AWI.21
In 1940, according to the SSA estimates, replacement rates at the normal retirement age(NRA), measured as irst-year beneits divided by career-average wage-indexed earnings, ranged rom
18 percent or a high earner to 29 percent or a low earner (Figure 8). For those retiring currently, Social
Security beneits are considerably higher in real terms and are about twice as high when measured as a
percentage o average earnings than when Social Security began to pay beneits. Beneits are projected to
replace 36 percent o career average earnings or a high earner and 58 percent o career average earni ngs
or a low earner. Under current law, these replacement rates are projected to remain stable into the
uture.22
FIgURe
Social Security Replacement Rates over Time for Representative Workers
Estimates of replacement rates (first-year benefits relative to average indexed earnings) at normal retirement age;
percentage of li fet ime earnings ; 1940203 5
Disabilitybenefits
Congressbeginstoincreasebenefitswithspecialacts
MedicareandMedicaid
COLAformulaadjusted
Automaticannualcostoflivingadjustments
(COLAs)
SocialSecurityamendmentscoverageoffederalemployeesincreasetaxratesfutureincreasestonormalretirementage
Repealofretirementearningstest
Medicareprescriptiondrugbenefit
Lowearner
Mediumearner
Highearner
Not:Lowmdiumandiarnrrfrtoscaldarninstatrflctpattrnsofworkandarninsforypotticalworkrsovrt
coursofacarrProjctionsassumnocanincurrntpolicy
Sourcs:ICIsummaryoflislativcansandSocialScurityAdministrationOfficoftCifActuary
7/30/2019 The Success of the U.S. Retirement System
21/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 19
SocialSecurityReplacementRatesbyHouseholdLifetimeEarnings
Social Security beneits vary with earni ngs and provide higher replacement rates or lower-earning
households. The replacement rates that SSA ca lculates are a useul measure to calibrate the evolution
o Social Security bene its over time (Figure 8). However, they are not an ideal measure o the range
o replacement rates that actual workers can expect to receive under the system, as they only measure
the replacement rates or three hypothetical workers.23 An alternative estimate o replacement rates is
produced regularly by the Congressional Budget Oice (CBO). Rather than create hypothetical workers,the CBO builds a model based on the actual earning records o a large sample o workers, which allows
or the calculation o replacement rates or a wider range o incomes.24 To analyze how Social Security
bene its di er by earnings, the CBO irst groups individuals by the decade in which they were born, and
then ranks individuals by household lietime earnings.
Social Security is designed to replace a greater portion o income or lower earners because it uses a
progressive ormula to determine beneits.25 This means that or lower earners, Social Security unctions
as a pension program that provides an annuity beneit that replaces a high percentage o their average
lietime earnings. For the cohort o individuals born in the 1940s, the CBO analysis shows that Social
Security beneits a re projected to replace 70 percent o average earnings or the typical individual in the
bottom 20 percent o individuals ranked by l ietime earnings (Fig ure 9). The replacement rate drops to
47 percent or the second quintile, and then declines more slowly as li etime earnings increase. Social
Security beneits a re projected to replace a considerable ract ion o earnings29 percentor even the
top 20 percent o earners.
FIgURe
Social Security Benefit Formula Is Highly Progressive
CBO estimates of first-year benefits relative to average indexed earnings by household lifetime ea rnings (median),
1940s birth cohort, percent
HighestFourthMiddleSecondLowest
29
3742
47
70
Lifetime earnings quintile
Sourc:ConrssionalBudtOffic(sConrssionalBudtOfficb)
7/30/2019 The Success of the U.S. Retirement System
22/64
20 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
ImpactofSocialSecurityonPoverty
The expansion o Social Security beneits likely was a major actor in the reduction in the elderly poverty
rate since the late 1960s. The Social Security program has expanded since its inception in 1935 both in
the scope o its coverage and in the generosity o its beneits.26 The current system now covers most U.S.
workers, and over the years has added spousal, survivors, dependent childrens, and disability beneits.27
In addition, government-provided medical bene its or low-income and elderly individuals were introduced
in the mid-1960s and subsequently expanded.
The generosity o the basic beneit or those attaining the Social Security NRAmeasured as the
share o a workers average lietime earnings replaced by i rst-year beneitsincreased substantially in
the early 1950s and again in the 1970s (Figure 8). When instituted, Social Security beneits were not
automatically adjusted or either in lation or wage growth. In the 1950s and 1960s, beneit increases
were made periodically by special acts o Congress. Legislation enacted between 1950 and 1954 raised
replacement rates or new retirees substantially. Between 1955 and 1969, the legislative adjustments kept
bene its airly stable relative to wages. Leg islation passed between 1969 and 1973 led to a second jump
in replacement rates. Further, legislation enacted in 1972 provided that bene its would be automatically
indexed or in lation through an annual cost o livi ng adjustment (COLA).28 However, a law in the
method used to calculate in lation-adjusted beneits led to beneits being over-indexed or in lation. A
new beneit ormula was adopted in 1977, which ixed the law and clawed back some o thepresumably
unintendedincrease in beneits caused by the method introduced in 1972.29 The 1977 beneit ormula is
still used today and, under current law, the Social Security replacement rate at NRA is projected to remain
stable or uture retirees.30
When Social Security was created in 1935, it was intended to be a modest source o retirement income. 31
However, rom its modest beginnings, Social Secur ity has evolved to become a system designed to be the
primary means o support or retirees with low lietime earni ngs. The progressive beneit structure o
Social Security provides workers with low lietime earnings an income stream that replaces a substantial
portion o their pre-retirement income. For workers with higher lietime earnings, it provides a loorbelow which ret irement income cannot all.
7/30/2019 The Success of the U.S. Retirement System
23/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 21
Medicre nd Medicid Provide Resources to Retirees
Other important government-sponsored resources or retirees include Medicare and Medicaid. Along
with Social Security, Medicare and Medicaid are part o Americas saety net or retirees.
Medicare
The Social Security Act o 1965 created Medicare, which is inanced by the ederal government and
individual premium payments.* Medicares purpose is to provide health insurance to people aged
65 or older regardless o income or medical history. Typically, to be eligible a person must be aged
65 or older, an American citizen or legal resident or ive years, and have paid Medicare taxes or
10 years. Medicare beneits have expanded over time, including the addition o Part D prescription
drug coverage. In 2011, 93 percent o people aged 65 or older were covered by Medicare. Medicare
provides widespread health insurance coverage, which is an important resource or retirees. In 2011,
only 2 percent o Americans aged 65 or older were uninsured compared with 18 percent o Americans
younger than 65.
Medicaid
Medicaid, also created as a par t o the Social Security Act o 1965, is a means-tested program that
provides access to health insurance or people with low income and low i nancial assets. Medicaid
is paid or by the ederal government, state governments, and some county governments. In 2008,
about 4.6 million low-income people aged 65 or older received beneits rom Medicaid. Medicaid is an
important source o unding or long-term care: in 2008, Medicaid covered nearly 41 percent o the
total costs o nursing acility care.
* For additional inormation, see U.S. Social Security Administration, Ofce o Retirement and Disability Policy 2012a. See DeNavas-Walt, Proctor, and Smith 2012.
See U.S. Social Security Administration, Ofce o Retirement and Disability Policy 2012b.
7/30/2019 The Success of the U.S. Retirement System
24/64
22 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
Homeownership
For many households nearing retirement, the home that they live in is one o their most valuable
assets. Not only do older households tend to be more likely to own their homes compared with younger
households, they also are less likely to have mortgages on their homes. In addition, among older
households with a mortgage, the amount outstanding tends to be small in comparison with the value o
the home. Homeownership is a key resource available to households in reti rement. Although owning a
home does not direct ly provide a stream o income that can be used to und consumption in retirement, it
reduces monthly housing expenses. That is, retirees can l ive in their home and not pay rent elsewhere. In
this way, homeownership reduces the need to generate regular monthly income in retirement.
HomeownershipbyHouseholdAge
Homeownership by Birth Cohort. The rate o homeownership tends to i ncrease rapidly with age and
then stabilize, i ncreasing rom around 30 percent or younger households to over 80 percent or older
households (Figure 10, top panel).* For example, in 1989, the homeownership rate was 30 percent or
households born between 1960 and 1969 (the 1960s birth cohort) and aged 20 to 29 at the time o the
survey; 71 percent or the 1940s birth cohort (aged 40 to 49 at the time o the survey); and 79 percent or
the 1920s cohort (aged 60 to 69 at the time o the survey). Similarly, in 2010, the homeownership rate was
58 percent or the 1970s birth cohort (aged 31 to 40 at the time o the survey); 76 percent or the 1950s birth
cohort (aged 51 to 60 at the time o the survey); and 83 percent or the 1930s birth cohort (aged 71 to 80 at
the time o the survey).
Supplementl Security Income
An important government-sponsored resource or retirees i s the Supplemental Security Income
(SSI) program.* SSI was enacted by the Social Security A mendment Act o 1972 to provide beneits
to people aged 65 or older, and blind or disabled adults and children. Eligibil ity and ederal beneits
are nationally un iorm, but may be supplemented by states. SSI is a means tested program or those
with little or no Social Security beneits or other resources, with payment levels determined by bothincome and wealth. In 2011, the ederal SSI beneit or those with no other income was $674 per month
or individuals and $1,011 per month or couples. In 2010, more than 2 million people aged 65 or older
received SSI beneits.
* For additional inormation, see U.S. Social Security Administration, Ofce o Retirement and Disability Policy 2012c.
See U.S. Social Security Administration, Ofce o Retirement and Disability Policy 2012d.
* Figures 10 and 11 present housing-related data or households grouped by the decade in which the head o the household was
born. Housing-related data are available on these households every three years rom 1989 to 2010 rom the SCF. For eachyear in which data are avai lable, the relevant statistics are plotted at the approximate midpoint o the age range or each10-year birth cohort. For example, in 1989, the 1950s birth cohort ranged in age rom 30 years (or those born in 1959) to 39
years (or those born in 1950), and the 1989 data point or th is group is plotted above age 35. In 2010, the 1950s cohort rangedin age rom 51 years to 60 years, and the 2010 data point or this group is plotted above age 56. Presented in this way, patternscan be di scerned both across households and over time. For the data plotted in Figure 10, see Figures A.2 and A.3 in theappendix. For the data plotted in Figure 11, see Figure A.4 in the appendix.
7/30/2019 The Success of the U.S. Retirement System
25/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 23
Impact of Developments Between 2007 and 2010 on Homeownership. Between 2007 and 2010, the
homeownership rate or younger birth cohorts increased less than would be expected based solely on the
experience o preceding birth cohorts at the same ages. However, the relative drop in homeownership
experienced during the great recession needs to be kept in perspective: as homeownership expanded prior
to 2007, the rate o homeownership at any given age was higher or each successive cohort. In general,
the drop in homeownership or younger birth cohorts between 2007 and 2010 simply brought their
homeownership rate more in line with their predecessors.
Changes in the homeownership rates or each bi rth cohort between 2007 and 2010 can be seen by
comparing the last two observations (the dark blue and purple dots) on each cohorts line (Figure 10, top
panel).* For example, the line or the 1970s birth cohort shows that their homeownership rate edged down
rom 60 percent in 2007 (when they were aged 28 to 37) to 58 percent in 2010 (when they were aged 31
to 40). However, the 58 percent rate o homeownership or the 1970s birth cohort in 2010 was not much
dierent than the 57 percent rate o the 1950s birth cohort in 1989 (when they were aged 30 to 39) or the
57 percent rate o the 1960s birth cohort in 1998 (when they were aged 29 to 38).
MortgageDebtbyHouseholdAge
Incidence of Mortgage Debt by Birth Cohor t.As households age, they are more likely to own a home with
no mortgage debt. In particular, there appears to be a st rong tendency within cohorts to pay o mortgage
debt ater reaching age 50. For example, among those in the 1930s birth cohort, 33 percent o households
(representing 42 percent o home-owning households in the cohort) 32 owned a home with no mortgage
debt in 1992, when they were aged 53 to 62 (Figure 10, lower panel). This compares with 44 percent o
households (representing 53 percent o home-owning households in the cohort)in the same 1930s birth
cohort in 2001, when they were aged 62 to 71; and 53 percent o households (representing 64 percent o
home-owning households in the cohort) in the same 1930s birth cohort in 2010, when they were aged
71 to 80.
Although the pat tern o paying o mortgage debt a ter reaching age 50 can be seen within cohorts, youngerbir th cohorts appear to be somewhat less likely to own a home with no mortgage debt. This can be seen
by compar ing di erent cohorts when they were similar in age. For example, in 1989, when they were aged
60 to 69, 54 percent o all households in the 1920s birth cohort (representing 68 percent o home-owning
households in the cohort)owned their homes with no mortgage debt (Figure 10, lower panel). This
compares with 42 percent o all households in the 1930s birth cohort (representing 51 percent o home-
owning households in the cohort) in 1998, when they were aged 59 to 68; and 35 percent o all households
in the 1940s cohort (representing 42 percent o home-owning households) in 2010, when they were aged
61 to 70.
* The lone exception is the 1920s birth cohort. The last two observations plotted or this group are t he 2001 and 2004 surveys.
7/30/2019 The Success of the U.S. Retirement System
26/64
24 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
FIgURe
Homeownership and Mortgage Incidence by Age
Percentage of U.S. households by 10-year birth cohort of the head of household; 19892010
Born
Born
Born
Born
Born
Born
Ageattimeofsurvey
Homeowninghouseholds
807570656055504540353025
Born Born
Born
Born
Born
Born
Ageattimeofsurvey
Homeowninghouseholdswithnomortgagedebt
Not:AisbasdontaoftadofousoldFordataplottdintisfiursFiursAandAintappndix
Sourc:InvstmntCompanyInstituttabulationsoftSurvyofConsumrFinancs
7/30/2019 The Success of the U.S. Retirement System
27/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 25
Impact of Developments Between 2007 and 2010 on the Incidence of Mortgage Debt. The tendency o
households to pay o their mortgages as they age held up even during the great recession. For example,
among all households in the 1940s birth cohort, the share who owned a home with no mortgage debt
increased rom 32 percent in 2007 to 35 percent in 2010 (Figure 10, lower panel). When the analysis is
narrowed to home-owning households in the 1940s bir th cohort, the share o homeowners with no
mortgage debt increased rom 37 percent in 2007 to 42 percent in 2010.
Loan-to-Value Ratios by Birth Cohort. For households that have not paid o their mortgage debt, mortgage
debt as a percentage o home value tends to decline with age (Figure 11). For example, in 2010, the median
loan-to-value ratio or homeowners with debt was 81 percent or households in the 1970s birth cohort
(aged 31 to 40 in 2010); 53 percent or households in the 1950s birth cohort (aged 51 to 60); and 35 percent
or homeowners in the 1930s birth cohort (aged 71 to 80).
Impact of Developments Between 2007 and 2010 on Loan-to Value Ratios.The eects o the great recession
and housing market developments can be seen by the act that all birth cohorts have an increase in the
median loan-to-value ratio or homeowners with debt between 2007 and 2010 (Figure 11). For example, the
loan-to-value ratio or the 1950s birth cohort increased rom 45 percent in 2007 to 53 percent in 2010. The
changes in loan-to-value ratios across the cohorts were dr iven primarily by large drops in home valuesrather than increases i n mortgage debt; overall mortgage debt actually declined between 2007 and 2010.33
FIgURe
Loan-to-Value Ratios Typically Are Lower Among Older Households with Mortgages
Medin lon-to-vlue rtio (percent) by 10-yer birth cohort of the h ed of household; 1989 2010
Born
Ageattimeofsurvey
0
10
20
30
40
50
60
70
80
90
0
10
20
30
40
50
60
70
80
90
807570656055504540353025
Born
Born
Born
Born
Born
Not:AisbasdontaoftadofousoldTloan-to-valuratioiscalculatdforUSousoldswitmortasFortdata
plottdintisfiursFiurAintappndix
Sourc:InvstmntCompanyInstituttabulationsoftSurvyofConsumrFinancs
7/30/2019 The Success of the U.S. Retirement System
28/64
26 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
HomeEquityAmongNear-RetireeHouseholds
Home equitythe value o the home less any mortgage debt owed on the propertyrepresents a large
portion o measured wealth or many U.S. households. The typical measure o household wealth is net
worth, which is the sum o the value o al l household assets (inancia l and noninancial) less household
debt. As a group, near-retiree households, deined in this paper as households headed by a working
individual aged 55 to 64,34 held 19 percent o their measured wealth in the orm o home equity in 2010.35
The median ratio o home equity to net worth or near-retiree homeowners was 32 percent (Figure 12).
In 2010, 82 percent o near-retiree households owned their homes (Figure 12). Among near-retiree
households, homeownership tends to rise with household income: 48 percent o near-retiree households
with income less than $30,000 owned their homes, compared with 86 percent o households with income
o $55,000 to $79,999 and 96 percent o near-retiree households with income o $150,000 or more.
Among near-ret iree households that owned their homes, 25 percent owned their homes with no mortgage
debt in 2010 (Figure 12). Lower-income near-retiree households are more likely to have paid o the
mortgages on their homes. Among home-owning near-retiree households with income less than $30,000,
46 percent had no mortgage debt, compared with ewer than 25 percent o other homeowners. Among all
near-retiree households with mortgage debt, the median loan-to-value ratio was 51 percent.
Although ewer lower-income near-retiree households owned their homes, or those that owned their
homes, home equity represented a much higher share o wealth. For example, the median ratio o home
equity to net worth or near-retiree home-owning households with income less than $30,000 was 58
percent, compared with 40 percent or homeowners with income o $55,000 to $79,999, and 19 percent
or homeowners with income o $150,000 or more (Figure 12).
Employer-Sponsored Retirement Plns nd IRA s
Employer-sponsored retirement plans (which include both private-sector employer and government
employer plans, as well as both DB and DC plans) and IRAs are an important part o the retirement
resource pyramid or many households. Accumulations in employer-sponsored retirement plans and IR As
provide income in retirement that supplements Social Security beneits. Consistent with their role as a
supplement to Social Security, they increase in importance among households or whom Social Security
bene its replace a smaller share o thei r average li etime earnings.
7/30/2019 The Success of the U.S. Retirement System
29/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 27
FIgURe
Homeownership and Mortgages Among Near-Retiree Households by Household Income
Ner-retiree households ,1 2010
Household income2 All ner-retiree
households
Less thn
$30,000
$30,000
$54,999
$55,000
$79,999
$80,000
$149,999
$150,000
or more
Percentage of near-retiree
households16 24 19 23 18 100
Household income (dollars)2
Avra 20,263 42,588 66,895 110,287 318,780 109,141
Mdian 21,347 42,693 67,090 108,766 219,566 67,090
Household net worth (dollars)3
Avra 161,980 297,237 322,175 682,864 3,176,997 890,664
Mdian 15,100 138,100 181,400 382,800 1,949,100 221,300
Housing
homownrsip rat (prcnt) 48 80 86 94 96 82
Avra om quity (dollars)4 39,776 85,810 89,261 154,281 495,615 168,950
Mdian om quity (dollars)4 0 56,000 75,000 99,000 323,000 79,000
Selected statistics for home-owning households
Avra ous valu (dollars) 135,834 174,815 193,554 280,906 729,339 320,853
Mdian ous valu (dollars) 100,000 135,000 160,000 225,000 550,000 200,000
Avra om quity (dollars)4 82,621 107,056 104,149 164,234 513,991 205,917
Mdian om quity (dollars)4 75,000 72,500 83,000 104,000 345,000 104,000
Mdian ratio of om quity to
nt wort (prcnt)3, 458 46 40 29 19 32
Incidnc of morta dbt
(prcnt)54 76 77 78 77 75
Mdian loan-to-valu ratio for all
omownrs (prcnt)18 39 39 44 28 35
Mdian loan-to-valu ratio for
omownrs wit dbt (prcnt)47 54 60 54 39 51
Nar-rtirousoldsarousoldswitaworkinadadtoxcludinttopandbottomprcntoftincom
distributionTotalisousoldincombfortaxsinhousoldntwortistdiffrncbtwnousoldrossassts(financialandnonfinancial)andliabilitishomquityistomvalulssanyoutstandinmortadbtontatom Sourc:InvstmntCompanyInstituttabulationsoftSurvyofConsumrFinancs
7/30/2019 The Success of the U.S. Retirement System
30/64
28 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
Near-RetireeHouseholdsRetirementAccumulations
Employer-sponsored retirement plans and IRAs provide retirement resources to about 80 percent o near-
retiree households (Figure 13). Employer-sponsored retirement plans and IR As act as a supplement to
Social Security. For the past two decades, about 80 percent o near-retiree households have consistently
accrued DB, DC, or both types o reti rement plan beneit (rom private-sector employer and government
employer plans), or IRAs (rollover and contributory). Despite the act that, among private-sector
employers, DC plans have grown relative to DB plans, the portion o near-retiree households with accruedbene its in employer-sponsored retirement plans or IRAs has remained about 80 percent si nce 1989. These
data suggest that the transition rom DB to DC plans has not reduced the share o the population that has
reached retirement with employment-based retirement accumulations. In 2010, 81 percent o near-retiree
households had either claims to beneits rom DB retirement plans (rom current and previous employers,
private-sector employer and government employer plans), reti rement assets (IRAs or DC account ty pe
pensions, such as 401(k) plans or si milar accounts rom current and previous employers, private-sector
employer and government employer plans), or both.
DBPlanBenefits
Workers with traditional DB plans earn beneits that are based on years o service and earn ings averaged
over a period o years (or example, the highest ive years, or the most recent ive years). Because o the
bene it ormula, traditional DB plans place a premium on hav ing both a long tenure at a single employer
and separating rom employment close to retirement age.36 Workers with cash balance plans, another
type o DB plan, earn retirement beneits that are expressed as a notional account balance. The notional
account typically i s increased each year by a percentage o the workers pay plus an increase based on
a speci ied rate o return. Workers with cash balance plans accrue retirement beneits more evenly
throughout their working careers.
Participation in DB plans never has been universal and the number o workers participating in private-
sector DB plans has declined over the past 20 years. In 2010, there were 17 million active participants
in private-sector DB plans, compared with 27 mil lion active participants i n 1989.37 Despite the act thatewer current private-sector workers are covered by DB plans, many households who are near retirement
have accrued beneits in DB pension plans, and thus DB plans wil l continue to be an important source o
retirement resources in the near uture. In 2010, 41 percent o near-retiree households had accrued DB plan
bene its (including accruals in both pr ivate-sector employer and government employer plans), compared
with 55 percent o near-retiree households who had accrued DB plan beneits i n 1989 (Figure 13).
7/30/2019 The Success of the U.S. Retirement System
31/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 29
FIgURe
Vast Majority of Near-Retiree Households Have Accrued Pension Benefits
Percentage of near-retiree households,1 19892010
Retirementassetsonly
BothDBbenefitsandretirementassets
DBbenefitsonly
Nar-rtirousoldsarousoldswitaworkinadadtointyarindicatdxcludinttopandbottomprcnt
oftincomdistribution RtirmntasstsincludDCplanassts((k)(b)triftandotrDCplans)andIRAs(traditionalRotSePSAR-SePand
SIMPLe)wtrfromprivat-sctororovrnmntmployrs DBbnfitsincludousoldscurrntlyrcivinDBbnfitsandousoldswittpromisoffuturDBbnfitswtrfrom
privat-sctororovrnmntmployrs
Not:Componntsmaynotaddtottotalbcausofroundin Sourc:InvstmntCompanyInstituttabulationsoftSurvyofConsumrFinancs
7/30/2019 The Success of the U.S. Retirement System
32/64
30 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
DCPlanBenefitsandIRAs
In a DC plan, there is no beneit ormula. The beneit is the balance in the account, which re lects
contributions made to the account and investment returns earned on the account. Contributions to DC
plans can be made by employers, workers, or both. A 401(k) plan is one type o DC plan. Workers who
participate in a 401(k) plan have the option to contribute a portion o their earnings into a retirement
account. In addition, employers may contribute to the employees retirement account, either through
automatic or matching contributions. When the worker reaches retirement age, their retirement beneitis the balance in the 401(k) account, and the worker may begin to withdraw money. 38 I a worker does
not stay with the same employer throughout their career, the DC account can be let at the old employer,
rolled over into an IRA, or shi ted to a new employer. In any case, the balance o the DC account remains
under the ownership and control o the worker.
DC pension plans have grown signi icantly, particularly 401(k) plans in the private sector, which had
$3.3 trillion in assets in mid-2012.39 In 1989, there were 17 mill ion active participants in 401(k) plans,
compared with 51 million in 2010.40 Because 401(k) plans were irst introduced in the early 1980s,41
workers who are currently near retirement will not have had the opportunity to participate in 401(k) plans
throughout their ul l careers. Employees o non-proit organizations, un iversities, and public schools
typical ly are oered another type o DC plan, a 403(b) plan; these plans held $771 billion in mid-2012.42
State and local government employees had accumulated $201 billion in 457 plans in mid-2012.43 All told,
DC plan assets totaled $4.7 trill ion in mid-2012 (Figure 4).
Created in 1974 under the Employee Retirement Income Security Act (ERISA), IRAs were designed with
two goals.44 First, they provide individuals not covered by workplace retirement plans with an opportunity
to save or retirement on their own. Second, through rollovers, they allow workers who are leaving a job
a means to preserve the tax beneits and growth opportunities that employer-sponsored retirement plans
provide. The IRA has proved successul in both roles, although contributions to IRAs decl ined ater 1986
when restrictions were placed on who could make deductible contributions to traditional IR As.45 More
recently, rolloversinclusive o rollovers rom DB and DC plans; private-sector employer and governmentemployer planshave been the primary source o unds lowing into IRAs.46 As o the end o June 2012,
IRA assets totaled $5.1 tril lion (Figure 4).47
The growth o DC plans and IRAs also can be seen on household balance sheets. In 2010, 71 percent o
near-retiree households had DC plan accounts or IRAs, compared with 60 percent in 1989 (Figure 13).
Among households with DC plan accounts or IRAs, the median amount o assets in these accounts (in
constant 2010 dollars) has increased to $101,350 in 2010, compared with $40,479 in 1989 (Figure 5) .
7/30/2019 The Success of the U.S. Retirement System
33/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 31
Near-RetireeHouseholdsRetirementAccumulationsbyHouseholdIncome
The portion o near-retiree households that have accrued beneits in employer-sponsored retirement plans
or IRAs is high across all i ncome groups. In 2010, 94 percent o near-retiree households with income o
$55,000 or more had DB plan beneits, DC plan accounts, or IRAs (Figure 14). More than 70 percent o
near-retiree households with income o $30,000 to $54,999 had such employer-sponsored retirement plan
bene its or IRAs. Among near-ret iree households with income less than $30,000, almost hal had such
retirement plan beneits or IRAs.
Employer-sponsored retirement plans and IRAs act as a supplement to Social Security. The ownership
pattern o employment-sponsored retirement beneits mirrors (or complements) the progressive Social
Security bene it. Social Security provides higher replacement rates or workers with lower lietime
earnings. Employer-sponsored retirement plans supply retirement beneits at al l income levels, but higher-
income households are more l ikely to accrue beneits in employer-provided retirement plans.
FIgURe
Near-Retiree Households Across All Income Groups Have Retirement Assets or DB Benefits or Both
Percentge of ner-retiree households1 by income group,2 2010
Retirementassetsonly
BothDBbenefitsandretirementassets
DBbenefitsonly
AllHigher
ormore
Upper-Middle
to
Middle
to
Lower-Middle
to
Lower
Lessthan
Householdincomegroup
Nar-rtirousoldsarousoldswitaworkinadadtoxcludinttopandbottomprcntoftincom
distributionTotalisousoldincombfortaxsin RtirmntasstsincludDCplanassts((k)(b)triftandotrDCplans)andIRAs(traditionalRotSePSAR-SePand
SIMPLe)wtrfromprivat-sctororovrnmntmployrsDBbnfitsincludousoldscurrntlyrcivinDBbnfitsandousoldswittpromisoffuturDBbnfitswtrfrom
privat-sctororovrnmntmployrs
Not:Componntsmaynotaddtottotalbcausofroundin Sourc:InvstmntCompanyInstituttabulationsoftSurvyofConsumrFinancs
7/30/2019 The Success of the U.S. Retirement System
34/64
32 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
Near-retiree households across a ll income groups part icipate in employer-sponsored retirement plans and
IRAs. As household income increases so too does the amount o retirement assets in DC plans and IRAs, as
well as the share o near-retiree households with DB beneits (Figure 15). The median ratio o DC account
and IRA assets to net worth var ies across income groups. For example, in 2010, near-retiree households
with income less than $30,000 that owned retirement assets had median reti rement assets o $18,000. One-
quarter o those households also had DB pension beneits. In comparison, near-retiree households with
income o $150,000 or more that owned retirement assets in DC accounts and IRAs had median retirementassets o $423,000, and almost hal o those households expected beneits rom DB pension plans. However,
the accumulation in DC accounts o near-retiree households today is not an indication o what uture
generations o workers wil l accumulate when exposed to DC plans or a ull working career.48 In 2010,
among near-retiree households with reti rement assets, 43 percent also had beneits rom DB plans.
As a share o net worth, retirement assets are most important to moderate-income households. In 2010, the
median ratio o retirement assets (DC accounts and IRAs) to net worth was 23 percent or households with
income less than $30,000 that owned reti rement assets (Figure 15). The median ratio o retirement assets
to net worth rises with income, reaching 40 percent or households with income between $80,000 and
$149,999 that owned retirement assets. Although the amount o retirement assets continues to increase
with income, these assets represent a smaller share o net worth or households with t he highest incomes.Among households with income o $150,000 or more that ow ned retirement assets, the ratio o retirement
assets to net worth was 24 percent.
Other Assets
In addition to Social Security, homeownership, employer-sponsored DB and DC plans, and IRAs, some
households own other assets upon which they can draw in reti rement. Other assets include inancial
and noninancial assets, such as nonresidential property, business equity, and stocks owned outside o
employer-sponsored retirement plans or IRAs.
Ownership o other assets is not evenly distributed across households. Bricker et al. (2012) ind thathouseholds with greater wealth and higher incomes are more likely to own other assets. For example,
stock ownership outside retirement accounts rises with wealth. In 2010, the SCF shows that 3 percent o
households with net worth in the bottom 25 percent o the population owned stocks outside o retirement
accounts, compared with 55 percent o households with net worth in the top 10 percent o the population.
Similarly, the 2010 SCF data show that 5 percent o households in the bottom 20 percent o the income
distribution owned business equity, compared with 38 percent o households in the top 10 percent o the
income distr ibution.
7/30/2019 The Success of the U.S. Retirement System
35/64
TheSUCCeSSOFTheUSReTIReMeNTSYSTeM 33
FIgURe
Retirement Plans and IRAs Among Near-Retiree Households by Household Income
Ner-retiree households ,1 2010
Household income2All ner-
retiree
households
Less thn
$30,000
$30,000
$54,999
$55,000
$79,999
$80,000
$149,999
$150,000
or more
Percentage of near-retiree
households16 24 19 23 18 100
Household income (dollars)2
Avra 20,263 42,588 66,895 110,287 318,780 109,141
Mdian 21,347 42,693 67,090 108,766 219,566 67,090
Household net worth (dollars)3
Avra 161,980 297,237 322,175 682,864 3,176,997 890,664
Mdian 15,100 138,100 181,400 382,800 1,949,100 221,300
Accrued pension benefits
Som typ of accrud pnsion
bnfit (prcnt)48 71 89 95 96 81
Rtirmnt assts only4 28 38 43 43 48 40
Bot DB bnfits and rtirmnt
assts4, 59 20 32 45 45 31
DB bnfits only5 11 12 14 7 3 10
Avra rtirmnt assts (dollars)4 17,431 51,048 83,780 206,002 654,545 196,735
Mdian rtirmnt assts (dollars)4 0 4,300 28,000 130,000 400,000 34,000
Selected statistics for households with retirement assets
Avra rtirmnt assts (dollars)4 47,224 87,741 112,341 233,042 704,385 277,064
Mdian rtirmnt assts (dollars)4 18,000 36,000 57,000 150,000 423,000 101,350
Sar tat also av DB bnfits
(prcnt)525 35 43 51 48 43
Mdian ratio of rtirmnt assts to
nt wort (prcnt)3, 423 27 32 40 24 29
Nar-rtirousoldsarousoldswitaworkinadadtoxcludinttopandbottomprcntoftincom
distributionTotalisousoldincombfortaxsinhousoldntwortistdiffrncbtwnousoldrossassts(financialandnonfinancial)andliabilitis
RtirmntasstsincludDCplanassts((k)(b)triftandotrDCplans)andIRAs(traditionalRotSePSAR-SePandSIMPLe)wtrfromprivat-sctororovrnmntmployrs
DBbnfitsincludousoldscurrntlyrcivinDBbnfitsandousoldswittpromisoffuturDBbnfitswtrfrom
privat-sctororovrnmntmployrs Sourc:InvstmntCompanyInstituttabulationsoftSurvyofConsumrFinancs
7/30/2019 The Success of the U.S. Retirement System
36/64
34 TheSUCCeSSOFTheUSReTIReMeNTSYSTeM
Estimting the Components of the Retirement Resource Pyrmid
The U.S. retirement system generally is successul at providing reti rement resources to older Americans.
The system consists o Socia l Security, which provides a wide base o resources to households;
homeownership; DB and DC employer-sponsored retirement plans and IR As, which complement Social
Security; and other assets. Households do not rely on each part o the retirement system equally in order
to maintain li estyles as they transition rom working lie to reti red lie. Some components o Americas
retirement system, such as Social Security, are designed to provide greater support to people with lower
lietime income. This means that Americans with higher levels o lietime income are more reliant on
employer-sponsored retirement plans, IRAs, homeownership, and other assets in retirement.
It is possible to estimate the retirement resource pyramid or U.S. households, but doing so requires
measuring the value o a households uture stream o Social Security and DB plan beneits. Gustman,
Steinmeier, and Tabatabai (2009) undertake this exercise using HRS data rom 2006. The analysis ocuses
on households approaching reti rementin this case, households with a member born between 1948 and
1953 (aged 53 to 58 in 2006). Their analysis is used to estimate the components o the retirement resource
pyramid or these households, with households grouped by their augmented wealth (Figure 16).
Re lecting the progress ive beneit ormula, households approaching retirement in the lowest augmented
wealth quintile (the lowest 20 percent o households approaching retirement ranked by augmented wealth)
rely heavily on Social Security beneits. In 2006, Social Security comprised 82 percent o total augmented
wealth or households approaching retirement who were in the lowest augmented wealth quintile
(Figure 16). Although Social Security typical ly replaces a high percentage o earnings or these households,
many also had equity in their homes, accumulated retirement beneits, and other assets.
In comparison with those with lower augmented wealth, households approaching reti rement in the middle
o the augmented wealth distribution rely more heavily on resources other than Social Security. Social
Security comprised a large portion o total augmented wealth (