Extreme heat events are becoming more frequent and intense with climate change. Children are particularly vulnerable to such environmental stressors. This column uses nationwide exam data from Japanese primary and secondary schools to explore how heat exposure affects student learning. Heat exposure lowers student test scores, with disadvantaged students experiencing a greater reduction in scores. Access to air conditioning in schools substantially offsets these adverse effects and especially benefits lower performers. Investing in school air conditioning could promote both efficiency and equity.
With climate change, extreme heat events are occurring with greater frequency and intensity. Children are especially susceptible to these environmental stressors, owing to their physiological immaturity. It is therefore vital to examine how classroom heat exposure affects learning, since cognitive skills are closely tied to later labour market outcomes and broader economic inequality.
In recent years, research has shown that climate change affects numerous societal outcomes, including health, agriculture, labour productivity, income, cognition, and conflict (Carleton and Hsiang 2016). Yet most studies measure only average impacts, leaving distributional effects – such as who disproportionately bears the burden of heat-related damage – and their implications for inequality largely unexplored. For example, Cho (2017) and Park et al. (2020) document the cumulative impact of heat exposure on student learning. However, both focus on high school students in Korea and the US who are preparing for college entrance and are more likely to come from higher socioeconomic backgrounds. This makes them less suitable for addressing a central question: who bears the greatest burden of heat-related learning loss?
To address this question, our new study (Akesaka and Shigeoka 2025) analyses test scores from nationwide exams in Japan between 2007 and 2019, covering virtually all public school students in grades six and nine. The dataset encompasses an extensive sample of approximately 22.8 million students. The combination of individual-level data and the exams’ national coverage provides a rare opportunity to examine, for the first time, how heat exposure affects students across socioeconomic status. We focus on whether the cumulative effects of heat fall more heavily on disadvantaged students, thereby widening pre-existing educational inequalities.
Why focus on classroom heat exposure?
Analysing the effects of classroom heat exposure on student learning is important for three reasons. First, children are particularly vulnerable to environmental stressors such as heat due to their physiological and neurological immaturity (Rowland 2008). Second, harm incurred during childhood can have long-term adverse consequences on educational attainment and labour market outcomes (Cunha and Heckman 2007). Third, improving the learning environment within schools – particularly through infrastructural investments – is a feasible and policy-relevant measure to mitigate the adverse effects of climate change (Biasi et al. 2025).
How to measure the impact of heat exposure on student learning
To measure academic performance, we use test scores from 2007 to 2019 for Japan’s National Assessment of Academic Ability, a nationwide exam given every April (National Institute for Educational Policy Research 2021). Administered by the Ministry of Education, Culture, Sports, Science and Technology, the test covers sixth graders in primary school and third-year students in secondary school (ninth-grade students), with participation from virtually every public school in the country. Our analysis focuses on combined scores in math and Japanese language, the two subjects that were measured consistently throughout the study period.
To assess the cumulative effect of heat – rather than the immediate impact of taking a test on a hot day – we use temperature data from the year preceding each exam. First, we matched each of the approximately 20,000 primary and 10,000 secondary schools to the nearest meteorological station with continuous data available from 2006 to 2018, yielding 899 usable stations. Then, for each school year, we recorded the number of school days with maximum temperatures in the following bins: ≤6°C, 6–10°C, 10–14°C, 14–18°C, 18–22°C, 22–26°C, 26–30°C, 30–34°C, and >34°C.
We then estimated the relationship between these temperature exposures and the subsequent year’s test scores, controlling for school fixed effects. This setup allows us to compare the test scores of students in the same school who experienced hotter summers with those exposed to milder conditions, helping us isolate the effect of temperature itself on learning.
Students with lower academic performance are disproportionately affected
Our analysis reveals a negative association between the number of hot days during the prior school year and student test scores. Notably, the detrimental effects are not uniformly distributed: students in the lowest academic decile experienced a reduction in scores approximately three times greater than those in the highest decile. Quantitatively, each additional day above 34°C during the school days reduced scores by 0.30% standard deviations among the bottom 10% of students, compared to 0.09% standard deviation for the top 10%.
The role of socioeconomic status
Lower-achieving students tend to come from lower socioeconomic backgrounds, typically defined based on parental education and household income (Figure 1). Our findings suggest that heat exacerbates existing educational disparities by disproportionately impacting students from disadvantaged households.
[Click to enlarge]
This interpretation is further supported by parental survey data from 2013 and 2017, conducted alongside the National Assessment of Academic Ability. These surveys show that children from higher socioeconomic backgrounds benefit from greater per-child educational expenditures and more time in extracurricular academic activities, such as tutoring schools (juku). These resource advantages may buffer the impact of adverse classroom conditions from summer heat.
Given the strong link between cognitive ability and future labour market success, the results raise concerns that climate change could reinforce intergenerational inequality by deepening the link between socioeconomic status and educational outcomes.
Air conditioning as an equalising intervention
Importantly, our study finds that the installation of air conditioning in regular classrooms significantly offsets the negative effects of heat. Air conditioning is the main technology for adapting to heat (Barreca et al. 2016), but its widespread adoption in public primary and secondary schools in Japan has occurred only recently. During the sample period from 2006 to 2018, air-conditioning coverage in public primary and secondary schools increased from approximately 10% to 50%.
Using data from 2018, the last year in the sample period, schools in municipalities with 100% installation were classified as ‘with air conditioning’ (43% of sample), while those with 0% were ‘without air conditioning’ (16%), as displayed in Figure 2.
[Click to enlarge]
In schools without air conditioning, each additional day above 34°C reduced scores by 0.56% standard deviation on average, while in air-conditioned schools, the reduction was only 0.15% standard deviation, representing a mitigation rate of approximately 73% (Figure 3).
Most importantly, the benefits of air conditioning were greatest among lower-performing students. To quantify this, we examined student performance by academic percentile (10th, 25th, 50th, 75th, 90th) within schools and analysed the impact of high-temperature exposure, stratified by the presence or absence of air conditioning. As Figure 4 shows, in schools without air conditioning, heat disproportionately harms lower-ranked students (panel A). But in schools with air conditioning, nearly all the negative effects disappear across ranks (panel B), resulting in a larger benefit for low-achieving students. These results underscore how adequate investment in school infrastructure can mitigate unevenly distributed damage caused by heat, thereby promoting both efficiency and equity.
[Click to enlarge]
Policy implications
This study contributes to our understanding of one of the mechanisms by which climate change may exacerbate economic inequality: through its effects on educational achievement. Our results highlight the potential for public investment in school infrastructure, especially classroom climate control, to serve as an effective policy response to these emerging challenges.
As of September 2024, the rate of air-conditioning installation in public primary and secondary school classrooms in Japan had reached 99%, reflecting a rapid policy shift in the country. While energy efficiency and cost concerns remain important, ensuring a safe and conducive learning environment is essential for protecting the educational opportunities – and thus the future potential – of the next generation.
Editors’ note: The main research on which this column is based first appeared as a discussion paper of the Research Institute of Economy, Trade and Industry (RIETI) of Japan.
This article first appeared on VoxEU on September 25, 2025. Reproduced with permission.
