The Sufficiency of Retirement Savings: A Comparison of Two Cohorts of Retired Workers at the Time of Retirement

1
The Sufficiency of Retirement Savings A
Comparison of Two Cohorts of Retired Workers at
the Time of Retirement

• Robert Haveman, Karen Holden, Barbara Wolfe, and
  Andrei RomanovUniversity of WisconsinMadison
• Presentation at APPAM, November 2nd, 2006

2
Goal of Paper

• Assess the adequacy of resources at the point of
  retirement for two cohorts of retirees
• Early 1980s
• Mid 1990s
• Compare adequacy and sources of change in
  adequacy
• Use a socially defined concept of adequacy
• Retirement is defined as the date of take-up of
  Social Security Retirement Benefits
• Only those eligible for SS are included in
  sample.

3
Related Literature

• Will those near retirement have adequate
  resources?
• Bernheim 1992 simulation model of optimal
  savings over life cycle.
• Moore and Mitchell 1997, 2000- simulation model
  of needed savings assuming retirement at 62 and
  65
• Gustman and Steinmeier 1998 calculate
  annuitized wealth at date of expected retirement
• Engle, Gale and Uccelo 1999 stochastic
  life-cycle model to geenrate optimal wealth which
  is then compared to actual wealth.
• Wolff 2002 expected retirement income based of
  annuitized wealth using CSF.
• Butrica, Iams and Smith 2003 Simulation model
  using Social Security's MINT model

4
Related literature continued

• How adequate are savings at actual retirement and
  how adequate are they as retirement continues?
• Munnell and Soto 2005 important role for
  pensions
• Cole and Milligan 2005- changes in wealth
  portfolios
• Johnson, Mermin and Uccello 2005- changes in
  wealth due to shocks
• Haveman, Holden, Wolfe and Sherlund 2006
  adequacy at retirement for cohort retired in
  early 1980s.
• Haveman, Holden, Wolfe and Romanov 2005 --Change
  in adequacy over the first decade of retirement
  for cohort retired in early 1980s.

5
DATA-1

• New Beneficiary Survey (NBS)
• Our sample Retired-Worker Social Security
  Beneficiaries (62-72)
• Applied for/received benefits 1980-1981
• Interviewed in 1982
• Follow-up interview in 1991
• Large sample (Number of observations 5,579)
• Full linkage to Social Security Records
• Covered Earnings Records and Benefits Records
• Comparable data on both spouses including record
  linkage

6
Data- 2

• Health and Retirement Study
• Our sample original sample first interviewed in
  1992 and every two years thereafter and Children
  of the Depression Age first interviewed in 1998.
• Minimum age 62 at time of retirement maximum
  age 72
• Receive Social Security by 1998
• Plan to do a follow-up study of adequacy over
  retirement
• Adjust age profile to mimic that of the NBS in
  analysis of distribution
• Linkage to Social Security Records
• Covered Earnings Records and Benefits Records
• Where individual did not grant permission, use
  self-reported Social Security information.

7
Comparison of Characteristics of Samples from NBS
and HRS (age-standardized)
NBS Sample NBS Sample NBS Sample HRS sample HRS sample HRS sample
Variable Means Married Couples Single Men Single Women Married Couples Single Men Single Women
Number of observations 5783 702 1381 1447 196 309
Age in 1982 65.6 65.9 66.5 64.6 66.2 65.1
Male 0.7 0.6
Nonwhite 0.1 0.2 0.2 0.1 0.2 0.2
Widowed 0.3 0.5 0.3 0.4
Separated or divorced 0.4 0.3 0.5 0.4
Respondent high school 0.3 0.2 0.3 0.3 0.3 0.3
Respondent some college 0.1 0.1 0.2 0.2 0.2 0.2
Respondent college or higher 0.1 0.1 0.1 0.2 0.2 0.2
8
Available Resources

• Total resources of households at the age of
  retirement equal to the sum of
• financial and property wealth,
• net value of own home,
• the present discounted value of current and
  future expected pension benefits, and
• the present discounted value of current and
  future expected Social Security retirement
  benefits.

9
Two measures of resources

• Wealth present value in 2004 dollars at time of
  retirement of all resources over expected
  remaining lifetime
• Discount inflation-indexed Social Security
  benefits by 2.75 to capture individual time
  preference.
• Assume annual inflation rate 4 over lifetime
• Assume pension streams only partially adjust for
  inflation (known nominal growth 3.25 or an
  erosion rate of .75 percent) so use 3.50 to
  discount pensions.
• Incorporate value of survivor benefits and
  survivor lifetimes into wealth calculation
• Annuitized Net Wealth (ANW) annual steady income
  stream paid from wealth as calculated above
• over expected life of the respondent and spouse,
  if married
• use life expectancies by age, gender, race
• assume total wealth annuitized over lifetime of
  respondent (and surviving spouse) even if
  survivor benefits not actually taken
• Assume an equivalent scale such that couple needs
  1.66 of single person

10
Mean Net Wealth, 2004 dollars
  All households White Nonwhite Single Married
NBS
Total net wealth 553,967 582,786 236,832 252,814 500,092
Financial/property 22.4 23.25 8.57 20.05 22.87
Housing 15.7 15.79 13.27 15.56 15.67
Pensions 14.3 14.33 14.60 15.30 14.17
Social Security 47.6 46.62 63.57 49.10 47.29
HRS
Total net wealth 890,918 962,547 402,290 459,569 1,095,624
Financial/property 42.2 43.8 16.76 37.61 43.13
Housing 12.8 12.78 13.01 15.67 12.22
Pensions 21.8 21.21 31.08 18.05 22.52
Social Security 23.2 22.24 39.15 28.67 22.14
11
Composition of average wealth at retirement NBS
mean 553,967 (all) 582,786 (whites) 236,832
(Nonwhites)HRS mean 890,918 (all) 962,547
(whites) 402,290 (Nonwhites)
12
Distribution of ANW
13
Comparison of ANW by age
14
Comparison of ANW by Race
15
Comparison of ANW by Marital Status and Sex
16
ANW by Education
17
Now turn to issue of Adequacy

• Ask question Do newly retired workers have
  sufficient ANW to enable them to escape poverty
  or near poverty during retirement?
• Use two standards
• Poverty line and Twice Poverty Line
• Employ PL suggested by NRC study of poverty.
  Single 9060 couples 12,649

18
ANW Relative to Standards (actual life
expectancy)

• NBS HRS
• ANW/Poverty line 3.52 5.40
• ANW/2 x Poverty line 1.76 2.70
• Percent not meeting
• Poverty line standard 4 8
• Twice poverty line standard 23 23

19
Those who do not meet poverty standard by
education
20
Nonwhites much more likely to not meet poverty
standard than whites.

• Percent not meet poverty standard by race
• White Nonwhite All
• NBS 3 16 4
• HRS 6 25 8

21
Correlates of Retirement Resource Adequacy (see
Table 4 regressions)

• More likely to have adequate resources if
• White vs. nonwhite
• More educated (if married, have more educated
  spouse)
• Retired at later age (except single men in HRS)
• If respondent does not have a health condition (
  if married, spouse does not have a health
  condition)
• If have private health insurance
• If have pension
• If own home

22
More likely to fail to meet poverty standard of
adequacy if (see Table 5 probits)

• Nonwhite
• Did not graduate high school
• Have more children if a married couple
• Longest job uncovered by Social Security in NBS
• A Never married single woman
• Respondent had a health condition

23
Conclusions

• Using the poverty line, 4 of the NBS cohort did
  not meet this adequacy standard but for the later
  cohort, this had percentage had nearly doubled.
  This suggests increasing inequality.
• Using the twice poverty line or near poverty
  standard the percentage failing to meet the
  standard was stable but high at more than 20
  percent.
• Those failing to meet the standards are
  concentrated among women, singles, nonwhites,
  those with low education and those who retire at
  early ages.
• Vulnerability in working years tied to
  vulnerability in retirement.
• The situation at retirement did not improve for
  these vulnerable individuals comparing 1980s to
  1990s and may indeed have gotten worse.
• Policy makers would do well to focus on how any
  change in Social Security would influence persons
  in these vulnerable groups.