The Agenda

Guest Post by Arpit Gupta: Response to Saez-Piketty on Inequality

Editor’s note: Here is the second of two posts from Arpit Gupta.

Economists Emmanuel Saez and Thomas Piketty have been very successful in highlighting an agenda focused on inequality and progressive taxation. Though valuable in terms of detailing empirical facts, Saez and Piketty’s research isn’t necessarily subject to simple interpretation, and deserves critical consideration in light of the prominence policymakers have attached to their work. I look here their joint work on inequality; and in a separate post I consider recent work by Saez and Diamond and others on progressive taxation.

The Composition of the 1%

Annie Lowrey, in a recent writeup of Saez and Piketty’s research in the New York Times, writes that, “[r]esearch done by Emmanuel Saez, left, and Thomas Piketty has shown that inequality among the middle class and the rich is nearly as acute as it was before the Great Depression.” What Piketty and Saez actually show is the income distribution for the vast majority of Americans have been relatively smooth, with only a tiny fraction of top earners pulling ahead:

Even among that top 1%, it seems as if a tiny fraction of top-earners are pulling ahead of the rest. The entire rest of the income distribution appears relatively smooth over time. 

That is, the whole concern about rising income inequality that comes through Piketty and Saez’s work is really about the growth in incomes among a very small handful of individuals — perhaps only a few thousand — who have grown their incomes quite substantially in recent years. It’s not obvious why the income dynamics of such a small group of people ought to be of such outsized importance in determining public policy or taxation; but let’s focus on the composition of that group.

As a guide to what’s going on here, Piketty-Saez find that top income shares have been rising across English-speaking countries, but not elsewhere:


The English-non-English dynamic is telling and possibly points to the role of financial and globalizing trends more pronounced in the Anglophone world. For instance, top performers in English-speaking countries — say entertainers, sports stars, etc. — can appeal to much broader audiences and thereby draw larger incomes as a result of the superstar effect. The resulting large-scale tournaments also tend to draw top talent from around the world, relative to say the Dutch-speaking world.

Another dynamic concerns the role of corporations. CEO pay has gone up in lockstep with firm market capitalization, suggesting that the payoff for delivering even marginal improvements to corporate profits is higher in areas with outsize corporations. Globalization and more interconnected markets have boosted the returns to top CEO pay, more so in English-speaking countries than elsewhere.  Consider for instance a recent NYT article on the decline of Sony, which mentions that no major electronics manufacturer had broken into the top group of Japanese electronics manufacturers in 50 years, while American upstarts routinely do so. As a result, America is home to a relatively unique group of highly compensated innovators, such as Amazon’s Jeff Bezos or Facebook’s Mark Zuckerburg.

Another explanation for the divergence in inequality across countries comes from Fatih Guvenen, economist at the University of Minnesota, and colleagues. They suggest that higher levels of progressive taxes in some countries dampen investment in education and other human capital — their results suggest that 76% of upper-end inequality (the difference between incomes at the 90-50 percentiles) can be attributed to different progressive taxation policies. This is particularly important when thinking about skill-biased technological change — the idea that higher levels of education can push countries towards greater overall technology and output levels. As a result, the highly taxed non-Anglophone countries have certainly experienced greater wage compression — but at the cost of lower aggregate investment and productivity.  

The broader point is that rather then worrying about inequality as a general phenomenon, it’s important to understand the nature and sources of income inequality. The evidence seems to suggest that the recent rise in income inequality is very localized among a small group of individuals, as opposed to being a general trend affecting many Americans. Additionally, it seems plausible that the core drivers of inequality come from individuals gaining additional compensation for generating additional productivity.

Looking at the overall social impact, the role of these productive 1%-ers seems fairly positive. Arguably less beneficial is the other big group in the top income bracket—financial managers earning outsize compensation. While I’ve suggested that these managers ought to be subject to the same income taxes as other rich individuals, rather than paying the capital gains rate; it’s hard to get too upset about this phenomenon. Researchers tend to suggest that financial managers earn outsize fees by taking advantage of other extremely rich individuals. It’s not obvious why such inter-rich distributional issues are a source of concern for the rest of society.

Shares of Income

In order to justify a focus on top incomes, Piketty and Saez routinely resort to language such as the the “Fraction of growth captured by the top 1%.” This sort of rhetoric has the effect of suggesting a causal relationship — that the income gains of the rich came at the expense of everyone else — though Piketty and Saez offer no good arguments for suggesting this is in fact going on.

Consider for instance Piketty and Saez’s most recent figures, which update top income information through the present day. Piketty and Saez find that the top 1% lost 36% of their income from 2007-2009; which supposedly accounts for half of the total output losses in this period. Would anyone suggest that the large losses by the rich buffered losses felt among ordinary Americans, thereby lessening the impact of the recession? Did anyone feel tangibly better off as the rich did worse; and would anyone have felt better had the rich done even worse still? These ideas sound preposterous on face. Yet that argument is identical to arguing that the gains of the rich on the way up had anything to do with the woes of the median worker during the boom. Unless you’re willing to believe that income losses by the rich during the Great Recession benefitted ordinary Americans; there’s no grounds for believing that their income gains during the years prior came at the expense of anyone. 

The large drop in incomes among the top 1% during the recent recession is consistent with other work by Fatih Guvenen and colleagues. Working with a random 10% sample of all male tax returns, they find that the top 1% have lost about 30% of their income in recent recessions, a pattern not seen in the past. The authors tantalizingly suggest:

For the first two recessions in our sample period, very high-income individuals fared better than anybody else in the population, whereas for the last two recessions, there has been a remarkable reversal of these fortunes and the highest- income individuals suffered the most. We conjecture that this new patterns may have to do with the rise of bonuses in the compensation of highly paid workers, which are more likely to be pro-cyclical than base-salary

To the extent that high-end incomes are simply more countercylical than others (perhaps due to the unique compensation patterns America’s dynamic capitalism fosters), perhaps we shouldn’t be as worried about what high-end incomes looked in the boom times of 2004-2007, the figures most prominently highlighted by the anti-inequality camp. A bonus-type structure to wages would make it appear as if the rich were hoarding all of the income; when in fact their income might simply be more variable than others. This whole exercise perhaps just goes to show the limitations of basing policy on the idiosyncratic behavior of a small number of people.

Another estimate that’s doing a lot of work for Piketty-Saez is their claim that the median worker has done fairly badly for the past several decades. The inference is that since ordinary workers did not share in recent growth, the gains of the rich must have been to blame in accounting for the difference. Yet as Terry Fitzgerald and Richard Burkhauser have argued using other data sources that the real compensation of ordinary workers has been steadily rising alongside GDP.

Burkhauser for instance has worked with Census CPS data to find that though he can replicate Piketty and Saez’s figures, a radically different picture emerges when more broadly accounting for economy-wide changes in the past thirty years. He presents the following table:

Here, Burkhauser is looking at total growth rates for household income over the past thirty years calculated in different ways. The 3.2% figure in the top left is comparable to the estimates that Saez and Piketty find — that American incomes have barely budged over a long time period. However, it’s essential to understand the basis on which that figure was computed — using IRS data that include pre-tax, pre-transfer income, while looking at the tax unit (who is listed on your 1040) as the unit of measure. Simply looking all members of the household broadly, but computing income the same, results in a 12.5% rise in income over this time period. Burkhauser perform a number of additional calculations, accounting for changing household sizes over time, the role of taxes and transfers in affecting household income, and the role of health insurance. Their calculation including all such estimates (bottom right) suggests that the households have in fact experienced compensation growth of 36.7% over the period in which Saez and Piketty claim that incomes were stagnant.   

The point here is not that the Piketty-Saez figures are wrong — just that these sort of household income comparisons involve complicated judgements and statistical care. The results that Piketty and Saez provide answer an important question — how has the pre-tax, pre-transfer market income of a typical worker changed over the past thirty years or so? It’s important to understand that gains have been minimal on this front.

However, we are frequently interested in a different question — how has household compensation fared over this time period, taking into account that we have tackled household incomes with policy adjusting taxes and transfers, and that a rising share of labor market compensation is paid out in fringe benefits rather than market wages? The answer to this question is very different — households appear to have done fairly well over this time period.

To see the tension between these calculations, consider the role of health insurance, which has dramatically risen in cost over the time period. Saez and Piketty might argue that though the value of health insurance benefits has been rising over time, the take-home pay of workers has been more stagnant, and people pay for other products with this cash. However, understanding and valuing the benefits of health insurance is critical to understanding how the compensation of workers has fared over this time period, and how the median worker has fared relative to the top 1%. If workers are being paid increasingly in the form of benefits rather than wages, this is important for some policy questions — we may want to further reduce cost inflation in healthcare for instance. But it’s misleading to argue that households are losing out because of inequality when they’re simply receiving compensation compensation in one form (health benefits) rather than another (cash wages).  

Political Capture

When you dig into it, many of the tangible harms that higher inequality is supposed to cause come through the political system. More money in the hands of the rich, it’s argued, furthers their political influence, which they are bound to use in a detrimental fashion. However, one wonders about problems going in the other way — could an excess focus on boosting tax revenue from the rich be counterproductive to constructing a sustainably financed welfare state?

Consider, for instance, that Nordic countries finance their welfare states through broadly regressive taxes, such as VATs, while actually maintaining relatively low taxes on capital to motivate investment. A broad tax base with minimal distortions is arguably necessary to sustain a large-scale provision of public goods.

Compare that to a country like Saudi Arabia, which uses tax revenues from oil production to avoid both taxation and representative institutions. Does an excessive dependence of taxes on the wealthiest limit the possible size of the state, along with lowering pressures for accountability? A common assumption among progressives is that the sheer wealth of the richest contributes to their political power. But is it possible that the rich might actually get more power if they are increasingly responsible for financing the government? Is it possible that the pressures of democratic oversight might diminish in a world with fewer taxpayers? America already collects around 40% of incomes taxes from the top 1%, and our overall tax structure is already the world’s most progressive (which by itself would seem to argue against the political power of the richest). How much higher taxes are possible or desirable, and what mix of factors ought to go into structuring the balance? It’s hard to settle these issues, but questions over designing a sustainable tax-and-spend system are critical and are not being widely discussed.

Multiple Unrelated Issues?

What makes the inequality story so complicated as that America simultaneously faces three different, more or less unrelated, issues at different levels of income. On the top end, the incomes of a select few have been rising as with many other English-speaking countries. Even with lower marginal tax rate schedules, this group now contributes a massive amount of tax revenue.

In the middle, though real compensation seems to be rising at the pace of growth, the take-home income of households does seem to be stagnating. The chief issues here relate to how median worker compensation isn’t translating into cash income, due to presence of taxes, the role of health care and education, paired with other family structure problems. Broadly, it seems that people are signing off increasing chunks of their paychecks to the government and other highly regulated industries, which are then spending this money on their behalf.

The really troubling aspects of inequality are on the bottom end — with the low apparent mobility of low-skilled, particularly male, workers. The story here is complex but relates to issues like the role of lower unionization, higher immigration, incarceration, high implicit marginal tax rates, etc. that have been heavily discussed in this space.

However, the dynamics of all three groups — the top, middle, and bottom — appear to be fairly distinct. At best, tackling high-end inequality won’t do anything to address the issues faced by other Americans; at worst it will worsen incentives and so hurt overall growth at the margin, while entrenching a wealthy-dominated method of public finance.


The Latest