The evidence

How much does heat cut worker productivity?

On the days heat actually gets to a crew, about 35%. On an ordinary warm day, closer to 10%. Those two numbers come from different studies measuring different things, and the gap between them is the part almost everyone gets wrong.

35%

less productive on a heat-affected day

Self-reported by 1,214 of 1,726 employed Australians. Those workers were affected on about ten days that year, not all summer.

2%

of task performance, per degree Celsius

Measured between 25°C and 32°C, with no detectable effect below 25°C. A standard hot day lands near 10%.

The physical ceiling

What the body can still do as the heat climbs.

Self-reports tell you what workers noticed. Chamber studies tell you what their bodies could still deliver. Foster and colleagues put 40 people through 338 work sessions and measured physical work capacity against wet bulb globe temperature.

Physical work capacity lost, by wet bulb globe temperature. WBGT folds in humidity, radiant heat and air movement, so it reads lower than the number on a thermometer: WBGT 30°C is a genuinely punishing day on an exposed site, not a warm one. Markers are the reported anchor points, not a fitted curve.

The studies

Six papers worth reading in full.

Field surveys, chamber physiology, factory output data and global projections. They disagree in useful ways, and every one of them is linked.

Field survey Nature Climate Change · 2015

Workers were 35% less productive on days they had suffered from heat.

Sample
1,726 employed Australians; 1,214 reported loss
Method
Recall survey, two seasons, health-economics framing
Cost
US$655 per person per year; 0.33 to 0.47% of GDP

The single most quoted number in this field, and the most misused. Seventy per cent of respondents lost productivity to heat on at least one day, but the average was about ten days across twelve months. The 35% is a rate for those days, not an annual drag. Absenteeism was rarer and sharper: 7% missed work, averaging 4.4 days.

Zander KK, Botzen WJW, Oppermann E, Kjellstrom T, Garnett ST. Heat stress causes substantial labour productivity loss in Australia. Nature Climate Change 5, 647–651. doi:10.1038/nclimate2623

Chamber physiology Int. Journal of Biometeorology · 2021

Physical work capacity falls 10% by WBGT 18°C and 78% by WBGT 40°C.

Sample
40 participants, 338 work sessions
Method
Climate chambers, self-paced work, sigmoidal models
Fit
Heat index predicted capacity best, R² = 0.96

The physiological backbone under the survey numbers, and the reason the 25 to 50% band at WBGT 30 to 35°C matters: it brackets the 35% self-report from an entirely independent direction. Notably, skin temperature predicted capacity better than core temperature, which suggests workers pace themselves off how hot they feel before they are in real danger.

Foster J, Smallcombe JW, Hodder S, Jay O, Flouris AD, Nybo L, Havenith G. An advanced empirical model for quantifying the impact of heat and climate change on human physical work capacity. Int J Biometeorol 65, 1215–1229. doi:10.1007/s00484-021-02105-0

Meta-analysis The Lancet Planetary Health · 2018

30% of workers reported productivity loss during a shift in heat.

Sample
111 studies, 30 countries, 447 million workers
Method
PRISMA systematic review and meta-analysis
Also found
35% experienced occupational heat strain

Read the units carefully. This 30% is the share of workers reporting any loss, not the share of output lost, and the neighbouring 35% is the share experiencing heat strain. Both get quoted as though they measured lost production. They did not. Cite this one for prevalence, and Zander or Foster for magnitude.

Flouris AD, Dinas PC, Ioannou LG, Nybo L, Havenith G, Kenny GP, Kjellstrom T. Workers' health and productivity under occupational heat strain: a systematic review and meta-analysis. Lancet Planet Health 2(12), e521–e531. doi:10.1016/S2542-5196(18)30237-7

Economic field data Journal of Political Economy · 2021

Factory output fell about 2% per degree Celsius, and worker productivity 4 to 9% per degree on hot days.

Sample
58,000+ Indian factories, plus high-frequency plant microdata
Method
Panel regression on annual output and daily worker records
Also found
Higher absenteeism; climate control offsets much of the loss

The strongest evidence that this shows up in real output and not just in how people feel, because nobody was asked anything. It also makes the case for intervention directly: plants with climate control lost substantially less, which is the same mechanism a cooling vest uses on a body instead of a building.

Somanathan E, Somanathan R, Sudarshan A, Tewari M. The Impact of Temperature on Productivity and Labor Supply: Evidence from Indian Manufacturing. J Polit Econ 129(6), 1797–1827. doi:10.1086/713733

Indoor and light work Lawrence Berkeley National Lab · 2006

Performance dropped 2% per degree from 25°C to 32°C, and not at all below 25°C.

Sample
Pooled studies of measured task performance
Method
Weighted meta-analysis of objective work output
Peak
Best performance around 22°C

The reason a warehouse or plant floor is not a construction roof. Light indoor work holds up until about 25°C, then degrades in a straight, predictable line. This is the study behind the 10% figure for a standard hot day, and it is deliberately the conservative end of the evidence.

Seppanen O, Fisk WJ, Lei QH. Effect of temperature on task performance in office environment. Lawrence Berkeley National Laboratory, LBNL-60946. Full paper (PDF)

Global projection International Labour Organization · 2019

2.2% of all working hours lost to heat by 2030, or 80 million full-time jobs.

Scope
Global, modelled on 1.5°C warming
Sectors
Agriculture 60% of lost hours, construction 19%
Cost
About US$2.4 trillion a year

Useful for scale, useless for a budget. This is a projection of aggregate global hours, weighted heavily toward agriculture in southern Asia and western Africa. It should never be presented as what heat costs a specific American crew this year, which is exactly how it usually gets used.

International Labour Organization. Working on a warmer planet: The effect of heat stress on productivity and decent work. Geneva, 2019. ilo.org

Read this before you quote any of it

Four ways these numbers get misused.

  • 01A rate is not an annual average. The 35% applies to days a worker was actually suffering from heat, which in the Australian data was about ten days a year. Spreading it across every warm day of a season inflates the answer several times over.
  • 02Self-report is not measured output. Zander's figure is what workers said about themselves, and the authors flag the caution themselves. It is corroborated by chamber and factory data, which is why it is usable, but it is not a stopwatch.
  • 03WBGT is not the temperature on your phone. Wet bulb globe temperature accounts for humidity, sun and wind. WBGT 30°C can occur well below 100°F air temperature in humid conditions, and a dry 100°F day in shade may sit far lower.
  • 04Heavy outdoor work is not light indoor work. A roofer and a picker in a warehouse lose capacity on different curves. Using one number for both will overstate one and understate the other.

In practice

What we actually put in the calculator.

Our heat cost calculator prices what heat takes from a crew. These are the two rates it runs on and where each one comes from, so you can argue with them.

35% of a paid day, on an extreme heat day. Zander 2015, bracketed by Foster 2021.
10% of a paid day, on a standard hot day. Seppanen 2006, the conservative end.
0% on days that are not hot. No hot days produces no cost, by design.

We split the rate rather than applying 35% to a whole season, because the research does not support that and a procurement team will catch it. You set your own hot-day and extreme-day counts, so the split moves with your climate rather than ours.

Price it for your crew

Questions

Common questions, answered straight.

How much does heat reduce worker productivity?

On days when workers are actually affected by heat, the best field evidence puts the loss around 35%. Zander et al. (2015, Nature Climate Change) surveyed 1,726 employed Australians and found the 1,214 who reported heat-related productivity loss were on average 35% less productive on those days. On an ordinary warm day the loss is far smaller, closer to 10%, based on Seppanen, Fisk and Lei (2006), who measured a 2% drop in task performance per degree Celsius between 25°C and 32°C.

How many days a year does heat actually affect workers?

Fewer than most people assume. In Zander et al. (2015), the workers reporting productivity loss were affected on about ten days in the previous twelve months, and 7% of the full sample missed at least one day of work because of heat, averaging 4.4 days. This is why a 35% loss rate cannot be applied to every hot day of the year.

How much physical work capacity is lost to heat?

Foster et al. (2021) measured 338 work sessions across 40 participants in climate chambers and found physical work capacity falls about 10% at WBGT 18°C, roughly 25 to 50% at WBGT 30 to 35°C, and 78% at WBGT 40°C. WBGT is wet bulb globe temperature, which accounts for humidity, radiant heat and air movement, so it runs lower than the air temperature on a sunny site.

Is heat productivity loss the same for indoor and outdoor work?

No. Outdoor and heavy physical work loses capacity fastest because the body is already generating heat it cannot shed. Indoor and light work degrades more gradually: Seppanen, Fisk and Lei (2006) found no measurable effect between 21°C and 25°C, then a steady 2% decline per degree from 25°C to 32°C. Somanathan et al. (2021) found Indian factory output fell about 2% per degree Celsius annually, with worker productivity on hot days falling 4 to 9% per degree.

What is the global cost of heat-related productivity loss?

The International Labour Organization projects that 2.2% of total working hours worldwide will be lost to heat in 2030, equivalent to 80 million full-time jobs and about US$2.4 trillion. Agriculture accounts for roughly 60% of those lost hours and construction about 19%.

Summer 2026, 500-unit first batch

Put a number on it for your own crew.

Open the heat cost calculator Talk to the team
Heat cost calculatorEarly access