Changing agents of deforestation: From state-initiated to enterprise driven processes, 1970–2000

Abstract

A meta-analysis of 268 case studies of tropical forest cover change indicates that deforestation shifted from a state initiated to an enterprise driven process between 1970 and 2000. During the 1970s state run road building and colonization programs opened up regions for settlement and deforestation throughout the tropics. By the 1990s these programs had all but disappeared. Meanwhile, enterprise driven processes, present in the 1970s, had both expanded and diversified by the 1990s. The implications of these findings for theories of forest cover change and for forest conservation policy are explored.

Introduction

For almost three decades researchers have investigated the people who destroy tropical forests out of a concern for the losses of biodiversity that occur when the forests disappear. To understand the political, economic, and social forces that drive forest destruction, analysts have investigated why rates of forest loss have varied so much from place to place, usually through quantitative, cross-national investigations. By the late 1990s social scientists had constructed more than 150 models of deforestation processes and published them in a wide variety of scientific journals (Kaimowitz and Angelsen, 1998). These studies outlined some of the basic features of societies that destroy forests rapidly, most notably the presence of powerful economic incentives to convert old growth forests into fields (Lambin et al., 2001).

While the quantitative models are useful for understanding the general features of situations that contribute to forest destruction, these analyses have a disembodied quality to them. They point to general features like levels of affluence and population growth that contribute to forest losses, but they neglect more specific, proximate and sometimes contextual factors like new land settlement schemes, road building, and topography that, when added to the more general factors, produce compelling explanations for forest losses in a place. The neglect of context extends to the historical dimensions of forest destruction. Typically, deforestation modelers engage in ‘comparative statics’, exhaustive analyses of variations in forest cover change at a single point in time.

More qualitatively inclined researchers have also produced a large volume of work on tropical forest losses. Their work typically takes the form of intensive case studies of forest losses in one place. The data for these studies come in diverse forms. Researchers collect information through household surveys, key informant interviewing, and satellite images of forest cover. An increasing number of the case study researchers employ a mix of all three types of data in analyzing forest cover losses (Downton, 1995). While the case studies produce plausible, even authoritative, accounts of deforestation in particular places, their external validity tends to be poor. In other words the complicated accounts about how forest cover losses occurred in one place do not apply particularly well to other places.

Under these circumstances it would be useful if we could aggregate the findings from the case studies across time and space to see if the agents of destruction emphasized in individual case studies form a more general pattern. Geist and Lambin (2001) have already begun this task with a valuable meta-analysis of approximately 150 case studies of forest cover change. In this article, I continue this analytic effort, drawing on a larger number of case studies and incorporating an historical dimension into the analysis. The database contains information about 268 case studies of tropical forest cover change between 1970 and 2000 in both humid and dry tropical forest settings in Africa, Asia, and Latin America. The analysis pays particular attention to shifts in the composition of the agents between the 1970s and the 1980s on the one hand and the 1990s on the other hand.