A Spatial Propensity Score Matching Evaluation of the Social Impacts of Sugarcane Growing on Municipalities in Brazil

Abstract

The expansion of sugarcane growing in Brazil, spurred particularly by increased demand for ethanol, has triggered the need to evaluate the economic, social, and environmental impacts of this process, both on the country as a whole and on the growing regions. Even though the balance of costs and benefits is positive from an overall standpoint, this may not be so in specific producing regions, due to negative externalities. The objective of this paper is to estimate the effect of growing sugarcane on the human development index (HDI) and its sub-indices in cane producing regions. In the literature on matching effects, this is interpreted as the effect of the treatment on the treated. Location effects are controlled by spatial econometric techniques, giving rise to the spatial propensity score matching model. The authors analyze 424 minimum comparable areas (MCAs) in the treatment group, compared with 907 MCAs in the control group. The results suggest that the presence of sugarcane growing in these areas is not relevant to determine their social conditions, whether for better or worse. It is thus likely that public policies, especially those focused directly on improving education, health, and income generation/distribution, have much more noticeable effects on the municipal HDI.

Sugarcane industry's socioeconomic impact in São Paulo, Brazil: A spatial dynamic panel approach

“…Scientific literature suggests that the expansion of Brazil's sugarcane industry has generated net benefits for the population, especially in the generation of employment and income, while acting to mitigate the effects greenhouse gas emissions Toneto-Jr et al, 2008;Hoffmann, 2006;Chagas et al, 2012;Satolo and Bacchi, 2013;Bacchi and Caldarelli, 2015). Since the sector's most evident impacts are felt at the local rather than national level, it is reasonable to expect added insights to come from municipality scale analyses, realizing that results of such analysis may not be consistent with national trends.…”

“…Employing the spatial propensity score matching method, these authors concluded that increased sugarcane cultivation in the evaluated municipalities is not relevant to determining social conditions, either for better or for worse. Their study did not consider the impact of sugarcane related industrial processing units (Chagas et al, 2012). Bacchi and Caldarelli (2015) undertook a panel data analysis to examine externalities arising from the expansion of the sugarcane industry in the state of São Paulo using the Municipality Development FIRJAN Index (IFDM).…”

“…An observation may be associated with another according to its spatial location due to the behavior of economic agents who define choice according to local market conditions, regional market conditions, or the distance between regions (Elhorst, 2014 pioneering studies that introduced spatial control panel models and municipal specific data into a study of the Brazilian sugarcane industry (Chagas et al, 2012;Satolo and Bacchi, 2013). These studies showed that the regional dispersion of sugarcane cultivation implies the existence of a spatial relationship between the analyzed municipalities.…”

“…Processing plants have been shown to be key driver of endogenous growth in many Brazilian municipalities (Shikida and Souza, 2009). In addition to the positive direct economic effect of a new processing plant brought about by job creation, there are also positive economic effects on the local business sector, the local service sector, and municipal tax revenue (Chagas et al 2010;Satolo and Bacchi, 2013;Shikida and Souza, 2009 (Chagas et al 2012). On a more regional level, Terciote observed that the implementation of an industrial plant creates demand for many steel industry products, such as machinery, tractors and construction materials, further stimulating the regional economy (Terciote, 2006).…”

Does regional science need an experimentalist buzz?

“…Chagas, Toneto, & Azzoni (2012) have done a first step towards this direction by using spatial regression techniques in the course of propensity score estimation. This may help to produce better matches of treated and control groups given the spatial pattern of many regional policy instruments.…”