U
NEQUAL
W
E
S
TAND
An empirical analysis of economic inequality in the
United States, 1967-2006
Jonathan Heathcote
Federal Reserve Bank of Minneapolis, and CEPR
Fabrizio Perri
University of Minnesota, Federal Reserve Bank of Minneapolis, CEPR and NBER
Gianluca Violante
New York University, CEPR and NBER
RED special issue
• Macroeconomics is expanding its focus from aggregate dynamics
to dynamics of entire equilibrium distributions across agents
• Representative-agent business-cycle literature built on well
defined set of facts about aggregate variables
• Heterogeneous-agents incomplete-markets literature lacks
systematic set of “uncontroversial" facts about cross-sections
• RED special issue: consistently document facts about key dimensions of cross-sectional inequality in several countries
I USA, UK, Canada - Italy, Germany, Spain, Sweden - Russia,
RED special issue
• Macroeconomics is expanding its focus from aggregate dynamics
to dynamics of entire equilibrium distributions across agents
• Representative-agent business-cycle literature built on well
defined set of facts about aggregate variables
• Heterogeneous-agents incomplete-markets literature lacks
systematic set of “uncontroversial" facts about cross-sections • RED special issue: consistently document facts about key
dimensions of cross-sectional inequality in several countries
I USA, UK, Canada - Italy, Germany, Spain, Sweden - Russia,
RED special issue
• Macroeconomics is expanding its focus from aggregate dynamics
to dynamics of entire equilibrium distributions across agents
• Representative-agent business-cycle literature built on well
defined set of facts about aggregate variables
• Heterogeneous-agents incomplete-markets literature lacks
systematic set of “uncontroversial" facts about cross-sections
• RED special issue: consistently document facts about key dimensions of cross-sectional inequality in several countries
I USA, UK, Canada - Italy, Germany, Spain, Sweden - Russia,
Economic inequality in the United States
1970 1975 1980 1985 1990 1995 2000 2005 0.25
0.3 0.35 0.4 0.45
Year
Variance of Log
Wage and Consumption Inequality
Men Hourly Wages(CPS)
Economic inequality in the United States
1970 1975 1980 1985 1990 1995 2000 2005 0.25
0.3 0.35 0.4 0.45
Year
Variance of Log
Wage and Consumption Inequality
Men Hourly Wages (CPS) Household Consumption (CEX)
From “endowments” to “welfare”
• Several intervening choices and institutions in between individual
wages and household consumption
1. individual labor supply
2. income pooling within family
3. government taxes and transfers
4. borrowing/saving through financial markets
• Some mechanisms acts as dampening forces, others as
amplifying forces • Aim of this paper: descriptive empirical study to shed light on the
From “endowments” to “welfare”
• Several intervening choices and institutions in between individual
wages and household consumption
1. individual labor supply
2. income pooling within family
3. government taxes and transfers
4. borrowing/saving through financial markets
• Some mechanisms acts as dampening forces, others as
amplifying forces
• Aim of this paper: descriptive empirical study to shed light on the
Organizing device: household budget constraint
c + a0 = a +
N
X
i=1
wihi + bP + d + bG − τ
• wi: individual hourly wage
• hi: individual hours worked
• N: number of workers in the household
• bP: net private transfers
• d: household net asset income
• bG: government transfers
• τ: taxes
Intervening choices and institutions
c + a0 = a +
N
X
i=1
wihi + bP + d + bG − τ
• wi individual wage
• wihi individual labor supply
• PN
i=1wihi family labor supply
• PN
i=1 wihi + b
P + d family/social networks + financial markets
• PN
i=1 wihi + b
P + d + bG − τ public transfers and tax system
Three micro data sets representative of US population
1. Current Population Survey (CPS), 1968-2007
• repeated cross-section: ASEC supplement (March) covering
60,000+ households
• key strength: its size
2. Consumer Expenditure Survey (CEX), 1980-2007
• repeated cross-section: Interview Survey covering 15,000+
households
• key strength: consumption data
3. Panel Study of Income Dynamics (PSID), 1968-97, 99, 2001, 03
• longitudinal study: SRC sample following 3,000 families
Three micro data sets representative of US population
1. Current Population Survey (CPS), 1968-2007
• repeated cross-section: ASEC supplement (March) covering
60,000+ households
• key strength: its size
2. Consumer Expenditure Survey (CEX), 1980-2007
• repeated cross-section: Interview Survey covering 15,000+
households
• key strength: consumption data
3. Panel Study of Income Dynamics (PSID), 1968-97, 99, 2001, 03
• longitudinal study: SRC sample following 3,000 families
Three micro data sets representative of US population
1. Current Population Survey (CPS), 1968-2007
• repeated cross-section: ASEC supplement (March) covering
60,000+ households
• key strength: its size
2. Consumer Expenditure Survey (CEX), 1980-2007
• repeated cross-section: Interview Survey covering 15,000+
households
• key strength: consumption data
3. Panel Study of Income Dynamics (PSID), 1968-97, 99, 2001, 03
• longitudinal study: SRC sample following 3,000 families
Sample selection
1. Sample A
• “Clean” version of raw data: drop record only if seriously
incomplete or implausible
• used for population-level statistics (like NIPA)
2. Sample B
• restrict to households where at least one member is of
working age 25-60
• used for household-level (earnings, income, consumption)
statistics
3. Sample C
• restricted to those who work at least 260 hours per year
Macro facts in micro data
1970 1975 1980 1985 1990 1995 2000 2005 9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8 9.9
10 Labor Income Per−capita
Year
Log − 2003 $
1970 1980 1990 2000
9.5 9.6 9.7 9.8 9.9 10 10.1 10.2 10.3 10.4
Personal Income Per−capita (NIPA) vs
Money Income Per−capita (CPS)
Year
Log − 2003 $
CPS NIPA CPS
NIPA
• Labor income p.c. in CPS aligns well with NIPA
Individual wage inequality
1970 1980 1990 2000 0.2 0.25 0.3 0.35 0.4 0.45 Year
Variance of Log Hourly Wages
1970 1980 1990 2000 0.26 0.28 0.3 0.32 0.34 0.36 0.38
Gini Coefficient of Hourly Wages
Year
1970 1980 1990 2000 1.8
2 2.2
2.4 P50−P10 Ratio of Hourly Wages
Year 1970 1980 1990 2000
1.8 2 2.2 2.4
Year
P90−P50 Ratio of Hourly Wages
Decomposing (male) wage inequality
1970 1975 1980 1985 1990 1995 2000 2005 0.25
0.3 0.35 0.4 0.45
Variance of Log Male Wage
Year
Raw Residual
Recap: individual wage inequality
1. Continuous increase since late 1960s
• 1970s: concentrated at the bottom
• 1980s: throughout the distribution
• 1990s: concentrated at the top
2. Striking similarity between men’s and women’s trends...
• ... except for 1970s
3. Two-thirds of the increase is residual
• virtually 100% residual in the 1970s
Recap: individual wage inequality
1. Continuous increase since late 1960s
• 1970s: concentrated at the bottom
• 1980s: throughout the distribution
• 1990s: concentrated at the top
2. Striking similarity between men’s and women’s trends...
• ... except for 1970s
3. Two-thirds of the increase is residual
• virtually 100% residual in the 1970s
Recap: individual wage inequality
1. Continuous increase since late 1960s
• 1970s: concentrated at the bottom
• 1980s: throughout the distribution
• 1990s: concentrated at the top
2. Striking similarity between men’s and women’s trends...
• ... except for 1970s
3. Two-thirds of the increase is residual
• virtually 100% residual in the 1970s
Role of individual labor supply: men
1970 1980 1990 2000 0.2
0.3 0.4 0.5
Variance of Male Log Wages
Year 1970 1980 1990 2000
0 0.1 0.2 0.3
Variance of Male Log Annual Hours
Year
1970 1980 1990 2000 −0.1
0 0.1 0.2
Correl. btw Log Hours and Log Wages
Year 1970 1980 1990 2000
0.4 0.5 0.6 0.7
Variance of Male Log Annual Earnings
Understanding individual earnings inequality
1970 1980 1990 2000 −0.5
0 0.5
Earnings Ranked by Earnings Decile
Year
Log Change
1970 1980 1990 2000 −0.4 −0.2 0 0.2 0.4 0.6
Wage Ranked by Earnings Decile
Year
Log Change
1970 1980 1990 2000 −0.8 −0.6 −0.4 −0.2 0 0.2
Hours Ranked by Earnings Decile
Year
Log Change
1970 1980 1990 2000 4
6 8
10 Unemployment Rate
Recap: men’s earnings inequality
1. Steady increase, until early 1990s
• During 1970s-1980s, rise is twice as big as for wages
2. It reflects rising correlation between wages and hours worked
Recap: men’s earnings inequality
1. Steady increase, until early 1990s
• During 1970s-1980s, rise is twice as big as for wages
2. It reflects rising correlation between wages and hours worked
Var. of log vs. Gini for household earnings
1970 1980 1990 2000 0.5
0.6 0.7 0.8
Variance of Log Equiv. Household Earnings
Year 1970 1980 1990 2000
0.32 0.34 0.36 0.38 0.4
0.42Gini Coefficient of Equiv. Household Earnings
Year
1970 1980 1990 2000 2.5
2.75 3 3.25 3.5
P50−P10 Ratio of Equiv. Household Earnings
Year 1970 1980 1990 2000
1.8 2 2.2 2.4 2.6 2.8
P90−P50 Ratio of Equiv. Household Earnings
Year
Var. of log vs. Gini for household earnings
1970 1980 1990 2000 0.5
0.6 0.7 0.8
Variance of Log Equiv. Household Earnings
Year 1970 1980 1990 2000
0.32 0.34 0.36 0.38 0.4
0.42Gini Coefficient of Equiv. Household Earnings
Year
1970 1980 1990 2000 2.5
2.75 3 3.25 3.5
P50−P10 Ratio of Equiv. Household Earnings
Year 1970 1980 1990 2000
1.8 2 2.2 2.4 2.6 2.8
P90−P50 Ratio of Equiv. Household Earnings
Year
Var. of log vs. Gini for household earnings
1970 1980 1990 2000 0.5
0.6 0.7 0.8
Variance of Log Equiv. Household Earnings
Year 1970 1980 1990 2000
0.32 0.34 0.36 0.38 0.4
0.42Gini Coefficient of Equiv. Household Earnings
Year
1970 1980 1990 2000 2.5
2.75 3 3.25 3.5
P50−P10 Ratio of Equiv. Household Earnings
Year 1970 1980 1990 2000
1.8 2 2.2 2.4 2.6 2.8
P90−P50 Ratio of Equiv. Household Earnings
Year
Cyclicality of household earnings inequality
1970 1975 1980 1985 1990 1995 2000 2005 −0.3
−0.2 −0.1 0 0.1 0.2 0.3 0.4 0.5 0.6
Equivalized Household Earnings
Year
Log (normalized to 0 in 1967)
CPS Percentiles
P5 P10 P25 P50 P95
P90 P75
Long-run impact of recessions on inequality
1970 1975 1980 1985 1990 1995 2000 2005 0
0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
Household Earnings
Year
Log (normalized to 0 in 1967) 90−10 Log Differential
Role of household labor supply
1970 1980 1990 2000 0.45
0.5 0.55 0.6 0.65 0.7 0.75 0.8
Variance of Log Earnings
Year 1970 1980 1990 2000
0.32 0.34 0.36 0.38 0.4 0.42 0.44 0.46
Gini of Household Earnings
Year
Main Earner Household Main Earner
Household
Forces impacting scope for within-household insurance
1970 1980 1990 2000 0.4
0.5 0.6 0.7 0.8
0.9 Var. of Log Household Earnings
Year
Singles Married
1970 1980 1990 2000 0.55
0.6 0.65 0.7 0.75
0.8 Fraction of Married Households
Year
1970 1980 1990 2000 0.55 0.6 0.65 0.7 0.75 0.8
Fraction of Two−Earner Married Households
Year
1970 1980 1990 2000 0.1
0.2 0.3 0.4
Within Married Household Correlation of Log Earnings
Recap: income pooling within the household
• Married households do significant income pooling
I but... increasingly assortative matching
• Women’s labor force participation increases potential role of
within-family insurance
I but... increasing fraction of single households
Recap: income pooling within the household
• Married households do significant income pooling
I but... increasingly assortative matching
• Women’s labor force participation increases potential role of
within-family insurance
I but... increasing fraction of single households
Recap: income pooling within the household
• Married households do significant income pooling
I but... increasingly assortative matching
• Women’s labor force participation increases potential role of
within-family insurance
I but... increasing fraction of single households
Role of private transfers
1970 1980 1990 2000 0.5
0.55 0.6 0.65 0.7 0.75 0.8 0.85
Year Variance of Log
1970 1980 1990 2000 0.33
0.34 0.35 0.36 0.37 0.38 0.39 0.4 0.41 0.42 0.43
Year Gini
Household Earnings + Private Transfers
Role of private transfers
1970 1980 1990 2000 0.5
0.55 0.6 0.65 0.7 0.75 0.8 0.85
Year Variance of Log
1970 1980 1990 2000 0.33
0.34 0.35 0.36 0.37 0.38 0.39 0.4 0.41 0.42 0.43
Year Gini
Household Earnings + Private Transfers
Role of asset income
1970 1980 1990 2000
0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8
Year
Variance of Log
1970 1980 1990 2000
0.34 0.36 0.38 0.4 0.42 0.44
Year Gini
HH Earn. + Priv. Transf. + Asset Income
Role of asset income
1970 1980 1990 2000
0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8
Year
Variance of Log
1970 1980 1990 2000
0.34 0.36 0.38 0.4 0.42 0.44
Year Gini
HH Earn. + Priv. Transf. + Asset Income
Role of public transfers
1970 1975 1980 1985 1990 1995 2000 2005 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 Year Variance of Log
Pre−Govt. Income + Govt. Benefits
1970 1975 1980 1985 1990 1995 2000 2005 0.33 0.34 0.35 0.36 0.37 0.38 0.39 0.4 0.41 0.42 0.43 Year Gini
• Public transfers greatly reduce level of inequality at the bottom
• Expansion in 1970s, tightening after 1994
Role of public transfers
1970 1975 1980 1985 1990 1995 2000 2005 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 Year Variance of Log
Pre−Govt. Income + Govt. Benefits
1970 1975 1980 1985 1990 1995 2000 2005 0.33 0.34 0.35 0.36 0.37 0.38 0.39 0.4 0.41 0.42 0.43 Year Gini
• Public transfers greatly reduce level of inequality at the bottom
• Expansion in 1970s, tightening after 1994
Role of public transfers
1970 1975 1980 1985 1990 1995 2000 2005 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 Year Variance of Log
Pre−Govt. Income + Govt. Benefits
1970 1975 1980 1985 1990 1995 2000 2005 0.33 0.34 0.35 0.36 0.37 0.38 0.39 0.4 0.41 0.42 0.43 Year Gini
• Public transfers greatly reduce level of inequality at the bottom
• Expansion in 1970s, tightening after 1994
Role of public transfers
1970 1975 1980 1985 1990 1995 2000 2005 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 Year Variance of Log
Pre−Govt. Income + Govt. Benefits
1970 1975 1980 1985 1990 1995 2000 2005 0.33 0.34 0.35 0.36 0.37 0.38 0.39 0.4 0.41 0.42 0.43 Year Gini
• Public transfers greatly reduce level of inequality at the bottom
• Expansion in 1970s, tightening after 1994
Role of tax system
19701975198019851990199520002005 0.3
0.35 0.4 0.45 0.5 0.55 0.6
Year Variance of Log
1970 1975 1980 1985 1990 1995 2000 2005 0.31
0.32 0.33 0.34 0.35 0.36 0.37 0.38 0.39 0.4 0.41
Year Gini
Pre−Tax. Income Disp. Income
REP DEM REP REP DEM REP
• Taxes greatly reduce level of inequality throughout the distribution
Role of tax system
19701975198019851990199520002005 0.3
0.35 0.4 0.45 0.5 0.55 0.6
Year Variance of Log
1970 1975 1980 1985 1990 1995 2000 2005 0.31
0.32 0.33 0.34 0.35 0.36 0.37 0.38 0.39 0.4 0.41
Year Gini
Pre−Tax. Income Disp. Income
REP DEM REP REP DEM REP
• Taxes greatly reduce level of inequality throughout the distribution
Recap: household income dispersion
• Private transfers somewhat reduce inequality at bottom (small)
• Asset income increases inequality at the top, but it is underestimated
• Public transfers play a significant role for redistribution and stabilization – less since 1994
Macro facts in micro data: CEX
1980 1985 1990 1995 2000 2005 9.6 9.7 9.8 9.9 10 10.1 10.2 10.3 Income Per−capita Year
Log scale (2003 $)
NIPA CEX
1980 1985 1990 1995 2000 2005 7.5 7.6 7.7 7.8 7.9 8 8.1 8.2
8.3 Food Consumption Per−capita
Year
Log scale (2003 $)
NIPA CEX PSID
Macro facts in micro data: CEX
1980 1985 1990 1995 2000 2005 9.6 9.7 9.8 9.9 10 10.1 10.2 10.3 Income Per−capita Year
Log scale (2003 $)
NIPA CEX
1980 1985 1990 1995 2000 2005 7.5 7.6 7.7 7.8 7.9 8 8.1 8.2
8.3 Food Consumption Per−capita
Year
Log scale (2003 $)
NIPA CEX PSID
Comparison CPS-CEX: household earnings
1970 1980 1990 2000
−0.2 −0.15 −0.1 −0.05 0 0.05 0.1
Variance of Log
Year
Series demeaned
CPS CEX
1970 1980 1990 2000
−0.04 −0.03 −0.02 −0.01 0 0.01 0.02 0.03 0.04
Gini
Year
Series demeaned
CPS CEX
Comparison CPS-CEX: disposable income
1970 1980 1990 2000 −0.1
−0.05 0 0.05
Variance of Log
Year
Series demeaned
CPS CEX
1970 1980 1990 2000 −0.04
−0.03 −0.02 −0.01 0 0.01 0.02 0.03
Gini
Year
Series demeaned
CPS CEX
From disposable income to consumption
1980 1990 2000
0.25 0.3 0.35 0.4 0.45 0.5 0.55
Variance
Year 1980 1990 2000
0.28 0.3 0.32 0.34 0.36 0.38
0.4 Gini Coefficient
Year Disp. Income
Nond. Cons.
Contrasting income smoothing at top and bottom
19801 1985 1990 1995 2000 2005 1.5
2 2.5 3 3.5 4 4.5 5 5.5 6
P50−P10 (ranked by earnings)
19801 1985 1990 1995 2000 2005 1.5
2 2.5 3 3.5 4 4.5 5 5.5 6
P90−P50 (ranked by earnings)
Saving and borrowing
• Difference between income and consumption inequality: saving
and borrowing
• Scope for self-insurance via saving/borrowing depends on
persistence of shocks
• Estimating persistence requires panel data: PSID
I Permanent-transitory model with non-stationarity in
Estimation of wage dynamics from PSID
• Focus on log male wages because closest definition to
“exogenous endowments”
• Permanent-transitory model for (residual) wage dynamics
wi,a,t = zi,a,t + εi,a,t
zi,a,t = zi,a−1,t−1 + ηi,a,t
• Time-varying variances of the three components: structural parameters are {σεt, σηt, σz0,t}
T t=1
• Minimum distance estimation:
1. moments in first-differences
Comparison CPS-CEX-PSID: male wages
1970 1980 1990 2000 −0.1
−0.05 0 0.05 0.1 0.15
Var. of Log
Year
Series demeaned
1970 1980 1990 2000 −0.04
−0.02 0 0.02 0.04 0.06 0.08
Gini
Year
Series demeaned
CPS PSID CEX
CPS PSID CEX
Results
1970 1975 1980 1985 1990 1995 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04 0.045 0.05
Variance of Permanent Innovation
Year 1970 1975 1980 1985 1990 1995
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Wage Variance Decomposition
Year Levels
Diff.
Total Var
Transitory Var: Levels Transitory Var: Diff.
Fit of moments in first differences
1970 1975 1980 1985 1990 1995 0.1
0.15 0.2 0.25 0.3
Variance of Wage Growth
Year Data & Diff. Levels
1970 1975 1980 1985 1990 1995 −0.15
−0.1 −0.05 0 0.05
Covariance of Wage Growth at Lag 1
Year Data & Diff. Levels
Fit of moments in levels
1970 1975 1980 1985 1990 1995 0.2 0.4 0.6 0.8 1 1.2
Variance of Wages (1969 Cohort)
Year Data
Levels Diff.
1970 1975 1980 1985 1990 1995 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Wage Correlation between age 25 and age t (1969 Cohort)
Year
Data Levels Diff.
• Estimation in differences spectacularly fails to match these
Recap: wage dynamics
• Misspecification of the permanent-transitory model
• Estimates in levels more reliable (misspecification error less
dramatic)
• What is the right stochastic process for log wages?
Autocovariance function of log wages
1970 1975 1980 1985 1990 1995 0.02
0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18
0.2 Autocovariance of Log Wages
Year
1969
1976
Net worth-income ratios by income deciles (SCF)
1990 1995 2000 2005 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4
Bottom Income Deciles
Year D1 D2 D3 D4 D5
1990 1995 2000 2005 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4
Top Income Deciles
Year D6 D7 D8 D9 D10
Conclusions
• Difficult to generalize about inequality without asking: Inequality in
what? Inequality measured how?
• Difficult to understand evolution of inequality or its implications for
welfare without modeling explicitly choices/institutions that
mediate between wages and consumption / leisure
• Inequality increases permanently in recessions ⇒ unified theory
of business cycles and long-run trends for inequality
• Permanent-transitory model misspecified, and bias enormous
when estimated in first-differences
• Rise in consumption inequality smaller than rise in income
Conclusions
• Difficult to generalize about inequality without asking: Inequality in
what? Inequality measured how?
• Difficult to understand evolution of inequality or its implications for
welfare without modeling explicitly choices/institutions that
mediate between wages and consumption / leisure
• Inequality increases permanently in recessions ⇒ unified theory
of business cycles and long-run trends for inequality
• Permanent-transitory model misspecified, and bias enormous
when estimated in first-differences
• Rise in consumption inequality smaller than rise in income
Conclusions
• Difficult to generalize about inequality without asking: Inequality in
what? Inequality measured how?
• Difficult to understand evolution of inequality or its implications for
welfare without modeling explicitly choices/institutions that
mediate between wages and consumption / leisure
• Inequality increases permanently in recessions ⇒ unified theory
of business cycles and long-run trends for inequality
• Permanent-transitory model misspecified, and bias enormous
when estimated in first-differences
• Rise in consumption inequality smaller than rise in income
Conclusions
• Difficult to generalize about inequality without asking: Inequality in
what? Inequality measured how?
• Difficult to understand evolution of inequality or its implications for
welfare without modeling explicitly choices/institutions that
mediate between wages and consumption / leisure
• Inequality increases permanently in recessions ⇒ unified theory
of business cycles and long-run trends for inequality
• Permanent-transitory model misspecified, and bias enormous
when estimated in first-differences
• Rise in consumption inequality smaller than rise in income