"Who's the Boss?" Problem Now Infecting Inequality Analysis

An NBER working paper has set off a cascade of takes about the nature of inequality in the US. The responses, however, seem to make some analytical mistakes.

From the abstract, here is what the paper purports to show:

Covering all U.S. firms between 1978 to 2012, we show that virtually all of the rise in earnings dispersion between workers is accounted for by increasing dispersion in average wages paid by the employers of these individuals. In contrast, pay differences within employers have remained virtually unchanged, a finding that is robust across industries, geographical regions, and firm size groups. Furthermore, the wage gap between the most highly paid employees within these firms (CEOs and high level executives) and the average employee has increased only by a small amount, refuting oft-made claims that such widening gaps account for a large fraction of rising inequality in the population.

To put it simply, the finding here is that the relative pay differences within firms (with firm boundaries defined in their technical legal sense) have not changed, but that firms increasingly have either a purely highly-paid workforce or a purely low-paid workforce. So, for example, whereas previously a big company might have had a full set of official employees, including low-paid janitors all the way up to chief executives, now that same company may consist of only highly-paid management with the janitorial work and temporary office work being outsourced to "different" firms.

This finding is interesting insofar as it shows how firms are evolving in the US, and the way in which companies are using outsourcing to get low-paid workers off their own payrolls and onto the payrolls of third parties. But it doesn't really negate the story that CEOs are jacking up their pay at much higher levels than their (actual, if not legal) workforce.

Nonetheless, that is exactly how it is being wrongly read. The most helpful of the incorrect reads thus far comes from Peter Orszag at Bloomberg View:

Consider Apple and McDonald’s. The Piketty perspective is that inequality rises because the top executives at both companies get much bigger raises than the rank-and-file. But the other possibility is that the average pay at Apple rises a lot relative to the average pay at McDonald’s, even as the wage gaps within each firm stay about the same.

What's so interesting about this incorrect read is that McDonald's is a perfect example of just how deceptive technical intrafirm inequality is. The vast majority of people that you and I would identify as McDonald's workers will not show up in the Master Earnings File as working for McDonald's. Instead, they will show up as working for one of tens of thousands of different franchise firms. (The same is also true, though in a more attenuated way, for Apple, whose products are manufactured by low-wage Chinese workers that are not technically employed by Apple, and obviously would not show up in US income records even if they were.)

So when calculating inequality on the firm level in this manner, you don't actually end up calculating what most people (including Orszag) think you are calculating. The pay of the McDonald's corporate CEO is not being compared to the average pay of all US employees of McDonald's franchises. Instead, it is being compared only to the average pay of the employees officially under the umbrella of the McDonald's corporate entity, which excludes at least 90% of what normal people would think of as "McDonald's employees."

Thus, this paper does not really challenge the argument that the pay of CEOs is rapidly outpacing the pay of their workforces. All that's happened is that the so-called "Who's the Boss?" problem (wherein complicated employment relationships are increasingly slicing and dicing workers into firms in ways that bear little relationship to who they are actually working for) has now come to corrupt even academic research into intrafirm distributions.

A CEO that cuts off (or never technically hires) the lowest-paid third of its (real) workforce, rehires them as workers of a separate firm, and then jacks up their own pay may not show up in the MEF as escalating their pay in ways totally detached from the average pay of their workforce, but that's what's really going on.