Occupational injuries impose substantial economic and social costs, including higher healthcare expenditures, absenteeism, early retirement, lost productivity, and diminished quality of life. According to the European Agency for Safety and Health at Work, work-related accidents and illnesses cost the European Union around 3.3% of GDP each year. Extreme temperatures are an increasingly important occupational hazard, but little is known about how temperature affects occupational safety, and more importantly, whether adaptation policies can effectively mitigate these risks.
In the Barcelona School of Economics Working Paper 1519, “When Policy Meets Weather: Extreme Temperatures and Workplace Safety”, Cristina Bellés-Obrero, Giulia Montresor and Catia Nicodemo estimate the causal effects of extreme temperatures on workplace accidents and assess how a reform to Spain’s heat-alert system led to a measurable reduction in accidents.
Both Cold and Heat Raise the Incidence of Work Accidents
The authors estimate a (Poisson) model on the number of workplace accidents, exploiting day-to-day variation in maximum temperature across provinces, within the same week, the same region-year-month, and the same province-year (i.e. sets of fixed effects), and controlling for other confounding factors. They document a U-shaped relationship between daily maximum temperature and workplace accidents: risks increase at both low and high tails of the temperature distribution (Figure 1).
Relative to 10–15°C days:
Extreme cold: Days with maximum temperatures below 0°C increase workplace accidents by 16%
- Who’s affected the most? Cold temperatures disproportionately increase accidents for women, older, native-born, and permanent workers, particularly during commutes
- Type of injuries: Extreme cold causes more severe but non-lethal accidents
Extreme heat: Days between 35-40°C increase workplace accidents by 5.5%
- Who’s affected the most? Extreme heat poses greater dangers for men, blue-collar workers, and those on temporary contracts, especially in outdoor settings.
- Type of injuries: Heat also drives sharp increases in fatal accidents and in temperature-specific injuries.
Do workers and firms adapt over time?
The paper finds no evidence that temperature-related accident risks declined over the period 2004 to 2019, even in regions more regularly exposed to extreme conditions, highlighting the importance of policy-led adaptation.

Notes: This figure plots the estimated coefficients from the Poisson fixed-effects model, estimated via pseudo-maximum likelihood. Coefficients are expressed as percentage changes relative to the baseline temperature bin (10–15°C) by exponentiating and subtracting one. The model includes controls for precipitation, wind speed, province-by-week fixed effects, region-by-year-month fixed effects, a national holidays indicator, and day-of-week fixed effects. Temperature bins range from 0°C and below to greater than 40°C in 5°C increments. The omitted category is the temperature bin with daily maximum temperatures between 10 and 15°C. Standard errors are clustered at the region-year level. Solid lines denote the 95% confidence intervals.
Spain’s 2015 Reform of the National Heat Alert System Reduced Workplace Accidents
Spain introduced a Heat Health Prevention Plan (HHPP) in 2004 after the 2003 European heatwave. A core element of the HHPP is the province-specific heat alert system, which operates from June 1 to September 15 and triggers alerts when forecast maximum/minimum temperatures exceed predefined thresholds.
What changed in 2015?
Before 2015, thresholds were based purely on climatology data (summer temperature percentiles). The 2015 reform updated these province-level thresholds to incorporate epidemiological (mortality-risk) evidence, aligning alerts more closely with actual health risks (Figure 2).
The authors estimate the impact of this reform using a difference-in-differences and event-study design, comparing accident counts in “treated” provinces (where thresholds were reduced) with those in “untreated” provinces (unchanged thresholds), focusing on summer days that are particularly hot.

Notes: This figure presents the minimum and maximum temperature thresholds that trigger heat-risk alerts under Spain’s National Heat Health Prevention Plan before and after the 2015 reform.
They find that:
- Workplace accidents fell by 7.6%, on average, because of the policy change
- Who benefited the most? The reform had larger effects for temporary workers (11.1%) and temperature-related injuries (9.7%)

Notes: This figure plots event study estimates from the Poisson pseudo–maximum likelihood. Coefficients correspond to interactions between the treatment indicator and year dummies, with 2014 as the omitted (baseline) category, and are expressed as percentage changes by exponentiating and subtracting one. The treated group comprises provinces that, following the 2015 reform, experienced a reduction in either the maximum or minimum temperature thresholds required to trigger heat-risk alerts. The model includes controls for precipitation, wind speed, province, year, month, day of the week, and national holiday fixed effects. Standard errors are clustered at the province level. Solid lines represent 95% confidence intervals. The p-value from a joint test of the null hypothesis that all pre-reform interaction coefficients are zero is 0.872.
Overall, this study provides robust evidence that both cold and heat increase occupational injury risk, but through different channels. Their findings also demonstrate that well-designed adaptation policies, such as Spain’s heat alert system reform, can mitigate the adverse effects of extreme temperatures on workplace safety. These results suggest that public health interventions can generate meaningful spillovers in occupational safety, reinforcing the case for targeted interventions as temperature extremes become more frequent.

