What Will Workers Give Up to Avoid an Employer-Set Schedule?

The headline from Mas and Pallais (2017) is easy to misquote. Roughly 7,000 applicants to a national call center were each shown the same baseline job, 40 hours a week, Monday to Friday, 9am to 5pm, on site, next to one randomly assigned alternative, with the wage gap between the two randomized across a wide grid. The average applicant gave up $3.41 an hour, 20.1 percent of the $17 average wage shown, to avoid the version where the employer controlled the schedule on short notice.

The authors then ran supplementary arms built to split that premium into its parts, and the split does not land where most operators expect. A job with irregular hours that were consistent and given before the job began, with no week to week variation and no short notice, commanded $3.42 an hour, also 20.1 percent, statistically indistinguishable from the short-notice job. A Thursday to Monday weekend schedule drew $3.27, or 19.2 percent, and a Monday to Friday noon to 8pm schedule drew $2.39, or 14.1 percent. A 7am to 3pm morning schedule went the other way entirely, with applicants preferring it to 9 to 5 by $1.09 an hour, 6.4 percent of wages.

The authors' own conclusion is that the distaste for employer discretion reflects aversion to working non-standard hours rather than unpredictability in scheduling. The sharpest number is the shift comparison: 3pm to 11pm against 7am to 3pm carried a $5.20 gap, 30.6 percent of wages, and even the quartile least bothered by the late shift required $1.43, about 8.4 percent, which lands at the top of the 5 to 10 percent second-shift premium typically found in employer surveys. Three quarters of applicants, in other words, wanted more than a standard differential pays. If what you want is a price for advance notice on its own, this experiment's answer is close to zero.

The same experiment prices the flexibility that gets sold as universally wanted, and the averages are small. Choosing your own days and times, with total hours held fixed, was worth $0.48 an hour, 2.8 percent of wages. Choosing your own number of hours was mildly disliked, at minus $0.22, or minus 1.3 percent. Working from home was worth $1.33, 7.8 percent. A combined option bundling all three came to $1.17, 6.9 percent, close to the $1.59 sum of its components, so the three did not reinforce one another.

The shape of the distribution matters more than its center. About 60 percent of workers placed no value at all on being able to make their own schedule, a mass point at zero in the authors' non-parametric model, and the median valuation of flexible scheduling was zero or close to it. When the wage gap was set to zero, only 60 percent of applicants picked the flexible alternative, and about 20 percent chose to work exclusively on site even with no wage penalty for doing so. A part-time replication at 20 hours a week put the median at $0.55, still indistinguishable from zero.

What sustains a compensating differential is the tail. The top quartile would give up at least 10 percent of wages for the ability to make their own schedule, and at least 14 percent, $2.45 an hour or more, to work from home. Almost 40 percent of applicants would not take the employer-discretion job even if it paid 25 percent more than the Monday to Friday 9 to 5 position. Mas and Pallais caution that mean willingness to pay may differ substantially from marginal willingness to pay, which is a direct warning against costing a policy off the average alone.

The applicant pool is narrow, so the authors checked it twice. Reweighting the sample to the March 2016 CPS hourly workforce left the estimates close to intact, at $3.75 for avoiding employer discretion and $1.35 for working from home. A separate discrete choice module fielded in the nationally representative Understanding America Study put the employer-discretion premium higher, at 29.3 percent of wages across 1,614 respondents, with a 25th percentile of 15.9 percent and a 75th percentile of 42.7 percent.

Two features of that replication are worth pausing on. Its flexible-scheduling estimate, 2.5 percent, sits almost on top of the field experiment's 2.8 percent, which argues against applicants shading their answers to look hireable. The valuation also did not depend much on what people already had: respondents in regular-schedule jobs valued avoiding employer discretion at 30.5 percent and those already in irregular-schedule jobs at 26.9 percent, a difference of 3.6 percentage points that was not statistically significant. Among workers already holding employer-discretion jobs, the top quartile would give up at least 43 percent of earnings for a Monday to Friday 9 to 5 job.

The gender result is partly real and partly not, and the difference is worth getting right. Women were willing to give up $4.27 an hour, about 25 percent, to avoid employer discretion, against $2.11, about 12 percent, for men, a gap of $2.16 that was significant at the 5 percent level. The work-from-home gap ran the same direction, $1.59 for women against $0.68 for men, but the $0.91 difference was not statistically significant and should not be repeated as a finding. Women with young children were not willing to take a pay cut for flexible scheduling on average and were no higher than men even at the upper quantiles, though in the survey replication they would give up over a third of wages to avoid an irregular schedule.

The wage gaps were assigned at random and the choices were consequential, made inside a real application that determined job matching, with applicants told the choice would not affect hiring. That makes the comparisons causal in the sense that matters for pricing. What the study measures, though, is valuations rather than outcomes: it reports nothing about turnover, health, productivity, or earnings once people are working the schedule, and it records what applicants said they would accept rather than what they did over a year.

The sample is entry-level phone interviewer roles at roughly $16 to $19 an hour, disproportionately female, average age 33, about half with some college. Carrying these dollar figures over to salaried professionals, clinicians, or hospitality work is an assumption rather than a result. Precision deserves care too, because the paper publishes standard errors rather than confidence intervals: the $3.41 estimate carries a standard error of $0.42, so an approximate 95 percent interval runs from about $2.59 to $4.23, or roughly 15 to 25 percent of wages.

Two further limits travel with the tail numbers in particular. Headline estimates are corrected for inattention, since placebo tests showed roughly 25 percent of applicants were inattentive, and quantile estimates are more sensitive to that correction than the mean or median are, so the 10 percent, 14 percent, and 43 percent figures rest on a modeling choice in a way the 20 percent does not. Fieldwork also ran in 2015 and 2016, before remote work became ordinary and before the wave of US predictive-scheduling laws, so the 7.8 percent work-from-home valuation is a pre-2020 estimate and no cited study here re-measures it.