There is a math problem with Uber’s most powerful argument for classifying its drivers as independent contractors, not employees.  That argument emphasizes how weakly attached to Uber most drivers are, as evidenced by how many drivers work very few hours per week.  Uber CEO Travis Kalanick recently defended the company’s labor practices by citing company data that over half of all its drivers worked for Uber under 9 hours per week.  The math problem is that this argument treats all drivers as equally important to Uber’s business, when in fact a small proportion of drivers are doing most of Uber’s work.  Uber wants to direct attention away from the bulk of its business (the dog) and onto the moonlighting outliers (the tail.)

Uber presents driving as a “secondary” gig that supplements earnings from other jobs or from other household members.  The news release accompanying its latest driver survey highlights that “more people are using Uber on the side” and emphasizes how many women, parents, and students drive for Uber.  This framing is familiar from debates over the minimum wage.  There, supporters routinely emphasize the prevalence of “breadwinners” affected.  Critics invoke teenagers and married women working part-time, summoning a long history of devaluing women’s economic contributions.  (See here for an alternative understanding of the minimum wage divorced from household income.)

Public policy relevance aside, the prevalence of driving “on the side” also supports Uber’s legal argument against classifying its drivers as employees.  Moonlighting drivers are less economically dependent on Uber than those who drive full-time, at least if part-time status is a matter of choice.  That assessment, and that assumption, drives the prominent recent proposal by Seth Harris and Alan Krueger for a new “independent worker” category for gig workers who, they argue, fit poorly into employee status.  Even courts and commentators who disagree with their conclusion acknowledge that low hours and intermittent driving weigh against employee status under current law; countervailing considerations may just be more important on balance.

The problem is that this entire discussion assumes that proper regulatory and legal analysis should focus on the typical driver.  But when hours of work are at issue, that framework can be deeply misleading.  Consider a stylized workforce of 100 drivers. Twenty of them work 50 hours per week.  Eighty of them work 5 hours per week. Although only 20% of the workforce, the 50-hours drivers do over 70% of the work (1000 out of 1400 hours.)  There are many reasons for the drivers who do most of the work to be the center of attention.  Their lives are more deeply affected by their working conditions.  Their working conditions drive Uber’s business model, and in particular its ability to undercut competitors on price, something Harris once argued was essential to wage and hour law.

Uber has chosen not to present its driver data in a way that makes clear what proportion of its services are provided by drivers working over 30 hours per week, or by those working fewer than 10.  The same is true for the one detailed study of Uber drivers, a collaboration between Krueger and Uber executive Jonathan Hall.  That study found that 51% of drivers worked 15 hours per week or less, and only 19% worked full-time, defined as 35+ hours per week.  By applying some reasonable assumptions to this data, we can estimate which drivers did most of Uber’s work.  The table below presents that analysis based on the assumption that the average driver within each bracket works for the midpoint number of hours.  In other words, drivers in the 1-15 hours bracket average 7.5 hours per week.  That assumption is probably generous to Uber, given the overall pattern of drivers being concentrated at lower numbers of hours.

Notice that the much-touted majority of Uber drivers working very few hours per week are performing far less than the majority of the work.  And the seemingly marginal group of full-time drivers actually are doing about half the work, far more than those driving the fewest hours.  The precise numbers are sensitive to how one subdivides the drivers (using 1-10, 11-29, 30-44, and 45+ hours per week might look worse for Uber) and to the distribution within each bracket, but the basic point is robust.

Exactly what to make of this is hardly obvious.  Arguably, the circumstances of full-time drivers should be given weight above and beyond their aggregate contribution to Uber’s business, precisely because of their greater dependence on Uber.  More generally, employee classification law has not developed a robust analysis of job categories that are heterogeneous with respect to considerations bearing on employee status.  What should be done if driving for Uber is a close case, tipping toward employee status if one focuses on workers driving 20+ hours per week and performing the vast majority of Uber’s work but tipping toward independent contractor status if one focuses on the majority of drivers working fewer than 10 hours per week?  There are good reasons to think that the answer should be the same for all workers. At a minimum, we should be asking these questions, and we should be less impressed when Uber excitedly wags its long tail.