Why it’s so hard to tell which climate policies actually work

The successful policies shared some similarities. Taxation was generally effective; so was mixing different interventions. In Britain, for example, the researchers reckoned that a range of policies introduced in the 2010s—including a minimum carbon price for power producers, the phase-out of coal plants and stricter rules about air pollution—achieved a 40% cut in emissions from the electricity sector.

Combined, researchers reckoned the 63 success stories reduced emissions by up to 1.8bn tonnes of carbon dioxide, more than the combined net total of Britain, France and Germany in 2023. That is commendable. But it is barely a sixth of what is needed to stop global temperatures from rising beyond 2°C above pre-industrial levels by the end of the century.

There is little evidence to suggest that the remaining 1,400 or so worked. That does not mean they were total failures. For one thing, the study in Science looked only at near-term effects; for another, the lack of available data meant significant sectors (such as agriculture) as well as vast regions (like most of Africa) were excluded. But their exact impact is unknown.

That ignorance is at odds with the speed and scale of the action required. It is partly the result of the field’s traditional focus on modelling science, rather than policy, explains Jan Minx, who leads the Applied Sustainability Science working group at the Mercator Research Institute in Berlin. Predictions about climate are routinely collated and evaluated in the vast “assessment reports" published by the Intergovernmental Panel on Climate Change (IPCC). Several influential international institutions, such as the OECD and the World Bank, review countries’ environmental efforts and make recommendations about how they might improve. But in general they do not analyse the actual impact specific interventions have on emissions. Nor do most government reports.

Things do not have to be this way. Academics in other fields regularly perform systematic policy reviews, in which a wide range of evidence is collected and analysed in a transparent and reproducible way. In clinical and public-health research this “has absolutely been the norm for the last 30 years," says Alan Dangour, the head of the climate and health team at Wellcome, a research-funding institute based in London. (By one reckoning, 80 systematic reviews relating to epidemiology were published per day in 2019 in English alone.)

Both Dr Minx and Dr Dangour are part of a growing effort to establish something similar in climate policy. The first step is persuading the right people. This June in Berlin, after more than two years of drumming up support, Dr Minx and his colleagues hosted What Works, the first international conference for evidence synthesis in climate policy. Among the 300-plus attendees were Jim Skea, the IPCC’s chairman; Jennifer Morgan, Germany’s climate envoy; plus representatives from Wellcome and the Bezos Earth Fund (a $10bn pot set up by Amazon’s founder). More meetings are planned.

The next step involves teaching climate researchers how to synthesise evidence in meaningful, standardised ways. Techniques that work in epidemiology, for example, which often examine limited regions over timescales of days or weeks, need adapting for global climate analyses spanning decades or centuries. The conference in Berlin was followed by two days of this type of training for attendees, and Dr Minx says the aim is to offer similar sessions to ever more researchers.

But it is also necessary to speed up the entire process. Artificial-intelligence (AI) models, which excel at repetitive and lengthy tasks like identifying and digesting relevant papers, can help. The research group behind the Science paper themselves used a combination of machine learning and statistical analysis to link emissions cuts with potential policies. A separate project in 2021 used AI models to discover that there was almost no scientific literature on climate change and maternal and child health, nor on studies focusing on regions like Africa and South America. Wellcome is now funding projects explicitly aimed at plugging those gaps.

AI models can also help keep the existing evidence bank up-to-date. Because scientific understanding of the climate system is still evolving—just how much warming should be expected from each extra unit of carbon dioxide in the atmosphere is still a hotly debated topic, for example—that would help policymakers make the best decisions possible.

There was a concerted effort to create such “living" platforms during the covid-19 pandemic. Dr Minx and Dr Dangour both think a similar version is needed for climate policies; and quickly. “We have 30 years left to get emissions down to net zero," Dr Minx says. “We really need to be efficient, we need to be thrifty and we need to apply rigour—and that starts in science and ends in policy."

© 2024, The Economist Newspaper Limited. All rights reserved. From The Economist, published under licence. The original content can be found on www.economist.com