Published on Development Impact

Who benefits from deforestation for infrastructure in India? Guest post by Sayantan (Sunny) Mitra

This page in:
MitraJMP

The potential trade-off between preserving natural resources and economic development depends in part on the balance between improved access to economic opportunities and the negative effects on resource-dependent populations. Resource-dependent populations are often the most vocal opponents of such projects, including in the important case of infrastructure expansion into tropical forests. In India, infrastructure expansion accounts for more than 80% of annual tree cover loss.  

In my job market paper, I estimate the causal effects of replacing forests with infrastructure on the local population as well as on different sub-populations based on income, education and historical forest dependence. Common infrastructure projects in this case include highways, railway lines and electricity transmission lines. These estimates help inform policy debates on the likely local consequences of such infrastructure projects and may also help rationalize local political support for these projects.

Analyzing local impacts of the Indian Government’s approvals of deforestation

In India, government approval is necessary to clear forest land permanently for a non-forest use. This implies that activities such as logging, firewood collection, and shifting cultivation do not require approval, but clearing forest permanently for infrastructure expansion requires approval. These approvals involve review by multiple officials at multiple levels as well as committees involving external advisers. Almost 40,000 instances have been approved since 1980, with an approval rate of 65%, and 98% of these have been for infrastructure projects. The pace of these approvals has grown recently with 80% of all approvals being after 2000, amounting to over 2 million hectares of deforestation.

I use the staggered timing of these approvals between 2014 and 2022 as the source of identifying variation in an event study framework with staggered adoption and heterogeneous treatment effects following Borusyak et al. (2024). Following this framework, I separate the testing of parallel trends and no anticipation from the treatment effect estimation to ensure that my treatment effects are point-identified, especially given that there are no never-treated units in this case. I also restrict the timeframe for estimation of treatment effects to 3 years after approval since there is an insufficient sample of never-treated units beyond that horizon to estimate an unbiased treatment effect.

In order to explore the local distributional impacts of these approvals on local households, I use data from the Consumer Pyramids Household Survey (CPHS) which is a household-level panel survey that collects data every 4 months on demographics, incomes, consumption and assets from more than 170,000 households. I restrict the primary sample to households located within 10 kilometers of projects that apply to clear forests for infrastructure, dropping the few households which are in the vicinity of multiple projects.

5 facts about how clearing forests for infrastructure affects different local households

1.        Forest-dependent tribal households experience a decline in employment in retail trade by 88% (14 p.p. lower than baseline of 16%) accompanied by a 32% increased engagement in subsistence agriculture (12 p.p. higher than the baseline of 37%).  This in line with them facing a negative shock to production inputs and consumption choices due to the loss of the forest.

Figure 1: Proportion of Workforce Engaged in Retail for Tribal vs Non-Tribal Households

 

MitraJMP

 

 

 

 

 

 

 

 

 

 

 

 

2.       Contrary to expectations, there is no change in household incomes and consumption on average, but there is considerable heterogeneity. Incomes rise by 34% for poorer households (Rs. 5537 / USD 66 higher than baseline of Rs. 16,284 / USD 194) but there are no significant effects for richer households. Poorer households are identified as those with below-median annual consumption level during the year before deforestation approval, and richer households as those with above-median annual consumption level during the same time period.

3.       Household size increases by 12% for poorer households (0.5 persons higher than a baseline of 3.63 persons) as 45% of migrants return (proportion of out-migrants is 8 p.p. lower than the baseline of 18%), but there are no significant effects for richer households. This is possibly driven by migrants being attracted back by the higher local incomes among poorer local households.

Figure 2: Monthly Incomes for Poorer vs Richer Households

 

MitraFig2

 

 

 

 

 

 

 

 

 

 

 

 

4.       Occupational structure changes as non-tribal households across the income distribution move into retail trade by a 39% increase in engagement in retail (9 p.p. higher than a baseline of 24%). Besides this occupational change, there is also an increase in business incomes for non-tribal households across the income distribution by 136% (Rs. 4885 / USD 58 higher than the baseline of Rs. 3592 / USD 43).

5.       Employment in skill-intensive services such as finance and education declines by 63% for educated households (32 p.p. lower than a baseline of 51%). This shift is also reflected in the decline in wages for educated households by 40% (Rs. 8978 / USD 107 higher than the baseline of Rs. 22,445 / USD 267).

The last two findings are consistent with the predictions of a Hecksher-Ohlin trade model with infrastructure-induced reduction in trade costs. For such a model, I assume high and low skilled workers to be the factors of production in a Local economy and a skill-abundant External economy. I also assume two tradable sectors in these economies: retail and services, with services being relatively skill-intensive. This set-up allows me to explain the observed occupational shift from services into retail and the observed reduction in skill premia for educated households.

When Do The Benefits Outweigh the Environmental Costs?

With no observed change in income or consumption for the average local household, there is no local average economic benefit from clearing forests for infrastructure. This raises questions regarding the justification for such projects given their large economic and environmental costs, and who these projects are intended for. 

I address the latter part of that question by identifying the only sub-population that benefits from such projects: local households with below-median baseline consumption. These poorer households experience a gain in income of Rs. 39,760 / USD 468 per household over the 3 years after the approval, amounting to a total benefit of around $8 million for poorer households within a 10 kilometer radius from the project.

Using the official US EPA social cost of carbon (SCC) of $190/ton reduces the net benefit for poorer households to around $6 million per project, while the distributionally weighted SCC of $1300/ton (more appropriate for an Indian context) from Prest et al. (2024)  reduces it further to a net cost of almost $1 million per project. I find that benefits (to local poorer households) outweigh the costs for any SCC below $1200/ton. Using the US EPA SCC, I also find that these benefits outweigh the costs only for projects smaller than 95 hectares in locations with more than 25 households per square kilometer.  It is important to caveat these estimates with the fact that I do not consider the broader impacts of these projects which may be dispersed spatially and temporally across larger areas over longer time periods.

In summary, while clearing forests for infrastructure in India has large environmental costs, I demonstrate that it also has progressive distributional benefits especially for smaller projects in densely populated areas. However, there are also sections of the local population that are significantly hurt as well: forest-dependent tribal households who undergo disruptive significant lifestyle changes, and educated households who are competed out of skilled jobs due to increased trade.

Sayantan (Sunny) Mitra is a PhD student at UC Berkeley [Twitter: @SunnyMitter]


Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000