www.bea.gov

Accounting for the Distribution of Income in the U.S. National Accounts

Dennis Fixler Bureau of Economic Analysis

David S Johnson

US Census Bureau

Prepared for BEA Advisory Committee Meeting 16 November 2012

1

www.bea.gov

Has income increased or not?

0

10,000

20,000

30,000

40,000

50,000

60,00019

80 1981

1982

1983

1984

1985

1986

1987

1988

1989

1990 1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

Real Median Household Income

Real Per capita GDP

2

www.bea.gov

Issue is that CPS income tracks National Accounts Personal Income until recently

40,000

50,000

60,000

70,000

80,000

90,000

100,000

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

CPS Mean

Adj Personal income per household

Since 1999

5.3%

-5.7%

3

www.bea.gov

GDP and Distribution Information

▪ Long recognized that in gauging economic performance GDP cannot stand alone; distribution information needed

▪ Is there a positive or negative correlation between income distribution and economic growth?

▪ Kuznets curve—upside down U

4

www.bea.gov

Evaluating the income distribution and its relationship to National accounts is not new

▪ 1st NBER volume - Mitchell, et al. 1921. Income in the United States: Its Amount and Distribution, 1909-1919

▪ CRIW volume 1943 - Income Size Distributions in the United States, Part I.

▪ CRIW Volume, 1975 - The Personal Distribution of Income and Wealth, James D. Smith, ed., 1975.

▪ Office of Business Economics (the predecessor to BEA) early reports - Goldsmith (1955) “Income Distribution in the United States, 1950-53,” Survey of Current Business.

5

www.bea.gov

Recent Emphasis on Distribution

▪ Stigliz et al: information on distribution serves as an important complement to GDP

▪ 2012 Economic Report of the President: distributional aspects of fiscal policy

▪ IMF Working Paper: “Innocent Bystanders? Monetary Policy and Inequality in U.S.” WP/12/199 August 2012

▪ Gordon “Misperceptions About the Magnitude and Timing of Changes in American Income Inequality” NBER Working Paper 15351

6

www.bea.gov

Is Inequality related to Growth

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

0.360.370.380.390.4

0.410.420.430.440.450.460.47

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

Gini Coefficient

Per capita GDP

7

www.bea.gov

Purpose of Research

• BEA FY11 budget proposal, which included producing “a decomposition of personal income that presents median as well as mean income…”

• Because survey data suffer from under-reporting, determine how to deal with measurement error in income

• Demonstrate that one can use NIPA data to adjust survey data to obtain alternative distributions and measures of inequality.

▪ Provide examples of the usefulness of the distribution measures on expenditure multipliers and social welfare measures

8

www.bea.gov

The Measurement of Income

▪ Use the Haig-Simons definition of income, Income (Y) = Consumption (C) plus change in wealth (∆W). Most studies do not implement this definition

▪ Census Money income is different (conceptually and empirically) than BEA Personal Income

▪ Issue is that there is underreporting of income in household surveys

▪ Key is that a common, consistent and accurate measure of income is important for understanding the distribution.

9

www.bea.gov

Alternative measures of Income

SOURCE Haig/ Simons Census PI (BEA) CBO SOI (AGI) Canberra

Employment income Yes Yes Yes Yes Yes Yes

Employer contribution to Soc Sec Yes No Yes Yes No Yes

Employer-provided benefits Yes No Yes Yes No Yes

Investment income Yes Yes Yes . Yes Yes Yes

Imputed investment income Yes No Yes No No No

Government cash transfers Yes Yes Yes Yes Yes (taxable) Yes

Employee contribution to Soc Sec Yes Yes No (subtract) Yes Yes Yes

Retirement income Yes Yes No (only int.) Yes Yes Yes

Cash assistance from others Yes Yes No Yes No Yes

Realized capital gains Yes No No Yes Yes No

Lump sum (IRA disbursements) Yes No No Yes Taxable Yes

In-kind government transfers* Yes No Yes Yes No No**

Other In-kind transfers* Yes No No No No No**

Home production Yes No No No No In concept

Imputed rent* Yes No Yes No No Yes

Unrealized capital gains Yes No No No No No

Savings withdrawals Yes No No No No No

10

www.bea.gov

Data and Methods

▪ Begin with Household Income from Current Population Survey, 1999-2010

▪ Obtain total income and components -- wages, business income, property income, retirement income, government transfers, other

▪ Use Adjusted Personal income (from Katz (2012)) to ratio adjust CPS income

▪ Adjust measures to 2010($) using PCE deflator ▪ Calculate Adjusted Gross Income ▪ Use SOI tables to ratio adjust the distribution of

income

11

www.bea.gov

Adjustments to Personal Income, Selected Years, in billions of 2010 dollars Adjustments to Personal Income, Selected Years

1999 2007 2010 Personal Income 10,030 12,546 12,374

Employer health benefits (450) (637) (620)

Employer pensions benefits (267) (396) (470)

Imputed interest (433) (480) (457)

Imputed rent for homeowners (187) (68) (236)

Government transfers in-kind (575) (919) (1,132)

Adjustment for social security contributions 428 526 514

Adjustments for pension treatment (148) 123 257

Other adjustments (100) (92) (167)

Total adjustments (1,731) (1,943) (2,311)

Adjusted Personal Income 8,299 10,603 10,062

Census Money Income 7,387 8,316 8,015

12

www.bea.gov

Ratio adjusting CPS income

▪ Ratio adjust CPS to NIPA totals by source ▪ This procedure increases each household’s

income by source, and then the new data is used to obtain distribution measures (the procedure yields a mean for each source that matches the NIPA totals).

▪ Because higher income households have more property income and business income, their income is adjusted higher.

13

www.bea.gov

Adjusting CPS to Personal Income

CPS Income Adjustment factors (i j) NIPA income

Wages,…

Business,…

Property,…

Retirement,…

Government transfers,…

Other,….

Wij (+ supplements)

Bij

Pij (+ imputed int)

Rij

Gij (+ health benefits)

Oij

Adj Wage

Adj Bus.

Adj Prop

Adj Retire

Adj Gov’t

Adj Other

14

www.bea.gov

Adjustment Factors

0

0.5

1

1.5

2

2.5

3

3.5

4

WagesBusinessPropertyRetirementGovernmentTotal

15

www.bea.gov

Income shares

0%10%20%30%40%50%60%70%80%90%

100%

less than50,000

50,000-200,000 200,000 ormore

All

RetirementIncomeGovernmenttransfersProperty Income

Business Income

Wages

16

www.bea.gov

Ratio adjusting CPS distribution of income using SOI table

▪ Ratio adjust CPS distribution by the SOI totals by source and income level

▪ This procedure increases each household’s income by source and by income level, and then the new data is used to obtain distribution measures (the procedure yields a mean for each source that matches the NIPA totals).

▪ Because higher income households have higher underreporting, their income is adjusted higher while middle income households are adjusted lower.

17

www.bea.gov

SOI Factors used to adjust CPS income (ratio of aggregate income by source for level of AGI), 2009

00.5

11.5

22.5

33.5

44.5

5

Wages Business Property Retirement

18

www.bea.gov

Summary of Results between 1999 and 2010

▪ Real mean household income fell 5.7 percent, while per capita personal income increased 11.1 percent

▪ Using a more comparable definition of income, the mean adjusted real personal income per household increased 5.3 percent.

▪ Taking into account differences in the price index, accounting for underreporting and incorporating distributional information from both the CPS and SOI data, we obtain an increase of 5.7 percent (between 1999 and 2009)

▪ Hence, difference of 17 percentage points falls to 0.4 percentage points.

▪ In addition, there are larger increases in the median, yielding larger increases in inequality and Gini index increases more

▪ However, including health benefits: employer provided health, Medicaid and Medicare increases means, but decreases the level of and change in inequality

19

www.bea.gov

Mean NIPA-adjusted income increases more than mean Household income

40000

50000

60000

70000

80000

90000

100000

110000

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Mean (NIPA adjusted with health benefits)

Mean - SOI dist adj

Mean - NIPA adjusted

Household Mean

9.6% 5.7% 5.4% -2.3%

20

www.bea.gov

Median NIPA-adjusted income increases more than median Household income, but less than Mean NIPA-adjusted

40000

50000

60000

70000

80000

90000

100000

110000

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Household Median

Median - NIPA adjusted

Median - SOI dist adj

Median (NIPA adjusted with health benefits)

9.1% 0.8% 1.9% -3.6%

21

www.bea.gov

0.39

0.4

0.41

0.42

0.43

0.44

0.45

0.46

0.47

0.48

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Gin

i Coe

ffic

ient

Household IncomeNIPA- Adjusted IncomeSOI Adjusted IncomeNIPA adj, with health, retirement and imputed interest

Levels and Trends in Inequality

22

www.bea.gov

Distribution of income and consumption and multipliers: an application

▪ How does the income distribution affect the Keynesian expenditure multiplier?

▪ Economic Report of the President, among others, suggests that because lower income categories have higher MPC, then a redistribution can increase the size of the multiplier.

▪ This is an old concern: Stone and Stone (1938), Goodwin (1949), Chipman (1950) and Conrad (1955)

▪ Consider a simple closed economy in which the autonomous expenditures include all expenditures except consumption

23

www.bea.gov

Calculating an expenditure multiplier

▪ Use Yi = Ai + ciYi, dYi = dAi +cidYi Where Yi denotes income, Ai autonomous expenditure,

and ci the marginal propensity to consume for the ith income class and

Where I is the identity matrix and C the diagonal matrix of the ci

▪ Using Dynan (2012) estimate of income elasticity of consumption, e, to obtain MPC, i.e., MPC=e*APC 24

www.bea.gov

Alternative APCs over time and across the distribution

0.7

0.75

0.8

0.85

0.9

0.95

1

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

2008

2010

Average APC using CE data and NIPA, 1980-2010 PCE/Personal disposable income

Fisher, et al. APC

Using the quintile distributions of income and consumption in McCully (2012), we obtain a an expenditure multiplier of 5.75. A constant MPC across income categories yields a multiplier of 5.48 (for a .27 difference)

0.00

0.50

1.00

1.50

2.00

2.50

bottom 2nd middle 4th top

APC by Quintile, 2010

Fisher, et al. (adj)McCullyCE Published

25

www.bea.gov

Social Welfare Function: An application

▪ Consider µ(1-G) as the SWF (as in Sen (1973)); µ is the mean income and G in the respective Gini coefficient

▪ Similar to Jorgenson (1990), Jorgenson and Slesnick (2012) and Jones and Klenow (2011)

▪ Larger increases in income yield larger increases in SWF, while larger increases in inequality diminish increases in SWF.

26

www.bea.gov

Changes in income, inequality and SWF

-6

-4

-2

0

2

4

6

8

10

12

HouseholdIncome

NIPA Adjusted NIPA Adjustedwith all

imputations

Per-capita GDP

IncomeInequalitySWF

27

www.bea.gov

Conclusion and Future Work

▪ Almost 60 years ago, Kuznets (1955) stated: “Today, there is increased concern about the skewed income distribution, and the increase in skewness over time.”

▪ We constructed two straightforward ways to provide a distribution to NIPA Personal Income

▪ We show that many subjective decisions are part of the transformation ▪ Future work involves analysis of the matched household data with the

tax records to more completely measure income underreporting. ▪ Multiplier analysis will be improved by incorporating similar

decompositions of PCE and personal income that rely on the distribution of the household survey data (as in McCully (2012))

▪ The results in this paper may provide a framework for developing measures of median personal income and their distribution that could be produced on a regular basis.

28