Published on Development Impact

The Hidden Local Costs of Deforestation in the Tropics. Guest post by Teevrat Garg

This page in:

This is the seventh post in our series of blogs by graduate students on the job market this year.

The debate over deforestation has traditionally weighed the tradeoffs between local economic benefits and the broader ecological footprint measured in carbon emissions (Alix-Garcia et al., 2013; Foster & Rosenzweig, 2003). Consequently, this framing has led to the creation of several multi-billion dollar programs under the umbrella of the United Nations known as REDD+ or Reduced Emissions from Deforestation and Degradation. The idea is simple: in exchange for forgoing the economic benefits of logging and forest land clearing, countries that preserve forests (particularly poorer countries) receive payments from richer countries that benefit from the reduced carbon emissions associated with deforestation.

The underlying principle is that effects of deforestation related emissions are global in nature.

To the contrary, my research finds that the effects of deforestation are substantially larger at the local level due to health externalities (particularly from increased malarial incidence).  I find that local health costs of deforestation in Indonesia are an order of magnitude higher than the global carbon externalities. Thus local institutions, as opposed to external governments, may have the strongest incentives for forest preservation. Furthermore, given the productivity, morbidity, mortality and fertility costs associated with malaria (Lucas 2013, Lucas 2010), there may be a double dividend from environmental conservation currently being ignored in policy formulation.

Indonesia offers an ideal case study for the significance of local externalities for two reasons. Between 2001 and 2008, Indonesia lost 4.8 million hectares of forest cover—or roughly the size of Vermont and New Hampshire combined. And the World Health Organization estimates that 44% of Indonesia’s population—roughly 130 million people—is at risk of malaria.

The Ecology of Deforestation and Malaria

In theory, deforestation can affect malaria through at least three possible channels (Pattanayak & Pfaff, 2009):

  1. Post-deforestation land-use change in the form of urbanization, construction and agricultural production has been associated with increased incidence of malaria.
  2. Deforestation is accompanied by migration, and migrants can act as latent hosts for infectious diseases.
  3. Deforestation alters the disease ecology of tropical malaria by reducing biodiversity, particularly of species that feed on anopheline larvae (i.e., the mosquitos that transmit malaria), and increasing puddle formation, which triggers greater anopheline growth.
The first two mechanisms are socio-economic while the third is ecological. My statistical analysis demonstrates that there is a pure ecological mechanism that connects deforestation to malaria. The two socio-economic mechanisms do not appear to exert any significant effect. This distinction is important since the mechanism directly impacts the policy implications. For instance, if the underlying mechanism were through post-deforestation land-use change, then the policy prescription would be to internalize the health costs of agricultural expansion or urbanization as opposed to that of the act of deforestation.

A Panel Data Approach using Satellite Data

I combine spatially disaggregated district-level satellite data on deforestation (Burgess et al., 2012) with administrative village-level data to estimate the causal effect of deforestation on the incidence of malaria. It’s important to use satellite data since 60-80% of all deforestation in Indonesia is illegal, so official records are likely unreliable. The malaria data is obtained from the village-census administrative data set (Podes) that is collected by Budan Pusat Statistik (BPS) every 2-3 years. It covers all of Indonesia’s 66,000 villages. Village heads report the disease data in response to three questions – was there an outbreak? If so, how many individuals were infected, and how many died? In any given year, on average, 19% of villages report having an outbreak with roughly 14-15 individuals infected per outbreak. While measurement error is often a concern in recall data, the implications for this study are negligible since measurement error in the dependent variable should only increase the standard errors on coefficients providing at best, an underestimate of the statistical significance of the key result.

The primary specification includes district fixed effects and island-year fixed effects to control for district-level time-invariant heterogeneity and allow for flexible time-varying heterogeneity in each of the islands. The rich nature of the administrative village-level data set allows me to control for a number of village-level confounding variables such as presence of and access to medical facilities, poverty indicators, distance from cities, sanitation facilities, dwelling conditions, population demographics, physical and geographic characteristics, etc.

A battery of robustness and placebo checks corroborate the core panel data results. Therefore, the strong effect I find linking deforestation to increased malaria outbreaks does not appear due to omitted or confounding variables that could otherwise be correlated with both deforestation and health.

Deforestation Dramatically Increases Malaria

Three key results stand out from the analysis:
  1. Deforestation substantially increases the incidence of malaria. In particular, the baseline estimate is that the within-sample average district-level deforestation (1000 hectares) increases the probability of malarial outbreak in each village in that district by 0.364 percentage points or roughly 2%. I subsequently estimate that deforestation in Indonesia between 2001 and 2008 led to almost one million additional malaria infections.
  2. The effect is ten times larger in forests intended for biodiversity conservation than in forests designated for the purposes of logging and land-use: conversion to agriculture or residential construction. This finding supports the inference that the underlying mechanism is an ecological response to deforestation.
  3. The effect is specific to malaria, as deforestation has no discernible effect on diseases such as diarrhea, measles, and respiratory infections—all of which have markedly different disease ecologies from that of malaria.
Implications for Local Conservation and Payments for Ecosystem Services

The implications of this research can affect ongoing and future policy instruments being used in Indonesia as well as countries with similar forest infrastructure such as those in the Amazon and Congo River basins.

First, Indonesia recently launched the Rainforest Standard, a first-of-its-kind payments for ecosystem services program designed to reduce deforestation. This research can help policymakers better internalize the local health benefits of avoided deforestation. We are likely underestimating the costs of deforestation by focusing exclusively on the associated carbon footprint.

Second, this research begs the obvious question – why do local communities and institutions not internalize these local costs of deforestation? Consequently, and in line with Burgess et al. (2012), this paper motivates further research on how local institutions can capture these sizable local benefits from avoided deforestation.

Finally, this research helps recast the conventional wisdom of the environment-economy tradeoff by accounting for the morbidity and productivity externalities associated with malarial infections. It appears that there are double dividends from environmental conservation – health and economic benefits – being ignored in present cost-benefit analyses.

Teevrat Garg is a PhD. Candidate in Environmental and Development Economics at the Charles H. Dyson School of Applied Economics and Management at Cornell University. 

Join the Conversation

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