Farmers with access to digital agricultural extension shared more information with their peers and became sought-after members of farmer peer groups
The spread of information and communications technologies (ICTs) throughout the world is revolutionising the delivery of agricultural information (Fabregas et al. 2019). Millions of farmers can now turn to their mobile phones for high-quality information relating to weather forecasts, recommended seed varieties, pest management, and more. But as farmers become less dependent on ‘traditional’ sources of information like their peers and input sellers, will ICTs also restructure their social interactions by also altering their in-person peer interactions?
This question is especially relevant to the study of information dissemination and in agriculture. A series of influential papers have combined complex models with rich datasets to demonstrate the power of ‘social learning’ in the adoption of new agricultural technologies (Munshi 2004, Foster and Rosenzweig 1995, Conley and Udry 2010).1 In this context, ‘social learning’ refers to an iterative process of local information exchange among farmers facing similar conditions. Given the wide array of variables (soil quality, pests, weather, etc.) that farmers must consider when they make decisions, the input recommendations and practices that make for successful agriculture in one region may not work in the next. The exchange of local, or ‘bottom-up’, information is therefore crucial and difficult to replicate with ‘top-down’ sources even when they are customisable. As such, if farmers are substituting information from ICTs for information from their peers this may influence the economic rationale behind peer interactions and have unintended downstream effects on technology adoption.
Experimental design and the Avaaj Otalo intervention
A recent study (Fernando 2021) builds on prior work with Shawn Cole (Cole and Fernando 2021) which shows that a mobile phone-based agricultural extension service reliably altered farmers’ sources of agricultural information and influenced input adoption. The study uses data from the same experiment that randomised access to a mobile phone-based extension service – Avaaj Otalo (AO) – among 1,200 cotton farmers in Gujarat, India. The service had a ‘push’ component that sent automated notifications to farmers on a weekly basis, and a ‘pull’ component through which farmers could call in to a helpline to receive expert advice from agronomists. Eight hundred farmers received toll-free access to the service, and the remaining 400 farmers served as a control group. In three surveys over a period of two years – a baseline at year zero, a midline at year one, and an endline at year two – participants were asked to provide a list of three peers (hereafter referred to as peers) with whom they frequently discussed their agricultural work. To gauge the usefulness of AO as an informational tool, participants’ scores on an index of agricultural knowledge were measured with a set of 44 questions regarding general agriculture, cotton, and other crops like wheat and cumin. The surveys also collected detailed information on farm operations. Lastly, the Becker-DeGroot-Marschak (BDM) mechanism is used to obtain and measure the willingness to pay for a 9-month subscription to the AO service.
Changes in sources of information and information exchange
Use of the AO service increased scores on the knowledge index and the treatment was associated with a modest increase in cotton-related knowledge among treatment group farmers. It was also observed that farmers substituted using their peers for advice (-0.14 standard deviations) for the use of the service’s information (1.14 standard deviations) across an index that captures the sources of information used in seven key categories related to the choice of inputs used in their agricultural practices.
Despite this preference for information from the AO system, farmers in the treatment group did not change the frequency with which they spoke to their peers. In fact, farmers who have access to AO were even more likely to share information with their peers than those without it. Treatment group farmers were more likely to share information with (6.6 percentage points) and recommend inputs to (7.9 percentage points) their peers than farmers in the control group.
Changes in the composition of farmer peer groups
In addition to being more likely to share information, treatment group farmers also became more popular among their farmer peer groups, a likely consequence of the increased value that all respondents placed on information from the service as suggested by the willingness-to-pay (WTP) experiments. Even study respondents who were indirectly exposed to the AO service valued it more. Control group respondents who were indirectly exposed to the treatment through a (treatment group) peer increased their willingness to pay (WTP) for the service by 46%. Importantly, the WTP responses use an incentive compatible mechanism (BDM) that requires farmers to pay for a subscription. In contrast to survey-based responses that may be biased by farmers telling us what we think we’d like them to say, the WTP exercises require farmers to ‘walk the talk’.
Consistent with this increased valuation, farmers seem to restructure their ‘peer groups’ to include more treatment group peers after – both indirect and direct – exposure to the treatment. Control group participants with treatment group peers at baseline are nearly 15 percentage points more likely to list a new treatment group peer – i.e. one they did not list at baseline – at the midline or endline. Treatment group farmers similarly seek out treatment group peers as a source of information and they are 4.2 percentage points more likely to list a new treatment group peer at the midline or endline.
The aggregate effect of these changes is that treatment group farmers are more likely be ‘peaks’ within their communities at the end of the experiment. That is, a farmer that is frequently mentioned as a source of information by their peers and likely has a reputation as a conduit of quality agricultural information.
Scaling and complementarities between the treatment group
A key question alluded to at the onset of this column is how the spread of ICTs might influence local information sharing at scale. Or put otherwise, what happens when everyone gets AO? While the experiment is unable to test this proposition directly, examining complementarities between treatment group respondents who have peers that also received the service can inform this question. Treatment group farmers who themselves have a treatment group respondent in their peer group both use the service more than treatment group farmers without a treatment group peer – they are in fact 11 percentage points more likely to call into the line and use the service for nearly two hours longer. In addition, treatment group farmers with treatment group peers are 10.8 percentage points more likely to visit the homes of their peers to discuss agriculture and increase their WTP for the service relative to a control respondent without any treatment group peers.
The spread of ICTs and unintended effects on the structure of peer interactions
Taken together, these findings suggest that the introduction of ICTs may induce persistent changes in the structure of peer groups. While the study does not find that these changes influence downstream input adoption, this is an intriguing topic for future research with designs that explicitly test for these effects. Overall, the findings suggest the importance of understanding how information interventions and the spread of technologies more generally influence the structure of social interactions and networks. Stated more generally, rather than considering peer networks as ‘fixed’, these results hint at their fluidity and contribute to a nascent literature that quantifies these effects (Comola and Prina 2021, Barsbai et al. 2020, Chandrasekhar et al. 2020).
Bandiera, O, and I Rasul (2006), "Social networks and technology adoption in northern Mozambique", The Economic Journal 116(514): 869-902.
Barsbai, T, V Licuanan, A Steinmayr, E Tiongson, and D Yang (2020), "Information and the acquisition of social network connections", NBER Working Paper 27346.
BenYishay, A and A M Mobarak (2019), "Social Learning and Incentives for Experimentation and Communication", Review of Economic Studies 86(3): 976-1009.
Banerjee, A V, A G Chandrasekhar, E Duflo, and M O Jackson (2020), "Changes in social network structure in response to exposure to formal credit markets", NBER Working Paper 28365.
Cole, S A, and A N Fernando (2021), "Mobile’izing Agricultural Advice Technology Adoption Diffusion and Sustainability", The Economic Journal 131(633): 192-219.
Comola, M, and S Prina (2021), "Treatment effect accounting for network changes", Review of Economics and Statistics 103(3): 597-604.
Conley, T G and C R Udry (2020), "Learning about a new technology: Pineapple in Ghana", American Economic Review 100(1): 35-69.
Fabregas, R, Kremer, M and Schilbach, F (2019), "Realizing the potential of digital development: The case of agricultural advice", Science 366(6471).
Fernando, A N (2021), "Seeking the treatment group: The impact of mobile extension on farmer information exchange in India", Journal of Development Economics 153: 102713.
Foster, A D and M R. Rosenzweig (2010), "Microeconomics of technology adoption", Annual Review of Economics 2(1): 395-424.
Munshi, K (2004), "Social learning in a heterogeneous population: technology diffusion in the Indian Green Revolution", Journal of Development Economics 73(1): 185-213.
1 See also Bandiera and Rasul (2006), BenYishay and Mobarak (2019), Beaman et al. (2018).