This study is a randomised evaluation of the effect of a payment for ecosystem services (PES) programme implemented in Western Uganda on forest cover. Specifically, it evaluated the impact of a payment made to private forest owners (PFOs), conditional on preserving the forest, on the change in tree cover during the study period measured using satellite imagery. It further investigated the cost-effectiveness of such programmes in terms of delayed carbon dioxide (CO2) emissions, as well as the social benefit of delayed CO2 emissions. Finally, the impact on the economic well-being of PFOs was assessed.
Deforestation in developing countries is a key priority for international climate policy, as it is the second largest source of carbon emissions. It is particularly an issue in Uganda, which had the third largest deforestation rates in the world between 2005 and 2010.
This study is relevant to Ugandan policy as deforestation rates in the country are even greater in privately owned land, and is mostly driven by subsistence agriculture and domestic demand for forest products. This programme could be a way to address this issue by reducing PFOs’ economic incentives to cut trees.
What is the impact of PES on tree-cutting?
Does PES incentivise shifting tree-cutting to other lands that are not covered by PES contracts?
Do PFOs reforest as a result of the programme?
Do the programme payments outweigh the opportunity cost of cutting trees, such that it benefits socioeconomic wellbeing of PFOs’ household?
What is the cost-effectiveness of this programme in terms of delayed CO2 emissions, and social benefits of delayed CO2 emissions?
The Chimpanzee Sanctuary and Wildlife Conservation Trust, a local non-profit organisation contracted by the Ugandan government’s National Environment Management Authority, implemented the intervention. They reached out to PFOs located in treatment villages and advertised the opportunity to enroll in the programme, offering 70,000 Ugandan shillings, or US$28 (2012 exchange rate) per year, if they complied with the contract. The contract required the PFO to conserve his/her entire forest and prohibited cutting down any medium-sized trees. Additional payment were also offered in exchange for planting tree seedlings. The Chimpanzee Sanctuary and Wildlife Conservation Trust conducted spot checks to monitor compliance with the programme once every one or two months.
Theory of change
There is a market failure due to the negative externality of cutting down trees, and this intervention intervenes as a way to capture this externality in economic terms. The theory of change underlying the intervention is that offering a PES payment to private forest owners provides an incentive to reduce deforestation. This relies on the assumption that PFOs will be aware that the programme is taking place and will agree to take part in it. It also relies on the assumption that these forest owners are actually engaging in deforestation activities and do not adopt a conservationist behaviour incentivised by the payment. If this assumption were wrong, the programme would not have the expected effect on the forest but may be a benefit for household outcomes. Finally, the expected impact of the programme relies on the assumption that under the PES contract, PFOs will not seek to cut trees from nearby land uncovered by the contract, which would simply shift deforestation from one place to another.
To design this study, authors randomly allocated the programme to 56 PFOs residing in the 60 treatment villages. The control group comprised of 535 other PFOs residing in the remaining 61 villages. They estimated the impact of the programme via ordinary least squares regression, controlling for baseline tree cover and a number of variables related to village- and PFO-level characteristics.
The authors selected 121 villages in Hoima and Kibaale districts of the Western region based on the number of PFOs living in each, and conducted a baseline survey with PFOs of each village. Using the baseline survey data, they created two lists of villages in each of the seven sub-counties (the above administrative unit) to ensure balance on a number of covariates, and held public lotteries to determine which of the two lists would receive the treatment.
Authors used satellite images to collect data on forest cover, and remote sensing analysis to improve the precision of the images. Each satellite object or pixel corresponded to a tree or non-tree. They also sampled geographic areas and conducted ground measurements. The unit of observation was the PFO, and they converted pixels or on the ground measurement to an aggregated, PFO-level data. Household outcomes and self-reported tree-cutting were measured using household surveys at baseline and endline. Finally, authors used data from the Chimpanzee Sanctuary and Wildlife Conservation Trust on enrollment, reforestation and payments.
The programme take-up was 32 per cent and authors used an intention-to-treat analysis to avoid selection bias. The main threat to the validity of this study was the risk of shifting tree-cutting to areas where the PES contract does not apply. Authors tested and measured all the potential forms of leakage, through expanding their sample to non-PFO contracted land, by conducting a village-level assessment and a heterogeneity analysis, controlling for distance between PFO forests and government forests. There were risks of spillovers with treatment PFOs cutting trees in control forests as well, which the randomisation at the PFO level partly mitigated. Additional analysis controlling for distance between treated and control villages allowed authors to measure potential increases in demand for timber market in control areas. Authors discuss potential biases coming from PFOs’ expectations of the programme and behaviours adopted as a result. They explore this through collecting endline data on PFOs’ expectations of the programme and controlling for this in their analysis.
Impact on forest cover: The results from the remote-sensing analysis demonstrate that the programme reduced deforestation by 0.27 hectares per PFO eligible for the programme. This result is robust to different specifications and is confirmed by household surveys’ tree-cutting self-report, which is statistically significant and demonstrates that treatment households were 14 per cent less likely to have cut any trees in the last year.
Impact on tree cutting on other lands: Authors found no evidence that PFOs shifted tree-cutting to lands outside the PES contracts.
Impact on reforestation: The programme offered an additional payment for additional trees planted. Authors estimated the impact on reforestation to explain 0.005 hectares of the forest gain.
Impact on socioeconomic well-being: The programme had no impact on socioeconomic outcomes. PFOs’ income was measured through expenditures, and authors found no impact of the programme on this metric. Similarly, loans and child health outcomes did not change between treatment and control PFOs.
Cost-effectiveness: Using results from the intention-to-treat analysis of the programme’s impact on deforestation, and the total cost of the programme including payments and monitoring costs, the cost per averted ton of CO2 emissions amounts to US$0.57.
The social cost of carbon, which represents the benefit of permanently averting carbon emissions, is calculated under different scenarios. Under the authors’ base case, assuming that the PFOs who received the programme catch up on their emission after the programme ends, by deforesting at a 50 per cent higher rate until they reached their level before the programme, this would result in a three-year delay in carbon emissions. They conclude that the benefit of delaying a ton of CO2 emissions’ worth of tree-cutting is US$1.11, which is twice as large as the programme cost.
Implications for policy and practice: Payment for ecosystems services in this context appears to be a cost-effective way to avert deforestation in a developing country. It does not create leakages that could dampen its cost-effectiveness and the social benefit of delaying a ton of CO2 emissions outweighs the cost of the programme.
Implications for further research: Further research is needed to assess longer-term impact of the programme.