Retirement Economics and Aging

1
Retirement Economics and Aging

• Robert J. Willis
• University of Michigan
• Conference on Neuroeconomics and Aging
• Stanford, March 31-April 1, 2006

2
Outline (30 minute talk)

• HRS Project briefly described
• Basic design and goal
• New content on biomarkers, performance measures,
  cognition
• International pandemic
• New Directions Cognition and Biomarkers
• ADAMS project and cognitive aging
• Potential for linkage with imaging studies
• P01 Projectlinking survey methods, psychology,
  economics
• Role of cognition/emotion in complex
  decisionmaking
• Pensions/Savings
• Portfolio Choice
• Survival Expectations
• Poverty Issues
• Application to Medicare Prescription Drug Program

3
The Health and Retirement Study
http//www.umich.edu/hrswww
4
Origins of the HRS

• 1985-90
• National Institute on Aging determined there was
  lack of data to study policy or scientific issues
  relevant to aging society.
• 1990
• NIA decided to invest in new longitudinal study
  of retirement
• 1990-92
• Institute for Social Research at UM won
  competition for HRS
• Interdisciplinary research team from entire U.S.
  designed HRS, launched in 1992

5
Evolution of HRS

• 1992 Original HRS cohort
• Persons age 51-61 in 1992 (born 1931-41)
• Plus spouses of age eligible persons
• N12,654 at baseline
• Longitudinal follow-up every two years
• 1993 Original AHEAD cohort
• Persons age 70 in 1993 (born 1890-1923)
• Plus spouses of age eligible persons
• N8223 at baseline
• Longitudinal follow-up every two years

6
HRS Steady State Design1998-2010 and Beyond

• 1998 Re-design
• Merged HRS/AHEAD into single study
• Add New Cohorts
• Children of Depression entering their 70s (b.
  1924-30)
• War Babies entering 50s (born 1942-47)
• Add New 6 Year Cohort Each 6 years
• 2004 Add Early Boomers (born 1948-53)
• 2010 Add Mid-Boomers (born 1954-59)

7
The HRS is a multidisciplinary effortthat began
with initial design in 1990
A large and diverse community of researchers
actively participated in the study design and
implementation

• Economics
• Sociology
• Demography

• Medicine
• Public Health
• Psychology

8
HRS Designed to Understand Decisions, Choices,
and Behaviors in Response to Policies

• Study families rather than individuals
• Gather integrated multidisciplinary information
  about all aspects of life
• Follow people over time as events happen and
  their choices get made
• Make the data available to researchers and
  policy-makers as quickly as possible
• Let the full power and creativity of Americas
  scientific community address the challenges of an
  aging population

9
HRS Studies All Cohorts Born between 1900 and 1953
10
The HRS Data System Scale of Measurement during
2000-05

• 59,718 interviews
• 20,129 in HRS-2004 Core
• 4,222 interviews with next of kin of respondents
  who died
• 32,138 mail surveys
• 21,228 Social Security linkages
• 18,688 Medicare linkages
• 1,430 in-home neuropsychological assessments on
  850 ADAMS respondents

11
HRS Longitudinal Sample Design
12
Scientific Productivity of HRSHRS Publications,
1993-2005 Cumulative Count by Type
Cumulative Count June 15, 2005 Journal
Publications 510 Books and Book
Chapters 83 Dissertations
59 Working Papers
326 Total
978 On average, one new journal article using
HRS is published every 5 days

13
Comparable International Surveys in 17 Countries

• ELSA (English Longitudinal Study of Ageing)
• Two waves completed (2002, 2004)
• SHARE (Survey of Health, Ageing and Retirement in
  Europe) supported by EU
• 11 Countries Completed first wave in 2004
• Belgium, Denmark, France, Germany, Greece, Italy,
  Netherlands, Spain, Sweden, Switzerland
• 4 Countries will begin in 2006
• Czech Republic, Israel, Ireland, Poland
• MHAS (Mexican Health and Aging Study)
• Two waves completed (2001, 2003)
• KLoSA (Korean Longitudinal Study of Aging)
• Japan and China may be next

14
New Directions for HRS 1. Biomarkers and
Physical Performance

• New measures
• Biomarkers
• Dried blood spots
• Collect and store genetic material
• Physical Performance Measures
• Measure grip strength, timed walk, lung capacity
• Measure blood pressure
• Collected in Enhanced Face-to-Face Interviews
  for half of sample each wave
• Each person measured once every 4 years

15
Selecting Biomarkers for HRS

• Multi-dimensional predictors of disease and
  dysfunction
• Genes C-reactive protein
• Major predictors of disease and death among
  elders
• Blood pressure cholesterol, HbA1c
• Candidates for assessing stress, and reflecting
  mental illness and some behaviors
• genes

16
New Directions for HRS 2. Measurement of
Cognition and Psychosocial Variables
17
Growing Complexity of Decisions Facing Older
Americans

• Examples
• Shift from defined benefit to defined
  contribution pension plans
• Contribution rates how much to save
• Portfolio choice how much risk to take
• Post-retirement Income Annuitize or not?
• Introduction of Medicare Prescription Drug Plan
• Choose among 40 or so alternative plans which is
  best?

18
ADAMS Study of Dementia

• ADAMS has established baseline measures of
  dementia and CIND (cognitive impairment, not
  demented) in substratified subsample of 850 HRS
  respondents
• In-home clinical assessment
• Plan to use longitudinal data to measure
  transitions into dementia and CIND
• Longitudinal data in HRS will allow measurement
  of costs and burdens of disease to individual,
  family and society

Langa,K, et. al., The Aging, Demographics and
Memory Study Study Design and Methods
Neuroepidemiology, vol. 25, pp. 181-191, 2005
19
Assessing and Improving Cognitive Measurements
in the HRS (AG-07407-14) John J. McArdle, USC

• Experimental measures in HRS
• Adaptive number series test in 2004
  (Woodcock-Johnson)
• repeat number series, add vocabulary,
  prospective memory
• How Do HRS Cognitive Measures Relate to
  Woodcock-Johnsnon?
• Use independent sample of 1000 people age 57
  from HRS screening survey in two wave
  longitudinal survey
• Administer HRS cognitive items, full
  Woodcock-Johnson (3hrs)
• Use results to re-engineer HRS cognitive
  measures, beginning in 2008

20
Behavior on Surveys and in the Economy Using
HRS (P01 AG02657) R. Willis, PI

• Goal to link psychology, survey methods,
  econometrics and economic theory
• Three Projects
• Probabilistic Thinking and Economic Behavior
  Willis, PI, Hurd, Manski
• Well Being and Utility in Psychology and
  Economics Kimball, PI Schwarz, Willis
• Explaining Household Portfolio Choice in Saving
  for Retirement (T. Shumay, PI Miles Kimball,
  Matthew Shapiro
• Data Innovation Core
• Networking Core

21
Psychosocial Leave Behind

• Psychosocial
• ELSA measures administered in HRS-2004 as
  leave-behind self-administered questionnaire
• Design new leave-behind SAQ based on advice from
  planned conference
• Vignettes
• Work-relate vignettes in 2004
• Health vignettes similar to SHARE

22
Retirement Behavior In International Perspective?
23
Retirement Patterns of the Men in the Original
HRS Cohort 1992-2002
By age 70 just under 80 percent were fully
retired and another 6-7 percent were partly
retired The process of retirement begins in the
early 50s, accelerates about 60, reaches a
sharp spike at age 62 and another sub-spike at
age 65
Source Gustman and Steinmeier (2004)
24
What Will Happen to Retirement for the Early
Boomers?

• Long term trend in U.S. from 1850 to 1990 has
  been toward earlier retirement.

25
International Trends Toward Earlier Retirement

• Trend toward lower labor force participation at
  older ages is much sharper in a number of
  European countries than in US

Source J. Gruber and D. Wise, Social Security
Programs and Retirement Around the World, U.
Chicago Press, 1999.
26
Retirement Policy Shapes Retirement Behavior
70
Belgium
60
France
Italy
Holland
UK
50
Germany
Percent Early Retirement
Spain
Canada
40
US
Sweden
30
20
40
60
80
100
Percent Penalty for Continued Work
Source J. Gruber and D. Wise, Social Security
and Retirement Around the World (NBER, 2000)
27
Downward Trend in Labor Force Participation in
U.S. Began to Reverse Itself around 1985. Is
the Reversal Permanent?
Males
Females
Source Technical Panel for the Social Security
Administration, 2003, based on research by Joseph
Quinn, Boston College.
28
What Do the Early Boomers Expect?

• HRS has pioneered asking questions about
  expectations on a wide variety of topics
  including survival to a given age, bequests,
  stock market returns and work and retirement
  expectations
• Researchers have used the work expectation
  questions in earlier waves and have found them to
  be useful predictors of actual work and
  retirement in later waves
• We can now obtain a look at what the Early
  Boomers, aged 51-56 in 2004 say they will be
  doing at age 62, 65 and their age at retirement
  as compared to earlier cohorts at the same age.
• Note that these questions offer a window into the
  future about ten years hence!

29
Signs of Reversal of Long Term Trend toward Early
Retirement

• Historical trend toward earlier
• Retirement from 1850s
• through mid-1980s
• Labor force participation
• at older ages flat since then
• Possible reversal
• Early boomer cohort in HRS
• (born 1948-53) expects to
• work more than earlier
• Cohorts

Source Health and Retirement Study
30
Are Retirement Savings Adequate?
31
Distribution of Wealth HRS1992
Source Scholz, Seshadri and Khitatrakun Are
Americans Saving Optimally JPE (forthcoming)
32
Predicted Wealth Based on Dynamic Optimization
Model vs. Actual Wealth
Optimal Net Worth
Most people have more net worth than optimal!
Observed Net Worth
SourceScholz, Seshadri and Khitatrakun
(forthcoming)
33
Inequality in Health, Wealth and Cognition
34
Mean Net Worth by Health of Husband and Wife

Strong Correlation Between Health and Wealth

• Understanding this correlation requires
  longitudinal data from
• Economics
• Health
• Sociology/Demography
• Psychology

Source Robert J. Willis, Theory confronts
data how the HRS is shaped by the economics of
aging and how the economics of aging will be
shaped by the HRS, Labour Economics, Volume 6,
No. 2, page 119, June, 1999.
35
Economic Consequences of a Husbands Death
Source P. Sevak, D. Weir and R. Willis, The
Economic Consequences of a Husbands
Death Evidence from the HRS and AHEAD. Social
Security Bulletin (2003/2004).
36

• Probability of Poverty Highest
• for those who are widowed
• earliest
• Poverty probability increases
• with age/duration of widowhood
• Underlying causation could be
• from various sources e.g.,
• poor selected into widowhood
• early death of husband is
• negative shock to income,
• saving

Source P. Sevak, D. Weir and R. Willis, The
Economic Consequences of a Husbands
Death Evidence from the HRS and AHEAD. Social
Security Bulletin (2003/2004).
37
Monitoring a Social Experiment The Medicare
Prescription Drug Program, Part D

• CMS (Center for Medicare and Medicare
  Administration) began
• publicizing new program in Fall 2005.
• Enrollment began Jan. 15, 2006, ends May 15,
  2006. Substantial lifetime penalty if plan not
  chosen before May 15
• Potentially Highly Complex Choices
• Each enrollee has choice of about 40 different
  plan providers from
• own state, each with three tiers of coverage and
  level of premium.
• Coverage has doughnut hole 2250 and 3600 of
  out-of-pocket cost
• in which there is no coverage.
• Formulary of plan may not match persons needs

38
HRS Prescription Drug Survey

• Mail Survey, mid-October to mid-December 2005
• Collect detailed med data, prior to
  implementation of Part D
• Link meds to external databases to get prices
• Assess state of knowledge, attitudes, and
  intentions about Part D
• Core Survey for HRS-2006
• Determine enrollment, premiums
• Name of plan, information used in choosing
• Confidence about making right decision

39
HRS Prescription Drug Survey (cont.)

• HRS/RAND Internet Survey, Phase 2
• Design meeting 3/20/06 sponsored by Willis P01
  and HRS/RAND Internet project to design
  questionnaire module
• Will go in field following core interview around
  November, 2006
• Can look choices during new enrollment period
• Can study help given by Rs lt65 to their parents
• Could do experiments
• HRS Mail Survey, 2007
• New Roster of Prescription Drugs to detect changes

40
Early Results from PDS-2005 Mail Survey

• Based on David Weir, The Unbalanced
  Distribution of Cognitive Demands and Cognitive
  Resources for Part D Decision-Making, (in
  progress)

41
PDS 2005 Response Rates by Race/Ethnicity, and
by Type of Contact
Source D. Weir, The Unbalanced Distribution of
Cognitive Demands and Cognitive Resources for
Part D Decision-Making.
42
People Love Their Meds4600 respondents listed
15,000 drugsFor each drug, we asked

• Is this medication very important for your
  health?
• 91 Agree
• 1 Disagree
• 8 Dont Know

Source D. Weir, The Unbalanced Distribution of
Cognitive Demands and Cognitive Resources for
Part D Decision-Making.
43
To understand who needs to do what, you need to
separate out five main categories of people.CMS
Outreach Five Target Groups

• Retiree Rx coverage
• Short term, stay put
• Medicaid
• Short term, automatic enroll, maybe wrong meds
• Medicare Advantage (HMO)
• Short term, dont need to do anything
• Other Low-income
• Have to apply for extra help, choose plan
• Other
• Need to decide whether to enroll, choose plan

Source D. Weir, The Unbalanced Distribution of
Cognitive Demands and Cognitive Resources for
Part D Decision-Making.
44
Groups Differ Widely by Economic Resources
(Median values, HRS2004)
Source D. Weir, The Unbalanced Distribution of
Cognitive Demands and Cognitive Resources for
Part D Decision-Making.
45
Groups Differ Widely in Cognitive Resources
Source D. Weir, The Unbalanced Distribution of
Cognitive Demands and Cognitive Resources for
Part D Decision-Making.
46
Groups Differ Widely in Health Status
Source D. Weir, The Unbalanced Distribution of
Cognitive Demands and Cognitive Resources for
Part D Decision-Making.
47
Stress Associated with High Demands, Low
Resources for Decision-Making
Source D. Weir, The Unbalanced Distribution of
Cognitive Demands and Cognitive Resources for
Part D Decision-Making.