| Roads, land use, and deforestation:
A spatial model applied to Belize
Rural roads promote economic development, but they also facilitate deforestation.
To explore this tradeoff, this article develops a spatially explicit model of land use
and estimates probabilities of alternative land uses as a function of land characteristics
and distance to market using a multinomial logit specification of this model.
Controls are incorporated for the endogeneity of road placement.
Download the full document (PDF , 358 Kb), click here ...
The loss of tropical forests is a global concern because of its impact on biodiversity and climate. Roads are viewed as having precipitated much of this loss by opening forest areas to logging and agricultural conversion. This view poses a dilemma: road construction has traditionally been one of the most important tools for rural development (Creightney 1993) and, moreover, is thought to favor the rural poor (Lipton and Ravallion 1995). It is therefore important to quantify the impact of road building on both deforestation and development in order to assess the severity of the tradeoff between environmental preservation and economic growth.
Although planners are aware of the deleterious effect of roads on forests, they have no empirical guidance on the extent and nature of that damage. How far from the road do conversion effects extend? What kind of conversion is induced? Whom does it benefit? In this article we propose a model that addresses these questions. Following von Thünen (1966), we hypothesize that land is devoted to the use that generates the highest potential rent. Roads play an important role in determining rent—and thus land use—by affecting agricultural output and input prices. But we hypothesize that the impact of roads will be strongly modulated by other factors affecting rent, including soil quality and distance from markets. If our hypothesis is correct—that road impacts are modulated by local conditions—then it may be possible to locate roads so as to spur development while minimizing induced deforestation.
Our goal is to develop a spatially explicit framework for testing this hypothesis. Spatially explicit models are appropriate for two reasons. First, they exploit rich spatial variation in variables of interest—variation that is obscured in aggregate data (for example, district-level means). Second, location matters. In general, we are interested not just in the physical extent of deforestation, but also the degree to which it affects critical habitats and watersheds. The model we present uses spatially disaggregated data, controls for a wide variety of land and soil characteristics, employs multiple land-use categories, and is embedded in an economic framework.
The model is estimated using 1989– 92 data from Belize. Still mostly forested, Belize is of great interest for conservation because of its rich biodiversity and of its relatively large tracts of contiguous forest. Despite its small size, Belize exhibits many different deforestation processes, including encroachment by swidden agriculturalists and forest conversion to pasture, citrus groves, and large mechanized farms. In addition, Belize has superb documentation of land use and land characteristics, which facilitates this kind of study.