Inequality is overwhelmingly between firms, not within

Firming up Inequality: 

Many people believe that the rise in income inequality in the United States is due to the wage earnings of executives ballooning far greater than their employees are. It is easy to think of examples of CEOs who make far more than the ordinary employee, and indeed the claim is cited by prominent economists. Politicians like Bernie Sanders rail against the top 1%. This is even agreed upon by some prominent economists. Piketty, for example, writes in “Capital in the 21st Century” that “the primary reason for increased income inequality in recent decades is the rise of the supermanager.” 

Song, Price, Guvenen and Bloom 2015 are here to tell you that that’s all wrong. Income inequality has increased because inequality between firms has increased, not because income within firms has increased. They are able to use the tax records of a 1/16th of a percent sample of all Americans, and 100% of firms, to definitively answer why inequality rose between 1982 and 2012. 

The increase in inequality has indeed been large. Since 1982, the median worker has seen their real pay rise 20%, while the pay of the top one percent has seen their real wage rise 94%. However, this has been due to changes in the fortunes of the firms for which they work. The median worker works at a firm which has seen its mean real wages increase 25%, while the top one percent works at a firm which has seen its mean real wage increase 105%. The bulk of inequality is explained by differences between firms, not within firms.

The authors note that there are two hypotheses to explain why this has happened. One is that workers have increasingly sorted between firms. The other is that the firms themselves have had differences in productivity. They are not, alas, able to distinguish between the two hypotheses with the data they have, but later work will suggest 40% is attributable to sorting, and 20% attributable to firm differences, with the 40% remaining being unexplained simply due to limitations in precision their approach has. (Card, Cardoso, Heining, and Kline 2016)

This does track with (you guessed it) Kremer 1993. The O-ring theory of production predicts that not only will an increase in the complicatedness of the economy increase income inequality, but that workers will find themselves increasingly sorted into firms of similar skill. The most productive firms – or those which can tolerate the least error – will hire the best people for every job and pay more, even if the tasks to be done are identical. 

The lesson from this, I think, is to not lose sight of magnitudes in public policy. Particular executives are extremely visible, but they are not numerous. The vast bulk of income inequality has been highly skilled, well-paid people earning more. The top 1%, after all, is over three million people. In 2012, that’s all people with an income above $400,000 a year, which sweeps up many lawyers, tech workers, and other such conventional workers. The case for interfirm inequality may get stronger as you restrict the cutoff even more, though. Even the top one percent saw mean wages at their firm increase by more than their own, and wages should be right-skewed.

To be honest I find it inspiring that, right now, more than three million people have an income exceeding $785,000 a year. The world has never been better for those willing and able to produce value for others.