Networks and low adoption of modern technology: the case of pearl millet in Rajasthan, India

Purpose - – This paper aims to study low adoption of modern technology for pearl millet in Rajasthan, India, from the perspective of social networks. The state has the lowest adoption of modern pearl millet seeds among Indian states. . In particular, this paper tries to identify the limitations of channels with endogenous effects, thereby limiting large-scale adoption of modern varieties that would require social multipliers. Design/methodology/approach - – Defining the network/reference groups in terms of social identity and geographical proximity, this paper utilizes the intensity of interaction with different network nodes to identify the presence of endogenous effects. In particular, this paper uses the interaction of intensity of social exchange with the group level adoptions to establish the presence of endogenous effects. With adequate controls, greater intensity of interaction having a bearing on technology choice can only happen when there exists social learning (endogenous effect) and cannot be associated with other forms of social effects (namely, exogenous and correlated effects). Findings - – This paper finds evidence for the existence of endogenous social effects in adoption but largely from exclusionary channels. A comprehensively mapped network is used with its intensity to explain the extremely low rate of adoption. Only close-knit networks that, with social fragmentation, limit benefits to few, affect adoption significantly. The non-functionality of less exclusionary information sources and services can be a factor underlying low adoption. Research limitations/implications - – The main limitation of the study is inability to control for unobserved individual heterogeneity because of the cross-sectional nature of data. Further, although an extensive mapping of individual networks has been done, it still cannot be guaranteed to be exhaustive. Practical implications - – With fragmentation, large-scale adoption programs would require networks, sources of information and services that are less exclusionary. Based on the survey data, media and non-religious organizations play a focal role here in the adoption of modern technology. This finding is extremely crucial for policy, as these channels comprise direct policy levers in a fragmented society like India. Indeed, several government programs in India have relied on these channels to run large-scale adoption programs. Their ineffectiveness could be a prime factor for such limited dissemination of technology in Rajasthan. Social implications - – In different settings, social fragmentation could be an important factor determining technology adoption outcomes. The evolving consensus in the literature based on several studies is that ethnic fragmentation has potentially negative consequences on macro-economic performance (Alesina and Tabellini, 1989 and Collier, 2000). In the literature on technology adoption, the role of fractionalization is somewhat under-studied. With fragmentation, there can be significant micro-level impacts (for instance, low technology adoption of a crop) if channels that are inclusive are not well developed. The finding that channels like extension services, media or organizations are not effective in determining choice of technology does not mean that they should not be tapped. The empirical findings suggest that, in their current form in the state of Rajasthan, the roles played by these are limited. The policy implications would be to develop these systems in a way that there is a greater uptake. Recall that less than 4 per cent of the respondents got information on seeds from media sources, an extremely low number. There is certainly scope for increasing the outreach of these channels that are much more important for spread of agricultural technology in a fragmented society. Originality/value - – This paper is an attempt to come up with an empirical strategy to mitigate the issues related to reflection problem. In the cross-sectional data itself, we use the interaction of group choices with intensity of interaction within the group to introduce a non-linearity that tries to bypass the identification issues as in reflection problem. This method of introducing non-linearity in cross-sectional data is a novel attempt to achieve identification of endogenous effects.