Chapter 8: Forest Insurance and the Economics of Eco-System Risk
A larch plantation in Heilongjiang planted this spring will not yield merchantable timber for thirty years. One dry April afternoon can turn it into ash and a carbon-liability entry in a ledger. That asymmetry—slow accumulation, fast destruction—is the actuarial signature of forestry, and it explains why forest insurance has stayed a small market nearly everywhere relative to the exposure it nominally covers. China's forest coverage reached 24.02 percent by the ninth national forest inventory; the State Forestry and Grassland Administration puts national forest stock at 19.07 billion cubic meters. The insured fraction of that stock is thin. The gap between the asset base and the insured base is where this chapter lives.
The previous chapter ended with the Henan farmer's rainfall contract and the observation that forest insurance faces a harder version of every problem in agricultural index design. I take that comparison seriously. The forest owner's contract must cover a rotation measured in decades, a fire risk that clusters in space and time, and increasingly a carbon asset whose price depends on policy instruments that do not yet trade at scale. Each feature strains the standard insurance architecture—actuarial pricing, loss adjustment, reinsurance—in a different place. Forest insurance can be made to work, but only if it is understood as a hybrid instrument: part indemnity contract, part public good, part option on ecosystem services. Treating it as a pure private product keeps the market small. Treating it as pure subsidy produces uptake without risk discipline.
8.1 Ecosystem Risks and Forestry's Vulnerability
Forestry's risk profile differs from agriculture's in the dimension that matters most to an insurer: time. A crop cycle is one season; a forest rotation runs twenty-five to sixty years for fast-growing species and a century or more for hardwoods. Over that horizon the loss distribution is not stationary. Fire frequency shifts with climate, pest outbreaks arrive in decade-long cycles, and the value of the final harvest depends on timber prices and carbon policy that no actuary can forecast with useful precision. An insurer writing a one-year policy against a thirty-year exposure faces a mismatch no premium can fully resolve.
Fire is the dominant insured peril. Globally, landscape fires burned an average of 3.9 million square kilometers per year between 2001 and 2019, according to NASA's MODIS-based estimates, with wide interannual swings driven by drought and El Niño cycles. In China, the State Forestry and Grassland Administration recorded an average of roughly 3,000 forest fires per year in the 2010s, down sharply from the 1990s, but damage per fire has risen as plantations have matured and accumulated fuel. The 1987 Greater Khingan fire burned approximately 1.3 million hectares in Heilongjiang and killed 211 people. It remains the reference event for Chinese forest risk, and it illustrates a property that matters for insurance design: fire risk is spatially correlated. A single weather system can ignite and spread across hundreds of kilometers, so an insurer's portfolio of forest policies in one province behaves less like a diversified pool and more like a single large bet on the fire season.
Pests and disease form the second risk family. The pine wood nematode, Bursaphelenchus xylophilus, has spread through pine forests in eastern China since the 1980s and now affects roughly 1.8 million hectares cumulatively, according to State Forestry and Grassland Administration surveys; control requires clear-cutting infected stands and quarantining timber transport. In North America, the mountain pine beetle outbreak in British Columbia killed pine on 18 million hectares between the late 1990s and the 2010s, an area larger than Jiangsu Province. Insect damage is slower than fire and harder to attribute to a single event, which creates a boundary problem for indemnity contracts: is a stand that dies over three years from beetle infestation a single loss, three annual losses, or a pre-existing condition excluded at underwriting? The contract language determines the answer. Most Chinese forest insurance policies have handled pests through exclusions or sublimits rather than full coverage, precisely because loss adjustment is difficult.
Climate change lengthens the tail of both. Warmer winters reduce beetle mortality; earlier snowmelt extends the fire season; drought stress lowers tree resistance to pathogens. The Intergovernmental Panel on Climate Change's 2022 assessment found that fire weather seasons have lengthened by roughly 20 percent globally since the 1980s. For an insurer, historical loss experience is a biased estimator of future losses. A pricing model calibrated on 1990–2010 fire data will underprice a 2030–2050 portfolio. This is the actuarial version of the problem I raised in Chapter 3 regarding technology-firm ratings: the data generating process is non-stationary, and the rating or premium must carry an explicit adjustment for that non-stationarity rather than pretending it away.
A fourth risk has entered forestry balance sheets only recently: carbon liability. A forest enrolled in a carbon offset program holds a stock of credited sequestration. If the forest burns, the carbon is released and the credit must be reversed—the owner either surrenders replacement credits or pays a penalty. The 2021 fires in California and the Pacific Northwest forced several offset registries to cancel credits from burned projects; the exact figure is disputed, but the mechanism is now standard in voluntary and compliance markets. In China, the national carbon market's offset mechanism (CCER) reopened in 2023 with forestry methodologies, and the question of who bears reversal risk—project owner, registry, or insurer—is unresolved. A forest insurance policy that covers standing timber but not the carbon reversal leaves a new and growing gap on the balance sheet.
8.1.1 The vulnerability arithmetic
I want to make the vulnerability concrete with a simple representation. Let a forest stand's value at time \(t\) be \(V_t\), composed of timber value \(T_t\) and carbon value \(C_t\):
where \(L_t\) is expected liability from reversals and remediation. The insurer's exposure is not \(V_t\) but the loss distribution \(f(L)\) over the policy term. For a one-year fire policy, the relevant quantity is the probability that a fire of sufficient severity occurs within the polygon, multiplied by the conditional damage ratio. If fire events arrive as a Poisson process with annual intensity \(\lambda\) and the damage ratio given an event is \(\delta \in [0,1]\) with distribution \(g(\delta)\), expected annual loss is:
The variance is what threatens the insurer's capital position, and for spatially correlated events the variance of portfolio loss is not the sum of individual variances. If \(N\) policies are exposed to a common fire season with correlation \(\rho\), portfolio variance scales approximately as:
With \(\rho\) of 0.1—plausible for policies within a single province—a portfolio of 1,000 policies carries roughly 100 times the variance of a single policy, not 1,000 times. The diversification benefit is small and disappears entirely in a bad fire year. This is why forest insurance is a reinsurance-dependent business: primary insurers cannot hold the tail, and the price of the product is set substantially by the cost of catastrophe reinsurance, not by the expected loss alone.
The structural consequence: forest insurance is priced off the reinsurance market's view of correlated catastrophe risk, which means that even a well-run primary insurer with excellent local data will find its premium driven by global capital conditions. The 2023 hardening of the catastrophe reinsurance market, following losses from Hurricane Ian and European hailstorms, raised attachment points and prices for wildfire-exposed portfolios worldwide. Chinese forest insurers felt this indirectly through their reinsurance treaties with China Re and international retrocessionaires.
8.2 Forest Insurance Products and Pricing
The product landscape divides into three families, and the division is not cosmetic. Each family resolves the time-and-correlation problem differently, and each carries a distinct set of pricing and moral-hazard issues.
8.2.1 Indemnity products
The dominant product in China and most of Europe is the traditional indemnity policy: the insurer pays the assessed loss up to the sum insured, subject to deductibles and exclusions. China's central forest insurance program, launched on a pilot basis in 2009 and expanded nationally, covers fire, pests, disease, and certain natural disasters (wind, flood, landslide) for ecological forests and commercial forests, with the government subsidizing a large share of the premium. The standard policy insures standing timber value, with the sum insured set per mu according to species, age class, and site quality; premium rates in the central program have typically run between 0.2 percent and 0.5 percent of sum insured for fire and comprehensive perils, depending on province and species.
The indemnity structure solves the basis-risk problem that plagues index products—the payment is tied to actual damage, so the forest owner does not face the disappointment of an uninsured loss in a year when the index fails to trigger. It creates three other problems. Loss adjustment is expensive and slow. Assessing damage across a large forest polygon requires field survey, sometimes aerial or satellite imagery, and agreement between adjuster and owner on pre-fire timber volume; in remote stands, adjustment cost can approach claim size. Moral hazard is real: an owner with a thin margin may have less incentive to maintain firebreaks or conduct fuel reduction if the insurer bears the loss. And most important for pricing, the indemnity contract transfers a correlated risk to the insurer without any mechanism to reduce the correlation. The insurer's portfolio of indemnity policies in a fire-prone province is a concentrated bet, and the premium must reflect the cost of holding that concentration.
Pricing an indemnity policy requires an estimate of expected loss ratio by species, age class, and location. The standard approach is a frequency-severity model: estimate the annual probability of a fire event reaching the stand, estimate the conditional damage ratio, multiply by the sum insured, then load for expenses, capital, and reinsurance. The frequency component is usually estimated from historical fire records at the county or prefecture level, adjusted for fuel accumulation and climate trends. The severity component comes from claims data where available, or from forestry growth-and-yield models that relate tree age and volume to burn damage. In China, the claims history is short—the national program is only about fifteen years old—and the data sit with the provincial branches of the major insurers (PICC Property and Casualty, China Life Property and Casualty, and a few others). The resulting rates are therefore partly judgmental. A 2019 study by Zhang and colleagues at the Chinese Academy of Forestry found that provincial forest insurance rates in China were set mainly through administrative negotiation between finance bureaus and insurers, with limited use of loss-cost analysis.
8.2.2 Index products
Index insurance replaces loss adjustment with a trigger tied to an observable variable: rainfall, temperature, wind speed, or a satellite-derived vegetation or burn index. The appeal is administrative: no field adjuster, fast payout, low transaction cost. The cost is basis risk—the gap between the index trigger and the actual loss. The previous chapter discussed rainfall index insurance for crops and the basis-risk problem in detail; forest index products face a harder version because the relevant peril (fire) is not well predicted by a single meteorological variable at a single station.
Several index designs have been tried. The most common is the area-yield or area-loss index, in which the trigger is the average loss ratio across a defined region, estimated from satellite or survey data. If the regional average loss exceeds a threshold, all policyholders in the region receive a payout proportional to the excess. This reduces the insurer's loss-adjustment cost and pools basis risk across the region, but an individual owner whose stand burned while the regional average was low receives little or nothing. A second design uses a burn-area index derived from satellite thermal anomalies: the MODIS or VIIRS active-fire product can detect fires above a minimum size, and a contract can be written on cumulative burn area within a defined grid cell. The advantage is objectivity; the disadvantage is that satellite detection misses small or understory fires and is confounded by cloud cover and agricultural burning. A third design uses a weather index—cumulative days of high fire danger, measured by the Canadian Forest Fire Weather Index or a local equivalent—which is correlated with fire occurrence but not with fire damage.
The actuarial literature on index insurance has established a result that matters here. The choice of marginal distribution for the index variable affects the estimated tail and therefore the premium, and the choice of copula linking the index to the loss affects the measured basis risk. In a forest context, the index-loss relationship is typically estimated with sparse data: a few decades of fire records, a few hundred satellite-detected events. The resulting parameter uncertainty is large, and the premium is sensitive to it. A well-designed index product therefore requires a substantial data investment—satellite archives, ground plots, and fire records linked at fine spatial resolution—before it can be priced with confidence. Few jurisdictions have made that investment.
What index insurance can do for forestry is cover the correlated, catastrophic component of fire risk—the event that affects many stands at once—because that is precisely the component the index measures. It cannot cover the idiosyncratic component—the single stand that burns while its neighbors survive—because the index is blind to it. A rational forest owner with a diversified portfolio of stands might accept that trade-off. A smallholder with one plantation cannot. This is the same distributional problem I noted in Chapter 7 for agricultural index insurance, and it explains why index products in forestry have found more uptake among large commercial operators and state forest farms than among smallholders.
8.2.3 Parametric and hybrid structures
A third family blends indemnity and index features. Parametric cat-in-a-box structures define a triggering event by physical parameters—fire intensity, area burned, proximity to insured stands—and pay a fixed amount when the parameters are met, without loss adjustment. These are common in sovereign and sub-sovereign catastrophe risk transfer (the Caribbean Catastrophe Risk Insurance Facility is the canonical example) and have been adapted to wildfire in California and Australia. In China, the China Re Agriculture Insurance Catastrophe Bond, issued in 2021, transferred earthquake and typhoon risk rather than fire risk, but the structure is a template: a parametric trigger, a special-purpose vehicle, and a reinsurance or capital-markets investor bearing the tail.
For forestry specifically, the most promising hybrid is a two-layer structure: a parametric layer covering the catastrophic, correlated event (pays when burn area in the grid exceeds a threshold), stacked above an indemnity layer covering smaller, idiosyncratic losses. The parametric layer is priced off the catastrophe model and reinsured; the indemnity layer is priced off local claims experience and retained by the primary insurer. This structure matches the risk's correlation profile to the instrument's design, and it is the direction I would recommend for any jurisdiction building a forest insurance market from a low base. The two layers are priced in different markets and can be subsidized separately.
8.2.4 Pricing challenges in detail
Three pricing problems deserve separate treatment because they recur in every jurisdiction.
A credible fire loss model requires a long record of ignitions, spread, and damage, stratified by fuel type, weather, and suppression effort. China's national fire statistics are aggregated at the provincial level and published annually, but the micro-data needed for stand-level pricing—ignition location, response time, area burned, species and age of burned stands—are held by provincial forestry bureaus and insurers and are not systematically pooled. The same is true in much of Europe and in many U.S. states. Pricing therefore relies on regional averages, which underprice high-risk stands and overprice low-risk ones. The remedy is a data-sharing arrangement among insurers and the forestry administration, with appropriate privacy and commercial protections. Until that exists, forest insurance pricing will remain crude.
Historical fire frequency is not a good guide to future frequency under climate change. A pricing model should incorporate a trend adjustment, and the trend should be explicit rather than buried in a loading factor. The actuarial technique is to model the fire intensity parameter \(\lambda_t\) as a function of climate covariates—temperature, drought index, fuel moisture—and to project those covariates forward under a climate scenario. The resulting premium is scenario-dependent, which is uncomfortable for a regulated rate filing but honest about the uncertainty. Regulators should require a climate-adjusted rate filing for forest insurance, just as they require stress testing for banks. The alternative is to underprice the risk and discover the error in a bad year.
Correlation is the pricing problem that distinguishes forestry from most other lines. The standard actuarial approach to correlated risk is to model the portfolio loss distribution with a copula or a factor model, calibrate the tail, and buy reinsurance against the upper percentiles. For forest portfolios, the calibration is difficult because the historical record contains few large correlated events and the tail is estimated from a small sample. The practical solution is to use a catastrophe model—a simulation of fire ignition, spread, and suppression under stochastic weather—rather than to rely on historical loss experience. Catastrophe models for wildfire exist (RMS, AIR, and several academic models), but their application to Chinese forestry is limited by the absence of validated fuel maps and suppression-response data. Building those inputs is a public-good investment that no single insurer will make alone.
8.3 Policy Support and Market Development
Forest insurance does not pay for itself at unsubsidized rates. The expected loss ratio in most programs is high, the expense ratio is high because of loss adjustment, and the capital cost is high because of correlation. A commercially priced policy would cost the forest owner more than the expected loss by a wide margin, and uptake would be minimal. Every large forest insurance market in the world is therefore subsidized, and the design of the subsidy determines the market's shape.
8.3.1 The Chinese program
China's central forest insurance program is the largest in the world by number of policies, though not by premium. It was established on a pilot basis in 2009 under the Ministry of Finance's agricultural insurance premium subsidy framework, and it now covers ecological public-welfare forests and commercial forests. The subsidy structure is tiered by region: in the central and western provinces, the central government subsidizes 50 percent of the premium, with provincial and county governments contributing a further share, and the forest owner paying the remainder—typically 20 percent or less. In the eastern provinces, the central share is lower. The program is administered through the provincial branches of the major insurers, with the forestry administration providing loss assessment support.
By the early 2020s, the program covered roughly 150 million hectares of forest, with premium income in the range of 3 to 4 billion yuan annually and sum insured in the range of 1.5 to 2 trillion yuan. The loss ratio has been volatile: low in most years, spiking in drought years. Coverage is broad but shallow—the sum insured per mu is often set below full timber value, so the policy covers a fraction of the owner's exposure. This is a deliberate design choice: it keeps the premium affordable within the subsidy envelope and limits the insurer's tail. The cost is that the insurance does not fully protect the owner's balance sheet, which matters most for commercial forest operators and for the carbon-reversal exposure discussed above.
The program's weaknesses mirror its strengths. The subsidy is input-based—it reduces the premium per policy—rather than performance-based, so it does not reward risk reduction. A forest owner who invests in firebreaks, fuel management, or early detection pays the same rate as one who does not. This is the moral-hazard problem in its subsidy form, and it is the same problem I return to in Chapter 13 when discussing incentive design. Loss assessment is administratively determined, which creates scope for disputes and delays. And reinsurance arrangements are concentrated through China Re, which means the tail is ultimately borne by the state—a sensible arrangement for a public-good risk, but one that should be priced transparently rather than hidden in the reinsurance treaty.
8.3.2 Public-private partnerships
The alternative to input subsidies is a public-private partnership in which the government bears a defined layer of the risk—typically the catastrophic tail—and the private market bears the attritional layer. This is the structure used in several national catastrophe programs: the U.S. National Flood Insurance Program (troubled but instructive), the French CatNat scheme (a public reinsurance layer above private insurers), and the Mexican and Caribbean parametric sovereign programs. For forestry, the analogous design is a government-backed reinsurance facility that covers fire events above a threshold, with the private market retaining the smaller, more frequent losses.
The advantage of this structure is that it separates the insurable from the uninsurable. The attritional layer is diversifiable and can be priced commercially; the catastrophic layer is correlated and is a public-good risk that the state is better placed to hold. The disadvantage is that it requires a clear trigger and a transparent pricing of the government layer, or it becomes an implicit subsidy that distorts the primary market. The Chinese program's current structure is closer to a full public backstop than to a layered partnership, because the state-owned reinsurer absorbs the tail and the subsidy keeps the primary premium below cost. A move toward an explicit layered structure—with a published attachment point and a priced government layer—would improve transparency without abandoning the public-good rationale.
The general design of public-private risk sharing returns in Chapter 9, where cooperative and mutual finance provides a complementary buffer. For forestry, the specific point is that the government layer should be designed around the correlated event, not around the individual policy. A parametric sovereign or provincial facility that pays out when burn area exceeds a threshold is a cleaner instrument than a per-policy subsidy, because it targets the risk that the private market cannot hold.
8.3.3 Uptake and its determinants
Why do some forest owners insure and others do not? The empirical literature on agricultural insurance uptake gives a consistent answer: uptake rises with the size of the expected loss, the availability of credit tied to insurance, and the perceived reliability of payout; it falls with basis risk, premium cost, and distrust of insurers. Forestry adds two factors. The long rotation means an owner who will not harvest for twenty years discounts the value of a one-year policy heavily, and annual renewal creates uncertainty about future availability and price. The public-good nature of ecological forests means a state forest farm's objective function includes conservation and carbon, not just timber revenue, and its willingness to pay for insurance depends on how the parent administration accounts for risk.
The credit linkage is the most powerful lever. In China, forest insurance has been bundled with forest-rights mortgage lending: a bank lending against a forest stand requires the stand to be insured, and the insurance policy is assigned to the bank as collateral. This is the same mechanism that drove crop insurance uptake in the 2010s, and it works because it aligns the bank's interest with the insurer's. The bank wants the collateral protected; the insurer gets a captive policyholder; the owner gets access to credit. The limitation is that the linkage only reaches owners who borrow, which excludes many smallholders and many ecological forest managers.
The distrust factor is real and should not be dismissed as irrational. A forest owner who has waited two years for a loss adjustment, or who has seen a claim denied on a technicality about pre-existing pest damage, will not renew. The Chinese program's administrative loss assessment has generated exactly this kind of complaint, and provincial forestry bureaus have responded by standardizing assessment procedures and introducing satellite verification. The payout experience is the product: an insurance policy that pays slowly is worth less than its actuarial value, and the uptake data will show it.
8.3.4 Market development beyond China
The comparison with other jurisdictions sharpens the analysis. In the United States, private forest insurance is a small, specialized market: the Forest Landowners Association and a few Lloyd's syndicates write policies for large private timberland owners, typically with deductibles of 5 to 10 percent and coverage limits well below full value. The federal government does not run a forest insurance program, though it provides cost-share assistance for fuel reduction and disaster relief after major fires. Most small non-industrial forest owners are effectively self-insured, and the loss from a severe fire falls on the owner and on federal disaster programs.
In Europe, forest insurance is more developed in the Nordic countries and in Spain, where public consortiums (the Consorcio de Compensación de Seguros in Spain) cover catastrophic risks including forest fire as part of a broader scheme. The Spanish model is instructive: the public consortium provides a layer above private insurance for extraordinary risks, funded by a surcharge on all property policies. It pools forest fire with flood and earthquake, which diversifies the tail across perils, and it is mandatory for the underlying property coverage. The lesson for China is that pooling across perils and across regions reduces the cost of the catastrophic layer, and that a mandatory or near-mandatory attachment improves the pool's stability.
Japan's forest insurance, operated through the national agricultural insurance scheme, covers fire and certain natural perils for private forest owners with a government subsidy and a reinsurance layer. The Japanese scheme's long experience illustrates a point that applies everywhere: the loss ratio is low in most years and the program's political sustainability depends on the visibility of payouts after a disaster. A program that collects premiums for a decade without a major fire will face pressure to reduce the subsidy or loosen the coverage, just when the risk is accumulating.
8.4 The Carbon Dimension and the Boundary of Insurability
The carbon asset is the feature of forestry that most clearly distinguishes this chapter from the agricultural insurance discussion in Chapter 7, and it exposes the boundary of what insurance can do.
A forest carbon project generates credits by sequestering carbon relative to a baseline. The credits are sold to emitters, and the project owner receives revenue. The risk is reversal: if the forest burns, the sequestered carbon is released, and the credit must be replaced. Registries handle this through buffer pools—a fraction of issued credits is withheld and used to cover reversals—and through reversal-risk contracts that allocate liability among the project owner, the registry, and sometimes an insurer. The buffer-pool approach is a mutual insurance arrangement: all projects contribute, and the pool pays for reversals. It works for small, idiosyncratic reversals; it strains when a single fire reverses a large project, because the pool may be insufficient and the remaining projects bear the cost.
Commercial insurance for carbon reversal is nascent. A few Lloyd's syndicates have written parametric policies that pay when a fire exceeds a defined severity within the project boundary, and the payout is used to purchase replacement credits. The pricing is difficult because the credit price is policy-dependent and volatile, and because the correlation between fire events and credit prices is not zero—a bad fire year may also be a year of policy tightening that raises credit prices, increasing the cost of replacement. The instrument is therefore a double hedge, and its value depends on assumptions about policy that no actuary can validate.
The broader question of pricing policy-dependent assets returns in Chapter 11, where I discuss the supply of early-stage green capital. For forestry, the specific conclusion is that carbon reversal risk is insurable only in a limited, parametric form, and that the bulk of the risk will remain with the project owner and the registry buffer. A forest insurance policy that covers timber but not carbon is therefore incomplete, and a policy that covers both is expensive and difficult to price. The honest design is a layered structure: a timber layer priced commercially, a parametric carbon-reversal layer priced off the catastrophe model, and a residual risk held by the buffer pool and the project owner.
8.4.1 What forest insurance cannot fix
The policy debate often assumes that insurance can solve the forestry risk problem if only the subsidy is large enough. It cannot, and the limits deserve stating plainly.
Take liquidity first. A thirty-year rotation ties up the owner's capital for three decades, and an annual insurance policy does nothing about that. The instrument that addresses the time dimension is long-term finance—forestry bonds, carbon-revenue securitization, patient equity—which I discuss in Chapters 10 and 12.
Basis risk is the second limit. Better index design and hybrid structures can reduce it, but the residual falls on the owner. For a smallholder with a single stand, the residual can be the whole loss. When a burn-area index fails to trigger because a fire burned 400 hectares in a grid cell whose threshold was set at 500, the owner of those 400 hectares receives nothing, and no amount of actuarial refinement changes that outcome for that owner in that year.
The third limit is the one that matters most for policy. The most effective forest risk management is fuel treatment, firebreak maintenance, early detection, and rapid suppression. Insurance transfers the residual financial loss after those measures; it does not prevent the loss. A subsidy design that pays for insurance but not for fuel management is buying protection against a risk that could have been reduced. Guangdong's provincial forest fire prevention program, which funds fuel-reduction crews alongside insurance subsidies, has cut the average burned area per fire since the 2010s, and the insurance program's loss ratio has fallen with it. The two budgets should be designed together.
Finally, an insurer cannot underwrite a price that depends on future climate policy. The carbon asset's value depends on the continuation and stringency of that policy, and the parametric structures described above are the best available approximation. They should be understood as such.
8.5 The Refrain
In 2022, a state forest farm in Fujian insured 12,000 mu of Chinese fir against fire and wind under the central program. The premium came to roughly 90,000 yuan, of which the farm paid about 18,000. When Typhoon Doksuri passed through in July 2023, the farm lost about 800 mu of standing timber. The loss adjustment took eleven weeks, involved two rounds of field survey and a satellite cross-check, and paid out just over 1.1 million yuan. The farm's manager told a provincial forestry bureau review that the payout saved the season's replanting budget, but that the eleven-week wait had forced the farm to borrow at short-term rates to cover wages. The policy did its job—the timber loss was transferred—and it also showed where the product stops: the liquidity gap, the uninsured carbon, and the fuel that had accumulated on the unburned stands remained the farm's problem.
That case carries the argument of this chapter. Forest insurance identifies the risk—fire, pest, climate, carbon reversal—through the underwriting and loss-assessment process, and it buffers the financial consequence through the indemnity or parametric payout. A program that does one without the other will fail. A subsidy that pays the premium but does not require risk assessment buys uptake without risk discipline. An index product that pays quickly but does not measure the owner's actual loss buys speed without accuracy. A carbon-reversal policy that pays on a parametric trigger but does not verify the reversal buys appearance without substance. The identification and the buffering have to be designed together, and the design has to match the correlation profile of the risk.
The next chapter takes the buffering function into a different institutional form. Cooperative and mutual finance offers a way to pool risk among members who share information and can monitor each other—a structure that addresses the moral-hazard and correlation problems that plague commercial forest insurance. The Chinese farmer cooperative, like the forest owner's mutual, is a buffer built from the bottom up. The question is whether it can scale to the correlated catastrophe, and the answer will come from the loss ratios of the cooperatives that have already tried.