Inequality Testimony before the Joint Economic Committee

 

Chairman Brady, Vice Chair Klobuchar, and Members of the Committee, thank you for inviting me to appear today to discuss the topic of income inequality in America. With long-term unemployment historically high and still-pervasive economic insecurity in the wake of the Great Recession, it is understandable that many Americans have grown more concerned about the nation’s levels of inequality. Too many families struggle in poverty, too many workers have given up on finding fulltime work, and too many young adults have graduated into a weak economy that will lower their lifetime earnings.

At the same time, it is important to note that it is the fragility of the economy that lies behind concerns over inequality. Inequality was high and rising during the late 1990s, but because the growing economy was largely benefitting everyone, few people were worried about income concentration at the top. As the economy continues to recover and unemployment continues to fall, concern about inequality will recede.

As I will show, in long-run perspective, living standards have improved for the poor and middle class even as income inequality has grown. In part, that is because inequality has increased less than most analysts suggest. Furthermore, there is little compelling evidence that the gains at the top have reduced income growth lower down. And contrary to claims that rising income inequality has hurt inequality of opportunity, the evidence of a link between the two is weak. In part, that is because intergenerational mobility has not declined much—if at all—as income inequality has grown.

However, if intergenerational mobility is no worse today than it was decades ago, nor is it any better. We should not be satisfied as a nation with the limited upward mobility facing poor children today, and fifty years after Lyndon Johnson’s declaration of war on poverty, we should establish a second front against immobility. Attacking income inequality, however, is unlikely to reduce poverty or to promote equal opportunity; emphasizing it is, in fact, a distraction from the task at hand.

 

The increase in income inequality has been overstated, and the rise in incomes among the poor and middle class have been understated.

 

During the 1950s and 1960s, incomes in America rose dramatically, and more among the poor and middle class than among the rich. Beginning in the 1970s, income growth slowed, and it has remained sluggish for poor and middle class Americans relative to this earlier “Golden Age.” However, as shown in Figure 1, inequality within the bottom 80 percent has grown only modestly, primarily during the 1980s, and hardly at all since the 1980s.

[2]

Indeed, a wealth of research indicates that earnings inequality and household income inequality between the middle class and poor has not risen since the mid-to-late 1980s.

[3]

This literature includes a paper on earnings trends coauthored by University of California-Berkeley economist Emmanuel Saez, whose “top one percent” income concentration estimates have been cited ubiquitously. It also includes a book by economist and University of Wisconsin chancellor Rebecca Blank, who oversaw the Census Bureau at the start of the Obama presidency, as well as a report by the Congressional Budget Office.

Nor can the modest rise in middle-poor inequality be attributed to income stagnation: average income in the middle fifth rose 66 percent between 1969 and 2007 and 55 percent in the bottom fifth. It is true that rising work among married women has contributed to this income growth. However, median male earnings did not fall between 1969 and 2007, so it is not the case that married women in the middle class increasingly took jobs to prevent family income from falling.

[4]

Indeed, in industrialized nations around the world, women have become more interested in work for the fulfillment it provides, increasingly earning postsecondary degrees, delaying marriage, and delaying and reducing childbearing. Work has risen most among women with the best-educated husbands, not among wives of lower-skilled men, who are likely to have more employment problems and lower wages.

[5]

The decisions of married women to work have become less sensitive to husband’s earnings over time, not more.

[6]

While the hours of married men have declined over the long run, those of single men have not, suggesting that husbands have become more economically comfortable, not less.

[7]

 

Figure 1.

 

Note: For methodological details, see the appendix.

 

 

Even when it comes to income concentration at the top, there are good reasons to believe that the increase has been overstated, particularly since the 1980s. The oft-cited estimates of Saez and French economist Thomas Piketty indicate that the share of income received by the top one percent rose from 10 percent to 24 percent between 1979 and 2007, reaching an all-time high (see Figure 2).

[8]

But there are several problems with those figures:

[9]

·        

They look at incomes before taxes and do not include government transfers or employer-provided fringe benefits. CBO figures that remedy these issues indicate the top one percent’s share grew from 9 percent in 1979 to 19 percent in 2007 before taking taxes into account and from 7 percent to 17 percent after taxes. From 2000 to 2010, it fell from 15 percent to 13 percent.

[10]

·        

They exclude non-taxable capital gains (the vast majority of housing gains), exclude all capital gains until an asset is sold, and then include the lifetime gains from an asset in the year it is sold. The CBO figures share this problem. In 2007, 20 percent of realized capital gains were on assets that had been held for 20 years or more, 41 percent were on assets held for ten years or more, and 60 percent were on assets held for five years or more.

[11]

That means that often income from capital gains counted as being received in the previous year actually accrued year-to-year for decades. The Piketty/Saez estimates that do not include capital gains show an increase in the top share from 8 percent to 18 percent between 1979 and 2007.

[12]

CBO pretax estimates after excluding capital gains suggest an increase from 7 percent to 14 percent.

[13]

·        

The estimated trends are affected by changes in the taxation of stock options and businesses incorporated under different parts of the code. These and other tax-code changes greatly complicate the measurement of top incomes. Earnings concentration figures from another paper coauthored by Saez avoid business incorporation issues but switch from households to individuals and are available only through 2004.

[14]

They show the top one percent’s share rising from 7 percent to 13 percent from 1979 to 2004.

[15]

About one-third of this increase came between 1986 and 1988 in the wake of major tax reform legislation. That legislation resulted in more income from the exercise of stock options showing up on W-2 forms (and thus appearing in the Saez data).

·        

A recent paper by Cornell University economist Richard Burkhauser and two of his students, Phillip Armour and Jeff Larrimore attempted to address most of these problems—in particular the treatment of capital gains.

[16]

It found that the share of income received by the top five percent fell between 1989 and 2007. The CBO pretax estimates after excluding capital gains suggest an increase from just 10 percent to 14 percent over this period. Saez’s estimates for earnings show an increase from 11 percent to 13 percent from 1989 to 2004.

Figure 2.

 

 

The case that any increase in the share of income received by the top has come at the expense of the middle class or the poor is weak.

 

The increase in income concentration at the top is regularly blamed for the slowdown in the incomes of the poor and middle class, but the case rests on fragile grounds. As Figure 1 shows, income growth began to slow in the 1970s, before the run-up in income concentration gathered steam. It did so around the industrialized world—the primary culprit was slower productivity growth. In the U.S., inequality between the poor, middle class, and rich fell during the 1970s, and the decade was not especially generous to the top. Since the 1970s, the incomes of Americans at the top and bottom of the ladder have risen and fallen together over time, and if the large losses experienced by the top during the Great Recession did not help the poor and middle class, it stands to reason that the large gains at the top during expansionary periods may not have hurt them.

A number of researchers have conducted naïve analyses purporting to show that absent rising inequality, the incomes of the poor and middle class would be thousands of dollars higher today. Even if we ignore the data problems noted above that likely overstate growth in income concentration, these analyses still must assume that the size of the economic pie would have grown the same amount (or more) in the absence of rising inequality. However, if, for instance, we had kept top marginal tax rates at their high 1970s levels, entrepreneurs might not have worked as hard, investors may not have taken as much risk, and the economy might have grown less than it did. The poor and middle class could have received bigger slices of a smaller pie.

For instance, CBO data indicate that the middle fifth’s share of after-tax income fell from 16.5 percent in 1979 to 14.3 percent in 2007. The estimates suggest that if its share of after-tax income had stayed at the 1979 level and overall income growth had stayed the same, then middle-class households would have had an average income $9,000 higher in 2007 than what they actually received.

However, imagine that the rising inequality we experienced actually raised overall income growth. In that case, preventing inequality from rising would have lowered income growth correspondingly. If the increase in inequality between 1979 and 2007 expanded the size of the economic pie by 15 percent (0.5 percent per year), then holding the middle class at its 1979 share of income would have left it no better off than it actually was in 2007. The middle class’s share of the economic pie would have been maintained, but the pie would have been smaller.

Is it credible to think that rising income concentration between 1979 and 2007 might have raised income growth by 15 percent? As a crude check, consider that the top one percent’s after-tax income share in the CBO data rose from 7 percent to 11 percent from 1979 to 1989. Rough estimates from one study suggest that that four-point increase in the top’s share, sustained for 18 years, raised gross domestic product by 9 percent in 2007.

[17]

The further increase in the top one percent’s share from 11 percent to 15 percent between 1989 and 2000 would have raised GDP by another 3 percent in 2007, bringing the total to 12 percent.

If the increase in the top one percent’s share raised aggregate household income by 12 percent, then preventing that share from rising by leaving income shares at their 1979 levels would have left the middle fifth not $9,000 richer but $1,500 richer. However, preventing the top one percent’s share from rising might not have left the middle fifth’s share at its 1979 level; perhaps the middle fifth’s share would still have fallen while households between them and the top one percent would have increased their share. If the middle fifth’s share would have fallen from 16.5 percent in 1979 to just 16.1 percent in 2007, then combined with the smaller size of the economic pie, the middle fifth would have been no better off for having restrained the top one percent’s share.

Perhaps rising inequality increased aggregate incomes by less than 12 percent between 1979 and 2007. If rising inequality raised aggregate household income by 8 percent instead, then the middle class would have been no better off restraining the top one percent’s share if their own share had fallen from 16.5 percent to just 15.5 percent.

These back-of-the-envelope calculations demonstrate how rising income concentration might not have hurt income growth at the bottom or in the middle if it increased the size of the economic pie enough. It is true that not all research finds that inequality increases economic growth. But contrary to many assertions, the academic literature offers little support for the idea that inequality harms growth. The most-cited study, by International Monetary Fund economists, finds slower recovery from economic downturns in countries with high inequality levels, but its boosters generally fail to note that only developing nations are examined.

[18]

In recent years, no fewer than three reports from the liberal Center for American Progress concede that the evidence fails to show an inequality-growth connection.

[19]

Of course, if income concentration rises and growth does not, that means that the incomes of the poor and middle class will be lower than they would have been in the absence of rising inequality. I am aware of only one paper that directly examines whether inequality affects income growth lower down. Cross-national comparisons by sociologist Lane Kenworthy indicate that countries with a larger increase in the top one percent’s share of income have median income gains no lower than countries where the top’s gains are smaller.

[20]

This finding is consistent with the premise that economic growth has benefitted from rising inequality in the way I have outlined above, leaving poor and middle class households with a smaller share of a bigger economic pie and no worse off for it.

 

The case that any increase in income inequality has reduced equality of opportunity is weak.

 

A running theme of President Obama’s administration over the past two years has been the deep connection between income inequality and inequality of opportunity. This theme taps into an intuition captured by the evocative but ultimately misleading metaphor suggesting that if the rungs of the economic ladder grow further apart it becomes more difficult to climb. In truth, how unequal the rewards are to different economic destinations and how unequal are the chances to reach those destinations comprise two distinct features of a nation’s political economy. High income inequality is compatible with both high and low inequality of opportunity. Whether rising income inequality affects inequality of opportunity is an empirical question that hinges on the relationship of parental income to child outcomes and on how the incomes of rich, middle class, and poor parents change over time.

As I have already summarized, one reason why rising income inequality might not have diminished equality of opportunity is that the former has not grown much across the broad swath of poor and middle class families. And even if income concentration at the top has increased, there is another empirical result that casts severe doubt on claims that this type of inequality has reduced opportunity: intergenerational mobility has probably not declined much—if at all—in the past three decades.

The academic literature on intergenerational income mobility fails to reach consensus regarding trends. No research shows a sizable increase in mobility since the mid-twentieth century, but the most common finding is a change so modest as to be statistically indistinguishable from no change at all.

[21]

This tendency is all the more striking because the vast majority of these studies use a mobility measure that mechanically declines if income inequality accelerates, even if the relationship between child and parent income rankings stays constant. In my own forthcoming research, I find that today’s thirty year olds—who experienced rising income inequality between the middle class and poor during early childhood and have witnessed rising income concentration at the top through their entire lives—have experienced no less mobility than did thirty year olds in the mid-1970s.

Faced with such an unsupportive research base, some proponents of the view that inequality has hurt mobility have turned to cross-national evidence, a line of argument that culminated in the Obama Administration’s popularization of the “Great Gatsby Curve”.

[22]

This chart, showing a strong statistical relationship between countries’ inequality levels and the extent of mobility their citizens enjoy, has been used to project that today’s children will experience less mobility than earlier generations.

However, the nations in the Gatsby chart differ in myriad ways that could be driving the inequality-mobility link. Kenworthy noted that Scandinavian countries have low inequality and high mobility but also distinctive education policies; it could be that these policies produce high mobility and that the low inequality levels are incidental.

[23]

Perhaps nations’ cultures determine both how strongly they combat unequal opportunity and how unequal they are willing to let adult incomes become. In this scenario, the Great Gatsby Curve primarily might reflect deep-seated preferences for more or less state intervention, with inequality per se having no causal impact on mobility. My colleague Jim Manzi even found that the relationship between a country’s population size and its mobility was as strong as the inequality-mobility correlation.

[24]

The relationship between inequality and mobility across countries also fails to recur looking across labor markets in the United States. Income concentration at the top does not predict lower mobility across over 700 such local labor markets, and inequality across the middle of a labor market's population has no relationship to mobility across the 100 largest metropolitan areas.

[25]

Finally, it is worth emphasizing that the Great Gatsby Curve uses the conventional measure of mobility that mechanically indicates lower mobility when the growth in income inequality is greater. If nations with high inequality have experienced high growth in inequality, they will mathematically have lower mobility than other nations even if the relationship between parental income ranking and child income ranking is the same across countries. In other words, a relationship between inequality and mobility is to some extent baked into the Great Gatsby Curve. Recent research indicates that when mobility is measured using an indicator that focuses solely on parent and child income rankings, Sweden and the United States have the same mobility levels.

[26]

These two countries had the lowest and highest levels of inequality, respectively, in the Great Gatsby Curve. The implication is that there may be no cross-national relationship between inequality and mobility when the relationship is not baked into the mobility measure itself.

 

While rising income inequality has not hurt opportunity, and opportunity has not grown more unequal, upward mobility remains far too limited.

 

Income inequality has become a convenient bogeyman for a host of economic problems that are real and that demand our attention. Promoting economic growth and reducing joblessness should be at the top of our list of immediate priorities, goals that are not effectively advanced in Washington (or New York) by decrying inequality levels.

Equally important over the long run is ensuring the vitality of the American Dream. A divisive populism that casts income inequality as the source of all our problems may make for effective politics, but it will not show the way toward expanded opportunity for children facing long odds of economic success. Upward mobility has not diminished over time, but for decades it has been stuck at unacceptably low levels. If past patterns hold, 70 percent of poor children today will fail to make it to the middle class as adults. Four in ten will be mired in poverty themselves in midlife.

[27]

The fact that mobility is no worse than in the past—perhaps even no worse than in Sweden—is hardly reason to shrug it off. These are not the kind of odds those of us solidly in the middle class would accept for our children. Kids do not pick their parents, childhood communities, genes, or schools, and they do not choose the time and place into which they are born. The American Dream is in poor health if children who grow up in the bottom can aspire only to fill the same sorts of jobs as their parents hold, even if they will be better paid in those jobs.

The challenge is to identify real solutions to the problem of limited upward mobility. Democrats look skeptically when Republicans express interest in mobility. Ultimately, Republicans will and should be judged on deeds and not words. Expanded opportunity cannot be done on the cheap. However, by framing our mobility challenge as primarily a problem of income inequality, Democrats evade hard policy realities. Immobility is not primarily a problem of some parents having more money than others. For that matter, policies to invest in the human capital of children and to promote opportunity-enhancing behaviors among disadvantaged parents and children have a lousy track record. It will be hard work discovering how to expand upward mobility. The task is not well served by pretending that income redistribution would solve our problems.

 

Appendix: Methodological Details for Figure 1

 

For income growth between 1948 and 1969 for the bottom 80 percent, estimates come from Economic Policy Institute tabulations of data from the Census Bureau’s Current Population Survey. The figures indicate annual change in the family income of families (rather than the family income of individuals). Household income data only go back to 1967, and while family income growth was probably stronger than household income growth from 1948 to 1969, using the former as a proxy for the latter is unlikely to distort the basic trend. “Income” includes cash transfer income from federal programs (such as Social Security, Unemployment Insurance, Supplemental Security Income, and Temporary Assistance for Needy Families) but does not include non-cash benefits from employers or government (such as health insurance, food stamps, or housing subsidies). These sources of income were relatively unimportant prior to the 1970s; Medicare and Medicaid were enacted in 1965 and the food stamp program expanded nationally only in the early 1970s. Incomes are measured before taxes. See Economic Policy Institute (2012). “Mean family income, by income group” [data table]. The State of Working America. Washington, D.C.: Economic Policy Institute. https://www.epi.org/files/2012/data-swa/income-data/Mean%20family%20income,%20by%20income%20group.xlsx.

For income growth within the bottom 80 percent between 1969 and 2007, the estimates are my own, using the Current Population Survey microdata. I show changes in the household income of households (rather than the family income of families or the household income of individuals). Since households became smaller over time as marriage and fertility declined and more people chose to live independently, I adjust income for household size. From 1979 forward it is possible to make other improvements. I add to income noncash federal benefits such as food stamps, Medicaid, and Medicare, and employer-sponsored health benefits. Because there are two different ways of valuing Medicaid and Medicare, which produce very different growth estimates for the bottom fifth, I average across the two approaches for each year before computing income growth. I also use estimates in the CPS data to account for federal and state income taxes (before credits, except that the Earned Income Tax Credit is included in 1989 and 2007), payroll taxes, and property taxes. As with the EPI estimates, capital gains are excluded from these figures. Following the Congressional Budget Office, I also drop households with negative incomes, because such households experienced business or investment losses and are likely to have considerable wealth from which to draw down. Some households with business losses that left them with low but non-negative incomes remain in the data.

The Current Population Survey cannot be used to validly produce estimates of income growth for the richest Americans, so for estimates of top-five-percent income growth across the entire 1948-to-2007 period, the figures of Thomas Piketty and Emmanuel Saez are used. These estimates indicate the change in the mean income of “tax units” (essentially, tax returns, after accounting for a small number of non-filers) rather than of families, households, or individuals. Income is measured before taxes are deducted, and no government or employer benefits are included. See Thomas Piketty and Emmanuel Saez (2007). “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 and English-Speaking Countries (Oxford: Oxford University Press). Updated figures at https://elsa.berkeley.edu/~saez/TabFig2012prel.xls. The figures cited here are from their series that excludes capital gains, for consistency with the other data sources in the chart.

All income figures were adjusted back to nominal dollars and then inflated using the Bureau of Economic Analysis Personal Consumption Expenditures deflator. Periods are chosen to begin and end with a business cycle peak to ensure comparability.