Tax Units vs. Households

by Reihan Salam

There’s this funny tensions in our ongoing conversations about income inequality. Those who are alarmed by growing inequality and the living standards of households in the lower half of the income distribution tend to think that we should do more through taxes and transfers to make up for the deterioration of the labor market position of less-skilled and mid-skilled workers. That is a perfectly respectable position. But it is important to have a clear picture of how much we already do via taxes and transfers. That is where a 2011 working paper by Richard Burkhauser, Jeff Larimore, and Kosali Simon (“A ‘Second Opinion’ on the Economic Health of the American Middle Class”) comes in handy.

Those of you who follow inequality debates are familiar with the work of Piketty and Saez, both of whom were profiled just this week by Annie Lowrey. There is much that Piketty and Saez say that I disagree with, e.g., I’m less sanguine about the potential impact of much higher marginal tax rates (see Arpit Gupta’s classic post on labor supply and taxes). But they have made a significant contribution to our understanding of how taxable income has changed over time across tax units: 

The two economists’ project of mapping income inequality started two decades ago, when Mr. Saez was teaching at Harvard and Mr. Piketty teaching down the road at the Massachusetts Institute of Technology.

Their innovation was to measure American income inequality historically. Existing data went back only to the 1970s. Tedious archival research at the Internal Revenue Service allowed them to stretch the data all the way back to 1913.

Once they had collected the data, the computation was easy. They figured out the benchmark for various income levels — the top 10 percent, top 1 percent and top 0.1 percent of earners, for instance — and calculated what share of income each group took each year.

What they found startled them. As in other industrially advanced countries, income inequality in the United States fell after World War II, a period that economic historians call the “Great Compression,” and remained stable through much of the 1970s.

But then inequality started increasing again, with the top 1 percent of earners drawing a bigger and bigger share of overall income. Their graph showing the trend became well-known: a deep U, with inequality as acute today as it was just before the depression.

It is now entirely controversial that inequality has increased significantly. It would be hard to imagine that it would not, given the acceleration of routine biased technical change, the increased strength of managers relative to less-skilled and mid-skilled workers, increasing returns to experience and education, etc. 

But it is important to have a clear sense of what we’re talking about. Do we care about inequality of tax units or inequality of households? As Burkhauser, Larrimore, and Simon (BLS) emphasize, focusing on tax units rather than households yields very different results regarding growth in middle class incomes. In the view of BLS, households are the more appropriate sharing units because family structure has changed and the rise of cohabitation means that adults sharing a household might be in separate tax units. This is not an entirely uncontroversial view. For example, there could be a material difference between how a cohabiting couple shares resources and a married couple shares resources. It does, however, seem reasonable to believe that there is some sharing of resources, and that looking at tax units would give us a less than complete picture.

The trouble, however, is that the IRS data on tax units has generally been more useful in understanding broad trends about income than the Current Population Survey data that tracks households. BLS have taken various steps to address these concerns, in this paper and in a new paper by Burkhauser, Shuaizhang Feng, Stephen Jenkins, and Jeff Larrimore that draws on CPS data concerning top incomes that is not publicly available (to protect the privacy of extremely high-earners) and in doing so successfully reconciles CPS household data with IRS tax unit data.  

To return to the BLS paper, the authors demonstrate that plausible assumptions about household sharing units suggest that middle incomes have grown much more than the tax data would suggest. “Deductible consumption,” in the form of employer-sponsored medical insurance, also improves the picture somewhat. Middle income households have also received larger transfers over time and they’ve seen their tax burden decrease. Taken together, we’ve seen a modest improvement in post-tax-and-transfer disposable income over the period from 1979 to 2007. Richard Burkhauser offers a very thoughtful discussion of the underlying issues on a recent episode of EconTalk, which is how I first encountered his work. James Pethokoukis recently interviewed Burkhauser on the same theme.

My interpretation of BLS is somewhat different from Pethokoukis. One of the ironies with Burkhauser’s finding is that changes in taxes and transfers really have made a significant difference for middle income households during a period in which the labor market position of many workers deteriorated, yet at least some people on the center-left are reluctant to acknowledge the extent to which this is true. Part of the reason might be that acknowledging the modest success of this redistributive efforts would undermine the political urgency of increasing transfers even more. Those on the right, meanwhile, are both inclined to see the BLS findings as an implicit rebuke to a narrative of middle class stagnation and decline, yet we are also less enthusiastic about transfers and steeply progressive income taxes, which are both an important part of why we’ve seen some improvement in disposable incomes at the middle. I am sympathetic to the view that there is a counterfactual world in which a more lightly regulated economy and a more efficient public sector would have yielded better results for households in the lower half of the income spectrum in the absence of increased transfers, but this is necessarily speculative. 

So in a sense BLS have left us with an awkward story that neither the left nor the right really wants to claim.