In a groundbreaking study, researchers have harnessed the power of machine learning to evaluate approximately 1,500 climate policies across 41 countries, identifying which ones have been most successful in reducing carbon emissions. The study, published in Science, found that 63 interventions across 35 countries led to significant reductions in emissions, with an average cut of 19%.
This AI-enhanced analysis offers a more comprehensive approach than previous studies, which often focused on a narrow set of prominent policies. The researchers combined machine learning with statistical analysis to uncover which policies and combinations were most effective in four high-emission sectors: buildings, electricity, industry, and transport.
One key finding is that the most effective reductions occurred when multiple policy tools were used in tandem. For example, the UK’s success in phasing out coal-fired power stations was largely due to its combination with pricing mechanisms, such as a minimum carbon price. Similarly, Norway’s ban on combustion engine cars saw greater success when paired with price incentives that made electric vehicles more affordable.
The study, led by Annika Stechemesser from the Potsdam Institute for Climate Impact Research, emphasizes the importance of the right mix of policies rather than the sheer number of policies. In high-income countries, pricing interventions like energy taxes were particularly effective for reducing emissions from electricity generation. However, in the building sector, policies that phased out or banned emissions-generating activities more than doubled the effectiveness when combined, as opposed to when implemented individually.
Interestingly, taxation was the only policy that achieved nearly equal or larger emission reductions as a stand-alone measure across all four sectors analyzed.
The implications of this study are profound, highlighting the need for countries to re-evaluate and optimize their climate strategies. The world’s annual emissions are projected to be 15 gigatonnes (Gt) of CO2 equivalents higher by 2030 than the levels required to keep global warming below 2 °C above pre-industrial levels, according to the United Nations. As such, this AI-driven analysis provides valuable insights that could help nations adjust their policies to meet global climate goals more effectively.
