The OWID Brief: Global food trade, heat deaths, paying for CO2 emissions, and more
1. Key Themes
Carbon Pricing Is Largely Symbolic at Current Levels
The gap between existing carbon prices and the threshold needed to drive behavioral change is vast. Most carbon markets are operating far below the level required to matter: "At $5 per tonne of CO₂ — typical of many markets — the cost of a liter of petrol rises by less than 1%. That's nothing compared to the regular ups and downs of oil prices." Only a handful of countries exceed the $100/tonne threshold where prices begin to register (~15% increase on petrol), and those nations "account for less than 0.5% of global emissions." Meanwhile, "71% of emissions have no price at all."
Temperature-Related Mortality Is Misunderstood and Underestimated
The public mental model of heat and cold deaths is wrong in a way that likely causes the risk to be underappreciated by policymakers and investors in climate adaptation. "Almost no one has 'heat' or 'cold' written on their death certificate, but researchers estimate that sub-optimal temperatures lead to several million premature deaths every year" — deaths attributed instead to cardiovascular disease, kidney disease, respiratory infections, and diabetes. The mechanism is hidden, making the mortality burden easy to undercount.
Global Food Systems Are Deeply Interdependent
Supply chain concentration in food is a systemic risk for investors and operators in agriculture, logistics, and food retail. The article describes "bananas from South America, cocoa beans from West Africa, or citrus fruits from Southern Europe" as routine, and frames the new interactive tool as a way to understand which countries are structurally dependent on which trading partners — a direct lens on supply chain fragility.
AI Dominance Is Geographically Concentrated
The article flags a stark data insight with major geopolitical and investment implications: "US and Chinese companies train almost all of the world's most-used AI models." This concentration suggests both regulatory risk and continued compounding advantages for incumbents in those two nations.
The Global Poverty Bar Is Set Too Low to Drive Meaningful Policy
The World Bank's extreme poverty line creates a false binary that obscures hundreds of millions of near-poor people. Economist Lant Pritchett argues the line should sit between $20–$40/day: "The point is not to conflate different degrees of deprivation, but to ensure that the near-poor and the moderately poor are not simply written out of the global development agenda." OWID founder Max Roser independently converges on $30/day as the appropriate threshold.
2. Contrarian Perspectives
Carbon pricing is largely theater — not climate policy. The consensus narrative is that carbon markets represent meaningful climate progress. The data says otherwise. "Only a few countries charge more than $100 [per tonne], and they account for less than 0.5% of global emissions. Most of Europe sits in the $65–90 range, and most carbon markets charge less than $10." If $100/tonne is "roughly the median estimate of the damage caused by a tonne of CO₂," then virtually the entire global carbon pricing apparatus is operating below the threshold of real-world impact. For investors pricing in carbon policy as a transformative force, this is a material correction to that assumption.
Heat and cold kill through chronic disease, not acute events — meaning the death toll is far larger than reported figures suggest. The intuitive model (heatstroke, hypothermia) dramatically underestimates the true burden. "Almost no one has 'heat' or 'cold' written on their death certificate, but researchers estimate that sub-optimal temperatures lead to several million premature deaths every year." This has implications for how climate risk is priced in healthcare, insurance, and infrastructure — sectors currently not accounting for this hidden mortality channel.
AI fact-checking bots are already operating at massive scale but perform only marginally better than chance. Against the optimistic narrative of AI as a misinformation solution, the evidence is sobering. Across nearly 19 million messages sent to Grok and Perplexity on X, "the bots agreed with professional fact-checkers just over half the time, short of the 64% agreement among the fact-checkers themselves." The scale is real; the reliability is not yet.
3. Companies Identified
| Company | Description | Why Mentioned | Quote |
|---|---|---|---|
| Grok (xAI / X) | AI chatbot integrated into X (formerly Twitter) | Used as a primary case study in large-scale AI fact-checking behavior; central to the research on social media misinformation | "People are increasingly turning to AI chatbots such as Grok, which can be tagged in posts on X to reply there." |
| Perplexity | AI-powered answer engine | Co-analyzed alongside Grok in the fact-checking study as one of the two bots receiving millions of fact-check queries | "They analyzed almost 19 million messages sent to Grok and another bot, Perplexity, on X between February and September 2025." |
4. People Identified
| Person | Description | Why Mentioned | Quote |
|---|---|---|---|
| Hannah Ritchie | Researcher/author at Our World in Data | Authored two articles: one on carbon pricing (with Pablo Rosado) and one on temperature-related mortality and climate change projections | "Hannah Ritchie wrote an article that examines how many people die from heat and cold each year." |
| Pablo Rosado | Researcher at Our World in Data | Co-authored the carbon pricing article with Hannah Ritchie | "Learn more in the new article by Hannah Ritchie and Pablo Rosado." |
| Sophia Mersmann | Data visualization developer at Our World in Data | Built the interactive global food trade visualization tool | "Our colleague Sophia Mersmann built an interactive visualization that shows the flow of different types of food from producer to consumer." |
| Lant Pritchett | Economist | Argues for a higher global poverty line; co-authored a paper proposing a $20–$40/day threshold | "As Pritchett puts it, 'The point is not to conflate different degrees of deprivation, but to ensure that the near-poor and the moderately poor are not simply written out of the global development agenda.'" |
| Martina Viarengo | Economist, co-author with Pritchett | Co-authored the paper proposing an elevated global poverty line | "In a recent paper with Martina Viarengo, Pritchett proposes adding a higher global poverty line." |
| Max Roser | Founder, Our World in Data | Independently arrived at a $30/day poverty line, aligning with Pritchett's range | "Our founder, Max Roser, has also emphasized the need for a higher poverty line, arriving at $30 a day." |
| Thomas Renault, Mohsen Mosleh, David Rand | Researchers, co-authors of AI fact-checking working paper | Conducted the first large-scale analysis of AI chatbot fact-checking on social media | "A working paper by Thomas Renault, Mohsen Mosleh, and David Rand offers the first large-scale look at this practice." |
5. Operating Insights
For operators in carbon-intensive industries: don't over-index on carbon pricing as a near-term cost threat. The data shows current carbon prices are functionally immaterial in most markets. "At $5 per tonne of CO₂ — typical of many markets — the cost of a liter of petrol rises by less than 1%." Compliance costs at current price levels are noise relative to commodity price volatility. The real strategic question is when/if prices reach the $100+ threshold — at which point the competitive landscape changes sharply.
For operators building AI-powered trust or verification products: calibrate expectations and messaging carefully. The fact-checking accuracy data (just over 50% agreement with professional fact-checkers) suggests current AI models are not ready to be positioned as authoritative. The authors "call for clearer standards and safeguards" — creating an opening for operators who can build verifiable accuracy benchmarks into their products as a differentiator, especially as "models improve."
For food and agriculture operators: map your supply chain dependencies using emerging data tools. The OWID food trade visualization is a signal that granular, public-facing supply chain transparency tools are becoming available. Operators who proactively map and communicate their geographic sourcing dependencies will be better positioned for both risk management and regulatory scrutiny.
6. Overlooked Insights
Tobacco use in India has halved this century — briefly noted as a data insight bullet point, but significant. India is one of the world's largest tobacco markets, and a 50%+ reduction in prevalence within 25 years is a major public health and consumer behavior shift. For investors in consumer staples, healthcare, or emerging market consumer trends, this signals that large-scale behavioral change in developing markets is achievable and potentially underway in other categories (e.g., obesity, alcohol).
Obesity rates in Pakistan have tripled in the last 20 years — a one-line data point with substantial investment implications. Tripling obesity prevalence in a country of over 240 million people signals an emerging chronic disease burden and a structural shift in healthcare demand, pharmaceutical needs, and food consumption patterns in a market that is often overlooked in favor of India and China.