Minimizing Ecological Damage from Road
Improvement in Tropical Forests: The Case of
Myanmar
Susmita Dasgupta
David Wheeler
DECEE, The World Bank
2016
Overview




Motivation
Research Objective
A Composite Biodiversity Indicator
Road Improvement and Deforestation
 Descriptives
 Model Specification
 Data
 Estimation
 Implications for Forest Clearing
 Road Upgrading and Ecological Risk
 Identifying Critical Road Links
 Summary
Conservation Management of
Tropical Forests
 Traditional Measure: Protected Area Strategy
 Demarcation and protection of large areas deemed
critical for biodiversity conservation.
 Restrict infrastructure development in “Protected Areas”
that may increase the profitability of forest clearing.
 Problems
 Conflicts arise when forested areas have significant
agricultural potential.
 Failure of attempts to protect large areas with strong
agricultural potential.
 Protected areas may not coincide with the areas of
highest ecological value.
Underlying Issues
 Lack of understanding: Valuable economic and ecological
resources have
distributions.
non-uniform
and
overlapping
spatial
 Lack of understanding: Both development and conservation
may be hindered if a policy regime treats large areas as either
completely protected or completely open for development.
 Scarcity of information: Assessments of tradeoffs between
development and conservation objectives is often hindered by
limited information on critical ecosystem and biodiversity.
Research Objective
 Developed high-resolution map of potential ecological loss
from range maps and threat status data for 25,000+ species
provided by the IUCN and BirdLife International.
 Constructed a spatial panel dataset on tropical forest loss
from newly-available high-resolution measures of forest
cover loss since 2000.
 Developed and applied a spatial econometric model that
links road upgrading to forest clearing and biodiversity loss
in the moist tropical forests of Bolivia, Cameroon and
Myanmar.
 Provided ecological risk ratings for individual road corridors
to inform environmentally-sensitive infrastructure investment
programs.
Biodiversity Indicator
Composite Biodiversity Indicator
Composite Biodiversity
Indicator
Biodiversity Vulnerability
Biome Vulnerability
Incorporating Biodiversity
 Species Density
(Count of resident animal species for every 250m cell in the
study region from range maps provided by the IUCN and
BirdLife International)
 Species Vulnerability
 Geographic Vulnerability
(Endemicity: Proportion of each species’ range that lies within
each cell of our study region)
 Density of endangered and critically endangered species
(Count of endangered and critically endangered species in each
cell)
 Extinction Risk
(Probability of extinction over next 50,100, 500 years using
Mooers et al. 2008 and Isaac et al. 2007)
Biome Vulnerability
 Use of 825 terrestrial ecoregions* of the world provided by
the WWF.
 Computation of percentage of total moist forest area in a
country accounted by each ecoregion.
 Use of the inverse of the percentage of total moist forest
area in a country accounted by each ecoregion as the
vulnerability index to assign high values to cells in smaller
ecoregions, where clearing single cells may pose more
significant threat to biome integrity.
* Ecoregion: a large unit of land or water containing a
geographically distinct assemblage of species, natural
communities and environmental conditions.
Myanmar: Moist Forest Ecoregions
Myanmar: Biodiversity Indicator
Correlation Coefficients
Species
Count
Endangered
Species
Count
Endemicity
Isaac
Extinction
Risk
IUCN
50-Yr.
Extinction
Risk
IUCN 100Yr.
Extinction
Risk
Endangered Species Count
0.42
Endemicity
0.97
0.41
Isaac Extinction Risk
0.98
0.59
0.94
IUCN 50-Yr. Extinction Risk
0.64
0.92
0.61
0.78
IUCN 100-Yr. Extinction Risk
0.62
0.96
0.59
0.77
0.99
IUCN 500-Yr. Extinction Risk
0.71
0.88
0.66
0.83
0.91
0.95
Biome (Ecoregion) Risk
-0.26
-0.26
-0.26
-0.31
-0.39
-0.34
 Correlations calculated for 250m cells.
 Biome indicator is a distinct outlier
IUCN 500Yr.
Extinction
Risk
-0.27
Myanmar: Composite Biodiversity
Indicator
30-44
44-51
51-58
58-66
Myanmar
66-74
74-80
80-84
84-88
88-93
93-100
Road Network and Forest Clearing
Descriptives
Myanmar: Road Networks
Myanmar has 2,101 primary
road and 4,879 secondary
road links in 2014.
Source: Delorme, Inc.
Secondary road
Primary road (connector)
Primary road (highway)
Forest Clearing
 Data Source: High-resolution (30 m) estimates of global
forest clearing from Hansen, et al. 2013 and Hansen 30 m
estimates of tree cover in 2000.
 Annual files were created with cleared pixels = 1 in the year
when most clearing occurred. Uncleared pixels are
assigned the value 0.
 Hansen estimates are used to compute forest clearing rates
in 250 m (approximately 0.0025 decimal degree in length)
cells for expositional convenience, yielding cell areas of
7.75 hectares.
 Annual cumulative percent forest cleared is computed for
each cell through 2014.
 Focus is on areas defined as moist tropical forest
ecoregions by WWF.
Myanmar: Forest Clearing and
Road Networks, 2000-2014
% Cleared,
2000 - 2014
Myanmar: Change in Share of Forest Cleared
(2000-2014) vs. Distance from Road Segments
Mean forest clearing within half a km of the road increased
by 15 percent point from 2000 to 2014.
With increase in the distance from the road, there is a steep reduction in
mean % forest cleared for the first 5 km, a declining slope through 10 km,
and approximate flattening near zero beyond 10 km.
Road Network and Forest Clearing
Conceptual Framework
Road Construction /
Upgrading
Potential Gain
Increase in Trade
Potential Loss
Increase in
Agricultural Income
Loss of Biodiversity
Economics of Road Improvement
 Road improvement is a problem of competitive selection
among corridors that traverse the same region.
 Corridors differ in length, construction cost conditions,
biodiversity value and potential agricultural income.
 The optimum corridor choice reflects maximization of a social
utility function that values both income and biodiversity,
subject to

Fixed budget constraint - Feasible road quality improvement along each
corridor (reflecting the budget constraint)

Expected income growth from agriculture in the corridor

Expected income growth from trade between areas connected by the
corridor

Corridor specific impacts of road quality improvement on biodiversity loss
Economics of Forest Clearing
 The proprietor/ occupant of a forested area considers the
relative profitability of maintaining/ clearing the area.
 In each period, the present-value profitability of sustainably
harvested forest products is compared with the clear-cut
value of forest products and the cleared land’s present value
profitability in its best use (e.g., plantation, pasture,
smallholder agriculture, settlement).
 Generally there is a cost associated with forest clearing.
 Determinants of forest clearing highlighted in prior research:
Cost of Land, Expected Revenue from Production on
Cleared Land, Distance from Markets, Quality of Transport
Infrastructure, Cost of Capital, Agricultural Input Price,
Topography, Soil Quality and Forest Protection Measures.
Model Building Blocks

Trade between areas is encouraged by better road quality.

The proprietor of each road-front parcel confronts

Forest clearing cost, which is constant or increasing with the
distance from the road, depends on elevation and slope of the
terrain.

For agricultural production, commodity transportation costs that
increase with the distance from the road.

Improvement in road quality lowers transport costs and increases
potential profitability of agriculture along the corridor.

Expected profit for each roadside land parcel falls with increase in
road distance.

Expected profitability of agriculture increases with the size of the
cleared parcel of land.

Given a fixed road quality budget, road quality declines with road
length.
Variables of Interest
 Distance from the road
 Transport cost to the nearest market center
 Elevation of land
 Terrain slope
 Agricultural opportunity value of the land
 Legal protection status
 Road quality
Model Specification
Road Network and Forest Clearing
Econometric Estimation
Data
Variable
Source
Forest clearing
Hansen pixels cleared: per 250 m cell, the
percent of 30 m Hansen pixels cleared.
Distance from road segment
Distance from the centroid of each cell to the
nearest road segment, calculated in ArcGIS 10.
Distance traveled to the nearest
urban center via primary and
secondary road segments
Calculated in ArcGIS 10 from Delorme digital
road maps.
Elevation
Average elevation for a cell, calculated from the
CGIAR-SRTM dataset (3 seconds resolution)
Terrain slope
Standard deviation of pixel-level elevation
measures within a cell.
Agricultural opportunity value
Mean value for a cell, calculated from the highresolution global grid developed by Deveny, et
al. (2009).
Legal protection status
1 if the cell is in a protected area identified by the
World Database on Protected Areas (WDPA); 0
otherwise.
TRAVEL TIME MINIMIZING DISTANCE
NEAREST ROAD
Travel Time Minimizing Distance
PLOT
OF
FOREST
Quality of Road affects Travel Time
URBAN CENTER
POPULATION:
50,000
OR MORE
Issues Related to Estimation
 Simultaneity between forest clearing and to the urban
center:
 Problem is addressed via instrumental variables:
geodetic distance from each road increment midpoint
to the nearest urban center.
 Spatial autocorrelation:
 Problem is addressed with a spatial econometric
estimator using the inverse-distance specification of
the spatial weights matrix.
Myanmar: Estimation
 Sample size: 5.8 million cells.
1. Randomly-drawn sample of 10,000 cells: IV, IV with
bootstrapped standard errors calculated from moredispersed subsamples of 5,000 observations, spatial
econometric estimation using IV.
2. Randomly-drawn sample of 20,000 cells: IV and IV with
bootstrapped standard errors.
3. Full sample: IV and IV with bootstrapped standard errors.
Myanmar: Regression Results
Myanmar: Median Parameter Estimates
Variable
Distance from road
Median Parameter
Estimate
-0.181
Distance to nearest urban center (DU)
-0.336
DU x Primary road share
0.060
Slope (Std. dev. of elevation)
-0.467
Slope x Elevation
0.044
Protected area
-0.780
Agricultural opportunity value
0.075
Constant
1.324
Myanmar: Findings for DistanceRelated Variables
1. Forest clearing is inversely related to distance from road.
2. The same pattern holds for distance to the nearest urban
center.
3. The median elasticity for travel solely on secondary roads
is -0.336.
4. The median elasticity for travel solely on primary roads is
-0.321 (=-0.530+0.209).
In addition, Significant roles for Terrain Slope (-), Agricultural
Opportunity Value (+) and Protected Area Status (-).
Road Upgrading and Ecological Risks
Myanmar: All Secondary Roads
Improvement Impact
Area Cleared 80-100%
Moist Forest Ecoregion
Irrawaddy freshwater swamp forests
(thousand hectares)
After
2014
upgrading Change
Impact
ratio
1.04
2.98
1.95
2.9
Irrawaddy moist deciduous forests
55.13
91.78
36.65
1.7
Myanmar coastal rain forests
30.16
48.64
18.48
1.6
Chin Hills-Arakan Yoma montane forests
15.28
23.92
8.63
1.6
108.61
166.42
57.81
1.5
22.43
31.51
9.08
1.4
30.85
41.40
10.55
1.3
Northern Triangle subtropical forests
26.08
34.81
8.73
1.3
Kayah-Karen montane rain forests
Lower Gangetic Plains moist deciduous
forests
38.97
49.90
10.94
1.3
0.00
0.00
0.00
.
Northern Indochina subtropical forests
Tenasserim-South Thailand semievergreen rain forests
Mizoram-Manipur-Kachin rain forests
Myanmar: Predicted Impacts by
Moist Forest Ecoregions
 The greatest expansion of maximum (80-100%) cleared area is
expected to occur in the:
 Northern Indochina subtropical forests (eastern Myanmar), from 108,610 to
166,420 hectares;
 Irrawaddy moist deciduous forests (central Myanmar), from 55,130 to 91,780
hectares; and
 Myanmar coastal rain forests (western and southern Myanmar), from 30,160 to
48,640 hectares.
 Expansions in the range 8,000 - 11,000 hectares will occur in the:
 Chin Hills-Arakan Yoma montane forests (western Myanmar);
 Tenasserim-South Thailand semi-evergreen rain forests (southern Myanmar);
 Mizoram-Manipur-Kachin rain forests (western and northern Myanmar);
 Northern Triangle subtropical forests (northern Myanmar); and
 Kayah-Karen montane rain forests (east-central Myanmar).
Indicator: Ecological Risk of Road
Upgrading
Expected biodiversity loss for a cell is estimated as
(Biodiversity Indicator Value of the cell) x (Change in
the cleared forest percentage of the cell induced by
road upgrading).
Myanmar: Ecological Risk of Road
Upgrading
 Large losses (60-100) are
expected in the far north, a
band from the north to east
and scattered areas in the
west and south.
 Intermediate expected losses
(40-60) are expected in large
areas.
 Lowest expected losses are in
corridors flanking roads that
already have primary status.
Identification of Critical Road Links
 Road links are graded by expected biodiversity losses when
upgrading occurs (using the forest clearing impacts of
specific road links).
 Mean expected losses in corridors extending 10 km on
either side of the Delrome-identified secondary road links in
moist forest areas is computed.
 For ease of comparison, estimates are normalized to the
range 0 to 100.
 The highest four deciles were color coded for visual
comparison : purple (90-100); red (80-90); orange (70-80);
yellow (60-70)] along with primary road links where
upgrading does not occur (blue).
Myanmar: Ecologically High-Risk
Road Corridors
 Large
clusters
of
most
critical(purple) corridors are
visible in the east, with smaller
clusters linked to next-priority
(red) corridors in the north,
west and south.
 Clusters in the lower priority
categories
(orange
and
yellow) are widely scattered.
Summary
 This paper develops and applies a spatial econometric model
that links road upgrading to forest clearing and biodiversity loss
in moist tropical forests of Bolivia, Cameroon and Myanmar.
 Forest clearing is highly responsive to the distance to the
nearest urban market which comprises of the distance of the
parcel of land (cell) to the closest point on the nearest road and
the transport-minimizing route to the nearest urban market.
 Expected biodiversity loss from upgrading secondary roads to
primary status was estimated using forest clearing response
elasticities and a composite biodiversity indicator.
 The research provides ecological risk ratings for individual road
corridors for environmentally-sensitive infrastructure investment
programs.
Environmentally-Sensitive
Infrastructure Planning
Road upgrading will inevitably accompany rural development
programs.
Identification of ecologically-vulnerable areas
corridors can provide two valuable information:
and
road
 With limited budgets, it can help steer road upgrading
programs toward corridors where expected biodiversity
losses will be minimized.
 It can inform adoption of appropriate protection measures in
vulnerable road corridors and adjacent areas.
Relevance for Other Countries
 Use of global database that includes a larger set of
biodiversity measures.
 Use of globally-available road quality estimates.
 Use of high-resolution satellite data on forest cover change.
 Exclusive use of globally-available databases
ensures applicability of this empirical work to all
moist tropical forest countries.
Acknowledgement
 Knowledge for Change Trust Fund for funding the
research.
 Ms. Siobhan Murray for GIS support.
 Ms. Polly Means for help with graphics.
 Mr. Pritthijit (Raja) Kundu for help with the
Powerpoint presentation.