Chapter 8: Designing Incentives and Evaluating Performance in Policy-Guided Green Funds

A provincial green industry fund closed its first vintage in 2019 with 2 billion yuan of government anchor capital and a ten-year term. By the end of 2023 it had deployed 1.1 billion yuan across fourteen portfolio companies. Three had failed. Four were growing but would need follow-on rounds the fund could not supply from its remaining dry powder. The rest were profitable in the ordinary sense: they sold equipment, software, or services, and they paid taxes. What the fund's annual report could not tell its government sponsors was whether any of this constituted a green transition. The portfolio's aggregate emissions reduction appeared as a single figure, 410,000 tons of CO₂-equivalent, derived from company self-reports multiplied by sector-average emission factors. Nobody had audited it. The manager's carried interest was calculated on a standard 20 percent of realized gains above an 8 percent hurdle, with no environmental condition attached. The manager had every reason to exit the strongest financial performer early and every reason to ignore the weakest environmental performer as long as it did not lose money.

That gap is the subject of this chapter. Policy-guided green funds occupy an awkward institutional position: they are asked to earn commercial returns and to produce environmental outcomes that the market does not price. Development finance institutions have wrestled with the problem for decades. What is new is the scale at which green funds are being capitalized and the specificity of the environmental mandate. A fund told to "support green development" can satisfy its mandate under almost any interpretation. A fund told to reduce industrial process emissions by a measurable quantity in a defined region has a much harder task, and the incentive structure around it has to be built differently.

Three movements organize the discussion. The first is the incentive structure itself: how carry, hurdle rates, management fees, and non-financial compensation can be arranged so that the manager's private optimum sits closer to the social optimum. The second is measurement: which metrics can bear the weight of a performance evaluation, and how financial and environmental returns can be combined without collapsing one into the other. The third is the balancing act: how funds in practice trade profitability against impact, and which institutional arrangements make the trade-off explicit rather than accidental.

8.1 Incentive Structures for Green Funds

8.1.1 The agency problem at the heart of the fund

A policy-guided green fund is a principal-agent relationship with three parties rather than two: the government (or government-backed anchor investor), the private co-investors, and the fund manager. The manager's compensation typically has three components—a management fee on committed or invested capital, a carried interest on realized gains above a hurdle, and possibly a fixed salary or platform fee. The standard venture capital contract, as analyzed by Gompers and Lerner (1999) and by Kaplan and Stromberg (2003), solves a two-party problem: the limited partners want the general partner to exert effort and to avoid excessive risk, and they use carry, vesting, and clawbacks to align interests. Add a third party with a non-pecuniary objective and the contract becomes a multi-task agency problem of the kind formalized by Holmstrom and Milgrom (1991).

The multi-task insight applies directly. When an agent is asked to perform two tasks whose outputs are measured with different precision, the agent allocates effort toward the task that is measured better. If financial return is measured continuously by audited accounts and environmental return is measured by an annual self-report, the manager will optimize the former and satisfy the latter at minimum cost. This is not corruption; it is the predictable response to an asymmetric measurement system. Raising the precision of the environmental measure, or lowering the weight placed on the financial one, is the only lever available.

A second problem is subtler. Green technology investments have a risk profile that differs from conventional venture capital. Deep-tech decarbonization ventures—electrolysis, novel battery chemistries, industrial heat—have longer development cycles, higher capital intensity, and more pronounced valley-of-death problems than software. A standard ten-year fund with a five-year investment period is poorly matched to these timelines. If carry is calculated on a ten-year horizon and the environmental payoff arrives in year twelve, the manager faces a terminal mismatch that no amount of good faith can resolve. The incentive structure has to address the timing of surplus realization as well as its division.

8.1.2 Carry design: paying for what you want

Four carry structures have been used in practice, and each rewards something different.

The bluntest is a standard 20 percent carry above an 8 percent hurdle, released only if the fund meets a pre-specified environmental target—a minimum aggregate emissions reduction, say, or a minimum share of portfolio capital in taxonomy-aligned activities. A provincial fund in Anhui wrote exactly this clause into its 2021 limited partnership agreement, with a gate set at 300,000 tons of CO₂-equivalent avoided. The gate is binary: meet the target and receive full carry, miss it and receive nothing. That simplicity is also the weakness. A manager close to the threshold near the end of the fund's life has an incentive to reclassify a portfolio company, count an avoided emission that would have happened anyway, or push a marginal project over the line. Binary gates also create perverse risk-taking: a manager who expects to miss the gate has no reason to preserve financial value in the final years, because the carry is already lost.

A second option splits the carry into separate financial and environmental pools. Fifteen percent of gains above the hurdle is paid on financial performance; a further 5 percent is paid on verified environmental performance measured in tons of CO₂-equivalent avoided or sequestered, at a fixed price per ton. This converts the environmental objective into a shadow price and makes the trade-off explicit. The manager knows that a ton avoided is worth a defined amount of carry and can compare that amount against the financial return of the same capital allocation. The weakness is administrative: the shadow price has to be set by someone, and if it is set too low the environmental pool is decorative, while if it is set too high the fund becomes an emissions-reduction program with a venture capital hobby. The UK's Bridges Fund Management has run a version of this structure since its 2015 fund, with the environmental pool priced at £15 per verified ton—a figure the firm publishes and revises annually.

A third option defers the carry and vests it against environmental outcomes. The calculation runs on financial gains, but vesting is spread over time: one-third of the carry vests only after independent verification of environmental outcomes at years three, five, and seven post-exit. This addresses the timing problem, since the manager cannot take the money and run before the environmental effect is observable, and it aligns with the long-lived nature of green assets. The cost is administrative: verification at multiple points is expensive, and the deferral ties up manager capital, which raises the required gross return. A forestry fund in Yunnan adopted this structure in 2020 and found that the deferral added roughly 180 basis points to its required gross return—a real cost, but one the sponsor accepted because the carbon outcome was the point of the fund.

The fourth option is impact-linked carry with a clawback. Carry is paid on financial gains, but a portion is subject to clawback if verified environmental outcomes fall below a threshold. Clawbacks are familiar from the standard LP agreement, where they correct for early exits that later turn out badly. Extending the clawback to environmental outcomes punishes overstatement and rewards conservatism in reporting. The risk is that clawbacks are rarely enforced in practice because enforcement is costly and litigious; a clawback that is never triggered is not an incentive. The European Investment Fund's 2019 review of its own clawback provisions found that fewer than 3 percent had ever been invoked.

Which structure fits which fund depends on the mandate and the investor base. A fund anchored entirely by a government green bank with a long horizon can accept deferred vesting. A fund with private co-investors who need liquidity at year eight will resist deferral and may prefer the split pool. The general principle: the more precise the environmental metric, the more the carry can be conditioned on it without creating gaming incentives. Precision and conditionality are complements, not substitutes.

8.1.3 The hurdle rate and the cost of concessional capital

The hurdle rate is the minimum return the manager must earn before carry begins. In a conventional fund it reflects the opportunity cost of capital. In a policy-guided green fund, the anchor investor is often willing to accept a below-market return in exchange for environmental outcomes. Should the hurdle be lowered to reflect the concessional capital, and if so, who captures the benefit?

Consider the arithmetic. Suppose the private co-investors require a 10 percent net return, while the government anchor is willing to accept 4 percent. If the fund earns 8 percent gross, a single blended hurdle of 8 percent leaves no carry; the manager has no incentive to continue. A tiered structure—a 4 percent preferred return to the government tranche, then a 10 percent preferred return to the private tranche, then carry—matches the waterfall to the investors' actual required returns. This is standard practice in blended finance and it works, but it has an uncomfortable implication: the government tranche absorbs the first losses and the lowest returns, which means the private investors are effectively subsidized. That is the point of concessional capital, but the subsidy should be priced and disclosed, not buried in the waterfall.

A cleaner alternative is a first-loss guarantee provided by the government anchor, with the private tranche taking a standard senior position. The guarantee has a measurable cost—the expected loss on the guaranteed tranche—and that cost can be compared against the environmental outcome the fund is expected to produce. This is the logic of the "environmental return on concessional capital" metric discussed in Section 8.2. The guarantee is not free money; it is a purchase of environmental outcomes at a known price per unit.

A third approach, used in several Chinese provincial green funds, has the government anchor accept a capped return, with any surplus above the cap recycled into the fund or into a follow-on fund. The cap is set at, say, the one-year loan prime rate plus 200 basis points. This preserves the government's capital in real terms while directing upside to the private co-investors and the manager. It also creates a natural ceiling on carry, since the manager's share is calculated on a smaller surplus. The mechanism is administratively simple and politically defensible, which matters for funds that must report to a provincial finance bureau every year. The Jiangsu Green Development Fund has used a 3.85 percent cap since 2020—the LPR plus 100 basis points at the time of writing—and the provincial finance bureau cites the simplicity of the arrangement as its main advantage over a tiered waterfall.

8.1.4 Management fees and the deployment pressure

Management fees are the least discussed and most consequential part of the contract. A 2 percent fee on committed capital over a ten-year term is 20 percent of the fund's total capital, paid regardless of performance. For a fund with a strong environmental mandate, the fee creates a deployment pressure that conflicts with selectivity. The manager is paid to invest, not to wait for the right green project. In sectors where bankable green projects are scarce—early-stage carbon capture, green hydrogen, sustainable aviation fuel—the fee structure actively encourages the manager to lower standards.

Three modifications are worth considering. A step-down fee: 2 percent during the investment period, 1.5 percent during the harvest period, and a lower fee on committed-but-uninvested capital. This reduces the reward for mere deployment. A fee-at-risk: 25 percent of the management fee is held in escrow and released only on achievement of environmental milestones. The lever is small but it signals to the manager that the environmental mandate is not a side letter. An impact-linked fee adjustment: the fee is reduced by a defined amount for each year the fund's verified environmental outcome falls below its trajectory. None of these is a complete solution, but together they shift the manager's marginal calculus.

Deployment pressure is especially acute in funds with a fixed investment period and a mandate to invest in a defined region. A provincial fund told to invest 70 percent of its capital within its own province faces a thin pipeline. The manager can either hold cash and miss the deployment target, or invest in marginal projects. Most managers choose the latter, and the environmental quality of the portfolio declines. The fix is to allow the fund to co-invest outside the region, or to extend the investment period, or to reduce the fund size. The last option is rarely chosen because it reduces the manager's fee base, which is precisely why the fee structure matters. The 2022 restructuring of a Shanxi coal-transition fund cut its size from 1.5 billion to 900 million yuan after two years of marginal deals; the manager's fee fell by 40 percent, and the fund's average deal quality rose measurably in the following vintage.

8.1.5 Non-financial incentives: reputation, mandate, and career

Not all incentives are monetary. Fund managers in the green space often care about environmental outcomes directly—they chose the sector for a reason—and this intrinsic motivation can be leveraged. The problem is that intrinsic motivation is fragile. It erodes when the manager is forced to choose between the mandate and the carry, and it can be crowded out by excessive monitoring. The design task is to preserve the manager's discretion over how to achieve environmental outcomes while constraining the outcomes themselves.

Reputation is a powerful non-financial incentive. A manager with a track record of verified environmental performance can raise subsequent funds more easily, attract better co-investors, and command higher fees. If the evaluation framework produces a public, comparable rating of fund performance—financial and environmental—the reputational channel does real work. This is the logic behind the impact transparency initiatives discussed in Section 8.2, and it explains why funds resist standardized metrics: standardization makes comparison possible, and comparison is a discipline.

Career incentives matter too. A fund manager who spends five years on a green fund and then moves to a conventional fund needs the green experience to be legible. If the evaluation framework treats green fund performance as a distinct skill set—verified environmental outcomes, taxonomy alignment, stakeholder management—the manager's human capital is portable. If it treats green performance as a soft add-on to financial performance, the manager's green experience is discounted in the labor market, and the fund will struggle to retain talent. The International Finance Corporation's 2021 review of its own blended-finance staff found that fewer than one in five moved into conventional private equity roles, against more than half of those who had worked on standard infrastructure funds—a gap the review attributed to the difficulty of translating impact experience into conventional deal credentials.

8.1.6 A worked example: the tea cooperative fund

Take a concrete case from the agriculture and forestry side, which Chapter 10 develops in more detail. A provincial fund is set up to invest in tea cooperatives in a mountainous region. The cooperatives are small. The Mangjing village cooperative has 608 households and 2,568 people, 97 percent of them Blang ethnic minority, with 80 percent of income from tea. The fund's environmental mandate is to support the conversion of conventional tea plots to organic or eco-certified production, with an expected reduction in chemical fertilizer and pesticide runoff into the catchment.

The fund's investment is 300 million yuan over five years, deployed as equity in cooperative processing enterprises and as working capital loans to cooperatives. The manager's carry is 20 percent above a 6 percent hurdle, with 30 percent of the carry deferred and conditioned on verified conversion of at least 40 percent of the portfolio's tea area to certified organic production by year seven.

What does this structure do? It makes the manager care about conversion rates, not just processing margins. It defers a meaningful share of the reward until the conversion is observable. It sets a threshold that is ambitious but achievable—40 percent is above the regional average but below the best-performing cooperatives. And it prices the environmental outcome implicitly: the deferred carry is roughly 60 million yuan, against an expected conversion of perhaps 8,000 hectares, or about 7,500 yuan per hectare of converted tea. That figure can be compared against the cost of achieving the same conversion through a direct subsidy program, which gives the government sponsor a benchmark for whether the fund is a cost-effective delivery vehicle.

The structure also has a weakness. Conversion to organic certification takes three years, and the yield drop in the first two years can be 20–30 percent. The cooperative needs bridge financing to survive the transition. If the fund's working capital loans are priced at market rates, the cooperatives may not be able to service them during the transition, and the conversion fails. The incentive structure has to be paired with a risk-buffering instrument—weather index insurance, in this case, or a transition-period guarantee—or the environmental target is unreachable. This is the complementarity between the three legs of the mechanism that Chapter 9 develops.

8.1.7 The rent-seeking problem

One design issue deserves separate treatment: rent-seeking. Policy-guided funds allocate capital below its market price, which creates a queue. The queue is managed by the fund manager, who decides which projects receive capital. If the manager's compensation is not tied to outcomes, the manager may allocate on the basis of relationships, political pressure, or personal preference. The green mandate can become a cover: a project is labeled green, the capital flows, and the environmental outcome is never verified.

The standard remedy is competitive allocation with transparent criteria. The fund publishes its investment thesis, its eligibility criteria, and its decision process. Applications are scored against published metrics. Decisions are documented. This is unglamorous but effective. The deeper remedy is to make the manager's compensation depend on verified outcomes, so that the manager bears the cost of a bad allocation. A manager who receives carry on a project that fails environmentally has less incentive to allocate to it in the first place.

There is a tension here. Strict verification and competitive allocation reduce rent-seeking but also reduce the manager's discretion, and discretion is valuable in early-stage investing where the best projects do not fit the published criteria. The resolution is to separate the two functions: competitive allocation for the fund's core capital, discretionary allocation for a smaller "exploration" sleeve with stricter outcome conditions. The exploration sleeve accepts higher failure rates in exchange for option value on novel technologies, and its carry is conditioned on a higher environmental bar to compensate for the weaker selection discipline.

flowchart TD
    A[Government anchor capital] --> B[Fund manager]
    C[Private co-investors] --> B
    B --> D{Allocation decision}
    D -->|Core sleeve: competitive| E[Eligible green projects]
    D -->|Exploration sleeve: discretionary| F[Novel early-stage ventures]
    E --> G[Portfolio]
    F --> G
    G --> H[Financial returns]
    G --> I[Environmental outcomes]
    H --> J[Carry: financial pool]
    I --> K[Carry: environmental pool]
    J --> L[Manager compensation]
    K --> L
    I --> M[Independent verification]
    M --> K

Figure 8.1: Separation of allocation sleeves and carry pools in a policy-guided green fund.

8.2 Performance Metrics and Evaluation Frameworks

8.2.1 Why measurement is the binding constraint

Section 8.1 assumed that environmental outcomes can be verified. That assumption is doing a great deal of work. In practice, the measurement of environmental outcomes at the portfolio level is the weakest link in the entire incentive chain. A fund can only pay for what it can measure, and it can only measure what its portfolio companies report. The reporting is voluntary, unaudited, and inconsistent across sectors.

The ESG rating literature offers a cautionary tale. Billio, Costola, Hristova, Latino, and Pelizzon (2021) examined the agreement among major ESG rating providers and found that "heterogeneity in rating criteria can lead agencies to have opposite opinions on the same evaluated companies and that agreement across those providers is substantially low." Escrig-Olmedo, Fernández-Izquierdo, Ferrero-Ferrero, Rivera-Lirio, and Muñoz-Torres (2019) reached a similar conclusion: rating agencies "do not fully integrate sustainability principles into the corporate sustainability assessment process." If the rating agencies, with their data budgets and analytical teams, cannot agree on whether a company is green, a fund manager with fourteen portfolio companies and a two-person sustainability team has little chance of doing better without a standardized framework.

Measurement is not hopeless, but it is expensive and partial. The fund should choose metrics that are robust to disagreement, narrow in scope, and tied to the fund's specific mandate. A fund that invests in industrial energy efficiency should measure energy saved per unit of output, not a composite ESG score. A fund that invests in forestry should measure hectares under improved management and carbon stock change, not a biodiversity index. Narrow metrics are less comparable across funds but more reliable within a fund, and reliability is what the incentive structure needs.

8.2.2 The metric hierarchy

Metrics can be organized into a hierarchy, from the most robust to the least.

Activity metrics count what the portfolio company does, not what effect it has. Hectares converted to organic production. Megawatts of renewable capacity installed. Tons of waste diverted from landfill. Activity metrics are cheap to collect, hard to fake (the activity is observable), and easy to aggregate. Their weakness is that they say nothing about the environmental outcome: a megawatt of solar in a coal-heavy grid displaces more emissions than a megawatt in a hydro-heavy grid.

Physical outcome metrics estimate the environmental effect using a defined methodology. Tons of CO₂-equivalent avoided, calculated as baseline emissions minus project emissions. Cubic meters of water saved. Kilograms of nitrogen runoff avoided. These require a baseline counterfactual, which is the source of most disagreement. The baseline can be historical (what the company emitted before the project), sectoral (what a typical company in the sector emits), or regulatory (what the law requires). Each choice produces a different number, and the choice can be gamed. A fund that wants a large emissions reduction number will choose a generous baseline.

Monetized outcome metrics convert physical outcomes into money using a shadow price. The social cost of carbon, the avoided cost of water treatment, the value of ecosystem services. Monetization allows comparison across different environmental objectives and against financial returns, which is why it is attractive for evaluation. It also imports the uncertainty of the shadow price into the evaluation. The social cost of carbon has been estimated at values ranging from $50 to over $200 per ton, depending on the discount rate and the damage function (Rennert et al., 2022). A fund's environmental performance can swing by a factor of four depending on which price is used.

Composite scores aggregate multiple metrics into a single number, often weighted by materiality or by the fund's mandate. Composite scores are convenient for reporting and dangerous for incentives, because the weights are arbitrary and the aggregation hides trade-offs. A fund that improves its composite score by increasing renewable investment while increasing water use in a water-stressed region has not necessarily improved its environmental performance. Composite scores should be used for communication, not for carry calculation.

Pay carry on physical outcome metrics. Report activity and physical outcome metrics. Use monetized metrics for comparison against alternative uses of public money. Keep composite scores away from anything that affects compensation.

8.2.3 The evaluation framework: four dimensions

A workable evaluation framework for a policy-guided green fund has four dimensions, each with its own metrics and its own audience.

Financial performance is measured by the standard metrics: net internal rate of return (IRR), total value to paid-in capital (TVPI), distributions to paid-in capital (DPI), and the public market equivalent (PME). The PME is especially important for policy-guided funds because it benchmarks the fund against a public market index, correcting for the fact that a fund with a long horizon and illiquid assets should be compared against a diversified portfolio rather than against cash. A fund that returns 8 percent net over ten years while the relevant equity index returns 10 percent has underperformed on a risk-adjusted basis, even if the absolute return is positive.

Environmental performance uses physical outcome metrics tied to the fund's mandate, verified by an independent auditor, reported at the portfolio and fund level. The verification should be annual and should include a sample audit of portfolio company reports. The cost of verification—perhaps 0.5 percent of fund assets per year—is a legitimate fund expense and should be disclosed to investors.

Additionality is the counterfactual question: would the environmental outcome have happened without the fund? It is the hardest dimension to measure and the most important for a policy-guided fund, because a fund that finances projects that would have happened anyway is a subsidy without a return. Three approaches are standard. A control group compares the portfolio companies against similar companies that did not receive fund capital. A stated counterfactual asks the manager and the company what would have happened without the investment, and documents the answer. A market test asks whether the project could have raised capital from private investors on the same terms; if it could, it is not additional. None is fully satisfactory, but the combination is better than no assessment at all.

Systemic effects capture the fund's effect on the broader market: did it crowd in private capital, build a local green finance ecosystem, or create a demonstration effect that other investors followed? This dimension is the hardest to quantify and the most likely to be invoked as a justification after the fact. A disciplined approach measures it through follow-on capital raised by portfolio companies, the number of first-time green investors attracted to the region, and the diffusion of the fund's measurement practices to other funds. These are proxies, but they are observable.

8.2.4 A decision rule for the blended return

The four dimensions can be combined into a single framework without collapsing them into a single number. Define the fund's total return to society as:

where \(R_{\text{financial}}\) is the net financial return to investors, \(E_{\text{verified}}\) is the verified environmental outcome in physical units, \(\lambda\) is the shadow price per unit, \(C_{\text{concessional}}\) is the cost of the concessional capital (the difference between the market return and the actual return on the government tranche), and \(A_{\text{additional}}\) is a qualitative adjustment for additionality and systemic effects.

The formula is a decision rule, not an accounting identity, and the parameters have to be stated before the fund is capitalized. In the provincial fund described at the opening of this chapter, the government tranche of 1.2 billion yuan accepted a 4 percent return against a market return of 8 percent, for a concessional cost of 48 million yuan per year. The verified environmental outcome at the end of 2023 was 410,000 tons of CO₂-equivalent—assuming, for the sake of the calculation, that the figure survives audit. At a shadow price of 200 yuan per ton, the environmental benefit is 82 million yuan per year, against a concessional cost of 48 million. The fund clears the test, but narrowly, and the margin depends entirely on the shadow price. At 100 yuan per ton the fund fails; at 300 yuan it passes comfortably. The sponsor's willingness to pay per ton is therefore the single most important parameter in the evaluation, and it should be set by the finance bureau in advance, not negotiated with the manager after the fact.

The formula also clarifies a common confusion. A fund can have a positive financial return and a negative social return if the concessional cost exceeds the environmental benefit. A fund can have a negative financial return and a positive social return if the environmental benefit is large enough. The evaluation should report both, and the decision to continue or terminate the fund should be based on the social return, not on the financial return alone. This is obvious in principle and frequently ignored in practice, because the financial return is the number that appears in the budget documents.

8.2.5 The verification problem

Verification is the weak point of any environmental performance metric. The fund manager has an incentive to overstate, the portfolio company has an incentive to cooperate, and the verifier is paid by the fund. The standard remedies are independence, rotation, and sampling.

Independence means the verifier has no commercial relationship with the fund manager or the portfolio companies beyond the verification contract. Rotation means the verifier is replaced every three years, so that a cozy relationship does not develop. Sampling means the verifier audits a random subset of portfolio companies in depth rather than reviewing all reports superficially. The sample should be stratified by sector and by reported outcome, with over-sampling of outliers. A portfolio company that reports an unusually large emissions reduction should be audited with higher probability, because the large numbers are where the errors and the gaming concentrate.

The cost of verification is real but not prohibitive. For a fund with fourteen portfolio companies, an annual verification budget of 1–2 million yuan covers a desk review of all reports plus a deep audit of four to six companies. That is 0.05–0.1 percent of a 2 billion yuan fund, small relative to the management fee. The obstacle is not cost but governance: the verification contract must be held by the fund's investors or by an independent committee, not by the manager. If the manager controls the verifier, the verification is decorative. The Green Climate Fund's independent evaluation unit, which reports to the board rather than to the secretariat, is the clearest working example of this separation at scale.

8.2.6 Reporting to different audiences

A fund reports to at least four audiences, and each needs a different report.

The government sponsor needs the social return calculation, the additionality assessment, and the cost per unit of environmental outcome. The sponsor is deciding whether to renew the fund, expand it, or terminate it, and the decision requires a comparison against alternative uses of public money.

The private co-investors need the financial return, the risk metrics, and the environmental performance only insofar as it affects financial risk. A private investor in a green fund cares about the environmental mandate because it affects deal flow and exit options, not because the investor is paid on environmental outcomes.

The portfolio companies need clear, stable reporting requirements. A company that is asked to report twenty metrics in year one and a different twenty in year two will report badly. The fund should publish a reporting template at the start of the investment period and change it only with notice.

The public needs a short, comparable summary: the fund's financial return, its verified environmental outcome, and its cost per unit of outcome. The summary should be published annually and should be auditable. This is the reputational channel that Section 8.1.5 discussed, and it only works if the summary is standardized enough to allow comparison across funds.

8.2.7 Case: evaluating a forestry fund

A forestry fund with a mandate to support sustainable forest management in a mountainous province illustrates how the framework works in practice. The fund invests in plantation companies and in community forest cooperatives. The environmental mandate is to increase forest carbon stock and to reduce soil erosion.

The activity metrics are hectares under improved management and hectares certified to a sustainable forestry standard. The physical outcome metrics are carbon stock change (measured by a combination of remote sensing and field plots) and sediment load in downstream water bodies (measured by sampling stations). The monetized metric is the value of carbon sequestration at the social cost of carbon, plus the avoided cost of reservoir dredging.

The evaluation reveals a trade-off that the fund's marketing material had concealed. The plantation companies increase carbon stock rapidly, because they plant fast-growing species and harvest on short rotations. The community cooperatives increase carbon stock slowly, because they manage natural forest with selective harvesting. But the cooperatives reduce soil erosion much more effectively, because they maintain understory vegetation and avoid clear-cutting. On a carbon-only metric, the plantations win. On a combined carbon-and-erosion metric, the cooperatives win. The fund's mandate said "increase forest carbon stock and reduce soil erosion" without specifying the weights. The evaluation forced the weights into the open, and the fund's investment committee had to decide which outcome it valued more. That decision is a political and ethical choice, not a technical one, and the evaluation framework should not pretend otherwise.

8.2.8 The limits of metrics

The pressure to measure can distort the activity being measured. A fund that is paid on tons of CO₂ avoided will favor projects with large, easily measured emissions reductions—utility-scale solar, industrial efficiency—and neglect projects with smaller or harder-to-measure effects—soil carbon, biodiversity, community adaptation. This is Goodhart's law in its environmental form: when a measure becomes a target, it ceases to be a good measure.

The remedy is a portfolio of metrics, with some metrics explicitly protected from the incentive structure. A fund can designate a "learning portfolio" of projects that are evaluated on narrative and site visits rather than on standardized metrics, with the understanding that these projects will not contribute to the carry calculation. The learning portfolio generates the knowledge that the standardized metrics cannot capture, and it protects the fund from the tyranny of the measurable.

8.3 Balancing Financial and Social Returns

8.3.1 The trade-off is real

Sections 8.1 and 8.2 treated financial and environmental returns as separable, each measured on its own axis. In practice they interact. A fund that pushes for higher environmental performance will, at some margin, accept lower financial returns. The question is how steep the trade-off is, and whether the fund's institutional design makes the trade-off at the right point.

The trade-off is steepest in early-stage deep-tech. A green hydrogen electrolyzer company needs ten years and 2 billion yuan to reach commercial scale. The risk-adjusted financial return is uncertain and back-loaded. The environmental return, if the technology works, is large. A fund with a ten-year horizon and a 10 percent hurdle will underinvest in this segment unless the environmental return is monetized or the fund's horizon is extended.

The trade-off is shallower in mature segments. A fund that invests in energy-efficient building retrofits can earn a market return and a measurable environmental return at the same time, because the energy savings pay for the retrofit. The constraint is not the trade-off but the pipeline: there are only so many buildings, and the retrofit market is fragmented.

The trade-off is different again in agriculture and forestry. A tea cooperative that converts to organic production suffers a yield drop for two to three years and then earns a price premium. The financial return is negative in the short run and positive in the long run. The environmental return is positive throughout, because the reduction in chemical runoff begins immediately. A fund that evaluates on a ten-year IRR will see the conversion as marginal; a fund that evaluates on a twenty-year social return will see it as clearly positive. The horizon is doing as much work as the discount rate.

8.3.2 Three strategies for balancing

Funds in practice use three strategies to manage the trade-off.

Segmentation divides the fund's capital into pools with different return targets. A "commercial" segment, perhaps 60 percent of capital, targets market returns and accepts only projects with a clear financial path. An "impact" segment, perhaps 30 percent, accepts below-market returns in exchange for verified environmental outcomes. A "catalytic" segment, perhaps 10 percent, takes early-stage risk that no commercial investor would take, with the explicit expectation that most projects will fail. Each segment has its own hurdle, its own carry, and its own metrics. Segmentation makes the trade-off explicit: the government sponsor can see how much of its capital is being used for market-return investments and how much is being used as a subsidy. The China Development Bank's green bond program uses a version of this structure, with a 70/25/5 split across commercial, impact, and catalytic pools.

Blending combines commercial and concessional capital in a single project, with the concessional capital taking a junior position. A blended project can achieve a commercial return for the private tranche and a below-market return for the government tranche, while the environmental outcome is the same for both. Blending is more capital-efficient than segmentation because it uses the concessional capital as a lever rather than as a separate pool. It is also more complex to administer, because each project requires its own waterfall.

Sequencing invests in the early, risky stage and exits to private investors once the technology or the business model is proven. This is the classic venture capital model applied to green technology, and it works when the fund has the patience to hold through the valley of death. The fund's return comes from the exit, and the environmental outcome comes from the technology's diffusion after the exit. Sequencing requires a robust exit market, which in green technology often means a strategic buyer—an incumbent utility or industrial company—rather than a public listing.

8.3.3 The role of concessional capital pricing

The price of concessional capital is the key parameter in the balancing act. If the government tranche accepts a return too close to the market return, the fund cannot invest in the early-stage or low-return projects that the mandate requires. If the return is too low, the fund becomes a grant program with extra steps, and the private co-investors have no reason to participate.

The right price can be derived from the environmental outcome the government is buying. Suppose the government wants to reduce emissions by 1 million tons over ten years and is willing to pay 200 yuan per ton, for a total of 200 million yuan. The concessional capital is the vehicle for that payment. If the fund can deliver the reduction with 150 million yuan of concessional cost, the government has bought the outcome below its willingness to pay. If the cost is 300 million yuan, the fund is overpriced and the government should look for another delivery vehicle.

This calculation requires the government to state its willingness to pay for environmental outcomes, which is politically difficult but analytically necessary. Without it, the concessional capital is priced by negotiation, and the negotiation is dominated by the fund manager's estimate of what the government will accept. The result is a price that reflects bargaining power rather than the value of the outcome.

8.3.4 The exit problem

A green fund's financial return is realized at exit. The exit market for green assets is thinner than for conventional assets, for three reasons. The buyer pool is limited: strategic buyers in green sectors are fewer than in software or consumer goods. The assets are often capital-intensive and regulated, which reduces the universe of financial buyers. And the environmental attribute may not survive the exit: a buyer may reduce the environmental performance of an acquired company, which undermines the fund's impact thesis even as it realizes thefinancial return.

The exit problem has a design implication. Carry should be conditioned on the environmental outcome persisting after exit, not merely on the outcome measured at the moment of sale. A deferred carry tranche, released three years after exit and conditioned on the portfolio company maintaining its environmental performance, aligns the manager's interest with the persistence of the outcome. This is the logic of the deferred-vesting structure in Section 8.1.2, and it matters most for funds that exit to buyers with weak environmental commitments. The 2021 sale of a Zhejiang industrial-efficiency portfolio to a private equity buyer illustrates the risk: the buyer cut the energy-audit function within eighteen months of closing, and the verified savings that had justified the fund's environmental carry disappeared before the deferred tranche vested.

There is a second implication: the fund should cultivate the exit market. A green fund that builds relationships with strategic buyers, provides them with diligence information, and demonstrates the value of green assets is investing in its own exit liquidity. This is a form of market development that rarely appears in the fund's performance metrics, but it is a real source of additionality.

8.3.5 Governance: who decides the trade-off

The trade-off between financial and social returns is ultimately a governance question. Who decides how much financial return the fund should sacrifice for a given environmental outcome? The fund manager has an incentive to sacrifice too little, because the manager's carry is financial. The government sponsor has an incentive to sacrifice too much, because the sponsor does not bear the financial cost directly. The private co-investors have an incentive to sacrifice nothing.

The governance structure must therefore include a body that represents both the financial and the environmental interest and has the authority to set the trade-off. An investment committee with a mandate that explicitly states the trade-off—for example, "the fund shall target a net financial return of at least 6 percent and a verified emissions reduction of at least 500,000 tons, and shall not sacrifice more than 2 percentage points of financial return for an additional 100,000 tons"—gives the manager a boundary within which to operate. A committee told only to "maximize risk-adjusted returns while supporting green development" has been given no boundary at all, and will resolve the ambiguity in favor of whichever objective its members are held accountable for.

In practice, the composition of the committee matters as much as its written mandate. A committee dominated by finance ministry representatives will price environmental outcomes at zero. A committee dominated by environmental regulators will ignore the fund's solvency. The workable arrangement is a mixed committee with an odd number of members, a published voting record, and a requirement that any decision to deviate from the stated trade-off be justified in writing. The written justification is not a bureaucratic formality; it is the mechanism by which the trade-off becomes visible to the sponsor, the co-investors, and eventually the public.

8.3.6 The distributional dimension

There is a dimension of the balancing act that the financial-versus-environmental framing tends to hide: distribution. A green fund's investments create winners and losers. A fund that supports the conversion of tea cooperatives to organic production raises the incomes of cooperatives that can afford the three-year transition and disadvantages those that cannot. A fund that invests in utility-scale solar creates construction jobs in the province but may displace agricultural land. A fund that finances energy-efficient retrofits lowers the operating costs of building owners but may reduce the demand for the labor that maintained the old systems.

These distributional effects are not incidental. They determine whether the green transition is politically sustainable. A fund that produces a large aggregate environmental outcome while concentrating the costs on a vocal and organized group will face resistance that ultimately undermines the outcome. The evaluation framework should therefore include a distributional dimension: who bears the cost of the transition, who receives the benefit, and whether the compensation mechanisms are adequate.

The Mangjing tea case illustrates the point. The cooperative model, in which the village party branch holds the major decision rights, can raise the efficiency of the conversion in the short run. But as the earlier field observation put it, "this centralized approach will undoubtedly greatly improve the efficiency of cooperative operation and decision-making in a short period, but in the long run it will also wear down the enthusiasm of the masses." A fund that measures only conversion hectares and processing margins will not see the erosion of participation until it is too late. A distributional metric—the share of cooperative members who report that their voice in decisions has increased, the share of value added retained by smallholders—would make the erosion visible while there is still time to correct it.

8.3.7 When to walk away

A final question that funds rarely ask in writing: under what conditions should the fund decline an investment that meets its financial and environmental criteria? The answer is not obvious. A fund with a mandate to deploy capital will deploy it. But there are cases where the environmental outcome is real and the financial return is adequate, and the investment should still be declined—because it crowds out a private investor who would have made the same investment on the same terms, because it creates a dependency that the fund cannot sustain, or because the environmental outcome is achieved at an unacceptable social cost.

The discipline of walking away is easier to maintain if the fund's mandate includes an explicit additionality test at the deal level. Before committing capital, the investment team must document why the project cannot raise the same capital from private sources on comparable terms. If the answer is "it can," the fund declines. This test is uncomfortable because it reduces the deal flow, and a reduced deal flow is visible to the sponsor as underdeployment. The alternative—deploying capital into projects that do not need it—is worse, because it converts the fund into a subsidy for projects that would have happened anyway, and the environmental outcome is attributed to the fund when it should not be.

8.3.8 The evaluation cycle in action

In the third year of a Hunan forestry fund, the annual verification report landed on the investment committee's desk with a number that did not match the fund's own self-report. The manager had claimed 62,000 tons of CO₂-equivalent sequestered across the portfolio; the independent verifier, sampling six of the eleven holdings, could confirm 41,000 tons and could not confirm the rest because two plantation companies had used a biomass expansion factor that the fund's own methodology manual had retired in year one. The discrepancy was 34 percent. The committee had ninety days to decide what to do about it before the year's carry calculation was finalized, and what it did over those ninety days is the clearest illustration I can give of how an evaluation cycle actually functions when it is not a formality.

The cycle begins, in any well-run fund, with targets fixed before the year opens. The Hunan fund had set four: a net financial return of at least 5 percent on invested capital, a verified sequestration outcome of 55,000 tons, an additionality threshold requiring that at least 70 percent of new commitments pass the deal-level counterfactual test, and a distributional target on the share of cooperative members reporting increased decision-making voice. The targets were approved by the committee in January and disclosed to the provincial finance bureau. None of them was invented at year-end to match results, which is the first thing that separates a real cycle from a reporting ritual.

Monitoring ran quarterly and distinguished leading from lagging indicators. Hectares under improved management, a leading activity metric, tracked close to plan all year. Carbon stock change, the lagging physical outcome, could only be estimated until the verification window opened. The quarterly reports flagged the two plantation companies as outliers as early as the second quarter, when their self-reported sequestration rates ran 40 percent above the sectoral benchmark. The flag was noted and not acted on. That is the ordinary failure mode of monitoring: the signal was there, and nobody with authority read it until the verifier forced the issue.

Verification is where the cycle acquires teeth, and only if the verifier is outside the manager's control. In Hunan the contract sat with the finance bureau, not the fund manager, and the sampling protocol over-weighted outliers by design. The verifier's report was published alongside the manager's, with the discrepancy disclosed rather than smoothed. This is the single most important procedural choice in the whole cycle. A fund that publishes only its own number has no cycle; it has a newsletter.

Recalibration is the stage the Hunan committee reached in the ninety-day window, and it is where the incentive structure and the measurement framework meet. The committee had three options. It could accept the manager's 62,000-ton figure and adjust the methodology retroactively, which would have preserved the carry and destroyed the credibility of every future verification. It could hold the manager to the verified 41,000 tons, fail the environmental target, and withhold the environmental tranche of carry—a defensible outcome that would have punished the two plantation companies for a methodology error the fund's own manual had created. Or it could split the difference institutionally: hold the carry calculation to the verified number, but treat the methodology failure as a fund-level error rather than a manager-level one, restate the year-one baseline for all eleven holdings, and reset the year-four target to reflect the corrected baseline. The committee chose the third path, by a vote of four to three, with the written dissent from the finance ministry representative appended to the minutes. The dissent argued, correctly in my view, that the manager had signed off on the methodology manual and should bear the cost of its retirement.

What the episode shows is that the cycle is not a loop that runs itself. Every stage requires a decision about who bears the cost of an error, and those decisions are where the incentive structure actually lives. The published voting record and the appended dissent did more to align the manager's behavior in year four than any clause in the limited partnership agreement, because the manager could see that the committee would read the verification report and that a methodology shortcut would surface in public.

flowchart LR
    A[Target-setting] --> B[Monitoring]
    B --> C[Verification]
    C --> D[Recalibration]
    D --> A
    D --> E{Adjust strategy, incentives, or targets}
    E --> A

Figure 8.2: The annual evaluation cycle for a policy-guided green fund.

The cycle is not a substitute for judgment. It is a structure that forces judgment to be exercised at the right time, with the right information, by the right people. A fund that runs the cycle mechanically will produce reports that satisfy the sponsor and change nothing. A fund that uses the cycle to confront the trade-offs described in this chapter will, over time, develop a track record that is legible to the market, to the sponsor, and to the public.

8.3.9 What the provincial fund should have done

Return to the fund described at the opening of this chapter. It had 2 billion yuan of anchor capital, fourteen portfolio companies, a 20 percent carry above an 8 percent hurdle, and an unaudited aggregate emissions figure. A better contract would have split the carry: 15 percent on financial gains above a 6 percent hurdle, with one-third deferred and conditioned on verified environmental outcomes at years three, five, and seven post-exit. The lower hurdle would have reflected the concessional capital; the deferral would have matched the long-lived nature of the environmental outcome; the verification would have guarded against overstatement.

The environmental target would have been narrow and sector-specific rather than an aggregate emissions figure. The fund would have committed to a set of physical outcome metrics—tons of CO₂-equivalent avoided in the industrial portfolio, cubic meters of water saved in the agricultural portfolio, hectares under improved management in the forestry portfolio—each with a baseline methodology published in advance and an independent verifier appointed by the sponsor rather than the manager.

The capital would have been segmented: 60 percent in a commercial sleeve targeting market returns, 30 percent in an impact sleeve accepting below-market returns for verified outcomes, 10 percent in a catalytic sleeve for early-stage technologies, each with its own hurdle, metrics, and reporting. The sponsor would have stated its willingness to pay per ton of CO₂-equivalent avoided, and the fund's implicit cost per ton would have been calculated annually and compared against that willingness to pay. Two consecutive years above the willingness-to-pay threshold would have triggered restructuring. And every investment memo would have carried an additionality test documenting why the project could not raise comparable capital privately; deals that failed the test would have been declined, even at the cost of underdeployment.

None of these changes is radical. Each is a known technique, used in some funds and neglected in others. The obstacle to adopting them is not analytical difficulty but institutional inertia: the sponsor is measured on deployment, the manager is measured on financial return, and the environmental outcome is measured by nobody. The evaluation cycle is the device that makes the environmental outcome visible, and visibility is the precondition for accountability.

8.3.10 The limits of fund-level design

A candid assessment must acknowledge what fund-level incentive design cannot do. It cannot fix a pipeline problem: if there are no bankable green projects in the region, no carry structure will create them. It cannot substitute for a carbon price: if the externality is unpriced, the fund is a second-best instrument, and its cost per ton will be higher than a well-designed tax. It cannot overcome a weak verification infrastructure: if there are no credible auditors, the environmental metrics will be soft regardless of how the carry is structured.

These limits point to the complementarity that Chapter 9 develops. The fund is one leg of the triad. The rating system identifies which firms are genuinely transitioning; the insurance system buffers the risks that make green projects unbankable; the fund catalyzes the capital that the first two legs make investable. A fund that is well designed but isolated from the rating and insurance systems will underperform, because it will be asked to bear risks that those systems should absorb and to verify outcomes that those systems should certify.

The reverse is also true. A rating system without a fund has no capital to allocate; an insurance system without a fund has no projects to protect. The three legs are complements, and the design of each must take account of the others. This is why the evaluation framework in this chapter includes a systemic-effects dimension: the fund's performance depends partly on whether it strengthens the rating and insurance infrastructure that surrounds it.

8.3.11 A note on international comparison

The design choices described here are not unique to any one jurisdiction. The European Union's sustainable finance framework, with its taxonomy regulation and disclosure requirements, provides a reference point for what a standardized environmental metric looks like. The EU taxonomy's technical screening criteria are, in effect, a physical-outcome metric hierarchy written into law, and a fund that aligns its reporting with the taxonomy gains comparability at the cost of some local specificity. The experience of development finance institutions in blending concessional and commercial capital offers a second reference point: the multilateral development banks have decades of experience with first-loss guarantees, tiered waterfalls, and additionality testing, and their published methodologies are a useful template. The Asian Development Bank's 2019 additionality framework, in particular, is the most detailed public statement of how a deal-level counterfactual test can be operationalized.

The temptation is to import a framework wholesale. The better approach is to adopt the structure—tiered metrics, independent verification, explicit trade-offs—while adapting the specifics to the local pipeline and the local verification capacity. A province with a strong remote-sensing capability can use satellite-based forest carbon metrics; a province without one should use field plots and accept a higher verification cost. The structure is portable; the metrics are not.

8.3.12 Conclusion: the design as a whole

The chapter has moved from the contract to the metric to the trade-off, and the three are inseparable. A carry structure that pays for environmental outcomes requires metrics that can bear the weight of payment. Metrics that can bear that weight require verification that is independent of the manager. Verification that is independent requires a governance structure that separates the verifier from the verified. And the trade-off between financial and social returns requires a governance body with the authority to set it and the information to monitor it.

The provincial fund at the opening of this chapter had none of these. It had a mandate, a manager, and a number. The number was unaudited, the mandate was unenforceable, and the manager's compensation was indifferent to both. The fund may still have produced environmental benefits—some of its portfolio companies were surely genuine—but it could not demonstrate them, and it could not learn from them. The design failure was not a failure of good intentions. It was a failure to build the institutional machinery that converts intentions into verified outcomes.

That machinery is what this chapter has tried to specify. It is not glamorous. It consists of waterfalls, hurdle rates, sampling protocols, and committee voting records. But it is the difference between a fund that claims to support the green transition and a fund that can show, year by year, what it has actually changed. In the Hunan fund, the machinery produced a specific decision in year three: the committee voted four to three to restate the year-one baseline, reset the year-four target to 48,000 tons, and hold the carry calculation to the verified 41,000. The manager's environmental tranche was cut by 11 million yuan. The following year's verification report came in within 4 percent of the self-report, the tightest margin the fund had ever recorded. Chapter 9 takes that same machinery and asks how it interacts with the other two legs of the triad—the rating system that identifies green firms and the insurance system that buffers their risks—beginning with the credit-rating methodology that a technology firm must satisfy before a green fund will consider it.