The overwork of the employed part of the working class swells the ranks of the reserve, whilst conversely the greater pressure that the latter by its competition exerts on the former, forces these to submit to overwork and to subjugation under the dictates of capital. The condemnation of one part of the working class to enforced idleness by the overwork of the other part, and the converse, becomes a means of enriching the individual capitalists, and accelerates at the same time the production of the industrial reserve army on a scale corresponding with the advance of social accumulation.

—Marx, Capital

Nearly three years since the ostensible end of the recession, the United States is still beset by over 15 percent underemployment — millions of Americans who would like to work full time and either can’t find jobs, or can only find part time work. We know from a large body of research that unemployment and underemployment have many negative consequences, not only financial but physical and mental. But this plague of underemployment exists alongside the corresponding problem of over-employment.

Our pre-eminent scholar of overemployment is the criminally obscure labor economist Lonnie Golden. In the broadest definition, overemployment happens whenever people are working longer hours than they say they would prefer. In a paper from a few years back, Golden gives a narrower definition: “workers’ inability to obtain reduced hours despite a willingness to proportionately sacrifice income.”

Overemployment is difficult to estimate precisely, because it is dependent on the wording of survey questions and the precise definition used. Jeremy Reynolds, the leading sociologist of preference mismatches in working time, reports in a recent paperthat more than two thirds of workers would like to work fewer hours than they actually do. This was in response to the question “If you could work just the number of hours in paid employment that you would like, how many hours per week would that be?” That wording is highly unusual because it asks about a specific ideal number of hours, rather than just the general desire to work “less,” and because it does not say anything about a proportional reduction of income.

The lowest estimates of overemployment come from the US Government’s Current Population Survey, which asks people if they “given the choice, (would) opt for more income and more hours, less income and fewer hours or the same income and hours?” and gives an overemployment rate of around 7 percent, even during recessions. Golden, in the paper linked above, surveyed eight other studies and found a range of estimates of the overemployment rate, from as low as 14 percent to a high of around 70 percent. None of those surveys asked for a specific hours target, while some of them specified that a reduction in hours would be linked to a reduction in income.

The assumption that reductions in hours should be linked to reductions in pay is in some sense a political one. It’s not common now, but demands like “thirty hours work for forty hours pay”have a long history in the labor movement. Asking for a cut in hours with no cut in pay is, in the end, just another way of asking for a raise.

But what’s really remarkable is that even when it’s presented as a pure trade-off between time and money, so many workers say they prefer fewer hours. Curiously, economists have chosen not to pay attention to this gross distortion of the labor market. Golden identifies the macro-economic causes of overemployment by analogy with theories about underemployment:

Structural Overemployment due to “structural incentives inherent in labour-market-related institutions or work organization that lengthen hours demanded per worker.” For example, the American system of employer-provided health insurance means that there are large fixed costs to hiring new workers. So it is cheaper to demand more hours from existing workers than to hire new workers, which creates a bias toward overemployment.

Cyclical Overemployment when “hours demanded per worker are rising faster than workers’ desired hours.” Economic analysts look to increases in hours worked as a leading indicator for increases in hiring in the future, and this should be regarded as a kind of cyclical overemployment.

Frictional Overemployment due to “barriers to full, perfect information among employers about their employees’ preferences and among worker applicants about job requirements and alternative jobs.” As someone who would generally prefer to take raises in the form of increased time rather than increased income, I can attest that it’s difficult to figure out whether and where such jobs exist.

In terms of politics and public policy, clearly the most significant factor is the structural one. And while liberals have worked tirelessly to kill off zombie tales about structural unemployment, there’s good reason to think we have a long-standing problem of structural over-employment.

If anything, the analysis of survey data understates the problem of overemployment. A worker’s preferences, after all, aren’t purely exogenous to their situation; they may “prefer” to work more because they feel it would be advantageous given their social and workplace context, but changing the context can change the preferences. In another paper, Golden and Morris Altman describe several factors that can shape worker preferences:

Signaling in the workplace. Workers may feel that they need to stay in the office as long as their coworkers, even if this extra time is not actually productive, in order to win promotions or avoid layoffs; this is sometimes referred to as “presenteeism.” It can lead to a straightforward collective action problem: most workers would be happier at a shorter-hours equilibrium, but cannot get there unless all of them move together.

Consumption of positional goods. Economists like Robert Frank have argued for the importance of “context externalities,” in which you judge the value of your consumption relative to those around you. This leads to an escalation of spending as everyone tries to make sure that their car, house, etc. are better than those of their neighbors. This higher spending creates a need for higher income, and hence a preference for longer hours. The result is a different kind of collective action problem, this one located in the consumption sphere rather than at the point of production.

Adaptive preferences. Even if the demand to work more hours is initially just an imposition from the boss, worker preferences may adapt over time. This can be a purely psychological process, but it also has a material dimension. If you work all the time, you will tend to pay for things that you otherwise might have done yourself: child-care, housecleaning, cooking, and so on. Once you become used to living this way, you need a higher level of income, and thus prefer longer hours of work, than would otherwise be the case.

Another explanation for the evolution of preferences is originally due to Ed Glaeser, et al:

The Social Multiplier. This is just the idea that leisure time is more valuable when one’s friends and family are also not working. Glaeser invoked this idea to explain the lower annual working hours of Europeans, relative to Americans.

Collective action problems in the formation of worker preferences can be overcome if workers can coordinate with each other, either by collectively bargaining through labor unions or by imposing state regulation on the labor market. Strongly enforced overtime laws with low thresholds can discourage presenteeism. Robert Frank has proposed progressive consumption taxes to dissuade the consumption arms race. And the existence of a social multiplier would mean that once a lower-hours equilibrium is established, the increased value of non-work time will tend to reshape worker preferences in the direction of shorter hours. Thus in combating overemployment, we will also tend to move society toward a lower-commodification equilibrium. (As discussed here and here).

Talking about overemployment in a jobless recovery might seem to confuse the issue. Is our problem that people should be working more, or working less? To steal a line from JW Mason, perhaps “some should do one, others should do the other.”