Chapter 2: The Financial Ecosystem of Zero-Carbon Transition

Green credit identifies, and it identifies well. It does not buffer risk, and it does not catalyze. Chapter 1 left the argument at that point, with a single sentence that deserves to be unpacked before anything else in this book proceeds: the missing functions are missing markets, and the missing markets are missing links between institutions that already exist.

That sentence is the thesis of this chapter. A zero-carbon transition is not a credit problem. It is a portfolio problem distributed across institutions with different mandates, different time horizons, and different tolerances for loss. Banks hold the identification machinery. Insurers hold the buffering machinery. Capital markets hold the risk-pricing machinery. Funds hold the patience. None of them, alone, can carry a hard green project from laboratory to commercial operation. The ecosystem, not the institution, is the unit of analysis.

What follows is a map of that ecosystem, drawn as a functional description of which institution does what and where the links between them break. A catalogue of instruments would be easier to write and mostly useless. The organizing question is functional: for each institution, what does it identify, what does it buffer, what does it catalyze? Where an institution performs one function well and the others poorly, that is a link to be built. Where it performs none, that is a market to be created.

2.1 From Banks to Capital Markets

2.1.1 The arithmetic of bank balance sheets

Start with a number that is often quoted and rarely examined. The International Energy Agency's 2021 estimate of cumulative investment needed for net-zero emissions by 2050 was roughly USD 130 trillion—about USD 4 trillion a year, sustained for three decades. Some of that will come from public budgets, some from households, most from private capital. The question is which private institution can hold what.

A commercial bank cannot hold a thirty-year pre-revenue technology bet on its balance sheet. The constraint is not appetite; it is the structure of bank funding. Deposits are short-term and redeemable. Loans are long-term and illiquid. The maturity transformation that makes banking profitable also makes it fragile, and regulators have spent a century building rules to limit how much of it a bank can do. Basel III's liquidity coverage ratio, the net stable funding ratio, and the leverage ratio all push in the same direction: banks must hold liquid assets against short-term liabilities, and must fund long-term assets with stable capital. A thirty-year green hydrogen project is the definition of an asset that does not fit.

This is not a criticism of banks. The identification function that green credit performs is genuinely valuable, and Chapter 1 argued that it is undervalued. The point is narrower. The banking system can identify a bankable green project and lend to it. It cannot create a project that is not yet bankable, and it cannot absorb the loss if the technology fails. Those are different functions, and they require different institutions.

The arithmetic is stark. Suppose a green steel plant requires USD 2 billion in capital expenditure, has a ten-year construction and ramp-up period, and a 40 percent probability of technical success at commercial scale. A bank lending against this project at investment grade would need to price the expected loss into the spread. With a 40 percent failure probability and a 60 percent recovery rate, expected loss is 16 percent of exposure—far above what any investment-grade loan book can absorb. The project is not bankable at any spread a borrower would accept. It becomes bankable only when the risk is restructured: equity absorbs the first loss, insurance or guarantees absorb specific hazards, and debt sits senior to both. That restructuring is the work of the ecosystem.

2.1.2 Equity: the first-loss absorber

Equity is the natural home for technology and horizon risk. An equity investor accepts the possibility of total loss in exchange for upside. A green venture fund that holds twenty early-stage companies expects most to fail and a few to return multiples. That is a different loss model from a bank's, and it is the correct one for pre-commercial green technology.

The scale of green equity is growing but remains small relative to need. Morningstar data cited by Hale (2021) put US sustainable fund flows at USD 51 billion in 2020, up from USD 5.4 billion in 2018—a ninefold increase in two years. Global sustainable fund assets passed USD 2 trillion in 2021, according to Morningstar's quarterly sustainable fund flows report. These numbers are large in absolute terms and small relative to the USD 4 trillion annual investment gap. More importantly, most sustainable fund assets sit in listed equities, not in the early-stage private companies where green technology is actually developed. A fund that buys shares in a wind turbine manufacturer is not financing the next wind turbine design.

The gap between listed sustainable equity and early-stage green venture capital is one of the ecosystem's structural weaknesses. Listed markets price existing assets; venture capital creates new ones. A zero-carbon transition that relies on listed sustainable funds will finance the deployment of mature technologies and will underfinance the development of immature ones. The institutions that fill this gap—government-guided funds, corporate venture arms, specialized green private equity—are the subject of Chapter 11. Here the point is structural: equity is the correct instrument for first-loss absorption, and the equity available for early-stage green technology is thinner than the equity available for listed green exposure.

2.1.3 Bonds: the middle of the risk spectrum

Bonds sit between bank loans and equity on the risk spectrum. A green bond is a debt instrument whose proceeds are earmarked for environmental projects, and the market has grown from USD 11 billion in 2013 to over USD 500 billion in annual issuance by 2021, according to Climate Bonds Initiative data. The European Investment Bank issued the first green bond in 2007, and the market has since developed a set of voluntary standards—the Green Bond Principles, published by the International Capital Market Association (ICMA) in 2014—that define what counts as green.

The green bond market performs a useful function: it allows institutional investors with long-duration liabilities, such as pension funds and insurance companies, to hold green assets that match their horizons. A twenty-year green bond financing a solar farm is a better duration match for a pension fund than a twenty-year government bond, and it carries a green label that satisfies the fund's mandate. The market has grown because it solves a real matching problem.

It also has limits. Green bonds finance projects that are already bankable at investment grade. The issuer must be able to service the debt from project cash flows, which means the project must have a revenue stream and a technology that works. Green bonds do not finance pre-commercial technology, and they do not absorb technology risk. They are, in functional terms, a way of moving bankable green projects off bank balance sheets and into institutional portfolios. That is valuable—it frees bank capital for other lending and it gives pension funds a green asset—but it is not catalytic.

The transition bond, a newer instrument, extends the bond market to companies that are not yet green but are committed to a transition path. The first transition bond was issued by the Italian energy company Enel in 2019, and the market has since grown in Japan and Southeast Asia. Transition bonds address a real problem: most high-carbon companies cannot issue green bonds because their current operations are not green, and excluding them from sustainable finance leaves their emissions unaddressed. The instrument is controversial—critics argue that it allows companies to claim credit for commitments they may not meet—but the underlying logic is sound. A transition bond with credible, verifiable milestones is a way of financing the decarbonization of existing assets, which is where most near-term emissions reductions must come from.

2.1.4 Alternative finance: the channels that do not fit the categories

Between bank lending, equity, and bonds, a set of alternative channels has grown up. Each addresses a specific gap.

Private debt funds lend directly to mid-sized companies that are too large for bank relationship lending and too small for public bond markets. In green finance, private debt funds have financed rooftop solar, energy efficiency retrofits, and small-scale renewable projects that fall below the threshold of public markets. The funds raise capital from institutional investors and deploy it through direct loans, often with covenants tied to environmental performance. The market is small relative to bank lending but growing; Preqin estimated global private debt assets at USD 1.2 trillion in 2021, with a green share that is difficult to measure but clearly rising.

Infrastructure funds hold operating assets—wind farms, solar plants, transmission lines—and earn returns from stable cash flows. They are the natural buyers of projects once construction is complete and technology risk has resolved. The infrastructure fund market has become the exit route for green project developers: a developer builds a wind farm, proves the technology and the cash flow, and sells it to an infrastructure fund that holds it for twenty years. This division of labor—developer bears construction risk, fund bears operating risk—is one of the ecosystem's working links, and it is worth studying as a model for other green sectors.

Yieldcos, a structure popularized in the 2010s, are listed vehicles that hold operating renewable assets and distribute most of their cash flow to shareholders. The first yieldco, NRG Yield (now Clearway Energy), was spun out of NRG Energy in 2013, and several others followed. The structure solves a specific problem: it gives retail and institutional investors exposure to operating renewable assets with a predictable yield, and it gives developers a lower-cost exit than a private sale. The yieldco boom of 2013–2015 was followed by a bust as rising interest rates compressed valuations, and the structure has since become less common. The episode is instructive: a financial structure that works in a low-rate environment can fail when rates rise, and green finance is not immune to the interest-rate cycle.

Securitization, the pooling of loans into tradable securities, has been applied to green assets with mixed results. Green securitization of residential solar loans, energy efficiency loans, and property assessed clean energy (PACE) assessments has grown, allowing banks to originate green loans and sell them to institutional investors. The market remains small—measured in tens of billions of dollars globally—and it depends on standardized loan documentation and reliable performance data, both of which are still being developed. The 2008 financial crisis cast a long shadow over securitization, and green securitization has grown cautiously as a result.

The common thread across these alternative channels is that each performs a specific function that neither banks nor public markets perform well. Private debt finances the mid-sized project. Infrastructure funds hold the operating asset. Yieldcos distribute the cash flow. Securitization moves the loan off the balance sheet. None of them is a substitute for the others, and none of them alone can carry a green project from conception to operation. The ecosystem works when they are linked in sequence.

2.1.5 The functional map

Figure 2-1 lays out the sequence. A green project moves through stages, and each stage requires a different institution. The developer and venture capital finance the early stage. Private equity and project finance carry construction. Infrastructure funds and green bonds hold the operating asset. The links between stages are where the ecosystem succeeds or fails.

flowchart TD
    A[Early-stage R&D] --> B[Pilot and demonstration]
    B --> C[Commercial construction]
    C --> D[Operating asset]
    D --> E[Refinancing and exit]
    A -.->|Venture capital,<br/>government grants| A
    B -.->|Private equity,<br/>project finance| B
    C -.->|Bank debt,<br/>green bonds| C
    D -.->|Infrastructure funds,<br/>yieldcos| D
    E -.->|Green securitization,<br/>listed equity| E
    style A fill:#e8f4e8,stroke:#2d6a2d
    style B fill:#e8f4e8,stroke:#2d6a2d
    style C fill:#fff4e8,stroke:#8a5a2d
    style D fill:#e8eef4,stroke:#2d4a6a
    style E fill:#e8eef4,stroke:#2d4a6a

The figure makes the gap visible. The early-stage box is thin: venture capital for hard green technology is scarce, and government grants are limited. The construction box is where bank debt and green bonds enter, but only for projects that have passed the bankability threshold. The operating box is well served by infrastructure funds and yieldcos. The ecosystem is thick at the operating end and thin at the early end. That asymmetry is the subject of Chapters 10 and 11.

2.2 Intermediaries and Specialized Institutions

2.2.1 Rating agencies: the identification infrastructure

A rating agency does one thing: it translates complex credit information into a letter grade that investors can use to compare risks across issuers and instruments. That translation is the foundation of capital market function. Without it, a pension fund cannot compare a green bond from a Portuguese utility with a green bond from a Japanese manufacturer, and the market fragments into bilateral relationships.

The rating process is more mechanical than it appears. S&P Global Ratings, Moody's, and Fitch each use a two-step methodology: first, they assess the stand-alone credit profile of the issuer or instrument; second, they adjust for external support, such as government guarantees or parental support. The output is a rating that maps to a probability of default, though the agencies are careful not to state the mapping explicitly. S&P's 2020 annual study, published in April 2021, reported global corporate average cumulative default rates by rating from 1981 to 2020. The investment-grade versus speculative-grade split is stark: investment-grade issuers default at a fraction of the rate of speculative-grade issuers, and the gap widens over longer horizons.

The recovery rate is the second input. When a borrower defaults, creditors recover some fraction of their exposure, and the recovery depends on the seniority of the claim, the assets available, and the legal structure. A senior secured creditor recovers more than a subordinated unsecured creditor. The expected loss of a transaction is the product of three terms:

where PD is the probability of default, LGD is the loss given default (one minus the recovery rate), and EAD is the exposure at default. This formula, the foundation of credit risk measurement, is the subject of Chapter 3. Here the point is institutional: the rating agencies supply the PD and LGD estimates that make the formula operational, and without them the capital markets cannot price green risk.

The rating agencies have been slow to incorporate environmental factors into their methodologies. S&P, Moody's, and Fitch all publish ESG assessments alongside credit ratings, but the assessments are separate from the rating and do not mechanically affect it. The agencies argue that environmental factors matter only insofar as they affect creditworthiness, and that a separate ESG score would confuse investors. Critics argue that the separation allows environmental risk to be underpriced. The debate is unresolved, and Chapter 4 takes it up in detail. For the ecosystem map, the relevant fact is that the identification infrastructure exists and is beginning to incorporate environmental data, but the incorporation is partial.

2.2.2 Insurers: the buffering infrastructure

Insurance does something banks cannot: it transfers a defined risk to a counterparty that holds a diversified portfolio of similar risks. A bank that lends to a thousand farmers faces correlated weather risk; a single drought can impair the entire portfolio. An insurer that underwrites weather risk across a continent holds a portfolio in which droughts in one region are offset by floods in another. The insurer's diversification is the source of the buffering function.

Weather index insurance is the clearest example. Instead of indemnifying a farmer for actual crop loss—which requires costly loss assessment and creates moral hazard—the policy pays out when a weather index (rainfall, temperature, wind speed) crosses a threshold. The farmer receives a payout based on the index, not on the observed loss. The design eliminates the assessment cost and the moral hazard, at the price of basis risk: the index may trigger a payout when the farmer has no loss, or fail to trigger when the farmer has a loss. Chapter 7 examines the basis-risk problem in detail.

The scale of weather index insurance is modest but growing. The World Bank's Global Index Insurance Facility has supported programs in over thirty countries, and India's national crop insurance scheme, Pradhan Mantri Fasal Bima Yojana, covers tens of millions of farmers. The programs have mixed records. Some have reduced farmer vulnerability; others have suffered from high basis risk, low uptake, and fiscal strain. The lesson for the ecosystem is that insurance can buffer weather risk, but only when the index is well-designed and the subsidy structure is sustainable.

Forest insurance is a newer instrument. A forest is a long-duration asset with multiple risks: fire, pest, disease, storm, and drought. The insurance market for forests is thin because the risks are correlated and the loss assessment is difficult. Chapter 8 examines the economics of forest insurance and the role of public reinsurance in making the market viable. The ecosystem point is that forest insurance is a buffering instrument that does not yet exist at scale, and its absence is a missing link in the green finance chain.

Credit insurance and guarantees are a third category. A guarantee from a public institution—a development bank, an export credit agency, a national guarantee fund—can lift a green project from speculative to investment grade by transferring a defined risk to the guarantor. The European Investment Bank's guarantee programs, the World Bank's Multilateral Investment Guarantee Agency (MIGA), and national green banks all perform this function. The guarantee does not eliminate risk; it reallocates it to an institution with a different risk tolerance and a different mandate. The guarantee reallocates risk to an institution with a different mandate.

2.2.3 Funds: the patient capital

A fund pools capital from many investors and deploys it according to a mandate. In green finance, the relevant funds fall into three categories.

Green private equity and venture capital funds invest in early-stage and growth-stage companies. The market is small. The International Energy Agency estimated that venture capital investment in clean energy technology was roughly USD 5 billion a year in the late 2010s, a fraction of the USD 4 trillion annual need. The funds that do invest—Breakthrough Energy Ventures, founded by Bill Gates in 2015 with over USD 2 billion under management, is the most prominent—concentrate on technologies with large potential impact and long development horizons. The fund model works for early-stage green technology because it allows the fund manager to hold a portfolio of bets and to accept that most will fail.

Infrastructure funds invest in operating assets and earn returns from stable cash flows. The market is large: infrastructure fund assets under management passed USD 1 trillion in the early 2020s, with a significant and growing green share. The funds are the natural holders of operating renewable assets, and they provide the exit that allows developers to recycle capital into new projects. The infrastructure fund model works for green assets because the assets have predictable cash flows once built, and the fund's holding period matches the asset's life.

Sovereign wealth funds and pension funds are the ultimate sources of long-duration capital. Norway's Government Pension Fund Global, the largest sovereign wealth fund in the world with over USD 1.3 trillion in assets, has gradually increased its exposure to renewable infrastructure and has set environmental criteria for its holdings. The Dutch pension fund ABP, one of the largest in Europe, has committed to reducing the carbon intensity of its portfolio. These funds do not typically invest directly in early-stage green technology—their mandates and liquidity needs preclude it—but they provide the capital that flows through infrastructure funds and green bonds. The ecosystem depends on them, and their allocation decisions shape the market.

The fund layer of the ecosystem is where patience resides. A venture fund holds a company for seven to ten years; an infrastructure fund holds an asset for twenty. That patience is the scarce resource in green finance, and the institutions that supply it are the ones that make the long-horizon projects possible.

2.2.4 Specialized institutions: green banks and development finance

Between the private institutions and the public regulators, a set of specialized institutions has emerged. Green investment banks—the UK Green Investment Bank (now privatized as the Green Investment Group), the Connecticut Green Bank, the New York Green Bank, Japan's Green Investment Promotion Organization—are public or quasi-public entities that use limited public capital to mobilize private capital. They do this through co-investment, guarantees, and warehousing: the green bank takes a first-loss position or provides a guarantee that makes a project bankable for private investors.

The model has a track record. The Connecticut Green Bank, established in 2011, reported that every USD 1 of public funds it invested mobilized roughly USD 6 of private capital by 2020. The UK Green Investment Bank, before its privatization in 2017, reported a similar multiplier. The numbers are not directly comparable across institutions—definitions of mobilization vary—but the direction is consistent: a small amount of public capital, deployed at the right point in the capital structure, can unlock a larger amount of private capital.

Development finance institutions—the World Bank Group's International Finance Corporation, the European Bank for Reconstruction and Development, the Asian Development Bank—perform a similar function at a larger scale. They provide long-tenor debt, equity, and guarantees to green projects in emerging markets, often in partnership with private investors. The institutions are not green-specific, but their green allocations have grown: the IFC's climate-related investments reached USD 4.6 billion in fiscal 2021, and the EBRD's green economy transition approach targets a majority of its annual investment.

The specialized institutions matter for the ecosystem because they fill the gap between what private capital will do and what the transition needs. A private infrastructure fund will not take construction risk on a first-of-a-kind green technology; a green bank or development finance institution will, if the public mandate justifies it. The specialized institution is the bridge between the private and public sectors, and its effectiveness depends on its mandate, its capital, and its independence from political cycles.

2.3 Policy and Regulatory Enablers

2.3.1 The regulatory architecture

Regulation shapes the ecosystem in three ways: it defines what counts as green, it requires disclosure of environmental information, and it sets the capital rules that determine how much risk institutions can hold. The three interact, and the interaction is where the ecosystem's development is decided.

The European Union has gone furthest in building the architecture. The 2018 Action Plan on Financing Sustainable Development set out the Commission's strategy, and a series of legislative instruments followed. The Taxonomy Regulation (EU) 2020/852, adopted in June 2020, establishes a classification system for environmentally sustainable economic activities. The Sustainable Finance Disclosure Regulation (EU) 2019/2088, adopted in November 2019, requires financial market participants to disclose how they integrate sustainability risks and to report on adverse sustainability impacts. The Benchmark Regulation (EU) 2019/2089, adopted in the same month, creates EU Climate Transition Benchmarks and EU Paris-aligned Benchmarks. The Non-Financial Reporting Directive, amended in 2014 and revised as the Corporate Sustainability Reporting Directive in 2021, requires large companies to disclose environmental and social information.

The architecture is ambitious and incomplete. The Taxonomy defines what is green, but the definition is technical and contested; the inclusion of natural gas and nuclear power in the taxonomy, decided in 2022, was politically divisive and illustrates the difficulty of drawing a line around "sustainable." The Disclosure Regulation requires disclosure, but the data quality is uneven and the comparability across firms is limited. The Benchmark Regulation creates labels, but the labels' environmental integrity depends on the underlying data. The architecture is a work in progress, and its effectiveness will depend on enforcement and on the quality of the data that flows through it.

China built its framework along different lines. The People's Bank of China and the China Banking and Insurance Regulatory Commission developed a Green Credit Statistics System, a green bond endorsement system, and a set of green finance pilot zones in Zhejiang, Guangdong, Xinjiang, Guizhou, and Gansu. The Chinese framework is more prescriptive than the EU's—it lists specific project categories that qualify for green credit—and it has driven rapid growth in green lending. China's green loan balance reached RMB 15.9 trillion (about USD 2.3 trillion) by the end of 2021, according to PBOC data, making it the largest green credit market in the world.

The two frameworks differ in philosophy. The EU's is principles-based and disclosure-driven; China's is list-based and target-driven. Both have strengths. The EU's approach is more adaptable to new technologies; China's is more operational for banks that need clear rules. The divergence matters for cross-border green finance, because a green bond issued under one framework may not qualify under the other. The International Platform on Sustainable Finance, launched by the EU in 2019, is an attempt to bridge the gap, but harmonization remains distant.

2.3.2 Disclosure: the data problem

Every green finance instrument depends on data. A green bond needs a verification that the proceeds are used for green projects. A green loan needs a measurement of the environmental performance of the borrower. A green fund needs a measurement of the carbon intensity of its portfolio. Without data, the labels are claims, and claims without verification are marketing.

The disclosure landscape is fragmented. The Task Force on Climate-related Financial Disclosures (TCFD), established by the Financial Stability Board in 2015, published recommendations in 2017 for voluntary climate-related financial disclosures organized around governance, strategy, risk management, and metrics and targets. The TCFD recommendations have been widely adopted—over 2,600 organizations had endorsed them by 2021—but adoption is voluntary and the quality of disclosure varies. The International Sustainability Standards Board (ISSB), established by the IFRS Foundation in 2021, is developing global baseline sustainability disclosure standards, with the first standards expected in 2023.

The data problem runs deeper than standards; it is a problem of measurement. A company's Scope 1 emissions (direct emissions from owned sources) are relatively easy to measure. Scope 2 emissions (from purchased electricity) are measurable if the electricity supplier discloses its carbon intensity. Scope 3 emissions (from the value chain, including suppliers and customers) are difficult to measure and often estimated with wide error bands. For a green finance ecosystem that depends on accurate environmental data, the Scope 3 problem is a serious constraint. A bank that lends to a manufacturer cannot easily measure the emissions embedded in the manufacturer's supply chain, and a fund that holds the manufacturer's shares cannot easily verify the company's climate claims.

The measurement problem is the subject of Chapter 5. Here the point is that the regulatory architecture is being built on a data foundation that is still being laid. The EU's Corporate Sustainability Reporting Directive, adopted in 2022, requires large companies to report according to European Sustainability Reporting Standards, which will be developed by the European Financial Reporting Advisory Group (EFRAG). The standards will cover climate, biodiversity, and social issues, and they will apply to roughly 50,000 companies in the EU. The scale is large, and the implementation challenge is correspondingly large.

2.3.3 Capital rules: the prudential dimension

A bank's capital rules determine how much of a given risk it can hold. The Basel III framework, agreed by the Basel Committee on Banking Supervision in 2010 and revised in 2017, sets minimum capital requirements for credit, market, and operational risk. Green finance enters the framework through the risk weights assigned to green assets. If a green loan has a lower risk weight than a comparable brown loan, a bank can hold more of it for the same amount of capital, and the incentive tilts toward green.

The evidence on green asset risk weights is contested. A 2021 study by Baker, Bergstresser, Serafeim, and Wurgler, published in the Journal of Financial Economics, found that green municipal bonds in the United States trade at a small premium—a "greenium" of a few basis points—relative to comparable conventional bonds. Other research, including work by Larcker and Watts (2020) on municipal bonds and by Tang and Zhang (2020) on corporate green bonds, has found no significant difference once issuer characteristics are controlled for. The European Banking Authority published a report in 2021 on the prudential treatment of ESG risks and found that the evidence for differentiated risk weights was not yet sufficient to justify a change. The Basel Committee has taken a similar position: it has focused on disclosure and risk management rather than on differentiated capital requirements.

The debate matters because capital rules are a powerful lever. A one-notch reduction in the risk weight of green corporate loans would free up capital for green lending at a scale that no subsidy could match. But the lever is powerful precisely because it is blunt: a lower risk weight for green assets would also apply to assets that are labeled green but are not, and the misallocation could be large. The prudential authorities are moving cautiously, and the caution is justified. Chapter 13 takes up the design of incentives without distortion.

2.3.4 Central banks and the monetary dimension

Central banks have entered the green finance debate through two channels: their asset purchases and their supervisory mandates.

The European Central Bank's asset purchase programs have included green bonds since 2016, when the ECB began buying corporate sector bonds under the Corporate Sector Purchase Programme. The green share of the ECB's corporate bond holdings has grown, though it remains a small fraction of the total. The ECB has also accepted green bonds as collateral in its refinancing operations, which gives them a liquidity advantage. The Bank of England has similar arrangements. The People's Bank of China has included green bonds in its medium-term lending facility collateral and has established a green finance evaluation scheme for banks.

The supervisory channel is more consequential. The ECB's 2020 guide on climate-related and environmental risks requires banks to assess and disclose their exposure to climate risk, and the ECB's 2022 supervisory review found that most banks had not yet integrated climate risk into their strategies. The Bank of England's 2021 climate stress test—the first of its kind—tested the resilience of major UK banks and insurers to climate scenarios, and found that the losses could be significant, particularly for banks with concentrated exposure to carbon-intensive sectors. The stress test is a supervisory tool that translates climate risk into capital planning, and its results feed into the banks' Internal Capital Adequacy Assessment Process (ICAAP).

The central bank dimension matters for the ecosystem because it changes the price of carbon-intensive assets. If a bank must hold more capital against a carbon-intensive loan, the loan becomes more expensive, and the borrower faces a higher cost of capital. The mechanism is indirect, but it is powerful. The ECB's climate stress test, the Bank of England's scenario analysis, and the PBOC's green finance evaluation all push in the same direction: they make climate risk a factor in the cost of capital.

2.3.5 The limits of regulation

Regulation can create markets, but it cannot create demand for green projects or supply of green technology. The EU's taxonomy defines what is green, but it does not make a green project profitable. The disclosure rules require data, but they do not make the data reliable. The capital rules tilt incentives, but they do not eliminate the risk that makes green projects hard to finance.

The limits are visible in the numbers. Despite a decade of regulatory effort, the EU's sustainable finance market remains a fraction of the total. The green bond market is large in absolute terms—over USD 500 billion in annual issuance—but it is still less than 5 percent of total bond issuance. The sustainable fund market is growing, but it is concentrated in listed equities and underweight in the private markets where green technology is developed. The regulatory architecture has built the plumbing; the water is not yet flowing at the scale the transition requires.

The gap between regulation and outcome is the central puzzle of green finance policy. The puzzle has several explanations. The transition is expensive, and no amount of labeling changes the underlying cost. The risks are real, and no amount of taxonomy changes the probability that a green technology fails. And the ecosystem's links are weak: the institutions exist, but they do not interact in the sequence that would carry a project from laboratory to operation. The first two explanations are constraints that policy cannot remove. The third is a coordination problem that policy can address, and it is where the rest of this book focuses.

2.3.6 The ecosystem as a system

I want to close this chapter by returning to the sentence from Chapter 1 and examining what it implies for the ecosystem map.

The missing functions are missing markets, and the missing markets are missing links between institutions that already exist. The map in this chapter shows the institutions: banks, rating agencies, insurers, funds, green banks, regulators. Each performs a function. The links between them are where the ecosystem fails.

Consider a specific case. A start-up develops a new battery chemistry. It needs USD 50 million to build a pilot plant. A venture fund provides USD 20 million in equity. A government grant provides USD 10 million. A green bank provides a USD 15 million loan with a guarantee from a development finance institution. The pilot plant succeeds, and the company needs USD 500 million to build a commercial plant. The venture fund can provide USD 100 million in follow-on equity. A project finance bank provides USD 300 million in debt, but only if the offtake is contracted and the technology is proven. A green bond issued by the company provides USD 100 million. The commercial plant succeeds, and the company sells it to an infrastructure fund for USD 800 million, recycling the capital into the next project.

Every step in this sequence requires a different institution, and every link between steps is a potential failure point. The venture fund may not have follow-on capital. The project finance bank may not accept the technology risk. The green bond market may not be open to a first-time issuer. The infrastructure fund may not pay the price the developer needs. The ecosystem works when the links hold, and it fails when any one of them breaks.

The policy implication is that the ecosystem's development depends on strengthening the links, not on expanding any single institution. A larger green bond market does not help if the venture capital stage is underfunded. A larger venture capital market does not help if the infrastructure fund stage is closed. The links are the leverage points, and the rest of this book examines them one by one.

Chapter 3 takes up the first link: the credit risk measurement that underlies the identification function. Without accurate risk measurement, the rating agencies cannot price green risk, the banks cannot allocate capital, and the insurers cannot underwrite the hazards. The measurement is the foundation, and it is where the ecosystem's technical work begins.

2.3.7 A note on the Chinese case

The Chinese green finance experience deserves separate attention because it illustrates both the power and the limits of a policy-driven ecosystem.

China's green credit balance reached RMB 15.9 trillion by the end of 2021, the largest in the world. The growth was driven by a combination of regulatory targets, window guidance, and a green finance evaluation scheme that scores banks on their green lending. The PBOC's evaluation scheme, introduced in 2018 and revised in 2021, assigns grades to banks based on the share of green loans in their portfolios and the environmental performance of their borrowers. The grades affect the banks' access to central bank refinancing and their regulatory standing.

The results are impressive in scale and mixed in quality. The green loan statistics include loans to projects that are green by the PBOC's definition, which is broader than the EU taxonomy and includes clean coal and energy efficiency improvements in coal plants. The inclusion is controversial—clean coal is not zero-carbon—but it reflects a pragmatic judgment that a transition from coal to cleaner coal is a necessary step for a country that generates most of its electricity from coal. The judgment is defensible on transition grounds and problematic on green labeling grounds, and the tension is not unique to China.

The Chinese case also illustrates the importance of the links beyond banking. China's green bond market is the second largest in the world, but the market for early-stage green technology venture capital is thin. The government has established guidance funds—state-backed funds of funds that invest in strategic sectors—and the guidance funds have allocated capital to green technology, but the allocation is small relative to the need. The infrastructure fund market is underdeveloped, and the exit routes for green project developers are limited. The ecosystem is strong at the banking end and weak at the venture and exit ends, which is the mirror image of the US ecosystem's strength in venture capital and weakness in green banking.

The comparison is instructive. Neither ecosystem is complete. The US has deep capital markets and a thin green banking system; China has a deep green banking system and thin capital markets for early-stage green technology. A complete ecosystem would have both, and the policy challenge is to build the missing links without dismantling the existing strengths.

2.3.8 The scale of the gap, revisited

I began this chapter with the IEA's estimate of USD 130 trillion in cumulative investment needed for net-zero by 2050. The number is large enough to be numbing, and numbing numbers are not useful for policy. A more useful number is the gap between current investment and the required investment.

The IEA's 2021 World Energy Investment report estimated that clean energy investment was roughly USD 1.1 trillion in 2020, against a required USD 4 trillion a year. The gap is roughly USD 2.9 trillion a year. Some of the gap will be closed by falling technology costs—solar and wind costs have fallen by 80–90 percent since 2010—and some by policy. The financial ecosystem's contribution is to mobilize private capital at the scale the gap requires, and the ecosystem's current capacity is far below that scale.

The gap is not evenly distributed. It is concentrated in emerging markets, where investment needs are large and capital is scarce. The IEA estimated that emerging market and developing economies need roughly USD 1 trillion a year in clean energy investment, and current flows are a fraction of that. The ecosystem's institutions are concentrated in advanced economies, and the links to emerging markets are weak. The development finance institutions are the primary channel, and their scale is limited.

The geographic gap is the ecosystem's largest missing link. A green bond issued in London can finance a wind farm in Scotland; the same bond cannot easily finance a solar plant in Kenya, because the currency risk, the political risk, and the offtake risk are different. The institutions that bridge the gap—MIGA guarantees, IFC loans, green banks in developing countries—exist but are small relative to the need. Chapter 11 takes up the question of how to scale them.

2.3.9 What the ecosystem map shows

The map in this chapter has three layers: institutions, links, and enablers. The institutions exist in every advanced economy and in most emerging ones. The links are uneven. The enablers are being built.

The functional test I set out at the start of the chapter gives a clearer picture than acatalogue would. Identification is well served: banks, rating agencies, and disclosure regimes all perform it, and the EU taxonomy and China's green credit statistics have made it operational. Buffering is partially served: insurers, guarantee funds, and development finance institutions perform it, but the instruments are thin in exactly the sectors—agriculture, forestry, first-of-a-kind technology—where the risks are hardest. Catalysis is poorly served: the institutions that create new projects rather than finance existing ones are scarce, and the links between the research base, the venture funds, and the project finance banks are weak.

That pattern—strong identification, partial buffering, weak catalysis—is the ecosystem's signature. It explains why green finance has grown rapidly in volume and slowly in impact. The volume is in the identification layer, where banks and bond markets operate. The impact depends on the catalysis layer, where the ecosystem is thin.

2.4 The Links That Fail

A functional map is only useful if it identifies where the links break. Four failure points stand out, each of which the later chapters address.

The valley of death between demonstration and commercial scale is the first. A technology that works in a pilot plant needs a first commercial plant to prove it works at scale, and the first commercial plant is the hardest thing to finance. It is too large for venture capital and too risky for project finance. The institutions that fill this gap—green banks, development finance institutions, corporate venture arms, and government procurement—exist but are small. The gap is not a shortage of capital in aggregate; it is a shortage of capital willing to take first-commercial-plant risk.

Then there is basis risk in insurance. A weather index insurance policy that pays out when the index crosses a threshold may not pay out when the farmer suffers a loss, and the farmer who experiences a loss without a payout does not renew the policy. The result is low uptake and a market that does not reach scale. Chapter 7 examines the design problem in detail. An insurance instrument that does not buffer the risk it is meant to buffer is not a working link.

Exit is the third break. A venture fund that invests in a green technology company needs an exit—an initial public offering, a trade sale, or a secondary sale—to return capital to its investors and raise a new fund. The exit routes for green technology companies are narrower than for software companies, because the public markets are less receptive to pre-revenue companies and the strategic buyers are fewer. Chapter 12 takes up the exit question. A venture capital market without exits does not recycle capital, and a market that does not recycle capital does not grow.

The fourth break is data. Every link in the chain depends on environmental data: the bank needs it to identify, the insurer needs it to underwrite, the fund needs it to report, the regulator needs it to supervise. The data is fragmented, inconsistent, and often estimated. The ISSB standards and the EU's Corporate Sustainability Reporting Directive are attempts to fix the fragmentation, but the fix will take years. Chapter 5 examines the measurement problem. A link built on unreliable data will fail under stress.

2.5 The Policy Question

The policy question that follows from the map is not "how do we expand green finance?" It is "which links do we strengthen, and in what order?"

The answer depends on the stage of the transition. In the early stage, the binding constraint is the valley of death, and the policy priority is to fund first commercial plants. In the middle stage, the binding constraint is the cost of capital for operating assets, and the policy priority is to deepen the green bond and infrastructure fund markets. In the late stage, the binding constraint is the decarbonization of existing assets, and the policy priority is to finance the transition of high-carbon industries.

The stages overlap, and a country may face all three constraints at once. The policy design must therefore be sequenced: strengthen the link that is binding now, and build the link that will bind next. The sequencing is the subject of Chapter 15, and the chapters in between examine each link in turn.

I will return to the functional test throughout the book. For each institution and instrument, I will ask what it identifies, what it buffers, and what it catalyzes. The answers will not always be flattering. Some instruments that are celebrated as green finance innovations perform none of the three functions well; they are marketing. Others that are unglamorous—guarantees, reinsurance, procurement—perform the functions quietly and at scale. The test is a way of distinguishing the two.

2.6 The Ecosystem in One Page

The chapter's argument can be compressed into a single table. It is not a substitute for the analysis, but it is a useful reference for the chapters that follow.

Table 2-1: The green finance ecosystem by function.

The table makes the chapter's central claim visible. The ecosystem has institutions but lacks links between them. Every function in the left column has at least one institution performing it. Every failure in the right column is a failure of connection—between the data and the rating, between the index and the loss, between the demonstration and the commercial plant, between the listed market and the emerging market.

The links are where the policy work is, and the links are where the rest of this book goes.

2.7 Coda: The View from Zhaodong

In August 2011, I conducted field research at the modern agriculture demonstration zone in Wuliming Town, Zhaodong City, Heilongjiang Province. The fields were computer-controlled, the irrigation was precision-managed, and the corn yield reached one ton per mu—roughly 15 metric tons per hectare, well above the regional average. The technology worked. The farmers who adopted it earned more. And the finance that made it possible was a patchwork: a bank loan for the equipment, a cooperative for the operating capital, an insurance policy for the weather risk, and a government subsidy for the infrastructure.

No single institution financed the transition. The bank identified the creditworthy cooperative; the insurer buffered the weather risk; the cooperative pooled the farmers' savings; the government catalyzed the infrastructure. The patchwork worked because the links held. When one link failed—in a bad weather year when the insurance payout did not cover the loss—the whole structure strained.

I have thought about that field ever since. The lesson is not that Chinese agriculture is promising, though it is. The lesson is that the financial ecosystem for a transition is a patchwork of institutions, and its strength is in the stitching. The same is true for the zero-carbon transition at national and global scale. The institutions exist. The stitching is the work.

Chapter 3 begins the stitching at the foundation: the measurement of credit risk that makes identification possible. Without it, the rating agencies cannot rate, the banks cannot lend, and the insurers cannot underwrite. The measurement is where the ecosystem's technical work starts, and it is where the next chapter goes.