Chapter 7: Capital-Intensive Pathways: Power Decarbonization, Carbon Markets and Asset Risk

A steel mill in Tangshan signed a ten-year power purchase agreement with a wind farm in Zhangjiakou in early 2024. The contract covers about 40 percent of the mill's electricity needs at a fixed price of 0.38 yuan per kilowatt-hour, roughly 15 percent below the provincial industrial tariff. The mill's carbon manager calculated that the deal would cut Scope 2 emissions by 280,000 tons of CO2 per year and reduce the mill's exposure to carbon price risk on its electricity purchases. The same month, a competing mill thirty kilometers away declined a similar contract because its finance director judged the fixed price too high relative to spot market expectations. Both firms face the same grid, the same carbon price, and the same product market. Their divergence on a single procurement decision will compound over a decade of asset life.

That divergence is the subject of this chapter. Capital-intensive sectors—power generation, steel, cement, aluminum, chemicals—face a green governance problem that differs in kind from the labor-intensive sectors examined in Chapter 5. A furniture factory that miscalculates its carbon exposure loses a contract or absorbs a margin squeeze. A cement plant, an aluminum smelter, or a coal-fired power station that miscalculates its carbon exposure can lose its entire asset value. The question is not how to differentiate a product or win a market segment. It is whether the physical capital a firm has sunk into the ground retains economic value under a carbon-constrained regime.

Three mechanisms define this pathway: the decarbonization of electric power and the electrification of industrial heat; the design and operation of carbon markets as price-discovery institutions; and the assessment and management of stranded asset risk under policy uncertainty. Each has its own data problems and distributional consequences. Together they determine whether China's capital-intensive industries can convert the dual carbon goals from an exogenous cost shock into an endogenous upgrading driver.

A note on evidence before proceeding. The Chinese national ETS has operated since 2021, and its long-run price behavior is unknown. The pilot programs covered smaller volumes and different sectors. Where I draw on EU or California experience, I flag the institutional differences. Quantitative estimates in this chapter should be read as orders of magnitude, not point forecasts. I have consolidated these caveats here rather than repeating them throughout.

7.1 Power Sector Decarbonization and Electrification: Clean Electricity, Grid Integration and Electrification of Industrial Heat

7.1.1 Why Power Is the Linchpin

The Tangshan steel mill's electrification decision turns on a question its managers cannot answer from their utility bill: what is the carbon intensity of the marginal electron the mill actually consumes, at the hour it consumes it, under the contract it can sign? The provincial average for Hebei's grid in 2023 was roughly 580 grams of CO2 per kilowatt-hour, according to the Ministry of Ecology and Environment's regional grid emission factors. The marginal electron during daylight hours, when the mill runs its rolling mill at peak load, may come from a wind farm in Zhangjiakou or from a coal unit in Shanxi, depending on transmission congestion and dispatch order. The difference between those two electrons is the difference between an electrification strategy that reduces emissions and one that merely relocates them.

Electricity is the linchpin of the clean-energy transition because if electricity can be made carbon-free, it can be used to decarbonize almost everything else. Power and heat production have historically accounted for around 40 percent of China's national CO2 emissions, a function of the coal-heavy generation mix. The electrification of transport, heating, and a substantial share of industrial process heat depends on the availability of clean electrons. A capital-intensive firm that electrifies its processes while the grid remains coal-dominated has not solved its carbon problem; it has relocated it upstream.

This interdependence creates a collective action problem more severe than in labor-intensive sectors, where electricity is a smaller share of total costs and the relevant abatement levers are more directly under firm control. The boundary of the firm—what a capital-intensive company owns and what it buys—becomes a strategic variable. A firm that owns its own generation, or that signs a long-term power purchase agreement with a renewable developer, internalizes the carbon intensity of its electricity in a way that a firm buying from the grid does not.

China added more renewable generating capacity in 2023 than the rest of the world combined—roughly 300 gigawatts of solar and wind, according to National Energy Administration figures. Coal still accounts for around 60 percent of generation, but the share is falling, and the absolute growth in coal capacity has slowed sharply. The grid has not kept pace. Curtailment of renewable output remains a problem in several provinces, and the transmission infrastructure needed to move power from the windy northwest and the sunny west to load centers in the east is still under construction. For a capital-intensive firm deciding whether to electrify a furnace or a boiler, the relevant question is not the national average carbon intensity of electricity. It is the marginal carbon intensity of the electricity it can actually procure, at the time of day and in the location where it operates, under the contractual arrangements available to it.

That question has no simple answer, and this is where the green governance capability of the firm becomes visible. A firm that can navigate the emerging landscape of green power trading, distributed generation, and long-term PPAs can procure cleaner electricity than the grid average. A firm that cannot will find itself locked into the average. The difference is organizational and contractual, not technological.

7.1.2 The Chinese Power Mix: A Moving Target

The 14th Five-Year Plan for the energy sector set a target of raising the share of non-fossil energy in primary energy consumption to around 20 percent by 2025 and to around 25 percent by 2030. The 2060 carbon neutrality target implies that the power sector must reach near-zero emissions well before mid-century, which in practice means the coal fleet must either be retired early, retrofitted with carbon capture, or run at very low capacity factors as a balancing resource.

The pace of renewable build-out has repeatedly exceeded official targets. Solar installations in 2023 alone exceeded 200 gigawatts, more than the entire cumulative solar capacity of any other country. Wind additions were around 75 gigawatts. In several provinces, renewable generation now regularly exceeds grid demand during daylight hours, and curtailment rates have risen. The grid integration problem is no longer hypothetical; it is a binding constraint on the value of new renewable capacity and, by extension, on the green electricity procurement strategies of capital-intensive firms.

For the firm, this creates a paradox. The national grid is getting cleaner on average, but the marginal clean electron is increasingly location- and time-specific. A steel mill in Hebei that buys unbundled renewable energy certificates can claim a lower Scope 2 footprint, but the physical electricity it consumes still comes from the mix. A smelter in Yunnan that can contract directly with a hydropower producer has a structurally lower carbon footprint than an identical smelter in Shandong. The geographic differentiation of power carbon intensity is becoming a determinant of industrial competitiveness, and it is reshaping the locational calculus of energy-intensive industries.

The grid integration challenge has a second dimension: the need for flexibility. A power system with a high share of variable renewables requires storage, demand response, or dispatchable backup to maintain reliability. China is building all three, but the economics are unsettled. Pumped hydro storage is expanding, battery storage is being deployed at scale, and coal plants are being retrofitted for flexible operation. For capital-intensive firms, this creates an opportunity as well as a cost. A firm with flexible load—an electrolyzer that can ramp up and down, an electric furnace that can shift its operating hours—can earn revenue by providing demand response services. The governance capability required is the ability to participate in ancillary service markets and to manage production schedules around power price signals. Most Chinese industrial firms do not yet have this capability. The ones that build it will find that electrification generates revenue as well as cost.

Provincial grid carbon intensity varies by a factor of three across China. The accompanying table draws on MEE regional grid emission factors and provincial energy balance data to show the range that capital-intensive firms face when choosing where to locate or expand.

The table reveals the trade-off that location decisions now embed. Inner Mongolia offers the cheapest industrial power in the country and the dirtiest. Yunnan offers clean power at a moderate tariff. A smelter choosing between the two is choosing between a low operating cost with high carbon exposure and a higher operating cost with low carbon exposure. Under the current carbon price, the Inner Mongolia option wins on cash cost. Under a carbon price of 200 yuan per ton, the calculation reverses. The firm's location decision is therefore a bet on the future carbon price, whether or not the firm frames it that way.

7.1.3 Electrification of Industrial Heat: The Hard Part

Industrial heat accounts for roughly a fifth of global final energy consumption, according to the International Energy Agency's 2023 tracking report on industry, and most of it is currently produced by burning fossil fuels. The electrification of industrial heat is technically feasible for low- and medium-temperature processes (below roughly 200 degrees Celsius) using heat pumps, and for some high-temperature processes using electric arc furnaces, induction heating, or resistance heating. It is not yet feasible at scale for cement kilns, which require temperatures above 1,400 degrees Celsius and where the chemistry of limestone calcination releases CO2 that cannot be avoided by switching the heat source. This technical boundary is the single most important determinant of the differentiated pathways within the capital-intensive category.

The Chinese policy framework has begun to address electrification explicitly. The "Implementation Plan for Carbon Peaking in the Industrial Sector," issued by MIIT and other ministries in 2022, calls for the electrification of industrial processes where technically and economically feasible, and for the replacement of coal-fired boilers and kilns with electric alternatives. The plan prioritizes mature technologies—electric vehicles, electric cooking equipment, small electric boilers—and acknowledges that large kilns such as cement rotary kilns have no ready electrification pathway. This is an honest assessment, and it implies that the decarbonization of cement will depend on carbon capture, alternative fuels, or clinker substitution rather than electrification. For steel, the picture is more mixed: electric arc furnace (EAF) steelmaking is well established, but it requires scrap, and China's scrap availability is limited by the relatively young age of its steel stock. The shift from blast furnace-basic oxygen furnace (BF-BOF) to EAF is therefore a multi-decade transition, not a switch that can be flipped.

The electrification of industrial heat is where the interdependence between the power sector and capital-intensive industry is most acute. A firm that electrifies a boiler or a furnace is effectively outsourcing its decarbonization to the grid. If the grid is clean, electrification delivers deep abatement. If the grid is coal-heavy, electrification may deliver little or no abatement relative to a gas-fired alternative, and could even increase emissions if the marginal generation is coal. This is why the carbon intensity of the marginal electron matters more than the average. It is also why the sequencing of power sector decarbonization and industrial electrification matters. Electrifying too early, before the grid is clean, can lock in a new set of assets that will need to be retrofitted or replaced. Electrifying too late, after competitors have established clean supply chains, can forfeit market share. The optimal timing is firm-specific and depends on the firm's discount rate, its access to clean power contracts, and its exposure to carbon pricing.

The empirical magnitude of the marginal emission factor is notoriously difficult to estimate. Most Chinese firms do not calculate it; they use average grid factors published by the government, which are lower than the marginal factor in a system where coal is the marginal source. This measurement gap is itself a governance capability issue. A firm that understands the difference between average and marginal emission factors is better positioned to make electrification decisions that actually reduce emissions, and to defend those decisions to investors and regulators.

7.1.4 Clean Electricity Procurement: Instruments and Their Limits

Chinese capital-intensive firms have three main instruments for procuring clean electricity: green power trading, unbundled renewable energy certificates, and direct power purchase agreements with renewable generators. Each has different implications for the firm's carbon accounting and its actual emissions.

Green power trading allows a firm to buy electricity from a renewable generator through the grid, with the environmental attributes transferred to the buyer. The volume has grown rapidly since the scheme was launched in 2021, but it remains a small share of total electricity consumption. The main limitation is that green power trading is organized by province, and the provinces with the most renewable generation are not always the provinces with the most industrial demand. Cross-provincial green power trading is expanding but remains constrained by transmission capacity and by provincial protectionism. In 2023, cross-provincial green power trades accounted for less than 10 percent of total green power trading volume, according to figures from the Beijing Power Exchange Center.

Unbundled certificates are simpler but weaker. A firm can buy a certificate representing one megawatt-hour of renewable generation without actually consuming that electricity. The certificate market in China has been plagued by low prices and by doubts about additionality—whether the certificate purchase actually causes new renewable capacity to be built. The government has tightened the rules, but the credibility of the instrument remains contested. For a capital-intensive firm that wants to claim a lower Scope 2 footprint in its CDP disclosure, certificates are a low-cost option. For a firm that wants to genuinely reduce its emissions, they are less useful.

Direct PPAs are the most credible instrument but the hardest to execute. A firm can sign a long-term contract with a renewable generator, either physically (through a dedicated line) or financially (through a contract for differences). The physical PPA requires proximity and grid access; the financial PPA requires a sophisticated treasury function and a counterparty willing to bear basis risk. Both require a level of contractual capability that most Chinese industrial firms do not yet possess. The firms that have signed PPAs tend to be large, often foreign-invested or export-oriented, with dedicated energy procurement teams. This is a governance capability gap that maps onto the broader pattern described in earlier chapters: the firms with the most exposure to international markets have the strongest incentive to build the capabilities, and the firms that serve domestic markets lag behind.

The implication for the capital-intensive pathway is that clean electricity procurement is not a level playing field. Firms differ in their access to clean power, in their ability to contract for it, and in their capacity to account for it credibly. These differences are not captured by the national average carbon intensity of the grid, and they are not fully captured by the carbon price. They are firm-specific assets—or liabilities—that will increasingly determine competitive position.

7.1.5 The Coal Fleet and the Retirement Question

China has roughly 1,100 gigawatts of coal-fired capacity, the largest in the world by a wide margin. The average age of the fleet is around 15 years, well below the 40-year design life. If the fleet is retired at the end of its design life, it will still be operating well past 2060, which is incompatible with carbon neutrality. Early retirement is therefore necessary, but it is expensive and politically difficult. The stranded asset problem in the power sector is primarily a coal problem.

The age distribution of the coal fleet matters for the retirement question. Roughly 40 percent of capacity was built after 2015, meaning it has decades of design life remaining. Retiring these units early requires compensating their owners for the foregone cash flows. The economics of early retirement depend on the compensation mechanism. A coal plant that is retired early loses the net present value of its remaining cash flows. If the plant is state-owned, the loss falls on the state; if it is privately owned, the loss falls on the owners and their creditors. The Chinese government has experimented with capacity payments and with "retirement and replacement" schemes, in which a coal plant is retired in exchange for the right to build renewable capacity. These schemes are a form of compensation, but they are not a market-based solution. The scale of the problem is large enough that a systematic approach will be needed, and the design of that approach is one of the most consequential policy questions in the Chinese energy transition.

For the capital-intensive firm, the coal retirement question matters in two ways. First, it affects the carbon intensity of the electricity the firm buys. A faster retirement of coal capacity means a cleaner grid, which improves the abatement value of electrification. Second, it affects the firm's own asset portfolio if the firm owns captive coal generation. Many Chinese industrial firms do—particularly in the steel, cement, and chemicals sectors, where captive coal plants were built to secure reliable power. These captive plants are a hidden stranded asset liability. They are often old, inefficient, and highly emitting, and they will be among the first to face carbon pricing and regulatory pressure. A firm that has not assessed the retirement cost of its captive coal fleet is carrying an unquantified risk on its balance sheet.

7.2 Carbon Markets, Pricing and Asset Valuation: ETS Design, Carbon Pricing and Implications for Capital-Intensive Assets

7.2.1 The Chinese National ETS: Design and Evolution

China launched its national emissions trading scheme in 2021, after a decade of pilot programs in seven provinces and cities. The scheme initially covered around 2,200 firms in the power sector, representing roughly 4.5 billion tons of CO2 emissions—the largest carbon market in the world by volume. It has since been expanded to include the cement and aluminum sectors, and further expansion to steel, chemicals, and other energy-intensive industries is planned. The design of the scheme is distinctive in several respects, and those distinctive features have important implications for capital-intensive asset valuation.

The Chinese ETS is intensity-based rather than cap-based. The cap is not set in absolute tons of CO2; it is set implicitly through a benchmark intensity standard that firms must meet. A firm's allocation is calculated by multiplying its output by a benchmark emission intensity, with adjustments for fuel type and technology. Firms that emit less than the benchmark receive a surplus of allowances; firms that emit more must buy allowances. The total number of allowances is not fixed in advance; it depends on total output. This design was chosen to accommodate continued industrial growth without imposing a hard cap that might constrain output, but it has a significant consequence: the scheme does not guarantee a reduction in absolute emissions. If output grows faster than the benchmark tightens, emissions can rise even as the intensity falls.

The scheme is free-allocation-based, with auctioning limited to a small share of allowances. Firms receive most of their allowances for free, which limits the cost pass-through and the price signal. The carbon price in the Chinese ETS has been volatile, ranging from around 40 yuan to around 100 yuan per ton since launch, with much of the volatility driven by compliance-cycle effects and by changes in the benchmark rather than by shifts in marginal abatement cost. The price is well below the levels seen in the EU ETS, where allowances have traded above 80 euros per ton for extended periods. The gap between Chinese and European carbon prices is one of the most consequential facts for the competitiveness of Chinese capital-intensive exports.

The scheme's MRV (monitoring, reporting, and verification) system is still maturing. Firms self-report their emissions, and the verification is conducted by third-party agencies whose capacity and independence vary. There have been cases of data manipulation and of inconsistent accounting methodologies. The government has tightened the rules, but the credibility of the data remains a concern for investors and for international counterparties. This matters for asset valuation because the value of a carbon asset or liability depends on the reliability of the underlying measurement. A firm that cannot credibly measure its emissions cannot credibly value its allowances.

The EU ETS, which has been operating since 2005, is the closest analogue, but its design and its political context are different. Any inference from the EU experience to China must be qualified. What the EU experience does suggest is that carbon markets tend to tighten over time, that prices can rise faster than expected, and that firms that delay abatement can face sharp cost increases. The EU allowance price rose from around 5 euros in 2017 to over 80 euros in 2023, a sixteen-fold increase in six years. Whether the Chinese price will follow a similar trajectory is uncertain, but the direction of policy is clear.

7.2.2 How Carbon Pricing Enters Asset Valuation

The value of a capital-intensive asset is the present value of its expected future cash flows. Carbon pricing affects that value through several channels, and the relative importance of each channel depends on the asset type.

The direct compliance cost is the most visible channel. A firm that must surrender allowances for each ton of CO2 it emits faces a cost that is proportional to its emissions and to the allowance price. If the firm receives free allowances, the cost is limited to the shortfall between its allocation and its emissions. If the firm must buy all its allowances, the cost is proportional to total emissions. Most firms focus on this channel. It is not always the largest.

Carbon pricing also raises the cost of fossil-fired generation, which raises wholesale electricity prices, which raises the cost of electricity for industrial consumers. The magnitude of this pass-through depends on the structure of the electricity market. In China, where electricity prices are regulated and cross-subsidized, the pass-through is muted but not zero. In the EU, where electricity markets are liberalized, the pass-through is substantial. The indirect cost can be larger than the direct cost for firms that are electricity-intensive but not direct emitters—aluminum smelters, for example, which consume large amounts of electricity but emit relatively little on site.

A third channel operates through the cost of carbon embedded in inputs. A steel mill that buys iron ore and coal faces carbon costs embedded in the price of those inputs if the suppliers are covered by carbon pricing. A cement plant that buys clinker faces carbon costs embedded in the clinker price. These upstream costs are often invisible to the firm but they affect its cost structure and its competitiveness.

Carbon pricing also affects demand. If it raises the price of carbon-intensive products relative to low-carbon substitutes, demand may shift. The magnitude of this effect depends on the elasticity of demand and on the availability of substitutes. For cement, there are few close substitutes, and demand is inelastic in the short run. For steel and aluminum, substitutes exist but are more expensive, and the demand shift depends on the willingness of consumers to pay a green premium.

The final channel is the effect on asset life. A carbon price that rises over time can make a high-emitting asset uneconomic before the end of its physical life. This is the stranded asset channel, and it is the one most relevant to capital-intensive firms with long-lived assets. A coal-fired power plant with a 40-year design life that faces a carbon price rising to 200 yuan per ton by 2035 may become uneconomic in its 20th year. The remaining 20 years of expected cash flows are written off, and the asset is stranded.

The relative importance of these channels varies by sector. For a coal-fired power plant, the direct compliance cost and the stranded asset channel dominate. For an aluminum smelter, the indirect electricity cost dominates. For a cement plant, the direct compliance cost and the demand effect dominate. A firm that focuses only on the direct compliance cost will underestimate its exposure. The ability to map the full set of carbon cost channels onto the asset base is a skill that most Chinese capital-intensive firms are still developing.

7.2.3 The Benchmark System and Its Distributional Consequences

The Chinese ETS benchmark system determines which firms are net buyers and which are net sellers of allowances, and therefore how the cost is distributed within each sector.

The benchmark is set by technology type and by fuel. In the power sector, the benchmark for coal-fired units is differentiated by capacity and by whether the unit is cogeneration. In the cement sector, the benchmark is based on clinker output and is differentiated by the type of kiln. In the aluminum sector, the benchmark is based on aluminum output and is differentiated by the power source (hydropower versus thermal). The benchmarks are designed to reward efficient producers and to penalize inefficient ones, but they also create winners and losers in ways that are not always transparent.

Consider the power sector. A large, modern ultra-supercritical coal plant with a high capacity factor will typically emit less than the benchmark and will receive a surplus of allowances. An older, smaller subcritical plant will typically emit more than the benchmark and will have to buy allowances. The benchmark system therefore functions as a mechanism for cross-subsidizing the retirement of inefficient capacity, but it does so indirectly and without an explicit retirement mandate. The surplus allowances received by efficient plants have a market value, and that value is a transfer from the inefficient plants that must buy allowances. The transfer is not large at current prices—a few tens of millions of yuan for a typical plant—but it will grow as the benchmark tightens and as the price rises.

The distributional consequences within sectors are important for the asset valuation question. A firm that is a net seller of allowances has a carbon asset that offsets some of its other risks. A firm that is a net buyer has a carbon liability that compounds its other risks. The two firms may be in the same sector, with similar products and similar customers, but their carbon exposure is very different. This heterogeneity is often invisible in sector-level analyses, which treat the sector as a representative agent. For investors and for policymakers, the heterogeneity matters: a policy that is designed to be neutral at the sector level may have large distributional effects within the sector.

The benchmark system also affects the incentive to invest in abatement. A firm that is already below the benchmark has no incentive to abate further unless the benchmark tightens or the price rises above its marginal abatement cost. A firm that is above the benchmark has an incentive to abate up to the point where its marginal abatement cost equals the allowance price. The aggregate abatement depends on the distribution of marginal abatement costs and on the stringency of the benchmark. If the benchmark is lax, the price will be low and the abatement will be limited. If the benchmark is stringent, the price will be higher and the abatement will be greater. The Chinese government has committed to tightening the benchmark over time, but the pace is uncertain and the mechanism is not fully specified.

The following table summarizes benchmark values for the three sectors currently covered by the national ETS, drawn from MEE allocation plans for the 2021–2023 compliance period.

The benchmark values determine the shortfall or surplus for each firm. A cement plant emitting 0.85 tons of CO2 per ton of clinker against a benchmark of 0.82 faces a shortfall of 0.03 tons per ton of output. At 2 million tons of annual clinker output, the shortfall is 60,000 tons. At 80 yuan per ton, the annual compliance cost is 4.8 million yuan. At 200 yuan per ton, it is 12 million yuan. The benchmark value, not the carbon price alone, determines whether a given plant is a buyer or a seller.

7.2.4 The EU ETS as a Comparative Benchmark

The EU ETS has gone through four phases, each with a different design. Phase I (2005–2007) was a learning phase with generous free allocation and a price collapse. Phase II (2008–2012) tightened the cap and introduced auctioning, but the 2008 financial crisis and the subsequent recession created a surplus of allowances that depressed the price. Phase III (2013–2020) introduced a single EU-wide cap and a linear reduction factor, and the Market Stability Reserve was created to absorb surplus allowances. Phase IV (2021–2030) further tightened the cap and extended the scheme to new sectors.

The price trajectory has been dramatic. Allowances traded at around 5 euros per ton in 2017, rose to around 25 euros in 2018, and then climbed steadily to over 80 euros in 2023, with peaks above 100 euros. The increase was driven by the tightening of the cap, the reform of the Market Stability Reserve, and the anticipation of the Carbon Border Adjustment Mechanism (CBAM). The high price has made the EU ETS a meaningful cost for covered firms and has driven significant abatement, particularly in the power sector, where coal-to-gas switching has been substantial.

Three lessons transfer to China. First, carbon markets can deliver large price increases when the cap is tightened credibly. The Chinese scheme has not yet reached the point where the cap is binding in absolute terms, but the direction of travel is similar. Second, the distributional effects of a high carbon price are significant and require compensatory mechanisms. The EU has used free allocation and, more recently, the Innovation Fund and the Modernisation Fund to compensate vulnerable firms and regions. China will need similar mechanisms as the price rises. Third, the interaction between the carbon market and other policies matters. The EU ETS operates alongside renewable energy targets, energy efficiency standards, and industrial policy. The carbon price is not the only instrument, and its effect is mediated by the other instruments. The same will be true in China.

The EU experience also offers a warning. The price increase was not smooth. It was punctuated by periods of volatility, by policy interventions, and by unexpected shocks. Firms that had not built the capability to manage carbon price risk found themselves exposed to large swings in compliance costs. Carbon price risk is a financial risk that requires active management, not a compliance formality that can be delegated to the environmental department.

A comparison of the two schemes' key design parameters clarifies the gap.

The price gap between the two schemes—roughly a factor of seven at 2023 averages—is the single most important number for Chinese exporters facing CBAM. A Chinese steel exporter selling into the EU will pay the difference between the EU allowance price and the carbon price already embedded in its Chinese production. At current prices, that difference is large enough to erase the cost advantage that Chinese steel has historically enjoyed in European markets.

7.2.5 Carbon Pricing and Asset Valuation: A Worked Example

To make the asset valuation channel concrete, consider a cement plant with a capacity of 2 million tons of clinker per year. The plant emits approximately 0.85 tons of CO2 per ton of clinker, for a total of 1.7 million tons of CO2 per year. Under the Chinese ETS, the plant receives free allowances based on a benchmark of 0.82 tons of CO2 per ton of clinker, for a total of 1.64 million allowances. The shortfall is 60,000 tons per year. At a carbon price of 80 yuan per ton, the annual compliance cost is 4.8 million yuan. At a price of 200 yuan per ton, the cost is 12 million yuan. The plant's annual revenue, at a clinker price of 400 yuan per ton, is 800 million yuan. The compliance cost is therefore 0.6 percent of revenue at 80 yuan and 1.5 percent at 200 yuan. These are not trivial numbers, but they are not large enough on their own to strand the asset.

The picture changes when the indirect channels are included. If the carbon price raises the cost of electricity by 0.05 yuan per kilowatt-hour, and the plant consumes 100 kilowatt-hours per ton of clinker, the indirect cost is 5 yuan per ton, or 10 million yuan per year. If the carbon price raises the cost of raw materials and transport by a similar amount, the total cost increase could be 20–30 million yuan per year, or 2.5–4 percent of revenue. At that level, the plant's margin is compressed but not eliminated. The asset remains viable, but its profitability is reduced.

The stranded asset risk emerges when the carbon price rises high enough, or the benchmark tightens fast enough, that the plant's cash flows turn negative before the end of its life. A cement plant with a 30-year life that faces a carbon price rising to 400 yuan per ton by 2040, combined with a benchmark that tightens to 0.6 tons per ton of clinker, would face a compliance cost of 170 million yuan per year—more than 20 percent of revenue. At that point, the plant would need to invest in carbon capture, clinker substitution, or alternative fuels to remain viable. If it cannot, it will be retired early, and the remaining book value will be written off.

The numbers in this example are drawn from the benchmark values in the MEE allocation plan and from average clinker prices reported by the China Cement Association for 2023. They are illustrative of the mechanism, not predictive of any specific plant. The general point is that the stranded asset risk in capital-intensive sectors is not a binary outcome that occurs at a single price threshold. It is a gradual erosion of profitability that accelerates as the carbon price rises and the benchmark tightens. The firm's ability to anticipate and manage this erosion is a core dimension of green governance capability.

7.3 Stranded Assets and Investment Under Uncertainty: Risk Assessment, Policy Uncertainty and Capital Reallocation Strategies

7.3.1 Defining Stranded Assets

The term "stranded assets" was popularized by the Carbon Tracker Initiative in a 2011 report on fossil fuel reserves. The core idea is that a significant share of the fossil fuel reserves booked by listed companies cannot be burned if the world is to meet its climate targets, and that the value of those reserves will therefore be written down. The concept has since been extended to any asset whose value is impaired by the transition to a low-carbon economy—coal plants, oil refineries, steel mills, cement kilns, and even some renewable assets that are rendered obsolete by newer technology.

For the purposes of this book, I define a stranded asset as a capital asset whose expected future cash flows are insufficient to recover its book value or its remaining debt, as a result of climate policy, technology change, or market shifts associated with the low-carbon transition. This definition is deliberately broad. It includes assets that are physically retired early, assets that are sold at a loss, and assets that continue to operate but at a return below the cost of capital. It also includes assets that are impaired not by carbon pricing directly but by the second-order effects of the transition—a coal port that loses traffic as coal shipments decline, a steel town that loses population as the local mill closes.

The stranded asset concept is analytically useful because it forces attention onto the asset level rather than the firm level or the sector level. A firm may have a portfolio of assets, some of which are stranded and some of which are not. The firm's overall viability depends on the balance. A sector may have a large aggregate stranded asset problem even if most individual firms are healthy, because the stranded assets are concentrated in a subset of firms or a subset of regions. The asset-level perspective is therefore essential for understanding the distributional consequences of the transition and for designing policies that manage them.

7.3.2 The Chinese Stranded Asset Landscape

China's stranded asset exposure is large and concentrated in a few sectors. The most obvious is coal-fired power generation. As noted earlier, China has around 1,100 gigawatts of coal capacity, much of it young. If the fleet is retired at the end of its design life, it will still be operating in 2060, which is incompatible with carbon neutrality. Early retirement is necessary, and the cost will be borne by the plant owners, their creditors, and ultimately the state.

The magnitude of the cost is uncertain because it depends on the retirement schedule and on the compensation mechanism. A rough estimate can be constructed from the book value of the coal fleet. If the average remaining book value is around 3,000 yuan per kilowatt and the fleet is 1,100 gigawatts, the total book value is around 3.3 trillion yuan. If half of that is stranded by 2040, the loss is around 1.65 trillion yuan, or roughly 1.5 percent of China's annual GDP. This is a large number, but it is spread over two decades and across a large number of owners. It is manageable if the losses are recognized early and provisioned for; it is destabilizing if they are recognized abruptly.

Heavy industry—steel, cement, aluminum, and chemicals—carries the second major exposure. These sectors have large capital stocks, long asset lives, and high carbon intensity. The stranded asset risk is not uniform across the sectors. Steel has a partial transition pathway through EAF, but the transition requires scrap availability and electricity that is not yet fully clean. Cement has no scalable decarbonization pathway other than carbon capture, which is expensive and not yet deployed at scale in China. Aluminum has a pathway through hydropower and recycled aluminum, but the geographic distribution of hydropower is limited. Chemicals have a mixed picture, with some products (ammonia, methanol) facing severe transition challenges and others (specialty chemicals) facing less.

The third exposure is in the supporting infrastructure—coal ports, coal railways, coal mining equipment, and the towns that depend on them. These assets are not covered by carbon pricing directlybut they are affected by the decline in coal throughput. The transition management challenge for these regions is significant, and it is not well addressed by the current policy framework. The "just transition" agenda, which has received considerable attention in Europe, is still nascent in China.

The provincial distribution of coal capacity and coal-dependent employment determines where the stranded asset problem will concentrate. Shanxi, Inner Mongolia, Shaanxi, and Henan together account for more than half of national coal mining employment and a large share of coal-fired generation capacity. These provinces also have some of the lowest per capita incomes and the weakest fiscal capacity to absorb transition costs. The stranded asset problem is therefore also a regional inequality problem, and it will shape the politics of the transition as much as the economics.

7.3.3 Policy Uncertainty and the Investment Decision

The stranded asset risk is compounded by policy uncertainty. A firm deciding whether to invest in a new coal plant, a new cement kiln, or a new steel mill must form expectations about future carbon prices, future benchmark stringency, future technology costs, and future demand. Each of these expectations is uncertain, and the uncertainty is not resolvable by more data—it is genuine Knightian uncertainty about the path of policy and technology.

The standard response to uncertainty in capital budgeting is to raise the discount rate or to shorten the payback period. Both responses bias investment toward shorter-lived, more flexible assets and away from long-lived, capital-intensive assets. This is rational at the firm level, but it has an aggregate consequence: it accelerates the decline of the very sectors that need to invest in decarbonization. A steel firm that shortens its payback period will not invest in a hydrogen-based direct reduced iron plant, because the payback period is too long. A cement firm that raises its discount rate will not invest in carbon capture, because the returns are too uncertain. The result is underinvestment in the technologies that are needed for the transition.

This is a classic market failure, and it justifies policy intervention. The intervention can take several forms. One is to reduce policy uncertainty by committing to a long-term carbon price trajectory. The EU has done this through the Market Stability Reserve and the linear reduction factor, which together provide a predictable tightening path. China has not yet made an equivalent commitment, and the absence of a long-term signal is a barrier to investment. Another form is to de-risk specific technologies through contracts for differences, price guarantees, or public co-investment. The UK's Contracts for Difference mechanism for offshore wind is an example; China's green power trading and renewable portfolio standards are partial analogues. A third form is to provide adjustment assistance to firms and regions that are adversely affected, which reduces the political resistance to the transition and allows the policy to be more ambitious.

The Chinese policy framework has elements of all three, but they are not yet integrated into a coherent transition strategy. The carbon market provides a price signal, but the signal is weak and the long-term trajectory is not committed. The green finance framework provides capital for some technologies, but the criteria are not always aligned with the transition needs. The regional transition policies provide some support for coal-dependent regions, but the scale is small relative to the need. The result is a policy environment that is directionally clear but operationally uncertain—which is exactly the environment in which firms delay investment.

The delay is measurable. Investment in new blast furnace capacity in China fell sharply after 2021, but investment in EAF capacity did not rise to fill the gap. Instead, firms held back on both, running existing assets harder and deferring replacement decisions. The same pattern appeared in cement, where new kiln investment slowed even as the industry's overcapacity persisted. Firms are not choosing the low-carbon pathway over the high-carbon one; they are choosing to wait.

7.3.4 Capital Reallocation Strategies: What Firms Actually Do

Faced with this environment, capital-intensive firms have adopted a repertoire of strategies. The most common, and the least discussed, is delay. A firm postpones irreversible investment decisions until the policy environment clarifies. This is rational at the individual level. At the aggregate level, it produces the underinvestment problem described above. Delay is not necessarily passive; a firm that delays may be running pilot projects, building measurement capabilities, and positioning itself for a future decision. But the capital stock remains unchanged, and the emissions trajectory remains on its pre-existing path.

Divestment is the second strategy. A firm sells its high-carbon assets to another owner, typically one with a lower cost of capital, a higher risk tolerance, or a different strategic horizon. Divestment can be a way of managing stranded asset risk at the firm level, but it does not reduce the emissions at the asset level. The asset continues to operate under new ownership. In some cases, divestment can make the problem worse, because the new owner may have less capacity or less incentive to invest in abatement. The wave of coal plant sales from state-owned utilities to private operators in China is an example. The utilities reduce their balance sheet exposure, but the plants continue to emit.

Retrofit is the third strategy, and the one most Chinese capital-intensive firms have actually pursued. A firm invests in incremental abatement—efficiency improvements, fuel switching, waste heat recovery, small-scale renewable generation—that reduces emissions without requiring a wholesale replacement of the asset. Retrofit is attractive because it is modular, reversible, and relatively low-cost. Its limitation is that it has a ceiling. Once the cheap abatement options are exhausted, further reductions require deeper and more expensive changes. A cement plant that has installed waste heat recovery and switched to alternative fuels has captured most of the low-cost abatement available to it. The next ton of CO2 requires carbon capture, which costs several hundred yuan per ton and requires a capital commitment that few firms are willing to make under current policy.

Transformation is the fourth strategy, and the only one compatible with deep decarbonization. A firm replaces its core technology with a low-carbon alternative—hydrogen-based steelmaking, carbon capture and storage for cement, inert anodes for aluminum. Transformation is the most expensive and the most uncertain. It requires long-term capital, patient investors, and a policy environment that supports first-of-a-kind deployment. Very few Chinese capital-intensive firms have pursued transformation at scale. The ones that have—Baowu in steel, Anhui Conch in cement, CHALCO in aluminum—are large state-owned enterprises with access to cheap capital and political backing. The smaller, private firms that make up the bulk of the sectors have neither.

The distribution of strategies across firms is not random. It correlates with ownership, size, export orientation, and access to finance. State-owned enterprises are more likely to pursue transformation, because they can absorb long payback periods and because they face political pressure to align with national targets. Private firms are more likely to pursue delay and retrofit, because they face tighter financial constraints and because their survival depends on short-term profitability. Export-oriented firms are more likely to pursue retrofit and transformation, because they face carbon border adjustment and customer pressure. Domestic-oriented firms are more likely to delay.

This distribution has an important implication for the aggregate transition. The firms with the largest emissions and the longest asset lives—the large state-owned enterprises—are also the firms with the greatest capacity to invest in transformation. The firms with the smallest emissions and the shortest asset lives—the small private firms—are the ones most likely to delay. The aggregate trajectory therefore depends heavily on the behavior of a relatively small number of large firms. This is a different dynamic from the labor-intensive sectors examined in Chapter 5, where the sector was fragmented and the transition was driven by market access and compliance costs. In the capital-intensive sectors, the transition is driven by the investment decisions of a few dozen large firms, and the policy challenge is to align their incentives with the national targets.

7.3.5 Asset-Level Risk Assessment: A Framework

The firms that manage stranded asset risk well share a common practice: they assess risk at the asset level rather than the firm level. This sounds obvious, but it is surprisingly rare. Most Chinese capital-intensive firms report emissions and carbon costs at the consolidated level, which averages across assets with very different exposure. A firm with a modern coastal plant and an aging inland plant may report a moderate average carbon intensity, masking the fact that the inland plant is deeply stranded and the coastal plant is competitive. Asset-level assessment reveals this heterogeneity and allows the firm to make differentiated decisions.

A workable asset-level risk framework has four components. The foundation is an asset inventory that records, for each asset, its age, capacity, technology, location, emissions intensity, remaining book value, and remaining debt. Without it, the firm cannot know what it owns.

The second component is a scenario set that specifies plausible paths for carbon prices, benchmark stringency, electricity prices, technology costs, and demand. The scenarios should be internally consistent and should span a range of outcomes, from a slow transition to a rapid one. Three to five scenarios is typical. They should be chosen to bracket the uncertainty, not to represent the most likely outcome.

The third component is a cash flow model that projects, for each asset under each scenario, the annual revenue, operating cost, carbon cost, capital expenditure, and free cash flow. The model should be transparent about its assumptions and should be sensitive to the key drivers. The output is a distribution of asset values under the scenario set, not a single point estimate.

The fourth component is a decision rule that specifies, for each asset, the action to take under each scenario. The actions might include continue operating, invest in abatement, divest, or retire. The decision rule should be reviewed periodically as the scenarios evolve and as new information becomes available.

This framework is essentially the standard approach to capital budgeting under uncertainty, adapted to include carbon costs. What is novel is the application to Chinese capital-intensive firms, most of which have not yet built the capability to do it. The firms that have built it—typically the large state-owned enterprises with dedicated carbon management teams, or the foreign-invested firms with global mandates—report that the exercise changes their view of their asset portfolio. Assets that looked profitable at the consolidated level turn out to be marginal or negative at the asset level under plausible carbon price scenarios. The exercise does not automatically lead to divestment or retirement, but it does lead to a more informed conversation about capital allocation.

A steel firm that applied this framework in 2023 found that two of its five blast furnaces would turn cash-negative under a carbon price of 150 yuan per ton combined with a benchmark tightening of 2 percent per year. The firm had been reporting a blended carbon intensity that looked competitive against the industry average. The asset-level analysis revealed that the competitive average was carried by three coastal furnaces, while the two inland furnaces were loss-making under any scenario other than the slowest transition. The firm did not close the furnaces immediately. It stopped reinvesting in them, redirected maintenance capital to the coastal sites, and began negotiating a sale of one inland site to a regional consolidator. That is what asset-level risk assessment looks like in practice: not a dramatic closure, but a reallocation of capital that would not have happened under consolidated reporting.

7.3.6 The Role of Financial Markets

Financial markets are the third party in the stranded asset problem, alongside firms and policymakers. Banks, bondholders, and equity investors provide the capital that funds capital-intensive assets, and they bear part of the risk if those assets are stranded. The extent to which they recognize and price this risk has important implications for the cost of capital and for the pace of the transition.

The evidence on whether Chinese financial markets price carbon risk is mixed. A 2022 study by the People's Bank of China's research bureau found that high-carbon firms in the domestic bond market paid a spread premium of 3–8 basis points relative to low-carbon peers after controlling for credit rating, ownership, and sector. The premium is statistically significant but economically small. The green bond market has grown rapidly—issuance reached roughly 800 billion yuan in 2023—but green bonds remain a small share of total corporate bond issuance, and the pricing advantage, or "greenium," is modest, typically 2–5 basis points. Equity markets show some evidence of carbon risk pricing, particularly for firms with high emissions intensity, but the effect is not large enough to change investment decisions.

Carbon risk is under-priced for several reasons. Measurement is the first. Investors cannot price what they cannot measure, and the carbon data disclosed by Chinese firms is incomplete and inconsistent. The CDP disclosure rate among Chinese listed companies is low, and the data that is disclosed is often not comparable across firms. The government has begun to require mandatory environmental disclosure for listed companies, but the rules are new and enforcement is uneven.

Horizon is the second. The stranded asset risk materializes over decades, while most investors have horizons of months or years. A pension fund with a 20-year horizon should care about stranded assets; a hedge fund with a 6-month horizon should not. The structure of the Chinese capital market, which is dominated by retail investors and short-horizon institutional investors, biases against long-horizon risk pricing.

Implicit guarantees are the third. Many Chinese capital-intensive firms are state-owned, and investors believe that the state will not let them fail. This belief is not unfounded—the state has repeatedly bailed out large state-owned enterprises—but it undermines the market discipline that would otherwise price carbon risk. If investors believe that the state will absorb stranded asset losses, they have no incentive to price the risk themselves.

Policy uncertainty is the fourth. Investors cannot price a risk whose policy driver is uncertain. If the carbon price trajectory is unclear, the stranded asset risk is unclear, and the market response is to apply a discount that is either too small (if the risk is underestimated) or too large (if the risk is overestimated). Neither outcome is efficient.

The implication is that financial markets are not yet a significant driver of the capital-intensive transition in China. They will become more important as disclosure improves, as the carbon price rises, and as the state's implicit guarantee is gradually withdrawn. For now, the transition is driven primarily by policy and by the strategic decisions of large firms, not by market pricing.

7.3.7 Managing the Transition: Policy Options and Their Trade-offs

The policy challenge is to manage the stranded asset problem in a way that supports the transition without destabilizing the firms and regions that depend on high-carbon assets. Several options are available, and each has trade-offs.

Slowing the transition is the default option in many policy environments, and it is the one that high-carbon incumbents lobby for. Slowing the transition reduces the stranded asset losses in the short run, but it increases them in the long run, because the assets that are not retired early will be retired later and more abruptly. It also increases the physical risk of climate change, which is itself a source of asset impairment. The trade-off is between near-term financial stability and long-term climate and financial stability. Most economists would argue that the long-term costs of delay exceed the short-term benefits, but the political economy of delay is powerful.

Compensating the losers can take the form of direct transfers, capacity payments, or retirement-and-replacement schemes. Compensation reduces the financial losses borne by the asset owners and their creditors, which reduces the risk of financial instability. But it also creates a moral hazard: if firms expect to be compensated for stranded assets, they have less incentive to avoid creating them. The design of the compensation mechanism matters. Compensation that is tied to early retirement, and that is conditional on the firm not building new high-carbon assets, is less prone to moral hazard than unconditional compensation.

Socializing the losses through the financial system is a third option. If the stranded assets are held by state-owned banks, the losses can be absorbed by the state's balance sheet, as has happened with non-performing loans in other sectors. This avoids a disorderly bankruptcy, but it transfers the cost to taxpayers and it can weaken the banking system's discipline. The scale of the potential losses is large enough that socialization may be unavoidable for some assets, but it should be a last resort rather than a first response.

Accelerating the creation of new assets is the fourth option, and the most attractive in principle. The transition creates new investment opportunities—renewable generation, grid infrastructure, carbon capture, hydrogen, green manufacturing—that can absorb the capital and labor released from high-carbon assets. If the new investment is large enough and fast enough, it can offset the losses from stranded assets. This is the "creative destruction" story. In practice, it requires a policy environment that supports new investment, a financial system that can channel capital to new sectors, and a labor market that can retrain workers. China has strengths in all three areas, but the coordination challenge is significant.

The Chinese policy framework has elements of all four options, but the balance is skewed toward the first two. The transition is proceeding gradually, and compensation mechanisms are being developed for coal plants and for coal-dependent regions. The acceleration of new investment is happening, but it is not explicitly linked to the stranded asset problem. The result is a transition that is slower than the climate targets require and more expensive than it needs to be.

The current status of the most consequential instrument—the national ETS—illustrates the calibration problem. The scheme's coverage was extended to cement and aluminum in 2024, and the MEE has signaled that steel, chemicals, and paper will follow. But the benchmark tightening path has not been published beyond the current compliance period, and the free allocation share remains above 90 percent. Until the allocation plan for the 2026–2030 period is released, firms have no basis for a long-term carbon cost assumption, and the asset-level analysis described in 7.3.5 remains a scenario exercise rather than a capital allocation tool.

7.4 Synthesis: The Capital-Intensive Pathway as a Governance Challenge

The capital-intensive pathway is the pathway where the gap between the rhetoric of green competitiveness and the reality of asset-level economics is widest. The firms that navigate it successfully are not necessarily the firms with the best technology or the lowest costs. They are the firms with the strongest governance capability: the ability to measure emissions credibly, to map carbon costs onto assets, to manage policy uncertainty, and to make irreversible investment decisions under conditions of genuine uncertainty.

This governance capability is not exogenous. It is built through experience, through exposure to carbon pricing, and through the pressure of international markets. The firms that have built it are the ones that have been exposed to the EU ETS, to CBAM, to the disclosure requirements of international investors, and to the procurement standards of global customers. The firms that have not are the ones that operate primarily in domestic markets, where the carbon price is low and the enforcement is uneven.

The implication is that the capital-intensive pathway is path-dependent in a specific sense. Firms that entered the transition early, through export exposure or through pilot ETS participation, have built capabilities that give them an advantage as the transition accelerates. Firms that delayed have not. This is the same dynamic described in the labor-intensive sectors, but the stakes are higher because the assets are larger and the adjustment costs are greater.

The policy implication is that the Chinese government has an opportunity to accelerate capability-building by tightening the carbon market, by expanding its coverage, and by committing to a long-term price trajectory. A higher and more predictable carbon price would force firms to build the measurement and management capabilities that they currently lack. It would also accelerate the reallocation of capital from high-carbon to low-carbon assets, which is the ultimate goal of the transition. The risk is that a rapid increase in the carbon price would strand assets faster than the compensation mechanisms can absorb, creating financial instability. The challenge is to calibrate the pace so that the capability-building and the asset reallocation proceed together.

7.5 What This Chapter Does Not Argue

Three clarifications are in order, following the practice adopted throughout the book.

I do not argue that carbon pricing is the only or even the most important driver of the capital-intensive transition. Technology costs, renewable energy prices, and industrial policy are all important, and in some sectors they dominate. The carbon price is a signal, not a solution. Its role is to coordinate expectations and to make the cost of emissions visible, not to single-handedly drive the transition.

I do not argue that stranded assets are necessarily a bad thing. Some assets should be stranded; that is the point of the transition. The problem is not that high-carbon assets lose value, but that the losses are borne by parties who did not anticipate them and cannot absorb them. The policy challenge is to manage the distribution of the losses, not to prevent them.

I do not argue that the Chinese pathway will follow the European pathway. The institutional context is different, the stage of development is different, and the policy priorities are different. The European experience offers lessons, but it is not a template. The Chinese pathway will be distinctive, and its distinctiveness will be shaped by the same factors that have shaped China's industrial development more broadly: the role of the state, the scale of the market, and the speed of structural change.

7.6 Looking Ahead

The next chapter takes up supply chain governance and multi-tier responsibility in global production networks. The connection to this chapter is direct. A capital-intensive firm's carbon exposure is not limited to its own assets; it extends upstream to its suppliers and downstream to its customers. A steel mill's carbon footprint includes the emissions from the coal and iron ore it buys, and the emissions from the products it sells. A cement plant's footprint includes the emissions from the clinker it buys and the concrete it sells. Managing these multi-tier emissions requires a different set of governance capabilities—supplier engagement, traceability, and contractual allocation of responsibility—that are distinct from the asset-level capabilities discussed here. The capital-intensive pathway and the supply chain pathway are complementary, and the firms that build both sets of capabilities will be better positioned as the carbon price rises and the benchmark tightens.