Federal Redistribution:
Supplementing Point-in-Time
Analyses with a Lifetime View
Presentation for Tax Policy Center Event
“Measuring the Distribution of Spending and Taxes”
Melissa Favreault
The Urban Institute
December 3, 2013
URBAN INSTITUTE
Motivation
• Well-being is better measured based on lifetime
earnings/income rather than point-in-time
earnings/income
• “Insurance company with a standing army”:
analysts often evaluate insurance through
expected value (actuarial fairness, insurance value)
– Requires looking at tax contributions (payroll,
personal income, other) and benefits over a lifetime,
not just at a point in time
– Separate analyses by age partially address this
URBAN INSTITUTE
Poverty at a point in time and
over longer periods
Shares in Poverty Using Different Accounting Periods
60
51.4
50
40
31.6
30
20
15.0
10
0
Annual 2012
At least 2 months over 3 years
(2009 - 2011)
At some time before age 65
Sources: DeNavas-Walt, Proctor and Smith (2013), Rank and Hirschl (1999)
URBAN INSTITUTE
Transitions between quintiles similarly
differ by accounting period
Current Earnings Quintile Compared to Last
Year:
Men Ages 30-54
100%
90%
0.0
18.7
13.9
10.0
100%
28.7
80%
70%
18.5
29.0
46.7
40.4
70%
60%
82.8
50%
63.9
69.2
71.0
20%
76.3
60%
75.5
30%
Move up
50%
Stay same
40%
Move down
30%
54.5
44.9
Move up
39.9
Stay same
Move down
53.3
20%
10%
0%
0.0
90%
80%
40%
Current Earnings Quintile Compared to
Average of Last 10 Years:
Men Ages 30-54 Who Worked at Least 6 of
Last 10 Years
17.4
17.0
0.0
14.6
17.1
10%
0%
19.7
26.0
26.9
23.6
0.0
Source: Favreault and Haaga (2013) based on Survey of Income and Program Participation matched to earnings records
URBAN INSTITUTE
Why do relatively few analysts
examine redistribution over a lifetime?
• Enormous data demands
• A girl born in 2000 can expect to live for 85.7 years
• Few panel data sources
• Can use administrative data; privacy concerns limit
• Families/tax units change over time
• Sensitive to assumptions, including about many
things that have not yet happened
• Current law vs. current policy
• Interest (“discount”) rates, mortality, earnings inequality
• Inclusion of migrants, those who die early, etc.
URBAN INSTITUTE
Mortality differs greatly by social
status, affecting lifetime redistribution
Life Expectancy Difference by Economic Status
9
8
7.8
Education
Difference in years
7
6
Community
income
5.4
5
4.1
4
3
2
1
0
Men
Women
combined
at age 25
at age 15
Sources: Meara et al. (2008); Singh and Siahpush (2006)
URBAN INSTITUTE
Example: Contrast Social Security at a
point in time vs. on a lifetime basis
• Just one component of federal budget, but large
• Tax side
– Flat payroll tax (subject to maximum, $117,000 in 2014)
– Taxation of benefits for higher income beneficiaries
• Benefit side
– Retirement, survivor, and disability benefits
• Marriage-based benefits and children’s benefits
– Mortality (increasingly disparate by income)
• Will look by race/ethnicity
URBAN INSTITUTE
A cross-sectional view of Social Security
redistribution
Aggregate Benefit to Tax Ratios by Year and Race/Ethnicity
1.9
1.8
1.7
1.6
1.5
1.4
Benefit to Tax Ratios
1.3
1.2
1.1
1
Non-Hispanic white
0.9
Non-Hispanic black
0.8
Hispanic
0.7
Non-Hispanic other
0.6
0.5
0.4
0.3
0.2
0.1
0
URBAN INSTITUTE
Year
Source: Based on Steuerle, Quakenbush, and Smith (2013)
A lifetime view of Social Security
redistribution
Lifetime Benefits to Contributions by Birth Year and Race/Ethnicity
Ratio of lifetime Social Security benefits to contributions
1.4
1.3
1.2
1.1
1
0.9
0.8
Hispanic (all)
0.7
Non-Hispanic black
0.6
Non-Hispanic white
0.5
0.4
0.3
0.2
0.1
0
1951-1955
1956-1960
1961-1965
URBAN INSTITUTE
Source: Computations from DYNASIM3
1966-1970
1971-1975
Distribution versus mean or median:
lifetime perspective
Distribution of the Ratio of Lifetime Benefits to Contributions
Share with benefit-to-contribution ratios of this amount or higher
Ratio of 1 or higher indicates lifetime Social Security benefits exceed lifetime payroll taxes
100
90
80
70
60
50
Hispanic
Non-Hispanic black
40
30
20
10
0
URBAN INSTITUTE
Non-Hispanic white
Conclusions
• Looking at cross-sections alone gives an
incomplete picture of redistribution
– Many of us are likely to be poor at some point, to
change income or earnings quintiles, and to face risks
– Lifetime calculations are difficult but necessary
• Looking at measures of central tendency (mean or
median) alone masks diversity within groups like
quintiles or broad age groups
– A high group-specific mean may result from a subset of
that group having very high outcomes , even though
many or even most individuals in the group do not
URBAN INSTITUTE
Interpretation
• Big-ticket budget items (Medicare, Social Security,
and Medicaid)
– Manage lifetime risks, especially from very large shocks
to income
• Health crises that will incur massive debts absent insurance
• Permanent disability
• Long-term unemployment
– Affect multiple generations
• Aged “pre-paid” benefits with decades of payroll taxes (“earned
right”): majority have at least 35 years of contributions
• Public redistribution to aged reduces need for private transfers
URBAN INSTITUTE
Disclaimer
All views are my own and not those of the
Urban Institute, its board, or sponsors.
Please visit us at:
www.urban.org and www.retirementpolicy.org
URBAN INSTITUTE