There is a vigorous ongoing debate about inequality in the United States. Battle cries range from “It’s my money” to “We are the 99 percent.” There are reasons why inequality matters, and why it should be the subject of careful analysis more than sloganeering. Let’s look at some of the many sides of this issue.
First, to motivate the discussion: Why should greater inequality bother us? There are several legitimate reasons. First, to put it in the mildest terms (using the word of two theoreticians in taxation going back more than half a century), extreme inequality is “unlovely.” Few – even those who take great satisfaction in their own relative achievement – like to be confronted with others living painful lives. And there is even an element of self-interest in shared success. If you believe that our society will respond to people living below some level of income – even if you may not agree with that response – your own tax burden will be less if everyone succeeds in some measure.
Greater concentration of income means a narrower consumer base; and anything that stands on a narrower base becomes less stable. In this connection, fewer decisions by fewer people – or fewer unexpected declines in people’s prospects – could have a bigger impact on the prospects for the whole economy. There can even be an economic case that the decline of a mass market reduces the need for investment, which over time snowballs negatively into continuing reductions of income and output (relative to what otherwise could be achieved).
And over the very long haul, increasing concentration of income and wealth can reduce the initiative and motivation of generations of the most successful – the very people who might be expected to lead the way to continuing growth and prosperity for society as a whole. The extreme vision would be an uninvolved economic aristocracy, living luxuriously just by clipping coupons. But we could imagine a far less outrageous but nearly as damaging pattern in political “rent seeking,” in which those who have made fortunes from the innovations of yesterday use their wealth to seek legislative and regulatory protection against competition from the innovations of today and tomorrow. There are many examples of such behavior in the past and around the world. We 21st century Americans tend to think of ourselves as exceptional, and immune to such temptations that would lead to hardening of the societal arteries. But the history books would tell us that we would need to be truly exceptional to avoid that fate.
And at the very least, perceptions of growing inequality have had a corrosive effect in the public square. The shouting that we hear can drown out rational discussion and debate over very real national problems in the economy and other spheres. It can make adversaries of people who should be allies and partners, before they even meet.
So there is valid reason to be concerned about excessive inequality – understanding that “excessive” is an arbitrary standard. And for that matter, there is ambiguity in any measure of inequality – which ambiguity is worth some exposition.
In recent years, the most commonly referenced measure of the distribution of income has come from the Congressional Budget Office (CBO). In years gone by, scholars used figures from a Census Bureau annual survey. With all respect to the good people at Census, their numbers suffered from both the voluntary nature of their survey, and its uniform sample of the population. As a result, few people with very high incomes would be included in the survey in the first place, and at least some would refuse to participate. The Census Bureau would make every feasible statistical adjustment and still be far from a detailed picture because of pure lack of information.
The Congressional Budget Office instead marries to the Census data an unidentified sample of federal income tax returns. Filing an income tax return is not voluntary (as you probably know). And the sample of tax returns is stratified by income, which is to say that it includes more observations with very high incomes (with reduced statistical weights, to keep the result representative), and fewer observations (with larger statistical weights) of the about-average-income, two-spouse, two-kids, one-dog, almost-all-income-from-wages-and-salaries households. The average-income families tend to be quite similar one to the other, and so fewer statistical observations provide all of the information needed. In contrast, the few higher-income households have unique individual characteristics that cannot be as well represented by averages. Still further down the income scale, many households with modest incomes are not required to file income tax returns, but the CBO method uses the information from the Census sample to represent them. Thus, the CBO data have distinct advantages in investigating inequality, even with the qualification that they are formed by a statistical (shotgun) marriage between two independent data sources.
And what CBO has found, over the lifetime of their data file (which extends back to 1979), is a distinct increase in income inequality (as shown in the following chart). Between 1979 and 2007 (hold the thought of 2009 for a moment), the share of total income of the top 1 percent of households more than doubled – from 8.9 percent to 18.7 percent. And as you might imagine from that enormous jump, it really was “the 1 percent” of popular rhetoric where all of the action occurred. The share of those from the 96th through the 99th percentiles grew, but only from 11.3 percent to 12.7 percent. All of the other shares, for every group from the 95th percentile on down, declined (by the arithmetic mirror image of the increases in these two higher income groups).
Looking at these data in another way, the average income (in inflation-adjusted 2009 dollars) within the 96th to 99th percentile group grew from $166,800 in 1979 to $315,800 in 2007 – a healthy near-doubling. But the growth among the highest 1 percent was from $523,300 to $1,917,200 – a near-quadrupling. (The average growth for the population as a whole was from $61,900 to $101,000, or 63 percent – obviously driven by these upper-income groups, with much less growth lower on the income scale.) So again, the top of the tree grew very rapidly over those years – strikingly so, to just about everyone; offensively so, to some.
But should we take those figures at face value? Are there reasons to question their significance? There are, and we should discuss just a few.
For perhaps the most fundamental consideration, each year of data is a snapshot. The people in the top 1 percent – and the bottom 20 percent – in one year are not necessarily the same people in the following year, or five or ten years hence. Some would go so far as to say that the income ranking in any one year is irrelevant, because people may be on randomly different rungs of the ladder in the next year.
If a household’s income in each year were a random draw, that argument would carry enormous weight. However, incomes are quite far from random from one year to the next. For example: A disproportionate share of the lowest income quintile is elderly households. People who are too old to chart upwardly mobile careers and have little income other than Social Security are unlikely to bounce up the income rankings in a later year. Similarly, a disproportionate share of the income of those at the top of the ladder is income from capital (as opposed to income from labor). Although a household that has accumulated considerable wealth could conceivably lose it, the probability of that unhappy occurrence is relatively low. Some mobility would come from, for example, young families with one or two adults living on limited incomes while attending graduate or professional school. (I have memories.) But the subsequent advancement of such families would be quite predictable – many would have parental roots at higher levels of income – and as they progressed in future years they would be replaced at the bottom of the ladder by others following the same path.
The U.S. economy certainly does have meaningful mobility, beyond the predictable ascensions and re-ascensions just described. There are plenty of bootstrap anecdotes to discuss, and there are subtleties in these data as well. But recent evidence suggests that mobility in the United States actually is less than in most other developed countries (see here and here for examples). Looking between generations, parental success (or the lack thereof) is a strong predictor of the next generation’s educational and economic success (or the lack thereof). So the snapshot income-distribution data should be understood to have conceptual limits, but should not be disregarded on that ground.
There are at least two further limitations of these income distribution data. One is the handling of employer- and government-provided in-kind benefits, most notably health insurance. CBO attributes to households the value of such health benefits. This is eminently reasonable, especially in assessing the relative well-being of two households at the same time; if one household receives health coverage from a breadwinner’s employer and another does not, a simple comparison of the cash income of the two would clearly understate the standing of the former. But including the employer contributions may not comport with people’s conceptions of their own well-being over time. An employee who was told that the cost of his health insurance was to increase substantially in the next year, but that his employer would pick up the tab, surely would be grateful. But if the same employee then were told that as a result his cash wage would be unchanged – or even decrease – in the next year, he or she quite likely would not feel better off. People tend to incorporate their health insurance (and their good health) in their personal “baselines,” and to require an increase in spendable cash income on top of that to conclude that their standards of living have improved.
The significance of this factor is that this simple parable of rising employer-paid health-insurance premiums coupled with near zero – or even negative – increases in cash wages is not far from the truth. By CBO’s measure, middle-range incomes have not increased very much in recent years – but even this limited increase includes employers’ increased contributions toward employee health insurance. Thus, during the hours when the typical U.S. private-sector employee is not studying the footnotes and methodological appendices to CBO documents, he or she probably feels much less optimistic about his or her economic lot than CBO’s income distribution statistics would suggest. For those at the top of the income tree, the cost of an annual health-insurance policy is far less material to perceived well-being. In this respect, therefore, middle-class demoralization about slow income growth, and even potential resentment about widening income inequality, may be worse than the numbers would suggest.
There is at least one more knotty methodological conundrum behind these income distribution data. Much of the cash flow of very-high-income households comes from sales of assets, including both capital gains and capital losses. CBO includes these net realized capital gains in a household’s income for purposes of the income distribution statistics. This is a highly imperfect measure. A household can choose either to realize an accrued capital gain or not. Arguably, the household is equally well off in either case. However, for purposes of the CBO statistics, a household that chooses to realize an accrued capital gain is shown to have a higher income than another that might not realize the same accrued gain.
The Census Bureau, in its long-standing income-distribution data series, does not even ask for the amount of realized capital gains. That is not the correct methodological answer either. A household with more wealth than God that supported itself solely by trading growth stocks would honestly answer the Census Bureau’s survey questions by saying that it had an income (as defined by Census) of zero – obviously not informative.
The right answer would seem to be to include in measured income the increase (or decrease) – realized or not – in the value of the household’s total portfolio over the year. But such information will never be available; it is not required on the income tax return, and it would be incredibly burdensome (and likely never reported accurately) in a household survey. (The Federal Reserve does undertake a detailed survey that includes asset values, but they do so only every three years, and the sample size is smaller than the Census Bureau’s.)
So with that limitation in mind, note that the CBO data showed a big drop in the income of the top 1 percent between 2007 and 2008-2009 – and that much of that drop came from a sharp reduction of realized capital gains. That surely came in part because of the drop in financial market values; there certainly was less accrued gain to realize. But at least in theory, some of that drop could have come from wealthy households simply holding on to well-selected assets, rather than selling them. It also could come from such households realizing capital losses; even though the net losses (over $3,000) would not be deductible in the initial year, those losses could be carried forward to offset gains in later years. And market values are what they are; a household that traded one asset that had declined in value for a similar but not identical asset that also had declined in value would not be losing any market value, in the greater scheme of things, while it would acquire the tax loss to carry forward.
The fluctuation of realizations of capital gains is a reason why these income distribution figures should be followed over time, without excessive weight given to the numbers for any one particular year. The sharp decline of incomes in the upper tail of the distribution during the financial crisis likely will have been reversed in the market recovery of the last few years once the numbers are available. Cycles should not be assumed to be eternal trends.
Any discussion of the ins and outs of income distribution statistics could go on far longer, and touch many additional conceptual and methodological issues. But considering just the points above, if anyone were truly to obsess over the inequality statistics, he or she might conclude from those last two years of the CBO data that a crash is a good thing. After all, it did reduce the relative share of total income of the “1 percent.” Of course, in so doing, it reduced the absolute levels of income of the other 99 percent as well. Historically, the period of the strongest sustained income growth for all groups – the 1990s – still saw growth in the relative share of the highest 1 percent. So what do we want: superficially more-equitable income distribution statistics, or a higher standard of living for all Americans?
At bottom, we all are in this together. Economic growth is not a zero-sum game. As rank-and-file workers benefit from the creation of new jobs, those who create those jobs will benefit as well. That is not a bad thing.
Even many economists who feel genuine concern about the trend toward greater inequality think long and hard before recommending remedies. Interventions in markets to reduce competitive high returns can entail serious unintended consequences – potentially far beyond merely dulling incentives to work and invest. Entire industries and channels of innovation can be slowed, to the ultimate detriment of the entire economy.
We certainly need a fair income tax, so that people on different rungs of the income ladder are treated equitably relative to one another. That entails judgment; there are no simple mathematical formulas that state the one right degree of progressivity in our tax system. We need to have that conversation; perhaps now, given the broad consensus that the current income tax is both unfair and detrimental to economic growth, we can begin to tackle this tough issue.
There is no objective measure, but to some pairs of ears, the dissension over the very nature of our economic system now is more discordant than at any time in memory. Our proximity to economic and financial crisis leaves open the risk that the nation could act in desperation and haste, and do serious and perhaps irreparable damage to our prospects and our prosperity. CED’s Trustees have undertaken a project on “sustainable capitalism” that we hope will help to head off this and other risks. Thanks to all of our Trustees for their efforts on behalf of our mission of equal economic opportunity and long-term economic growth.