Food policy in a warming world

In the face of extreme heat damage to agricultural production, governments intervene to protect domestic consumers from high food prices. Examining the interaction between environmental shocks and food policy across the globe, this article shows that such intervention redistributes the economic losses from climate extremes and exacerbates the regressivity of global warming, which already disproportionately harms lower-income regions.

A version of this article previously appeared on VoxDev.

In March 2022, a heat wave in India reduced expected wheat output by over 100 million tonnes. On 13 May, the Indian government banned wheat exports, citing concerns that high wheat prices threatened food security. The policy was met with immediate controversy. Farmers criticised the government for increasing the financial burden of the drought by cutting off their ability to sell at high international prices (Yasir and Kim 2022). International observers noted that, by pushing up global prices, India’s policy might threaten food security elsewhere in the world (Lockett and Fildes 2022).

This story is not unique to India. Just in 2023, similar stories of extreme heat shocks, policy interventions, and controversy unfolded in Indonesia, Myanmar, Spain, Turkey, Kenya, Tanzania, and Morocco (Ghosal et al. 2023). Moreover, as planetary warming leads to more frequent and severe episodes of extreme heat, governments will likely face more such decisions in the coming decades. These policy responses will critically shape global capacity to adapt to a warming world.

In our research (Hsiao, Moscona and Sastry 2025), we investigate the interaction between environmental shocks and food policy around the world. Does food policy systematically react to these shocks, as the anecdotes above suggest? If so, how and why? And what do these policy responses imply for the consequences of climate change?

Extreme heat exposure and agricultural market interventions

To answer these questions, we compile a global dataset of extreme heat exposure and agricultural market interventions since 1980. We measure extreme heat exposure with global data on daily temperature realisations, combined with expert-elicited estimates of the maximum growing temperatures for specific crops. Our measure varies across time and space, given changes in the distribution of temperature shocks, and it varies across crops, given differential sensitivities of plant species to temperature shocks. We measure crop-specific agricultural policy interventions using data collected for the World Bank’s Distortions to Agricultural Incentives project (Anderson 2009). For a collection of markets, which span 85% of global agricultural production, the dataset summarises agricultural policies that include border taxes, quantity restrictions, and domestic support programmes in the form of the ‘nominal rate of assistance’ (NRA), a statistic that describes the net effect of policy on domestic prices.

Our main finding is that domestic extreme heat shocks systematically induce consumer assistance through policy interventions that reduce food prices. Figure 1 reports our baseline findings. The empirical methodology absorbs country-level and crop-level trends in extreme heat or policy, exploiting only country-by-crop-by-year variation. We find larger effects when focusing on economically important crops and staple crops. A change from the first to fourth quartile of extreme heat exposure results in a more than 30 percentage point reduction in NRA. That is, a country with no initial policy distortion would introduce a 30% consumer subsidy.

Figure 1. Extreme heat induces consumer assistance

Notes: (i) This figure displays the relationship between quartiles of extreme heat exposure and NRA. (ii) The unit of observation is a country-crop-year, and we include all possible two-way fixed effects. (iii) Each set of three bars corresponds to estimates from a single regression, and displays the effect on NRA of being in each quartile of extreme heat exposure, relative to the bottom quartile. The sample of crops included in each regression is noted below each set of bars. (iv) We report 90% confidence intervals. A 90% confidence interval indicates that if the study was repeated many times, the true effect would fall within this range in 90% of the cases.

Figure 2 illustrates the patterns in the raw data that underlie our finding, focusing on two major crops in India. The two panels show extreme heat exposure and the NRA for wheat and rice, staple crops in different parts of the country. Within India, our analysis implies that government intervention typically responds to lower prices after bad harvests in a targeted way. Our full analysis uses global variation in policy and heat exposure, both within and across countries, to show that this pattern is a systematic feature of agricultural policymaking.

Figure 2. Extreme heat exposure and agricultural policy in India

Note: The figure plots the changes in the NRA (left vertical axis) and extreme heat exposure (right vertical axis) in India from 1980-2020 for wheat and rice.

We next study how and when governments intervene. Going beyond our summary measure of price intervention, we show that extreme heat leads to lower border tariffs (as measured in the UN Trade Analysis Information database), and a higher likelihood of export restrictions (as measured in the Global Trade Alert database). Focusing on political incentives, we find that policy responses are significantly larger during election years. This finding is consistent with a large literature in political economy, which suggests that politicians may be especially willing to spend money to win the support of constituents near elections (for example, Alesina and Roubini 1992). We also show that policy responds not only to year-to-year fluctuations but also to slower-moving climate trends at the decadal frequency.

Finally, we study how policy responds to food shortages that originate abroad. We construct measures of global agricultural distress, as well as more targeted measures of extreme heat shocks to a given country’s import suppliers. In both cases, we find that foreign extreme heat shocks induce producer assistance that increases the price of food – that is, threats to food security originating overseas have the opposite effect of those originating domestically. These results are consistent with a model of government decision-making that is driven by domestic distributional considerations. However, they are inconsistent with a model in which governments’ primary goal is to solely reduce price fluctuations faced by consumers, regardless of the origin of these fluctuations. This finding also runs counter to popular narratives of food policy “contagion” (Ghosal et al. 2023) and accounts of food policy restrictions that spread from one country to another (de Guzman 2022).

Aggregate welfare effects and cross-country inequalities

How do these policy responses shape the aggregate welfare effects of extreme heat shocks? To address this question, we combine our empirical estimates with a quantitative model of trade, production, and policymaking. We focus on the historical extreme heat shocks in our study sample. In markets hit by an extreme heat shock, responsive policy shields domestic consumers from surplus losses. Domestic prices increase by 7.1% when policy is fixed, but by 5.9% when policy responses are incorporated – an 18% reduction in the price response. At the same time, policy interventions amplify producer surplus losses, which increase from 3.3% to nearly 8.8% – a nearly 200% difference. Foreign consumers also suffer due to higher equilibrium prices.

Policy responses to extreme heat shocks thus reshape the global distribution of economic damages, both within and across markets. What do these results imply for the distribution of temperature damages across countries? Figure 3 maps the effect of policy responses on total country-level welfare, which we compute as the sum of producer surplus, consumer surplus¹, and government revenue. There are major winners (blue-shaded countries) and losers (green-shaded countries) from responsive policy. In many cases, responsive policy increases or decreases the welfare impact of climate change by over 25%, and in some cases by over 100%.

Figure 3. The consequences of responsive policy for country-level welfare

Notes: (i) This figure maps extreme heat shock-induced changes in welfare under responsive policy, reported as a percentage difference relative to shock-induced changes in welfare under unresponsive policy. (ii) We aggregate to the country level by summing welfare across crops and year. (iii) Countries are color-coded based on the difference between their welfare in the observed world in which we take their policy responses as given, compared to a hypothetical world in which these policy responses are shut down. In blue-shaded countries, responsive policy increases welfare while in green-shaded countries, responsive policy reduces welfare.

Moreover, these policy responses exacerbate inequality across countries. The poorest countries and countries that are already most adversely affected by extreme heat are also those that lose most from responsive policy. These countries include those in South Asia, Southeast Asia, East Africa, and South America. On the other hand, rich countries and countries that are more shielded from extreme heat shocks – including the US, Canada, Australia, and much of Europe – benefit from responsive policy. The differences in these country-level policy effects are determined by the extent to which policy responses amplify baseline distortions. For example, in countries with pro-consumer policy at baseline –including many lower-income parts of the world like India and much of Africa – extreme heat shocks lead to even larger pro-consumer distortions, which reduce welfare. In countries with pro-producer policy at baseline – including many higher-income parts of the world like much of Europe and the US – temperature shocks reduce producer support, leading to lower distortions and higher welfare.

Climate damages depend crucially on government action

Global leaders in policymaking and diplomacy argue that “food security rests on trade” (Gurria and da Silva 2019). Yet our research suggests that governments are willing to alter food policy and restrict trade in response to extreme heat shocks to agricultural production. While there is a political logic to these interventions, and consumers in the most-affected places may stand to benefit, they come with potential costs to producers and consumers around the world. Policy changes thus redistribute the economic losses from climate extremes and, according to our estimates, exacerbate the regressivity of global warming, which already disproportionately harms lower-income parts of the world. Climate damages depend crucially on government action. Better understanding the political realities of this action will be an important avenue for future research.

Note:

Producer surplus is the difference between the price sellers receive and the minimum they’re willing to accept, while consumer surplus is the difference between what buyers are willing to pay and what they actually pay.