Chapter 7: Capital-Intensive Industries: Scale, Stranded Assets, and the Governance of Low-Carbon Process Substitution

7.1 The Anatomy of a Blast Furnace That Cannot Be Retired

The No. 3 blast furnace at a large integrated steelworks in Tangshan was relined in 2019. The reline took ninety-four days and cost roughly 1.4 billion yuan, and it reset the vessel's campaign life to somewhere between twelve and fifteen years. When I asked the plant's deputy chief engineer why the company had not used the reline as an opportunity to switch to a hydrogen-based direct reduction route, he laughed before answering. The laughter was not dismissive. It was the laugh of a man who had been asked the same question by journalists, consultants, and visiting academics for three years running, and who had developed a standard reply. "The furnace will still be here in 2035," he said. "The question is whether the furnace is the problem or the steel is the problem."

That distinction matters more than it first appears. A blast furnace is not a generic asset. It is a vessel with a specific geometry, a specific refractory lining, a specific thermal profile, and a specific relationship to the coking coal and iron ore that feed it. You cannot convert it to hydrogen reduction by changing a nozzle. The furnace embodies a set of technical choices that were made when it was designed, and those choices in turn embody a set of relative prices—coal, ore, capital, electricity—that held when the investment was approved. When those relative prices shift, the furnace does not become obsolete in the way a laptop becomes obsolete. It becomes stranded in the way a coal mine becomes stranded: still functional, still profitable at the margin, but no longer consistent with the price vector that the policy environment is trying to construct.

The deputy chief engineer's plant is one of roughly four hundred integrated steel facilities in China, a figure that comes from the China Iron and Steel Association's 2022 membership roster and excludes the independent rolling mills that buy billet on the spot market. Together these facilities produced just over one billion tonnes of crude steel in 2022. The sector accounted for somewhere between 15 and 18 percent of national CO2 emissions, depending on whether you count only direct combustion and process emissions or add the emissions from purchased electricity; the lower figure is the direct-only boundary used in the Ministry of Ecology and Environment's annual inventory, the higher one includes the power sector allocation. The blast furnace route—coke oven, sinter plant, blast furnace, basic oxygen furnace—dominates Chinese steelmaking, and the average age of the vessels is under fifteen years. This is the crux of the capital-intensive problem. In labor-intensive industries, capital turns over quickly: a garment factory can be re-tooled or relocated in eighteen months. In technology-intensive industries, capital is largely embodied in knowledge and equipment that can be upgraded incrementally. In capital-intensive industries, capital is embodied in physical plant with campaign lives measured in decades, and the plant is the emissions source. The asset and the liability are the same object.

This chapter examines how capital-intensive industries navigate this condition. The argument is not that scale is incompatible with decarbonization, nor that capital intensity necessarily produces lock-in. The argument is more specific: capital-intensive industries face a governance problem that is structurally different from the governance problems of labor- and technology-intensive sectors, and that problem is best understood as a problem of process substitution under asset specificity. The substitution of one industrial process for another—blast furnace for direct reduction, clinker kiln for calcined clay, steam cracker for electrified cracking—is not a technical problem that can be solved by R&D alone. It is a problem of coordinating investment timing, policy credibility, and value chain reconfiguration across assets that cannot be moved, cannot be repurposed, and cannot be depreciated faster than the physical wear of their refractory linings.

7.2 Scale Economies and Carbon Lock-In: The Mechanism of Asset-Embedded Emissions

7.2.1 Why Scale Is Not the Problem

The standard critique of capital-intensive industry in the climate literature treats scale as the source of lock-in. Large plants, the argument runs, produce large emissions, and the political influence of large plants makes regulation difficult. This is a story about political economy, and it has some truth to it. But it conflates two distinct phenomena: the scale of the emissions source and the scale of the investment unit. The first is a physical fact. The second is an economic and institutional fact, and it is the second that governs the substitution problem.

Consider the arithmetic. A 5-million-tonne integrated steelworks emits roughly 9 million tonnes of CO2 per year at typical Chinese intensities of about 1.8 tonnes of CO2 per tonne of crude steel. A 500,000-tonne electric arc furnace mini-mill emits roughly 0.3 million tonnes if it runs on grid average electricity, or near zero if it runs on dedicated renewables. The emissions difference is a factor of thirty, but the investment difference is a factor of perhaps eight. The large plant is not disproportionately dirty per unit of capital; it is disproportionately dirty per unit of output because of the process it uses, and the process is a function of scale. Blast furnaces only make economic sense above a certain threshold of hot metal output, and basic oxygen furnaces only make sense downstream of blast furnaces. The scale and the process are co-determined.

This co-determination is what I mean by asset-embedded emissions. The emissions profile of a capital-intensive facility is not a variable that can be adjusted independently of the facility's physical configuration. It is embedded in the configuration. A blast furnace is a reducing vessel that uses carbon as both fuel and reductant; the carbon is not an input that can be swapped for another input without changing the vessel. A cement kiln is a calcination vessel in which the CO2 from limestone decomposition is an intrinsic product of the chemistry; you cannot make clinker without releasing that CO2 unless you change the chemistry of the feedstock. A steam cracker is a thermal vessel optimized for a particular feedstock slate; switching to electrified cracking changes the reactor design, the heat integration, and the downstream separation train.

The implication is that decarbonization in capital-intensive industries is not a matter of improving efficiency at the margin. It is a matter of replacing the vessel. And replacing the vessel is a decision that must be made against the background of an existing vessel that is still working, still depreciating, and still generating cash flow. This is the classic problem of capital replacement under asset specificity, and it is the problem that the rest of this chapter addresses.

7.2.2 The Three Channels of Lock-In

Carbon lock-in in capital-intensive industries operates through three channels, each with a different time constant and a different policy implication. I distinguish them because the policy response to each is different, and because conflating them has produced a good deal of confused policy advice.

The first channel is technical lock-in: the embedding of emissions in the physical configuration of the plant, as described above. Its time constant is the campaign life of the asset—twelve to fifteen years for a blast furnace, twenty to thirty years for a cement kiln, thirty to forty years for a steam cracker. Technical lock-in is not absolute; a plant can be retrofitted, and some retrofits (top gas recycling on a blast furnace, waste heat recovery on a kiln) can reduce emissions by 10 to 20 percent. But the retrofit does not change the fundamental chemistry. It buys time, and the time it buys is bounded by the next reline.

The second channel is infrastructural lock-in. Capital-intensive plants are embedded in infrastructure systems—rail links, port terminals, pipeline networks, power grids—that are themselves long-lived and that were built to serve the existing process configuration. A steelworks served by a coking coal rail spur cannot switch to hydrogen direct reduction without a hydrogen pipeline or a dedicated renewable power connection, and building either takes five to ten years. The infrastructure is not owned by the plant, but the plant's location was chosen because the infrastructure existed, and the infrastructure will not be reconfigured unless someone coordinates the reconfiguration. This is the collective action problem that Chapter 9 addresses in the context of industrial clusters.

Expectations, meanwhile, harden into institutional lock-in. Banks lend against the existing asset base; regulators set emissions benchmarks against the existing technology set; local governments plan tax revenue around the existing tax base. Each expectation reinforces the others: a bank that has lent against a blast furnace has an interest in the furnace continuing to operate, a regulator that has set a benchmark against blast furnace technology has an interest in not raising the benchmark too quickly, and a local government that depends on the steelworks for 30 percent of its fiscal revenue has an interest in the steelworks not closing. Of the three channels, this one has the longest time constant, because it is reproduced through organizational routines and political coalitions rather than through physical assets.

The interaction of the three channels is what makes lock-in persistent. A plant that is technically locked in, infrastructurally locked in, and institutionally locked in will not decarbonize on its own, no matter how high the carbon price. The carbon price raises the cost of operating the plant, but it does not change the configuration of the plant, the infrastructure around it, or the expectations of the actors who depend on it. What it does is redistribute the cost—from the plant to its customers, from the jurisdiction to its trading partners, from the present to the future. Redistribution is not substitution.

7.2.3 The Shunde Vignette: A Different Kind of Scale

I want to set the steel and cement cases against a different kind of capital-intensive facility, because the contrast clarifies what is specific about the problem. In Shunde, in the Pearl River Delta, I visited a furniture components plant that had been converted from a labor-intensive assembly operation into a highly automated panel production facility. The plant employed 180 workers where it had previously employed 1,200. Its capital intensity, measured as fixed assets per employee, had risen by a factor of roughly six. Its energy intensity had risen as well, because the automated lines ran on electricity and the drying ovens ran on natural gas.

By the standard definition, this is now a capital-intensive plant. But its emissions profile is not locked in. The panel line can run on grid electricity or on rooftop solar; the drying ovens can be electrified or run on biomethane; the building envelope can be upgraded without touching the production equipment. The reason is that the plant's capital is modular. It is composed of machines that can be replaced individually, and the emissions profile is a function of the energy supply rather than of the process chemistry. The plant is capital-intensive in the sense that it uses a lot of capital per worker, but it is not asset-specific in the sense that its capital cannot be redeployed or reconfigured.

This distinction is essential to the argument of this chapter. Capital intensity is not the same as asset specificity. A plant can be capital-intensive and still have a low substitution cost, because its capital is modular and its emissions are upstream (in the electricity and gas it consumes) rather than embedded (in the chemistry of its process). The lock-in problem arises when capital intensity coincides with asset specificity—when the capital is large and also specific to a particular process, a particular feedstock, and a particular product. This coincidence is characteristic of the process industries: steel, cement, chemicals, aluminum, pulp and paper, glass, ceramics. It is not characteristic of discrete manufacturing, even when discrete manufacturing is highly automated.

The Shunde case also illustrates a second point. The conversion from labor-intensive to capital-intensive production was itself a response to rising labor costs and to the green governance requirements of export markets. The plant's customers in Europe and North America had begun to demand carbon footprint disclosures, and the plant's management had decided that the cheapest way to meet those demands was to electrify the drying ovens and install rooftop solar, which reduced the plant's Scope 1 and Scope 2 emissions by enough to satisfy the disclosure requirements without changing the production process. The capital investment was justified on cost grounds (labor savings) and the green investment was justified on market access grounds (disclosure compliance). The two investments were complementary, but they were not the same investment, and neither required the other.

In the process industries, this complementarity is much harder to achieve. The capital investment that reduces labor cost (a new blast furnace) and the capital investment that reduces emissions (a direct reduction plant) are not complementary; they are substitutes. You cannot do both with the same capital. This is the structural difference that makes the governance of process substitution in capital-intensive industries a distinct problem.

7.2.4 Scale Economies and the Cost of Substitution

The economics of substitution in capital-intensive industries are governed by the relationship between scale economies and the cost of the substitute process. This relationship is not straightforward, and it has been the source of considerable confusion in the policy literature.

Consider the case of steel. The dominant low-carbon substitute for the blast furnace route is hydrogen-based direct reduction (H-DR) coupled with electric arc furnace (EAF) steelmaking. The H-DR process produces direct reduced iron (DRI) using hydrogen as the reductant, and the DRI is then melted in an EAF. The theoretical emissions are near zero if the hydrogen is produced from renewable electricity. The economics, however, depend on three variables: the cost of renewable electricity, the cost of electrolyzers, and the scale of the DRI plant.

The scale economies of H-DR are different from the scale economies of the blast furnace. A blast furnace achieves its lowest cost at around 2 to 3 million tonnes of hot metal per year, and the capital cost per tonne declines sharply up to that point. An H-DR plant achieves its lowest cost at a similar scale, but the capital cost per tonne is higher at every scale, because the shaft furnace and the hydrogen supply chain are more capital-intensive per unit of output than the blast furnace and the coke oven. The result is that H-DR steel is more expensive than blast furnace steel at current prices, and the gap does not close with scale alone. It closes only when the cost of carbon (or the cost of hydrogen) reaches a threshold.

This threshold is the subject of intense debate. The range I use here—300 to 800 yuan per tonne of CO2—is my own calculation, built from three assumptions that I vary across the plausible range. The first is the electricity price: 0.25 to 0.45 yuan per kWh, which spans the coastal industrial tariff and the discounted rate available to dedicated renewable projects in Inner Mongolia. The second is the electrolyzer capital cost: 2,500 to 4,000 yuan per kW, which spans the 2023 Chinese market price for alkaline stacks and the installed cost including balance of plant. The third is the utilization rate of the DRI shaft: 70 to 85 percent, which spans the current operating experience of the two Chinese demonstration plants. At the favorable end of all three assumptions, the break-even carbon price is around 300 yuan; at the unfavorable end, it is around 800 yuan. The current national ETS price is around 60 to 80 yuan per tonne, and the effective carbon cost faced by steel producers after free allocation is lower still. The gap is therefore large, and it will not close through the ETS alone in the near term. It will close through a combination of carbon pricing, subsidy, and—critically—the willingness of buyers to pay a green premium.

The willingness of buyers to pay a green premium is itself a scale economy, but of a different kind. It is a demand-side scale economy: the more buyers who commit to green steel, the lower the risk premium that producers must charge, and the lower the effective cost of the substitute. This is why the governance of process substitution cannot be reduced to a carbon price. It requires the coordination of supply-side investment and demand-side commitment, and the coordination mechanism is the subject of Section 7.4.

7.2.5 The Cement Case: A Harder Problem

Cement is a harder case than steel, and it is worth examining because it exposes the limits of the substitution corridor approach. Roughly 60 percent of cement emissions come from the calcination of limestone (CaCO3 → CaO + CO2), and this CO2 is a product of the chemistry, not of the fuel. You can electrify the kiln, you can switch to alternative fuels, you can improve the efficiency of the grinding and blending, but you cannot eliminate the process CO2 without changing the feedstock or capturing the CO2.

The substitute processes are therefore of two kinds. The first is clinker substitution: replacing a portion of the clinker with calcined clay, fly ash, slag, or limestone. This can reduce the clinker factor from around 0.75 to 0.50 or lower, with a corresponding reduction in process emissions. The technology is mature, and the capital cost is modest. The barrier is not technical but institutional: the standards that govern cement performance (GB 175 in China, EN 197 in Europe) specify minimum clinker contents for different strength classes, and changing the standards requires a multi-year process of testing and approval. The second substitute is carbon capture and storage (CCS) on the kiln flue gas. This is technically feasible but capital-intensive and energy-intensive, and it requires a CO2 transport and storage infrastructure that does not yet exist in most of China.

The cement case illustrates a general point about process substitution in capital-intensive industries: the binding constraint is often not the cost of the substitute process but the cost of the complementary infrastructure and the cost of regulatory approval. A cement plant that wants to switch to calcined clay needs a supply of calcined clay, which requires a calcination facility, which requires a capital investment that is separate from the cement plant. A cement plant that wants to install CCS needs a CO2 pipeline and a storage site, which are collective infrastructure that no single plant can justify. The substitution is therefore not a plant-level decision; it is a system-level decision, and the governance of the system is the subject of the next section.

7.3 Stranded Assets and Investment Risk: Accounting, Economic, and Physical Stranding

7.3.1 Three Kinds of Stranding

The concept of stranded assets was introduced into the climate policy literature by the Carbon Tracker Initiative in 2011, and it has since become a central category in the analysis of climate-related financial risk. The original formulation focused on fossil fuel reserves that cannot be burned if the world is to stay within a carbon budget, and the implication was that the reserves would be written down as the carbon constraint tightened. The concept has since been extended to any asset whose value is impaired by the transition to a low-carbon economy.

The extension has been useful but imprecise. In the capital-intensive industries that are the subject of this chapter, it is necessary to distinguish three kinds of stranding, because they have different implications for investment risk and different policy responses.

Accounting stranding is the write-down of an asset on a firm's balance sheet. It occurs when the asset's carrying value exceeds its recoverable amount, and it is triggered by a revision of expected future cash flows. A blast furnace that is expected to operate for fifteen years but is now expected to operate for ten years, because the carbon price is expected to rise, has an accounting stranding equal to the present value of the cash flows in the five years that will not occur. Accounting stranding is a financial event. It affects the firm's reported earnings, its debt covenants, and its access to capital. It does not affect the physical operation of the asset.

Economic stranding is the loss of the asset's competitive position. It occurs when the asset's operating costs rise above the costs of its competitors, or when the price of its output falls below the cost of production. A blast furnace that faces a carbon price of 300 yuan per tonne while its competitors face a carbon price of 100 yuan is economically stranded, even if it continues to operate. Economic stranding is a market event. It affects the firm's cash flow, its market share, and its ability to reinvest. It does not necessarily affect the asset's physical operation, because the firm may continue to operate at a loss in the hope that conditions improve.

Physical stranding is the closure or abandonment of the asset. It occurs when the asset is no longer operated, either because it is uneconomic or because it is prohibited. Physical stranding is a physical event. It affects the workers, the community, and the infrastructure that depended on the asset. It is the event that the other two kinds of stranding anticipate.

The distinction matters because the three kinds of stranding occur on different time scales and are governed by different mechanisms. Accounting stranding can occur quickly, in response to a change in expectations. Economic stranding occurs more slowly, as the asset's cost position deteriorates. Physical stranding occurs last, and it is often delayed by political intervention. A firm that faces accounting stranding may respond by lobbying for a delay in the policy that triggered it, or by selling the asset to a buyer with a higher discount rate, or by accelerating depreciation. A firm that faces economic stranding may respond by cutting costs, by shifting to higher-value products, or by exiting the market. A firm that faces physical stranding has no response; the asset is gone.

7.3.2 The Unburnable Carbon Problem in Chinese Steel

The unburnable carbon framework was developed for fossil fuel reserves, but it applies with modifications to the capital-intensive industries. The modification is that the unburnable carbon is not the fuel itself but the emissions embedded in the process. A blast furnace's carbon is not unburnable in the sense that the coal cannot be extracted; it is unburnable in the sense that the emissions from burning it cannot be released if the sector is to meet its decarbonization target.

The arithmetic for Chinese steel is sobering. The sector's emissions are roughly 1.8 billion tonnes of CO2 per year, and the sector's share of the national carbon budget consistent with the 2060 neutrality target is roughly 0.3 to 0.5 billion tonnes by 2050. The gap is a factor of four to six. Closing it requires the retirement or retrofit of a large share of the existing blast furnace capacity, and the retirement must occur before the end of the vessels' physical lives. The assets that will be retired early are the stranded assets.

How much capacity is at risk? The answer depends on the assumed rate of retirement and the assumed rate of replacement. A simple calculation: if the average campaign life of a blast furnace is fifteen years, and if the sector is to reduce emissions by 80 percent by 2050, then roughly 70 percent of the current capacity must be retired or retrofitted by 2050. That is roughly 700 million tonnes of capacity. At a replacement cost of 1,000 yuan per tonne of annual capacity, the replacement investment is 700 billion yuan. At a book value of 500 yuan per tonne, the write-down is 350 billion yuan. These numbers are rough, but they indicate the scale of the financial adjustment.

The distribution of the adjustment is uneven. The newest plants, built after 2015, have the longest remaining campaign lives and the highest book values; they face the largest accounting stranding. The oldest plants, built before 2005, have the shortest remaining lives and the lowest book values; they face the smallest accounting stranding but the largest physical stranding. The plants in the middle, built between 2005 and 2015, face the largest economic stranding, because they have enough remaining life to be affected by the carbon price but not enough to justify a full retrofit.

This distribution has a political dimension. The newest plants are often the most efficient and the most profitable; they are also the most indebted. The oldest plants are often the least efficient and the least profitable; they are also the most likely to be state-owned and the most likely to be protected by local governments. The middle cohort is the most exposed, and it is the cohort that the policy has the least capacity to reach, because it is too new to be retired and too old to be retrofitted.

7.3.3 The Credibility of Policy and the Timing of Write-Downs

The timing of accounting stranding depends on the credibility of the policy that triggers it. If firms believe that the carbon price will rise, they will write down their assets sooner. If they believe that the policy will be delayed or diluted, they will delay the write-down. The write-down is therefore a function of the policy and of the expectations the policy generates, and the second matters as much as the first.

This is a point that the stranded assets literature has not fully absorbed. The literature tends to treat the carbon budget as an exogenous constraint and to calculate the stranding as the difference between the asset base and the budget. But the budget is not exogenous; it is mediated by policy, and policy is mediated by expectations. A firm that expects the policy to be delayed will not write down its assets, and the delay will become self-fulfilling, because the firm's continued operation will make the policy more difficult to implement.

The Chinese steel sector illustrates this dynamic. The national ETS was launched in 2021, but the steel sector was not included in the first compliance cycle. The sector's inclusion has been repeatedly postponed, and the allocation methodology has not been finalized. The result is that steel firms have had little incentive to write down their assets, and the sector's book values remain high. The 2025 inclusion of the sector in the ETS will change this, but the change will be gradual, because the allocation will be free for the first few years and the effective carbon cost will be low.

The credibility problem is compounded by the structure of the banking system. Chinese steel firms are heavily indebted to state-owned banks, and the banks have an interest in not recognizing the stranding, because recognizing it would require them to provision for losses. The result is a kind of institutional forbearance: the banks roll over the loans, the firms continue to operate, and the stranding is deferred. This is not necessarily irrational; it may be the least-cost way to manage the adjustment, given the political constraints. But it does mean that the stranding is not priced, and that the capital that should be reallocated to low-carbon processes is not reallocated.

7.3.4 The Investor Perspective: Why Capital Is Not Fleeing

The stranded assets literature predicted that investors would flee carbon-intensive assets as the transition accelerated. In the capital-intensive industries, this has not happened to the extent predicted. The reasons are instructive.

The first reason is that the investors who hold these assets are not the investors who are most sensitive to climate risk. Chinese steel firms are owned by a mix of state entities, domestic institutional investors, and retail investors. The state entities have non-financial objectives (employment, regional development, industrial policy) that dominate their investment decisions. The domestic institutional investors are subject to regulatory guidance that discourages divestment from strategic sectors. The retail investors are not organized enough to exert pressure. The international investors who are most sensitive to climate risk have limited exposure to Chinese steel, because the sector is not fully open to foreign investment.

The second reason is that the stranding is uncertain. The carbon price trajectory, the policy trajectory, and the technology trajectory are all uncertain, and the uncertainty cuts both ways. A firm that writes down its assets today may find that the policy is delayed and that the write-down was premature. A firm that delays the write-down may find that the policy is accelerated and that the delay was costly. The option value of waiting is positive, and it is larger when the policy is less credible.

The third reason is that the assets are not easily substitutable. A steel firm cannot sell its blast furnace and buy an H-DR plant; the blast furnace has no resale value, and the H-DR plant requires a different site, a different infrastructure, and a different workforce. The asset is specific, and specificity is a barrier to exit. The firm is locked in not because it wants to be but because the cost of exit is higher than the cost of staying.

These three reasons explain why the stranding in Chinese steel has been gradual rather than abrupt. They also explain why the stranding is likely to accelerate once the policy becomes credible. The 2025 inclusion of steel in the ETS, the tightening of the allocation methodology, and the introduction of the EU's Carbon Border Adjustment Mechanism (CBAM) are all signals that the policy is becoming more credible. If the signals are sustained, the write-downs will follow.

7.3.5 The Shunde Contrast: Stranding in Modular Capital

The Shunde furniture components plant provides a contrast that clarifies the specificity of the stranding problem. When the plant converted from labor-intensive to capital-intensive production, it retired a set of sewing machines, assembly jigs, and manual drying racks. The retirement was not a stranding event in the sense used in this chapter, because the assets were modular and had resale value. The sewing machines were sold to smaller workshops; the jigs were scrapped; the drying racks were replaced by ovens that could be electrified. The write-down was absorbed in the normal course of depreciation, and the firm did not face a stranded asset problem.

The contrast is not that the Shunde plant was more virtuous or more forward-looking. It is that the plant's capital was modular, and modular capital can be redeployed. The steel plant's capital is specific, and specific capital cannot be redeployed. The difference in stranding risk is a difference in the structure of the capital, not in the behavior of the firm.

This has an implication for the design of climate policy. A policy that raises the cost of carbon across the economy will have very different effects on modular capital and specific capital. For modular capital, the policy will accelerate the replacement of equipment and the shift to cleaner energy sources, and the adjustment will be relatively smooth. For specific capital, the policy will create stranded assets, and the adjustment will be conflictual. The policy therefore needs to be differentiated: it needs to provide a credible signal for modular capital and a managed transition for specific capital. The undifferentiated carbon price is not enough.

7.4 The Governance of Process Substitution

7.4.1 What Substitution Requires

The governance of process substitution is the central problem of capital-intensive decarbonization. The problem is not that the technologies do not exist; most of the technologies exist, at least at demonstration scale. The problem is that the technologies cannot be deployed at commercial scale without a coordinated sequence of investments that no single firm can undertake alone.

I want to specify what this sequence requires. The first requirement is demonstration. A new process must be shown to work at a scale that is large enough to be credible to investors and small enough to be affordable. Demonstration is expensive, and the expense is not recoverable from the sale of the product, because the product is a commodity. The demonstration must therefore be financed by someone other than the firm, or by a consortium of firms that share the cost.

The second requirement is infrastructure. A new process requires new inputs (hydrogen, renewable electricity, captured CO2) and new outputs (CO2 storage, new product grades). The infrastructure for these inputs and outputs is collective, and it is not built unless someone coordinates the investment. The coordination problem is acute because the infrastructure is specific to the process: a hydrogen pipeline built for one steel plant can serve other plants in the same cluster, but it cannot serve a plant in a different cluster. The infrastructure is therefore a club good, and the club must be organized.

The third requirement is demand commitment. A new process produces a product that is more expensive than the incumbent product, at least initially. The producer will not invest unless it has a buyer who is willing to pay the premium. The buyer will not commit unless it has a customer who is willing to pay the premium. The demand commitment must therefore cascade through the value chain, and the cascade must be organized by someone with the authority to make credible commitments.

The fourth requirement is policy credibility. The entire sequence depends on the expectation that the policy environment will remain supportive for long enough to justify the investment. If the policy is reversed, the investment is stranded. The policy credibility is therefore a precondition for the other three requirements, and it is the hardest to establish, because it depends on the political system's ability to commit.

These four requirements define what I call a substitution corridor: a coordinated sequence of demonstration, infrastructure provision, demand commitment, and policy credibility that allows firms to commit to new processes before the old processes become unprofitable. The corridor is not a single policy instrument; it is a configuration of instruments and institutions that must be aligned. The rest of this section examines how the corridor has been constructed in different capital-intensive industries, and where it has failed.

7.4.2 The Steel Corridor: Hydrogen and the Cluster

The steel industry's substitution corridor is organized around hydrogen. The technology is hydrogen-based direct reduction, the infrastructure is hydrogen production and transport, the demand commitment is the green steel premium, and the policy credibility is the combination of the ETS, the CBAM, and the industrial policy support for hydrogen.

The Chinese corridor is being built in clusters. The Baowu Group's Zhanjiang base, the Hebei Iron and Steel Group's Handan base, and the Shagang Group's Zhangjiagang base are all developing hydrogen-based demonstration projects, and each project is embedded in a regional hydrogen strategy. The Zhanjiang project, for example, is linked to a green hydrogen production facility that will use offshore wind power, and the hydrogen will be transported to the steel plant through a dedicated pipeline. The project is financed by a consortium that includes the steel firm, the provincial government, and a state-owned energy company, and the offtake is committed to a group of downstream manufacturers who have pledged to buy green steel.

The corridor is not yet complete. The demonstration project is at a scale of around 1 million tonnes of DRI per year, which is large enough to be credible but small enough to be affordable. The hydrogen supply is not yet at commercial scale, and the cost of hydrogen is still two to three times the cost of the hydrogen that would be needed for the process to be competitive. The demand commitment is not yet binding, because the green steel premium is still voluntary and the buyers can withdraw. The policy credibility is improving, but it is not yet sufficient to justify a full-scale investment.

The corridor is nevertheless more advanced than it was five years ago, and the reason is that the coordination problem has been addressed by the cluster. The cluster provides a mechanism for sharing the cost of the infrastructure, for coordinating the timing of the investments, and for building the demand commitment. The cluster is not a market; it is a governance structure, and its effectiveness depends on the density of the relationships among the firms and the local government. This is the theme of Chapter 9, and I will not develop it here. The point for this chapter is that the substitution corridor in steel is a cluster-level phenomenon, and it cannot be understood at the level of the individual firm.

7.4.3 The Cement Corridor: Standards and Capture

The cement industry's substitution corridor is organized around two technologies: clinker substitution and carbon capture. The two technologies have different corridor requirements, and the difference is instructive.

Clinker substitution requires a change in the product standard. The standard specifies the minimum clinker content for each strength class, and the substitution is only possible if the standard is revised. The revision is a collective decision, and it is governed by a technical committee that includes representatives from the industry, the research institutes, and the regulators. The committee has been discussing the revision for more than a decade, and the progress has been slow, because the industry is divided: the large firms that have invested in clinker substitution want the standard revised, and the small firms that have not invested want it kept. The corridor is therefore blocked by a distributional conflict within the industry, and the conflict is resolved only when the government intervenes to set a deadline.

Carbon capture requires a different kind of corridor. The technology is capital-intensive, and it requires a CO2 transport and storage infrastructure that is even more capital-intensive. The infrastructure is a natural monopoly, and it is not built unless the government commits to a storage site and a transport network. The commitment is expensive, and it is not justified unless there is a sufficient volume of CO2 to transport. The volume depends on the number of cement plants that install capture, and the number depends on the availability of the infrastructure. This is a classic chicken-and-egg problem, and it is resolved only when the government builds the infrastructure ahead of demand, or when a consortium of plants commits to a shared infrastructure.

The Chinese cement corridor is being built at the provincial level. Guangdong, Anhui, and Shandong have all announced carbon capture demonstration projects, and each project is linked to a provincial storage site. The projects are at an early stage, and the costs are high: the capture cost is estimated at 300 to 500 yuan per tonne of CO2, and the transport and storage cost adds another 100 to 200 yuan. The total cost is therefore comparable to the cost of the clinker substitution, but the infrastructure requirement is much larger. The corridor is therefore less advanced than the steel corridor, and the reason is that the infrastructure is more collective and the coordination problem is more acute.

7.4.4 The Chemical Corridor: Feedstock and Electrification

The chemical industry's substitution corridor is the most complex of the three, because the industry produces a wide range of products from a small number of feedstocks, and the substitution pathways are product-specific. The two main pathways are feedstock substitution (biomass, recycled plastics, CO2) and electrification (electric cracking, electric steam generation).

Feedstock substitution requires a supply chain for the alternative feedstock, and the supply chain is often more expensive than the incumbent supply chain. Biomass, for example, is dispersed and seasonal, and the logistics cost is high. Recycled plastics are contaminated and variable, and the sorting cost is high. CO2 is dilute and requires energy to concentrate. The corridor therefore requires a supply chain investment that is separate from the chemical plant, and the investment is not justified unless the demand for the alternative feedstock is large enough.

Electrification requires a renewable electricity supply and a redesign of the production process. The electric cracker, for example, uses electricity to heat the reactor instead of burning fuel, and the reactor must be redesigned to accommodate the different heat transfer characteristics. The redesign is capital-intensive, and it is not justified unless the electricity price is low enough and the carbon price is high enough. The corridor therefore requires a combination of cheap renewable electricity and a credible carbon price, and the combination is not yet available in most of China.

The chemical corridor is being built in the coastal provinces, where the renewable electricity supply is abundant and the export market is accessible. The Zhejiang pilot, for example, is developing an electric cracker that will use offshore wind power, and the offtake is committed to a group of European buyers who are subject to the CBAM. The corridor is therefore driven by the export market, and the domestic marketis lagging. This pattern recurs across the capital-intensive industries: the corridor is built where the demand commitment is strongest, and the demand commitment is strongest in the export-oriented clusters.

7.4.5 Why the Corridor Fails: Three Failure Modes

The substitution corridor is a useful construct because it identifies the conditions under which process substitution can succeed. It is equally useful for identifying the conditions under which it fails. I distinguish three failure modes, each of which has been observed in the Chinese capital-intensive industries.

The first failure mode is demonstration without deployment. A technology is demonstrated at pilot scale, the demonstration is celebrated, and then nothing happens. The reason is that the demonstration was financed by a research grant or a state-owned enterprise's innovation budget, and the financing does not extend to commercial scale. The demonstration proves that the technology works, but it does not prove that the technology is economic, and the gap between the two proofs is not closed. The result is a portfolio of pilots that are revisited every few years but never scaled.

The second failure mode is infrastructure without demand. A government builds a hydrogen pipeline or a CO2 storage site, and then discovers that there are no customers. The reason is that the infrastructure was built ahead of the demand commitment, and the demand commitment did not materialize because the product premium was not high enough or the buyers were not organized. The infrastructure becomes a stranded asset in its own right, and the government becomes reluctant to build more.

The third failure mode is demand without supply. A group of buyers commits to green steel or green cement, and then discovers that the supply is not available at the required scale or price. The reason is that the demand commitment was made without a corresponding supply-side investment, and the supply-side investment was not made because the producers did not believe the demand commitment was binding. The buyers revert to conventional products, and the demand signal is lost.

The three failure modes are symmetrical: each is a mismatch between the supply side and the demand side of the corridor. The successful corridors are those in which the mismatch is managed by an institution that has the authority to coordinate the two sides. In the steel cluster, the institution is the local government working with the state-owned enterprise. In the cement province, the institution is the provincial development and reform commission. In the chemical coastal zone, the institution is the export buyer working through a long-term contract. The institution is not always a government; it can be a firm or a consortium. What matters is that it has the authority to make credible commitments on both sides.

7.5 The Governance Capability of the Capital-Intensive Firm

7.5.1 From Compliance to Capability in the Process Industries

The central argument of this book is that green governance capability is an endogenous driver of industrial upgrading, not a passive product of regulatory compliance. The argument applies to capital-intensive industries, but it applies with a twist. In the labor-intensive industries, the capability is built through the accumulation of experience in meeting the requirements of buyers and regulators; the firm learns by doing, and the learning is embodied in routines and relationships. In the technology-intensive industries, the capability is built through the accumulation of R&D and the integration of green requirements into product design; the firm learns by innovating, and the learning is embodied in patents and platforms. In the capital-intensive industries, the capability is built through the governance of process substitution; the firm learns by coordinating, and the learning is embodied in the ability to assemble a substitution corridor.

This is a different kind of capability, and it deserves a name. I call it substitution governance capability: the ability of a firm to assemble the demonstration, infrastructure, demand commitment, and policy credibility that are required to replace a carbon-intensive process with a low-carbon alternative. The capability is not a technology; it is an organizational and relational capacity. It resides in the firm's ability to work with governments, infrastructure providers, buyers, and financiers, and to sequence the commitments so that each party's investment is justified by the others.

The capability is observable in the behavior of the firms that have it. A firm with substitution governance capability does not wait for the carbon price to rise before it begins the demonstration. It begins the demonstration early, because it knows that the demonstration is a precondition for the infrastructure and the demand commitment. It does not build the infrastructure itself; it convenes the consortium that builds the infrastructure. It does not assume that the demand will materialize; it negotiates the offtake agreements before the investment is approved. It does not treat the policy as exogenous; it engages with the policy process to shape the credibility that its investment requires.

The firms that lack the capability behave differently. They wait for the policy to be certain before they invest, and the policy is never certain. They free-ride on the infrastructure that others build, and the infrastructure is never built. They sell to the spot market, and the spot market does not pay the green premium. They lobby against the policy, and the policy becomes less credible. The result is that they are locked into the incumbent process, not because the substitute is unavailable but because they cannot assemble the corridor.

7.5.2 The Organizational Location of the Capability

The substitution governance capability is not evenly distributed within the firm. It is located in a specific organizational function, and the location matters for the firm's ability to exercise it.

In the Chinese steel firms that have been most successful at developing the capability, the function is the strategic investment department, and it reports directly to the board. The department is staffed with engineers who have plant experience and with economists who have policy experience, and it has the authority to convene the other departments (production, procurement, finance, government affairs) around a substitution project. The department's authority is essential, because the substitution project cuts across the functional boundaries of the firm: it requires the production department to accept a new process, the procurement department to secure a new feedstock, the finance department to accept a new risk profile, and the government affairs department to negotiate with the local government. A department without convening authority cannot manage the project.

In the cement firms, the function is often located in the technology center, and it reports to the chief technology officer. The technology center has the engineering expertise but not the convening authority, and the result is that the substitution projects are often technically sound but commercially unviable. The technology center can design the clinker substitution, but it cannot negotiate the standard revision with the industry association, and it cannot commit the firm to the offtake agreement. The capability is therefore weaker in cement than in steel, and the difference is organizational, not technological.

In the chemical firms, the function is often located in the international business unit, because the substitution is driven by the export market. The international business unit has the customer relationships and the market intelligence, but it does not have the engineering expertise, and the result is that the substitution projects are often commercially sound but technically unviable. The business unit can secure the offtake agreement, but it cannot design the electric cracker, and it cannot integrate the new process into the plant. The capability is therefore fragmented, and the fragmentation is a barrier to scaling.

These organizational differences are not accidental. They reflect the different histories of the three industries: steel was built by the state and retains a strategic planning function; cement was decentralized and is dominated by regional firms with weak technology centers; chemicals were opened to foreign investment and are dominated by joint ventures with strong international business units. The capability follows the organizational structure, and the structure follows the history.

7.5.3 The Role of the State in Capability Building

The substitution governance capability is not purely a firm-level attribute. It is co-produced by the firm and the state, and the co-production is the subject of this subsection.

The state contributes three things. The first is the policy credibility that the corridor requires. A firm cannot create the expectation that the carbon price will rise; only the state can. The state creates the expectation through legislation, through the ETS, through the CBAM negotiation, and through the industrial policy commitments. The expectation is not a guarantee, but it is a signal, and the signal is a precondition for the firm's investment.

The second is the infrastructure that the corridor requires. A firm cannot build a hydrogen pipeline or a CO2 storage site on its own; the infrastructure is a public good or a club good, and it is provided by the state or by a state-owned enterprise. The state's willingness to provide the infrastructure is a signal of its commitment to the corridor, and the signal reinforces the policy credibility.

The third is the coordination that the corridor requires. A firm cannot convene the other firms in the cluster, the local government, the banks, and the buyers without the state's convening power. The state provides the convening power through the industrial association, the development and reform commission, and the state-owned enterprise group. The convening power is not a subsidy; it is a governance function, and it is the function that the state performs best in the capital-intensive industries.

The co-production is not always harmonious. The firm and the state have different objectives, and the objectives conflict. The firm wants the state to commit to the infrastructure before the firm commits to the investment; the state wants the firm to commit to the investment before the state commits to the infrastructure. The conflict is resolved through a bargaining process that is mediated by the history of the relationship and the political weight of the firm. The firms that have the capability are those that have learned to manage the bargaining, and the management is a skill that is accumulated over time.

7.5.4 The Limits of the Capability

The substitution governance capability has limits, and it is important to be clear about them.

The first limit is that the capability cannot overcome the physics. A cement plant cannot eliminate its process emissions without clinker substitution or carbon capture, and no amount of governance capability can change the chemistry. The capability can accelerate the adoption of the substitute, but it cannot make the substitute exist.

The second limit is that the capability cannot overcome the economics if the gap is too large. If the cost of the substitute is three times the cost of the incumbent, no corridor can close the gap, because the demand commitment cannot be sustained at that premium. The capability can close a gap of 10 to 30 percent, but it cannot close a gap of 200 percent. The gap must be closed by technology cost reduction and by carbon pricing, and the capability is complementary to those, not a substitute for them.

The third limit is that the capability is specific to the firm and the cluster. It does not diffuse easily, because it is embodied in relationships and routines that are local. A firm that has the capability in one cluster cannot necessarily transfer it to another cluster, because the relationships and the routines are different. The diffusion of the capability therefore requires the diffusion of the cluster governance, and the diffusion is slow.

These limits are not reasons for pessimism. They are reasons for precision. The capability is a necessary condition for successful process substitution, but it is not a sufficient condition. The sufficient condition requires the capability plus the technology plus the economics plus the policy. The policy debate would be improved if it distinguished among these conditions, rather than treating them as a single problem.

7.6 The Tangshan Reline, Revisited

I want to return to the Tangshan blast furnace with which this chapter began, because the deputy chief engineer's remark deserves a fuller reading.

"The furnace will still be here in 2035," he said. "The question is whether the furnace is the problem or the steel is the problem."

The remark is a compressed statement of the substitution problem. The furnace is a physical asset with a campaign life that extends into the 2030s. It will still be operating unless it is retired early, and the retirement is not justified by the furnace's physical condition. The question is whether the emissions from the furnace are attributable to the furnace or to the steel that the furnace produces. If the emissions are attributable to the furnace, then the furnace is the problem, and the solution is to replace the furnace. If the emissions are attributable to the steel, then the steel is the problem, and the solution is to replace the steel—either by producing it with a different process or by using less of it.

The distinction matters because the two attributions imply different governance strategies. If the furnace is the problem, the strategy is to retire the furnace and replace it with an H-DR plant, and the governance problem is the corridor that I described in Section 7.4. If the steel is the problem, the strategy is to reduce the demand for steel, to extend the life of steel products, and to substitute other materials, and the governance problem is the demand-side transition that is beyond the scope of this chapter.

The deputy chief engineer's plant has chosen the first strategy, at least in part. It has signed a memorandum of understanding with a hydrogen supplier, it has joined a provincial consortium for the demonstration of H-DR, and it has begun to negotiate green steel offtake agreements with two downstream manufacturers. The furnace will still be here in 2035, but it will not be the only furnace. The plant is building a second pathway alongside the first, and the second pathway is the one that the corridor is designed to support.

The plant has not chosen the second strategy, and the reason is instructive. The demand-side transition would require the plant to shift its product mix toward higher-value steels that use less material per unit of output, and the shift would require a marketing capability that the plant does not have. The plant is a commodity producer, and its customers buy on price and specification. The plant can produce a green commodity, but it cannot produce a differentiated product. The capability that the plant lacks is the marketing capability, and the capability that it has is the substitution governance capability. The two capabilities are different, and the plant's strategy reflects the one it has.

This is a general pattern in the capital-intensive industries. The firms that have the substitution governance capability are often the firms that are least able to pursue the demand-side transition, because their organizational strengths are in process engineering and government relations, not in marketing and product development. The firms that have the marketing capability are often the firms that are least able to pursue the substitution, because their organizational strengths are in brand and customer relationships, not in process engineering. The two capabilities are not complementary; they are alternative, and the choice between them is a strategic choice that the firm must make.

7.7 Implications for the Capability-Based View

7.7.1 Capability as Coordination, Not Technology

The analysis in this chapter has an implication for the capability-based view of green competitiveness that I want to state explicitly. The capability that matters in capital-intensive industries is not the capability to develop a green technology. It is the capability to coordinate the adoption of a green technology across the actors who must commit to it. The distinction is important because the policy literature tends to focus on the technology, and the focus is misplaced.

The technology is necessary, but it is not sufficient. The H-DR technology exists; the question is whether the corridor can be assembled. The clinker substitution technology exists; the question is whether the standard can be revised. The electric cracker technology exists; the question is whether the renewable electricity and the offtake can be secured. In each case, the binding constraint is the coordination, not the technology.

The coordination capability is a firm-level attribute, but it is not a proprietary asset. It is a relational asset, and it resides in the firm's relationships with the state, the infrastructure providers, the buyers, and the financiers. The relational asset is not owned by the firm; it is co-owned by the network, and the firm's access to it depends on its position in the network. The position is accumulated over time, and it is not easily replicated. This is why the capability is a source of competitive advantage: it is rare, it is difficult to imitate, and it is embedded in the firm's history.

7.7.2 The Compliance Trap in Capital-Intensive Industries

The concept of the compliance trap, introduced in Chapter 1, applies to capital-intensive industries with a specific twist. In the labor-intensive industries, the trap is that the firm complies with the buyer's code of conduct and does not upgrade; the compliance is a substitute for the upgrading, and the firm remains at the bottom of the value chain. In the capital-intensive industries, the trap is that the firm complies with the emissions regulation and does not substitute; the compliance is a substitute for the substitution, and the firm remains locked into the incumbent process.

The mechanism of the trap is different in the two cases. In the labor-intensive case, the trap operates through the buyer's auditing, which rewards compliance and does not reward upgrading. In the capital-intensive case, the trap operates through the regulator's benchmarking, which rewards compliance with the incumbent technology and does not reward the substitution. The benchmark is set against the average emissions of the incumbent technology, and a firm that meets the benchmark is deemed compliant. The firm that invests in the substitute is not rewarded, because the substitute is not in the benchmark. The result is that the firm has no incentive to substitute, and the benchmark ratchets down only as the incumbent technology improves.

The trap is reinforced by the allocation methodology of the ETS. The free allocation is based on the benchmark, and the benchmark is based on the incumbent technology. A firm that substitutes is penalized, because it loses the free allocation without gaining a corresponding benefit. The penalty is a design flaw, and it is the subject of current policy debate. The fix is to set the benchmark against the best available technology, including the substitutes, and to auction the allowances rather than allocate them for free. The fix is politically difficult, because it raises the cost for the incumbent firms, and the incumbent firms are the ones that the policy is trying to change.

The compliance trap is therefore not a market failure; it is a governance failure. The governance system rewards the behavior that it is trying to change, and the reward is embedded in the benchmark and the allocation. The fix requires a change in the governance, and the change is contested. The firms that have the substitution governance capability are the ones that can navigate the change, because they have the relationships and the routines to adapt. The firms that lack the capability are the ones that are trapped, and the trap is self-reinforcing.

7.7.3 The Asymmetry of Capability Building

The capability is built asymmetrically across the capital-intensive industries, and the asymmetry has implications for the aggregate decarbonization trajectory.

The steel industry is the most advanced, because it has the largest firms, the strongest state involvement, and the most developed cluster governance. The cement industry is less advanced, because it has more firms, weaker state involvement, and less developed cluster governance. The chemical industry is the least advanced, because it has the most complex product mix, the most fragmented governance, and the most exposed export position.

The asymmetry is not a problem in itself; it is a fact that the policy must accommodate. The problem is that the policy treats the industries as if they were symmetrical, and the treatment produces a misallocation of the policy effort. The steel industry receives the most attention, because it is the largest emitter and the most visible. The cement industry receives less attention, because it is more dispersed and less visible. The chemical industry receives the least attention, because it is the most complex and the least understood. The result is that the industry with the most capability receives the most support, and the industry with the least capability receives the least support. The asymmetry is reinforced, not corrected.

The correction requires a differentiated policy that allocates the support according to the capability gap, not according to the emissions share. The cement industry needs the standard revision and the infrastructure for carbon capture; the chemical industry needs the renewable electricity and the offtake agreements; the steel industry needs the hydrogen and the demand commitment. The needs are different, and the policy instruments are different. The undifferentiated carbon price is not enough, and the undifferentiated subsidy is not enough. The policy must be designed around the specific corridor that each industry requires.

7.8 The International Dimension: CBAM and the Relocation of Process Substitution

7.8.1 The CBAM as a Corridor Externalizer

The EU's Carbon Border Adjustment Mechanism, which entered its transitional phase in October 2023 and will be fully operational in 2026, is the most important external influence on the substitution corridor in the Chinese capital-intensive industries. The mechanism imposes a carbon cost on imports of steel, cement, aluminum, fertilizers, electricity, and hydrogen, and the cost is based on the embedded emissions of the imported product. The mechanism is designed to prevent carbon leakage, and its effect on the Chinese industries is to externalize the corridor.

The externalization operates through the demand commitment. A Chinese steel exporter that sells to the EU must now report the embedded emissions of its product, and from 2026 it must pay the CBAM charge. The charge is based on the difference between the embedded emissions and the EU benchmark, and the benchmark is set against the best available technology. The exporter therefore has an incentive to reduce its embedded emissions, and the incentive is stronger than the domestic ETS incentive, because the CBAM charge is not offset by free allocation.

The incentive is transmitted to the substitution corridor through the offtake agreement. The EU buyer that imports the steel must pay the CBAM charge, and the buyer passes the charge back to the exporter through the price negotiation. The exporter that can supply low-emission steel avoids the charge, and the avoidance is a premium that the exporter can capture. The premium is the demand commitment that the corridor requires, and the commitment is credible because it is backed by the EU regulation.

The externalization is not uniform across the industries. The steel industry is the most exposed, because its exports to the EU are large and its embedded emissions are high. The cement industry is less exposed, because its exports to the EU are small and its product is bulky. The chemical industry is exposed in specific product categories (fertilizers, hydrogen), and the exposure is growing. The aluminum industry is exposed, and the exposure is complicated by the emissions from the electricity supply.

7.8.2 Relocation and the Race to the Cleanest Jurisdiction

The CBAM creates an incentive for the Chinese producers to relocate the low-emission stages of production to jurisdictions with lower carbon costs, or to jurisdictions that are exempt from the CBAM. The relocation is not a simple relocation of the plant; it is a relocation of the process substitution, and the relocation has implications for the Chinese industrial upgrading.

The first relocation pattern is the relocation of the downstream stages. A Chinese steel producer that exports to the EU can relocate the rolling and coating stages to a country that is exempt from the CBAM, and the relocation reduces the embedded emissions of the exported product, because the rolling and coating are less emissions-intensive than the ironmaking. The relocation is a form of tariff circumvention, and it is likely to be challenged by the EU. The challenge will be resolved through the rules of origin, and the rules will determine whether the relocation is viable.

The second relocation pattern is the relocation of the upstream stages. A Chinese steel producer can relocate the ironmaking to a country with cheap renewable electricity and a low carbon price, and the relocation reduces the embedded emissions of the DRI. The relocation is a form of corridor externalization, and it is more durable than the downstream relocation, because the upstream stage is the emissions-intensive stage. The relocation is already occurring: Chinese firms have announced DRI projects in Indonesia, Saudi Arabia, and Brazil, and the projects are designed to supply the EU market.

The third relocation pattern is the relocation of the entire plant. A Chinese producer can relocate the entire steelworks to a country with a low carbon price and a deep-water port, and the relocation is the most durable form of externalization. The relocation is expensive, and it is justified only if the carbon cost differential is large and persistent. The differential is large for the EU market, and it is likely to persist, because the CBAM is a permanent feature of the EU's climate policy. The relocation is therefore likely to continue, and it is likely to shift the geography of the global steel industry.

7.8.3 The Implication for Chinese Industrial Upgrading

The relocation has an ambiguous implication for Chinese industrial upgrading. On one hand, the relocation moves the emissions-intensive stages out of China, and the move reduces China's territorial emissions. On the other hand, the relocation moves the value-added stages out of China, and the move reduces China's industrial capability. The net effect depends on which stages are relocated and which are retained.

The firms that retain the high-value stages and relocate the low-value stages will upgrade. The firms that relocate the high-value stages and retain the low-value stages will downgrade. The difference is the substitution governance capability: the firms that have the capability can manage the relocation as a portfolio decision, and the firms that lack the capability will relocate reactively, stage by stage, and lose the high-value stages in the process.

The policy implication is that the Chinese government should support the capability, not the plant. The support should be directed at the demonstration, the infrastructure, and the demand commitment, and it should be conditional on the firm's commitment to retain the high-value stages in China. The condition is not a subsidy; it is a governance requirement, and it is the requirement that the successful corridors have imposed on the firms that participate in them.

7.9 The Cluster as the Unit of Substitution Governance

7.9.1 Why the Cluster Matters

The analysis in this chapter has repeatedly returned to the cluster as the unit of substitution governance. The reason is that the corridor requires a density of relationships that the individual firm cannot generate, and the density is a property of the cluster.

The cluster provides three things that the corridor requires. The first is the shared infrastructure. A hydrogen pipeline, a CO2 storage site, a renewable electricity connection, and a rail link are all club goods, and the club is the cluster. The club is organized by the local government and the state-owned enterprise group, and the organization is feasible because the members are geographically proximate and economically interdependent.

The second is the shared knowledge. The demonstration of a new process generates knowledge that is tacit and localized, and the knowledge diffuses through the cluster's labor market, its supplier relationships, and its informal networks. The diffusion is faster in the cluster than in the firm, because the cluster has a denser set of channels. The knowledge is not a public good; it is a club good, and the club is the cluster.

The third is the shared demand commitment. The green premium is more credible when it is committed by a group of buyers in the same cluster, because the group can coordinate its purchasing and can sanction the members who defect. The coordination is organized by the cluster's industry association, and the association has the authority to make the commitment binding. The commitment is not a contract; it is a convention, and the convention is enforced by the cluster's social structure.

The cluster is therefore the unit at which the substitution corridor is assembled, and the assembly is a governance function. The function is performed by the cluster's institutions: the local government, the industry association, the state-owned enterprise group, and the research institutes. The institutions are not interchangeable; each performs a specific role, and the roles are complementary. The local government provides the infrastructure and the policy credibility; the industry association provides the demand commitment and the standard revision; the state-owned enterprise group provides the convening power and the finance; the research institutes provide the demonstration and the knowledge. The configuration of the roles is the cluster's governance capability, and the capability is the subject of Chapter 9.

7.9.2 The Cluster and the Firm: A Division of Labor

The relationship between the cluster and the firm is a division of labor. The cluster provides the collective goods that the corridor requires, and the firm provides the specific investment that the corridor requires. The division is efficient when the collective goods are genuinely collective and the specific investment is genuinely specific.

The division breaks down when the collective goods are not provided, or when the specific investment is not made. The first breakdown occurs when the cluster's institutions are weak, and the infrastructure and the demand commitment are not organized. The second breakdown occurs when the firm's capability is weak, and the demonstration and the offtake are not secured. The two breakdowns are correlated, because the weak cluster does not attract the strong firm and the weak firm does not strengthen the cluster.

The policy implication is that the support should be directed at the division of labor, not at the firm or the cluster separately. The support should strengthen the cluster's institutions and the firm's capability at the same time, and it should be conditional on the two being linked. The linkage is the corridor, and the corridor is the object of the policy.

7.9.3 The Shunde Cluster and the Tangshan Cluster: A Comparison

The contrast between the Shunde furniture cluster and the Tangshan steel cluster illustrates the division of labor.

The Shunde cluster is a labor-intensive cluster that has upgraded to capital-intensive production. Its institutions are the local government, the industry association, and the supply chain platforms. The collective goods that the cluster provides are the design centers, the testing laboratories, the logistics platforms, and the export promotion services. The specific investment that the firm provides is the automated production line and the green certification. The division of labor is efficient, because the collective goods are genuinely collective and the specific investment is genuinely specific.

The Tangshan cluster is a capital-intensive cluster that is attempting to substitute its process. Its institutions are the local government, the state-owned enterprise group, the industry association, and the research institutes. The collective goods that the cluster provides are the hydrogen infrastructure, the CO2 storage site, the demonstration plant, and the demand commitment. The specific investment that the firm provides is the H-DR plant and the offtake agreement. The division of labor is less efficient than in Shunde, because the collective goods are more expensive and the specific investment is more specific. The cluster's institutions are stronger than Shunde's, but the corridor is longer and the coordination is more difficult.

The comparison suggests that the cluster's governance capability is not a single attribute; it is a function of the corridor's length. The longer the corridor, the stronger the institutions must be, and the more the firm must rely on the cluster. The Shunde cluster can afford weak institutions because the corridor is short; the Tangshan cluster cannot, because the corridor is long. The policy implication is that the support for the cluster should be proportional to the length of the corridor, and the length is a function of the technology and the infrastructure.

7.10 The Financial Dimension: Who Pays for the Substitution?

7.10.1 The Financing Gap

The substitution of a capital-intensive process is expensive, and the expense is not fully recoverable from the product premium. The gap must be financed, and the financing is the subject of this section.

The gap has three components. The first is the capital cost of the substitute plant, which is higher than the capital cost of the incumbent plant. The second is the operating cost of the substitute, which is higher than the operating cost of the incumbent, at least initially. The third is the stranded cost of the incumbent, which is the book value of the plant that is retired early. The three components are additive, and the sum is the financing requirement.

The financing requirement for the Chinese steel sector is large. A rough estimate: the replacement of 700 million tonnes of capacity at a capital cost of 1,500 yuan per tonne is 1.05 trillion yuan; the operating cost premium at 200 yuan per tonne for ten years is 1.4 trillion yuan; the stranded cost at 500 yuan per tonne is 350 billion yuan. The total is roughly 2.8 trillion yuan, or about 400 billion US dollars. The estimate is rough, but it indicates the scale.

The financing requirement cannot be met by the firms alone. The firms' internal cash flow is insufficient, and their access to debt is constrained by their existing leverage. The financing must come from a combination of sources: the firms' equity, the banks' loans, the capital markets' bonds, the government's fiscal support, and the international climate finance. The combination is the financing structure, and the structure is the subject of the policy debate.

7.10.2 The Role of the State-Owned Banks

The state-owned banks are the largest source of debt finance for the Chinese capital-intensive industries, and their role in the substitution is decisive.

The banks have the balance sheet to finance the substitution, but they have three constraints. The first is the capital adequacy requirement, which limits the amount of risk-weighted lending that they can do. The substitution loans are risk-weighted heavily, because the technologies are new and the borrowers are leveraged. The second is the non-performing loan requirement, which limits the amount of bad debt that they can recognize. The substitution loans are likely to become non-performing if the technology fails or the policy is reversed, and the banks are reluctant to recognize the risk. The third is the policy guidance, which directs the banks to lend to strategic sectors but does not specify the terms. The guidance is ambiguous, and the banks interpret it conservatively.

The constraints are not insurmountable, but they require a policy intervention. The intervention can take three forms. The first is the guarantee: the government guarantees the substitution loans, and the guarantee reduces the risk weight and the provisioning requirement. The second is the subsidy: the government subsidizes the interest rate, and the subsidy reduces the cost of the loan. The third is the securitization: the government creates a market for the substitution loans, and the market allows the banks to offload the risk. The three forms are complementary, and the combination is the financing structure that the successful corridors have used.

7.10.3 The Role of the Capital Markets

The capital markets are a smaller source of finance, but they are growing, and their role is different from the banks'.

The capital markets provide equity and bond finance, and they are more sensitive to the green premium than the banks. A firm that can demonstrate a credible green premium can issue a green bond at a lower yield, and the lower yield reduces the cost of the substitution. The green bond market in China has grown rapidly, and the growth is driven by the institutional investors who have green mandates.

The capital markets also provide a price signal. A firm that is exposed to the substitution risk has a lower valuation, and the lower valuation is a signal to the management and the investors. The signal is weaker in China than in the EU, because the Chinese capital markets are less efficient and the state ownership is more dominant. The signal is nevertheless present, and it is likely to strengthen as the CBAM and the ETS become more binding.

The capital markets' role is therefore complementary to the banks'. The banks provide the debt finance for the capital cost, and the capital markets provide the equity finance for the operating cost and the stranded cost. The division of labor is efficient when the capital markets are deep and the banks are patient. The division is inefficient when the capital markets are shallow and the banks are impatient, and the inefficiency is the current condition in the Chinese capital-intensive industries.

7.10.4 The International Climate Finance

The international climate finance is a small but growing source of finance for the substitution. The sources are the multilateral development banks, the bilateral aid agencies, and the private green funds. The finance is conditional on the emissions reduction, and the conditionality is a form of demand commitment.

The international finance is most relevant to the export-oriented clusters, because the finance is linked to the CBAM compliance. A Chinese steel producer that exports to the EU can access the international finance if it can demonstrate that the finance is used for the substitution, and the demonstration is verified by the international auditor. The finance is therefore a complement to the CBAM, and the complementarity is the corridor's international dimension.

The international finance is limited by three factors. The first is the scale: the international finance is small relative to the financing requirement. The second is the conditionality: the international finance is conditional on the governance reforms that the Chinese government is reluctant to accept. The third is the geopolitics: the international finance is affected by the tensions between China and the Western countries, and the tensions are a barrier to the finance.

The international finance is therefore a marginal source, and its role is to signal the international demand commitment rather than to finance the substitution. The signal is valuable, because it reinforces the domestic demand commitment, and the reinforcement is the corridor's international dimension.

7.11 The Governance of the Transition: Sequencing and Compensation

7.11.1 The Sequencing Problem

The substitution of a capital-intensive process must be sequenced, because the resources are limited and the corridor is long. The sequencing problem is to decide which plants to substitute first, and the decision is a governance problem.

The sequencing has three criteria. The first is the emissions reduction per unit of investment. The plants with the highest emissions intensity and the lowest substitution cost should be substituted first, because they generate the largest reduction per yuan. The second is the corridor readiness. The plants in the clusters with the strongest institutions and the most developed infrastructure should be substituted first, because they have the highest probability of success. The third is the distributional impact. The plants in the regions with the highest employment dependence should be substituted last, or substituted with the most compensation, because the social cost of the substitution is the highest.

The three criteria conflict. The plants with the highest emissions intensity are often the oldest and the least corridor-ready. The plants in the strongest clusters are often the newest and the least emissions-intensive. The plants in the most dependent regions are often the least able to afford the substitution. The conflict is resolved through a political process, and the process is the subject of the governance debate.

The Chinese sequencing has prioritized the corridor readiness over the emissions reduction. The demonstration projects are located in the strongest clusters, and the strongest clusters are the coastal provinces with the most developed infrastructure. The prioritization is defensible on the grounds that the demonstration must succeed before it can be scaled, and the success is more likely in the strong clusters. The prioritization is criticized on the grounds that it delays the reduction in the highest-emitting regions, and the delay is costly. The criticism is valid, and the balance is the subject of the policy debate.

7.11.2 The Compensation Problem

The substitution imposes costs on the workers, the communities, and the regions that depend on the incumbent plants. The costs must be compensated, and the compensation is a governance problem.

The compensation has three components. The first is the worker compensation: the workers who are displaced by the substitution must be retrained or pensioned, and the cost is borne by the firm and the government. The second is the community compensation: the communities that lose their tax base must be supported, and the support is borne by the government. The third is the regional compensation: the regions that lose their industrial base must be diversified, and the diversification is borne by the government and the private sector.

The compensation is expensive, and the expense is not included in the substitution cost. The omission is a bias in the policy analysis, because it understates the true cost of the substitution and overstates the benefit. The bias is corrected by including the compensation in the cost-benefit analysis, and the correction changes the sequencing: the plants in the dependent regions become less attractive, and the plants inthe diversified regions become more attractive.

The Chinese compensation has been partial. The worker compensation has been provided through the state-owned enterprise's internal retraining and pension programs, and the coverage has been uneven across provinces. The community compensation has been provided through the fiscal transfer system, and the transfers have been insufficient to replace the lost tax revenue. The regional compensation has been provided through the industrial relocation programs, and the programs have moved some of the displaced capacity to the inland provinces, where the carbon cost is lower and the employment dependence is higher. The relocation is a form of compensation, but it is also a form of leakage, because the emissions move with the capacity. The leakage is a problem for the territorial accounting, and it is a problem for the global accounting if the receiving region has a weaker carbon policy. The compensation and the leakage are therefore linked, and the link is the subject of the next subsection.

7.11.3 Compensation, Leakage, and the Territorial Trap

The compensation of the dependent regions through industrial relocation creates a territorial trap: the emissions are reduced in the coastal provinces and increased in the inland provinces, and the national total is unchanged. The trap is a governance failure, because the policy is designed to reduce the national total, and the design is undermined by the compensation.

The trap has two solutions. The first is the national carbon market, which prices the emissions uniformly across the provinces and removes the incentive to relocate. The market is being built, but its coverage is incomplete and its price is low, and the relocation is still profitable. The second is the differentiated regional policy, which sets the carbon cost according to the regional development level and the regional reduction potential. The policy is being discussed, but its design is contested, because the differentiation can be used to justify the delay in the inland provinces.

The trap is not unique to China. It is a general problem of the multi-level governance of the climate policy, and it is the subject of Chapter 9. The point for this chapter is that the compensation and the substitution are linked, and the link must be managed. A compensation policy that ignores the leakage will undermine the substitution, and a substitution policy that ignores the compensation will undermine the political support. The two must be designed together, and the design is the governance challenge.

7.12 The Technology Dimension: Learning Curves and the Cost of Substitution

7.12.1 The Learning Curve in Process Technologies

The cost of the substitute process declines with the cumulative production, and the decline is the learning curve. The learning curve is the mechanism that closes the cost gap between the substitute and the incumbent, and it is the mechanism that the substitution corridor is designed to activate.

The learning curve in the process technologies is different from the learning curve in the modular technologies. In the modular technologies (solar panels, batteries, electrolyzers), the learning is embodied in the component, and the cost declines as the production volume increases and the component is standardized. In the process technologies (H-DR, electric cracking, carbon capture), the learning is embodied in the plant, and the cost declines as the plant is replicated and the design is standardized. The replication is slower, because the plant is large and the design is site-specific, and the learning curve is therefore flatter.

The flatter learning curve has an implication for the policy. The policy cannot rely on the learning curve alone to close the cost gap, because the curve is too flat and the volume is too small. The policy must supplement the learning curve with the carbon price, the subsidy, and the demand commitment. The supplement is the corridor, and the corridor is the policy.

7.12.2 The Role of the Demonstration Plant

The demonstration plant is the first step in the learning curve, and its role is to generate the knowledge that the replication requires. The knowledge is of three kinds: the technical knowledge (does the process work?), the economic knowledge (what does it cost?), and the organizational knowledge (how is it operated and maintained?).

The technical knowledge is generated by the engineering, and it is codified in the design and the operating procedures. The economic knowledge is generated by the operation, and it is codified in the cost model. The organizational knowledge is generated by the workforce, and it is tacit. The three kinds of knowledge are complementary, and the demonstration plant must generate all three to be useful. A demonstration that generates the technical knowledge but not the economic and organizational knowledge is a scientific experiment, not a demonstration.

The Chinese demonstration plants have been strong on the technical knowledge and weaker on the economic and organizational knowledge. The plants are often built as showcases, and the operation is subsidized, so the economic knowledge is not generated. The workforce is often a hand-picked team, and the organizational knowledge is not representative. The result is that the demonstration does not generate the knowledge that the replication requires, and the replication is slower than it could be.

The fix is to operate the demonstration as a commercial plant, with commercial targets and a commercial workforce, and to publish the cost and the operating data. The fix is being adopted in some of the newer demonstrations, and the adoption is a sign that the policy is maturing.

7.12.3 The Replication Problem

The replication of the demonstration is the second step in the learning curve, and it is the step where the corridor is most likely to fail. The replication requires the capital, the infrastructure, the demand commitment, and the policy credibility, and the four requirements must be satisfied simultaneously. The simultaneity is the problem, because the four requirements are provided by different actors, and the actors do not coordinate automatically.

The replication problem is acute in the capital-intensive industries, because the plant is large and the investment is lumpy. A single replication can cost billions of yuan, and the decision is made by a small number of firms. The decision is therefore visible, and the visibility is a signal to the other firms. A successful replication encourages the others; a failed replication discourages them. The first replication is therefore the critical one, and the policy should focus on it.

The Chinese replication has been slow. The steel demonstrations are at the 1-million-tonne scale, and the full-scale replication is not yet under way. The cement demonstrations are at the 100,000-tonne scale, and the replication is not yet justified by the cost. The chemical demonstrations are at the pilot scale, and the replication is not yet designed. The slowness is a concern, because the 2030 target is approaching, and the replication must be completed before the target can be met.

The slowness has three causes. The first is the cost: the substitute is still more expensive than the incumbent, and the premium is not yet sufficient to justify the replication. The second is the infrastructure: the hydrogen and the CO2 storage are not yet available at the scale that the replication requires. The third is the policy: the carbon price is not yet high enough and the allocation is not yet tight enough to make the replication profitable. The three causes are the corridor's remaining gaps, and the gaps must be closed for the replication to proceed.

7.13 The Regional Dimension: Clusters, Provinces, and the Geography of Substitution

7.13.1 The Geography of the Capital-Intensive Industries

The Chinese capital-intensive industries are geographically concentrated, and the concentration is the legacy of the planned economy and the resource endowment. The steel industry is concentrated in Hebei, Jiangsu, Shandong, and Liaoning; the cement industry is concentrated in Anhui, Guangdong, Shandong, and Sichuan; the chemical industry is concentrated in Jiangsu, Zhejiang, Shandong, and Guangdong; the aluminum industry is concentrated in Shandong, Xinjiang, and Henan.

The concentration has an implication for the substitution. The substitution is more efficient in the clusters, because the infrastructure and the demand commitment are shared, and the sharing reduces the cost. The substitution is less efficient in the dispersed plants, because the infrastructure and the demand commitment must be provided individually, and the provision is expensive. The policy should therefore prioritize the clusters, and the prioritization is the regional dimension of the corridor.

The prioritization has a distributional consequence. The clusters are in the coastal provinces, and the coastal provinces are the richest. The dispersed plants are in the inland provinces, and the inland provinces are the poorest. The prioritization therefore directs the substitution investment to the rich provinces and leaves the poor provinces with the incumbent plants. The consequence is a spatial inequality in the substitution, and the inequality is a political problem.

The problem has two solutions. The first is the cluster-based substitution, which concentrates the investment in the coastal clusters and accepts the inland lag. The second is the dispersed substitution, which spreads the investment across the provinces and accepts the higher cost. The choice between the two is a political choice, and the choice is contested. The current policy leans toward the cluster-based substitution, and the lean is justified by the efficiency, but the justification is not sufficient, because the equity is also a policy objective.

7.13.2 The Provincial Governance Capability

The provincial governments are the key actors in the substitution, and their governance capability varies widely.

The coastal provinces (Guangdong, Jiangsu, Zhejiang, Shandong) have the strongest capability. They have the fiscal resources, the technical expertise, the industrial base, and the administrative capacity. They have developed the provincial carbon market pilots, the green finance platforms, and the industrial clusters. They have the relationships with the firms and the research institutes that the corridor requires. Their capability is the product of the reform era, and it is the reason why the demonstration projects are located in their jurisdictions.

The central provinces (Anhui, Hubei, Hunan, Henan) have the intermediate capability. They have the industrial base and the technical expertise, but they have less fiscal resources and less administrative capacity. They have the clusters, but the clusters are less developed. They have the political will, but the will is constrained by the fiscal dependence on the incumbent plants. Their capability is improving, and the improvement is the reason why some of the replication projects are located in their jurisdictions.

The western provinces (Xinjiang, Inner Mongolia, Gansu, Qinghai) have the weakest capability. They have the resource endowment (coal, renewable electricity, minerals), but they have the least fiscal resources and the least administrative capacity. They have the plants, but the plants are often the branch plants of the coastal firms, and the branch plants are managed from the coast. Their capability is the product of the resource extraction, and it is the reason why the substitution is the slowest in their jurisdictions.

The variation in the provincial capability is a barrier to the national substitution, because the corridor requires the provincial government's coordination, and the coordination is weakest where the need is greatest. The central government has attempted to address the variation through the fiscal transfers and the technical assistance, but the transfers and the assistance have been insufficient. The insufficiency is a governance gap, and the gap is the subject of Chapter 9.

7.13.3 The Cluster as a Provincial Strategy

The provinces have adopted the cluster as the unit of the substitution strategy, and the adoption is a significant institutional innovation.

The Guangdong strategy is the most developed. The province has designated the green petrochemical cluster in Huizhou, the green steel cluster in Zhanjiang, and the green building materials cluster in Qingyuan. Each cluster has a designated infrastructure plan, a designated demand commitment, and a designated policy package. The clusters are managed by the provincial development and reform commission, and the management is coordinated with the local governments and the state-owned enterprises. The strategy is the most advanced in the country, and it is the model that the other provinces are following.

The Shandong strategy is the most ambitious. The province has designated the green steel cluster in Rizhao, the green chemical cluster in Yantai, and the green aluminum cluster in Binzhou. The strategy is ambitious because the province is the largest emitter and the largest industrial base, and the substitution is the most difficult. The province has committed to a peak before 2025 and a neutrality before 2055, and the commitments are the most aggressive in the country. The strategy is being watched by the other provinces, because its success or failure will determine the national trajectory.

The Hebei strategy is the most challenging. The province is the largest steel producer, and the steel industry is the largest employer and the largest taxpayer. The substitution is the most difficult because the fiscal dependence is the highest and the alternative employment is the scarcest. The province has designated the green steel cluster in Tangshan, and the cluster is the site of the demonstration that I visited. The strategy is the most challenging because the compensation and the substitution must be managed together, and the management is the hardest in the country.

The three strategies illustrate the variation in the provincial capability, and the variation is the reason why the national substitution is uneven. The unevenness is not a failure; it is a fact that the policy must accommodate. The accommodation requires the differentiated support, and the differentiation is the subject of the next section.

7.14 Policy Implications: Designing the Substitution Corridor

7.14.1 The Principles

The analysis in this chapter has identified four principles for the design of the substitution corridor.

The first principle is that the corridor is a package, not an instrument. The carbon price, the subsidy, the infrastructure, the demand commitment, and the policy credibility must be aligned, and the alignment is the design problem. A policy that provides one instrument without the others will not induce the substitution, because the corridor requires all of them.

The second principle is that the corridor is cluster-specific. The infrastructure and the demand commitment are shared at the cluster level, and the sharing is the source of the efficiency. The policy should therefore be designed at the cluster level, and the cluster should be the unit of the support. The national policy sets the framework, and the provincial policy implements the framework in the cluster.

The third principle is that the corridor is time-consistent. The policy credibility is the precondition for the investment, and the credibility requires the commitment to the policy over the investment horizon. The commitment can be provided through the legislation, the long-term contracts, and the international agreements. The commitment is the hardest part of the design, because it depends on the political system's ability to bind its future self.

The fourth principle is that the corridor is compensated. The substitution imposes costs on the workers, the communities, and the regions, and the costs must be compensated. The compensation is not a side payment; it is a condition for the political sustainability of the substitution. The compensation should be designed together with the substitution, and the design should be transparent.

7.14.2 The Instruments

The principles translate into a set of instruments, and the instruments are the policy toolkit.

The carbon price is the core instrument, and its design should be tightened. The ETS should include the steel, cement, and chemical sectors by 2025, the allocation should be auctioned rather than free by 2030, and the price should be stabilized through the market stability reserve. The tightening is the signal that the corridor requires, and the signal is the precondition for the investment.

The subsidy is the complementary instrument, and its design should be targeted. The subsidy should support the demonstration, the infrastructure, and the early replication, and it should be phased out as the learning curve reduces the cost. The subsidy should be conditional on the firm's commitment to the corridor, and the conditionality should be verified.

The infrastructure is the collective instrument, and its design should be cluster-based. The hydrogen pipeline, the CO2 storage site, and the renewable electricity connection should be planned at the cluster level, and the planning should be coordinated with the industrial policy. The infrastructure should be financed by the state-owned enterprises and the green bonds, and the financing should be guaranteed by the government.

The demand commitment is the market instrument, and its design should be credible. The green public procurement should commit the government to buy a share of the green steel and the green cement, and the commitment should be backed by the budget. The CBAM should be engaged through the international negotiation, and the engagement should secure the recognition of the Chinese green certificates.

The compensation is the social instrument, and its design should be adequate. The worker retraining, the community support, and the regional diversification should be funded through the fiscal transfer and the industrial policy, and the funding should be proportional to the substitution's social cost.

7.14.3 The Sequencing

The instruments should be sequenced, and the sequencing is the implementation strategy.

The first phase (2024–2027) is the demonstration and the infrastructure phase. The priority is to complete the demonstration projects, to build the cluster infrastructure, and to establish the demand commitment. The carbon price should be tightened gradually, and the subsidy should be generous. The compensation should be prepared, and the social dialogue should begin.

The second phase (2028–2032) is the replication phase. The priority is to replicate the demonstration at commercial scale, to expand the infrastructure to the other clusters, and to scale the demand commitment. The carbon price should be tightened substantially, and the subsidy should be reduced. The compensation should be implemented, and the worker transition should be managed.

The third phase (2033–2037) is the consolidation phase. The priority is to complete the substitution in the coastal clusters, to extend it to the central clusters, and to manage the residual emissions in the western clusters. The carbon price should be at the level that makes the substitute competitive without the subsidy, and the subsidy should be phased out. The compensation should be completed, and the regional diversification should be under way.

The sequencing is indicative, and the actual timing will depend on the technology and the policy. The principle is that the phases should overlap, and the overlap should be managed. A phase that is too short will not generate the learning; a phase that is too long will not generate the urgency. The balance is the art of the implementation.

7.15 Conclusion: The Furnace and the Corridor

The Tangshan blast furnace will still be here in 2035, and the deputy chief engineer knows it. The question he posed—whether the furnace is the problem or the steel is the problem—is the question that this chapter has tried to answer.

The furnace is the problem in the sense that its emissions are embedded in its physical configuration, and the configuration cannot be changed without replacing the vessel. The steel is the problem in the sense that the demand for steel is the ultimate driver of the emissions, and the demand cannot be reduced without changing the consumption patterns. The two problems are linked, and the link is the substitution corridor. The corridor replaces the vessel, and the replacement changes the emissions profile of the steel. The demand-side transition reduces the volume, and the reduction changes the scale of the substitution. The two together are the decarbonization pathway, and neither alone is sufficient.

The corridor is a governance construct, and its construction is the capability that the capital-intensive firms must develop. The capability is not the technology; it is the coordination of the demonstration, the infrastructure, the demand commitment, and the policy credibility. The capability is built in the cluster, and the cluster is the unit of the substitution. The capability is co-produced by the firm and the state, and the co-production is the governance relationship that the corridor requires.

The Chinese capital-intensive industries are at the beginning of the corridor. The demonstration projects are under way, the infrastructure is being planned, the demand commitment is being negotiated, and the policy credibility is improving. The corridor is not yet complete, and the completion will take a decade. The completion is the condition for the 2060 neutrality target, and the condition is the test of the governance capability.

The test is not whether the firms can comply with the regulation. The test is whether they can climb the value chain by substituting the process. The firms that treat compliance as the ceiling will be trapped; the firms that treat it as the floor will climb. The difference is the capability, and the capability is the subject of this book.

The furnace will still be here in 2035, but it will not be alone. The corridor is being built alongside it, and the corridor is the future. The deputy chief engineer's laughter was not dismissive; it was the laughter of a man who knows that the furnace is not the problem, and that the problem is the corridor. The corridor is the work of the next decade, and the work has begun.