Chapter 4: The Green Premium as a Diagnostic Metric for Capital Allocation

A cement plant in Anhui province produces a tonne of clinker for about 320 yuan. Retrofitted with carbon capture and fed by green electricity, the same tonne costs roughly 760 yuan. The gap is 440 yuan per tonne, and it has a name that matters more in credit committees than most analysts admit: the green premium. It is the price of a decision not yet made.

Most discussions of green finance begin with money—how much is needed, where it will come from, who bears the risk. These questions are second-order. Before capital can be allocated, someone has to answer a prior one: what does "green" actually cost relative to the alternative? A bank that designates a loan as green without knowing the premium it finances is guessing. A fund that excludes carbon-intensive assets without measuring the premium of the substitutes it favors is performing a ritual.

The green premium is a number, or a family of numbers, with units, a measurement method, and known limitations. Measured properly, it tells a financier whether a technology sits close to commercial parity and how much policy support or concessional capital is required to close the gap. It also tells a research director where the marginal yuan of R&D buys the most abatement. This chapter defines the metric, distinguishes it from carbon pricing, and works through its consequences for credit decisions, portfolio construction, and the design of support instruments.

4.1 Concept and Measurement of Green Premium

4.1.1 A working definition

The green premium is the difference between the cost of producing a good or service with a zero-carbon (or near-zero-carbon) technology and the cost of producing the same good or service with the incumbent fossil-based technology. It is expressed per unit of output: per tonne of cement, per tonne of steel, per megawatt-hour of electricity, per tonne-kilometer of freight.

Three properties of the definition are easy to blur, and each changes the number.

The premium compares production costs; market prices are a separate variable. If a government subsidizes coal-fired electricity, the observed price gap between coal power and solar understates the true cost gap. The premium should be computed on a cost basis and then adjusted for explicit transfers, so that the analyst can see how much of the observed price gap is engineering and how much is fiscal.

The premium is technology-specific and sector-specific. There is no single green premium for the economy. A solar plant in Gansu faces a different premium from a cement kiln in Anhui. Aggregating them into one national figure is possible, but it destroys the information a capital allocator needs.

The premium is time-varying. Solar photovoltaic module costs fell roughly 90 percent between 2010 and 2023, and the green premium for solar electricity in most of China is now negative: new solar is cheaper than new coal on a levelized basis. That fact arrived recently and did not arrive evenly across geographies.

Let \(C_g\) be the levelized cost of the green technology per unit of output and \(C_b\) the levelized cost of the benchmark fossil technology. The green premium \(P\) is:

When \(P > 0\), the green technology is more expensive and requires support to be adopted at scale. When \(P \leq 0\), it is at or beyond parity and should be adopted on commercial grounds alone.

The levelized cost is a discounted cash flow per unit of output:

where \(I_t\) is capital expenditure in year \(t\), \(O_t\) is operating and maintenance cost, \(F_t\) is fuel or feedstock cost, \(Q_t\) is output, and \(r\) is the discount rate.

The choice of \(r\) can double the computed premium for a capital-intensive plant. A private developer using a 12 percent hurdle rate will compute a much larger premium for a nuclear plant or a carbon-capture retrofit than a state-owned utility using a 5 percent social discount rate. The same physical project yields two different premiums, and the difference is not a rounding error. For a capture retrofit with front-loaded capital and low operating cost, moving from 5 to 12 percent raises the levelized cost by roughly a third over a 20-year life; applied to the Anhui plant, that is the difference between a premium near 300 yuan per tonne and one near 440.

4.1.2 Why "premium" and not "cost gap"

The word is deliberate. In insurance, a premium is what you pay to transfer risk. In finance, a premium is what you pay for an asset that offers something extra. The green premium is what society pays, per unit of output, to avoid a unit of emissions. Framing it this way forces the analyst to ask: premium relative to what, paid by whom, for what benefit?

The framing has a direct consequence for capital allocation. Take the Anhui cement plant: a premium of 440 yuan per tonne, a social cost of carbon of 200 yuan per tonne, and process emissions of 0.6 tonnes of CO2 per tonne of cement. The social benefit of the green tonne is 120 yuan. The premium exceeds the benefit by 320 yuan. Someone must pay that 320 yuan, or the green tonne does not get produced. That "someone" is the policy designer's problem and the financier's.

Bill Gates popularized the term in How to Avoid a Climate Disaster (2021), arguing that the path to net zero runs through innovation that drives green premiums to zero. The formulation converts an abstract climate goal into sector-specific cost targets. Its weakness is that it treats the premium as a single number to be minimized. In practice the premium is a distribution with a mean, a variance, and a trajectory. A technology with a high but rapidly falling premium may be a better investment target than one with a low but stagnant premium.

4.1.3 Measurement in practice: four sectors

The measurement method differs by sector. Four cases show the range.

Electricity. The green premium for utility-scale solar or wind compares the levelized cost of energy (LCOE) of a new renewable plant with that of a new coal or gas plant, or with the marginal operating cost of an existing fossil plant. In China, the premium for new solar has been negative since roughly 2021 in most provinces. Offshore wind remains positive, at 0.1 to 0.3 yuan per kWh depending on location and water depth. Measurement is relatively clean: output is homogeneous, and the LCOE methodology is standardized.

Cement. Cement is harder. The premium depends on the decarbonization route—clinker substitution, alternative fuels, or carbon capture—and on the capture rate, the energy penalty of capture, the cost of CO2 transport and storage, and the system boundary. A plant capturing 90 percent of process emissions faces a different premium from one capturing 50 percent. The China International Capital Corporation (CICC) research report Carbon Neutrality: The Green Premium (2021) estimated the green premium for building materials, glass and cement included, at roughly 138 percent, meaning the green product costs about 2.4 times the conventional one. For a tonne of cement selling at 400 yuan, that implies a premium above 500 yuan. The Anhui figure of 440 yuan per tonne is my own calibration against plant-level cost data for a 1-million-tonne line with a 90 percent capture retrofit, and it sits at the low end of the CICC range.

Steel. Steel has two main green routes: electric arc furnace with green hydrogen-based direct reduced iron (DRI), and blast furnace with carbon capture. Published estimates for the green steel premium in China cluster between 1,000 and 2,000 yuan per tonne, against a conventional hot-rolled coil price near 4,000 yuan. The range is wide because the cost of green hydrogen is itself uncertain and location-dependent. A plant in Inner Mongolia with access to cheap renewable electricity faces a lower premium than one in Hebei relying on grid power.

Shipping. The green premium for shipping is expressed per tonne of CO2 avoided or per tonne-nautical-mile. Green methanol and green ammonia currently cost two to four times conventional bunker fuel. The premium per tonne of fuel is straightforward; the premium per unit of transport service requires assumptions about energy density and engine efficiency.

These cases share a structure. In each, the premium is the difference between two levelized costs, and in each the result is sensitive to the discount rate, the assumed utilization rate, and the cost boundary. The sensitivity is information. A project whose premium swings widely with the discount rate is one whose economics depend on the cost of capital, which means financial policy—guarantees, concessional loans, green bonds—can move the needle. A project whose premium is insensitive to the discount rate is one where only technology cost or carbon price will help.

4.1.4 The premium curve and the abatement cost curve

The green premium is closely related to the marginal abatement cost curve (MACC), used in climate policy since the early 2000s. The MACC plots the cost per tonne of CO2 abated against cumulative abatement potential, ordered from cheapest to most expensive. The green premium is roughly the MACC value for a specific technology, converted from per-tonne-of-CO2 to per-unit-of-output.

If a technology emits \(e_b\) tonnes of CO2 per unit of output in the baseline and \(e_g\) in the green case, abatement per unit is \(e_b - e_g\), and the abatement cost per tonne is:

The Anhui plant, with a premium of 440 yuan per tonne and abatement of 0.6 tonnes of CO2 per tonne of cement, has an abatement cost of about 733 yuan per tonne of CO2. That number can be set directly against the carbon price or the social cost of carbon.

The MACC and the green premium curve differ in one respect that matters for allocation. The MACC is usually built at the level of the economy or the sector, aggregating across technologies and assuming the cheapest abatement is pursued first. The green premium is built at the level of the individual project. Capital is allocated project by project, so the project-level number is the one a credit officer can use.

A second distinction: the MACC typically assumes a fixed baseline and a fixed green alternative. The green premium, properly measured, should account for the fact that the baseline itself is changing. A coal plant built today may be retired early; a cement plant built today may operate for 40 years. The premium should be computed over the life of the asset. A green premium that looks small today may look large tomorrow if the baseline becomes more expensive through carbon pricing, or if the green technology becomes cheaper through learning. A financier who computes a static premium and stops there will systematically underinvest in technologies with steep learning curves.

4.1.5 What the premium does not capture

Three limitations define the metric's proper use.

The green premium is a private cost metric. It excludes externalities. A coal plant's premium relative to solar does not include the health costs of particulate matter, the climate damages from CO2, or the water depletion from cooling. Were these included, the premium would be smaller, possibly negative. The premium measures the gap private capital faces, not the gap society faces. Confusing the two leads to underinvestment.

The premium is static in most applications. It compares two technologies at a point in time and does not capture the option value of waiting, the value of modularity, or the risk of technological lock-in. A technology with a slightly higher premium but a much steeper learning curve may be the better bet. The premium is a screening tool, not a decision rule.

The premium is technology-specific but not firm-specific. Two cement plants with the same technology may face different premiums because of differences in energy prices, logistics costs, and utilization rates. A sector-level premium is an average, and the dispersion around it can be large. For credit decisions, the firm-level premium is what matters, and it requires firm-level data.

flowchart TD
    A[Define output unit and system boundary] --> B[Estimate baseline cost C_b]
    A --> C[Estimate green cost C_g]
    B --> D[Choose discount rate r]
    C --> D
    D --> E[Compute levelized costs]
    E --> F[Premium P = C_g - C_b]
    F --> G{P greater than 0?}
    G -->|Yes| H["Compute abatement cost A = P / (e_b - e_g)"]
    G -->|No| I[Technology at or beyond parity]
    H --> J[Compare A with carbon price and social cost of carbon]
    I --> K[Commercial deployment without support]
    J --> L[Design support: subsidy, concessional capital, or carbon price]

4.2 Green Premium vs. Carbon Pricing

4.2.1 The EU ETS price swing

In 2017, a tonne of CO2 on the EU Emissions Trading System cost about 5 euros. By early 2023 it cost over 90 euros. A cement producer in Hebei watching that series would have seen a price that moved by a factor of eighteen in six years, driven by cap adjustments, the Market Stability Reserve, fuel-switching decisions, and speculation. The producer's cost of capture did not move by a factor of eighteen over the same period. It fell slowly, along its own learning curve.

That divergence is the reason the green premium and the carbon price should not be treated as substitutes. They answer different questions.

The carbon price answers: what does a tonne of CO2 cost today in a market where permits or taxes are traded? It is a market price, set by the interaction of supply (the cap or the tax rate) and demand (the abatement opportunities available to covered entities). It is observed, not computed.

The green premium answers: what does a unit of output cost to produce with the green technology, given the current cost structure? It is a computed cost gap, specific to a technology and a location.

A carbon price of 100 yuan per tonne does not tell a cement producer whether to invest in carbon capture. The producer needs the premium of capture relative to the carbon price, multiplied by the emissions intensity. If the premium implies an abatement cost of 733 yuan per tonne and the carbon price is 100 yuan, the producer will not invest, however high the carbon price is relative to some abstract benchmark. The relevant comparison is between the abatement cost derived from the premium and the carbon price.

A green premium of 440 yuan per tonne of cement does not tell a policymaker what the carbon price should be. The premium is a cost gap; the carbon price is an instrument. The policymaker needs to know how the premium responds to the carbon price, which depends on the elasticity of substitution, the availability of abatement technologies, and cost pass-through to product prices.

The two metrics are complementary. The carbon price is a signal; the green premium is a map. The signal tells firms that emissions are costly. The map tells them where the cheapest abatement is.

4.2.2 Why the premium is the better primary metric for allocation

Three properties of the premium make it the better primary diagnostic for capital allocation.

Granularity. A carbon price is economy-wide or sector-wide and applies uniformly to all covered entities, with limited differentiation. The green premium is technology-specific and location-specific. A bank lending to a steel plant in Hebei needs the premium for the specific decarbonization route available to that plant, not the national carbon price. The premium gives the bank a number it can put in a credit model; the carbon price gives it a signal for scenario analysis.

Stability. The EU ETS swing from 5 to 90 euros is the extreme case, but the Chinese national ETS has also been volatile in relative terms, trading in a range of roughly 50 to 80 yuan per tonne with thin volumes and occasional sharp moves. A capital allocator who relies on the carbon price as the primary metric is relying on a variable that can change by a factor of ten within a few years for reasons unrelated to technology costs. The green premium is anchored in engineering cost estimates, which change slowly. It moves when a plant's cost structure changes, not when a trading desk changes its view.

Actionability. A carbon price tells a firm how much it pays per tonne emitted. It does not tell the firm what to do. The green premium tells the firm which technology to adopt and what it will cost to adopt it. A firm facing a carbon price of 80 yuan per tonne and a green premium implying an abatement cost of 733 yuan per tonne has a clear answer: do not invest in the expensive abatement yet; wait for the premium to fall or the carbon price to rise. A firm facing the same carbon price and a premium implying an abatement cost of 60 yuan per tonne has a different answer: invest now.

4.2.3 The case against

The case against the premium as the primary metric is serious enough to state at length.

The premium is hard to measure. It requires engineering cost data that are often proprietary or unavailable. It requires assumptions about discount rates, utilization rates, and technology performance that are contested. It requires a system boundary that is arbitrary at the margins. A carbon price is a single number read off a screen.

The premium is static and partial. It compares two technologies at a point in time. It does not capture the full portfolio of abatement options, the interactions between them, or the general equilibrium effects of widespread adoption. A carbon price, in principle, captures all of these, because it is the outcome of a market that aggregates dispersed information.

The premium is not a policy instrument. A policymaker cannot set the premium; they can only influence it through subsidies, taxes, or R&D support. A carbon price can be set, adjusted, and enforced. The premium tells you where the gap is; the carbon price is one way to close it.

These arguments suggest a division of labor. The green premium is the diagnostic metric for capital allocation: it tells a financier where the gap is, how large it is, and how it is likely to evolve. The carbon price is the policy instrument for emissions reduction: it tells emitters that emissions are costly and lets them decide how to respond.

4.2.4 A comparison table

A policymaker who wants to reduce emissions in cement has two levers: raise the carbon price, which increases the cost of the baseline, or lower the green premium, which decreases the cost of the alternative. The first is a price instrument; the second is a technology instrument. The green premium tells the policymaker which lever is more effective, and by how much.

4.2.5 The interaction: carbon pricing changes the premium

A carbon price changes the green premium because it changes the cost of the baseline. If the baseline emits \(e_b\) tonnes of CO2 per unit and the green technology emits \(e_g\), a carbon price \(\tau\) adds \(\tau e_b\) to the baseline cost and \(\tau e_g\) to the green cost. The premium becomes:

The premium falls as the carbon price rises, at a rate equal to the abatement per unit of output. A carbon price of 100 yuan per tonne, applied to a cement plant with abatement of 0.6 tonnes per tonne, reduces the premium by 60 yuan per tonne. If the original premium was 440 yuan, the new premium is 380 yuan. The carbon price has done some work, not enough.

This interaction is the basis for the "carbon price plus" argument. A carbon price alone will not close the premium in hard-to-abate sectors, because the premium is too large relative to any politically feasible carbon price. Additional instruments—subsidies, concessional capital, public procurement—are needed to close the remaining gap. The green premium tells you how much additional support is required, and for how long.

The interaction also explains why carbon pricing has been more effective in some sectors than others. In power generation, where the premium for renewables is small or negative, a modest carbon price tips the balance. In cement and steel, where the premium is large, a carbon price at current levels is a rounding error. The sectoral difference reflects the underlying cost structure, not a failure of carbon pricing.

4.2.6 The social cost of carbon as a benchmark

A third metric is sometimes introduced: the social cost of carbon (SCC), the estimated monetary damage from an additional tonne of CO2. The SCC is used to evaluate whether a given abatement cost is worth incurring. If the abatement cost derived from the green premium is below the SCC, the abatement is worth doing on welfare grounds, regardless of the carbon price.

The SCC is controversial because it depends on assumptions about discount rates, climate sensitivity, and the valuation of future damages. The US Interagency Working Group's 2021 estimate put it at 51 dollars per tonne; the UK government's 2022 central case was 250 pounds per tonne. The range is wide enough to accommodate almost any policy conclusion, which limits its usefulness as a decision metric.

For capital allocation, the SCC is a useful ceiling: it tells the financier the maximum abatement cost that can be justified on welfare grounds. If the green premium implies an abatement cost above the SCC, the project is not worth doing at current technology costs, and the capital should go elsewhere, or into R&D to lower the premium. If the abatement cost is below the SCC, the project is a candidate for support, even if the carbon price is low.

The green premium, the carbon price, and the SCC form a triangle. The premium tells you the cost; the carbon price tells you the current market signal; the SCC tells you the welfare benchmark.

flowchart LR
    A[Green premium P] --> D[Abatement cost A]
    B[Emissions intensity e_b - e_g] --> D
    D --> E{A vs carbon price tau}
    D --> F{A vs social cost of carbon SCC}
    E -->|A less than tau| G[Invest now on commercial grounds]
    E -->|A greater than tau| H[Support needed or wait]
    F -->|A less than SCC| I[Welfare justifies support]
    F -->|A greater than SCC| J[Invest in R&D to lower P]

4.3 Implications for Financial Allocation

4.3.1 From premium to credit decision

The green premium enters a credit decision through the borrower's cost structure, the collateral value of the asset, and the policy support the borrower can expect.

Consider a bank lending to the Anhui cement producer for a carbon capture retrofit. The retrofit costs 400 million yuan and raises production cost by 440 yuan per tonne. The bank's credit model needs to know whether the borrower can service the additional debt.

The plant produces 1 million tonnes per year. At a premium of 440 yuan and a carbon price of 80 yuan, the net cost increase is 360 yuan per tonne, or 360 million yuan a year. Before the retrofit, a plant of this size with a typical EBITDA margin of 15 percent on 400 yuan per tonne generates about 60 million yuan of EBITDA. The retrofit would consume six times that. The debt service coverage ratio, on any plausible loan structure, falls below 1.0. The loan does not get approved without a subsidy or a long-term offtake agreement at a green premium price.

Put the numbers differently. A 400 million yuan loan at 4 percent over 10 years requires annual debt service of about 49 million yuan. The retrofit's net cost increase of 360 million yuan a year exceeds the plant's entire pre-retrofit EBITDA. The credit conclusion is not a judgment call; the arithmetic settles it. Only a subsidy covering most of the premium, or an offtake contract at 700 yuan per tonne or above, makes the DSCR work.

The example generalizes. The green premium is a parameter in a firm-level credit model, not an abstraction sitting above it. A bank that incorporates the premium into its credit assessment will make different decisions from one that does not. The difference determines whether a green project gets financed.

The second channel is collateral. A green asset—a capture facility, a green hydrogen plant, an offshore wind farm—may have a lower collateral value than a conventional asset, because the market for second-hand green assets is thin and technology risk is higher. The green premium proxies this risk: a high premium means the asset is far from commercial parity, which means its resale value is uncertain. A bank lending against a green asset with a high premium should apply a larger haircut, which raises the effective cost of capital and widens the premium. This is the green collateral trap: the premium raises the cost of capital, which raises the premium.

Policy can break the trap. A government guarantee on green loans reduces the collateral haircut, which lowers the cost of capital, which narrows the premium. The size of the guarantee needed is a function of the premium: the larger the premium, the larger the guarantee. The premium becomes a tool for sizing policy support.

4.3.2 Portfolio construction: the premium as a screening variable

For an equity investor or a fund manager, the green premium is a screening variable. It identifies which green technologies are close to parity and therefore investable on commercial terms, and which are far from parity and therefore dependent on policy or philanthropy.

A simple screening rule ranks technologies by abatement cost, the ratio of the green premium to abatement per unit, and allocates capital to those with the lowest ratio, adjusted for learning potential. A technology with a premium of 100 yuan per tonne and abatement of 1 tonne of CO2 per unit has an abatement cost of 100 yuan per tonne; a technology with a premium of 500 yuan and abatement of 0.5 tonnes has an abatement cost of 1,000 yuan per tonne. The first is the better bet at current carbon prices, unless the second has a steeper learning curve.

The rule ignores scale. A technology with a low abatement cost but a small market may be less attractive than one with a higher abatement cost and a large market. The premium should be combined with a market-size estimate to produce a premium-weighted abatement potential, the total abatement available at a given premium level. That is the area under the premium curve, not the height of a single point.

The rule also has a time dimension. A technology with a high premium today but a steep learning curve may be the better long-term bet. The learning rate—the percentage cost reduction per doubling of cumulative production—is the key parameter. Solar PV has a learning rate of roughly 20 percent; offshore wind, roughly 10 percent; carbon capture, roughly 5 to 10 percent. A technology with a 20 percent learning rate sees its premium halve with every four doublings of capacity. A technology with a 5 percent learning rate takes much longer. The premium trajectory, not just the premium level, should drive allocation.

4.3.3 Policy design: sizing support and exit

The green premium is a tool for policy design as well as capital allocation. It tells a government how much support is needed, for how long, and when to exit.

The "how much" question is answered by the premium itself. If the premium for green steel is 1,500 yuan per tonne and the carbon price is 100 yuan, the gap is 1,400 yuan per tonne. A subsidy of 1,400 yuan closes the gap; a subsidy of 700 yuan closes half, leaving the rest to the producer or the buyer. The premium tells the government the size of the check.

The "how long" question is answered by the premium trajectory. If the premium is falling at 10 percent per year, a subsidy covering the full premium today covers only 35 percent in ten years. The government can commit to a declining subsidy schedule, which gives investors certainty while limiting fiscal exposure. The schedule is derived from the learning curve, not from a political negotiation.

The "when to exit" question is answered by the parity condition. When the premium reaches zero, the subsidy can be withdrawn. The exit condition should be defined in terms of the premium, not a calendar date, because the premium may fall faster or slower than expected. A subsidy with a fixed end date may exit too early, if the premium is still positive, or too late, if the premium has turned negative.

The European Union's contracts for difference (CfD) for renewable energy are an example of a premium-contingent instrument. The CfD pays the difference between a strike price and the market price, which is essentially a subsidy equal to the premium when the market price is below the strike. The strike price is set through an auction, which reveals the premium investors require. As the premium falls, the strike price falls and the subsidy shrinks. The UK's CfD auctions saw strike prices for offshore wind fall from 119 pounds per MWh in 2015 to 37 pounds per MWh in 2022, a decline reflecting both technology cost reductions and competitive pressure.

China's approach has been different. The national green bond catalogue and the People's Bank of China's re-lending facility for carbon reduction provide subsidized capital rather than a price guarantee. The subsidy is implicit in the interest rate: a loan at 2 percent instead of 5 percent reduces the cost of capital, which reduces the premium. The size of the implicit subsidy depends on loan tenor and the share of the project financed by debt. For a capital-intensive project with a 15-year loan covering 70 percent of the cost, a 3 percentage point interest rate reduction can reduce the levelized cost by 15 to 20 percent, which may be enough to close a moderate premium.

The two approaches are not equivalent. A price guarantee transfers risk to the government; a subsidized loan transfers risk to the bank, unless the loan is guaranteed. The choice depends on the structure of the financial system and the government's risk appetite. The green premium is the common metric that allows comparison: both instruments should be sized to close the same gap.

4.3.4 The premium and the cost of capital

A green project with a high premium is, by definition, a project whose economics depend on factors outside the project: policy, carbon prices, technology learning. That dependence raises the risk premium investors demand, which raises the cost of capital, which raises the premium. The feedback loop is real and can be large.

Consider two projects with the same engineering cost. Project A has a premium of 50 yuan per tonne and a stable policy environment. Project B has a premium of 500 yuan per tonne and an uncertain policy environment. The cost of equity for A might be 8 percent; for B, 15 percent. The difference in the discount rate changes the levelized cost by 20 to 30 percent, which changes the premium by a comparable amount. The high-premium project is penalized twice: once by the engineering cost, once by the risk premium.

This is the green premium discount rate trap. It explains why some green technologies struggle to attract private capital even when their engineering costs are falling. The trap can be escaped through de-risking instruments: guarantees, offtake agreements, long-term contracts, and policy stability. Each reduces the risk premium, which reduces the cost of capital, which reduces the green premium. The premium is a financial parameter as much as a technology one, shaped by the institutions that govern risk.

The policy implication is that the cost of capital should be treated as a policy variable, not a market given. A government that wants to accelerate the green transition can do so by reducing the risk premium on green projects, through guarantees or stable policy frameworks, at a lower fiscal cost than direct subsidies. The green premium tells the government how much the risk premium matters: the higher the premium, the larger the benefit from de-risking.

4.3.5 The premium as a communication device

The green premium also functions as a communication device, and this has consequences for financial allocation.

It converts an abstract debate about climate policy into a concrete conversation about costs. A policymaker who says "we need to decarbonize cement" is making a statement that is easy to endorse and hard to operationalize. A policymaker who says "the green premium for cement is 440 yuan per tonne, and we propose to cover 300 yuan through a combination of carbon pricing and subsidies, leaving 140 yuan to be passed through to consumers" is making a statement that can be debated, costed, and implemented.

The communication function shapes investor expectations. An investor who understands the premium and the policy path to closing it can price the risk. An investor who faces only vague commitments will demand a higher risk premium, which raises the cost of capital and widens the premium. Clear communication about the premium and the policy path is a form of de-risking, with measurable effects on the cost of capital.

The premium also communicates across sectors. A steel producer and a cement producer face different premiums, but they can compare them and learn from each other's experience. A policymaker can compare the premium across sectors and allocate support to where it is most needed. A bank can compare the premium across borrowers and adjust its portfolio. The premium is a common language, and common languages reduce transaction costs.

4.3.6 A worked example: green steel in Hebei

Consider a green steel project in Hebei province using a hydrogen-based direct reduced iron (DRI) process, with green hydrogen produced from renewable electricity. The conventional route is a blast furnace-basic oxygen furnace (BF-BOF) with coal.

Conventional hot-rolled coil sells for about 4,000 yuan per tonne. The BF-BOF route emits about 2.0 tonnes of CO2 per tonne of steel. The hydrogen DRI route, with green hydrogen, emits about 0.3 tonnes, an abatement of 1.7 tonnes. Green steel costs about 5,500 yuan per tonne to produce, a premium of 1,500 yuan per tonne. The abatement cost is therefore 1,500 / 1.7 = 882 yuan per tonne of CO2.

The carbon price in China's national ETS is about 70 yuan per tonne, and steel is not yet fully covered. Even if it were, the carbon price would reduce the premium by 70 × 1.7 = 119 yuan, leaving a gap of 1,381 yuan per tonne. The social cost of carbon, at 200 yuan per tonne, would justify an abatement cost of up to 200 yuan per tonne, far below 882. On a strict welfare calculation, the project is not yet justified at current technology costs.

The calculation is static. The cost of green hydrogen is falling, from about 40 yuan per kg in 2020 to 25 yuan per kg in 2023, with projections of 15 yuan per kg by 2030 in favorable locations. The green premium for steel is on a downward trajectory. If the premium falls to 500 yuan per tonne by 2030, the abatement cost falls to 294 yuan per tonne, within the range of plausible carbon prices and social cost estimates.

The financial implication is that the project should be financed with a structure that bridges the gap until the premium falls. A combination of concessional debt at 2 percent instead of 6 percent, a carbon contract for difference guaranteeing a carbon price of 200 yuan per tonne, and an offtake agreement with a green steel buyer paying a premium of 500 yuan per tonne could close the gap. The green premium analysis tells the financier the size of each instrument and the horizon over which it is needed.

This is the practical value of the metric. It does not tell the financier whether the project is a good idea in the abstract. It tells them how much support the project needs, from which instruments, and for how long. Without the number, the negotiation over support is a negotiation over beliefs. With the number, it becomes a negotiation over parameters.

4.3.7 The premium across the portfolio: concentration and correlation

A bank or fund does not hold a single green project. It holds a portfolio, and the green premium matters at the portfolio level in a way that project-by-project analysis misses: it determines the correlation structure of returns.

Consider a portfolio of green steel, green cement, and green hydrogen projects. Each has a premium that depends on the price of green electricity, the price of carbon, and the cost of capital. If the price of green electricity rises, all three premiums rise together. If the carbon price rises, all three premiums fall together. The premiums are correlated, and the correlation is driven by common factors.

A portfolio of green projects with highly correlated premiums is riskier than a portfolio with independent premiums, even if the average premium is the same. The bank's capital charge should reflect this correlation. A naive approach that treats each project's premium as independent will understate the portfolio risk.

The correlation also affects the design of support instruments. A carbon price that rises with the electricity price would stabilize the premium, because the two effects offset. A carbon price fixed in nominal terms would not. Green premium analysis can be used to design hedges: a contract that pays when the electricity price rises, or a carbon contract for difference that pays when the carbon price falls. These hedges reduce the correlation of premiums across the portfolio, which reduces the capital charge.

The portfolio dimension is often neglected in policy discussions, which focus on individual technologies. For a financial institution, the portfolio is the unit of decision. The green premium is a project-level metric; its portfolio-level implications require an additional step, the correlation of premiums across projects. That step is where the metric becomes a tool for risk management, not just project selection.

4.3.8 The premium and the rating of technology firms

The book's later chapters develop a rating methodology for technology firms that incorporates environmental performance. The green premium is a natural input, because it measures the distance betweena firm's technology and commercial parity.

A technology firm with a product that has a zero or negative green premium is, by definition, commercially viable. Its environmental performance is aligned with its commercial performance. A firm with a product that has a high premium depends on policy support or on a future cost decline. Its environmental performance is not yet aligned with its commercial performance, and its credit rating should reflect the dependence.

The rating implication is not that high-premium firms are bad credits. It is that their credit quality is contingent on factors outside their control: policy, carbon prices, technology learning. A rating methodology that incorporates the premium should therefore distinguish between the level of the premium and the trend. A firm with a high but falling premium may be a better credit than one with a low but rising premium, because the trend determines the trajectory of support needs.

This is the bridge to Chapter 5. The green premium is the diagnostic; the rating methodology is the identification mechanism. A rating that incorporates the premium can distinguish between firms that are green because they are commercially viable and firms that are green because they are subsidized. The two types of firms have different risk profiles, and the distinction matters for capital allocation.

4.3.9 Objections and limits

Three objections deserve a response before the chapter closes.

The first is that the green premium is too uncertain to be useful. The uncertainty is real. Cost estimates for carbon capture vary by a factor of two or more; green hydrogen cost projections vary by a factor of three. But uncertainty is a reason to present the premium as a range, not a point estimate, and to update it as data arrive. A range of 300 to 600 yuan per tonne is more useful than a single number of 440, because it tells the financier the scale of the risk. Proceeding without a cost estimate is worse, because it substitutes implicit assumptions for explicit ones.

The second objection is that the premium is a private cost metric and therefore ignores the social benefits of decarbonization. This is correct, and it is why the premium should be compared with the social cost of carbon, not used in isolation. The premium tells you the private cost; the SCC tells you the social benefit. The gap between them is the case for public support. A financier who uses only the premium will underinvest in technologies with large social benefits and small private returns. A policymaker who uses only the SCC will overcommit to technologies that are not yet commercially viable.

The third objection is that the premium is static in a dynamic world. The response is that the premium should be computed as a trajectory, not a snapshot. The trajectory requires assumptions about learning rates, scale economies, and input prices. Those assumptions are uncertain, but they are explicit and can be tested against data. A premium trajectory updated annually is a living document. The static version is a first approximation; the trajectory is the operational version.

A fourth limit is worth stating plainly. The green premium is a metric for mitigation technologies, not for adaptation. It measures the cost of reducing emissions, not the cost of adapting to a changed climate. Adaptation has its own economics: the benefits are avoided damages, not avoided emissions, and the beneficiaries are often diffuse. The green premium framework does not apply directly. This is a limitation of scope, and it should be acknowledged.

4.3.10 What the premium cannot do

The green premium cannot tell a government whether to prioritize mitigation or adaptation. It cannot tell a bank whether to lend to a green project or to a health clinic. It cannot tell a society how much to spend on climate policy relative to other public goods. These are questions of value, not of cost.

What the premium can do is narrow the range of disagreement. A debate about whether to decarbonize cement is often a debate about costs in disguise. If the participants agree on the premium, the debate shifts to the willingness to pay. If they disagree on the premium, the debate is empirical and can be resolved with data. The premium does not settle the normative question. It clarifies the empirical one, and that is the whole of its contribution.

4.3.11 The premium as a bridge between chapters

The book's argument is that financial support for the green transition requires an integrated mechanism: rating-based identification, insurance-based risk buffering, and sci-tech finance catalysis. The green premium is the metric that connects the three.

For rating-based identification, the premium measures the distance between a firm's technology and commercial parity, a key input to the credit assessment. For insurance-based risk buffering, the premium measures the cost gap that insurance can help bridge, by stabilizing revenues or covering catastrophic losses. For sci-tech finance catalysis, the premium measures the return on R&D: a technology with a high premium and a steep learning curve offers a large payoff to innovation, because a small reduction in the premium translates into a large increase in addressable market.

Without the premium, the rating methodology has no anchor, the insurance design has no target, and the catalysis has no metric of success. With it, each component of the chain has a number to work with.

The next chapter takes up the first component: how to incorporate environmental performance, including the green premium, into the credit rating of technology firms. The premium provides the link between the environmental dimension and the financial dimension. A firm whose technology has a high premium is a firm whose environmental performance is expensive; a firm whose technology has a low premium is a firm whose environmental performance is cheap. The rating should reflect the difference, because the credit risk reflects the difference.

4.3.12 A note on data and method

The premium estimates in this chapter come from a mix of sources: engineering cost studies, industry benchmarks, and my own calculations. The CICC figures for building materials and non-passenger transport, from the 2021 report Carbon Neutrality: The Green Premium, are sector averages, not project-specific numbers. The cement premium of 440 yuan per tonne is an illustrative figure, calibrated to the CICC range and to plant-level cost data from Anhui. The steel premium of 1,500 yuan per tonne sits at the upper end of published estimates, reflecting a green hydrogen price of about 30 yuan per kg; at 20 yuan per kg, the premium falls to roughly 1,000 yuan per tonne.

The discount rate used in the levelized cost calculations is 8 percent for private projects and 5 percent for projects with public support. The choice is arbitrary at the margin, but it is stated explicitly so that readers can adjust it. For a capital-intensive project with a 20-year life, a 1 percentage point change in the discount rate moves the levelized cost by roughly 0.3 percent, so a 3 percentage point difference changes the premium by about 1 percent of levelized cost. That is small next to the uncertainty in the engineering cost estimates, which is why the discount rate debate, however heated, rarely decides a project on its own.

The premium estimates should be treated as order-of-magnitude figures. Their value lies in the comparison across sectors and the trajectory over time, not in the decimal places. A financier who uses the premium to rank technologies will make better decisions than one who uses it to justify a predetermined conclusion.

4.3.13 The premium and the cost of inaction

The green premium is usually framed as the cost of acting. It is equally a measure of the cost of not acting, because the baseline technology is not free. A coal plant that is not retrofitted continues to emit, and those emissions carry a cost through carbon pricing, litigation, and stranded asset risk. The premium is the difference between two costs, and the baseline cost is rising as carbon pricing expands and as public opinion shifts.

This reframing changes the sign of the comparison. A green premium of 440 yuan per tonne looks large against a carbon price of 70 yuan. But if the baseline cost is rising at 10 percent a year and the green cost is falling at 5 percent a year, the premium closes within a decade even without additional policy. A financier who computes the premium at a single point in time will miss this. A financier who computes the trajectory will see that the cost of inaction is rising, and that the green project is a hedge against that rise.

A high premium means the baseline is cheap today. It does not mean the baseline will be cheap tomorrow.

flowchart TD
    A[Green premium diagnostic] --> B[Rating-based identification]
    A --> C[Insurance-based buffering]
    A --> D[Sci-tech finance catalysis]
    B --> E[Credit assessment of technology firms]
    C --> F[Risk mitigation for eco-agriculture and forestry]
    D --> G[Capital for green innovation]
    E --> H[Integrated mechanism]
    F --> H
    G --> H
    H --> I[Capital allocation to green transition]

The figure summarizes the logic. The premium is the entry point. It feeds the three components of the chain, which together determine the allocation of capital. The chain is not a sequence; it is a set of parallel channels that share a common diagnostic. The premium is what makes the channels comparable, because it measures the same thing—the cost gap—across different technologies and different financial instruments.

One caution on the politics of the metric. The green premium is attractive to policymakers because it appears technical. It is a number, and numbers are easier to defend than judgments. But the premium is not politically neutral. The choice of discount rate, the boundary of the cost accounting, and the treatment of subsidies all embed value judgments. A premium computed with a 5 percent discount rate will justify more public support than one computed with a 12 percent rate. A premium that excludes the health costs of air pollution will understate the case for decarbonization; one that includes them will overstate it relative to a narrow cost-of-production definition.

The response is not to abandon the metric but to be explicit about its assumptions. A premium estimate should come with a statement of the discount rate, the system boundary, and the treatment of externalities. A reader who disagrees with the assumptions can recompute the number. That is how a technical metric earns its place in a political debate: by being transparent, not by being neutral.

The green premium answers a simple question: how much more does it cost to do the right thing? The answer differs by sector and location, and it moves. But it is knowable, at least approximately. The next chapter puts the number to work, as an input to the credit rating of technology firms—the first link in the identification-buffering-catalysis chain.