Attributing changes in food insecurity to a changing climate.

It is generally accepted that climate change is having a negative impact on food security. However, most of the literature variously focuses on the complex and many mechanisms linking climate stressors; the links with food production or productivity rather than food security; and future rather than current effects. In contrast, we investigate the extent to which current changes in food insecurity can be plausibly attributed to climate change. We combine food insecurity data for 83 countries from the FAO food insecurity experience scale (FIES) with reanalysed climate data from ERA5-Land, and use a panel data regression with time-varying coefficients. This framework allows us to estimate whether the relationship between food insecurity and temperature anomaly is changing over time. We also control for Human Development Index, and drought measured by six-month Standardized Precipitation Index. Our empirical findings suggest that for every 1 [Formula: see text] of temperature anomaly, severe global food insecurity has increased by 1.4% (95% CI 1.3-1.47) in 2014 but by 1.64% (95% CI 1.6-1.65) in 2019. This impact is higher in the case of moderate to severe food insecurity, with a 1 [Formula: see text] increase in temperature anomaly resulting in a 1.58% (95% CI 1.48-1.68) increase in 2014 but a 2.14% (95% CI 2.08-2.20) increase in 2019. Thus, the results show that the temperature anomaly has not only increased the probability of food insecurity, but the magnitude of this impact has increased over time. Our counterfactual analysis suggests that climate change has been responsible for reversing some of the improvements in food security that would otherwise have been realised, with the highest impact in Africa. Our analysis both provides more evidence of the costs of climate change, and as such the benefits of mitigation, and also highlights the importance of targeted and efficient policies to reduce food insecurity. These policies are likely to need to take into account local contexts, and might include efforts to increase crop yields, targeted safety nets, and behavioural programs to promote household resilience.