Introduction

The awareness of the many services provided by tropical forest (Costanza et al. 1997) and the rapid increase in tropical deforestation has put forest at the center stage of the agenda for developers, conservationists, and policy-makers. Researchers have studied many of the factors influencing deforestation, such as the opening of new roads (Chomitz & Gray 1996; Reid 2001), land property right issues (Deacon 1999; Alston et al. 2000; Godoy et al. 2001), the spread of industrial cash cropping (McMorrow & Talip 2001), and slash-and-burn agriculture, cattle ranching, and logging activities (Hecht & Cockburn 1989; Palm et al. 2005).

Researchers have found that the various factors that cause deforestation are woven together in a complex net of interactions (Angelsen & Kaimowitz 1999). Furthermore, authors find that the causes of deforestation could, in turn, result from deforestation. Because of the complexity of the issue, the variability across sites, and the lack of reliable empirical information, there is little consensus on an overall mechanism to explain deforestation (Kaimowitz & Angelsen 1998). But the lack of empirical information is one of the main drawbacks on research on deforestation. Kaimowitz and Angelsen (1998) reviewed 146 econometric models explaining deforestation and found that 24% were based on simulations and 23% draw on analytical models, i.e. theoretical mathematic equations including no empirical data.

Furthermore, among the 53% of the studies based on empirical data, 38 drew on secondary, national-level data. Only nine of the models reviewed (or 6% of the total) used household-level empirical data. The authors suggest that future studies of the causes of deforestation should focus on either household or regional-level data, because studies without a strong micro-level empirical Cash cropping and deforestation base are of little value: “There is no substitute for careful, quantitative micro-level empirical research, and the volume of such studies is not impressive. Plausible theoretical mechanisms are often found to be of little empirical relevance” (Kaimowitz & Angelsen 1998: p. 99).

This study has two parts. In the first part, we draw on household data from Tsimane’ Amerindians, a horticultural and foraging society in the Bolivian Amazon, to assess how cash cropping by smallholders affects the clearance of neotropical forests. We study clearance of fallow and old-growth forest because previous research suggests that fallow forests harbor substantial biological diversity (Finegan 1996; Silver et al. 1996; Smith et al. 1999), but we pay special attention to old-growth forests.

In the second part of the paper, we use the same data to simulate the consequences of having the rural poor cash crop their way out of poverty. We focus on the consequences for the total area of forest clearance, for household labor requirements, and for fallow duration. We pay special attention to fallow duration because previous research suggest that increased land scarcity reduces the fallow period (Coomes et al., 2000). This work contributes to the debate on the causes of deforestation in several ways.

First, we use household-level data which, according to Kaimowitz and Angelsen (1998) is of great value and still relatively rare. Second, we document deforestation by indigenous peoples. Indigenous peoples do not account for a large share of deforestation, but this share could grow as indigenous people become more integrated into the market economy (Godoy 2001) or in response to population growth (Picchi 1991). Third, we contribute to the debate on the effects of agricultural technology innovations on Cash cropping and deforestation 5 deforestation. The results of the simulations help formulate recommendations on the type of technologies needed to reconcile development and conservation of tropical forests.