Forest road
Forest road
Infrastructure expansion into forests substantially degrades biodiversity. Yet, institutions that centre the voices of forest-dependent communities during project planning can help mitigate species loss.
Although crucial for economic growth, infrastructure expansion narrows the frontier between human activity and fragile ecosystems. In 2015, over half of global infrastructure spending was earmarked for projects in the tropics, a region home to two-thirds of Earth’s biodiversity (FAO and UNEP 2020). The ecological threat from infrastructure encroachment is exacerbated by the displacement and exclusion of millions of tribal communities, who have supported biodiversity for centuries.
While much of economic research seeks to understand the underlying environment-economy trade-offs in developing countries (Asher et al. 2020, Garg and Shenoy 2021, Baehr et al. 2021, Burgess and Jack 2023), biodiversity is often overlooked due to its elusive measurement (Frank 2025, Frank and Schlenker 2016). There has also been less focus on the role of institutions in mitigating the trade-off (Gulzaar 2024), highlighting an important area for research.
In my new paper (Madhok 2025), I aim to fill this gap by: 1) quantifying the extent to which infrastructure drives species loss, and 2) investigating the role of inclusive forest governance in mitigating the trade-off. This study is set in India, a global biodiversity hotspot, home to 150 million forest-dependent tribal people, yet where 20% of annual deforestation is due to clearing forests for infrastructure development projects (Madhok 2025).
Infrastructure and biodiversity conservation efforts in India
Figure 1: Infrastructure encroachments and eBird activity
Infrastructure encroachments and eBird activity
Note: Panel A is a map of 2015 forest cover (Townshend et al. 2017). Pixels are shaded by percent forest cover. Panel B maps the number of infrastructure projects approved for construction between 2015-2020. Panel C shows GPS coordinates of all birdwatching trips recorded through eBird during the study period.
India’s Forest Conservation Act of 1980 protects forests from “conversion to non-forest uses”. When non-forest sites are unfeasible, firms can site projects in forests pending an ecological assessment. While this process historically ignored tribal land claims, the landmark Forest Rights Act (FRA) granted tribal forest rights in 2006, including the right to informed consent. A deforestation permit is issued to developers after passing review and documenting FRA compliance.
To measure infrastructure encroachment, I digitised all deforestation permits granted between 2015 and 2020 and aggregated by cumulative district-monthly forest area diverted for development. Projects divert forests in nearly every district, especially in North and West India (Figure 1A, 1B).
Using birdwatching notebooks to create ecosystem data
To measure biodiversity, I obtained one million geocoded birdwatching notebooks from eBird, the world’s most popular birding app (Sullivan et al. 2009). Ecologists cite birds as the best proxy species for biodiversity (Fraixedas et al. 2020), and have found that eBird’s crowd-sourced data correlates strongly with structured bird censuses (Callaghan et al. 2018). I count the number of species observed on each user’s birdwatching trips, generating a spatially explicit biodiversity dataset spanning 95% of districts (Figure 1C).
My empirical strategy is based on the fact that development projects fragment district forests, and eBird users venture to these districts to record birds. To estimate the infrastructure-biodiversity trade-off, I use a two-way fixed effects design that compares species diversity within user’s trips as they travel across different districts over six years. This approach accounts for some districts being more popular than others, seasonal bird migrations, and heterogeneous abilities across birdwatchers. This analysis yields three key insights.
The impacts of infrastructure development on biodiversity
1. Infrastructure development triggers substantial species loss.
I find that an additional 10 km² of district forest diverted for development leads eBird users to observe one fewer species. To put this in perspective, this implies that infrastructure expansion accounted for nearly 20% of the total observed species decline during the study period. Decomposing by project category, resettlement camps, transport, and irrigation projects threaten species diversity the most (Figure 2B). Mining has a surprisingly small effect, mainly due to selection bias from few eBird users travelling to remote mining districts.
I also classify each species by habitat type in order to investigate which species are most threatened by infrastructure. Perhaps unsurprisingly, forest birds are the most threatened by infrastructure encroachment into forests, whereas there is no impact on wetland birds.
Figure 2: The infrastructure-biodiversity trade-off in India’s forests
The infrastructure-biodiversity trade-off in India’s forests
The infrastructure-biodiversity trade-off in India’s forests
Note: The outcome is mean species richness across users’ trips in a district-month. Panel A shows coefficients on cumulative area of infrastructure encroachments in a district-month. Panel B is a single regression with deforestation decomposed into project categories. All regressions include user-year, district, and state-month fixed effects and control for temperature, rainfall, traveling trips, log nightlights, log duration, log distance, log experience, log group size, and log spatial coverage. Standard errors clustered by biome.
2. Species diversity is more resilient in intact forests.
Next, I investigate impacts by baseline ecosystem quality, which has implications for whether conservation should target intact or fragmented landscapes. I find that species are more resilient to infrastructure development in intact forests. The adverse impact on species diversity in Figure 2 is halved in districts with one standard deviation higher initial forest cover. Since species loss is largest when baseline ecosystem quality is low, these results provide support for stronger protections for degraded landscapes.
3. Species diversity does not rebound in the medium-run.
Lastly, I estimate a cumulative lag model to trace the evolution of species diversity in the medium-run. Species decline is triggered soon after habitat loss and then persists (Figure 3). The net impact up to two years later is similar to the instantaneous effect, suggesting that once forests are diverted**,** species diversity does not recover in the medium-run.
Figure 3: Species impacts in the medium-run
Species impacts in the medium-run.
Species impacts in the medium-run.
Note: “Baseline” repeats the main result. “Sum L0- L1” adds the first lag of forest diversion to the main specification and reports the sum of coefficients on the first lag and baseline effect. “Sum L0-L3” sums up to the third lag, and so on. Controls and fixed effects are the same as Figure 1.
What is the role of local institutions in reducing species loss?
Having established that infrastructure degrades biodiversity, next I investigate whether participatory institutions that emphasise inclusive decision-making can help mitigate the infrastructure-biodiversity trade-off. I draw on the fact that millions of tribal people in India, who have stewarded biodiversity for millennia, are excluded from development planning.
Relying on data from Banerjee and Iyer (2005), I indicate whether district institutions favour elites (extractive) or are more inclusive of the masses. If tribal groups can better protect their livelihoods in inclusive districts, which hinges on protecting forests, then we would expect better conservation outcomes there.
Using the same two-way fixed effects specification with an interaction for institutional type, I find that infrastructure development is over 70% less harmful to biodiversity in inclusive districts compared to extractive ones.
How do inclusive institutions promote conservation?
I explore two channels through which inclusive institutions promote conservation: informed consent and greater environmental scrutiny on developers. The digitised project permits provide data on both. Although the FRA (2006) mandates informed consent between developers and the concerned tribal council, I find that less than half of projects in my sample follow this mandate, providing the necessary variation to evaluate mechanisms.
Matching project permits with the inclusive-extractive dummies, I compare outcomes across projects in the two types of districts. Developers in inclusive districts are 8 percentage points (pp.) more likely to obtain informed consent from tribes compared to extractive districts (Figure 3). Forest officers in these districts are also 7 pp. more likely to commission additional cost-benefit reports before permits are awarded. Lastly, projects are 1 pp. less likely to be sited near protected areas. These mechanisms contribute to lower biodiversity loss in inclusive districts and highlight the importance of engaging forest-dependent communities, along with more stringent checks-and-balances during project approval, for protecting biodiversity.
Figure 4: Mechanisms mitigating the infrastructure-biodiversity trade-off
Mechanisms mitigating the infrastructure-biodiversity trade-off
Mechanisms mitigating the infrastructure-biodiversity trade-off
Note: Data are at the project level. Bars are coefficients from Equation (10). Sample is restricted to 163 districts in Banerjee and Iyer (2005). Informed consent indicates whether the FRA was followed. Cost-benefit analyses indicates whether one was done during project review. Protected Area equals one if the project is sited in or near one. All specifications control for: project size, tribal population share, baseline forest cover, latitude, altitude, coastal dummy, and district area. Grey bars are 95% confidence intervals.
Policy implications for promoting development and conservation
The findings of this paper are especially relevant as emerging economies beginning to prioritise the types of infrastructure studied here. In places where institutions favour the economically advantaged, infrastructure development is associated with more biodiversity loss. This highlights the need for people-centred conservation policy. In a previous VoxDev article, Gulzaar et al. (2025) note that over a hundred countries have implemented inclusive institutions with the aim of bolstering the voice of marginalised communities.
India has made strides with the FRA (2006), which promises forest rights to tribal communities and their inclusion in development decisions. Yet, almost two decades later, half of all forest rights claims remain legally unrecognised and face other forms of weak enforcement (Ministry of Tribal Affairs 2022). I find that upholding the FRA helps neutralise the infrastructure-biodiversity trade-off. In sum, inclusive institutions and procedural justice are crucial for meeting the dual objectives of development and biodiversity conservation.
References
Asher, S, T Garg, and P Novosad (2020), “The ecological impact of transportation infrastructure,” The Economic Journal, 130(629): 1173–1199.
Baehr, C, A BenYishay, and B Parks (2021), “Linking local infrastructure development and deforestation: Evidence from satellite and administrative data,” Journal of the Association of Environmental and Resource Economists, 8(2): 375–409.
Banerjee, A, and L Iyer (2005), “History, institutions, and economic performance: The legacy of colonial land tenure systems in India,” American Economic Review, 95(4): 1190–1213.
Burgess, R, and K Jack (2023), “Environmental economics and policy in low- and middle-income countries,” VoxDevTalk Podcast.
Callaghan, C T, J M Martin, R E Major, and R T Kingsford (2018), “Avian monitoring—comparing structured and unstructured citizen science,” Wildlife Research, 45(2): 176–184.
FAO and UNEP (2020), "State of the world’s forests: Forests, biodiversity and people," FAO.
Fraixedas, S, A Linden, M Piha, M Cabeza, R Gregory, and A Lehikoinen (2020), “A state-of-the-art review on birds as indicators of biodiversity: Advances, challenges, and future directions,” Ecological Indicators, 118: 106728.
Frank, E G, and W Schlenker (2016), “Balancing economic and ecological goals,” Science, 353(6300): 651–652.
Frank, E (2025), “The economics of ecosystems: How nature and economies interact,” VoxDevTalk Podcast.
Garg, T, and A Shenoy (2021), “The ecological impact of place-based economic policies,” American Journal of Agricultural Economics.
Gulzar, S, A Lal, and B Pasquale (2025), “Can political representation boost forest conservation?” VoxDev Article.
Gulzar, S, A Lal, and B Pasquale (2024), “Representation and forest conservation: Evidence from India’s scheduled areas,” American Political Science Review, 118(2): 764–783.
Madhok, R (2025), “Infrastructure, institutions, and the conservation of biodiversity in India,” Journal of the Association of Environmental and Resource Economists, Forthcoming.
Ministry of Tribal Affairs (2022), "Monthly update on status of implementation of the Scheduled Tribes and Other Traditional Forest Dwellers (Recognition of Forest Rights) Act, 2006."
Sullivan, B L, C L Wood, M J Iliff, R E Bonney, D Fink, and S Kelling (2009), “eBird: A citizen-based bird observation network in the biological sciences,” Biological Conservation, 142(10): 2282–2292.
Townshend, J, M Hansen, M Carroll, C DiMiceli, R Sohlberg, and C Huang (2017), "Annual global automated MODIS vegetation continuous fields (MOD44B) at 250 m spatial resolution for data years beginning day 65, 2000-2014, collection 5 percent tree cover, version 6," University of Maryland, College Park.
India biodiversity conservation ecosystems