EVALUATING A PAYMENT FOR ECOSYSTEM SERVICES PROGRAM IN UGANDA Joost de Laat (PhD) World Bank July 2015.

EVALUATING A PAYMENT FOR ECOSYSTEM SERVICES PROGRAM IN UGANDA

Joost de Laat (PhD) World Bank

July 2015

Research Question and Design What is the impact of the PES program in

Uganda? Our primary variables of interest:

Rate of deforestation Land use behaviours (tree cutting,

agricultural practices) using household surveys

We divide the forest owners into treatment and control groups randomly, and compare (intermediate) outcomes

Outline

Project Description and Target Population Evaluation Design Description of Sample Population Evaluation activities

Project Location and Target Population

Project Location

2 districts: Hoima and Kibaale

8 sub-counties 140 villages, incl. 4

pilot villages Approx. 1400-2000

forest owners

Project and Evaluation Timeline Sept 2010 – Feb 2011: Census of study region Mar 2011: Selection of sample villages Apr-May 2011: Baseline survey of 1,275 forest

owners May 2011-Jan 2012: Baseline satellite images Jul - Dec 2011: Public lotteries to draw treatment

and control villages Aug 2011 – Feb 2012: Rollout of PES program Feb - Mar 2012: Land mapping and midline survey Aug - Dec 2012: Year 1 payments Jan – Mar 2013: Endline satellite images Jul – Nov 2013: Endline survey of forest owners

Private Forest Owners Generally Poor

1,275 baseline respondents in Hoima and Kibaale

Average household size: 7 93% of household heads are male 43% of household heads completed primary

school Average per capita weekly income: 4 USD Self-reported forest ownership: avg 5.5 ha,

median: 2 ha

Baseline attitudes and behaviour 85% report having cut trees in past year

To sell for money To clear land for cultivation

CSWCT PES Implementation and Participant Characteristics

Description of PES Program Implemented by Chimpanzee Sanctuary and

Wildlife Conservation Trust (CSWCT) Aug 2011 – Feb 2012: CSWCT holds meetings

in each village to explain the program Each forest owner signs a contract with

CSWCT Compliance monitored by community

monitors Payments made to forest owners who meet

the terms of the contract

PES Contract Details

2 components: conservation and reforestation Forest owners must agree to conserve their entire forest Optional to dedicate additional land for reforestation

Annual payments for 2 years, approx $33 per hectare

Conditions: Cannot cut medium-sized trees (10-50 DBH) Can only cut mature trees (>50 DBH) as specified,

depending on number of each species present Cutting small trees (<10 DBH) and gathering firewood

from fallen trees allowed If reforesting, plant seedlings provided by CSWCT

Who joined the PES program? Outreach: 63% of PFOs surveyed by IPA at

baseline participated in CWSCT meetings

Take-up rate: ~ 25-40% of PFOs signed contracts

Why did PFOs join PES program?

Those PFOs that signed up report having done so*: 39% because PES payment better than

other uses 92% because care about the environment 15% to increase tenure security 34% because would have conserved

anyways

*IPA Endline survey data 2013

Why did other PFOs NOT join? 40% simply did not know the program

well by endline (recall: 63% outreach penetration)*

Among those who knew about the program but did not sign a contract*: 6% had disputed/multiple ownership of land 18% wanted to cut trees / payments too

low 34% found contract too complicated*IPA Endline survey data 2013

PES results: relatively high contract compliance

CSWCT Project Monitoring Data: Year 1:

338 plots/PFOs Average 4.8 ha, median 1.4 ha Reforestation: 55% of plots have conservation

only (no reforestation); average area share for reforestation is 19%

82% received some payment (279 PFOs) and 46% (130 PFOs) received 100%

Average payment is UGX 311k, and UGX 177k when < UGX 10 mln payments

PES results: relatively high contract compliance

CSWCT Project Monitoring Data: Year 2:

287 plots/PFOs remain Average 4.7 ha, median 1.4 ha

93% received some payment (266 plots/PFOs), 51% (147 PFOs) received 100% or 125%

Average payment is UGX 325k, and UGX 187k when < UGX 10 mln payments

Measuring the Impact of PES(in progress: results not for citation)

Research Design: Randomized Control Trial

RCT: compare randomly assigned treatment and control villages

For each of the 8 sub-counties, treatment and control villages

Public lottery to draw the treatment group

65 Treatment and 71 Control

Uganda Payment for Ecosystem Services Project

19

Satellite imagery

Tasked Quickbird satellite, before and after program rollout

Resolution: 2.4m Villages are treatment

or control Village-level analysis:

Use government administrative boundaries

Remote sensing Analysis

Remote sensing analysis led by Prof. Eric Lambin and Nancy Thomas at Stanford University

Classify land by forestation status (object-based classification)

Calibrate image area using sampled ground measurements by local forestry partner, NAHI

Calculate biomass, incorporating ground measurements of species type, tree diameter, etc.

From pixels to polygons

Ground-based forest surveys

24

Satellite image Tree Cover Classification

Results: Tree Cover Classification

Tree cover map shows 3 classes: tree cover, other vegetation, and non-vegetated areas

Results: Tree Cover Changes

25

Time 1 (June 6, 2011) Time 2 (February 23, 2013)

Tree Cover Change Map (draft)

Purple = Tree Loss

Green = Tree Gain

Time 1 (June 6, 2011) Time 2 (February 23, 2013)

Tree Cover Change Map (draft)

PES reduced self-reported tree cutting and clearing

Example: self-reported tree cutting reduced by 12.6 percentage points in treatment villages relative to control villages, where 46.7% report tree cutting

PES PFOs reduce community access to lands

Example: PFOs report reducing access by other people to collect firewood by 16.2 percentage points in treatment villages relative to control villages, where 42.1% report allowing others

Satellite measures detect substantial forest loss Baseline and endline forest coverages

across the 136 villages:

PES seems to result in a change in distribution of forest losses/gains

Note: -0.5 = 5%, -.1 = 10%, etc.

Measured impact of reduced deforestation consistent but statistically insignificant

Example: Column 1: the rate of deforestation is 2.59 percentage points lower in treatment villages relative to control (1.51 pp per annum), a reduction of 27% relative to control villages

Note: Column 1 and 2 are weighted regressions using the % of land w/o cloud coverage

• Compare with ~34% take-up rate; would imply a treatment on the treated reduction of 27/0.34 = 79%

Can we get more precise estimates using PFO level forest change?

Mapping an individual’s land

PFO level analysis

Key Findings on Program Impact Very high rates of forest loss in absence of the

program Measures of self-reported reduction in tree cutting as

a result of PES consistent with CSWCT monitoring data

Measures of self-reported land use point to PES increasing private use of forests

Satellite images detect a village level effect size for PES reduced deforestation that is consistent with PES take-up rates and self-reported changes in behavior. Point estimate of effect not significant, however, but low statistical power

Annual cost of payment for all 136 villages: ~ USD 70k for 20,000 ha of forests

Thank you! (also from the field)