income inequality in america: fact and fiction

8/12/2019 Income Inequality in America: Fact and Fiction

1/35

PublishedbytheManhattanInstitu

te

No.

1May

2014

INCOME INEQUALITY IN AMERICA

Fact and Fiction

e21ISSU

E

BRIEF

e21ISSU

E

BRIEF


2/35


3/35

Introduction

Diana Furchtgott-Roth, e21 Director, Manhattan Institute

Chapter 1: Should We Care About Income Inequality?

Scott Winship, Manhattan InstituteChapter 2: The Importance of Consumption for Measuring Inequality

Bruce Meyer, University of Chicago

Chapter 3: How the Tops Share of Income Changes with Comprehensive Measurements

Philip Armour, Cornell University

Chapter 4: How Changing Demographics Affect Inequality

Diana Furchtgott-Roth, Manhattan Institute

Chapter 5: Historical Perspectives on Income Inequality and Mobility

Gerald Auten, U.S. Department of the Treasury

Chapter 6: Relative and Absolute Mobility Across GenerationsScott Winship, Manhattan Institute

Chapter 7: Has Worker Compensation Tracked Productivity?

James Sherk, Heritage Foundation

Chapter 8: Inequality and Risk-Taking in a 21st Century Economy

Edward Conard, American Enterprise Institute

About the Authors

1

3

7

9

12

15

20

22

27

30

CONTENTS


4/35


5/35

Income Inequality in America: Fact and Fiction

The presence ofand, in some countries, increase inhouseholdincome inequality has become a flash point in public policy andpolitical discussion.

For Winston Churchill, such inequality was an unavoidable part of economiclife in capitalist societies. The main vice of capitalism, remarked the British

Prime Minister, is the uneven distribution of prosperity. The main vice ofsocialism is the even distribution of misery.

But for President Barack Obama, income equality is not only a pressingproblem, it is the defining challenge of our time. Against this backdrop,e21 brought together leading economists to provide a primer on ways tothink about income inequality.

The extent of inequality differs with the measure used. An InternationalMonetary Fund report published in February 2014, for instance, measuresinequality using market incomedefined as income before taxes on upper-

income individuals are removed and transfers to lower-income individualsadded back.

This pre-tax, pre-transfer measure of inequality is, however, misleading be-cause it fails to properly measure well-being. Upper-income individuals can-not spend the money that is taken away in taxes, so it gives them no benefit(other than, perhaps, higher social status). On the other hand, lower-incomeindividuals clearly benefit from more spending power with, among others, the

INCOME INEQUALITY IN AMERICA: Fact and Fiction

INTRODUCTIONDiana Furchtgott-Roth, e21 Director, Manhattan Institute


6/35

e21

Issue

BriefNo.

1

May 2014

2

academic research. In chapter 2, Bruce Meyer dis-cusses the strengths and weaknesses of income andconsumption data, showing that the living standardsof the poor are, in fact, better captured by the lat-terand that consumption inequality has risen lessthan income inequality.

Philip Armour follows, in chapter 3, by identifyingseveral shortcomings of conventional measures ofincome concentration, providing evidence that whenremedied, income concentration actuallyfell between1989 and 2007.

In chapter 4, I describe how demographic changesaffect income comparisons, showing that inequalityin per capita expenditures between the top and bot-

tom fifths has not grown in 25 years. In chapter 5,Gerald Auten uses IRS data to examine patterns ofeconomic mobility, exploring movement over indi-viduals lifetimes, historical trends, and mobility intoand out of, the top one percent.

In chapter 6, Scott Winship clarifies the distinctionbetween absolute and relative economic mobility,arguing that absolute intergenerational mobility re-mains strongand that relative mobility has changedless and may be closer to levels in other countries than

widely believed. In chapter 7, James Sherk disprovesthe conventional wisdom that worker compensationhas stagnated as productivity has increased.

And, in chapter 8, Edward Conard raises the radicaleffect technological innovations have had on incomeinequality, discussing the implications of this para-digm shift.

In compiling this volume, we hope to shed light onthis important subject by debunking some of thepopular myths about inequality.

Supplemental Nutrition Assistance Program, Medic-aid, housing vouchers, and unemployment insurance.As such, it would be inaccurate not to include thelatter in measures of their well-being.

Demographics also affect inequality. Things like in-creasing life expectancy, greater likelihood of divorce,and the rising percentage of births to single mothersall affect the distribution of inequality. For example,every time two earners marry or divorce, the distribu-tion of inequality is affected.

No less important in measuring inequality is therespective stage in the life cycle of different indi-viduals. A graduate student, say, embarking on hercareer has no income, and probably some debt, but

also a decent prospect of landing a job. If she mar-ries another similarly-placed, penniless graduatestudent, they might, before long, transform into atwo-income family located in the middle or upperincome quintiles.

This is the natural life-cycle progression, and thegraduate students income should not be a socialpolicy problem. Conversely, when the student retiresafter 50 years in a successful career, she might haveassets but little income and return back, again, to the

bottom quintile. This, likewise, is not a social policyproblem in need of correction.

Lack of mobility between income groups isa policyproblem, however, and we devote time in thesepages to discussing it: has mobility indeed changedover time, and, if so, is it intra-generational or inter-generational?

Scott Winship begins this volume by observing, inchapter 1, that much of the current conventionalwisdom on income inequality is not supported by


7/35


Among commentators today, primarily on the left,there is a sense that President Obama put it mild-ly when he declared inequality the defining challengeof our generation. Nobel Laureate Joseph Stiglitz haswritten that, with inequality at its highest level sincebefore the Depression, a robust recovery will be dif-ficult in the short term, and the American dream agood life in exchange for hard work is slowly dy-ing. New York Times columnist Paul Krugmanalso a Nobelistcalls inequality toxic.EconomistThomas Piketty, whoseincome concentrationfigures are ubiquitouslycited, recently inveighedagainst tax cuts by sayingthey will:

eventually con-tribute to rebuild aclass of rentiers in theUnited States where-by a small group ofwealthy but untal-ented children controls vast segments of the U.S.economy and penniless, talented children simplycannot compete.

But claims about the supposed harm done by rising

income inequality are rarely substantiated, and acomprehensive read of the evidenceas to inequal-itys consequences offers little cause for alarm. In-equality has been rhetorically linked to stagnantincomes among the poor and middle class, toslow economic growth, to diminished opportunityamong children lower down the income ladder, tomacroeconomic financial instability and household

indebtedness, and to political inequality. The stateof our knowledge about these links, however, sug-gests little reason to prioritize income inequality asa national problem.

Consider first the relationship between inequal-ity and the living standards of the poor and middleclass. It is widely believed that incomes below thetop have stagnated as the rich have pulled away, tak-ing an outsized share of income growth with them.

However, the truth isthat there is no incon-sistency between the topreceiving a large shareof income gains and thepoor and middle classseeing significant in-

come growth.

To be sure, income growthbelow the top has slowedsince the Golden Age ofthe 1950s and 1960s. As

the chart below shows, annual income gains among thebottom 80 percent of households were stronger thanamong the top five percent between the peak years of1948 and 1969, and gains were strongest among thebottom fifth (details on data sources here). Since the

1960s, income growth below the top has slowed. How-ever, an indication that rising inequality has not beenprimarily responsible for this slowed growth is the factthat diminished income growth preceded the rise in in-come concentration at the top. The 1970s were a lousydecade for the poor, middle class, and rich alike. Thereason? A slowdown in productivity across the industri-alized world.

Claims about the supposed harm

done by rising income inequality

are rarely substantiated,

and a comprehensive read of

the evidence as to inequalitys

consequences offers little

cause for alarm.

CHAPTER 1:SHOULD WE CARE ABOUT RISING INCOME INEQUALITY?Scott Winship, Manhattan Institute
http://opinionator.blogs.nytimes.com/2013/01/19/inequality-is-holding-back-the-recovery/?_php=true&_type=blogs&_r=0http://opinionator.blogs.nytimes.com/2013/01/19/inequality-is-holding-back-the-recovery/?_php=true&_type=blogs&_r=0http://krugman.blogs.nytimes.com/2013/09/11/toxic-inequality/http://www.nytimes.com/2003/04/20/business/economic-view-efficiency-and-equity-in-the-same-breath.htmlhttp://www.nytimes.com/2003/04/20/business/economic-view-efficiency-and-equity-in-the-same-breath.htmlhttp://www.brookings.edu/~/media/research/files/articles/2013/03/overstating%20inequality%20costs%20winship/overstating%20inequality%20costs%20winship.pdfhttp://www.brookings.edu/~/media/research/files/articles/2013/03/overstating%20inequality%20costs%20winship/overstating%20inequality%20costs%20winship.pdfhttp://www.nytimes.com/2003/04/20/business/economic-view-efficiency-and-equity-in-the-same-breath.htmlhttp://krugman.blogs.nytimes.com/2013/09/11/toxic-inequality/http://opinionator.blogs.nytimes.com/2013/01/19/inequality-is-holding-back-the-recovery/?_php=true&_type=blogs&_r=0


8/35

e21

Issue

BriefNo.

1

May 2014

4

Income inequality did not increase that much in the1970s, and while it increased thereafter, within the

bottom 80 percent, the rise in inequality was confinedto the 1980s. Overall, inequality within the bottom80 percent has increased only modestly since the1960s. In 1969, the middle fifth of households had anaverage income 2.5 times that of the bottom fifthwith the ratio rising only to 2.9 times larger in 2007.

Income concentration at the top rose after the 1970s,though even here, the increase shown in the chartis overstated. The figuresfrom Piketty and col-league Emmanuel Saezdo not account for public

transfers, the value of non-wage employer benefits,or redistribution occurring through the tax code.They also focus on tax returns instead of householdsand do not adjust for declining household size. Tworoommates constitute one household but two taxreturns, and teens and college students who workpart-time and file their own tax returns are count-ed independently of their parents. Estimates from

the Congressional Budget Office that correct theseshortcomings show smaller increases in income

concentration over time. Figures from Piketty andSaez that focus on earnings and exclude investmentincome indicate even smaller increases in incomeconcentration. While the figures in the chart abovesuggest that the share of income received by the topfive percent rose from 21 percent to 34 percent from1979 to 2007, the estimates using earnings showanincrease from 17 percent to 25 percent. (See alsoPhilip Armours contribution to this primer.)

Note, too, that incomes below the top 5 percent did

not stagnate. Incomes among the bottom fifth wereone-third higher in 2007 than in 1979, and thoseamong the middle fifth were 43 percent higher.

Turning to the relationship between inequality andeconomic growth, despite the attention given to anInternational Monetary Fund paper purporting tofind that countries with more inequality experience

Figure 1. Annual Household Income Growth, 1948-2007

Source: For income below the top five percent, 1948-69, Mean family income, by income group Table 2.1, The State of WorkingAmerica. Washington, D.C.: Economic Policy Institute, 2012, and author calculations. For 1969-2007, Current Population Survey andauthor calculations. For top five percent, Thomas Piketty and Emmanuel Saez, Income and Wage Inequality in the United States,1913-2002, in A.B. Atkinson and Thomas Piketty, eds., Top Incomes Over the Twentieth Century: A Contrast Between European andEnglish-Speaking Countries(Oxford: Oxford University Press, 2007), updated athttp://elsa.berkeley.edu/~saez/TabFig2012prel.xls, and author calculations. For details, see Scott Winship, Choosing Our Battles: WhyWe Should Wage a War on Immobility Instead of Inequality, Testimony Before the Joint Economic Committee, January 16, 2014.
http://www.jec.senate.gov/republicans/public/?a=Files.Serve&File_id=4472dcdb-4bc9-40e0-974d-dc050719891ehttp://elsa.berkeley.edu/users/saez/TabFig2012prel.xlshttp://www.imf.org/external/pubs/ft/sdn/2011/sdn1108.pdfhttp://www.imf.org/external/pubs/ft/sdn/2011/sdn1108.pdfhttp://elsa.berkeley.edu/users/saez/TabFig2012prel.xlshttp://www.jec.senate.gov/republicans/public/?a=Files.Serve&File_id=4472dcdb-4bc9-40e0-974d-dc050719891e


9/35


weaker recoveries from recessions, the academic lit-erature comes to no consensus. Studies are as likelyto find that more inequality corresponds with highergrowth, as they are to find a negative relationship.The IMF study was primarily focused on develop-ing countries, while including only a few industrial-ized nations in Asia. The liberal Center for AmericanProgress has published no fewer than three studiesconcluding that, among the studies which directlyexamine the question, there is little evidence thathigher inequality harms growth.

Of course, if inequality rises and economic growthdoes not increase, then the implication is that growthdoes not benefit the middle class or poor. But researchby sociologist Lane Kenworthy shows that across in-

dustrialized countries, increases in inequality do notcorrespond with lower median income growth. Astudy by Christopher Jencks and his colleagues sug-gests that in recent decades, greater inequality growthcorresponds with stronger economic growth, whichwould be consistent with increases in median income.

Income distribution is not a zero-sum game. Yeteven if possible, limiting income concentration after1979by keeping the shares received by the bot-tom, middle, and top at their 1979 allocation

might have reduced subsequent income growth. If itwould have done so by 0.5 percent per year or more,then preventing inequality from rising would havesuccessfully kept the share of income received by thetop from rising, but would have left the middle fifthno better off than they actually were in 2007. Theywould have received a larger slice than they actuallydid, but of a smaller pie.

What about the impact of inequality on opportu-nity? The conclusion of mobility expert and soci-ologist Michael Hout in 2004 remains true today:[The] literature to date has offered surprisingly littleevidence that links intergenerational difference andpersistence (mobility and immobility) to economicor other inequality. Economists Alan Krueger andMiles Corak have argued in recent years that the pos-itive association across countries between income in-equality and intergenerational mobility implies that

the former diminishes the latter. However, this cor-relation has not held up across labor marketswithinthe United States, suggesting that cultural and otherdifferences across countries may be responsible forany correlation between inequality and mobility. In-deed, the relationship is as strong between mobilityand population size as between mobility and incomeinequality, and there is no association between mo-bility and wealth inequality.

Furthermore, the mobility measure used by Kruegerand Corak indicates less mobility when inequalityincreases more. When a measure of mobility is usedthat is unaffected by changes in inequality and fo-cuses on whether parental income rank is related tothe income rank of adult children, Sweden and the

United States have the same mobility (despite hav-ing very low and very high inequality, respectively).The implication is that when inequality is not bakedinto the mobility measure, there may be little cor-relation between inequality and mobility, let alone acausal relationship.

Does income inequality lead to financial crises andindebtedness? While former Labor Secretary RobertReich and Raghuram Rajan, chairman of Indias cen-tral bank, have suggested that high and rising inequal-

ity led to both the Great Depression and the GreatRecession, the most rigorous research on this ques-tion indicates that this is a classic case of an omittedvariable. Economists Michael Bordo and ChristopherMeissner looked at financial crises across countriesand over time, and found that rising inequality wasincidental. Rather than causing crises, rising inequalitytends to co-occur with the inflation of credit bubbles.But it is the credit bubbles that lead to financial crises.

Another argument, made most prominently byeconomist Robert Frank, is that when inequalityincreases, people below the top feel pressured tooverspend to keep up with the Joneses. The not-quite-rich spend more to keep pace with the rich,the upper-middle-class follow suit, and so on, all theway down to the poor. States and counties that sawbigger increases in inequality during the 1990s alsosaw more growth in bankruptcy filings. However, we
http://www.americanprogress.org/wp-content/uploads/2013/12/Howell-Inequality-report.pdfhttp://www.americanprogress.org/wp-content/uploads/2013/12/Howell-Inequality-report.pdfhttp://www.degruyter.com/view/j/bejeap.2011.11.issue-1/bejeap.2011.11.1.2617/bejeap.2011.11.1.2617.xmlhttp://www.degruyter.com/view/j/bejeap.2011.11.issue-1/bejeap.2011.11.1.2617/bejeap.2011.11.1.2617.xmlhttp://economics21.org/commentary/great-gastby-curve-revisited-part-1http://economics21.org/commentary/how-ratchet-down-great-gatsby-curvehttp://economics21.org/commentary/how-ratchet-down-great-gatsby-curvehttp://www.nber.org/papers/w17896http://www.nber.org/papers/w17896http://www.nowpublishers.com/articles/review-of-behavioral-economics/RBE-0003http://www.nowpublishers.com/articles/review-of-behavioral-economics/RBE-0003http://www.nber.org/papers/w17896http://economics21.org/commentary/how-ratchet-down-great-gatsby-curvehttp://economics21.org/commentary/great-gastby-curve-revisited-part-1http://www.degruyter.com/view/j/bejeap.2011.11.issue-1/bejeap.2011.11.1.2617/bejeap.2011.11.1.2617.xmlhttp://www.americanprogress.org/wp-content/uploads/2013/12/Howell-Inequality-report.pdf


10/35

e21

Issue

BriefNo.

1

May 2014

6

do not know that the stronger rise in bankruptcy fil-ings in these counties was concentrated among thepoor or middle class.

In another study,middle class households in stateswith more inequality also spent more relative to theirincomes than their counterparts in other states. Thatcould have reflected greater spending out of housingwealth as income concentration bid up the value ofhomes, although middle-class households in high-inequality states were also more likely to say they wereworse off financially than a year ago. The specific ar-eas of spending that rose disproportionately amongthe non-rich in response to income gains at the topmostly involved personal appearance and home main-tenance, suggesting that if income concentration in-

creases spending, the additional dollars are spent ondiscretionary items rather than necessities. These factshardly constitute a case for public policy to interveneto save consumers from themselves.

Finally, concern about whether income inequalityleads to political inequality is also excessive relativeto existing evidence. Here, too, the alarmists caseis built on inapplicable research and selective cita-tion. Many commentators cite the work of econo-mists Daron Acemoglu and James Robinson, who

show that inequality in developing countries is as-sociated with the creation and maintenance of po-litical institutions that redistribute money into thehands of elites. However, Acemoglu and Robinsonprovide no evidence that this occurs in modern in-dustrialized democracies.

Commentators worried about inequality often citethe research of Larry Bartels and of Martin Gilens,who separately find evidence that political outcomesare more aligned with the preferences of the rich,than with those of the poor or middle class. Howev-er, Robert Erikson and Yosef Bhatti find no evidenceof unequal representation in research that directlyaddresses the Bartels paper.

In 2004, a task force on inequality convened by theprofessional association of academic political scien-

tists concluded, We know little about the connec-tions between changing economic inequality andchanges in political behavior, governing institutions,and public policy. More recently, a 2011 book sum-marizing a political science conference on unequalrepresentation, summarized the conference findings:

We discovered that no real consensus exists onhow different groups [including those definedby income] influence policy. Not only were theredebates about differences between groups, therewere also serious disagreements about whetherthese differences matter....[T]he conference madeclear that we do not yet have a good answer to thequestion of who gets represented.

The authors of the book found that the policy pref-erences of poor, middle class, and upper-incomeAmericans do not differ all that much, and wherethere are differences, they are long-standing onesthat have not changed as income inequality has risen.

The case that inequality has substantial costs is sim-ply overstated by conventional wisdom. That is notto say that new evidence will not emerge that couldchange this conclusionand it may be that despitethe weak evidence, rising inequality really has been

problematic. But policymaking must be based onevidence, not biases or hunches (unless, of course, itis argued that inequality is just wrong even if it hasno costs). Looking at the facts, it is difficult to seewhy we should focus on inequality over any numberof other potential policy issuesif we are concernedabout the poor and middle class, or the state of theeconomy, or of our democracy.

In this volume, we evaluate several facets of the in-come inequality debate. Empirical analysis showsthat many commonly accepted ideas about incomeinequality are false or overstated. The debate overeconomic mobilityand how income inequalitycontributes to itis an important debate. However,if policy recommendations are to be effective, theymust be informed by an accurate picture of the cur-rent situation.
http://www.nber.org/papers/w18883http://www.nber.org/papers/w18883


11/35


CHAPTER 2:THE IMPORTANCE OF CONSUMPTION FOR MEASURING INEQUALITYBruce Meyer, University of Chicago

The most well-known work on inequality has fo-cused on the distribution of wages, earnings, or tax-able income, leaving consumption out of the discussion.There are several reasons why consumption is a criticaltool in measuring inequality and well-being. Consump-

tion reflects typical income and is thus more permanentthan measured income, which often fluctuates fromyear to year. This permanence is because people tend tosmooth their consumption over time, as income fluc-tuates, by borrowing or saving. Measures of income alsodo not reflect all availableresources, including wealth,that improve well-being. Forinstance, more than 80 per-cent of people over 65 owna house and may make small

or no payments on it. This isnot factored into measures ofincome. Similarly, measuresof income do not reflect ac-cess to credit, which differsacross groups, nor do such measures reflect price chang-es in assets, such as houses and stocks.

There are also difficulties in effectively measuringincome for the purposes of inequality. Measures ofincome usually do not account for taxes and trans-fers, which have large effects on income inequality.For those near the bottom of the income distribu-tion, income tends to be underreported, partly dueto the exclusion of many non-cash transfers fromthe government.

There are also challenges to measuring consumptionaccurately, but many of the supposed weaknesses of

consumption data are overstated or wrong. At leastfor those at the bottom of the distribution, con-sumption is more accurately captured in householdsurveys than income.

In the Consumer Expenditure Survey, the accuracyof the consumption measure depends on what typeof consumption is being measured. For instance,the accuracy is high for rent, utilities, groceries, carsand gasoline. As might be expected, people are less

accurate in reporting theirconsumption on alcohol,clothing, and furniture.

My coauthor, James Sul-livan of Notre Dame, and

I used information from asubset of total consump-tion that includes im-portant spending catego-ries that tend to be well

reported, including housing and utilities, food athome, and transportation. These categories makeup most of non-medical consumption, and relativeto overall price changes, those for this core grouphave been small.

Even when using consumption rather than income,the question still remains: has inequality changedin the past 50 years? Looking at the ratios of theincomes of the 90th, 50th, and 10th percentilescompared to each other, the data do present a storyof generally widening inequality, increasing fromthe 1970s on, even when taxes and non-cash ben-efits are factored in. However, when these ratios are

The level of inequality is

much lower for consumption

than income, and since 1980,

consumption inequality hasrisen considerably less than

income inequality.


12/35

e21

Issue

BriefNo.

1

May 2014

8

considered using comparisons based on consump-tion, a very different story emerges. This shows thatinequality in consumption rose at a much slowerrate than income inequality over the past severaldecades. In recent years, even the direction of thechange has differed.

Since 2006, the ratio of the 90th percentile to the10th percentile shows rising inequality when in-comes are compared, but a declinein inequality whenconsumption is examined. What exactly caused thisshift is not entirely clear, but changes in housingand financial asset values can explain some of thepattern. Furthermore, one observes lower absolutelevels of inequality when looking at consumption.A similar phenomenon of lower inequality is seen

when comparing the 50th percentile and the 10thpercentile, or the 90th and 50th percentiles.

When talking about consumption, a common ques-tion is Have people over-consumed? This questionis largely irrelevant. For measurement purposes, itdoes not matter. A consumption measure shows uswhether a person drives a car or eats a meal regardlessof other factors, and spending cut-backs are record-ed as well. The financial crisis and recession forcedmany people to adjust their consumption plans

when their incomes and asset values changedthereare, perhaps, better questions to be asked when itcomes to how people are spending.

One such question is, Does the balance sheet ofthe population look worse than in the past? Ourresearch provides the answer. In these terms, the bot-tom 20 percent of the population has little assets ordebts, so there is little to examine. The middle 20percent has seen a rise in debt, but this level is now

well below its peak. Moreover, the net worth of thisquintile is also above what it has been in the past,except for the years around the peak of the housingbubble. It would seem, then, that the balance sheetof the population does not look worse than it has inthe past.

Over the past five decades, both income and con-sumption inequality have risen. The level of inequal-ity is much lower for consumption than income,and since 1980, consumption inequality has risenconsiderably less than income inequality. Incomeinequality has generally increased episodically, withconcentrated spurts in the late 1970s, early 1980s,and in the last several years. And since 2006, thoughincome inequality has been rising, consumption in-

equality has been falling.

The causes of these differences are somewhat un-clear. Demographic changes can account for some ofthe changes in consumption and income inequality,particularly in the 1980s, but account for few of thechanges overall. The quality of the income data atthe bottom may also explain some of the differences.Changes in asset prices could play the biggest rolein explaining the difference, at least in recent years.

As for inequality being the defining issue of our time,our research indicates that when looking at inequal-ity from a more holistic perspective, and when mea-surement tools are thoroughly analyzed, there maybe more facets to inequality than commonly con-sidered. While there is evidence of income inequal-ity increasing over the past five decades, the increasehas only partially affected consumption inequality,which is where policy-makers should concentratetheir efforts.


13/35


CHAPTER 3:HOW THE TOPS SHARE OF INCOME CHANGES WITHCOMPREHENSIVE MEASUREMENTSPhilip Armour, Cornell University

Few serious scholars believe that middle class andpoor households have seen the income growthexperienced by top earners in recent decades. Boththe ubiquitous estimates from economists Thomas

Piketty and Emmanuel Saez and figures from theCongressional Budget Office show dramatic in-creases in the share of income received by the richestone percent of Americans. Between 1979 and 2007,the Piketty-Saez numbers rise from 10 percent to 24percent, and the CBO share increases from 7 to 17percent. Attempts to deny that the top has pulledaway have, generally, been wholly unpersuasive.However, this does not mean the conventional wis-dom is correct.

My colleagues, Richard Burkhauser (Cornell Uni-versity) and Jeff Larrimore (Joint Committee onTaxation), and I use data from the Bureau of LaborStatisticss Current Population Survey to mimic theapproach used by Piketty-Saez, which focuses on theincomes on tax returns, before accounting for taxesand transfers. Doing so shows that from 1979-2007the bottom quintiles income fell 33.0 percent, themiddle quintiles income rose just 2.2 percent, andthe top quintiles income rose 32.7 percent. If thesenumbers are representative of the economic situa-

tion in the United States, an argument can be madethat there is cause for concern.

But are tax returns the best way to measure incometrends? Examining size-adjusted household incomeis far better. Tax units are neither individualsa married couple filing jointly is a tax unitnorhouseholds (two roommates filing returns constitute

two tax units). Furthermore, household needs havedeclined with household size.

Using households, and adjusting incomes for theirsize, lowers but does not eliminate the measuredincrease in income inequality. All groups experi-enced income growth with this methodology: theincomes of the bottom quintile rose 16.5 percent;those of the middle rose 20.2; and those of the toprose 41.0 percent.

But using pre-tax, pre-transfer incomealso knownas market incometo evaluate effects of moreprogressive taxation and redistribution is pointless.If only market income is used, then no matter how

much is redistributed through taxation and spend-ing, inequality will be unaffected. When taxes andtransfers are included in household income, it be-comes clear that government programs have beensuccessful in mitigating income concentration.

Non-cash transfers, including SNAP benefits (foodstamps) and government health coverage, financedby taxes on the wealthy, are major tools used to com-bat inequality. Taxes, it is true, have declined acrossthe board, increasing take-home pay. Additionally,

more and more compensation going to middle classworkers has come in the form of employer-providedhealth insurance.

Once these factors are taken into account, incomesof those in the bottom quintile improved 31.8 per-cent, those in the middle quintile saw incomes rise34.4 percent, and those in the top saw an increase
http://elsa.berkeley.edu/~saez/TabFig2012prel.xlshttp://elsa.berkeley.edu/~saez/TabFig2012prel.xlshttp://www.cbo.gov/publication/43373http://www.cbo.gov/publication/43373http://elsa.berkeley.edu/~saez/TabFig2012prel.xls


14/35

e21

Issue

BriefNo.

1

May 2014

10

of 54 percent. While inequality did increase, every-one is now substantially better off. For those at thebottom, an increase of 31.8 percent in income is fardifferent than a decrease of 33 percent (Piketty-Saezmethodology).

From 1989 to 2007 (again comparing peak years),the bottom fifth, middle fifth, and top 5 percent sawgains of 26 percent, 20 percent, and 17 percent, re-spectively, indicating a decline in inequality. A prob-lem with these estimates, however, is that they do notinclude any income from capital gains. Capital gainsderive from assets that appreciate in value, and sinceassets are especially unequally distributed, inequalityestimates that leave capital gains out are potentiallyproblematic. We can incorporate capital gains into our

income measure, startingin 1989, by imputingamounts to householdsusing information fromother datasets.

To facilitate comparabil-ity with the CBO andPiketty-Saez figures,gains that are both tax-able and realized must

be added in first. Theseare gains which result intaxable income directlyreceived as a consequence of selling assets in a givenyear. After doing so, the data show the top 5 percentpulling away from everyone else, even from 1989to 2007. The poorest fifth of Americans saw theirincomes rise by 28 percent, the middle fifth by 22percent, and the top 5 percent by 52 percent.

CBOs estimates confirm both the decline in poor-middle inequality and the disproportionate rise inincome at the top. After applying the same cost-of-living adjustment as in our paper, CBO income fig-ures show gains for the bottom fifth, middle fifth,and top 5 percent of Americans of 31 percent, 23percent, and 81 percent from 1989 to 2007. Whilethe CBO figures combine tax return and CPS data,Piketty-Saez rely entirely on tax return data. They

ignore the bottom and middle fifth but show an 87percent increase in the income of the top 5 percentof tax units. CBO and Piketty-Saez both showbigger income gains for the top 1 percent (116percent and 123 percent, respectively). The surveywe used cannot reliably capture changes in incomein the top 1 percent, but it is safe to say that if itcould, it would show a rise similar to that in theCBO figures.

But this is not the last word, either. The CBO andPiketty-Saez income figures are only able to accountfor capital gains that are both taxable and realized.We point out two big problems with this restriction.First, tax-exempt, realized capital gains are ignored,including those from the sale of homes. These con-

stitute a large share ofcapital gains received bythe non-rich, so ignoringthem overstates the risein inequality. Another is-sue related to tax exemp-tion is that savvy taxpay-ers at the top can altertheir asset allocations sothat more or fewer oftheir realized gains are

taxable in response to taxlaw changes.

Second, and more important, there is a conceptualproblem including realized capital gains in in-come, but not the gains that accrue on assets thatare held rather than sold. For one, the distinctionis immaterial. Gains that accrue each year add tothe resources available for consumption or saving,whether they are realized or not. No less than real-ized gains, accrued gains not realized constitute partof the annual flow of resources properly conceivedas income (as distinguished from the stock ofaccumulated resources properly thought of aswealth). In addition, investors strategically chooseto realize capital gains depending on the state of as-set markets and on changes in the tax treatment ofdifferent assets. Realization of gains accrued overmany years tends to show up in tax return data in

A shocking result emerges: from

1989 to 2007, the incomes of the

bottom and middle fifth rose

(by 13 percent and 6 percent),

but the income of the top 5

percent declined by 5 percent.

Inequalityeven between the

top and everyone elsefell.


15/35


lumpy ways, as Cato Institute scholar Alan Reyn-olds has argued. A sizable share of the capital gainsaccruing to middle class households builds up overadulthood in accounts, such as IRAs and 401(k)s,and is not realized until retirement.

When we incorporate estimates of all capital gainsaccrued by a household over the past yeartaxableor tax-exempt, realized or notfrom investments inpublic companies and housing into our findings, ashocking result emerges. From 1989 to 2007, theincomes of the bottom and middle fifth rise (by 13percent and 6 percent), but the income of the top 5percent declines by 5 percent. Inequalityeven be-tween the top and everyone elsefalls. The declineis even more pronounced when we incorporate gains

from privately-held businesses.

Should our results be discounted because they can-not capture the incomes of the very rich, as Saez ar-gues? This is surely a relevant question. Note, how-

ever, that the top 5 percents income growth takingonly realized capital gains into account is eliminatedby taking into account how much smaller total ac-crued gains were in 2007, than in 1989. At the veryleast, then, the income growth of the top 1 percent,or the top 1 percent of the top 1 percent, also wouldbe expected to be significantly lower after account-ing for accrued gains. Furthermore, our imputationof accrued gains draws from the Survey of ConsumerFinances, which is designed to give reliable estimatesfor the top 1 percent.

Is our research the final word on inequality trends?Of course not. Of necessity, we use non-ideal im-putation strategies to assign accrued gains to peo-ple. Our findings cannot tell us very reliably what

happened to incomes at, say, the 99.9th percentile.There is little to suggest, however, that the ideal setof estimates would look qualitatively different fromour results. The rise in income concentration hasbeen drastically overstated.


16/35

e21

Issue

BriefNo.

1

May 2014

12

CHAPTER 4:HOW CHANGING DEMOGRAPHICS AFFECT INEQUALITYDiana Furchtgott-Roth, e21 Director, Manhattan Institute

Economists often divide households into incomequintiles (fifths) and measure the differences intheir incomes. However, the demographic charac-

teristics of these quintiles have been changing overtime, so comparisons of quintiles are misleading.Quintiles differ in the number of people per house-hold, as well as in the number of earners per house-hold. Table 1 shows that in 2012, households in thelowest fifth had an average of 1.7 people, and in halfthese households, there were no earners. The highestfifth, however, had 3.1 persons per household, withtwo earners.The lowest-income group contains at least three

significant groups of individuals. Some have lowincomes because of lack of employment and aresearching for jobs, or better-paying jobs. A secondgroup comprises elderly people who may have smallamounts of retirement income, but substantial as-sets, such as stocks and a home. These individualsare not in the labor force. A third group consists ofstudents or recent graduates whose education lev-

els ensure that they will have a prosperous future.Clearly, the first group is a social problem in need ofa solution, but not the other two.

In 1990, median income for a family with one earn-er was about $41,800. In 2012, median income forthat one-earner family rose to about $43,300, a 4percent difference. But the increase between a familywith two earners in 1990 and 2012 was far greater.That familys income rose from about $71,000 toabout $82,600, a 16.5 percent difference, resultingin a measured increase in inequality.

This reflects the contribution of the second earner,

which comes from increasing womens wages in thejob market, as young women have invested in theireducation in preparation for a full-time career. Ifthere were more one-earner households, the distri-bution of income would be far more even.

Another change is the shrinkage in household sizeat the bottom of the income scale, adding to a false

Figure 2. 2012 Consumer Units by Income Quintile

Source: U.S. Department of Labor, Bureau of Labor Statistics, Consumer Expenditure Survey, September 2013.

2.0


17/35


perception of increased inequality. This is due tothe increased longevity of todays seniors and to thehigher numbers of divorced people and single-par-ent households.

In 1960, 13 percent of households had just one per-son. By 2011, 27 percent of households, more thandouble the previous share, had one person.

It is notable that 30 percent of female householdswithout a husband are living in poverty. In contrast,only 7 percent of married couples and 17 percent ofmale households without a wife are poor.

Census data in Table 2 showthat men and women living

alone are most likely to be inthe lowest-income quintiles.Some 46 percent of womenliving alone were in the bot-tom quintile in 2012, and72 percent of women liv-ing alone were in the bot-tom two quintiles. Only 3.4 percent of women livingalone were in the top quintile. The trends are similarfor men. Some 60 percent of men living alone were inthe bottom two quintiles, and only 6.8 percent were

in the top quintile.In contrast, married couples are more likely to be inthe top quintiles. Some 32 percent of married cou-

ples were in the top quintile, and 58.4 percent werein the top two quintiles.

Another factor, which can influence measures of in-equality, is change to the tax code. The Tax ReformAct of 1986 lowered the top individual tax rate to28 percent, and the corporate rate to 35 percent.In 1986, the top individual rate was 50 percent,and the top corporate rate was 46 percent, so smallbusinesses would pay tax at a lower rate if they in-corporated and filed taxes as corporations. With theimplementation of the Tax Reform Act of 1986,the top individual tax rate of 28 percent meant that

small businesses were of-ten better off filing underthe individual tax code.

Revenues shifted from thecorporate to the individualtax sector. In the late 1980sand 1990s, that made it ap-pear as though people hadsuddenly become better offand income inequality had

worsened. This had not happened; rather, incomethat had been declared on a corporate return wasbeing declared on the individual return. This makesany comparisons between pre-and post-1986 re-

turns meaningless.

A more meaningful measure of inequality thatavoids changes in tax laws and changes in demog-

Figure 3: Percentage of Households within Each Income Quintileby Type of Household, 2012

Differences in per-person

spending, from the lowest-income fifth to the highest,

are not different from 25

years ago.


18/35

e21

Issue

BriefNo.

1

May 2014

14

raphy comes from an examination of spending, asBruce Meyer notes in chapter 2.

Differences in per-person spending, from the lowest-income fifth to the highest, are not different from25 years ago. These measures of spending show lessinequality than do measures of income. Spendingis vital because it determines our current standardof living and our confidence in the future. It showshow much purchasing power individual Americanshave.

I calculate spending on a per-person basis in order toproduce comparable measures. These data are con-verted into 2012 dollars using the Bureau of LaborStatistics Consumer Price Index for all urban centers.

It is important to compute spending on a per-personbasis because the number of people in a householdvaries by quintile. For a given level of income, a fam-ily is better off with fewer people.

Table 3 shows that the average annual spendingfor a household in the lowest quintile in 2012 was

$13,032 per person. In contrast, the average spend-ing for a household in the top quintile was $32,054per person.

On a per-person basis, the new Department of La-bor numbers show that in 2012, households in thetop fifth of the income distribution spent 2.5 timesthe amount spent by the bottom quintile, as can beseen in Table 3. That was about the same as 25 yearsago. There is no increase in inequality. In addition,the overall level of inequality is remarkably small. Aperson moving from the bottom quintile to the topquintile can expect to increase spending by only 146percent.Many commentators today bemoan a supposed in-

equality in the United States. Much of this concernis a problem in search of reality, caused by prob-lems of measurement and changes in demographicpatterns over the past quarter-century. Governmentdata on spending patterns show remarkable stabilityover the past 25 years and, if anything, a narrowingrather than an expansion of inequality.

Figure 4. Annual Expenditures by Income Quintile 2012


19/35


CHAPTER 5:HISTORICAL PERSPECTIVES ON INCOME INEQUALITY

AND ECONOMIC MOBILITYGerald Auten, U.S. Treasury

Widely-cited figures from Thomas Piketty, Em-manuel Saez, and the Congressional BudgetOfficeshowing that the top one percents share ofincome has increased over timehave been influen-tial in debates over income inequality. The CBO trendis slightly lower than the Piketty-Saez trend because ituses a broader measure of income, adjusts for fam-ily size and makes other important improvements. Ingeneral, most studies have found a long-term trend ofrising inequality. RichardBurkhauser, Philip Ar-mour, and Jeff Larrimore(in research described inchapter 3) found thatthe trend flattens out af-

ter 1990 once employerprovided health care isaccounted for as part ofearnings.

These studies comparecross-section views of thepopulation over time. The actual experiences of in-dividuals over time can be quite different. Thus it isimportant to also consider income mobility, variabil-ity, and other aspects of income dynamics over time.

Increasing inequality potentially affects mobilitybecause it means that the steps on the incomeladder are further apart, making upward mobilitymore difficult and potentially leading to reducedmobility over time. The ratio of income cutoffs forthe top one, five and 20 percent income classesto median income has generally grown over time.

For example, the income cutoff for being in thetop one percent rose from about 6.5 to 8.8 timesmedian income, from 1987 to 2011. Similarly, theratio rose from about 3.0 to 3.7 for the cutoff forthe top five percent.

Because of the discussions about widening incomegaps and the fact that the experiences of individu-als and families can differ from cross-section results,

it is important to lookat income mobility andhow this has changedover time.

The income measure

used in this analysis isintended to provide abroad measure of currentyear cash income usingconsistent tax data start-ing in 1987. In additionto the regular income

that is reported on returns, the measure adds backnon-taxable social security benefits, which is the sin-gle largest transfer payment program, and tax-exemptinterest. It also adjusts for net operating losses, which

are really business losses in previous years, and makesother adjustments to arrive at this years cash income.

While the tax filing rate for individuals age 25 to 65is quite high, one problem with using tax data is howto account for non-filers. For the results from the2013 paper (intergenerational mobility), income ofnon-filers was estimated using information returns

A significant part of the income

mobility story is the life cycle of

income: incomes are typically

lowest for people in their firstjobs, rise with seniority and

promotions, and then decline

after retirement.


20/35

e21

Issue

BriefNo.

1

May 2014

16

filed with the IRS (i.e., W-2s, K-1s, and other types

of income information returns).

It is common to examine income mobility by track-ing peoples positions in the income distributionover a 10-year period. There are three ways of look-ing at mobility: mobility relative to the comparablepopulation; mobility relative to the initial sample;and absolute income mobility. Figure 5 shows theresults of the first approach by tracking individuals,age 25 and older, from their initial income quintilein 1996 to their 2005 quintile. In this figure, theyare compared to the population that was 25 and

older in 2005.

Auten and Gee (2009) found that 43.7 percent ofthose initially in the bottom quintile were in the bot-tom quintile 10 years later as well (represented in the

upper left cell). This means more than half (56.3 per-

cent) moved up to a higher income quintile. About4.5 percent had moved to the top quintile. About 69percent of the top income quintile remained in thetop income quintile again 10 years later.

Of those that were in the top 1 percent in 1996,fewer than half, only 41.5 percent, were there 10years later. In the tax data, the analysis was able tolook at the top 0.1 percent and the top 0.01 percent,where the percentages of people who remain in thetop groups were even lower (less than 25 percent forthe top .01 percent). This suggests that top income

earners are not a static group, year-after-year.

Notice that the totals in the bottom row are notequal to 20 percent for each quintile. This is be-cause there are new people in the population 10

Figure 5. 10-Year Mobility Relative to the Total Population,Age 25+, 1996-2005

Source: Auten and Gee, National Tax Journal, June 2009.

Figure 6. 10-Year Mobility Relative to the Original Sample,Age 25/35+, 1996-2005



21/35


years later: new immigrants and new 25 year-oldswho, typically, have lower incomes. In addition, thepopulation is growing 1 percent per year, because ofimmigration and births, so that there are more slotsin the top 1 percent. As a result, those in the initialpopulation tend to move up in the overall distribu-tion: 27.6 percent of the initial population was inthe top income quintile by 2005.

Figure 7 again compares everyone 25 and olderin 1996but instead compares them to the same

group 10 years later. Under these conditions, it ismore difficult to move up in the population becausethere are no new entrants to the comparison popula-tion. The numbers are going to show more peoplestaying at the bottom, but also fewer people stayingat the top. This is the traditional way mobility stud-ies have been done, but it ignores population growthand new entrants to the population.

In order to see whether real, family-size adjusted in-comes are increasing or decreasing, we need to con-sider the third way we looked at mobility: absoluteincome mobility. As shown in Figure 8, real incomesrose 77 percent for people initially in the bottomquintile and by 8.6 percent for those initially in thetop quintile. The higher you are in the distribution,the less your relative income increase. Real incomesof those initially at the top of the distribution tendto decline. The real income of the median taxpayer

in the top 1 percent fell by nearly 31 percent overthis period. For those initially in the top 0.1 percentand 0.01 percent, the drops in median income areeven more dramatic: around 70 percent for the top0.01 percent.

The median real income rose by 22.7 percent overthis period, in contrast to some comparisons of cross-sections over time. Thus, real incomes were found tobe rising for a majority of the population when theincomes of specific individuals were examined.

A significant part of the income mobility story is thelife cycle of income: incomes are typically lowest forpeople in their first jobs, rise with seniority and pro-motions, and then decline after retirement. That iswhy we would expect people age 46 to 55 to be over-represented in the top income groups.

One way of seeing the effects of the life cycle of in-come is to follow the Baby Boomers, Generation Xand the other generations as some of them occupypositions in the top one percent. Back in 1987, 79percent of the top 1 percent was occupied by theGreatest Generation and the Silent Generation. Overtime their share declined, however, and was down to22 percent by 2010. On the other hand, in 1987,the Early Baby Boomers (then 32 to 41) accountedfor only 16 percent of the top classthough theirshare rose to a peak (reaching 33 percent) in their

Figure 7. Absolute Income Mobility, Age 25+, 1996-2005

Source: Auten and Gee, National Tax Journal, June 2009.Notes: Cash income is adjusted for family size and inflation. Primary and secondary taxpayersare followed separately.


22/35

e21

Issue

BriefNo.

1

May 2014

18

early 50s. In 2010, the combined early and laterBoomer generations occupied 59 percent of the topone percent. The changing occupation of the top bynew generations provides another illustration of theturnover at the top of the income distribution.

In 2011, 3.2 percent of Baby Boomers were in the top1 percent. (As a ratio, this means Baby Boomers were3.2 times the random probability of being in the top1 percent.) In 2011, Baby Boomers, then in the 56-65age group, were just past their peak earnings. Thus, ifwe are looking for villains, we can blame the BabyBoomers: they have more than their fair share of topincomes, at least for now. This is another way to seethe effect of the life cycle of income on ones positionin the overall income distribution.

Mobility is mostly concerned with longer-termmovements, but we also want to think about short-term income variability and turnover at the top aswell. Forty-one percent of those in 2005 were thereagain in 2010, but only 25 percent of them had beenthere every year. Most people in the top one percentin a given year are likely to be there only once, or fora few years, and others bounce in and out of the top

one percent. Its not, in short, the same people in thetop one percent every year.

One of the important mobility questions is whethermobility has decreased as income gaps have widenedand the steps on the ladder are further apart. In anexamination of two 10 year periods, 1987 to 1996and 1996 to 2005, there was identical mobility fromthe bottom quintile in the two periods, as 43.7 per-cent in the bottom quintile remained there after 10years. In other words, 56.3 percent moved up. Over-all, there is slightly more upward mobility into thetop income groups in the more recent period, eventhough the income gaps were wider. For example,2.6 percent from the bottom quintile rose to the topquintile, as compared to 2.1 percent in the earlier

period. But since this earlier analysis did not haveaccess to the information returns of non-filers andthe differences are small, it is safer to conclude thatincome mobility has basically stayed the same overthe two periods.

How is it possible that relative income mobilityremained the same even though the income gapswidened over time? Figure 8 shows the percentage

Figure 8. Absolute Income Mobility Over Time: 1987-1996 vs 1996-2005



23/35


changes in median cash income in the two time pe-riods by income quintile and for the top 1 percent.In every quintile, the percentage increases in realfamily-size adjusted income were larger (or more

positive) in 1996-2005, than in the earlier period.The change in median income for those in the topquintile went from negative to positive. In otherwords, absolute income mobility increased as therewas more upward change in real income in the morerecent periodand this offset the wider steps in theincome ladder. As a result, relative income mobilityremained about the same.

The results in Auten and Gee (2009) are not the onlyones to find income mobility relatively unchanged.

At a presentation at the recent American EconomicAssociation meetings (available as NBER WorkingPaper 19844), Raj Chetty and his co-authors pre-sented results indicating that intergenerational mo-bility has remained relatively unchanged among the

more recent birth cohorts. Adding their results tothose of earlier studies, they conclude that intergen-erational mobility did not change significantly be-tween the 1950 and 1970 birth cohorts.

Conclusions

It is important to keep in mind that peoples incomesand positions in the income distribution spectrumcan change considerably over time: over a life cycle,across generations, and even in the short-term. Manystudies have shown that there is considerable incomemobility and opportunity for low-income people tomove up in the income distribution. While the out-comes are not fully distributed equally, many people

in the bottom do indeed move upand some moveup to the top income groups. People looking at thesame data may, however, disagree on whether the ob-served income mobility is sufficient or fair, as well ason what policies are most appropriate.


24/35

e21

Issue

BriefNo.

1

May 2014

20

CHAPTER 6:RELATIVE AND ABSOLUTE MOBILITY ACROSS GENERATIONSScott Winship, Manhattan Institute

Intergenerational mobility has emerged in recentyears as a bipartisan policy priority. To the extentthat upward mobility rates are lowrelative to thepast, to other countries, or simply to our intuitionsabout appropriate levelsthe health of the Ameri-can Dream is called into question. However, thereare many misperceptions about the state of economicmobility in the United States, and a fuller apprecia-tion of the various facetsof the issue can aid poli-cymakers as they devel-op proposals to promoteupward mobility.

While it is often useful

to think about mobil-ity over the course of apersons adult life (see Gerald Autens contributionin Chapter 5), discussions about opportunity gener-ally have in mind intergenerational mobility: howmuch adult outcomes do, or do not, depart fromthose of their parents. Even focusing on intergenera-tional mobility leaves open what outcomes shouldbe considered. Here I will discuss mobility in termsof family income, but mobility researchers have alsoconsidered earnings, educational attainment, occu-

pational status, and wealth, among other outcomes.

Even restricting the focus to income still leaves twoways of thinking about mobilityin relative or abso-lute terms. Relative mobility is about how the rank-ing of adults against theirpeers is (or is not) tied to theranking of their parents against their peers. That is tosay, ignoring dollar amounts, did adults who rank high

or low in the income distribution also have parentswho ranked high or low? Careful research by the PewEconomic Mobility Projectdivides adult and parentalincome distributions into five equally-sized groups, orquintiles, ranked by income adjusted for family size.It then asks how likely it is that children starting ina given quintile end up in each quintile themselvesas adults. If there were no connection between par-

ent and child incomes,then a child growing upin any of the quintileswould have a 20 percentchance of ending up inany of the quintiles asadults. And for todays

forty-somethings whogrew up in the middle

fifth around 1970, that is close to what we see: 19percent ended up in the top fifth, 23 percent in themiddle fifth, and 14 percent in the bottom fifth (Fig-ure 3 in the Pew report).

But children who grow up poor or rich see morelimited mobility. Among those raised in the bot-tom fifth, 43 percent remain there as adults. Just 30percent made it to the top three-fifths (whereas 60

percent would have if parental income had had norelationship to child income), and only 4 percentmade it to the top fifth. Mobility among todaysadults raised in the top fifth displays the mirror im-age: 40 percent remain at the top, 37 percent fall tothe bottom three-fifths, and only 8 percent fall tothe bottom-fifth. (These figures indicate less mo-bility than in Gerald Autens contribution because

While relative mobility in

the United States appears

uninspiring, the same cannot be

said of absolutemobility.
http://www.pewstates.org/uploadedFiles/PCS_Assets/2012/Pursuing_American_Dream.pdfhttp://www.pewstates.org/uploadedFiles/PCS_Assets/2012/Pursuing_American_Dream.pdfhttp://www.pewstates.org/uploadedFiles/PCS_Assets/2012/Pursuing_American_Dream.pdfhttp://www.pewstates.org/uploadedFiles/PCS_Assets/2012/Pursuing_American_Dream.pdfhttp://www.pewstates.org/uploadedFiles/PCS_Assets/2012/Pursuing_American_Dream.pdf


25/35


they average multiple years of income together tonet out volatility.)

While relative mobility in the United States appearsuninspiring, the same cannot be said of absolutemobility. Absolute mobility ignores rankings andsimply considers whether adults tend to have higher,size-adjusted incomes than their parents did at thesame age, after taking into account increases in thecost of living. The answer is, unambiguously, yes.Fully 84 percent of todays forty-somethings havehigher size-adjusted family incomes than their par-ents did at the same age (Figure 1). Because poorparents have especially low income, adults who grewup in the bottom fifth are most likely to experienceupward absolute mobility; 93 percent of them have

higher incomes than their parents, compared with70 percent of adults who grew up in the top fifth.

On the one hand, upward absolute mobility oughtto affect how we think about limited relative mobil-ity. A person stuck in the bottom fifth may endup much better off in absolute terms than her par-ents were. A forty-something in a family of four to-day could have $56,000 in income and be in thebottom-fifth of size-adjusted income (Figure 4).Butthat income would have put them squarely in the

middle-fifth of size-adjusted income among parentsaround 1970. The richest forty-something in thebottom-fifth today is 85 percent richer than the topparent in the bottom-fifth was back then. Similarly,the forty-something, in the middle of the middle-

fifth today, is 89 percent richer than the parent, inthe middle of the middle-fifth of parents, was then.

On the other hand, even though those stuck at thebottom are better off than their parents were, theystill occupy the lowest rungs of the economic lad-der. If todays security guards or food service workerswanted to be lawyers or architects growing up butwere unable to, that should temper our enthusiasmfor their upward absolute mobility.

Of course, as many observers have pointed out, thefact of limited relative mobility does not necessarilyindicate that opportunity is unequally distributed,and it is difficult to objectively assert what the idealrate of upward or downward mobility would be.

Reihan Salam put it well, noting that a world of per-fect relative mobility is one in which no matter howhard you work to provide your children with everyadvantage in life, theyre just as likely to sink to thebottom of the heap as to rise to the top.

Still, relative mobility has not improvedover time aswe have become richer, and the United States has nobetter relative mobility than other industrialized na-tions that are less wealthy than we are (and probablyworse mobility than some). No one in the middle and

upper-middle classes would accept it if their childrenhad a 70 percent chance of dropping out of the mid-dle class. We should resist accepting that poor chil-drenwho do not choose their parentshave onlya 30 percent chance of making it to the middle class.
http://www.pewstates.org/uploadedFiles/PCS_Assets/2012/Pursuing_American_Dream.pdfhttp://www.pewstates.org/uploadedFiles/PCS_Assets/2012/Pursuing_American_Dream.pdfhttp://economics21.org/commentary/persistent%E2%80%94if-insufficient%E2%80%94american-dreamhttp://economics21.org/commentary/how-ratchet-down-great-gatsby-curvehttp://economics21.org/commentary/how-ratchet-down-great-gatsby-curvehttp://ftp.iza.org/dp1938.pdfhttp://www.pewtrusts.org/uploadedFiles/wwwpewtrustsorg/Reports/Economic_Mobility/PEW_EMP_US-CANADA.pdfhttp://www.pewtrusts.org/uploadedFiles/wwwpewtrustsorg/Reports/Economic_Mobility/PEW_EMP_US-CANADA.pdfhttp://ftp.iza.org/dp1938.pdfhttp://economics21.org/commentary/how-ratchet-down-great-gatsby-curvehttp://economics21.org/commentary/persistent%E2%80%94if-insufficient%E2%80%94american-dreamhttp://www.pewstates.org/uploadedFiles/PCS_Assets/2012/Pursuing_American_Dream.pdfhttp://www.pewstates.org/uploadedFiles/PCS_Assets/2012/Pursuing_American_Dream.pdf


26/35

e21

Issue

BriefNo.

1

May 2014

22

CHAPTER 7:HAS WORKER COMPENSATION TRACKED PRODUCTIVITY?James Sherk, Heritage Foundation

The debate over the minimum wage really gets tothe heart of one of the basic differences in howthe Political Left and Right see the world. It is, infact, one of the core misperceptions underlying the

Lefts economic prescriptions.

I read a recent New York Times op-ed describingwhy we need to raise the minimum wage to $10.10an hour. Figure 9 presents the core of the argument.Productivity since 1973 has gone up 100 percent,

according to the Bureau of Labor Statistics. Whenyou adjust payroll survey wages for inflation usingthe CPI, it looks as if average wages have gone downabout 7 percent.

The core of the New York Times argument is: busi-nesses have been making more and more money;workers have become more productive but are notsharing in the fruits of their labor; instead, greedycapitalists have expropriated it; something has to be

Figure 9. Comparing Productivity with Hourly Wages, Inflation-Adjustedwith the Consumer Price Index (CPI)


27/35


done so that companies give workers the raises theyhave earned. This is not welfare, since businesses areunfairly keeping what workers have earned fromthem. The left repeats these claims over and overagain. Productivity has gone up, but workers do notenjoy the fruits of their labor. So the governmentmust step in. This would be a compelling argu-mentif only it were true.

This analysis makes an apples to oranges compari-son. These figures come from different data sets thatmeasure different things, using very different meth-odologies. What I find in Productivity and Com-

pensation: Growing Together is exactly what manyof the Federal Reserve Banks, Martin Feldstein, andeven liberal economist Dean Baker have found.When researchers compare the numbers carefully,they canmake an apples to apples comparison. Suchcomparisons find pay and productivity track eachother very closely.

One thing missing from Figure 9 is that the pay fig-ure only counts about 60 percent of workers in theeconomy. Basically, it only measures hourly earners,not salaried employees. Also, there is something ex-cluded from hourly wages: benefits. Figure 10 showswhat happens when one includes total compensa-tion, as well as the average hourly compensation ofthe entire economy. It shows that total compensationhas not gone down, but actually risen 30 percent.

However, this adjusts for inflation using the Con-sumer Price Index. The CPI has a lot of problems,including formula issues. Indeed the CPI uses what

econometricians and statisticians know to be a lessaccurate methodology when calculating inflation.If analysts use a better measure of inflation, the Per-sonal Consumption Expenditures Price Index, thiseliminates some, but not all, of the bias in the infla-tion measurements. Using the PCE to calculate total

Figure 10. Comparing Productivity with Total Compensation, Inflation-Adjusted with the Consumer Price Index (CPI)


28/35

e21

Issue

BriefNo.

1

May 2014

24

compensation, as shown in Figure 11, finds average

hourly compensation has increased 56 percentvery different from the 7 percent drop the Left cites.

But still, even the PCE is not strictly comparable.The Bureau of Labor Statistics uses another measureof inflation to calculate productivity changes overtime. The Bureau of Labor Statistics, quite sensibly,measures productivity inflation using changes in thecosts of the goods a company sells.

If a person runs ACME widget factory and sellswidgets in America, it really does not matter to that

person how the price of oil imported from SaudiArabia changes. That has no impact on their abilityto pay employees wages or benefits. The only thingthat matters, in terms of compensation decisions, ishow much they can sell widgets for. The price ofwidgets determines the productivity of employeesat this company. If people will pay more money for

the widget, then effectively workers have become

more productive. If people pay less for the widget,the value of their labor has gone down. Changes inprices of imported foreign goods really have no im-pact on the ability of companies to compensate theiremployees.

The BLS calculates inflation in its productivity fig-ures by only looking at changes in the prices of goodsthat American workers produce. These prices differfrom the price of goods Americans consume becauseof imports and exports.

BLS does this with an inflation measure called theImplicit Price Deflator for Non-Farm Business. Basi-cally, it only looks at the changes in goods producedby non-farm businesses in the United States. Asshown in Figure 12, total compensation, on an ap-ples to apples comparison, has increased 77 percentover the past 40 years. This is, again, very different

Figure 11. Comparing Productivity with Total Compensation, Inflation-Adjustedwith the Personal Consumption Expenditures Price Index (PCE)


29/35


from the very first slide which showed wages falling7 percent. Even this remaining 23 percent gap is ex-aggerated, for two reasons. The government produc-tivity is mismeasured and overstated, because BLS isdoing an incomplete job accounting for the prices ofimported goods that have been used in production.Consider a widget factory that starts using cheaperinputs from China. The BLS does not capture thefull cost-savings to employers. It appears to BLS thatfactories are producing more widgets at lower cost.This looks like increased productivity. However, thegains are coming from trade and the lower pricedforeign goods, not worker productivity. This may ac-count for about half of the remaining 23 point gapseen in Figure 12.

Another factor that has changed is depreciation. To-day, businesses use a lot more computers and robotsin production than 40 years ago. Buildings last for awhile, while depreciation rates on buildings have notchanged much. If a company built a manufacturing

factory 20 years ago, that business could still use thesame building and some of the heavy machinery to-day. What about computers? Would anyone want touse a 1994 computer to run a manufacturing facili-ty? Computers and software depreciate. They do notlast as long and they become obsolete faster. Evenif a manufacturer had a 1994 computer in perfectcondition, he still would not want to use it.

Depreciation rates have increased. But BLS measuresproductivity on the basis of gross productivity. Theydo not factor in changes in depreciation rates, onlyhow many gross widgets get produced. If a companyhas to replace computers and machines every fiveyears instead of every eight years, this does not af-fect gross productivity. However, when it comes to

measure earned income, whether of investors, busi-ness owners, or workers, what matters is individualnet productivity: what remains after the machineryand equipment used up in the process get replaced.Depending on how it is measured, depreciation ac-

Figure 12. Comparing Productivity with Total Compensation, Inflation-Adjustedwith the Implicit Price Deflator (IPD)


30/35

e21

Issue

BriefNo.

1

May 2014

26

counts for another five points or so of the remaininggap between productivity and compensation.

So even the much smaller gap remaining in Figure12 is exaggerated. Workers are indeed enjoying thefruits of their labor.

This is both good and bad news. Good because itmeans companies are not exploiting their workers.Bad because it means there are no easy solutions. Ifthe minimum wage could simply be raised to $10 anhour, and people couldjust tell businesses, Youare making more, youhave tons of profits, justpay more out of your

profits, this would beeasy to fix by changingthe law.

In the actual world, thereason minimum wageworkers do not earnmuch is that workers inthose jobs have not become that much more pro-ductive over time. The data goes back to 1987, andif productivity in the fast food sector is examined,

pay and productivity track almost one to one. Both

have gone up about ten percent in inflation adjustedterms. If businesses have to pay employees $10 anhour when employees do not produce that value,they will spend a lot of time and money investing inlabor-saving technologies and finding ways to makedue with fewer workers.

The actual way to increase living standards is by in-creasing productivity. Things such as charter schoolsincrease their students productivity and economicmobility. There are studies about people who went

to charter schools, inthe mid-and late-1990s,and who are now intheir 30s. Such peopleare experiencing mea-

surably higher incomesthan their peers whodid not attend charterschools. The govern-ment should pursuethese sorts of solutions.Make workers moreproductive, and com-

petitive forces in the economy will force business topay more. It would be nice if we could just raise theminimum wage and give workers morethe world,

alas, does not work that way.

The left repeats these claims over

and over again. Productivity has

gone up, but workers do notenjoy the fruits of their labor.

So the government must step

in. This would be a compelling

argumentif only it were true.


31/35


CHAPTER 8:INEQUALITY AND RISK-TAKING IN A 21st CENTURY ECONOMYEdward Conard, American Enterprise Institute

Understanding inequality requires proper mea-surement of both income and consumption,along with an understanding of the ever-changing de-terminates of income and its distribution. These de-

terminates have changed significantly since the 1950s.

Information technology has opened an enormouswindow of investment opportunities and changedthe nature of investmentfrom labor and capital-intensive manufacturing carried out by large compa-nies, to idea-intensive opportunities like WhatsAppand Instagram. These opportunities require littleunskilled labor and capital, and can be created andowned by individuals.

The rewards for properly trained talent have risendespite a large productivity-driven increase in theirsupply. This indicates investment opportunities havegrown even faster than the productivity-enhancedsupply of talent.

At the other end of the pay scale, circumstances thatincreased middle class pay have run their course. Wesaturated the population with education and discov-ered a large pool of talented, but uneducated workerswhose productivity and pay rose substantially once

educated. Today, the potential for further saturationis smaller. When technology hollowed out agricul-ture, it drove the rural population to the cities wherethey became much more productive. That one-timemigration is over. And while the capital intensity ofmanufacturing continues to rise, it is no longer in-creasing the productivity of a growing proportion ofunskilled workers.

At the same time, the United States has moved froma shortage of unskilled workers to a surplus. A lackof births during the Great Depression gave way tothe baby boom after World War II and increased

participation of women in the workforce. Over thisperiod, a growing trade deficit exported jobs, andmassive migration to the U.S. added to the sup-ply of domestic workers. Today, the United Stateshas 37 million foreign-born adults and 16 millionnative-born adult children of foreign-born parents.This growth in supply put pressure on the wages ofunskilled workers.

Nevertheless, it is inaccurate to conclude that themiddle and working classes have not benefited from

innovation. The U.S. economy has grown about75 percent since 1991; U.S. employment grew 50percent since 1980. Over these same periods, theFrench and German economies grew by less thanhalf that amount, Japan by less than a third. TheU.S., moreover, achieved this growth with medianincomes that were already 25 to 30 percent higher.In addition, had the United States not contributed adisproportionate share of global innovationwhichhelped Europe and Japan grow faster than they oth-erwise would havethe latters growth relative to

the U.S. would have been even slower.

Indeed, with a more restricted supply of labor, wagesin the U.S. likely would have grown more. Yet evenso, median incomes have grown nearly 40 percentsince 1979 when healthcare and government ben-efits, like social security, are properly counted, andproper adjustments are made for family size.


32/35

e21

Issue

BriefNo.

1

May 2014

28

Given this pronounced growth differential, it isfar-fetched to claim rising crony capitalism in theUnited States accounts for growing income inequal-ity. Had a misallocation of resources been the driv-er of greater U.S. income inequality, U.S. growthshould have slowed relative to Europe and Japan,with more equally distributed incomes. Relativegrowth, however, accelerated.

Nor is there any credible evidence that U.S. incomemobility has declined (another possible symptom ofcrony capitalism). In fact, U.S. mobility is identical tocountries such as Denmark (with more equally distrib-uted incomes)except for the countrys bottom 20percent, where we seelower test scores, high-

er dropout rates, andmuch greater decima-tion of the family (fac-tors which probablytranscend economics).

More likely, incomeshave risen because wehave seen a rise in theopportunity cost ofdeploying a scarcity

of properly trained talent. CEO pay has risen, forexample, but not relative to the pay of the 0.1 per-cent, or relative to the pay of CEOs of privatelyowned companies.

In the United States, we also find that growth in theshare of GDP earned by the one percent has largelycome at the expense of the share of GDP earnedby capital and not the share earned by the 99 per-cent. In both the United States and Germany, forexample, the 99 percent earns about half of GDP,despite the fact that the one percent in the UnitedStates earns a larger share of GDP than does the onepercent in Germany. In the United States, however,capital earns a smaller share of GDP than it earns inGermany. This may indicate that properly trainedtalent in the United States, which is more produc-tive than its counterparts in Germany and elsewhere,may rightly have more negotiating leverage over in-

vestors. Regardless, the relative success of the onepercent does not seem to have significantly affectedthe share of GDP earned by the 99 percent.

If circumstances have benefited the one percent dis-proportionately, why not tax and redistribute theirgood fortune? The slow growth of Europe and Ja-pan should give one pause for concern. Neverthe-less, there are two opposing theories. One side ar-gues that Americans are inherently entrepreneurialand will continue to innovate, no matter the payoffs.The other side claims payoffs drive risk-taking likeany game of chance; as such, culture is largely a by-product of incentives.

Proponents of tax in-

creases often point tothe 1990s as evidencethat taxes dont affectentrepreneurialism.They forget, however,that the invention ofthe Internet drove theNASDAQ from 800to 4500. The resultingincrease in the payofffor risk-takingand

we saw a large increase in entrepreneurialismthustrumped any increase in the tax rate.

Proponents similarly point to the growth of SiliconValley in California, where higher tax rates reign.Again, however, the payoffs for working in such acommunity of experts likely trumps any differencein state tax rates. (Where the payoffs are higher, weconsistently see more entrepreneurialism.)

Consider, further, the compounding effect on statelotteries. When the size of the pool rises, ticket salesincrease exponentially. The U.S. has undoubtedlybenefited from the compounding effect on the pay-offs for risk-taking.

Success is relative: one persons success raises the barfor others. Our most talented workers are workinglonger hours than their counterparts in Europe, and

The U.S. economy has grownabout 75 percent since 1991; U.S.

employment grew 50 percent since

1980. Over these same periods,

the French and German economies

grew by less than half that amount,

Japan by less than a third.


33/35


with higher productivity than their counterparts inJapan. Our best students no longer want to be doc-tors and lawyers. They are going to business school.

Their success creates companies such as Googleand Microsoft, as well as communities of experts,like Silicon Valley, which give our workers far morevaluable on-the-job training. That training increasestheir chances of success and, in turn, the payoff forprudent risk-taking.

Meanwhile the success of these companies puts eq-uity into the hands of successful entrepreneurs, whoare more willing to underwrite the risks that produceinnovation, and who are more skillful at choosingwhich risks to underwrite. It is no coincidence that

the United States has more equity per employee andper dollar of GDP, while also growing faster thanboth Europe and Japan.

But this process is gradual and compounds slowlyover time. The resulting increase in payoffs is af-fected by far more than just the tax rate. Without

these compounding benefits, it is not as though Eu-ropeabsent similar infrastructurecan slash taxrates and immediately start growing as fast as theUnited States.

The financial crisis increased the importance of eq-uity and risk-taking. With a finite capacity for bear-ing risk, when the economy eventually reawakenedto the fact that banks were more unstable than ex-pected, the economy, accordingly, dialed back risk-taking elsewhere to compensate. As a result, growthslowed, unemployment rose, and incomes declined.

Risk-averse savings sat unused for want of more eq-uity to underwrite the risks of putting such savingsback to work. Unconventional monetary policy, a

growing level of public debt relative to GDP, and anincreased regulatory burden, also added to the risksfaced by our economy. Going forward, unless wemitigate some of these risksor accumulate moreequity per dollar of GDP and per employeeeco-nomic activity will, in the future, likely grow moreslowly from a permanently lower base of demand.


34/35

e21

Issue

BriefNo.

1

May 2014

30

ABOUT THE AUTHORS

Philip Armour Cornell University

Philip Armour is currently a PhD candidate at Cornell University, jointly located in the Economics Department andthe Department of Policy Analysis and Management. Prior to Cornell, he worked as a researcher at the Federal Re-

serve Bank of San Francisco, earned a Masters in Econometrics from the London School of Economics, and was a

summer associate at the Congressional Budget Office. With Richard Burkhauser and Jeff Larrimore, Armour stud-

ies how different definitions of income can result in differences in trends in income growth and inequality in the

US during the 20th and 21st centuries. Additionally, he is an expert in Social Security policy, in particular disability

programs in the US, on which he wrote his dissertation. Armour will be receiving his PhD from Cornell this spring

and will start as an Associate Economist at the RAND Corporation this coming fall.

Gerald Auten U.S. Department of Treasury

Gerald Auten is a Senior Research Economist in the Office of Tax Analysis, U.S. Treasury Department. Previous posi-

tions include Senior Economist at the Council of Economic Advisors (2004-2005), Professor of Economics at Bowl-

ing Green State University, Assistant Professor at the University of Missouri-Columbia, and Brookings EconomicPolicy Fellow (1978-79). He has a Ph.D. in Economics from the University of Michigan (1972). Autens research

interests and publications include behavioral effects of income taxes, charitable contributions, capital gains, and

income mobility.

Edward Conard American Enterprise Institute

Ed Conard is a visiting scholar at the American Enterprise Institute. Conard is the author of the top-10 New York

Times best seller, Uninte

income inequality in america: fact and fiction

Documents