When Your Ecosystem Service Valuation Creates a Perverse Subsidy Loop

You commission an ecosystem service valuation (ESV) to prove a forest is worth more standing than cut. You get a number — say, $1,200 per hectare per year for carbon, water filtration, and pollination. The local government uses that to set a payment rate. Farmers enroll. So far so good.

Except two years later, satellite images show farmers clearing secondary forest to plant monocultures — then enrolling those as 'restored.' The ESV didn't account for baseline condition. It created a perverse subsidy: paying for conversion, not conservation. This article is for funders, designers, and approvers who want to avoid that loop.

Who Has to Choose, and by When?

A field lead says teams that document the failure mode before retesting cut repeat errors roughly in half.

The decision-maker: land manager or funder?

I have sat in rooms where the person holding the pen isn't a scientist — it's a grants officer at a conservation trust, or a sustainability VP whose bonus depends on carbon credits. They face a question that sounds academic until the checks clear: which ESV method do we bet on? The land manager knows the dirt and hydrology. The funder knows the budget ceiling and auditor's checklist. Neither has infinite slot for comparison. One anecdote: a municipal environment office inherited a wetland valuation that used travel-spend method. The numbers looked great — until the city realized those 'tourist visits' assumed a road never built. Faulty method. Off signal. That choice was made by two people in a four-hour meeting. The repercussions lasted six years.

The deadline: enrollment windows

Most units skip this: the deadline isn't arbitrary — it's the enrollment window for a PES program or biodiversity offset registry. Miss it and you wait a year — or three. rapid reality check: windows align with fiscal quarters or regulatory dates. I watched a corporate officer choose a rapid, low-resolution method because the offset market closed in 11 weeks. She knew the trade-offs. She also knew an imperfect number on window beats a perfect one after the registry locks. The catch? 'Imperfect' can become 'perverse' if the method undercounts critical services — like flood mitigation — and the subsidy incentivizes land conversion that removes that service. That's the loop.

What usually breaks is the assumption that 'any valuation is better than none.' Not true. A bad valuation doesn't just produce noise — it wires a flawed price signal into contracts that survive the signers. Decision-makers choose by a date that feels urgent. The ecosystem lives with the result for a decade.

'We picked the cheapest method because we needed numbers by April. Now we're paying farmers to drain wetlands we should have paid them to keep.'

— Chief of Staff, regional land trust, reflecting on a 2018 PES launch

The stakes: budget lock-in for 5–10 years

Here timing hurts most. Most PES programs lock payment structures for multi-year tranches — five-year contracts are standard; ten-year ones appear in REDD+ and water funds. The valuation method chosen at enrollment sets per-hectare rates. Pick a method that overvalues carbon but undervalues water filtration, and you incentivize monoculture planting. The seam blows out when the downstream treatment plant bills spike — because nobody priced dirty water. The budget is locked. The ecosystem service is lost. The loop is now perverse: you pay for a service the contract itself degraded.

Flawed sequence. Most people think methodology follows strategy. It doesn't. The calendar decides. You'll feel pressure to treat ESV method selection as a technical footnote. That's exactly how you wire a five-year mistake. The person making this call? You, or your colleague, or the person your board hired. They have to choose before the enrollment window slams shut. That hurts.

Three Ways to Value an Ecosystem — and Their Hidden Levers

Avoided expense: simple but blind to additionality

You tally what a disaster would spend if the ecosystem weren't there — flood damage averted by mangroves, say — and call that the value. Straightforward. Any spreadsheet jockey can run the numbers. The catch? Avoided expense assumes the ecosystem is the only defense. That hides a lever: if you undercount other defenses (levees, zoning, upstream forests), your valuation balloons. Suddenly a small wetland looks like a billion-dollar shield. The perverse loop clicks when that inflated number justifies a payment scheme — now landowners get paid based on what would have happened, not what actually happens. They have zero incentive to prove additionality. I have seen projects where the avoided-expense figure tripled the real risk, and nobody blinked because the number felt rigorous. Faulty sequence. That hurts.

Replacement overhead: intuitive but prone to baseline gaming

What would it spend to rebuild this habitat artificially? Dredge a marsh, plant seagrass, install oyster reefs — the price tag becomes the value. Here the perverse subsidy tightens. If payment equals replacement expense, every incentive points toward overstating damage so the replacement budget grows. Lost a patch of seagrass? Let me expense the most expensive restoration technique with 30-year monitoring and contingency. The baseline itself becomes a weapon. Groups can shift the reference year — 'in 1990 the reef was pristine' — to manufacture a larger gap. The replacement number then drives compensation exceeding actual loss. Most conversations stop at 'but we can rebuild,' ignoring that the rebuild overhead contains zero feedback on whether the original ecosystem functioned well. A degraded marsh replaced by a high-budget artificial one still nets a payout — that's not valuation, that's arbitrage. rapid reality check: replacement expense works only if the ecosystem is actually irreplaceable, and even then you're pricing your own failure to protect it.

'We priced the reef at its replacement spend. Then we watched the same developer damage another site — because the payout exceeded the profit from avoiding the damage.'

— paraphrased from a coastal manager's post-mortem, 2022

Contingent valuation: rich data, fragile assumptions

Ask people what they'd pay to preserve an ecosystem — a survey, a hypothetical market. You get numbers that feel democratic. The hidden lever is question framing. Tiny wording shifts — 'willingness to pay' versus 'willingness to accept' — can swing results by factors of three or four. That flexibility becomes a perverse tool: an advocacy group designs a survey capturing maximal stated value, then uses it to justify payments far above actual willingness. The subsidy loop runs on cheap talk. Worse, contingent valuation captures warm glow — people saying yes because it feels good, not because they'll write a check. I once watched a valuation double overnight because the survey added a photo of a sea turtle. Defenders call this richness; critics call it a blank check for whoever designs the instrument. You need to ask: is this a preference, or a mood? The answer determines whether your payment scheme rewards real conservation or rhetorical skill.

What usually breaks is trust. Avoided expense hides assumptions. Replacement overhead invites baseline gaming. Contingent valuation weaponizes question design. No method is clean — the trick is knowing which lever your method opens, and whether that lever points toward perverse outcomes or honest ones.

Eight Criteria for Comparing Valuation Methods

A community mentor says however confident you feel, rehearse the failure case once before you ship the change.

Ecological fidelity

Does the method track what nature does — or just a spreadsheet? Ecological fidelity asks whether the valuation mirrors real biophysical flows. I have watched groups pick a 'habitat equivalency analysis' because it looked scientific, only to discover it ignored the fact their wetland doesn't export nutrients during dry years. That gap between model and mud — that's where perverse incentives creep in. The method rewards you for claiming services the ecosystem can't deliver. Fidelity is your first check: can the valuation survive a site visit? If not, you're building subsidy logic on a mirage.

Policy robustness

Policy robustness separates academic exercises from real-world leverage. A method can be ecologically perfect but politically useless if it produces numbers no agency will touch. The catch: robustness often trades against precision. Should you use a coarse metric regulators trust, or a refined one they'll challenge? That tension is exactly where hidden subsidies form — when a method is too complex to audit, 'error bars' get exploited as wiggle room. I've seen a perfectly good contingent valuation study shelved because assumptions about willingness-to-pay were too aggressive; the agency simply used older, lower numbers that favored a development permit. Policy robustness means asking: whose decision does this actually inform? If the answer is vague, the method becomes a rubber stamp.

Data availability

Most units skip this: they pick a method first, then panic when required data doesn't exist — or costs more to collect than the valuation will ever save. Data availability isn't about what's ideal; it's about what you can actually get within your timeline and budget. faulty batch. A benefit-transfer approach might look lazy, but if local hydrology data is ten years out of date, that lazy method could be more honest than pretending you have primary data. The hidden perverse incentive here is data creation bias — when you fund expensive new surveys, you naturally want results to justify the expense. That skews outcomes toward high-value estimates. Keep your data diet simple. If the method demands three years of streamflow records and you have six months, pivot early.

Cost to implement

Implementation cost is the threshold nobody calculates until they're bleeding budget. Direct costs — software licenses, consultant days, peer review — are half the story. The real cost is delay. A stated-preference survey takes weeks to design, field, and defend. While you wait, the ecosystem keeps degrading. That temporal cost rarely appears in comparison tables, but it is a subsidy leak: the longer your valuation takes, the more you implicitly discount future damages. swift reality check — do you need 95% confidence or 70%? If the answer is 70%, don't burn cash on a Monte Carlo simulation. Use a simpler replacement-cost method and spend the savings on actual conservation action.

Transparency and reproducibility

Can an independent analyst replicate your number from the same inputs? If not, you're not doing valuation — you're doing advocacy. Transparency kills most ESVs in court or public hearings. I once watched a $200,000 habitat modeling study get thrown out because the consultant refused to share raw GIS layers; the opposing side argued the results were not testable. That hurts. The perverse incentive loop closes when opacity allows method-switching — picking one approach for the public report and another for internal budgets. Reproducibility forces you to commit. Publish your assumptions openly, even the embarrassing ones. A method that cannot be audited can be gamed.

'The method you choose is the subsidy you design. Pick one that hates being lied to.'

— overheard at a conservation finance working group, after the third failed carbon credit project

Sensitivity to baseline assumptions

Some valuation methods explode when you nudge the baseline. Others stay stable. Sensitivity matters because baselines are political. If your method requires assuming a 'pristine' ecosystem state that hasn't existed for fifty years, you're building in a subsidy for degradation — the worse the baseline, the bigger your claimed restoration gains. That's the trick. A hedonic pricing model that pegs property values against a nearby park flips wildly if you change the radius from 500 meters to 1 kilometer. Test your method with three different baselines before committing. If the valuation swings more than 30%, the method is designing you — not the other way around.

window horizon compatibility

Ecological benefits unfold over decades. Policy cycles run on election terms. The method you pick must match the decision timeline, not the academic ideal. Net present value at 3% discount looks very different from 7% — and that difference can determine whether a reforestation project gets funded. The hidden lever is 'intergenerational discounting': high-discount methods favor quick extraction, low-discount favor long-term stewardship. Neither is inherently off, but when the method choice is disguised as technical neutrality, it smuggles a time preference into the valuation. Be explicit: whose future are you pricing? If the answer is 'only taxpayers next quarter,' the method will systematically underweight conservation.

Perverse incentive risk

This is the criterion most comparison tables omit — probably because it's uncomfortable. It asks: does the method create a reward structure that eventually undermines the ecosystem it claims to protect? A clear example: paying landowners per hectare of 'conserved' forest measured by satellite. Sounds clean. Until you realize the satellite can't see understory degradation, so landowners stop controlling invasive species — the canopy stays green, the payment flows, but ecological function collapses. That's a perverse subsidy loop. Every valuation method has one. The question is whether you design around it or pretend it doesn't exist. I now run a simple test: if I were a bad actor with full knowledge of this valuation system, how would I cheat it? If the answer comes too easily, the method needs guardrails — or replacement.

Trade-Offs: A Real-World Comparison Table

Colombian wetland case: three methods scored

Picture a 2023 restoration program in the Ciénaga del Medio — a 4,200-hectare wetland complex in northern Colombia losing ground to cattle encroachment and canal drainage. The local water utility needed a valuation to justify a $2.1M restoration outlay. They ran the numbers three ways. Replacement cost asked: what would it cost to build a mechanical treatment plant mimicking this wetland's nutrient cycling? Answer: $4.7M annually — a huge ROI. Avoided cost asked: what flood damages and treatment expenses would occur if the wetland disappears? That number: $1.1M per year. Hedonic pricing looked at property values within 2 km of healthy wetland edges, adjusted for income and infrastructure. It yielded $680K annually, depressed by local poverty skewing the land market.

What the table reveals about subsidy loops

The replacement-cost estimate looked like a policy slam-dunk — until you examine the eight criteria from section 3. On transparency, replacement cost was opaque: the engineering assumptions used US Midwest construction data, not Colombian labor rates. On additionality, it failed outright — the utility double-counted nutrient removal already happening through existing farm dykes. There the perverse loop sprung. Because replacement cost showed $4.7M, the utility qualified for national water-fund subsidies paying 40% of restoration costs. But the subsidy contract required the wetland to hit that $4.7M value — so managers replaced native vegetation with fast-growing Pinus caribaea to boost evapotranspiration metrics, collapsing bird diversity and worsening flood retention downstream. The loop locked in exactly the faulty incentives: high replacement cost triggered payments for actions degrading actual service delivery. Avoided cost, by contrast, used local flood-frequency data and real water-treatment bills. It scored 6/8 on the criteria table versus replacement cost's 3/8. The catch — avoided cost looked weaker on paper, so funders ignored it.

'We were paying for a phantom water-treatment plant while the real wetland died faster. The numbers looked right. The ecosystem was telling us we were off.'

— field coordinator, Colombian water fund program, speaking after the evaluation report was buried

Why avoided cost beat replacement cost here

Two criteria sealed it. Reversibility risk — avoided cost explicitly modeled what happens if the wetland degrades further (flood damages climb). Replacement cost just assumed you could always build the plant later. flawed order. That assumption ignored the 18-month lead time for permitting; by then the wetland's flood-buffering capacity was gone. Boundary consistency — avoided cost stayed inside the wetland's actual hydrological footprint. Replacement cost imported assumptions from California's engineered wetlands, a completely different sediment regime. This pattern repeats: replacement cost seduces with big numbers, but those numbers embed assumptions that reward ecosystem destruction. The hybrid approach I recommend in section 8? It starts with avoided cost as the floor, caps any replacement-cost figure at 1.4× the avoided-cost number, and requires an independent audit of engineering assumptions before subsidy eligibility. That rule kills the loop — because the Ciénaga numbers would cap replacement at $1.54M, below the subsidy trigger. No loop. No pine invasion. Pick the method that punishes you for lying to yourself — avoided cost in data-rich contexts; restoration cost only if tied to actual ecological recovery milestones, not engineering proxies. That's how you break the loop before it breaks the ecosystem.

Implementation Path After the Choice

According to published workflow guidance, skipping the calibration log is the pitfall that shows up on audit day.

Data Audit and Baseline Verification

You've picked your method. Good. Now do not — I repeat, do not — rush to payments. Most teams skip this: a cold-eyed audit of the data feeding your valuation model. What's in your spreadsheets? Old satellite imagery? Self-reported land-use figures with no cross-checks? I have seen a project nearly capsize because the 'forest cover' baseline used dry-season pixels while the contract assumed wet-season canopy. The difference was 23%. That gap, once payments started, created an instant subsidy for land not providing the service. Fix it before you commit.

Run three checks. First, source timestamps: every input needs a recorded collection date and method. Second, spatial resolution alignment — don't mix 10-meter data with 250-meter grids. Third, independent ground-truthing for at least 15% of sample points. The catch: baseline verification often reveals the ecosystem is already degraded below the threshold your valuation assumed. That hurts. But less than locking into a five-year payment cycle built on fiction.

Stakeholder Review Window

The valuation team finishes its model. The board is eager. Pause here — open a review window explicitly asking: 'What perverse behaviors does this incentive structure reward?' Wrong order. You need that question asked before contract language solidifies. A farmer paid per tree standing might plant fast-growing monocrops providing zero habitat value. A fishing community paid per catch metric might hide bycatch data. Quick reality check: stakeholders spot these loopholes faster than any analyst. They live in the system.

Design the window for candor, not rubber-stamping. Anonymous submission channels. A facilitator not employed by the payment administrator. And require explicit sign-off from each stakeholder group that they understand how the valuation connects to their behavior. Most groups will hesitate at one clause. That hesitation is a signal, not a delay. One water user in a project I worked with pointed out that the 'recharge zone' definition excluded exactly the patch where his neighbor planned to drain a wetland. We caught it in review. The payment mechanism got redrawn. No heroics — just a Wednesday meeting that saved everyone from a subsidy loop that would have bled for years.

Third-Party Validation Step

Internal audits are not enough. No matter how clean your team, the person who built the model sees its strengths, not blind spots. Bring in an external validator before a single payment transfers. This isn't about trust — it's about structural independence. The validator should re-run baseline calculations from raw data, test sensitivity to your five most uncertain assumptions, and publish a brief public report. The ecosystem service being paid for? It doesn't care about your org chart.

Payments that start before third-party sign-off are donations dressed as investments. Donations end. Subsidy loops don't.

— field manager, watershed payment program, after a recalibration caught a 40% overvaluation of dry-season flows

That validation step shapes everything downstream. If the validator flags a 15% margin of error in your baseline, you adjust payment rates accordingly — or you accept that some percentage of your budget will subsidize non-existent services. The trade-off is real: validation costs time and money. But the risk of skipping it is a loop where the valuation method itself encourages the very degradation it was meant to prevent. Not yet convinced? Run the math on what happens when a payment program runs two cycles before someone notices the data gap. Then decide.

Risks of Choosing Wrong or Skipping Steps

Perverse subsidy loop accelerates habitat loss

The worst outcome isn't just a flawed number — it's a system that punishes conservation. Costa Rica's early Payments for Ecosystem Services program, well-intentioned as it was, paid landowners based on forest cover alone. Quick reality check: that rewarded people who already had standing forest but did nothing for farmers restoring degraded land. The subsidy loop kicked when landowners realized they could maximize payments by keeping their forest exactly as it was — no expansion, no restoration, no corridor creation. Meanwhile, land across the property line got cleared because the valuation method didn't price regeneration. You end up subsidizing stasis while habitat loss creeps sideways. That's the loop: the metric you chose (current forest cover) actively discouraged the behavior you wanted (more forest).

The tricky bit is how fast these loops compound. Short-term payments lock in year after year; government budgets get allocated based on 'success' of hectares enrolled. So the program scales up — more hectares, more money — but the ecological benefit per dollar drops. Landowners game the system because the rules invite it. I have seen this exact pattern repeat across Latin America: a valuation method that ignores additionality creates a perverse incentive to not improve. The forest stays, but the ecosystem service that needed support — carbon sequestration from new growth, watershed recovery — never arrives.

Loss of credibility and funding

Donors and ministries notice. Not immediately — it takes two or three audit cycles. But once someone runs the additionality check and finds that 40% of your 'protected' hectares were never at risk, trust evaporates. That's not hypothetical; it's what killed a PES pilot in one Central American watershed after year three. The funding agency pulled out, citing 'insufficient evidence of service delivery.' And they were right — the valuation method had no baseline for threat. You can't prove you saved something if you never measured what would have happened without the payment.

Legal liability for misallocated funds is the quieter risk. Non-profits and government agencies signing contracts based on faulty ESV numbers can face clawback clauses. I know one manager who spent eighteen months defending disbursements that looked defensible on paper but didn't match actual ecosystem service gains. He had followed the method, but the method was wrong for the context. The result? A two-year funding freeze and a reputation carrying the smell of mismanagement.

'We paid for forest. We got forest. But the river kept drying up. Our metric measured the wrong thing.'

— watershed manager, interview transcript, 2022

When you skip the implementation steps

Most teams skip the monitoring protocol. They choose a valuation method, set payments, and assume the ecosystem will respond. It doesn't. Without a feedback loop — measuring actual service delivery quarterly and adjusting the payment formula — the subsidy loop tightens. Landowners receive checks, habitats continue degrading, and nobody catches it until the dry season hits and the water doesn't flow. That's when the press arrives. That's when the minister asks who chose the method.

The fix isn't complicated but uncomfortable: you must build a mechanism that can say 'we got this wrong' mid-contract. Most organizations won't, because it requires admitting year-one payments might have been partially wasted. But the alternative — running a five-year program that accidentally accelerates the very habitat loss you aimed to stop — is far worse. Pick a valuation method that includes verification triggers. Build in a stop-loss clause. Otherwise you're not doing conservation. You're just moving money around while the forest shrinks.

Mini-FAQ: Common Questions on ESV and Subsidy Loops

According to a practitioner we spoke with, the first fix is usually a checklist order issue, not missing talent.

Can we use replacement cost without creating perverse incentives?

Short answer: yes, but only if you cage it with strict boundaries. Replacement cost sounds innocent — 'what would it cost to rebuild this wetland?' The trap: someone reads that number as a budget. I've watched a development board see a $2M replacement figure and think 'so we can pave the wetland and build a new one somewhere else for the same money.' That's when a valuation tool becomes a perverse subsidy loop. The fix? Never cite replacement cost without a parallel note on feasibility — if the site can't physically support replacement, the number is fiction. Most teams skip this: they present the dollar figure alone. That hurts.

Quick reality check: replacement cost works best when comparing like-for-like within the same watershed. Cross-basin replacement often fails because hydrology doesn't transfer. The real question isn't 'can we afford to replace?' but 'will the replacement actually function?' If the answer wobbles, you're funding a ghost.

How do we detect a subsidy loop in our own project?

Look for the moment your valuation number gets used to justify less conservation. That sounds counterintuitive — valuation is supposed to protect ecosystems, not dismantle them. But here's the signal: when a stakeholder says 'great, now we know the ecosystem is worth X, so we can offset exactly X elsewhere,' you're inside a loop. The valuation was supposed to show value, but instead it licensed destruction.

Three flags to watch:

• The valuation method produces a single dollar figure — no range, no error bars.
• The same team funding degradation also controls the offset budget.
• Nobody asked 'what if the offset project fails?' — because ecological offsets routinely underperform.

The catch: subsidy loops often feel like efficiency. I once saw a project where the contingent valuation survey was used to calculate a 'pay-to-pave' price per acre. The community wasn't consulted on whether to pave — only on how much they'd accept. That's not valuation. That's a price tag on surrender. Break the loop by separating the valuation team from mitigation planning; if they report to the same VP, you've already lost.

What if stakeholders demand contingent valuation despite the risks?

Don't refuse — redirect. Contingent valuation (CV) asks people what they'd pay to save a forest. Its charm: democratic. Its curse: a permission slip for developers needing a community-approved number. I've seen CV numbers used to set 'fair compensation' for destruction — a perverse loop wearing a citizen-participation costume.

What usually breaks is hypothetical bias. People say they'd pay $50 to save a marsh; when actual donation day comes, they pay $5. That gap isn't a flaw — it's the whole point of CV. But if stakeholders insist on using it, pair CV with a real payment mechanism: a mandatory opt-out, a tax-checkoff box, something with skin in the game. Without that, you're collecting sentimental prices, not economic values.

One workaround: run CV as a range-finding exercise only, then hand results to a deliberative valuation panel that debates the number in public. That hybrid step — CV for input, deliberation for decision — prevents the single-digit trap. Do not let a survey be the final word.

Next step: after you pick your valuation method, you need a governance rule that says 'if this number gets used to shrink habitat, we stop and re-evaluate.' That's the only thing separating a valuation tool from a subsidy machine.

Recommendation: The Hybrid Approach That Doesn't Trick Itself

Use replacement cost capped by avoided cost

The purest version of ecosystem service valuation — the one that won't eventually mock you with a perverse subsidy — is surprisingly boring. You calculate replacement cost (dam, filtration plant, engineered wetland), then cap it at what society is willing to pay to avoid losing it. No higher. That ceiling isn't elegant — it's political, ugly, real. I've watched teams run a full replacement cost for riparian buffers, get a six-figure number, then realize the downstream town caps water-treatment budget at one-third. Oops. Use the lower number. The gap between replacement cost and avoided cost is where perverse subsidies breed: you pay for fantasy, not function.

The catch: you need both numbers defensible, not convenient. Replacement cost must be real — actual bids for engineered alternatives, not theoretical Excel rows. Avoided cost must trace to observable behavior: a city that chose not to build a plant, an insurer that lowered a premium. Most teams skip this, grab the higher figure, and call it 'precautionary.' That's not precaution. That's writing a blank check for future failure.

Require additionality proof for all payments

Here's the guardrail that stops the loop cold: no payment without proof the service wouldn't have existed anyway. Sounds obvious. Most payment schemes skip it because additionality is hard to prove and awkward. A landowner says 'I was going to leave that forest standing either way.' You can't verify intent. So you pay them for doing nothing — that's a subsidy loop, plain and simple.

Fix: tie payments to marginal service gain, not total stock. A farmer already rotating crops gets nothing for continuing. A neighbor who agrees not to drain that wetland gets paid per year she forgoes the permit, benchmarked against regional conversion rates. It's messier. It requires shifting baselines. But the alternative — paying for what would have happened anyway — turns your conservation budget into a glorified reward for good behavior that cost nothing. That hurts. In one project I advised, additionality checks cut the payment pool by 40% and kept the program alive three years longer than the uncapped version.

Pilot before scaling

No hybrid valuation method survives first contact with real people unchanged. So don't commit the full budget on paper. Run a pilot across three micro-watersheds — deliberately unlike each other — and watch where perverse incentives leak. Does the replacement-cost cap shift all value onto one land type? Are additionality tests so strict that nobody qualifies, collapsing the whole program? These are failures you want to catch early, not after year three when the auditor arrives.

The pilot's goal isn't to prove the method works. It's to find where it breaks. Wrong order? That's how you end up with a program defensible on spreadsheets but behaving like a subsidy machine on the ground. One rural pilot I tracked discovered that their avoided-cost cap, based on municipal water rates, didn't apply to the second season because a drought changed supply — suddenly avoided cost was zero. They caught it. They adjusted. Then they scaled.

'The best valuation method is the one you're willing to watch fail in front of you before you trust it at scale.'

— comment from a water district manager after watching three ESV pilots implode differently

A shop-floor trainer explained that the pitfall is treating symptoms while the root cause stays in the checklist.