do households have a good sense of their retirement preparedness_center for retirement research...

7/29/2019 Do Households Have a Good Sense of Their Retirement Preparedness_Center for Retirement Research REPORT

1/9

August 2008, Number 8-11

DO HOUSEHOLDS HAVE A GOOD SENSE

OF THEIR RETIREMENT PREPAREDNESS?

* Alicia H. Munnell is the Director o the Center or Retirement Research at Boston College (CRR) and the Peter F. DruckerProessor o Management Sciences at Boston Colleges Carroll School o Management. Francesca Golub-Sass is a researchassociate at the CRR. Mauricio Soto and Anthony Webb are research economists at the CRR. The authors would like tothank Gary Burtless and Robert Clark or helpul comments. The Center grateully acknowledges Nationwide Mutual In-surance Company or its exclusive nancial support o the National Retirement Risk Index (NRRI). This brieprovides thelatest analysis using the NRRI; prior NRRI publications are available at http://crr.bc.edu/special_projects/national_retire-ment_risk_index.html.

Introduction

The National Retirement Risk Index (NRRI) mea-sures the percentage o working-age households whoare at risk o being nancially unprepared or retire-ment today and in coming decades. The calculationsshow that even i households work to age 65 and an-nuitize all their nancial assets, including the receiptsrom reverse mortgages on their homes, 44 percentwill be at risk o being unable to maintain theirstandard o living in retirement. An extension o theanalysis to account explicitly or health care costs inretirement raises the share o at risk householdsrom 44 percent to 61 percent.

This brieexamines whether households have agood sense o their own retirement preparedness do their retirement expectations match the reality thatthey ace? Do people at risk know that they are atrisk? The rst section summarizes the NRRI andcompares households sel-assessed preparedness tothe objective measure provided by the NRRI. Thesecond section describes the characteristics o house-holds associated with being too optimistic or too

pessimistic. The last section o this brieintroduceshealth care costs into the analysis.

The NRRI

To quantiy the eects o the changing landscape,the National Retirement Risk Index (NRRI) providesa measure o the percent o working-age Americanhouseholds who are at risk o being nanciallyunprepared or retirement. The Index calculatesor each household in the 2004 Survey o ConsumerFinances a replacement rate projected retirementincome as a percent o pre-retirement earnings and compares that replacement rate with a target re-placement rate derived rom a lie-cycle consumptionsmoothing model. Those who ail to come within 10percent o the target are dened as at risk, and theIndex reports the percent o households at risk.

The results as updated to 2006 show that 44percent will be at risk o being unable to maintaintheir standard o living in retirement. An analysis byage group indicates that the situation gets more seri-ous over time (see Table 1 on the next page). About35 percent o the Early Boomers (those born between1948 and 1954) will not have an adequate retirementincome. This share increases to 44 percent or the

Late Boomers (those born between 1955 and 1964),and then rises to 48 percent or the Generation Xers(those born between 1965 and 1974).1

By Alicia H. Munnell, Francesca Golub-Sass, Mauricio Soto, and Anthony Webb*
http://crr.bc.edu/special_projects/national_retirement_risk_index.htmlhttp://crr.bc.edu/special_projects/national_retirement_risk_index.htmlhttp://crr.bc.edu/special_projects/national_retirement_risk_index.htmlhttp://crr.bc.edu/special_projects/national_retirement_risk_index.html


2/9

Households Sel-Assessment

o Retirement PreparednessThe Survey o Consumer Finances (SCF), which is thesurvey used to construct the NRRI, asks each house-hold to rate the adequacy o their anticipated com-bined income rom Social Security and pensions ona spectrum rom one to ve, with one being totallyinadequate, three being enough to maintain livingstandards and ve being very satisactory.5 Thus,by NRRI standards, any household that answers oneor two considers itsel to be at risk.

Table 3 shows the results on an aggregate basis.Forty-eight percent o the households in our sample,in responding to this SCF question, consider them-selves to be at risk o not having enough to maintaintheir standard o living in retirement slightly abovethe NRRI calculation o 44 percent at risk. The our-percentage-point dierence might be attributed to theact that, in the NRRI, households are not considered

Center or Retirement Research2

This pattern o increasing risk refects the chang-ing retirement landscape.2 The length o retirementis increasing as the average retirement age hovers at63 and lie expectancy continues to rise. At the sametime, replacement rates are alling or a number oreasons. First, at any given retirement age, SocialSecurity benets will replace a smaller raction opre-retirement earnings as the Full Retirement Agerises rom 65 to 67. Second, while the share o theworkorce covered by a pension has not changed overthe last quarter o a century, the type o coverage hasshited rom dened benet plans to 401(k) plans. Intheory 401(k) plans could provide adequate retire-ment income. But individuals make mistakes at every

step along the way and the median balance or house-hold heads approaching retirement is only $60,000.3Finally, most o the working-age population savesvirtually nothing outside o their employer-sponsoredpension plan.

These estimates, however, do not explicitly ac-count or health care expenses in retirement. In-cluding health care raises the percent o householdsat risk that is, not capable o maintaining theirpre-retirement standard o non-health care consump-tion rom 44 percent to 61 percent. Becausehealth care costs are rising rapidly and the retirementincome system is contracting, a much larger percent

o later cohorts will be at risk than earlier ones. TheNRRI rises rom 50 percent or Early Boomers to 68percent or Generation Xers. The pattern also variesby income class, with a much larger share o thoseat risk in the bottom third than in the top third (seeTable 2).4

Table 1. Percent of Households At Risk by BirthCohort and Income Group, 2006

Source: Munnell, Golub-Sass, and Webb (2007).

Income group

All 44 35 44 48

Top third 37 33 36 41

Middle third 41 28 44 46

Bottom third 54 45 54 60

% %%%

EarlyBoomers

1948-1954

GenerationXers

1965-1974

LateBoomers

1955-1964All

Table 2. Percent of Households At Risk by BirthCohort and Income Group, Including HealthCare Expenses, 2006

Source: Munnell et al. (2008).

Income group

All 61 50 61 68

Top third 53 48 52 59

Middle third 57 44 57 67

Bottom third 72 58 74 80

% %%%

EarlyBoomers

1948-1954

GenerationXers

1965-1974

LateBoomers

1955-1964All

Table 3. Percent of Households At Risk byIncome Group, 2006

Sources: Authors calculations based on Center or Retire-ment Research at Boston College (2006) and U.S. Boardo Governors o the Federal Reserve System, Survey oConsumer Finances, 2004.

Income group

All 48 44

Top third 41 37

Middle third 46 41

Bottom third 58 54

% %

Sel-assessment NRRI


3/9

Issue in Brie 3

at risk i their replacement rate is within 10 percento their target the replacement rate needed tomaintain their standard o living, whereas no suchcushion exists or the sel-reported responses.6

The sel-assessment o retirement prepared-ness shows the same pattern displayed by the NRRI(Tables 3 and 4). Lower income and younger house-holds are more likely to report being at risk. Thusit seems that despite substantial gaps in nancialknowledge, as shown in recent literature, in the ag-gregate, households have a good gut sense o theirnancial situation.7

Table 4. Percent of Households At Risk by BirthCohort, 2006

Sources: Authors calculations based on Center or Retire-ment Research at Boston College (2006) and 2004 SCF.

Cohort

All 48 44

Early Boomers 44 35

Late Boomers 50 44

Generation Xers 51 48

% %

Sel-assessment NRRI

port not having enough resources to maintain livingstandards and the NRRI says they are at risk, or theyreport being adequately prepared and the NRRI saysthey are not at risk. Fity-seven percent o house-holds appear to know how they will are in retire-

ment.8 Quadrant II shows households that appear tobe more concerned than needed they report beinginadequately prepared but the NRRI says that they arenot at risk. Twenty-our percent o the householdsall into this category. Quadrant III shows that only19 percent o households seem to be less worriedthan they should be. That is, they report havingenough resources to maintain living standards whenthe NRRI says they are at risk.

The question is what characteristics cause ahousehold to be too worried or not worried enoughas opposed to getting it right.9 Figures 1 and 2 reportthe results o a multinomial logit equation thatexamines the impact on putting households in onecategory or another o the ollowing variables: riskaversion, income group, birth cohort, household type,dened benet coverage, home ownership, educa-tion, and health status.10 Figure 1 shows the impact

Table 5. Households At Risk and Not At RiskAccording to the NRRI and IndividualResponses, 2006

Sources: Authors calculations based on Center or Retire-ment Research at Boston College (2006) and 2004 SCF.

At risk Not at risk

NRRI

Figure 1. Effect of Each Variable on Being in the Too Worried Group

Source: Authors calculations.

Risk averse

Middle income

Own house

Have dened benet

Married one-earner household

Generation Xer

Late Boomer

High income

College degree

Good health

3.6%

-4.3%

1.1%

1.4%-1.3%

11.4%

12.9%

-1.4%

6.0%

1.3%

-10% 0% 10% 20%

Statisticallysignicant

Not statisticallysignicant

Households Sel-Assessmentvs. the NRRI

Even i aggregate perceptions match the NRRI, it ispossible that households assessment o retirementpreparedness diers rom the NRRI on a household-by-household basis. Table 5 examines how wellhouseholds are able to perceive their retirement risk.

Quadrants I and IV show the households whosesel-assessment agrees with the NRRI they re-

Household response

At risk 25%(quadrant I)

24%(quadrant II)

Not at risk 19%(quadrant III)

32%(quadrant IV)


4/9

o each variable on the probability o ending up in thetoo worried group as opposed to getting it just right;Figure 2 shows the impact on the probability o end-ing up in the not worried enough group.

The gures tell an intuitive story.

Risk aversion . The measure o risk aversion comesrom an SCF question that asks households abouthow much nancial risk they are willing to takein order to get higher investment returns.11 I ahousehold is not willing to take any nancial risk,it is classied as risk averse. As one would expect,a risk averse household is more likely to end upas too worried, and less likely to end up as notworried enough.

Income group . Being in the highest income cat-

egory does not increase the probability o beingeither too worried or not worried enough; thesehouseholds are most likely to get it right. Middle-income households are much less likely thanlow-income households to be in the too worriedgroup, suggesting that low-income householdsare most likely to worry too much. Low-incomehouseholds are nancially pressed while workingand may be earul o retirement. However, SocialSecuritys progressive benet structure providesrelatively high replacement rates to these house-holds, so they end up in better shape than theyear.

Cohort . The younger the cohort, the more likelythe household is to be in the not worried enoughgroup. Essentially, the result says that youngercohorts are less well-inormed about the trends in

the retirement income system and less likely tooresee the potential or those trends to put themat risk.

Household type . Married one-earner householdsare signicantly more likely to be in the too wor-ried group and signicantly less likely to be in thenot worried enough group than other householdtypes. This strong nding probably refects theact that Social Security provides a spouses benetequal to 50 percent o the benet o the higherearning spouse and many couples may not beaware o it beore they claim benets. As a result,

married one-earner households end up withhigher replacement rates than other householdtypes and nd themselves in a better position thanexpected.

Have defned beneft plan . Households with adened benet plan are signicantly less likely tobe in the not worried enough category than thosewithout such a plan. These households presum-ably know that they have a guaranteed income orlie and probably are not at risk both in the NRRIcalculation and in terms o their sel-assessment.


Figure 2. Effect of Each Variable on Being in the Not Worried Enough Group

Source: Authors calculations.

Risk averse

Middle income

Own house

Have dened benet

Married one-earner household

Generation Xer

Late Boomer

High income

College degree

Good health

-0.1%

0.8%

-0.8%

1.9%

2.6%

-8.4%

-5.2%

-8.5%

-3.9%

-2.3%

-10% 0% 10% 20%

Statisticallysignicant

Not statisticallysignicant


5/9

Issue in Brie 5

Own house . Owning a house signicantly in-creases the likelihood o being in the too worriedgroup and signicantly reduces the likelihoodo being in the not worried enough group. TheNRRI shows most o these home-owning house-holds as not being at risk because it has them tak-ing a reverse mortgage on their home and annui-tizing the proceeds. On the other hand, surveysshow that households do not plan to tap homeequity to support general consumption in retire-ment, so these households with homes underesti-mate their potential well-being, and all in the tooworried category.12

College degree . Having a college degree signi-cantly increases the probability o alling into thetoo worried group and signicantly reduces the

probability o being in the not worried enoughgroup. The implication is that education increaseshouseholds time horizon and the probability thatthey think about their well-being in retirement.Predictably, this perspective means that they aremuch less likely to be caught unaware, but sur-prisingly it also suggests that they might worry toomuch.

Good health . Being in good health has a small butstatistically signicant negative eect on the prob-ability o being classied as not worried enough.Perhaps the best way to interpret this result is that

those in good health have the energy to ocus onnancial issues and are less likely to be adverselysurprised by their preparedness or retirement.

Overall, the results suggest that not only do a highpercentage o households have a good sense abouttheir preparedness or retirement, but those house-holds with incorrect perceptions do so or predictablereasons. The real danger in terms o mispercep-tions is being not worried enough. One actor thatputs households in this category is age; youngerhouseholds are simply much less aware than olderhouseholds o the challenges they will ace. But being

anything but a one-earner married couple, not havinga dened benet plan, not owning a home, and nothaving a college degree also increases the probabilityo being in the not worried enough group. These 19percent o households that do not recognize that theyare at risk are unlikely to undertake remedial action.Unortunately, it is not clear that the 25 percent thatcorrectly perceive themselves to be at risk will takeaction either.

The Role o Health Care

As noted above, the NRRI was re-estimated to explic-itly account or health care expenses in retirement.Including health care raises the percent o householdsat risk that is, not capable o maintaining theirpre-retirement standard o non-health care consump-tion rom 44 percent to 61 percent. Table 6 showshow well households are able to perceive their retire-ment risk when aced with the higher hurdle. Again,quadrants I and IV show that 57 percent o house-holds appear to know how they will are in retirement.Quadrant II shows households that are too worriedand quadrant III those not worried enough.

Table 6. Households At Risk and Not At RiskAccording to the NRRI with Health Care and

Individual Responses, 2006

Sources: Authors calculations based on Munnell, et al.(2008) and 2004 SCF.

At risk Not at risk

NRRI

Table 7 compares the classication results or

the NRRI with and without explicitly recognizinghealth care expenses. Interestingly, the percent ohouseholds that misperceive their situation remainsvirtually the same 43 to 44 percent but the com-position o that 43 to 44 percent changes dramatically.Under the original NRRI, most o those householdswith an inaccurate perception were worrying morethan necessary. Once health care is introduced into

Table 7. Original NRRI and NRRI with HealthCare by Individual Responses, 2006

Sources: Authors calculations based on Munnell, et al.(2008) and 2004 SCF.

QuadrantI(At risk/at risk)

II(Too

worried)

III(Not worried

enough)

IV(Not at risk/not at risk)

health care

% % % %

Household response

At risk33%

(quadrant I)16%

(quadrant II)

Not at risk28%

(quadrant III)24%

(quadrant IV)

NRRI

Original 25 24 19 32

With 33 16 28 24


6/9


the calculation, the majority o those who got it wrongwere not worried enough. The clear message is thathouseholds tend to underestimate the adverse impactthat health care costs will have on their standard o

living in retirement.

Conclusion

Despite recent literature indicating that householdssuer large gaps in their nancial knowledge, mosthave a good gut sense o their nancial situation.Households sel-assessments closely mirror theresults produced by the NRRI. Even on a household-by-household basis, almost 60 percent o householdssel-assessments agree with their NRRI predictions.Moreover households that get it wrong do so or

predictable reasons. Households that are risk averse,that include a one-earner married couple, and thatown their own home are more likely to worry toomuch. Being a two-earner couple or single person,not owning a home, not being covered by a denedbenet pension, and not having a college degree allincrease the probability o not worrying enough.

Explicitly recognizing health care costs and settingthe goal o maintaining non-health care consumptionraises the hurdle on being prepared and increasesthe percent o households at risk. Including health

costs does not increase the percent o householdsthat misperceive their situation; almost 60 percentcontinue to understand their situation. Among thosewho misperceive their retirement outlook, however,the majority now all into the category o not worriedenough.

Classiying households by the accuracy o theirperceptions about retirement security does not an-swer the question o whether they are likely to takeremedial action. Under any circumstance, thosehouseholds that worry too little are the least likely tochange their saving or retirement plans. This groupaccounts or 28 percent o households once health

care costs are recognized explicitly, which means thata signicant portion o the population needs to get abetter assessment o their retirement income needs.


7/9

Issue in Brie 7

Endnotes

1 This sample does not include Generation Xers bornater 1974.

2 For more detail on the changing retirement land-scape, see Center or Retirement Research at BostonCollege (2006).

3 This amount includes Individual Retirement Ac-count (IRA) balances, because most o the money inIRAs is rolled over rom 401(k) plans. For urtherdetails on 401(k) missteps, see Munnell and Sundn(2006).

4 As discussed in earlier bries, part o this pattern re-fects the act that low-income households rely almost

exclusively on Social Security benets, which arescheduled to decline sharply relative to pre-retirementearnings. But health care spending is also a powerulorce putting large numbers o low-income house-holds at risk. This impact occurs despite the actthat households in the bottom third o the incomedistribution only spend about 70 percent o whatmiddle-income households spend, partly becausesome households in this group have their premiumsand copayments covered by Medicaid.

5 The NRRI calculations take all assets into accountwhen assessing retirement preparedness and do not

solely look at Social Security and pension income.This discrepancy is not signicant because, or mosthouseholds, Social Security and pension income arethe main sources o retirement income. On average,based on NRRI data, Social Security and pension in-come account or about 75 percent o expected retire-ment income. The SCF asks the ollowing question:

Using any number rom one to ve, where oneequals totally inadequate and ve equals verysatisactory, how would you rate the retirementincome you (receive or expect to receive) rom

Social Security and job pensions? (include 401(k)accounts and all other types o pensions)1. Totally inadequate2.3. Enough to maintain living standards4.5. Very satisactory.

6 In addition, the NRRI calculations take into ac-count all resources including other nancial assetsand the proceeds rom a reverse mortgage whenassessing retirement preparedness, while the sel-

assessment question reers solely to Social Securityand pension income. As noted, however, or most

households Social Security and pension income arethe main sources o retirement income.

7 See Gustman and Steinmeier (2004) and Lusardiand Mitchell (2007).

8 The NRRI relies on sel-reported income andwealth data to determine whether households areat risk. Many studies have shown that these dataaggregate well to national averages. But an unknownpercentage o households may mis-report incomeor wealth, and the NRRI may thereore incorrectlyassign their at risk status, and thus their sense o

their retirement preparedness, while at the sametime correctly measuring the overall percentage atrisk. Another explanation or the discrepancy is thatindividual households may apply a dierent yardstickin assessing their nancial preparedness than the oneembodied in the NRRI.

9 A potential concern is that our analysis o thisquestion may be identiying an incorrect assignmento at risk status instead o lack o awareness o thehouseholds situation. We think this situation isunlikely as most o the variables that we nd to besignicant are unrelated to inputs to the NRRI algo-

rithm, and thereore plausibly uncorrelated with anyclassication errors.

10 The regression uses the 2004 NRRI classication under which 43 percent (rather than 44 percent) othe households are at risk because the responsesto the retirement preparedness sel-assessment ques-tion rom the SCF were given in 2004.

11 The SCF asks the ollowing question to gaugea households level o risk aversion: Which o theollowing statements comes closest to describing theamount o nancial risk that you are willing to takewhen you save or make investments?

Take substantial nancial risks expecting to1.earn substantial returns.Take above average nancial risks expecting to2.earn above average returns.Take average nancial risks expecting to earn3.average returns.Not willing to take any nancial risks.4.

12 Munnell, Soto, and Aubry (2007).


8/9


Reerences

Center or Retirement Research at Boston College.2006. Retirements At Risk: A New National

Retirement Risk Index. Chestnut Hill, MA.

Gustman, Alan and Tom Steinmeier. 2004. WhatPeople Dont Know about Their Pensions andSocial Security. In Private Pensions and PublicPolicies, eds. William Gale, John Shoven and MarkWarshawsky, 57-125. Washington, DC: BrookingsInstitution.

Lusardi, Annamaria and Olivia S. Mitchell. 2007.Baby Boomer Retirement Security: The Roleo Planning, Financial Literacy, and Hous-ing Wealth.Journal o Monetary Economics 54:205-224.

Munnell, Alicia H., Francesca Golub-Sass, andAnthony Webb. 2007. What Moves the NationalRetirement Risk Index? A Look Back and an Up-date. Issue in Brie7-1. Chestnut Hill, MA: Centeror Retirement Research at Boston College.

Munnell, Alicia H., Mauricio Soto, and Jean-PierreAubry. 2007. Do People Plan to Tap Their HomeEquity in Retirement? Issue in Brie7-7. Chest-nut Hill, MA: Center or Retirement Research at

Boston College.

Munnell, Alicia H., Mauricio Soto, Anthony Webb,Francesca Golub-Sass, and Dan Muldoon. 2008.Health Care Costs Drive Up the National Retire-ment Risk Index. Issue in Brie8-3. Chestnut Hill,MA: Center or Retirement Research at BostonCollege.

Munnell, Alicia H. and Annika Sundn. 2006.401(k) Plans Are Still Coming Up Short. Issue inBrie43. Chestnut Hill, MA: Center or RetirementResearch at Boston College.

U.S. Board o Governors o the Federal Reserve Sys-tem. Survey o Consumer Finances, 2004. Washing-ton, DC: U.S. Government Printing Oce.


9/9

About the CenterThe Center or Retirement Research at Boston Col-lege was established in 1998 through a grant rom theSocial Security Administration. The Centers missionis to produce rst-class research and orge a stronglink between the academic community and decisionmakers in the public and private sectors around anissue o critical importance to the nations uture.To achieve this mission, the Center sponsors a widevariety o research projects, transmits new ndings to

a broad audience, trains new scholars, and broadensaccess to valuable data sources. Since its inception,the Center has established a reputation as an authori-tative source o inormation on all major aspects othe retirement income debate.

Aliated InstitutionsAmerican Enterprise InstituteThe Brookings InstitutionMassachusetts Institute o TechnologySyracuse UniversityUrban Institute

Contact InormationCenter or Retirement ResearchBoston CollegeHovey House140 Commonwealth AvenueChestnut Hill, MA 02467-3808Phone: (617) 552-1762Fax: (617) 552-0191E-mail: [email protected]: http://www.bc.edu/crr

2008, by Trustees o Boston College, Center or Retire-ment Research. All rights reserved. Short sections o text,not to exceed two paragraphs, may be quoted without ex-plicit permission provided that the authors are identied andull credit, including copyright notice, is given to Trustees oBoston College, Center or Retirement Research.

The research reported herein was pursuant to a grant romNationwide Mutual Insurance Company. The ndings andconclusions expressed are solely those o the authors anddo not necessarily refect the views o Nationwide MutualInsurance Company or the Center or Retirement Research atBoston College.

do households have a good sense of their retirement preparedness_center for retirement research...

Documents