Study Finds China’s Carbon Market May Be Boosting Output—and Emissions—at Its Most Efficient Coal Plants

China’s first nationwide carbon trading program, launched in 2021, was billed as a major climate policy milestone: it applies to the world’s largest carbon dioxide emitter and, in principle, should push the power sector toward cleaner generation. But new empirical evidence suggests the structure of the market—built around emissions rates rather than a firm cap—may be delivering a counterintuitive early result: higher output and higher emissions from the most efficient coal plants.

Unlike cap-and-trade systems that limit total emissions, China’s national scheme is “rate-based.” Power plants are evaluated against benchmarks tied to the amount of CO₂ emitted per unit of electricity produced. That design choice matters. Economic theory on rate-based regulation predicts that while such systems can encourage lower emissions intensity, they can also effectively subsidize production—because a plant that beats the benchmark can expand output without facing a rising total emissions constraint.

Shuang Zhang and Douglas Almond examine the early performance of this policy using plant-level evidence, drawing on satellite imagery combined with machine-learning methods to measure electricity generation and associated CO₂ emissions. Their focus is especially consequential because the market regulates nearly all coal-fired power plants in China, making coal a central testing ground for how the system performs in practice.

The study’s main result is stark: after the national market began, high-efficiency power plants increased electricity generation and CO₂ emissions by about 5% compared with low-efficiency plants. That divergence is consistent with the notion that more efficient facilities—those better positioned to comply under an intensity benchmark—responded by producing more power.

A key question is whether this shift in output came with a compensating improvement in carbon intensity. The authors test for a differential decline in emissions intensity between higher- and lower-efficiency plants. They do not find evidence of a distinct reduction in intensity for the more efficient group, though they caution that these estimates are imprecise.

To strengthen the interpretation, Zhang and Almond also compare China’s coal plants with facilities in Indonesia and the Philippines—countries without a comparable carbon market during the same period. In that comparison, China’s high-efficiency plants again show emissions increases, while the estimated change for China’s low-efficiency plants is small and statistically insignificant. Taken together, the pattern suggests that extra emissions from ramped-up high-efficiency plants outweighed any substitution away from less efficient generation.

The implications extend beyond carbon accounting. Using downwind air-quality patterns, the authors find that local pollution worsened more in areas affected by emissions from high-efficiency plants than in areas downwind of low-efficiency plants. In effect, the production boost to cleaner-by-design coal units still translated into additional pollution, degrading air quality in parts of China that had been relatively clean.

The findings point to a policy dilemma. If the aim is to cut China’s total CO₂ output, a rate-based market can fall short when the incentives encourage higher electricity production. The authors conclude that China may need to tighten the benchmarks within the current framework—or move toward an alternative approach that does not simultaneously reward expanding generation.

Why it matters

China’s national carbon market is often treated as a bellwether for global climate policy because of the country’s outsized role in emissions and energy demand. If the system’s intensity benchmarks continue to spur production at efficient coal plants, overall emissions reductions could be delayed even as plants “comply” on paper. Policymakers and investors will be watching whether China tightens benchmarks or shifts toward a design that constrains total emissions rather than emissions per unit of output.