Chapter 11: Government-Guided Funds and the Market Failure in Early-Stage Green Capital

The loan officer in Chongqing had 14 million yuan on the line and a credit score of 412 out of 1000. The borrower was a nine-person firm with a patented process for recovering rare-earth oxides from electronic waste. The pilot line ran at 60 percent of design capacity. The process worked. The firm needed the money to finish a second line, and its collateral was a warehouse lease, some secondhand equipment, and a patent no branch in the city had the capacity to value. Cash flow was negative. The score was the score.

She declined. Any loan officer at that branch would have declined, and that is the point worth holding onto. The decline was not a judgment about the technology. It was the output of a model that prices collateral, cash flow, and default history, and that has no line item for avoided emissions, no way to value a process that displaces virgin mining, and no mechanism for holding a ten-year environmental payoff against a three-year loan. The firm's technology might be exactly what the transition needs. The system, as built, cannot see it.

What follows is about that gap and about the instrument governments have built to fill it. The market failure in early-stage green capital turns out to be several failures stacked on one another, and they reinforce each other in a way that no single private actor can break. A guided fund can break the loop, but only if its design is aimed at the specific failure rather than at the general problem. The international record is more useful for its failures than its successes, because the failures are where the design parameters show their teeth.

One scope note before the argument. This chapter does not argue that governments should pick technologies. It argues that the early-stage green capital market has structural features that private capital alone will not finance, and that a guided fund can change the terms on which private capital enters. The fund is a catalyst. The capital required for the transition is orders of magnitude larger than any public fund, and a country that relies on guided funds to finance the transition will not finance it.

11.1 Market Failure in Green VC

11.1.1 What "Early-Stage Green" Means in Financial Terms

Venture capital runs on a stage model, and the stage model maps to how risk is priced. Pre-seed and seed money translates a concept into a prototype. Series A pays for the first commercial deployment. Series B and later pay for scale. Each stage carries a different risk profile, a different investor base, a different expected holding period. The taxonomy is not academic; it tells a general partner what she is buying and when she can sell it.

Early-stage green investment sits awkwardly in this model. Take a green technology firm at the moment it needs its first substantial external capital. It has a working prototype. It has not demonstrated commercial-scale production. Unit economics are projected, not realized. Its principal asset is a process or a design, protected imperfectly by patents. Its customers, if it has any, are utilities, municipalities, or industrial firms whose procurement cycles run 18 to 36 months.

Compare it to a software firm at the same stage. The software firm's marginal cost of production is near zero. It can acquire a hundred customers on a cloud platform for the cost of a sales team. Its failure is cheap and reversible. The green firm's marginal cost of production is the cost of a factory. Its failure is a stranded asset. On a spreadsheet the two firms can look identical—both pre-revenue, both carrying technology risk—but the capital needed to resolve the uncertainty is an order of magnitude apart.

That gap is the first structural feature: the capital intensity of the validation step.

Table 11-1: Validation Cost and Time by Venture Type

A venture fund with a ten-year life and a 2/20 fee structure can hold a software position from seed to exit in five years. A green hardware position may not reach commercial validation until year five, leaving no time to scale before the fund must return capital. The general partner faces three bad options: hold past fund life, which limited partners resist; sell early at a discount, which destroys the return; or avoid the position. Most avoid it.

The avoidance is rational, and it aggregates into a pattern. In the decade to 2023, global venture capital deployed roughly $600 billion a year at peak. Clean energy and industrial decarbonization captured a small single-digit share, and within that share the overwhelming majority went to Series B and later—to firms that had already resolved the validation risk. The pre-Series A green segment, where the validation risk lives, stayed thin.

11.1.2 The Four Failures

The thinness is not one failure. It is at least four, and they compound.

The first is information asymmetry of an unusual kind. In the standard lemons model the seller knows more about quality than the buyer. In early-stage green investment, neither side knows. Whether a technology performs at scale is genuinely uncertain, and disclosure does not resolve it. A battery chemistry that performs in the laboratory may fail in a 40°C warehouse after 800 cycles. No amount of due diligence settles the question; only operation settles it. The investor is therefore not choosing between good and bad firms but between overlapping distributions of technical outcomes. This is why the standard remedies for information asymmetry—mandatory disclosure, certification, rating—help less here than they do in credit markets. A rating can summarize known information. It cannot manufacture the operating history that does not yet exist.

The second failure is a maturity mismatch between environmental payoffs and financial returns. A green technology's environmental benefit accrues over the life of the asset it displaces: twenty years for a solar installation, thirty for a building retrofit, indefinitely for a process that avoids mining. The financial return accrues over the fund's holding period. The private investor discounts the environmental benefit to near zero because it cannot be monetized inside the holding period. Investors are not short-sighted by temperament. The instrument through which they are compensated has a horizon, and the environmental payoff exceeds it.

Coordination is the third. Many green technologies require complementary infrastructure before they are worth anything. A firm producing green hydrogen needs electrolyzers, renewable power, and offtake contracts, and none of these exist at the required scale until the others do. A firm producing low-carbon cement needs customers willing to specify it, standards that recognize it, and logistics that can deliver it. Each link in the chain waits for the others. Private capital will finance a single link if the other links are visible. When they are not, it waits. The waiting is individually rational and collectively costly.

The fourth failure is the absence of exit. Venture capital is compensated at exit, by acquisition or public listing, and the exit market for green hardware is thin. Strategic acquirers are few, because the industrial incumbents that would buy a green technology firm are often the firms being disrupted. Public markets are demanding: a pre-profit hardware firm with a ten-year horizon struggles to list. Without a credible exit the fund cannot underwrite the position, whatever its environmental merit.

These four interact. Information asymmetry raises the required return. The higher required return shortens the acceptable holding period. The shorter holding period makes coordination harder to wait out. The harder coordination is to wait out, the thinner the exit market. The thin exit market raises the required return again.

flowchart TD
    A[Technology uncertainty<br/>unresolved by disclosure] --> B[Required return rises]
    B --> C[Acceptable holding<br/>period shortens]
    C --> D[Coordination risk<br/>cannot be waited out]
    D --> E[Exit market thins]
    E --> B
    B --> F[Private capital<br/>withdraws from pre-Series A]
    F --> G[Validation gap persists]
    G --> A

The loop is the chapter's central object. Everything that follows is an attempt to break it at a specific point.

11.1.3 Why the Market Does Not Self-Correct

The standard objection: if the returns are there, capital will find them. The objection assumes the four failures are temporary frictions a liquid market would arbitrage away. Some are. Information asymmetry about a mature technology is arbitrageable; a rating agency can resolve it. But the failures described above are features of the asset class, not frictions in an otherwise complete market.

Ask what a private investor would need to underwrite the Chongqing firm. A technical assessment of the process at scale, which requires operating the second line, which requires the investment. An offtake contract with a buyer committed to recovered oxides at a fixed price, which requires the buyer to believe the process works. A policy environment that prices virgin mining at its environmental cost, which is a public good. An exit route, which requires all of the above. The investor needs the investment to have happened in order to evaluate whether the investment should happen. That is a coordination problem, and no private actor solves it alone.

The public instrument that addresses coordination problems is the guided fund. Its logic is to break the loop at one point: finance the validation step so the information asymmetry resolves, at a scale and horizon private capital cannot match. Whether it succeeds depends entirely on design.

11.2 Design of Government-Guided Funds

11.2.1 The Anatomy of a Guided Fund

A government-guided fund pools public and private capital, with the public money accepting terms that improve the private money's risk-return profile. The term covers a wide range of structures. At one end sits the direct investment fund, in which a state-owned manager makes equity investments. At the other sits the fund-of-funds, in which the government commits to private venture funds that then invest. In between are co-investment schemes, matching guarantees, and hybrids.

The design choices are not cosmetic. They determine which failure the fund addresses and which it ignores. Five parameters carry most of the weight, and they are best understood as a bundle rather than a menu.

Mandate comes first, because a fund told to support green technology will drift toward the firms that are easiest to support, which are the firms closest to commercial viability. A mandate narrow enough to name the stage and the sector—pre-revenue firms in hard-to-abate industries, defined by a published technology taxonomy—will accept higher loss rates and will actually address the information asymmetry. The mandate sets the target loss ratio, and the target loss ratio sets everything downstream.

Loss sharing is the second parameter, and it is the one most often fixed by political negotiation rather than analysis. Public capital typically accepts first-loss or mezzanine positions to attract private money. How much matters. At 100 percent first-loss coverage the private partner bears no downside, has no reason to conduct due diligence, and will invest in anything the fund approves. At 20 percent the private partner bears enough risk to discipline its selection. My own calibration, from the structure of European guarantee products discussed below, puts the workable band at 10 to 30 percent of first-loss coverage: enough to change the investment decision, not so much that screening effort collapses.

Pricing of the public capital is the third. Charge the same return as private capital and the fund is not a subsidy and changes nothing. Charge zero and it is a grant dressed as equity, attracting private money with no intention of holding the position. The workable structure is a return below market but above zero, with the gap calibrated to the specific failure. A fund addressing maturity mismatch should price with a long tenor and a low coupon, not with a low expected return.

Governance of investment decisions is the fourth, and its failure mode is political capture: the fund becomes a vehicle for directing capital to firms with connections rather than firms with technology. The standard remedy is an independent investment committee with a professional majority and a published investment thesis. The remedy is imperfect—committees can be captured too, and the appointment process is itself political—but the alternative, ministerial approval of individual deals, is worse, and the record of funds that adopted it is unambiguous.

The fifth parameter is the exit and recycling rule. A fund that must return capital to the treasury at the end of its life will behave like a private fund and avoid the long-horizon positions it was created to hold. A fund that can recycle returns into new investments has a longer effective horizon. The design question is whether the charter permits recycling and whether the accounting treatment rewards or penalizes it.

Table 11-2: Fund Design Parameters and the Failures They Address

No single parameter addresses all four failures. A fund built to resolve information asymmetry through a narrow mandate and independent screening will still fail on maturity mismatch unless its pricing and recycling rules are set correctly. Design is a bundle, and the bundle has to be coherent.

11.2.2 The Mandate Problem: Additionality Versus Selection

The central design tension is between additionality and selection. Additionality means the fund finances firms that would not otherwise be financed. Selection means the fund finances firms likely to succeed. The two pull in opposite directions, because the firms least likely to be financed privately are often the firms with the highest technical risk.

Maximize additionality and the fund holds a portfolio of high-risk firms and reports a high loss rate. Maximize selection and the fund finances the firms private capital would have financed anyway, reports a low loss rate, and achieves no additionality. The political system rewards the second and punishes the first, which is why so many guided funds converge on the second.

The remedy is to measure additionality directly rather than infer it from returns. For each investment the fund can require a documented statement of why private capital declined and what the fund's participation changes. "The firm was too early for private capital" is additionality. "The firm was financeable but the fund offered better terms" is not. The documentation requirement is not a formality. It forces the investment committee to name the failure being addressed, and it creates a record against which performance can be assessed later.

The Chongqing firm illustrates the distinction. A fund that invested because the technology was promising but pre-commercial would be exercising additionality. A fund that invested because it offered a below-market rate to a firm a bank would have financed at market would be exercising selection. The two investments look identical on the balance sheet. They are not identical in what they achieve.

11.2.3 The Loss-Sharing Calibration

Loss sharing deserves separate treatment because it is the parameter most often set by negotiation and because getting it wrong produces a specific, predictable failure.

Take a fund offering private investors a 100 percent first-loss guarantee on green investments. The private investor's downside is gone. The rational response is to invest in the highest-expected-return opportunity available, regardless of risk, because the risk has been transferred. The portfolio becomes a collection of high-risk bets selected by investors who bear none of the losses. The expected loss to the public is the average loss of the riskiest segment of the market—precisely the segment the fund was meant to make investable, not to absorb wholesale.

Take the opposite: no loss sharing. The private investor bears the full risk and invests only where the risk-adjusted return clears its hurdle. The public capital is a co-investor on market terms, and additionality is zero.

The workable structure sits between. The private investor bears a meaningful first-loss tranche, enough that its selection effort is disciplined, and the public capital absorbs losses above that tranche up to a cap. The calibration question is the size of the private tranche. Ten to 20 percent is sufficient to preserve screening incentives while improving the risk-return profile enough to attract capital; below 10 percent screening weakens, above 30 percent the guarantee is too thin to change the decision. That band is my reading of the guarantee products the European Investment Fund has run since the late 1990s, which I take up in Section 11.3.2.

A second calibration question is whether sharing is deal-by-deal or portfolio-wide. Deal-by-deal sharing preserves the strongest screening incentive, because the private investor loses on each bad deal. Portfolio-wide sharing pools the risk, which weakens the incentive to screen individual deals but protects against the tail outcome that would otherwise deter a risk-averse investor. For early-stage green investment, where individual outcomes are highly uncertain and the portfolio logic is strong, portfolio-wide sharing with a deal-level deductible is the more defensible structure.

11.2.4 Governance and the Capture Problem

Governance is where design meets politics. The fund exists because the market failed; the fund is funded by the state; the state has preferences about which firms should benefit. The preferences are not always corrupt. A regional government may legitimately want to direct green investment to a declining industrial region. The same mechanism that permits legitimate direction permits capture.

The pattern in the record is consistent: funds with ministerial approval of individual investments perform worse than funds with independent committees, controlling for mandate and vintage. The mechanism is straightforward. Ministerial approval selects for political return, political return is uncorrelated with technical merit, and the firms that benefit are those with the best connections rather than the best technology. The portfolio underperforms, and the underperformance is invisible in the early years because the losses take time to surface.

The remedy is structural. An independent investment committee with a professional majority, a published investment thesis, and a charter that excludes ministerial approval of individual deals is the minimum. Three practices help beyond the minimum. Committee members should be appointed for fixed terms that do not coincide with electoral cycles. Decisions should be published with a lag, so the pattern of investment can be assessed. And the fund should be audited on additionality, not only on returns.

None of this eliminates capture. It raises the cost. A fund that requires a published rationale for each investment and an independent committee's approval makes it harder to direct capital to a connected firm without a defensible technical case. Harder is not impossible, but in institutional design harder is the achievable goal.

11.2.5 A Reference Design

The parameters can be assembled into a reference design. It is not the only workable design, but it is coherent, and it can be used to assess the international cases that follow.

The mandate is narrow: pre-revenue and early-revenue firms in hard-to-abate sectors, defined by a published technology taxonomy. The fund operates as a fund-of-funds, committing to specialist venture funds with demonstrated technical capability rather than investing directly. Public capital takes a 20 percent first-loss position on a portfolio basis, with a 5 percent deal-level deductible. Pricing is below market but above zero, with a tenor matched to the technology's validation timeline, seven to ten years. Investment decisions are delegated to the venture managers, selected competitively, whose carry vests only on exit. The charter permits recycling of returns for fifteen years. Additionality is measured by a required statement, for each portfolio company, of the private capital gap the investment fills.

flowchart LR
    G[Government<br/>commitment] --> FOF[Fund of funds<br/>public manager]
    FOF --> VF1[Specialist VC fund A<br/>energy storage]
    FOF --> VF2[Specialist VC fund B<br/>industrial process]
    FOF --> VF3[Specialist VC fund C<br/>circular materials]
    VF1 --> P1[Portfolio firms<br/>pre-revenue]
    VF2 --> P2[Portfolio firms<br/>early revenue]
    VF3 --> P3[Portfolio firms<br/>pre-revenue]
    PVT[Private LPs] --> VF1
    PVT --> VF2
    PVT --> VF3
    FOF -.->|20% first loss<br/>portfolio basis| PVT

The logic is clean. The fund-of-funds structure delegates technical selection to managers who can assess it. The first-loss position is large enough to change the private investor's calculation and small enough to preserve screening. The long tenor and recycling rule address the maturity mismatch. The additionality statement forces the fund to name the failure it addresses. Whether it works in practice depends on the quality of the specialist managers and the independence of their selection, which is the next section's subject.

11.3 International Experiences

11.3.1 The Israeli Model: Yozma and Its Successors

The most cited guided fund is Israel's Yozma program, launched in 1993. The government committed $100 million to a fund-of-funds that invested in ten new venture capital funds, each managed by a private partnership. The government took a 40 percent share of each fund and offered the private partners an option to buy out the government's stake at cost plus a modest interest rate within five years. The private partners had an incentive to perform: if the fund succeeded they captured the upside, and if it failed the government bore the loss.

The results are documented. Within five years the ten funds had raised additional private capital and invested in more than 200 startups. The buyout option was exercised in most cases, and the government's share was recycled. By the early 2000s Israel had a venture capital industry that no longer needed the program.

Three features of the Yozma design are worth isolating. The buyout option converted a subsidy into a temporary equity stake and gave the private partner a clear path to full ownership. Competitive selection of fund managers sent the capital to teams with demonstrated capability. And the scope was narrow in the right way: Yozma did not pick technologies. It picked managers and let them pick technologies.

The model has been adapted widely, and the adaptations that preserved these features have outperformed those that did not. The usual caveat applies: Israel's defense technology base and its immigration of skilled engineers are not replicable, and part of what looks like program design is actually those endowments.

11.3.2 The European Experience: EIF, InnovFin, and the Guarantee Model

Europe took a different route. The European Investment Fund operates as a fund-of-funds and guarantee provider, committing to venture funds and providing first-loss guarantees to banks lending to small firms. The InnovFin program, launched in 2014 under Horizon 2020, extended the model to research-intensive firms including green technology firms, with guarantees covering up to 50 percent of loan principal.

Scale and standardization are the European model's strength. EIF guarantee products are standardized, which allows rapid deployment across member states. The same standardization is the weakness: a product calibrated to the average small firm is not calibrated to the early-stage green firm's risk profile. The green firm's validation risk is not the average SME's credit risk, and a guarantee covering 50 percent of a loan to a firm that cannot service the loan does not make the loan serviceable.

The European experience also illustrates the additionality problem. Evaluations of EIF guarantee programs have repeatedly found that a substantial share of guaranteed loans would have been made without the guarantee—the finding recurs across programs and countries, though the estimated share varies. Guarantees improve terms more than they expand access. For the green transition, where the access margin is the binding constraint, a guarantee that improves terms for firms that already have access is not enough.

11.3.3 The German and Japanese Models: Public Banks and Directed Credit

Germany and Japan rely on public development banks rather than fund-of-funds. Germany's KfW and Japan's Development Bank of Japan provide long-term debt and equity to firms the commercial banks will not finance, with mandates that include environmental objectives. KfW's energy transition programs have committed tens of billions of euros to building retrofits and renewable generation.

The public bank model has a clear advantage: it can hold long-tenor assets that a fund cannot, and it can price below market without the appearance of a subsidy because its funding cost is lower. It has an equally clear disadvantage: it is a bank, and banks are organized around credit risk, not technology risk. A public bank lending to a pre-revenue green firm is making an equity-like bet through a debt instrument, which misprices the risk and produces either a loss or a credit committee that declines the loan.

The German and Japanese models work best at deployment—financing proven technologies at scale—and least well at validation. That is consistent with their design. The public bank's comparative advantage is low-cost long-term funding, which is exactly what a proven technology needs and exactly what a pre-commercial technology cannot use.

11.3.4 The Chinese Experience: Guided Funds at Scale

China runs the largest guided fund sector in the world. By the mid-2020s, government-guided funds at the national, provincial, and municipal levels held committed capital in the trillions of yuan, with a substantial share directed to strategic emerging industries including green technology. The structures vary. Some are direct investment funds managed by state-owned capital operation companies. Some are fund-of-funds committing to private venture funds. Some are co-investment schemes with commercial banks.

The scale is the achievement. No other country has deployed public capital at this magnitude into early-stage technology, and the sector has financed firms that would not otherwise have been financed. The design challenges are equally instructive, because they are the challenges any large guided fund program faces, magnified.

Mandate drift comes first. A fund mandated to support early-stage green technology faces pressure to invest in later-stage firms that offer visible returns and political credit. The drift is hard to resist because the later-stage investments perform better on the metrics that get reported. The remedy is a mandate that specifies the stage and a reporting framework that measures additionality rather than returns.

The second challenge is the relationship between the fund and the banks. Much Chinese guided fund capital is deployed through banks, which means the investment decision is made by a credit committee applying credit criteria. The early-stage green firm fails those criteria and the capital does not reach it. The structural fix is to separate the equity decision from the credit decision, which requires the fund to have an equity investment capability rather than a lending capability.

Exit is the third. Chinese guided funds often require portfolio firms to list on domestic exchanges within a specified period. The requirement is meant to ensure recycling, but it selects for firms that can list, which are not the firms with the longest validation timelines. A green hardware firm with a ten-year horizon is excluded by a five-year listing requirement.

These challenges are not unique to China. They are the challenges the reference design in Section 11.2.5 is built to address. The Chinese experience is valuable precisely because it shows what happens when the design parameters are set by scale and political urgency rather than by the failures being addressed.

11.3.5 What the Cases Teach

Four conclusions survive the comparison.

The fund-of-funds structure with competitively selected specialist managers performs better than direct investment by a state agency. The reason is capability: technology assessment is a specialized skill, and the state agency does not have it. The Israeli and European models both delegate selection, and both outperform the direct-investment models on additionality.

The loss-sharing structure must preserve the private partner's screening incentive. European guarantee programs with high coverage show weaker additionality than those with lower coverage. The Israeli buyout option is a different mechanism with the same logic: the private partner keeps the upside only if the fund performs, which disciplines selection.

The public bank model suits deployment and not validation. Germany and Japan have used it effectively to finance proven technologies at scale. Neither has used it effectively to finance pre-commercial technologies, because the instrument is a loan and the risk is equity.

Scale without design discipline produces mandate drift. The Chinese experience shows that a large guided fund sector will drift toward later-stage, lower-risk investments unless the mandate and the reporting framework prevent it. The drift is the predictable response to measuring the wrong thing.

11.4 The Catalytic Logic: How a Guided Fund Changes the Terms

11.4.1 The Catalyst Function

A guided fund can change the terms on which private capital enters, so that private capital finances the transition. That is the catalytic function, and it operates through three channels.

Risk transfer is the most direct and most widely used. By accepting a first-loss position the fund reduces the private investor's downside, moving the risk-return profile from below the hurdle rate to above it. The private investor enters. The channel works only if the risk transfer is large enough to change the decision and small enough to preserve screening, which is the calibration problem of Section 11.2.3.

Information production is slower and more durable. By financing the validation step the fund generates the operating history that resolves the technology uncertainty. That resolution is a public good: the information that a technology works at scale is valuable to every subsequent investor, and no single investor will pay to produce it because the benefit is shared. The fund internalizes the externality. The information persists after the fund's capital is withdrawn, which is why this channel outlasts the first.

Coordination is the most powerful and the least used. By committing capital to one link in a supply chain the fund makes the other links investable. A fund that finances a green hydrogen producer signals to electrolyzer manufacturers, renewable developers, and offtake buyers that the producer will exist, which changes their investment calculations. The signal is credible because the capital is committed. Using this channel requires the fund to invest with a view of the whole chain rather than a view of individual deals, and that is a harder discipline than deal-by-deal selection.

The three channels are not alternatives. A well-designed fund uses all three, with the mix set by the failure being addressed. A fund addressing information asymmetry leans on information production. A fund addressing coordination leans on the chain signal. A fund addressing the private investor's hurdle rate leans on risk transfer.

11.4.2 The Conditions for Catalysis

Catalysis is not automatic. Four conditions have to hold.

The private capital must be genuinely additional. If the fund's participation displaces private capital that would have entered anyway, the catalysis is illusory. The additionality statement described in Section 11.2.2 is the mechanism for verifying this, and it has to be enforced rather than treated as a formality.

The fund's capital must be patient. A fund with a five-year horizon cannot finance a validation step that takes seven years. The tenor of the public capital has to match the technology's timeline, which is why the recycling rule and the long-dated commitment are design requirements rather than conveniences.

The fund must have the capability to assess technology. A fund that cannot distinguish a working process from a plausible one will finance the plausible ones and lose money. The capability can be internal or delegated to specialist managers, but it has to exist. This is the strongest argument for the fund-of-funds structure.

The exit route must be credible. The fund can hold a position longer than a private fund, but it cannot hold forever. The exit can be a strategic acquisition, a listing, or a sale to a later-stage investor. The fund's design should specify which exit it is underwriting and should not assume an exit that does not exist.

11.4.3 The Counterargument: Crowding Out

The standard objection to guided funds is that they crowd out private capital. The objection takes two forms. The fund competes with private investors for deals, raising valuations and lowering returns, which deters private capital from the sector. Or the fund's subsidy allows inefficient firms to survive, which slows consolidation and reduces long-run productivity.

Both forms have empirical support in some contexts. The German evidence on public guarantees found that a share of guaranteed loans would have been made without the guarantee. The objection is not hypothetical.

Crowding out is a design failure, not an inherent property. A fund that invests on the same terms as private capital, in the same deals, crowds out. A fund that invests where private capital will not, on terms that improve the private capital's risk-return profile, does not. The distinction is the additionality condition, and it is why the condition must be enforced. A guided fund without an additionality test is a crowding-out machine.

The second form of the objection—that subsidies allow inefficient firms to survive—is harder to dismiss. A fund that finances a firm with a flawed technology delays the firm's failure and diverts capital from better firms. The remedy is not to avoid the risk but to price it: the fund should expect a higher loss rate than a private fund, and it should report the loss rate honestly. A fund with a zero loss rate has not succeeded. It has not taken the risk it was created to take.

11.5 The Interaction with Rating and Insurance

This chapter sits inside a book whose argument is that the transition requires three mechanisms in sequence: rating, buffering, and catalysis. The guided fund is the catalytic mechanism. Its interaction with the other two determines whether the catalysis is durable.

11.5.1 The Fund and the Rating

The rating mechanism, developed in Chapters 3 through 5, converts environmental performance into a form the credit model can read. A guided fund can use it two ways.

The rating can serve as a screening criterion. A fund that requires a minimum environmental performance rating for its portfolio firms ensures the capital goes to firms with measurable environmental benefit, not merely to firms that describe themselves as green. The requirement is only as good as the rating, which is why the methodology matters. A rating that can be gamed is a screening criterion that can be gamed.

The rating can also serve as a pricing input. A fund that prices its capital against the rating, with a lower cost for higher environmental performance, creates an incentive for firms to improve. The incentive is strongest when the rating is granular enough to reward improvement, which is the argument for a continuous score over a pass-fail label.

The interaction runs both ways. The fund's portfolio generates the operating data the rating model needs to calibrate its environmental performance component. A rating model developed without access to early-stage firms will misprice them; a fund that finances early-stage firms produces the data that corrects the mispricing. That is the information-production channel of Section 11.4.1, applied to the rating mechanism.

11.5.2 The Fund and the Insurance

The insurance mechanism, developed in Chapters 6 through 9, buffers the risks that would otherwise make green investment unfinanceable. For early-stage green firms the relevant risks are performance risk (the technology does not work as specified) and revenue risk (the offtake does not materialize).

A guided fund can require insurance as a condition of investment, which transfers performance risk to an insurer and reduces the fund's loss exposure. The requirement is workable only if the product exists and is priced at a level the firm can afford. For many early-stage green technologies the insurance market is thin, which is why the fund may need to subsidize the premium or work with insurers to develop the product.

The more important interaction is at the portfolio level. A fund holding a portfolio of firms exposed to the same weather or technology risk is undiversified. Insurance allows the fund to transfer the correlated risk, which reduces the portfolio's variance and improves its risk-return profile. The logic is the same one that applies to a bank's loan portfolio, applied to an equity portfolio.

11.5.3 The Sequence

The three mechanisms operate in sequence. The rating identifies the firm's environmental performance and its credit-relevant risk. The insurance buffers the risks the rating cannot eliminate. The fund catalyzes the capital that the first two mechanisms make investable.

The sequence matters because each mechanism depends on the ones before it. A fund that invests without a rating cannot measure what it has bought. A fund that invests without insurance bears a risk it cannot diversify. A rating without a fund identifies firms that no one will finance. The three are a system, and the system's output is a financeable green firm.

flowchart LR
    R[Rating<br/>environmental performance<br/>+ credit risk] --> I[Insurance<br/>performance + revenue<br/>risk buffered]
    I --> C[Catalysis<br/>guided fund lowers<br/>cost of capital]
    C -->L[Financeable<br/>green firm]
    L --> D[Private capital<br/>enters at later stage]
    D --> E[Exit / recycling]
    E --> C

The sequence is not a pipeline a firm passes through once. It is a loop. The fund's portfolio generates the data that improves the rating. The rating improves the insurance pricing. The insurance improves the fund's portfolio performance. The improved performance attracts private capital, which allows the fund to recycle into new firms. The loop turns, and the cost of the next turn depends on the numbers: a second rating on a firm with two years of operating data costs a fraction of the first, and a second insurance policy on a technology with a loss history prices at a fraction of the first premium.

11.6 The Chongqing Firm, Revisited

Return to the loan officer and the rare-earth recycling firm. The file that produced a 412 can be reconstructed under the three mechanisms, and this time the reconstruction carries numbers the first telling did not have.

The rating first. The firm's process avoids mining rare-earth oxides, and the avoided mining has a measurable carbon and land-use footprint. A rating model that incorporates environmental performance would score the firm on avoided emissions per unit of output, verified against the pilot line's process data. Suppose the model returns 58 out of 100 on the environmental component and 31 out of 100 on the credit component. The composite score moves the firm from 412 to roughly 470. That is not creditworthy on its own. It makes the environmental benefit legible, which is the precondition for everything after.

The insurance second. The firm's principal risk is that the second line does not perform as the pilot did. A performance policy underwritten against the pilot data and the equipment specification would cover the output shortfall for a defined period. The premium would be high—4 to 6 percent of insured value—because the risk is real and the actuarial base is thin. Say 5 percent on 14 million yuan, or 700,000 yuan a year. The fund can subsidize that premium, which is cheaper than bearing the performance risk on its own book.

The catalysis third. A guided fund built to the reference design would invest through a specialist venture fund, taking an equity position alongside a private co-investor who holdsthe 20 percent first-loss tranche. Suppose the fund takes 30 percent of the equity at a 12 million yuan pre-money valuation and the private co-investor takes another 20 percent. The fund's investment is conditioned on the rating and the insurance policy.

Now watch what the loan officer sees when the file comes back. The firm's cash flow is still negative. Collateral is still the lease, the secondhand equipment, and the patent. But two things have changed. A 700,000 yuan annual premium has moved the performance risk off the firm's balance sheet and onto an insurer, which means the downside case for the second line is bounded in a way the credit model can read. And 6.4 million yuan of committed equity sits ahead of the bank in the capital structure, which changes the recovery calculation on a 14 million yuan loan. The model reruns. The score moves from 470 to somewhere in the 540–560 range. The branch's threshold for a secured facility is 580. The loan still does not clear on the model.

Here is where the second telling diverges from the first. The score does not clear the threshold, and that is the honest outcome. What changes is the conversation. The loan officer now has a file she can defend: a rating that names the environmental benefit, an insurance policy that caps the technical downside, an equity cushion from a fund with a published thesis, and a 14 million yuan ask that can be restructured as an 8 million yuan senior facility against the equipment plus a 6 million yuan mezzanine piece the fund guarantees at 20 percent. She writes it up. The credit committee approves the senior facility at 6.8 percent, roughly 180 basis points above the branch's prime rate, and the mezzanine piece closes three weeks later.

The firm still might fail. The second line may underperform the pilot, the offtake may not materialize, the fund may have backed the wrong process. If it fails, it fails because the technology did not work. Under the system as it stood in 2023, it would have failed because the model could not see it. The first failure is a failure of the firm. The second is a failure of the financial system, and the system is the thing this chapter is about.

11.7 Measuring Whether the Fund Works

A guided fund that is not measured will drift, and the choice of metrics determines the direction of the drift. Measurement is a design decision, not an accounting afterthought.

11.7.1 The Three Tiers of Measurement

Financial performance is the first tier. The fund should report its internal rate of return, its loss rate, and its realized exits, against a benchmark that reflects the mandate. A fund built to finance pre-revenue firms should be benchmarked against a portfolio of comparable early-stage investments, not against a broad market index. The benchmark determines whether the fund's returns are attributable to the mandate or to market conditions.

Additionality is the second tier. The fund should report the share of its portfolio that would not have been financed without its participation, verified against the additionality statements described in Section 11.2.2. The verification should be independent. A fund that self-reports its additionality will report a high figure, because the incentive to do so is strong and the cost of overstating is low.

Environmental outcome is the third. The fund should report emissions avoided, resources conserved, or environmental benefit delivered by its portfolio, measured against a baseline. The measurement should use the same methodology as the rating mechanism, so that the fund's environmental reporting and the rating's environmental component are consistent. Consistency is what allows the two mechanisms to reinforce each other; without it, the fund reports one number and the rating scores another, and neither can be checked against the other.

11.7.2 The Metrics That Matter

Within each tier a small number of metrics carry most of the information.

Table 11-3: Core Metrics for a Guided Fund

The target ranges are indicative, not universal. What matters is that the ranges are stated in advance and the fund is assessed against them. A fund reporting a 5 percent loss rate against a 25–45 percent target has not outperformed. It has failed to take the risk it was created to take. A fund reporting private leverage of 1:1 has not catalyzed. It has co-invested.

11.7.3 The Reporting Cycle and the Political Economy

The measurement framework interacts with the political system in a way that is easy to miss. A fund that reports a high loss rate will be criticized, whatever its additionality. The criticism is rational from the critic's seat: the loss is visible and the additionality is not. Fund managers, anticipating the criticism, select safer investments, and the mandate dies by a thousand prudent decisions.

The remedy is to make additionality as visible as loss. The annual report should present the additionality figure beside the loss figure, with the same prominence. A 35 percent loss rate paired with a 70 percent additionality rate reads as a fund doing its job. The pairing is not spin. It is the correct frame for assessing a catalytic instrument.

A second remedy is to separate evaluation from the electoral cycle. A fund evaluated annually will optimize for annual results, which means avoiding the long-horizon positions that produce the additionality. A fund evaluated over a rolling five-year window, with interim reporting that does not trigger reallocation, has a better chance of holding its mandate.

11.8 The Limits of the Instrument

A chapter on guided funds that did not state the instrument's limits would be advocacy. The limits are real.

The first is fiscal. A guided fund's capital is finite and the transition's capital requirement is not. The fund can change the terms on which private capital enters; it cannot substitute for private capital at scale. The catalyst is a small quantity by definition.

The second is capability. The fund's performance depends on its ability to assess technology, and that ability is scarce. A country that creates a large guided fund without the capability to deploy it will produce a portfolio of plausible firms with flawed technologies, and the losses will discredit the instrument for a decade. Capability building is a precondition, not a complement.

The third is political. The fund is a public instrument and public instruments are subject to political direction. The design features described here raise the cost of capture; they do not eliminate it. A fund in a political environment that rewards connection over capability will be captured, whatever its charter says. The instrument's effectiveness is conditional on the institutional environment, and the instrument cannot fix the environment.

The fourth is the exit market. The fund can hold a position longer than a private investor, but it cannot hold forever. If the exit market does not develop, the fund's capital locks into its portfolio and stops recycling. The exit market depends on the broader capital market, which is beyond the fund's control. A guided fund in an economy with a thin exit market will underperform an identical fund in an economy with a deep one, and the gap will not be attributable to design.

These limits argue for using the instrument where it works, not against using it. The market failure in early-stage green capital is real, and the guided fund is the instrument that addresses it. The conditions under which it works are the subject of the next two chapters.

11.9 Design Decisions in Sequence

A checklist would duplicate Section 11.2 without adding analysis. What a fund designer actually faces is a sequence of decisions, each constraining the next, and the sequence is worth walking through once.

The first decision is the mandate, and it is made before any money is committed. The designer has to choose between a broad green mandate and a narrow early-stage mandate, and the choice sets the target loss rate. A mandate that names pre-revenue firms in hard-to-abate sectors implies a loss rate of 30 percent or more, and a treasury that will not accept that loss rate should not adopt that mandate. Most of the failures in the international record trace back to this mismatch: a narrow mandate with a loss tolerance calibrated to a broad one, which forces the fund to select safer firms and abandon the additionality it was created to deliver.

The second decision is the structure, and it follows from the first. A narrow early-stage mandate requires technology assessment capability, and the state rarely has it. The fund-of-funds structure with competitively selected specialist managers buys the capability. The direct-investment structure does not, unless the state builds a team that can compete with venture partners for technical talent, which few states can.

The third decision is the loss-sharing calibration, and it is where the political negotiation usually goes wrong. The private partner's first-loss tranche has to be large enough to discipline screening and small enough to change the investment decision. The 10–30 percent band is the workable range, and the deal-level deductible is the mechanism that keeps the private partner honest on individual deals while pooling the tail risk across the portfolio.

The fourth decision is pricing and tenor. Public capital priced at market changes nothing; priced at zero it is a grant. The workable structure is below market and above zero, with a tenor matched to the validation timeline. The recycling rule belongs here too, because a fund that must return capital at the end of its life will behave like a private fund and avoid the positions it was built to hold.

The fifth decision is governance, and it is the one the designer has least control over. An independent investment committee with a professional majority, fixed terms that do not coincide with electoral cycles, published decisions with a lag, and an audit on additionality rather than returns is the minimum. The designer who cannot get these should expect capture and should say so in the charter.

The sixth decision is measurement, and it closes the loop. Three tiers, target ranges stated in advance, additionality reported with the same prominence as loss. A fund that reports a loss rate of 5 percent against a target of 25–45 percent has failed, and the reporting framework should make that failure visible rather than letting it read as prudence.

The seventh decision is the interaction with the other two mechanisms. The fund should use the rating as a screening and pricing input, require insurance where the risk is transferable, and invest with a view of the supply chain rather than individual deals. A fund that ignores the rating cannot measure what it has bought. A fund that ignores insurance bears a risk it cannot diversify. A fund that invests deal by deal forgoes the coordination channel, which is the most powerful one it has.

A designer who works through these seven decisions in order will produce a fund that has a chance of addressing the market failure. A designer who starts with the money and works backward will produce a fund that drifts toward the firms that are easiest to finance, and the drift will be invisible in the reported returns.

11.10 Conclusion: The Catalyst and the Sequence

The market failure in early-stage green capital is a cluster of four failures—information asymmetry, maturity mismatch, coordination, and exit—that compound into a loop. Private capital alone will not break the loop, because breaking it requires financing the validation step that produces the information, and no private actor will pay for information that benefits everyone.

The guided fund is the instrument that finances the validation step. Its design determines whether it resolves the failures or subsidizes the firms that would have been financed anyway. The parameters that matter are the mandate, the loss-sharing structure, the pricing and tenor, the governance, and the recycling rule. The international record shows that the fund-of-funds structure with competitively selected managers outperforms direct investment, that loss sharing must preserve screening incentives, and that scale without design discipline produces mandate drift.

The catalytic function runs through three channels: risk transfer, information production, and coordination. The channels are complementary, and a well-designed fund uses all three. The conditions for catalysis are additionality, patience, capability, and a credible exit. A fund that meets the conditions can change the terms on which private capital enters. A fund that does not will report a low loss rate and achieve nothing.

The fund sits inside the book's three-mechanism sequence. The rating identifies the firm's environmental performance and credit risk. The insurance buffers the risks the rating cannot eliminate. The fund catalyzes the capital the first two mechanisms make investable. The sequence is a loop, and the loop tightens as the data accumulate: the fund's portfolio improves the rating, the improved rating improves the insurance pricing, the improved insurance improves the fund's performance, and the improved performance attracts the private capital that lets the fund recycle.

The Chongqing firm's second line may still not be built. The technology may fail, the offtake may not materialize, the fund may back a different process. What the sequence changes is the reason for failure. Under the system as it stood, the firm would have failed because the financial system could not see it. Under the sequence, it fails, if it fails, because the technology did not work. The loan officer's 540–560 score, the 700,000 yuan premium, the 6.4 million yuan of equity ahead of the bank—these are the marks of a system that let the technology prove itself. The rest is up to the technology.

The next chapter turns to evaluation. A guided fund that is not measured will drift, and the measurement framework is where the design's intentions meet the political system's incentives. The question is not whether to measure but what to measure, and the answer determines whether the fund's capital reaches the firms that need it or the firms that are easiest to finance.