Measuring impacts of conservation interventions on human wellbeing and the environment in Northern Cambodia

Researchers: Henry Travers, Tom Clements, E.J. Milner-Gulland

Period: April 2017 – January 2020

Funder: 3ie: International Initiative for Impact EvaluationCollaborators: Malyne Neang, Suon Seng

Collaborating Organisations: Wildlife Conservation Society Cambodia Programme, Sansom Mlup Prey, Cambodian Royal University of Agriculture, Multi-Angles Centre


Project overview:

International context

A critical question for environmental conservation policy is whether interventions incur net costs or provide net benefits to the local people who are the most directly affected. There is now widespread acceptance that environmental conservation policies should, at the very least, do no harm, and where possible should contribute to poverty alleviation (Convention on Biological Diversity, 2008). Protected areas (PAs) are one of the most widely adopted policies, covering >12% of the terrestrial land surface, but the debate around the impacts of PAs has been particularly contentious. Large numbers of case studies document costs that PAs have imposed on local people, such as restrictions on agriculture or access to natural resources. Consequently newer policies, such as payments for environmental services (PES), which provide benefits to local people conditional upon achieving an environmental outcome or a change in behaviour, have gained popularity. It is hypothesised that PES programmes improve human well-being and change behaviour to enhance conservation outcomes, yet this hypothesis has only rarely been tested with empirical data.

Rigorous impact evaluation methods are widely credited with having transformed development policy by quantifying the contribution that specific interventions make to improvements in human well-being. There have been calls for adoption of the same methods in environmental policy. Most published studies to date have focused on assessing environmental rather than social outcomes, for example, using impact evaluation methods to show that PAs and PES policies do indeed protect forests. Only three studies have evaluated the social impacts of PAs in developing countries, and in these cases PAs had no net impact or slightly positive impacts for local people. Few studies have investigated PES impacts on behaviour.


Evaluation questions of interest

This study will investigate the environmental and social outcomes of PAs and PES programmes in the Northern Plains landscape of Cambodia. The Northern Plains is an ideal location to test the impacts of PAs and PES because the interventions were initiated relatively recently, thereby allowing before-after comparisons to be made, and is a relatively homogenous landscape providing suitable control sites. The PAs were established first, in 2005, and included 16 resident villages. From 2008, three PES programmes were rolled out to complement the PA management: (1) direct payments to households for species protection; (2) community-managed ecotourism linked to wildlife and habitat protection; and (3) payments to households to keep within land-use boundaries (“Ibis Rice”).

The impact evaluation will quantify the impact of PAs and PES over time on a panel of intervention and matched control villages and households practicing a range of livelihood strategies in villages in the northern forests of Cambodia, continuing a long-term study that started in 2008.

The evaluation questions are:

  1. Do PAs and PES protect forests in comparison with controls?
  2. Do PAs have positive or negative impacts on human well-being?
  3. Do PES programmes deliver additional benefits to human well-being in comparison with controls?
  4. Do the different environmental conservation programmes have different impacts on different livelihood strategies, focusing on rice farmers, growers of cash crops and non-timber forest product collectors?
  5. Do households reduce land-clearing behaviours as a result of the payment programmes?

We expect the result to have relevance to the national debate around the costs and benefits of environmental conservation nationally in Cambodia, in Southeast Asia and globally. The PES programmes in particular have been widely publicized and are seen as best-practice examples both nationally and in the region, hence the results will have significant policy relevance for the Ministry of Environment, other countries and bilateral and multilateral funding agencies. The results will influence the practice of one of the world’s largest international conservation organization (the Wildlife Conservation Society) and be disseminated through the conservation community, guiding future initiatives.


Empirical Data and Methods

paddy field
Paddy field in Chhep Wildlife Sanctuary, northern Cambodia (Credit: H. Travers)

The study will use a mixture of evaluation techniques and both quantitative and qualitative data collection. Matching methods will be used to select villages and households within villages to form an appropriate comparison group. Panel methods will then be used to follow trends in environmental and social outcome measures for the same villages and households over time, using before-after data where it exists. Combining matching and difference-in-difference methods helps to control for both observable and unobservable sources of bias ensuring that valid comparisons are made. The environmental assessment will investigate deforestation rates around within-PA villages compared to matched controls, and comparing villages inside PAs with PES programmes to those without. Deforestation rates will be measured from 2001 to 2005 (the period before programme implementation) and from 2005 to 2017 (twelve years after programme implementation) at 1-km resolution for >3,000 grid squares around 29 study villages (16 within-PA and 13 matched controls).

The social assessment will investigate household poverty status, measured using the Basic Necessities Survey, for households from within-PA villages compared to matched controls, and for households that benefited from the PES programmes compared to matched controls. Social data exists for a panel of 709 households from within-PA and control villages, surveyed in 2008, three years after PA management started and the year the PES programmes were scaled-up, repeated again in 2011, 2014 and 2017. The study has sufficient power to assess changes in both the environmental and social outcomes. We will also investigate a key causal link in the theory of change, considering evidence for whether households altered land-clearance behaviours in response to the payments, using a long-term dataset from all households in within-PA villages, some of whom participated in the PES programmes and some of whom did not.

A randomised control trial design will be used to investigate the impacts of the PES programmes on human behaviour over the timeframe of the study. Qualitative data collection will be used to validate results, in particular concerning the attitudes, perceptions and behaviour of participating and non-participating households towards the PA and PES interventions.