Panel Study of Income Dynamics

1
Panel Study of Income Dynamics
Institute for Social Research University of
Michigan
2
Origins of the PSID

• U.S. War on Poverty in 1960s stimulated interest
  in understanding dynamics of income poverty
• PSID was established in 1968 as a five year
  project to study income dynamics within a
  national sample of 5,000 families
• Today, PSID includes 8,000 families is used to
  address a much broader set of issues

3
Outline of Presentation

• Sample design content
• Administrative funding structures
• Indicators of impact
• Areas of scientific advance
• Focus on contributions to three issues
• Changes in income dynamics, 1970s through today
• Intergenerational transmission of well-being
• Long-run effects of early life events

4
Study Design

• National sample of 5,000 families in 1968
• Now consists of 8,000 families
• Split-offs are followed
• 1968 family members are followed as they grow-up
  and establish own economic family units
• Therefore, multiple generations of family members
  are surveyed
• Key design elements
• Mode Telephone - CATI
• Frequency Every year, 1968-97 biennial since
  1997
• Incentive payment 60
• 96-98 wave-to-wave response rate
• Attritors are recontacted

5
Survey Content has ExpandedSignificantly Since
1997
Interview length in minutes
6
PSID Now Spans Numerous Domains

• Income, in detail
• Employment
• Computer Use
• Expenditures
• Program participation
• Housing
• Time spent on housework
• Child care

• Vehicle inventory
• Health status, health behavior, health
  insurance
• Wealth active savings
• Philanthropic giving volunteering Tsunami aid
• Marriage fertility
• Education

Content of the 2005 instrument
7
Data are Freely Accessedon the Internet

• Internet-based Data Center
• http//www.psidonline.isr.umich.edu
• Supports customized subsetting of all 34 waves of
  data
• Provides customized codebooks
• Allows variable searching and browsing
• Automatically conducts complex dataset merges
• Automatically makes data available in SAS, STATA,
  dBase, ASCII, Excel
• Entire dataset can also be downloaded directly

8
Administrative Funding Structures

• Researcher initiated grant-funded project
• Design collection led by faculty at the
  University of Michigan
• Roughly 20 full-time staff, plus 100 interviewers
  during the field period
• Decisions made by the Principal Investigators
  with input from the external Board of Overseers

9
Sponsorship of the PSID

• Major sponsors
• National Science Foundation
• National Institute on Aging
• National Institute of Child Health Human
  Development
• Assistant Secretary for Planning Evaluation,
  Dept of HHS
• Department of Housing Urban Development
• Indiana University Purdue University Center on
  Philanthropy
• Funding obtained through peer-reviewed grant
  competition for each funder

10
Number of Published Journal Articles Using PSID
Data has Increased Steadily Over Time
Year
11
Journals Most Commonly Publishing Articles Using
the PSID, 1995-2005

• American Economic Review
• Journal of Human Resources
• Journal of Marriage and Family
• Journal of Labor Economics
• Demography
• Review of Economics and Statistics
• American Sociological Review
• Journal of Public Economics
• Social Science Research
• Journal of Econometrics
• Journal of Political Economy
• Review of Income and Wealth/

12
Other Indicators of Use Significance

• Over 2,200 peer-reviewed publications using the
  PSID
• Currently one peer reviewed publication every 3.9
  days
• Tool for the U.S. federal government
• Treasury Department, USDA, HUD, HHS/ASPE, CBO
• Each year..
• 1.6 million hits to the PSID web site
• 5,000 unique IP addresses download data
• Named one of NSFs Nifty Fifty advances
• Replicated in numerous countries across the world

13
Selected List of Areas of Scientific Advance

• Retirement aging
• Wealth savings
• Socioeconomic inequality
• International comparisons
• Influences of family history
• Child development
• Income poverty dynamics
• Intergenerational studies
• Long-run effects of early life events

• Public transfer program participation
• Fertility, marriage, migration
• Neighborhood effects
• Health disparities
• Economic consequences of job loss
• Impact of family structure on wellbeing

14
Example 1Income Poverty Dynamics

• Panel design transformed research on poverty from
  a static view of "poor" and "rich" to a dynamic
  view in which families experience episodes of
  poverty
• Most spells of poverty are 1-2 years in length
• A smaller share of spells are chronic, lasting
  for many years
• One-third of children live in poverty before age
  18
• 70 of African-American children experience
  poverty before age 18
• Poverty spells are triggered by
• reduction/loss of work (66 of spells)
• divorce or separation (10 of spells)

15
Example 1 (Continued)Income Poverty Dynamics

• Panel design allows estimation of change in
  dynamics
• Rising volatility in family income over the past
  15 years
• Series of front page stories in the Los Angeles
  Times, October 10, 2004-December 30, 2004
• Published just prior to U.S. Presidential
  elections stimulated a series of lead stories
  in the NY Times and other media outlets

16
Example 2Intergenerational Studies

• It has been claimed that success in the U.S.
  labor market is determined largely by ones own
  abilities and energy, not by family background
• A series of studies has investigated this issue
  using the genealogical design of the PSID
• Findings.
• Much higher intergenerational correlation in
  economic status than previously believed
• Intergenerational correlation is no lower in the
  U.S. than other countries

17
Intergenerational Correlationin Economic Status
is High
18
Intergenerational Transmission of Wealth
Adult Childrens Wealth Position
Source Charles and Hurst (2003)
19
Correlation in Father-Son Earnings is No Lower
in U.S. than in Most Other Countries
20
Intergenerational Transmission in Obesity is
High Even Across Three Generations
21
Example 3Long-run Effects of Early Life Events

• Background
• Substantial literature from the UK indicating
  long reach of early-life events
• Objective
• Investigate, in the U.S., linkages between health
  economic status in initial stages of life, and
  health, education, income in adulthood

22
Unique Aspects of the Study

• Nationally representative data of the U.S.
• Childhood measures are not based on long recalls
• Superior measures of income in childhood
  adulthood
• Data spanning substantial share of life course
  35 yrs
• Sufficient sample of low-income minority
  population to examine disparities
• Comparisons within families sibling fixed effects

23
Research Questions

• Does low birth weight affect adult health,
  education, labor market outcomes?
• Does childhood family income affect adult health,
  education, and labor market earnings
• Does low birth weight interact with
• childhood family income, health insurance, and
  parental health behaviors

24
Summary of Findings

• Being born low weight.
• ages you by 12 years,
• increases the odds of dropping out of high school
  by 5 percentage points, and
• lowers labor force participation by 5 percentage
  points, and
• reduces labor market earnings by 14
• Effects on health are mitigated by having health
  insurance in childhood

25
Closing Comments

• None of these findings would have been possible
  if the PSID had only 10-15 years of data
• Even the 35-year period observed in the PSID may
  be too short
• Intergenerational correlations higher if
  son/daughter observed at older ages
• Preliminary evidence indicates that cumulative
  effects of early life events get larger, not
  smaller, with age
• Need to look beyond ages 30-40

26
EXTRA SLIDES
27
The Child Development Supplement

• The CDS is a nationally representative,
  longitudinal study of children and their
  families, examining a broad array of
  developmental outcomes within the context of
  family, neighborhood, and school environments

28
About the CDS

• CDS-I (1997) Initial Sample
• Selected 2,705 PSID families with children 0-12
  years
• Randomly selected up to two children per family
• Successfully interviewed 2,394 families (88) and
  3,563 target children
• CDS-II (2002) Re-Interview
• Recontacted 2,226 families in CDS-I who remained
  active in the PSID panel as of 2001 PSID
  interview
• Successfully interviewed 2,019 families (91) and
  2,907 target children

29
CDS Modules and Respondents

• In-Home interviews with the CDS child and family
• Telephone/personal interview w/ primary
  secondary caregivers
• Personal interview with child
• Child self interview (ACASI)
• Standardized assessments (Woodcock Johnson) for
  reading math skills
• Time diary accounting for one randomly selected
  weekday and one weekend 24-hour period

30
CDS Modules and Respondents

• Interviewer observations (HOME Scale)
• Telephone interview w/ elementary school teachers
• Curriculum data from course catalogs for
  middle/high school
• School administrative information via links to
  U.S. Department of Education data

31
CDS Content Domains

• Family environment
• Psychological social wellbeing
• Health health behaviors
• Time use measured by time diaries
• Schooling school environment
• Achievement
• Caregiver social psychological resources

32
Future Directions

• Transition to adulthood
• Special modules for CDS children 18-25
• Fielded for the first time in 2005
• Time diaries
• Proposal under review to study the connection of
  disability and time use among older married
  couples
• Biomarkers
• Early planning stages