[This article](https://www.swiftcentre.org/publicforecasts/global-coal-consumption-will-defy-expectations), predicting much greater coal consumption by China and India than the IEA, Yakov & Partners, and Sinopec, was posted to this subreddit before, but there was little discussion of the actual arguments made by the “superforecasters”, probably because they were hidden away in the graphs showing the predictions. Here they are (each forecaster gave six predictions in total, covering China and India’s coal consumption in 2026, 2030, and 2035) – do you think they are good arguments?
**From forecaster predicting 4,963 Mt of coal consumption in China in 2026**
> The main two questions dominating this forecast for me are general predictions of peak coal and the relationship between coal consumption and economic growth (coupled with forecasts of Chinese growth over the next years). First, China’s ‘Dual Carbon’ goals set out by Xi Jinping in 2020 outline peak carbon emissions by 2030 and carbon neutrality by 2060. Of course, coal is not the only contributor to Chinese carbon emissions, and this goal might be unrealistic given the growth of coal consumption since 2020. SINOPEC on the other hand puts the predicted date of peak coal at around 2025 (https://daxtromn-power.com/blogs/news/sinopec-forecasts-peak-in-china’s-coal-consumption-around-2025). This prediction is somewhat more optimistic than that of some recent academic research using optimized grey model that predicts that “coal consumption in China and India will continue to rise over the next five years” (https://www.sciencedirect.com/science/article/abs/pii/S0360544221030358), though this builds on 2020 data and as such may be consistent with the former. This rough peak date of mid-to-late 2020s is also supported by academic work using system dynamics analysis putting the predicted time at 2027 (https://www.frontiersin.org/articles/10.3389/fenvs.2022.974763/full). Second, there has been quite some discussion in the academic literature on the relationship between economic growth and coal consumption in China. One strand of work looks at intensity effects (investigating how changes in coal consumption may result from changes in coal intensity, i.e., the amount of coal units used per economics output). This paper (https://www.sciencedirect.com/science/article/pii/S0301479721019745?casa_token=JBy-zAuscS4AAAAA:2XeKft1HtziolUxdhzkDz0nr8B0ecjJvdCALy-G2IpXJmlZ1J1zfKv7kuIdGWy3KIIKoE3Rv2cA#bib55) finds a positive effect, suggesting that improvements in efficiency and secondary industries can reduce overall coal intensity. However, economic growth remains the main driver of this, even though the relationship between economic growth and coal consumption has weakened (though proper decoupling has not happened). This result has also been found by other papers (https://www.sciencedirect.com/science/article/pii/S0301421522000374?casa_token=ur4I9csWPxMAAAAA:XvDnZS6_X4l6EF3B_cUEFbj37kZI-yj05pZekJnnRrpOpaQ8oJir3EQyqQXWLHavl3USIpYsMtE). Overall, these results point to a consistent relationship between economic growth and coal consumption, though this relationship is expected to weaken as the Chinese economy advances technologically. Putting these together as a baseline with respect to our forecasting questions, I agree with the rough picture that peak coal before 2026 is very unlikely (10% at best). I also think that the academic results with respect to the weak(ening) relationship between economic growth and coal consumption are likely to hold. However, I do anticipate Chinese growth to slow down substantially over the next years, with there already being quite substantial heterogeneity as to their current growth rate, reported to be over 5%, though unlikely to actually be that. Additionally, while renewable energy is unlikely to replace coal consumption any time soon, their status in the Chinese economy is likely to have substantial effects from around 2030 onwards (both in terms of price pressures in the context of solar but also with respect to the country’s image and the subsequent subsidies afforded to the renewable sector and away from coal as citizen’s concerns for clean air are likely to become more relevant as the rest of the economy advances). For 2026, I anticipate a slowing but consistent growth to about 4950 Mt, with a substantially heavier right tail.
**From forecaster predicting 5,020 Mt of coal consumption in China in 2026**
> IEA’s consistent under forecast factor of 2.8%, as mentioned by my fellow forecaster, would yield a corrected forecast of 4,927 MT for 2026. The slow down in the Chinese economy, particularly in the housing market, should lead to a slow-down in demand. This is likely to be offset by an increase in military production and a post-Covid peak return to normalcy of movement within China. Paradoxically the move towards EVs will diminish gas consumption at the cost of electric energy demand, partially driven by coal. China plans to increase their coal plant capacity by ~1/3. Why would they do this is if they do not expect a significant increase in demand? The current average age of coal plants in China is 13 years. They already have this sunk cost, with nearly 3 decades of life remaining in the existing capacity. Anything short of massive calamitous events that force a political recalculation, are unlikely to see a short term decrease in coal consumption. Green energy capacity is increasing, but not at a rate to cover the additional demand for energy, especially in a warming world. https://globalenergymonitor.org/projects/global-coal-plant-tracker/dashboard/
**From forecaster predicting 5,022 Mt of coal consumption in China in 2026**
> In forecasting it’s actually a good thing if you can find a person or institution that is consistently wrong. Fivethirtyeight had this insight with certain pollsters, finding that if they adjusted for partisan biases, the pollsters became a lot more reliable.The China 2026 coal forecast is therefore fairly straightforward, since I found that the past forecasting errors on the part of IEA offer a remarkable consistency of about 2.8 percentage points per year. My guess is this is basically the optimism bias that one finds in forecasting, often by people who admittedly have good intentions. Using that method a floor would then be 4,927 Mt for 2026, assuming no errors for IEA’s 2023 estimate. Assuming that error does apply for 2023 (which would have been a winning strategy in past years) gets me to 5,022 for a 2026 median, which is higher than the IEA’s 4,535 Mt. Then I have the right tail be longer than the left, given the experience of the past couple years. So overall the compound annual growth rate (CAGR) for these years should come to around 2.6%. This makes intuitive sense, since the long-term CAGR through 2017 was 2.7%, with an acceleration to 4.0% for the 2017-22 period.
> According to globalenergymonitor.org, China has maintained a pace of issuing permits for 2 new coal power plants per week. Some of this is simply an offsetting of decommissioned old plants of course–according to the coal plant tracker dashboard, 2023 is looking like 136 GW in new construction for China but 120 GW of retirements. With China’s energy consumption growing by 6% in 2023 according to IEA though, coal will remain part of the mix for the time being.
> On the steel side of the ledger (this being the other major use of coal, to heat the metal), China’s steel industry has been in the doldrums for a few years now, with only a slight increase in production in 2023. There are no expectations by the steel industry of higher profits or more stimulus coming (source). So because of all that, I don’t expect much increase in coal consumption from that sector.
> Finally, it’s notable that ten years ago the forecasts for the peak were for around 2020, including by the Chinese government (see pages 3 and 45 here) At the moment Sinopec forecasts peak coal consumption for China to come in 2025. I would probably go more for the latter part of this decade given the momentums, but I think the key point is that the clean energy takeover in China has indeed been pushed back 5 to 10 years.
**From forecaster predicting 5,038 Mt of coal consumption in China in 2026**
> As another forecaster points out, China has been on a bit of a coal burning spree recently: the pre-2017 CAGR was 2.7%, but it accelerated to 4.0% for the 2017-22 period. This was partly due to power shortages that some provinces of China experienced, particularly in 2021 and 2022. Meanwhile, poor rainfall in recent years combined with a decline in Chinese dam projects since 2015 meant that hydropower was unable to pick up the slack; as a percentage of electricity production, hydro actually fell from around 19% in 2014 to less than 16% in 2022, and utilisation was close to historic lows. Rainfall picked up in the summer of 2023, however, and this may stay high in 2024, which is one reason we might expect coal consumption not to increase too sharply and why the IEA might even be right that coal consumption will have peaked in 2023.
> How did China increase its coal power generation? According to Global Energy Monitor’s Global Coal Plant Tracker, they revived previously cancelled and shelved projects and approved brand new ones. In 2023, it added almost 50,000 MW and retired less than 4,000 MW.
> But just as renewable electricity capacity is not the same as renewable electricity generation (it makes up over 50% of China’s capacity but only about 26% of its generation), additional coal capacity doesn’t necessarily mean that it will be drawn on, as Hannah Ritchie points out. Indeed, coal consumption and coal-based electricity generation did fall around 2014 despite much more coal power capacity having been added than retired in the years preceding this. If the increase in renewable generation can outpace the increase in power demand for the first time, this could happen again.
> That said, if Ritchie is right about coal plants being uneconomical (China is offering coal producers capacity payments to keep their plants online even when they’re not being used), this is arguably a sign that China is serious about utilising them at the expense of solar and wind (which have their own economic problems, absent good storage capabilities). Since power plant approvals are now decentralised, provincial leaders may judge that it’s better for the economy as a whole to avoid blackouts of a few hours even if coal plants become stranded assets. Most of the countries (Switzerland, France, Sweden, Iceland) that have managed to create a low-carbon or zero-carbon electricity system rely heavily on some mix of n******, hydro and geothermal, and China is only planning on producing 10% of its electricity from n****** by 2035.
> Also, the IEA has been too optimistic in the past as a colleague notes, and I’d give more credence to the analyses cited by my colleagues suggesting that coal consumption will peak later this decade (although some these analyses seem to think coal consumption will ‘peak’ at a figure lower than the 2022 and 2023 figures, which is a bit confusing). Finally, a conflict between China and Taiwan could cause demand to stagnate or fall, as could a lacklustre economy (which will also impact steel production). On the other hand, it could cause China to ramp up coal production.
> Overall, I don’t think coal consumption in 2026 will be as high as some forecasters are predicting (based on an extrapolation of recent trends), because hydropower may regress toward mean utilisation levels after falling close to historical lows during the drought in 2021 and 2022, while additional solar and wind capacity (even with low utilisation rates) may finally start to shoulder a significant part of the extra demand. However, I don’t think it will be as low as the IEA is projecting because solar and wind still have limitations (as discussed by others), n****** and hydropower probably aren’t sufficient for coal to peak in 2023, and the IEA has historically underestimated coal consumption. A peak in the latter half of the decade appears more plausible, especially as this is what China itself is signalling will happen.
*continued in comments*
Source: eldomtom2
2 Comments
**From forecaster predicting 4,759 Mt of coal consumption in China in 2030**
> I think China’s coal consumption should plateau in the 2026-27 time frame. Sinopec forecasts peak coal consumption for China to come in 2025, and generally one can assume an optimism bias with such forecasts. Professional forecasters everywhere from central bankers to pharmaceutical companies seem to have a consistently systematically forecasting error to the optimistic side, particularly in timeframes of 1-3 years, so it’s a fair estimate to add a year or two to Sinopec’s forecast. Also of course, previous forecasts of China’s coal peak (example) have erred on the optimistic side.
> Looking at coal’s competitors, gas & oil power installed capacity, mainly natural gas, is growing much faster than coal on a percentage basis (10% vs. 3%), though coal has a much higher installed capacity so it still wins on an absolute amount. The amount of hydropower added in recent years has been almost as much as coal (24 GW in 2022 vs. 27 GW for coal), so it’s another one to watch. And then renewables have grown by leaps and bounds, with solar’s installed capacity growing by 87 GW in 2022 and 217 GW in 2023 and wind by 76 GW in 2023 and 38 GW in 2022. (Source)
> When it comes to new energy projects, electric utilities look at the lifetime cost of the project, discounted back to present day, which they call the levelized cost of energy. This consists of: 1) Capital expenditures. Cost of a new coal plant vs. solar panels. 2) operations and maintenance, which are slightly more expensive for solar due to the electronics and the need for cleaning. 3) Fuel costs, which are $36 per megawatt-hour for coal as of the last numbers I saw, compared with about a third of that for n****** or literally $0 for solar or wind. 4) Transmission costs.
> These 4 points are then multiplied by a capacity factor, which is a measurement of the percentage of time a power plant is actually producing energy. With coal it’s below 50% compared with n******, which is 92.5% (source) However, the growth in n****** has been very slow in China, just as it has been in most countries, and I don’t assign much probability to n****** having accelerated growth, due to two reasons: 1) high initial capital costs, which in historical experience has always had cost overruns and delays. (Discussion of US here) 2) A long history of strong NIMBYism, even in China.
> So the main competitor that will be eating away at coal by 2030 will be especially solar and batteries. By then, solar is on pace to be 2/3rd cheaper compared with its 2022 price and batteries 50% cheaper compared with 2022. By 2030, coal consumption in China should be declining at around 1% a year or so.
**From forecaster predicting 4,853 Mt of coal consumption in China in 2030**
> For my 2030 forecast, much of my previous reasoning is still the same, with peak coal having been reached and a likely decoupling of economic growth and coal consumption having been achieved in some more significant ways. However, I do not expect a stark and immediate decline of coal consumption but rather a stagnation and slow decline over the years to 2030, leaving my forecast at very similar levels as the 2026 ones, though this time the two tails are more symmetrical, with slightly more mass on the left side, leaving my final median forecast at 4850 Mt.
**From forecaster predicting 5,000 Mt of coal consumption in China in 2030**
> As the Chinese government is signalling, I think coal consumption will have peaked by 2030 but that it’s unlikely it’ll be far below its peak because solar and wind will continue to pose the problems that others have highlighted and provincial leaders may judge that it’s just better to maintain it as the bulk baseload of the electricity system, especially as n****** is unlikely to play a significant role in 2030 and hydropower will be limited by the availability of good locations among other things. On the other hand, coal plants have been mandated to have ‘flexibility retrofits’, indicating that China does plan to move coal in the direction of meeting ‘peak demand’ which is what gas does in many other countries
**From forecaster predicting 5,098 Mt of coal consumption in China in 2030**
> Between 2026 and 2030 it is likely that coal consumption will be peaking. China currently leads the global renwable markets and has about 230GW of wind and solar installations. With continued reduction in cost of solar panels this increasing capacity will likely be able to offset coal, despite increasing energy demand due to the growing economy.
> I have a slight right tail for the possibility of a US-China war over Taiwan, and a left tail to account for global catastrophic risk, which I think is considerable over these time periods.
**From forecaster predicting 5,120 Mt of coal consumption in China in 2030**
> Although China aims to install massive amounts of renewables capacity, this is currently powered by (pandemic era) subsidies that may not be renewed. As land in eastern china is short, most of the solar there is installed inefficiently on rooftops rather than being concentrated. While solar is installed more efficiently in western china, transmission issues mean this is rarely utilized. It would not surprise me that even if these issues were fixed to allow more renewable energy to flow to eastern china, actual coal consumption doesn’t decline a great deal because of the growth in energy demand from more aircon and electric vehicles. I expect coal demand to have leveled off by this point (unlike India) but I am quite skeptical that coal demand will do, mainly because of intermittency costs and the lack of battery storage.
China and India have an energy policy called “anything but oil and gas”. That includes coal. If renewables grow fast enough, coal will be replaced as well, but gas is the first target, and both countries are willing to grow coal to phase out gas faster.