Chapter 10: Sci-Tech Finance Intermediaries and the Catalysis of Green R&D

A perovskite solar cell that passes 1,000 hours of damp-heat testing at 4 cm² will often delaminate at 400 cm². The chemistry is the same. The coating equipment is not. Between the laboratory record and the factory floor sits a stretch of work that no loan officer is trained to underwrite and no grant agency is mandated to fund: the engineering that turns a proven reaction into a reproducible process, the pilot line that turns a reproducible process into a manufacturable product, the first commercial plant that turns a manufacturable product into a business. Most green technologies die in that stretch. The institutions that move them across it are neither banks nor grantmakers. They are sci-tech finance intermediaries (科技金融中介), and their function is to make an innovation legible to capital while making capital patient enough to survive the innovation.

The previous chapter described a pairing that works at the scale where local information is cheap and local risk is pooled. A cooperative monitors its members. A mutual absorbs the shock. A federation aggregates field-performance data that a distant lender cannot see. That pairing handles the risk that sits close to the ground. It does not handle the risk that sits in a pilot line, a test cell, or a first-of-kind plant, because the information needed to price that risk is technical, expensive to produce, and worthless to anyone who cannot read a test report. This chapter is about the institutions that produce and read those reports, and about the conditions under which their work actually moves capital.

Three questions organize what follows. What do intermediaries do in innovation finance that banks and markets cannot do? Through which mechanisms does their work catalyze green R&D, and what do the named cases show about when the mechanisms work? Where do they fail, and what institutional design responds to the failures? The answers matter for the book's central argument because catalysis is the third leg of the integrated mechanism. Rating-based identification without a catalyst produces a label that no project can use. Insurance-based buffering without a catalyst protects adopters who have nothing to adopt. The intermediary is what turns identification and buffering into financeable projects.

10.1 What Intermediaries Do That Banks Cannot

10.1.1 Screening: information that is expensive to produce and cheap to reuse

Consider what a bank needs to know before it extends a working-capital loan. Will the borrower repay within twelve months? The information required is narrow, and the cost of producing it is recovered from a spread of perhaps 200 basis points. Now consider what an investor needs to know before funding a pre-revenue green chemistry. Will the technology work? Will it work at scale? Will a market exist when it does? Will the founding team still be there in five years? Each question requires a different kind of expertise, and the answers decay fast enough that the cost of producing them cannot be recovered from a single transaction.

The economics of that second kind of screening have a specific shape. Once information about a technology's viability exists, it can be reused at near-zero marginal cost. If one investor establishes that a perovskite cell survives 1,000 hours of damp heat, every subsequent investor benefits from the finding. The first producer bears the cost; the followers free-ride. Left to the market, too little screening gets produced. This is the standard justification for public co-investment in technology assessment, and it explains why institutions such as the European Institute of Innovation and Technology fund validation facilities that any startup can use. EIT InnoEnergy has supported more than 500 early-stage sustainable-energy ventures since 2010, and the due-diligence findings from its screening are shared across its co-investors. The sharing is the point, not a side benefit.

Screening does not produce a probability of success. It reduces the variance of the estimate. A pre-screening estimate of a green chemistry venture's five-year survival might carry a standard deviation of 40 percentage points; after screening, 15. The expected value may barely move. What moves is the confidence interval, and with it the willingness of downstream capital to price the risk. The screening intermediary's output is therefore not a loan but a signal, and the signal's credibility depends on the intermediary having its own capital at risk. A due-diligence report from an institution that bears no consequence for being wrong is worth roughly what it cost to produce.

10.1.2 Monitoring: watching a technology, not a balance sheet

Monitoring a green R&D project is not monitoring a borrower's financial statements. The relevant milestones are technical: a pilot line reaching 18% module efficiency, a fermentation titer reaching 80 g/L, a battery cell surviving 1,200 cycles at 80% depth of discharge. These milestones are observable, but only to someone who can read the test report and knows what the numbers mean. A bank can hire that competence for one large loan. It cannot hire it repeatedly for a portfolio of small ones.

The intermediary monitors by holding a seat on the technical advisory board, by tying disbursement to milestones, and by maintaining a standing relationship with the testing laboratory. The cost is borne once and spread across the portfolio. The same logic makes a venture capital firm's monitoring cheaper per dollar than a bank's, and it explains why portfolio size matters: below a certain scale, the fixed cost of technical competence cannot be recovered, and the intermediary either under-monitors or charges a fee the project cannot bear.

Green R&D adds a second monitoring task that deserves separate mention. A technology's claim to greenness is itself a variable that must be watched. A bio-based plastic that composts in an industrial facility but not in a home composter is green under one standard and not under another. An intermediary that finances the technology and also verifies its environmental performance is producing the data that the rating framework of Chapter 4 needs. Monitoring and identification are the same function seen from two ends: the milestone certificate that releases a tranche of funding is also the environmental performance record that a rating model ingests.

10.1.3 Networking: matching a technology to what it lacks

Green technologies rarely fail because the science is wrong. They fail because the wrong team is commercializing them, or the pilot is sited where the feedstock is expensive, or the offtake agreement is with a buyer who will not pay a green premium. The intermediary's third function is to match the technology to the complementary capabilities it lacks: a manufacturing partner, a site, a customer, a regulatory pathway.

This function is hard to price and easy to underprovide. It is also what most sharply distinguishes the sci-tech intermediary from the financial institution. A bank can lend and a venture fund can invest, but neither is organized to broker a partnership between a battery-recycling startup and an automotive OEM. The intermediary that does so is performing a coordination service, and the value it creates is the avoided cost of a failed match. Fraunhofer-Gesellschaft performs this function at scale in Germany: its institutes sit between university research and industrial application, and its contract research model, with roughly €700 million of industry-funded research per year in recent years, is a standing mechanism for matching technical capability to industrial need. Fraunhofer does not finance the match. It makes the match possible, and the financing follows.

10.1.4 Why the three functions cluster

Screening, monitoring, and networking could in principle sit in three separate institutions. In practice they cluster, because the information produced by screening is the input to monitoring, and the relationships formed during monitoring are the raw material of networking. An intermediary that screens but does not monitor loses the information it paid for. One that monitors but does not network wastes the relationships it built. The cluster is an efficiency, not an accident of history.

The clustering also explains why banks do not simply absorb the intermediary function. A bank's screening answers a repayment question, not a viability question. Its monitoring detects deterioration, not milestone completion. Its networking is with other lenders, not with manufacturers. A bank can buy the intermediary's output—a due-diligence report, a milestone certificate—but it cannot easily produce it in-house. That division of labor is what the rest of the chapter examines.

flowchart TD
    A[Green technology project] --> B{Screening intermediary}
    B -->|viability signal| C[Monitoring intermediary]
    B -->|rejection with feedback| A
    C -->|milestone verification| D[Networking intermediary]
    C -->|environmental performance data| E[Rating framework, Ch. 4]
    D -->|manufacturing partner| F[Scale-up capital]
    D -->|offtake match| F
    E -->|green identification| F
    F --> G[Bank debt / green bond / blended facility]
    G -->|repayment and returns| H[Intermediary reinvestment]
    H --> B

The return arrow from H to B is what makes the model work over time. The intermediary does not sit in a linear chain from lab to market. It sits in a cycle in which returns from one cohort of projects fund the screening of the next, and performance data from one cohort sharpen the identification applied to the next. A linear reading of the figure—grant, then venture, then debt—misses the feedback that gives the intermediary its unit economics.

10.2 How the Catalysis Works

10.2.1 Staged capital and the conversion of uncertainty into priced risk

The core mechanism is staging. A green technology's development is decomposed into stages, each with a distinct risk profile and a distinct owner of that risk. The decomposition follows the technology readiness level scale, which runs from TRL 1 (basic principles observed) to TRL 9 (competitive operation in the market). Financing instruments map onto the scale with reasonable consistency across countries, though the boundaries leak.

Read the table's TRL 3–4 row against its TRL 6–7 row and a persistent observation becomes legible: the valley of death is not one valley but two. The first sits where the science is proven and the engineering is not. The second sits where the pilot works, the first commercial plant does not yet exist, and the capital required jumps by an order of magnitude. The instruments needed at each bridge are different, and so are the institutions that supply them.

Translational funding bridges the first valley. It pays for the engineering work that turns a laboratory result into a pilot process. The U.S. Department of Energy's Advanced Research Projects Agency–Energy, established in 2009, is the clearest working model. ARPA-E makes small, time-limited awards, typically $500,000 to $10 million over two to three years, requires milestone reporting, and tracks the follow-on capital its projects raise. By ARPA-E's own accounting, its portfolio companies had raised more than $10 billion in follow-on private capital by the early 2020s against roughly $3 billion of program funding. That ratio is the catalytic multiplier, and it is the number that justifies the public expenditure. Note what ARPA-E is doing in the terms of the first part of this chapter: it screens, it monitors through milestones, and it publishes the follow-on record that lets the next investor free-ride on its screening. The agency is an intermediary with a grant instrument.

First-of-kind finance bridges the second valley, and it operates on different logic. A first commercial plant for a green technology—a green hydrogen electrolyzer factory, a carbon-capture retrofit, a bio-refinery—has a cost profile no venture fund can absorb and a technology risk no commercial bank will underwrite at investment-grade pricing. The instrument that has emerged for this stage is blended finance: a layer of concessional capital from a development bank or green fund that absorbs the first loss, sitting beneath a layer of commercial capital that would not otherwise be deployed. The European Investment Bank's InnovFin program, launched in 2014, provides risk-sharing guarantees and subordinated loans that let commercial lenders finance research-intensive firms at lower cost than they could achieve alone. Its Energy Demo Projects facility targets first-of-kind renewable-energy demonstrations specifically, with individual loans of €7.5 million to €50 million.

Staging works only if the handoffs between stages are reliable. A project that receives translational funding and then cannot find first-of-kind finance has been screened and then abandoned, which is worse than never screening it, because the founders have invested years and the public money is stranded. Handoff failure is the most common failure mode in green innovation finance. The third part of this chapter returns to it.

10.2.2 The signaling cascade and the certification of greenness

An intermediary's willingness to fund a project is observed by other investors, and the observation changes their behavior. The signal is credible only if the intermediary has skin in the game and a reputation to protect. This is why the identity of the first institutional investor in a green technology round matters more than the size of the check.

Green R&D gives the signaling cascade a specific structure, because the signal must certify two things at once: technical credibility and environmental credibility. A technology that works but is not green will not attract green capital. A technology that is green but does not work will not attract any capital. The intermediary that certifies both is performing a dual screening, and the certification is valuable precisely because the two dimensions trade off in ways that are hard to observe from outside. A battery chemistry using abundant materials is green but may have lower energy density. A chemistry with higher energy density may depend on cobalt. The trade-off is the substance of the screening, and the intermediary's certification resolves it for the downstream investor who cannot run the comparison.

The European Union's Innovation Fund, financed by auction revenues from the Emissions Trading System, certifies both dimensions. It funds large-scale demonstration projects for low-carbon technologies, scoring proposals on greenhouse-gas avoidance and on innovation maturity. Its first large-scale call in 2020 awarded €1.1 billion to seven projects; later calls were larger. The certification value lies less in the grant, which is a fraction of project cost, than in the signal that the project passed a technical and environmental screen designed by the European Commission and its external evaluators. That signal lowers the cost of the next layer of capital.

The signaling literature has a tendency to overstate this effect. A signal is only as good as the signaler's track record, and public intermediaries have mixed records. The European Court of Auditors' 2020 review of EU innovation funding found that the Innovation Fund's predecessor, NER300, had disbursed only a fraction of its allocated funds and that project selection had been slow. The signal was credible in principle and weak in practice. The lesson is that the signaling mechanism depends on operational competence, not just on mandate. A public intermediary that cannot disburse on schedule sends a signal that the market learns to discount.

10.2.3 Aggregation: turning small projects into investable pools

Many green R&D projects are too small to finance individually at reasonable cost. A 5 MW distributed solar-plus-storage project, a 10,000-tonne-per-year bio-based chemicals plant, a fleet of 200 electric buses: each is a real project, and each sits below the threshold at which a bond issuance or a syndicated loan makes sense. The intermediary aggregates them into a pool large enough to be investable and diversified enough to be rated.

Aggregation changes the risk profile in a way that is economically real, not a financial trick. A portfolio of 50 small projects with uncorrelated technical risks has a lower probability of aggregate loss than any single project, and the lower probability is what allows the pool to be financed at a tighter spread. The intermediary's job is to construct the pool so the risks are genuinely uncorrelated. A pool of 50 solar projects in the same irradiance zone is not diversified against weather. A pool of 50 bio-based chemical plants using the same fermentation technology is not diversified against technical failure. The arranger who ignores the correlation sells a diversification that does not exist, and the rating that results is wrong in a way that surfaces only in a bad year.

The model that has worked at scale is green securitization of distributed energy assets. In the United States, rooftop solar securitizations issued by SolarCity from 2013 and later by Sunrun and others pooled thousands of residential solar leases and power-purchase agreements into asset-backed securities. The market grew to several billion dollars annually by the late 2010s. The arranger performs the aggregation and the risk tranching; the originator did the underlying screening. The template extends to energy-efficiency retrofits, small-scale biogas, and distributed storage, and its constraint is the availability of standardized performance data—the same data problem Chapter 5 addressed.

China has performed the aggregation function more through policy-guided funds than through securitization. The National Integrated Circuit Industry Investment Fund (国家集成电路产业投资基金), established in 2014 with a first phase of RMB 138.7 billion, is the largest example of a state-guided fund aggregating capital for a strategic technology sector. Its green counterpart is less developed, but the structure—public anchor capital, private co-investment, professional management—is the one Chapter 11 examines in detail. For this chapter the point is narrower: aggregation requires an intermediary with the mandate and the competence to construct the pool, and the pool's investability depends on standardizing the underlying assets.

10.2.4 The Green Climate Fund and the cost of distance

The Green Climate Fund, established in 2010 under the UN Framework Convention on Climate Change, is the largest dedicated multilateral climate finance intermediary. Its mandate is to catalyze low-emission and climate-resilient investment in developing countries, and its instrument is the concessional layer: grants, concessional loans, equity, and guarantees channeled through accredited implementing entities. By the early 2020s it had committed roughly $10 billion across more than 200 projects.

The GCF shows both the power and the limits of the intermediary model at multilateral scale. The power sits in the concessional layer: a GCF grant or guarantee can make a project bankable that would otherwise not be, and the accreditation process certifies the implementing entity's fiduciary and technical competence. The limit sits in the distance between the fund and the project. The GCF does not screen projects directly. It screens accredited entities, which then screen projects. Each layer adds screening cost and delay, and the fund's independent evaluation reports have documented the result: the average time from concept note to first disbursement has been measured in years, not months.

The general lesson concerns the number of layers between capital and project, which is a design variable with real consequences. Too few layers and the intermediary lacks local information. Too many and screening cost and delay overwhelm the catalytic effect. The cooperative and mutual structures of Chapter 9 sit at the low-layer end. The GCF sits at the high-layer end. Where a given technology and market should sit on that spectrum depends on the cost of local information relative to the cost of delay, and the two costs are rarely measured against each other in practice.

10.2.5 China's technology credit score: strong screening, weak monitoring

China has run a sustained experiment in sci-tech finance intermediation since the 2010s, and its most concrete instrument is the technology credit score (科技信用评分). Developed in pilot zones first designated in 2016 across nine regions, including Beijing's Zhongguancun, Shanghai's Zhangjiang, and Wuhan's East Lake, the score combines a technology enterprise's intellectual-property portfolio, R&D expenditure, and government recognition status into a number a bank can use to underwrite a loan without conventional collateral.

The scale is significant. By the early 2020s the score had been applied to tens of thousands of firms, and technology-based loans outstanding in the pilot zones had reached hundreds of billions of RMB. The mechanism is an intermediary function performed by a public agency, the local science and technology bureau, on behalf of a private lender. The agency produces the screening information; the bank uses it to price and monitor. That division of labor is exactly the one described earlier in this chapter.

The instrument has a documented limitation. The score is a point-in-time assessment and does not track the technology after the loan is made. A firm can score well on its patent portfolio and still fail on its first commercial plant. The pilots have responded by adding post-loan monitoring through the same public agencies, but that monitoring is administrative rather than technical. It does not substitute for the milestone verification a specialist intermediary would perform. The technology credit score is a strong screening instrument and a weak monitoring one, and the asymmetry is instructive: the same public agency that can produce a score from existing administrative data cannot easily produce a test report from a pilot line.

10.2.6 The contract research organization as intermediary

Not every sci-tech intermediary is a financier. The contract research organization sells R&D services and, in doing so, performs screening and monitoring for the investors who rely on its results. In pharmaceuticals the model is mature; in green technology it is emerging.

The relevant examples are the applied research institutes that test and certify green materials and processes. The Netherlands Organisation for Applied Scientific Research, TNO, performs this function for the Dutch construction and chemicals sectors; its testing programs generate the performance data that lenders and insurers use to underwrite green building materials. Underwriters Laboratories certification for solar modules and battery storage performs the same function in the U.S. market: a UL listing is a screening signal a bank can rely on without running its own technical due diligence. The certifier's product is the reduction of information asymmetry, and it is sold by the test.

The CRO model solves the monitoring problem at lower cost than equity investment. A venture fund monitors by holding a board seat and reading reports; a CRO monitors by running the test. The test result is verifiable by third parties, which makes it a better signal than a board member's judgment. The limitation is that the CRO monitors only what it is paid to test, and choosing what to test is itself a screening decision the financier must make. A certification regime that tests the wrong parameter produces a confident signal about the wrong thing.

10.2.7 Catalyzing demand, not just supply

The mechanisms so far operate on the supply of capital. A fifth operates on demand, and it is the one most often neglected in policy discussion.

A green technology's commercial viability depends on a market for its output. A bio-based chemical costing 30% more than its petroleum equivalent will not sell without a carbon price, a green premium from buyers, or a mandate. The intermediary can catalyze demand by aggregating buyers, negotiating offtake agreements, and certifying the green attribute that justifies the premium. Corporate renewable-energy procurement intermediaries do exactly this: they bundle the electricity demand of multiple corporates into a power-purchase agreement large enough to finance a new wind or solar farm. In the United States, the aggregation model developed by companies such as LevelTen Energy let smaller corporates participate in procurement previously available only to the largest buyers.

The demand-side mechanism is catalytic in a precise sense. It converts a willingness to pay a green premium into a bankable revenue stream. Without aggregation, the premium is a diffuse preference no developer can capture. With it, the premium becomes a contract, and the contract becomes collateral. That is the mechanism, and it is enough: the offtake agreement is what the lender discounts, and the intermediary's work is done when the signature is on the page.

10.3 Failures and Institutional Design

10.3.1 The handoff problem

The most consequential failure in green innovation finance is the handoff between stages. A project screened and funded through TRL 6 cannot find first-of-kind finance at TRL 7, and the capital already invested is stranded. The failure is structural. The institutions that fund early stages—public agencies, angel investors, corporate venture units—are not the institutions that fund later stages—development banks, infrastructure funds, commercial lenders—and the two sets have different risk appetites, return requirements, and time horizons.

The handoff problem has a measurable signature. The International Energy Agency's tracking of energy-technology demonstration projects has repeatedly found the pipeline of first-of-kind projects thinner than the pipeline of pilot projects, with the gap widening for capital-intensive technologies. That gap is the valley of death, and it persists despite decades of policy attention because the institutions on either side have no incentive to bridge it. A grant agency that funds a pilot has discharged its mandate when the pilot reports. A development bank that funds a first plant needs a project that already looks bankable. Neither is rewarded for the project that falls between them.

The design response that has worked is an institution whose mandate spans the seam. The EIB's InnovFin Energy Demo Projects facility provides loans at TRL 7–8, the stage commercial banks avoid and venture funds cannot reach. The U.S. Department of Energy's Loan Programs Office performs the same function for the U.S. market, with a portfolio that has included first commercial-scale solar thermal plants, first advanced nuclear reactors, and first commercial cellulosic biofuel plants. The LPO's history is instructive in both directions. It suffered a high-profile failure when Solyndra, which received a $535 million loan guarantee, collapsed in 2011, and it supported a series of utility-scale solar projects that performed. The political controversy over the failure obscured the structural point: the office exists because the handoff would otherwise not happen, and the correct measure of its performance is the portfolio return, not the single failure. A handoff institution that never has a failure is not taking the risk it was created to take.

10.3.2 Measuring what counts as catalytic

An intermediary claims to be catalytic when its capital mobilizes additional capital that would not otherwise have been deployed. The claim is hard to verify because the counterfactual is unobservable. If a project would have been financed anyway, the intermediary's capital is not additional; it is a subsidy to an investor who did not need it.

Additionality is well known in development finance, and the multilateral development banks have built indicators around it. The most widely used is the mobilization ratio: private capital mobilized per dollar of public or concessional capital. Reported ratios vary widely by instrument. Guarantees mobilize more per dollar than direct loans; direct loans mobilize more than grants. The variation is informative. A guarantee mobilizes more because it changes the risk profile of the private investor's position rather than substituting for it. The design implication is that the catalytic instrument should be chosen for its leverage, not for its visibility. A ribbon-cutting on a grant photographs better than a guarantee, and the guarantee moves more money.

Green R&D adds a second dimension: environmental additionality. A project reducing emissions by 100,000 tonnes of CO₂ per year is additional only if the reduction would not have occurred without the intermediary's support. The accounting protocols developers use—the CDM methodologies, the Gold Standard, the Verra standards—are designed to establish this, and their track record is mixed. A substantial share of CDM projects were found in ex-post evaluations to have questionable additionality. The lesson for the sci-tech intermediary is that environmental additionality has to be built into screening criteria before the fact, not asserted after it. A baseline written after the project is commissioned is not a baseline.

10.3.3 Mission drift and capture

An intermediary with a public mandate and private co-investors faces two opposite governance failures. The first is mission drift: the intermediary gradually shifts toward projects that are profitable but not green, or green but not innovative, because those projects are easier to underwrite. The second is capture: the screening criteria are shaped by the incumbent firms the intermediary is meant to disrupt, and the disruptive technologies are screened out.

Mission drift is documented in the green bond market. Growth has come with persistent concern about greenwashing—bonds labeled green whose proceeds finance projects with marginal environmental benefit. The European Union's Green Bond Standard, proposed in 2021 and anchored to the Taxonomy Regulation, responds by requiring that proceeds be allocated to activities meeting the taxonomy's technical screening criteria and by requiring external verification. The standard is an attempt to fix the intermediary's screening criteria by regulation. Whether it works depends on the competence of the verification market, which is itself an intermediary and subject to the same incentive problems.

Capture is less documented and more consequential for R&D. An intermediary governed by a board dominated by large incumbents will tend to fund incremental improvements to existing technologies rather than disruptive alternatives. Germany's Energiewende support for rooftop solar is a case where policy favored a distributed technology over the incumbent utility model, and the incumbent utilities fought it through the political process and the courts. The design response is to constitute the board with a mandate that includes the disruptive alternatives and to separate the screening function from incumbent-dominated industry associations. Neither is easy, because the incumbents have the expertise the intermediary needs and the disruptors do not yet have the standing.

10.3.4 The missing performance record

Screening and monitoring depend on a performance record that, for most green technologies, does not yet exist. A bank underwriting a conventional loan can rely on decades of default data by rating grade. An intermediary underwriting a green technology has a sample of perhaps a few dozen commercial-scale installations, and the sample is contaminated by survivorship bias because failures go unreported.

This is the binding constraint on extending the rating framework of Chapter 4 to green technologies. The framework needs a performance history to calibrate the relationship between environmental performance and credit risk, and the history is thin. The institutional response has been to build the record deliberately. The demonstration projects funded by the Innovation Fund and the LPO are, in effect, data-generation exercises, and the requirement that they report performance for a decade after commissioning matters as much as the financing. A demonstration project that is not instrumented is a subsidy, not an experiment.

The cooperative and mutual structures of Chapter 9 contribute to the record at field level. A cooperative monitoring member-level green practices produces the field-performance data the intermediary needs, and the intermediary that can access those records has a screening advantage over one that cannot. That is the data bridge between the local buffer and the innovation catalyst, and it is the reason the two layers should be designed together. A cooperative that reports into a standard format gives the rating model its calibration sample; a rating model that publishes its criteria tells the cooperative what to measure.

10.3.5 Five design principles

The failures point to a set of design principles. They are stated as principles rather than recommendations because the evidence supports the direction more strongly than the magnitude.

An intermediary whose mandate stops at the pilot will strand its projects at the seam, so the mandate should span the handoff into first-of-kind finance, with the instrument at that stage sized to a capital requirement typically an order of magnitude larger than the pilot's. InnovFin and the LPO are the working models, and both were created because the private market would not bridge the seam on its own.

Instrument choice should follow leverage. A guarantee mobilizes more private capital per public dollar than a direct loan, and a subordinated tranche mobilizes more than a grant. The instrument mix should tilt toward the higher-leverage tools, and reporting should measure mobilization rather than disbursement, because disbursement measures activity and mobilization measures effect.

Environmental additionality belongs in the screening criteria, quantified against a stated baseline before the project is approved, with verification independent of the developer. The EU Green Bond Standard's verification requirement is the template, and the CDM experience is the cautionary case for what happens when additionality is assessed after the money is spent.

Governance should include the disruptors. A board dominated by incumbents will fund incremental technologies, so the design should reserve seats or voting rights for the technology providers the intermediary exists to support, and the screening criteria should be published so the selection can be contested. Published criteria are uncomfortable for the institution and useful for the market.

Finally, the performance record should be treated as a public good. Data from publicly supported demonstration projects should be published in a standardized form, because the screening value of the data is non-rival and the private incentive to publish is weak. The certifier model described earlier is the template: the test result is published, and every subsequent investor benefits from a test it did not pay for.

10.3.6 The boundary

The intermediary can make an innovation legible to capital. It cannot make the innovation profitable when the market does not exist. A green technology whose cost is 40% above the incumbent's, with no carbon price and no mandate, will fail regardless of how well it is screened and monitored. Catalysis operates on the supply of capital and on the coordination of demand. It does not substitute for the policy that creates demand, and an intermediary asked to perform that substitution will fail in a way that discredits the model.

This boundary links to the next chapter. The guided funds that supply early-stage green capital are Chapter 11's subject, and their design problem is the one identified here: how to allocate public capital to technologies whose private returns are uncertain without distorting the investment decisions of private co-investors. The intermediary performs the allocation; the guided fund supplies the pool. Designed together, they can carry a project across both valleys. Designed separately, the fund commits capital and the intermediary has nothing to move.

The chapter's claim, stated compactly: green R&D is financed through a sequence of stages, each with a different risk profile and a different owner of that risk, and the sci-tech finance intermediary is the institution that moves a project from one stage to the next by screening it, monitoring it, and matching it to the capabilities it lacks. The catalysis is real and bounded. It works when the mandate spans the handoff, the instrument is chosen for leverage, additionality is verified before the fact, governance includes the disruptors, and the performance record is treated as a public good. Where those conditions fail, capital does not flow and the valley of death persists.

One number from the cooperative layer makes the connection concrete. A field-performance record covering three seasons of a cover-crop practice is enough to calibrate a rating model's first environmental variable, and the cooperative that holds the record holds the raw material the intermediary needs. The rating framework converts that record into a price. The insurance layer keeps the adopter solvent through the weather shocks that would otherwise swamp the return. The intermediary moves the resulting project across the seam that no single instrument can cross. The next chapter turns to the guided fund as the supply-side institution that makes the catalysis possible at scale.