There are as many narratives about income inequality as there are papers. Over the last few days, two reports have appeared in stark contrast with one another. An 80-page note by Deutsche Bank sheds light on dramatically increasing income inequality in the U.S. and other Organization for Economic Cooperation and Development (OECD) economies.
In line with several earlier academic studies, it shows that the share of income going to the top 10 percent in the United States now exceeds 50 percent, and the top 1 percent receives about 23 percent of all income.
While all income groups have seen an increase in average real (after-tax) incomes, the largest increases have been at the top of the income distribution, where incomes have grown by more than 200 percent since 1979. In contrast, for low- and middle-income households, incomes have increased by less than 50 percent.
However, a more in-depth academic research paper by economists Gerald Auten and David Splinter finds little evidence of widening income inequality and questions the methodology used in earlier papers that find the opposite. So what narrative do we believe?
On the face of it, it is easy to accept the story that those with high incomes are doing much better than before, more so than middle- and low-income households. There is no denying that globalization and automation have hit low-skill, low-wage work more than they have affected better educated, skilled and high-wage workers.
Workers at the lower end struggle to keep up with changes in the labor market that they are less equipped to handle, as some specialized skills become obsolete and workers lack the skills to transition to new sectors.
But the reality is not that simple. A growing literature in economics has identified problems with measuring income accurately and completely in various datasets. If income is under-reported, particularly for low-income workers, then we may overstate the rise in inequality.
Moreover, income does not fully capture a household’s standard of living. Among other issues, at very young or very old ages, individuals may borrow or rely on lifetime savings to maintain their standard of living. Income may not perfectly capture how well off people are at different points in the life cycle.
Let’s begin with the measurement issues. Research in economics has shown that when households are surveyed, individuals don’t always accurately report benefits and transfer payments such as Medicare, Medicaid and Food Stamps.
However, such programs have grown in importance over the last several decades precisely to supplement incomes at the bottom of the distribution. Economists Bruce Meyer and James Sullivan show that when comparing data from the Current Population Survey to administrative data aggregates (the most accurate data), the ratio of reported benefits to actual benefits is 0.6 for Food Stamps and 0.5 for TANF.
In other words, receipts reported on household surveys are more than 40- to 50-percent lower than those in administrative data. Hence, measurement issues explain much of why trends in income inequality vary widely across different studies.
Using tax return data, Piketty and Saez show that between 1960 and 2015, the income share of those at the top increased by 11 percentage points. However, the recent paper by Auten and Splinter finds serious flaws with this analysis because tax return data similarly misses government transfer payments and non-taxable employer-provided benefits.
To overcome this challenge, Auten and Splinter use a broader measure of income, which they term “consistent market income,” that includes employer-paid payroll taxes and insurance. After adjusting for differential marriage rates between high- and low-income households and under-reported incomes, this paper finds that the income share of the top 1 percent increased by less than 4 percentage points over the same period.
The after-tax income share for the top 1 percent has grown even less, from 8.5 percent in 1960 to 10.2 percent in 2015. Hence, measurement issues can significantly complicate our understanding of trends in income inequality.
A second reason not to rely exclusively on income data is that income is not a great measure of welfare for most households. For one, it underrepresents the ability of households to rely on assets or savings in times of low income and the ability to save when incomes are high.
In other words, people may be better off than their incomes would suggest because their standard of living is propped up and smoothed over time by their assets and savings. More importantly, in light of the fact that income data often miss households’ ability to access the government safety net, consumption may better reflect households’ overall access to cash and benefits.
Meyer and Sullivan show that while overall income inequality has risen 29 percent over the past five decades, the increase in consumption inequality has been far more modest at 7 percent. Since 2005, the two inequality measures have moved in opposite directions, with income inequality rising and consumption inequality falling.
An earlier paper that I co-authored with Kevin Hassett also finds that consumption inequality showed little change through the entire period between 1984 and 2010, even though income inequality rose significantly over the same time frame.
In addition, using data from the Residential Energy Consumption Survey, we show that since the 1980s, many more low-income households now have access to air-conditioning and heating within their homes, internet, computers and printing facilities, microwaves, dishwashers and other household appliances.
In other words, access to material goods has increased significantly for the lowest income households, and the gap in access to these goods between high and low income households has narrowed over time.
While some may have already hit the panic button on widening income inequality narratives, serious research has yet to credibly confirm this view. Instead, the increasingly more prevalent view is that measuring household incomes accurately, and studying income inequality trends, is fraught with problems. Perhaps, if we truly care about capturing household well-being, consumption may be a better measure of standards of living.
But I think the narrative needs to move beyond the static concept of income and consumption inequality to the more dynamic concept of economic mobility. Are low-income people today able to access opportunities to move up the income ladder?
Can they access good schools for their children and good jobs for themselves? How do we make single-parent families more upwardly mobile? Our energies should be focused not, as they too often are, on why incomes are growing rapidly at the top and ways to constrain that growth, but on improving wage growth and mobility at the bottom of the distribution.
If we are successful at that, we may remain an unequal society, but inequality may not be as much of a dilemma as it appears today.