Redesigning payments for ecosystem services to increase cost-effectiveness

The study addresses how modifications to the design of Payments for Ecosystem Services (PES) can improve conservation outcomes and cost-effectiveness. While PES has been shown to reduce deforestation in many contexts, effectiveness depends on additionality and minimizing inframarginal payments. A key concern is that standard PES allows landholders to enroll only low-risk parcels they would have conserved anyway, while continuing to deforest non-enrolled parcels. The authors hypothesize that requiring full enrollment of all forested land by participants will reduce strategic enrollment, increase additionality, and improve cost-effectiveness, potentially also increasing total forest preserved despite stricter compliance requirements. The context is Mexico’s long-standing PSA program amid budget constraints and high deforestation pressure in Selva Lacandona.

The paper situates its contribution within a large PES literature emphasizing the importance of opportunity and transaction costs, contextual and implementation factors, and program design features. Prior empirical and experimental work has examined participation, perceptions of equity, and collective action, and recent randomized trials have evaluated PES against no-program baselines. However, few studies isolate contract design features’ causal effects on environmental outcomes; one exception varied payment structure to reduce crop residue burning in India. Prior work in Mexico and the Marqués de Comillas region finds PSA can reduce deforestation and deliver socio-economic co-benefits but often experiences partial enrollment and high deforestation on non-enrolled lands, implying substantial inframarginal payments. A study in Uganda suggested full-enrollment PES can yield higher additionality and cost-effectiveness, motivating this trial in Mexico.

Design: Randomized controlled trial comparing a modified full-enrollment PES contract to a standard partial-enrollment PES contract. Participants were randomized at the individual level with stratification by ejido. Setting and sample: Five ejidos in Marqués de Comillas (MdC), Selva Lacandona, Chiapas, Mexico. Eligible participants were 64 individual landholders who applied to Mexico’s PSA program in January 2021 but were rejected for budget constraints. Landholdings average 42 ha total, 19 ha forest. On average, landowners previously chose to enroll 51% and leave out 49% of forest in PSA applications. Intervention: Two one-year contracts (June/July 2021 to June/July 2022) at PSA’s payment rate (MX$1,000 per ha-year). Control (standard PES): allowed enrolling only the forest parcels indicated in their PSA application (partial enrollment). Treatment (full-enrollment PES): required enrolling all forested land within the individual’s holdings. Compliance monitoring used satellite imagery and, if needed, field verification. Enforcement differed from PSA by individual-level contracts and zero payment for any detected non-compliance. Outcomes and measurement: Primary outcome is deforestation of baseline forest from May 2021 to August 2022, measured with high-resolution Planet-NICFI imagery. A random forest classifier (100 trees, max depth 50, mtry=2) using RGB+NIR bands classified monthly pixels (4.59 m × 4.56 m) as forest/non-forest; a pixel was deemed deforested if classified non-forest in that month and the subsequent month. Model AUC = 0.9495. Analyses considered (i) all property forest, (ii) Conafor (PSA-enrolled) area, and (iii) non-Conafor area (parcels participants would have left unenrolled under standard PES). Statistical analysis: Pixel-level regressions of deforestation on treatment with ejido fixed effects and individual-clustered standard errors. Individual-level regressions assessed average treatment effects and heterogeneity by above-median baseline forest area. Compliance and satisfaction were also measured via surveys. Cost-effectiveness: Converted deforestation reductions to hectares using baseline forest areas (control 663 ha, treatment 968 ha). Assumed standard PES reduces deforestation by 2.2% per year on enrolled land (based on prior literature), equating to 1.1% on total land given 49% enrollment, with no spillovers—assumptions chosen to be conservative for treatment gains. Payments totaled MX$313,400 (control) and MX$591,000 (treatment). Costs per avoided hectare and per ton of CO2 were computed using MX$20.036/US$ exchange rate and 550 tCO2/ha carbon stock, with a 3% discount to convert delayed deforestation to permanent emissions equivalence.

• Deforestation reduction (all forest on property): Treatment reduced deforestation by 5.7 percentage points (pp) relative to control (SE 0.021; p = 0.01), from a control mean of 14.2%, i.e., 41% less deforestation.
• Conafor (PSA-enrolled) areas: No meaningful difference (Treat −0.4 pp; p = 0.62) with low control deforestation (1.9%).
• Non-Conafor (unenrolled under standard PES) areas: Large reduction (Treat −13.5 pp; SE 0.036; p < 0.001), or 47% less than control’s 28.8%.
• Individual-level heterogeneity: Averaging equally across landholders, overall difference is not significant; however, effects are concentrated among owners with above-median forest area (net −8.2 pp, p = 0.005). The improvement is driven by non-Conafor parcels.
• Compliance: Control 91% (30/33) vs treatment 71% (22/31); difference p = 0.04. Despite lower compliance under full enrollment, total deforestation fell due to higher averted deforestation per compliant participant.
• Cost-effectiveness: Assuming standard PES averts 1.1% of total land deforestation annually, full enrollment averts 6.8% (1.1% + 5.7%), implying 65.8 ha avoided under treatment vs 7.3 ha under control. Program outlays: MX$591,000 (treatment) vs MX$313,400 (control). Cost per avoided hectare: US$448.29 (treatment) vs US$2,143 (control), a 4.8× improvement in cost-effectiveness. Cost per metric ton of permanently averted CO2 equivalent: US$4.76 (derived from US$0.81/ton to delay emissions, with a 3% discount rate translating delay into permanent equivalent).
• Participant satisfaction: High in both arms; at endline, 100% (treatment) and 90% (control) of respondents expressed satisfaction and interest in participating again; bounding for non-respondents yields 84% vs 82%, respectively.

The trial demonstrates that requiring full enrollment in PES contracts reduces strategic selection of low-risk parcels, thereby increasing additionality and improving both environmental outcomes and cost-effectiveness. The observed 41% reduction in deforestation relative to standard PES arises entirely from parcels that would have been left out under partial enrollment, directly addressing the inframarginality problem. Despite lower compliance under the stricter contract, the net effect on total averted deforestation is positive because compliant participants conserved substantially more area. The results suggest broad relevance for PES programs that currently allow partial enrollment, offering a simple contractual adjustment rather than complex targeting mechanisms or elicitation of private opportunity costs. High participant satisfaction indicates acceptability of the stricter contract in this context. The magnitude of benefits likely reflects context-specific factors: sizable forest endowments leading to partial enrollment, high baseline deforestation pressure from cattle expansion, strong local legitimacy and trust in implementers, and effective monitoring and sanctioning.

This study provides the first randomized head-to-head evaluation of a full-enrollment PES contract versus a standard partial-enrollment design. Requiring full enrollment substantially reduced deforestation and increased cost-effectiveness by 4.8×, primarily by eliminating strategic non-enrollment of high-deforestation-risk parcels. The findings highlight that simple contract modifications can yield large gains in conservation per dollar without complex targeting. Future research should assess scalability in broader settings, test additional design innovations, evaluate multi-year contracts typical of government programs, and examine socio-economic outcomes and equity implications alongside environmental impacts.

• External validity and duration: One-year contracts may overstate compliance relative to multi-year PSA contracts; effects under five-year terms may differ.
• Sample size and setting: Small sample (64 landholders) in five ejidos within a single high-deforestation region; results may not generalize to other contexts.
• No pure control (no-PES) arm: Effects are measured relative to standard PES, not against a no-program baseline; cost-effectiveness relies on assumptions from prior literature for standard PES additionality.
• Implementation intensity: Comprehensive mapping and strict enforcement (zero payment for any non-compliance) may be challenging to scale administratively and politically.
• Outcome scope: Primary focus on deforestation; socio-economic impacts are not analyzed; some endline survey attrition may affect satisfaction estimates despite bounding analyses.