The earnings premium for overtime work revisited

by Jennifer Glass

Much has been written recently about the growing earnings premium for workers willing to work long hours, creating a new dimension on which (mostly male) professionals without care responsibilities can distinguish themselves and hoard positions of power and authority within organizations. These findings should be of serious concern to scholars of inequality, as well as those studying new forms of work organization or the persistence of gender stratification in the workplace.

One primary way for groups to hoard resources, of course, is to “move the goalposts” as disadvantaged groups gain leverage in high wage sectors of the economy. The history of racial exclusion is full of such attempts to increase qualifications or transform them so that racial preferences can be kept intact. Long work hours serve the same purpose in excluding workers with family responsibilities and their associated costs (lower availability, greater need for flexibility, etc.) from positions of power and authority.

But recently, my colleague Mary Noonan and I examined the issue in a different way by looking longitudinally within the careers of individuals who are salaried workers to find the premium for overtime work.

To our surprise, we discovered that actual reported overtime hours worked have little to do with earnings growth over the course of an employee’s career

Using fixed effects analysis with the NLSY 79 panel, observed from the start of their posteducational careers in the 1980’s until the pre-recession wave in 2008 when they were entering their 50’s, we find that the amount of overtime worked by individuals had very little impact on their subsequent earnings within a job spell with the same employer.

The amount of overtime work reported by salaried respondents during this period was not insignificant—they averaged around 5-6 hours of overtime per week during these years, using a standard 40 hour week as our benchmark.  Nor was there limited variation over the course of a job spell – many workers reported extensive overtime in some years and hardly any in others.

Often these hours were worked onsite, but many of these hours were worked from home and involved the use of computer mediated technologies (personal computers, email, smart phones, etc.). When we included the location of overtime work, the findings were even more striking.  Overtime worked from home or offsite had literally no significant dollar impact on subsequent earnings, while overtime worked onsite produced some subsequent earnings growth but nowhere near the hourly wage estimated for their first 40 hours worked each week.

Employers could reward workers and encourage longer hours by promoting them or paying them more after they show substantial increases in work hours. But this is not evident in the data. We incorporated markers of recession years in our model, changed the time lag to see future effects on earnings, and tried the usual array of other time-varying individual covariates of productivity (general health, childbearing for women, etc.) but the results were robust to their inclusion or exclusion.

What could possibly be happening here to explain why an overwork premium exists at the occupational level but that self-reported hours of overwork have virtually no impact on the wage trajectories of individuals?

We think the answer lies in decoupling hours worked and productivity. While some economists (and employers) continue to believe that face time equals productivity, many economists have argued for years that there is little evidence to suggest that productivity goes up monotonically with hours worked per week. In fact, marked inefficiencies start to occur in cognitive and mechanical tasks when workers extend their working time well beyond 40 per week.

Instead long hours may be best understood for their signaling function to employers and symbolic boundary-maintenance within work groups.

Thus, some occupations require consistently long hours (senior managers, physicians, associates in law firms) not because those are most efficient or necessary to ensure tasks get accomplished, but because requiring long hours sets a standard of commitment and exclusivity that most workers cannot meet.

We can (and should) explore historically how this standard came into being during the past 40 years; after all, “bankers’ hours” in the 1960’s were 9 a.m. to 3 p.m.! But the idea that overwork itself creates the “overwork premium” needs to be discarded. Clearly the heroic hours devoted to work at the upper echelons of corporate management have more to do with, say, rat race effects and sociologist Mary Blair-Loy’s cult of work devotion, not from any business necessity.

But if salaried workers are not better rewarded economically after a period of intense overwork, why then do salaried workers continue to perform this unnoticed overtime labor for their employers?

Perhaps workers perform overtime when they are simply given too much work to do, or are expected to be available during nonwork hours for electronic messages and unpredictable last-minute tasks. Perhaps employers successfully squeeze more work out of their remaining employees after layoffs or downsizing in the hopes of trimming costs rather than raising pay among those who are left. So their desire for job security results in acquiescence to employer demands.

Perhaps they face acute deadlines for project or team results where rewards are distributed without much consideration of individual time or effort.

But we need to begin to understand the forces driving work hours ever higher as a condition for access to earnings and authority.

Jennifer Glass is the Barbara Bush Professor of Liberal Arts in the Department of Sociology and Research Associate in the Population Research Center at the University of Texas, Austin. This article is summarizes findings from “Telecommuting and Earnings Trajectories Among American Women and Men 1989–2008” in the journal Social Forces.

Image: Konstantin Panphilov CC BY-SA 4.0 via Wikimedia Commons