How extreme heat determines diets in rural India

While the link between extreme heat and crop loss is well documented, less is known about how this translates to undernutrition – an outcome with long-term consequences for health and productivity. Analysing data from rural India, this article finds that it leads to an increase in the number of strongly undernourished households in terms of calories, iron, and other nutrients. While some households cope by buying food grown elsewhere, the poorest remain highly vulnerable.

Nearly half of the global population lives in households that rely on agriculture for their livelihoods (Davis et al. 2023). These households are highly vulnerable to climate change (Kala et al. 2023), especially with rising temperatures damaging crops and reducing yields (Schlenker and Roberts 2009). While the link between extreme heat and crop loss is well documented (Aragon et al. 2021), we know less about how this translates to undernutrition – an outcome with long-term consequences for health and productivity. In principle, rural households can cope by buying food grown elsewhere. But for low-income farmers, this option may be out of reach due to limited savings, credit access, or government support (Lane 2024).

To better understand how rural households respond to heat, we examine the effects of heat on food consumption and nutrition in India – a country where over 900 million people live in rural areas (World Bank, 2024), and where both undernutrition and extreme heat are already widespread.

Measuring diet quality using undernourishment thresholds

In Stainier, Shah and Barreca (2025), we measure diet quality using responses to India’s National Sample Survey (NSS) from over 300,000 rural Indian households from 2003 to 2012. Each survey includes a detailed list of food items consumed over the past 30 days. We combine the survey responses with data on the nutrient composition of foods (Gopalan et al. 1989) to calculate household-level consumption of calories and six key nutrients: protein, iron, zinc, thiamine, niacin, and riboflavin. We also calculate dietary requirements for each household based on the number of residents of different ages and genders (Indian Council of Medical Research , 2010).

Poor diet quality is prevalent in the sample population. We define a household as experiencing ‘strong undernourishment’ for a given nutrient if it consumed below 80% of its recommended amount in the 30 days prior to the survey, and ‘extreme undernourishment’ as below 60%. Overall, 20% of households in our sample experience strong undernourishment of calories, with even higher rates of strong undernourishment for other nutrients such as iron (55%) and riboflavin (52%).

By using these undernourishment thresholds as outcome variables, we can identify distributional effects that aggregate measures of diet quality (for example, total calorie consumption) might miss. For example, it may be the case that most households are unaffected by a hot growing season, but that poorer households, without the means to adapt, suffer a drop in diet quality. In this case, aggregate calorie consumption may not change, but the percentage of strongly undernourished households would increase.

How does extreme heat affect diet quality?

Households grow crops during the growing season of one year (generally June to December, though it varies by location) and sell or eat in the next year. A hot growing season in a given district in 2011 likely threatens the food and income sources of its households in 2012. This could result in a higher-than-normal percentage of households experiencing undernutrition in that district in 2012. To test the effects of heat on diet quality, we compare how the prevalence of strong caloric undernutrition changes in the year after a hotter growing season compared to the year after a more moderate growing season.

Figure 1 shows that hot weather during a growing season increases the percentage of strongly calorically undernourished households in the next year. One additional day above 110°F increases the prevalence of strong caloric undernourishment by 0.36 percentage points, equivalent to 3.1 million additional people. We find that heat also increases strong or extreme undernourishment in iron, zinc, thiamine, and niacin.

Figure 1. The effect of growing season temperature on strong caloric undernourishment in the next year

Notably, we find no effect of heat on less severe measures of diet quality, such as average consumption or the percentage of households consuming below 100% of the recommended amount. Together, these findings suggest that extreme heat primarily affects the diet quality of households who are already suffering from low levels of nutrition.

How do households adapt to a hot growing season?

A hot growing season damages local crops. Households who can afford to buy food grown elsewhere might do so to supplement their diets. To study this potential adaptation strategy, we test the effect of a hot growing season on calorie and nutrient consumption, separately for home-grown and purchased foods. On average, around 65% of households in the sample consume some amount of home-grown foods, and 28% of total calories come from home-grown foods.

After an unusually hot growing season, we find that households decrease the amount of food they consume from home-grown sources and increase the amount of food they consume from purchases. Figure 2 illustrates this dynamic in the case of calories. A day above 110°F leads to a reduction of 4kCal per person per day (roughly 1g of uncooked rice) from home-grown sources, and a 5kCal per person per day increase in purchased food. This pattern holds true across all the nutrients we study, except for riboflavin. An increase in food purchases helps to explain the fact that we find no average effects of extreme heat on diet quality.

Figure 2. The effect of growing season temperature on calorie consumption, home-grown and purchased, in the next year

How might households pay for this food? The role of labour allocation

Rural households increase food purchases after a hot growing season, raising the question of how they may be paying for it. One possibility is that they supplement income with work in non-agricultural sectors. We test this possibility with additional NSS survey data asking individuals what sector they are working in. Using this data, we create two additional outcome variables: whether an individual is working in agriculture or working in non-agriculture.

In Figure 3, we plot the results from estimating regressions of these outcomes on temperatures from the previous growing season. The year after a hot growing season, fewer people work in agriculture, which may be due to a decreased amount of labour needed to harvest and process crops. In contrast, heat results in more people working outside of agriculture in the next year. This finding suggests that, at least in part, people are making up for local crop losses by supplementing their income with non-agricultural work. However, households may also be sacrificing non-food purchases to buy adequate amounts of food – a possibility for future research to consider.

Figure 3. The effect of growing season temperature on sector of work in the next year

A) Working in agriculture B) Working in non-agriculture.

Policy and future research on climate adaptation

Our findings on adaptive responses suggest that reliable access to agricultural and labour markets is important in facilitating household adaptation to heat shocks. Agricultural markets allow households to access food grown in areas that may not have experienced a heat shock, and labour markets allow workers to find alternative sources of employment that are less directly impacted by extreme heat.

Despite finding evidence of adaptation and no aggregate effects of heat on diet quality, we still find substantial negative effects for less well-off households. This finding suggests that adaptation strategies available to some households may be prohibitively costly for many others. Future research should focus on identifying the most effective interventions for these particularly vulnerable groups. For example, both governments and non-profits have begun implementing programmes to assist people after extreme weather events, through direct cash assistance (Premand and Stoeffler 2022), microloans (Lane 2024), or insurance policies (Climate Resilience for All , 2024). Policy-interested research can play an important role in comparing the effectiveness of these different models in mitigating negative nutritional shocks related to heat.

This article first appeared on VoxDev.