Adoption of Pasture Management Practices in Rondônia, Brazil: Social Media as an Informational Influence

The welfare of rural households depends on income from agricultural production. The adoption of new agricultural practices can improve farmer welfare by increasing yield and/or lowering production costs, but farmers do not always adopt beneficial practices. I examine whether social media use influences pasture management practices among smallholder cattle farmers in Rondônia, Brazil. This Amazonian state is heavily deforested for use as farmland, especially pasture for beef and dairy cattle. Traditional pasture management degrades soil over time, requiring pasture productivity interventions or deforestation for new land. Nontraditional practices can reduce degradation. Agricultural technology adoption literature explores the influence of risk, credit access, and access to information from agricultural extension or neighbors. Farmers tend to trust experiential information from other farmers most. Social media connects farmers to a greater variety of other farmers than before, providing greater information access at a much lower cost than other information technologies. Social media use increases the potential to learn about and adopt new agricultural practices, but few researchers have investigated to what extent it influences agricultural decisions. Such a relationship may suffer from selection bias. Farmers who tend to adopt all kinds of new technologies may be more likely to use both social media and new pasture practices. I estimate the effect of social media use on the adoption of pasture management practices using probit regression with covariates, propensity score matching (PSM), and bivariate probit with instrumental variables. Data comes from a 2019 survey by the Connections between Water and Rural Production project, with responses from 1362 smallholder households. Probit and PSM show social media use correlates strongly with the adoption of any, traditional, sustainable, or complex management, by between 3 and 10 percentage points. This holds for probit when internet access is included. Only any and complex management are significant at 7.2 and 9.5 percentage points for PSM with internet access included. In the bivariate probit model with an instrumental variable, only complex management showed significant effects from social media use and only when internet access was included, at a 15.3 percentage point increase