What If Everything We Thought We Knew About Inequality Was Wrong?

A dominant political narrative might be almost entirely the product of misleading and inconsistent data.
January 12, 2018 • Commentary
This article appeared in UK Telegraph on January 12, 2018.

In economics, it is often said that when the United States sneezes, the rest of the world catches a cold. This might be even truer of political narratives or campaigns.

In recent years, the UK has imported some of the worst aspects of university student culture from the States. The concept of the living wage became a goal for campaigners here following US cities’ actions in the mid to late 2000s. But perhaps the biggest story shipped in from America has been the tale of vast and increasing income inequality.

Never mind that on most measures the income distribution is no more unequal today in the UK than it was over 30 years ago. The US debate brought a new target: the top 1pc. It is widely propagated that they have been motoring away from the rest, capturing most of the gains from growth over four decades.

The work of economists Thomas Piketty and Emmanuel Saez has been critical here. Assessing income using income tax data, they previously suggested that in the US, the income share of the top 1pc had increased hugely from 9pc in 1960 to 20.3pc in 2015. For the UK, they currently estimate the rise was less pronounced but still striking, going up from 9pc in 1960 to 14.5pc in 2013.

Now, I’m sceptical that aggregate snapshot inequality statistics tell us anything much useful in the first place. For a start, the top 1pc itself changes over time. And given that Walter Scheidel’s historical work suggests the only means of substantially slashing inequality have been mass mobilisation wars, communism, plagues and state failure, the cures may be somewhat worse than the disease.

But a new paper by economists Gerald Auten and David Splinter implies the conventional US narrative about the 1pc has been wrong all along. They show that tax returns may be a misleading source from which to judge trends, since they are strongly influenced by the pertaining tax code.

In the US, major reforms to business taxes in 1986 coupled with large cuts to top individual income tax rates from the Eighties onwards reversed incentives for business owners to operate as corporations and retain earnings inside the company, rather than paying out dividends or higher executive salaries.

The big increase in top incomes following the Eighties changes was then, to a large extent, a statistical phenomenon, driven by a bigger incentive to report income.

About 40pc of the increase in the top 1pc’s share of total income occurred just prior to or after the 1986 tax reform alone.

But that’s not the only problem with assessing inequality consistently from tax data. Such measures exclude government benefit payments and some of the non‐​taxable benefits provided to employees by employers. They also fail to account for changes in technical tax rules about who is required to file a return, and declining marriage rates, which change the composition of who actually does file.

To correct for all this, Auten and Splinter developed a measure of “consistent market income”, which includes retained corporate profits and adjusts for the 1986 tax reforms, employer‐​paid payroll taxes and insurance, and falling marriage rates. The idea is to get a measure of income that can be judged on a like‐​for‐​like basis over time.

The change to the story of the top 1pc is striking. Using this new measure, their income share in 1960 is found to be higher at 11.3pc (not 9pc), rising to just 14.9pc in 2015 (rather than 20.3pc).

In other words, correcting for data limitations and tax policy changes suggests the income share of the top 1pc has risen by only 3.6pc since 1960, much lower than the 11.3 percentage points calculated by Piketty‐​Saez. Even this ignores government benefits. Including these, the pre‐​tax income share of the top 1pc has increased by just 1.9pc since 1960, a figure 83pc lower than implied by Piketty-Saez’s methodology. This suggests that far from the top 1pc capturing 65pc of income growth since 1979 (as they believed), the real consistent figure (including taxes and benefits) is more like 10pc.

As far as I’m aware, nobody has attempted such a comprehensive analysis of how tax policy changes and marriage trends have affected income inequality statistics for the UK.

But a 2007 Anthony Atkinson paper estimated that for the Fifties and early Sixties, including retained company profits in income calculations increased United Kingdom top 1pc income shares by about half.

This suggests that here too the top 1pc’s increased income share might be far less pronounced than currently believed.

The debate over the appropriate data will rumble on, of course. But it’s remarkable that such a dominant political narrative might be almost entirely the product of misleading and inconsistent data. This is particularly pertinent as Democrats in the US and Labour in the UK have used these statistics as justification for major changes in policy, on everything from large increases in minimum wages through to higher taxes.

This new and important study adds grist to the mill for sceptics of using inequality stats to guide policy.

Inequality statistics are aggregates of millions of decisions, transfers, policies, inheritances and much else, and, as Auter and Splinter have shown, can be hugely shaped by big changes in policy, in turn distorting our true understanding of underlying economic trends.

Western economies such as the US and UK have big challenges, from low productivity through to unsustainable public finances. But in focusing on inequality, this work suggests that at least some of our political capital may have been invested in the wrong area.

About the Author
Ryan Bourne

R. Evan Scharf Chair for the Public Understanding of Economics, Cato Institute