Research indicates that heat and drought driven by global warming are currently costing corn, wheat, and soybean yields approximately $20 billion annually. This figure could potentially escalate to over $160 billion by 2100 unless significant emissions reductions are achieved.

Large agricultural countries like the United States will experience the greatest economic losses, but the most pronounced effects will be felt in low-income nations. Countries with a significant portion of the population in agriculture can expect profound impacts, warns Fan Yilin from the International Institute for Applied Systems Analysis (IIASA) in Austria. “The repercussions are even more acute in the least developed countries in Africa,” she cautions, with possible consequences including social unrest and increased migration.

Forecasting such economic impacts involves considerable uncertainty as it largely depends on how farmers will adapt to ongoing climate changes, whether through crop diversification or enhanced irrigation methods. The intention of this study is to raise awareness and promote adaptability, aiming to ensure that these predictions are overly cautious. As Kai Kornhuber at IIASA notes, “Our goal as climate scientists is to advocate for timely action; ideally, we want our predictions to appear overly pessimistic.”

The research team began by gathering data on the specific yields of maize, wheat, and soybeans from the United Nations Food and Agriculture Organization (FAO). They assessed drought conditions using a standard methodology that analyses historical climate data along with estimating soil moisture levels based on rainfall and evaporation.

By comparing historical incidents of extreme heat and drought to agricultural yields from 1974 to 2004, the team estimated the impact of these factors on yields. Statistical correlations were subsequently used to assess crop losses from 2007 to 2019, revealing that extreme heat and drought have reduced yields by 3.5% in comparison to the 1974-2004 baseline. “While a 3% reduction may seem minor, it has a significant impact,” Kornhuber noted, warning that “the global food market is at risk of triggering severe regional crises.”

The team then calculated economic losses based on FAO data outlining farmers’ production costs at that time. They utilized the same approach to project future losses across several different emissions scenarios, accounting for some expected adaptation efforts.

In a high-emissions scenario (SSP3-7.0), global agricultural yields are projected to decline by nearly 35% by 2100, resulting in annual losses exceeding $161 billion. “Production losses driven by heat and drought are estimated at approximately 855 million tonnes per year,” reports Fung, who shared his findings at the European Geosciences Union conference in Vienna in May. “This equates to the annual consumption of around 2 billion individuals.”

However, this analysis may not fully capture the extent of climate change’s impact, particularly as it only considers three crops and excludes damage from flooding, storms, and rainfall. Crop shortages could further result in substantial price hikes, as has been observed with coffee and cocoa prices.

Jonas Jagermeyer of Columbia University warns that the study’s reliance on statistical relationships may inaccurately overestimate impacts by 2100. “Statistical yield models are effective for current and near-past events but lose reliability when applied to significantly different future conditions,” he explains. He emphasizes that computer models predicting the effects of rising CO2 and temperatures on plants offer valuable insights into future agricultural scenarios.

Karine Chenu, a professor at the University of Queensland in Australia, echoes this sentiment. “While the model has limitations, it provides a solid basis for extrapolating future trends.” However, her research team recently published a study revealing that common wheat models exhibit significant inaccuracies, particularly regarding the simultaneous impacts of heatwaves and drought.

Despite the criticisms, Kornhuber stands by the team’s statistical methodology. “While the model serves as a valuable tool, some validation studies suggest reduced sensitivity to extreme conditions,” he asserts. “In our research, the focus was primarily on extreme phenomena, leading us to directly establish these correlations through statistical means.”