Meatpacking industry

Mapping slaughterhouses in Brazil, past and present

Challenge. Depicting the evolution of the slaughter industry is crucial to model and understand drivers of land use change. Yet the spatial and temporal dynamics of slaughterhouses across Brazil remain largely unknown due to data limitations.

Solution. We provide a novel method to map slaughterhouse expansion and contraction. Our method is based on data mining and emerging sources of big data. We use computer and human-based matching techniques and triangulation across data sources to produce the most accurate map of the slaughter industry currently available.

The data sources include a registry of 21 million companies, government records of 3 million slaughter transactions, high-resolution satellite imagery, and others. Our work is the first to include longitudinal information and all sizes of slaughterhouses.

Result. The data that we produced are being used to better understand the beef slaughter industry in Brazil. Applications include the classification of regions according to the distribution of plant type and size, and the assessment of the relationship between industry growth, cattle herds, and deforestation.

Starting from the state of Mato Grosso, we analyze the opening and closing of 133 plants between 1967 and 2016 and estimate the geographic locations and slaughter volumes. We are currently finishing the full Amazon and Cerrado regions of Brazil, for a total of over 3 thousand plants.

Mato Grosso

Cattle slaughtering was present since the 18th Century, and the oldest meat processing plant was registered in 1967. Currently, 72 plants operate through 52 holding companies. By measuring distances between active plants and pasture areas, we document a 29% increase in the density of plants in 2000-2016, showing a significant expansion of the cattle slaughter infrastructure. We identify three periods of expansion: 1967-1995, with 15.1% of the plant openings; 1996-2003, with 24.6%; and 2004-2016, with 60.3% of all openings. While closings likely occurred throughout the period studied, no data on closings prior to 2002 was available. We estimate a minimum value for the volume of uninspected slaughter (for food quality) as 2-3% for 2013-2016.