Chapter 9: The Integrated Mechanism: Synergies and Complementarities among Rating, Insurance, and Sci-Tech Finance

In the third year of a provincial green industry fund, an investment committee approved a follow-on round for a solar-thermal startup on the strength of a business plan and an emissions-avoidance estimate prepared by the fund's own analysts. Fourteen months later the company's manufacturing line was shut by a hazardous-waste violation the fund had never been told about, and the cleanup consumed the remaining capital. The fund's mandate had mentioned green transition on every page. It had commissioned no rating, required no insurance, and audited no environmental figure. Each of the three instruments the previous six chapters describe existed somewhere in the province. None of them reached this deal.

That failure is the starting point for this chapter because it is ordinary. The instruments are not scarce. Credit rating agencies have built environmental assessment into their methodologies; weather index and forest insurance products have been piloted across a dozen provinces; guided funds and venture capital have money to place. What is scarce is the linkage—the specific transfers of information and risk that make each instrument more accurate and more affordable than it would be alone. Integration consists of information transfers and contractual linkages that can be specified, audited, and priced. A rating produces a default probability that an insurer can price against. An insurance policy removes a tail risk that a rating would otherwise have to penalize. A venture fund supplies the equity cushion and the disclosure discipline that let both operate on firms too young to have a track record. Remove one leg and the other two do not collapse. They degrade, in ways that show up in loan pricing and in loss experience.

The chapter works through the mechanism in four movements. Section 9.1 examines the rating–insurance interface, where the two instruments exchange information about risk and change the lender's loss distribution. Section 9.2 places sci-tech finance inside the triad as the catalytic leg that converts a rated, insured project into an investable one. Section 9.3 works through four cases where the three instruments were combined, or where their absence produced the familiar failure. Section 9.4 states the conditions under which integration works, and Section 9.5 marks its boundaries.

9.1 Rating and Insurance: Two Instruments That Read Each Other

9.1.1 The information asymmetry both instruments address

Credit rating and insurance answer the same question from opposite sides. The lender cannot observe the borrower's risk, and the borrower has no incentive to reveal it. A rating agency compresses unobservable risk into a comparable ordinal scale—S&P's ratings track probability of default, Moody's expected loss, which is default probability multiplied by loss given default. An insurer prices a contractually defined contingency and transfers the residual to a portfolio diversified across many policyholders. Both are information technologies. They observe different things.

For green finance the difference is decisive. A rating agency reads the balance sheet, the cash flow, and, where it has built the capability, environmental performance as a factor in transition risk. An insurer reads the physical hazard: rainfall deficiency, fire probability, wind speed, seedling survival. A tea cooperative can be financially sound and physically exposed. A forest plantation can be physically secure and financially fragile. Integration begins with the admission that each instrument sees half the risk.

Take a single agricultural loan. Without insurance the lender's loss given default depends on the borrower's cash flow, which in a drought year approaches zero, and on collateral, which in a rural credit market is often land with thin resale value. Expected loss is

with \(PD\) the probability of default, \(LGD\) loss given default, \(EAD\) exposure at default. Insurance changes \(LGD\) directly, substituting an insurer's payment for the missing cash flow, and \(PD\) indirectly, because the borrower's repayment capacity is stabilized. A weather index policy that pays when rainfall falls below 60 percent of the historical average converts a binary default event into a partially recovered loss. If the policy covers 70 percent of expected revenue in the trigger year, effective \(LGD\) falls from roughly 0.9 to roughly 0.3. The borrower is not more creditworthy in any fundamental sense. The loss event has been truncated.

Here is the arithmetic that makes the interface productive. The insurer's actuarial model produces a trigger probability and a payout schedule. The rating agency can use that schedule to model a conditional default probability instead of an unconditional one. Neither institution needs to share confidential borrower data. They need to agree on the structure of the contingency, and that agreement has to be written down somewhere both can point to.

9.1.2 What rating contributes to insurance pricing

The flow runs both ways. Insurers underwrite physical risk, and they also carry counterparty risk—the chance that the policyholder, or the entity whose obligation backs the policy, fails. In green finance this exposure is not trivial. A forest policy written against a plantation company's harvest is only as good as the plantation company's solvency. If the company defaults before the policy matures, the insurer faces a claims dispute with a receiver.

A rating gives the insurer a way to price that counterparty exposure and to set policy limits. There is a second contribution that insurers rarely discuss in public. Agencies that have folded environmental performance into their assessments, as Chapter 3 describes, produce a signal insurers can use to segment policyholders. A technology firm with a strong environmental score and stable cash flow is a different underwriting proposition from a firm with identical physical exposure and a weaker balance sheet. The rating does not replace the actuarial model. It conditions the insurer's appetite.

Formally, let the insurer's expected cost on a policy be

where \(\pi\) is the trigger probability, \(\bar{L}\) the average payout conditional on trigger, \(\lambda\) a loading factor for capital and expenses, and \(c\) fixed administrative cost. The credit rating enters through \(\lambda\). A policyholder with a weak rating requires more capital held against counterparty risk, so \(\lambda\) rises and the premium rises with it. This is why the same rainfall trigger can carry different premiums for two cooperatives in the same district. The physical risk is identical. The counterparty risk is not.

Insurers seldom state the linkage explicitly. They embed it in underwriting guidelines and in the policy limits they will offer to firms below a rating threshold. The integration is real and implicit, which is a design weakness. When the linkage is undocumented, the policyholder cannot see why the premium differs, and the rating agency cannot see how its assessment is being used. Both lose the feedback that would improve the next round of assessment.

9.1.3 The combined effect on the lender's portfolio

The lender sits at the intersection. A bank making a green loan to an agricultural cooperative faces three risks: the borrower's operational risk, the physical hazard, and basis risk—the chance that the index triggers without a loss, or a loss occurs without a trigger. Rating addresses the first. Insurance addresses the second and, partially, the third. The combination changes the loan's risk weight and therefore its capital cost.

Suppose a bank holds \(n\) green loans, each with exposure \(E_i\), default probability \(PD_i\), and loss given default \(LGD_i\). Without insurance the portfolio's unexpected loss is approximately

where \(\sigma_i = \sqrt{PD_i(1-PD_i)} \cdot LGD_i\) and \(\rho_{ij}\) is the default correlation between loans. Insurance reduces \(\sigma_i\) by truncating the loss distribution. It reduces \(\rho_{ij}\) because the weather events that hit every borrower in a district are exactly the events the index policy pays out on. The correlation term is where the largest gain sits. A drought that defaults every cooperative in a county is a correlated loss for the bank. A drought that triggers every policy in the county is a diversified insurer's problem.

That is the analytic case for bundling. The empirical case is thinner, and it is worth saying so plainly. Few Chinese banks have the data systems to estimate \(\rho_{ij}\) at loan level, and fewer have integrated insurance payouts into their loss-given-default models. The mechanism is understood in principle and partially implemented in practice. Chapter 10 takes up the cases where it has been implemented.

9.1.4 Where the interface breaks

Basis risk is the first failure mode, and it deserves more than a sentence. An index policy that triggers on district-level rainfall may not pay a cooperative whose farm sits on a slope that sheds water. The rating then overstates the protection, and the lender prices the loan as if insured when it is only partly insured. Basis risk is not a reason to abandon index insurance. It is a reason to design the trigger at a finer spatial resolution, to disclose the residual exposure in the loan file, and to price it. The tea cooperative in Section 9.3.1 shows what happens when all three steps are skipped.

Two further failure modes are worth naming. The rating–insurance horizon mismatch is the second: a credit rating typically looks one to three years ahead, a weather index policy is annual, a forest policy may run five to ten years, and the lender's exposure may span all three. When the horizons do not align, protection lapses before the loan matures and the rating that assumed protection goes stale. The third is incentive. A borrower with full insurance has less reason to mitigate the hazard; a borrower with an index policy has less reason still, because the payout depends on the index and not on the loss. Rating agencies have begun to ask whether environmental mitigation is contractually obligated or merely asserted. The answer determines whether the rating reflects the insured state or the uninsured one. Where the obligation is absent, the integration is nominal.

9.2 Sci-Tech Finance as the Catalytic Leg

9.2.1 Why the triad needs a third leg

Rating and insurance together can make a mature, asset-backed green project bankable. They cannot make a pre-revenue technology firm bankable, because there is no cash flow to rate and no physical asset to insure. The firms that develop the technologies the transition depends on—next-generation photovoltaics, grid-scale storage, precision irrigation, bio-based materials—spend years in a state where their main asset is a patent and their main liability is a burn rate. Single-channel debt cannot reach them. This is the gap sci-tech finance fills, and it is why the triad is a triad rather than a pair.

The catalytic role is specific. Venture capital and policy-guided funds supply equity that absorbs losses without triggering default, and they supply governance and technical assistance that raise the firm's probability of reaching commercial scale. They also supply a signal. The fact of a reputable fund's investment reduces the information asymmetry facing later lenders and insurers. The signal is not a rating, but it performs a related function earlier in the firm's life.

Sequence matters. In a well-designed ecosystem, sci-tech finance enters first, rating enters when the firm has a revenue history, and insurance enters when the firm has an asset or a physical exposure worth transferring. The instruments are staged, not simultaneous. Integration means the stages are linked, so that information generated at each stage is available to the next.

9.2.2 The staging model

The sequence can be drawn as a set of handoffs.

flowchart TD
    A[Technology firm, pre-revenue] --> B[Sci-tech finance: equity, guided fund]
    B --> C[Commercial scale-up, revenue history]
    C --> D[Rating: environmental performance integrated]
    D --> E[Bank credit at risk-adjusted price]
    C --> F[Asset formation: plantation, equipment, inventory]
    F --> G[Insurance: weather index, forest, equipment]
    G --> D
    E --> H[Green transition investment]
    G --> H
    B --> H

The arrows from insurance back to rating and from rating to credit are the integration channels. The arrow from sci-tech finance to commercial scale-up is the catalyst. The diagram is deliberately simple, and the actual sequence varies by sector. A plantation company may form insurable assets before it has a rateable revenue history. A software firm may never form insurable assets and rely on rating and equity alone. The general point holds: each instrument covers a stage of the firm's life the others cannot reach.

The handoff from sci-tech finance to rating is the weakest link in most ecosystems. Venture-backed firms often emerge from their growth phase with a cap table opaque to a rating agency, a revenue history two or three years long, and an environmental claim that has never been audited. The agency either declines to rate or rates on financials alone, which defeats the purpose. The fix is not to ask venture funds to produce ratings. It is to require, as a condition of public co-investment, that portfolio companies adopt a standardized environmental disclosure from an early stage. Chapter 8 discussed the incentive design. Here the point is narrower: the disclosure requirement is what makes the later rating possible.

9.2.3 The catalytic multiplier

The catalytic effect can be stated as a multiplier. A unit of public sci-tech finance is meant to mobilize private capital beyond its own contribution. If a guided fund invests \(G\) across a portfolio of green technology firms, and each unit of \(G\) attracts \(\mu\) units of private equity and \(\kappa\) units of subsequent debt, total mobilized capital is

What determines \(\mu\) and \(\kappa\) is the policy question. The literature on government-guided funds is not encouraging. Many funds report a headline leverage ratio that counts co-investment commitments rather than disbursed capital, and few report the \(\kappa\) term at all. The integrated mechanism gives a reason why \(\kappa\) should be higher when portfolio companies are rated and insured: the later-stage lender faces less information asymmetry and less tail risk. That reason is a hypothesis until the data are assembled.

A second multiplier runs through the rating channel. A fund that requires environmental disclosure from its portfolio companies generates a dataset rating agencies can use to calibrate environmental assessments. The dataset is a public good. No single fund has an incentive to produce it at the socially optimal scale. This is the standard argument for public co-investment with disclosure conditions attached, and it is the strongest argument for the guided-fund model that does not rest on the fund's direct returns.

9.2.4 The risk of catalytic overreach

An objection deserves a hearing. The catalytic role is often invoked to justify public investment in firms private markets have declined to fund. Sometimes the decline reflects a market failure: the information asymmetry is genuine, the technology is real, the social return exceeds the private return. Sometimes the decline is a correct assessment. The integrated mechanism does not resolve the ambiguity. It makes the assessment more informed. A fund that invests in an unrateable, uninsurable firm and calls the investment catalytic has integrated nothing. It has substituted public capital for a missing market and labeled the substitution a strategy.

The discipline that prevents overreach is the discipline Chapter 8 described: audited environmental performance, enforceable milestones, compensation tied to outcomes. In the integrated framework the rating and insurance stages reinforce it. A firm that cannot reach the rating stage within a defined period is a firm whose commercial prospects are doubtful, and the fund should be able to exit or write down without political cost. A firm that reaches the rating stage but cannot obtain insurance for its physical assets carries an unmanaged exposure, and follow-on investment should be conditioned on resolving it.

9.3 Case Illustrations of the Integrated Mechanism

9.3.1 The tea cooperative: insurance without rating

The first case is partial, because partial cases are the most common and the most instructive.

A tea cooperative in a mountainous county of southwestern China farms roughly 400 hectares across 260 member households. It sells fresh leaves to a processing enterprise under an annual contract. In a drought year the leaves are smaller and yield falls by 30 to 50 percent. The cooperative has no irrigation and no storage. Its only financial buffer is the processing enterprise's advance payment, which is itself contingent on the harvest.

The cooperative bought a weather index policy tied to a rainfall station 12 kilometers away. The trigger was cumulative rainfall below 55 percent of the ten-year average during the March–May growing season. The provincial agricultural insurance program subsidized 70 percent of the premium; the cooperative paid the rest. In the policy's first year, rainfall fell to 61 percent of average—below the agronomic threshold for a bad harvest, above the insurance trigger. Yield fell by a quarter. The policy paid nothing.

Basis risk in its purest form. The index was calibrated to the district, the loss occurred at the farm, and the gap between the two swallowed the protection. The cooperative's loan from the rural credit cooperative went into arrears, and the lender—which had not integrated the policy into its loss-given-default model—treated the loan as uninsured.

Three changes would have altered the outcome. A rating that incorporated both the cooperative's financial history and its physical exposure would have flagged basis risk as a material weakness; the rating would have come in lower than the balance sheet alone suggested, and the loan would have been priced higher or sized down. An insurance design with a finer trigger—a grid of rainfall stations, or a remote-sensing index calibrated to the cooperative's own fields—would have narrowed the gap. And the lender's model would have recognized that even a policy that did not pay in year one reduced the probability of a catastrophic loss in year three.

The general point: insurance without rating does not improve credit access, because the lender cannot verify that the protection is adequate. Rating without insurance does not improve credit access either, because the lender still carries the tail. The two have to be designed together, and the design has to reach the lender in a form the lender's model can consume.

9.3.2 The forest plantation: rating and insurance in sequence

A state-owned forest farm in the northeast converted 8,000 hectares of degraded secondary forest into a mixed plantation of larch and Korean pine. The investment horizon is 25 years. The farm bought a forest insurance policy covering fire, windstorm, and pest infestation, with a payout schedule tied to stand age and timber volume. It also obtained a credit rating from a domestic agency, which assessed cash flow from thinning operations and environmental performance against the provincial forest carbon standard.

The sequence was the reverse of the tea cooperative's. The farm had a long operating history, a rateable cash flow, and an insurable asset. The rating came first, because the farm needed a bank loan to finance planting. Insurance came second, as a condition of the loan covenant. The lender required the policy to name the bank as loss payee and to maintain coverage at replacement value for the life of the loan.

Integration worked because the instruments were contractually linked. The loan agreement specified the insurance requirement, the rating trigger for the interest-rate step-up, and the reporting obligations that kept both current. When a windstorm damaged 300 hectares in the seventh year, the insurer paid within 60 days, the farm replanted with the proceeds, and the loan stayed current. The rating agency reviewed the event and confirmed the rating, noting that the insurance recovery had prevented a default.

The covenant structure carries the lesson. The loan agreement did not merely require insurance. It required insurance with a named loss payee, a defined coverage level, and a reporting schedule. These plumbing details determine whether integration is real. A loan that requires insurance but does not verify the policy's terms has a checkbox, not a risk transfer.

The forest case also shows the model's limits. The farm's rating rested on timber revenue and carbon credits. Carbon revenue was small and uncertain, and the agency treated it conservatively. Had the farm been a pure carbon project with no timber revenue, it would have been unrateable on a conventional scale and the loan would not have been made. The integrated mechanism can extend credit to projects with mixed revenue streams. It cannot manufacture a cash flow where none exists.

9.3.3 The battery-materials startup: the full triad

The third case is a privately held firm producing cathode material for lithium-ion batteries. Two materials scientists founded it, a provincial guided fund and two private venture funds financed its first three years, and it reached commercial production in its fifth year. At that point it sought bank credit to expand capacity.

Its path through the triad was explicit. The guided fund's investment agreement required annual environmental disclosure, including energy intensity per ton of output and the hazardous waste stream. A third-party verifier audited the disclosure from the second year. When the firm sought a rating in its fifth year, the agency had four years of audited environmental data, a revenue history, and a supply contract with a major battery manufacturer. The rating folded the environmental data in as a factor in transition risk: the firm's energy intensity sat below the industry median, which reduced its exposure to a carbon price.

The bank loan carried an insurance component less familiar than weather or forest cover—a credit insurance policy on the firm's receivables from the battery manufacturer, covering 80 percent of the receivable if the customer became insolvent. This is the trade-credit insurance described in the trade-finance literature, applied to a green supply chain. It cut the bank's loss given default on the working-capital portion of the loan and let the bank lend against receivables at a lower rate.

The triad operated in sequence. Sci-tech finance carried the firm through the pre-revenue and early-revenue stages, when neither rating nor insurance was possible. The disclosure condition attached to public co-investment generated the data that made the later rating credible. The rating made the bank loan possible at a price the firm could afford. The receivables policy removed the concentration risk that would otherwise have raised the loan's risk weight. The rating let the bank price the loan; the receivables policy cut the risk weight.

A worked example makes the magnitudes concrete. The figures below are hypothetical, constructed from the deal's structure rather than drawn from a published dataset, and the assumptions are stated so a reader can substitute their own. Assume the firm's unsecured borrowing rate would have been 9.5 percent. With a rating two notches above the sector median and receivables insurance covering 80 percent of the working-capital exposure, the effective rate was 6.8 percent. On a 200 million yuan facility the annual saving is 5.4 million yuan, roughly the cost of the environmental verification and the insurance premium combined, with a margin. The saving is not a subsidy. It is the price of reduced information asymmetry, and it would shrink or vanish if either the rating or the policy were removed.

9.3.4 A counter-case: the fund without instrumentation

The final case returns to the fund from the chapter's opening, because it shows what happens when the triad is invoked rhetorically but not built.

A provincial green industry fund with 5 billion yuan under management invested in 14 companies over three years. Its mandate mentioned green transition. Its investment committee reviewed financial projections and a slide on environmental benefits. No portfolio company was rated. None obtained insurance against physical or counterparty exposure. Environmental reporting consisted of an annual estimate of emissions avoided, prepared by fund staff from company-supplied figures that were never audited.

Two of the 14 failed. A solar-thermal startup had working technology, but its manufacturing process generated a hazardous waste stream the company had not disclosed to the fund; the cleanup cost consumed the remaining capital. An agricultural processing firm's plant was destroyed by flooding; it carried no insurance, and the fund's equity was wiped out. In neither case did the environmental mandate produce a risk assessment that would have flagged the exposure.

The counter-case is not an argument against guided funds. It is an argument against guided funds that treat the triad as a label. This fund had the resources to require disclosure, commission ratings, and mandate insurance. It did none of the three because its compensation and reporting were not tied to risk outcomes. The design failure is the one Chapter 8 identified: the fund's principals could not observe whether the fund was managing risk, so the fund did not manage risk.

9.4 The Mechanism Stated Formally

The mechanism can be stated as four conditions. They are not independent, and the order in which they bind varies by sector.

The first condition is that information flows from the earlier stage to the later stage. Environmental disclosure required by the sci-tech fund must be in a form the rating agency can use. The rating must be in a form the insurer and the lender can use. The insurance policy terms must be in a form the lender's loss model can consume. Each handoff requires a standard, and the standard must be specified before the instrument is deployed. The battery-materials case shows the handoff working: four years of audited energy-intensity and waste data went into the rating, and the rating's sector-relative conclusion went into the loan file. The counter-case shows the handoff missing: the fund's emissions estimate was prepared by its own staff from unaudited company figures and could not be used by anyone.

The second condition is that risk is transferred, not merely asserted. The insurance policy needs a trigger that correlates with the lender's loss, a payout that arrives in time, and a loss-payee arrangement that gives the lender control of the proceeds. The rating needs audited data behind it, not management projections. The venture fund's environmental claim needs third-party verification. The tea cooperative's policy failed this condition on the trigger dimension: the index was calibrated 12 kilometers from the farm, and the correlation between trigger and loss was too weak to protect the lender.

The third condition is that incentives are aligned across the stages. Fund manager compensation must depend on portfolio companies reaching the rating and insurance stages, not only on financial returns. The rating agency must be paid in a way that does not depend on the rating outcome. The insurer must retain enough of the risk to have a reason to underwrite carefully. This condition is the hardest to satisfy because it reaches into the compensation contracts of three separate institutions, and it is the one the counter-case violated most completely.

The fourth condition is that the stages are sequenced, not simultaneous. Sci-tech finance enters when the firm is unrateable. Rating enters when the firm has a revenue history. Insurance enters when the firm has an insurable asset or exposure. Attempting all three at once produces the counter-case: a fund that rates nothing, insures nothing, and reports estimates.

The four conditions interact. The first is a precondition for the second. The third is a precondition for the first, because no actor will invest in producing standardized information if the compensation system does not reward its use. The fourth recognizes that the instruments have different data requirements, and that forcing them onto the same timetable produces nominal compliance.

The mechanism's effect on the cost of capital can be decomposed. Let \(r_0\) be the unsecured lending rate for a firm with no rating and no insurance. The integrated mechanism produces a rate

where \(\Delta r_{rating}\) is the reduction from lower information asymmetry, \(\Delta r_{insurance}\) the reduction from lower loss given default, and \(\Delta r_{catalyst}\) the reduction from the governance and technical assistance the venture investor supplies. The terms interact in practice and are not strictly additive, but the decomposition identifies which channel is doing the work. In the battery-materials case \(\Delta r_{rating}\) was the largest term, because the firm's environmental performance was genuinely better than the sector median and the rating made that visible. In the forest case \(\Delta r_{insurance}\) was the largest, because the physical hazard dominated. In the tea cooperative none of the terms was large, because basis risk defeated the insurance and the cooperative had no rating.

9.5 Complementarities and Substitutes

The triad is sometimes described as if the three instruments were substitutes, any one of which could support the transition. The cases argue otherwise. Each instrument covers a risk the others cannot, and the complementarity is strongest where the risks are correlated.

Consider the correlation structure. A green technology firm faces technology risk, market risk, and physical risk. Sci-tech finance absorbs technology and market risk through equity. Rating addresses market risk and, where environmental performance is integrated, transition risk. Insurance addresses physical risk and counterparty risk. The three risks are not independent: a drought that raises agricultural prices can improve a precision-irrigation firm's market while destroying its customers' ability to pay. The integrated mechanism handles this correlation because the insurer's payout to the customer and the rating agency's assessment of the firm's customer concentration are both visible to the lender. A single instrument sees one side.

Substitutability exists at the margin. A firm with a strong rating can borrow without insurance if its balance sheet can absorb the tail. A firm with comprehensive insurance can borrow without a rating if the insurer's underwriting is transparent to the lender. A firm with ample venture capital can defer both. The substitution is real and expensive: the firm pays for the missing instrument through a higher risk premium or a larger equity cushion. The integrated mechanism is not the only way to finance a green transition. It is the cheapest way for firms that lack one of the three.

The policy implication is that the instruments should be deployed together where the firm's risk profile spans all three categories and selectively where it does not. A software firm developing grid-management algorithms needs sci-tech finance and a rating; it does not need weather insurance. A plantation company needs insurance and a rating; it may not need venture capital. Mandating all three universally produces the checkbox compliance the counter-case describes.

9.6 The Boundaries of the Mechanism

The triad cannot make a project with negative social return positive. If the technology does not work or the market does not exist, no combination of rating, insurance, and venture capital produces a viable investment. The mechanism improves the allocation of capital among projects. It does not create projects.

Nor can it eliminate basis risk. Index insurance will always leave a gap between trigger and loss. Finer triggers and better data narrow the gap; nothing closes it. The lender must price the residual, and the loan file must disclose it.

The mechanism also cannot solve the principal–agent problem inside a public fund. The four conditions describe what a well-designed fund would do. They do not describe how to make a poorly designed fund change. That is a political problem, and Chapter 12 takes it up.

Finally, the mechanism cannot operate without data. The rating stage requires audited environmental disclosure. The insurance stage requires historical weather and loss data. The venture stage requires technical due diligence. Where the data do not exist, the mechanism does not operate, and building the data infrastructure is the first task. This is unglamorous, and in most of the sectors the transition depends on it is the binding constraint.

9.7 What the Integration Produces

The three instruments produce more together than the sum of their separate effects, because each generates information that improves the others' accuracy. Rating without insurance misprices the tail. Insurance without rating misprices the counterparty. Sci-tech finance without either cannot hand its portfolio companies to the credit market. The cases show the mechanism working and failing: the forest plantation because the loan covenant linked rating and insurance in a verifiable way, the battery-materials startup because the guided fund's disclosure condition generated the data that made the rating credible, the tea cooperative because the index was too coarse and the lender never integrated the policy into its loss model, the provincial fund because none of the four conditions held.

The next chapter takes the mechanism to the field, with loan-level data that let us estimate the size of the rating and insurance effects rather than assert them. The chapter after that does the same for technology firms and venture finance. The purpose of the cases is not to celebrate successes but to identify which condition was binding in each instance, and therefore which policy lever would have made the difference.

The instruments have existed for decades. What is new is the recognition that they must be sequenced and linked, and that the linkages are where policy attention belongs. A green finance system that funds rating agencies, insurance pilots, and venture funds in separate silos will produce the counter-case. A system that requires the handoffs will produce the battery-materials case. The difference is institutional. It shows up in the next round of green finance pilot design, where the relevant question for each province is whether the pilot's reporting template requires portfolio companies to disclose audited environmental data before the fund's second close—a specific, checkable design choice that determines whether the later rating is possible at all.