US employees spent an estimated 3.5 additional minutes on non-work activities at their workplace, excluding eating, on days reaching at least 100°F (37.8°C), a University of Zaragoza study found. The result points to changes in how people use time at work, although it does not measure a decline in output.
The research compared those days with daily maximum temperatures of 75°F to below 80°F, approximately 23.9°C to 26.7°C. The association was more pronounced among women and employees paid by the hour.
Ignacio Belloc, José Ignacio Gimenez-Nadal and José Alberto Molina analyzed US time diaries from 2003–2019 for their paper in Economics & Human Biology, published online in July 2026. Their sample contained 31,035 observations across 402 counties.
For employers, the findings concern activity during the working day, beyond whether someone attends a shift. They also raise a measurement problem: time recorded as non-work is not necessarily time that has no value to the business.
A change in minutes, not a measured productivity loss
The researchers used the American Time Use Survey, in which participants describe their activities over a 24-hour period. Records of activity, duration and location allowed the team to identify non-work activities taking place at a workplace.
They matched those diaries with county-level weather information and accounted for factors including occupation, industry, personal characteristics and other weather conditions. Each participant supplied one diary, so the analysis compares different people rather than following the same employee through hot and cooler days.
The precise estimate was 3.52 extra minutes. The authors expressed that as 35.9% of the sample’s average non-work time excluding eating. It is not a 35.9% reduction in productivity, despite a university news release using a productivity-loss headline.
The broader measure also produced a different result. When eating time was included, extreme temperatures had no statistically significant association with total non-work time at the workplace.
Most of the additional time in the narrower measure fell within leisure and exercise activities. The classifications cannot establish whether each activity reduced useful output. Listening to a broadcast could supply job-related information, for example, while social interaction could serve a professional purpose.
Hourly workers showed a stronger response
Among hourly employees, days reaching at least 100°F were associated with roughly 4.2 additional minutes of non-work activity, excluding eating. The researchers found no statistically significant relationship at that threshold among salaried employees.
One possible explanation is scheduling flexibility. Workers who must remain on site for set hours may have fewer options to move tasks to cooler periods. The paper discusses that explanation, but the data do not establish why the groups responded differently.
Women recorded an estimated 4.38 additional minutes at the same temperature threshold. The corresponding estimate for men was not statistically significant. Those findings do not prove that heat leaves men unaffected or establish a biological explanation for the pattern.
The authors also examined historical climate and economic conditions. Some results were consistent with adaptation to hotter climates and greater freedom to adjust work during economic expansions. These are possible mechanisms, not direct observations of cooling systems or employees’ bargaining decisions.
Cooling and breaks belong in operational planning
Employers should be cautious about treating every heat-related pause as avoidable downtime. The US Occupational Safety and Health Administration’s heat guidance says workers need cool drinking water and a cool place to recover. It recommends increasing the length and frequency of rest breaks as heat stress rises.
The study’s authors propose flexible schedules, remote work where feasible and better workplace cooling as possible responses. The research did not test those interventions or calculate whether their financial benefits exceed their costs.
Outdoor weather was measured, but individual workplace conditions were not. The survey cannot show whether a respondent had effective air conditioning, worked beside a heat source or faced another form of exposure. Its pre-2020 data also omit the subsequent expansion of remote and hybrid work.
A business assessing its own exposure would need information beyond attendance or recorded pauses, including working conditions, completed tasks, errors and safety outcomes. The study identifies a change in time allocation; it leaves the effect on output, and the return on particular workplace improvements, unresolved.
