The Causal Impact of Information Crowd-sourcing Platform on Agricultural Supply Chain

Abstract

This thesis aims to study the impact of increased access to market information on the welfare of micro-firms in informal supply chains, as measured by revenue and price realization. In most informal supply chains, micro-firms receive access to only limited market information through their informal networks (e.g. SMS or Whatsapp). Previous literature finds that the introduction of mobile phones significantly improves market performance for perishable commodities but not so for non-perishable goods. We study whether tools that increase access to market information beyond informal information channels, e.g. through distributed price promotions or through crowdsourcing, can further improve market performance. By leveraging transactions data of an information-sharing platform used by micro-firms in informal supply chains in a developing country and data from a pre-intervention survey, we seek to investigate the causal impact of the platform on the revenue of the users.

We found that the app leads to a significant increase the price (0.43%), post grading price (0.4%), and revenue (0.5%) realization of the sellers. We observed significant heterogeneity in post grading price treatment effect and new buyer rate. The heterogeneity in the former is driven by seller’s learning, while the heterogeneity in the latter is driven by local information availability.

Difference in means analysis revealed that the difference in post grading price treatment effect between users with high and users with low experience is around 80% of the average treatment effect. Linear regression indicates that as the user builds familiarity in using the platform, the conditional average treatment effect can grow by up to 7 times (from 0.2% to 1.6%). We found no evidence that information availability or prior commitment to sell to certain buyers affect the post grading price treatment effect.

Difference in means analysis indicated that heterogeneity in new buyer rate is driven by two factors, local information availability and prior commitment of the seller to sell to certain buyers. We found that the probability of transacting with a new buyer increases with local information availability but decreases with the presence of prior commitment. Regression analysis found statistically non-zero positive effect for local information availability, but not for prior commitment.

The observation that local information availability increases new buyer rate but not post grading price treatment effect can be explained by the fact that user may opt to switch to a new buyer for reason other than better pricing, e.g. shorter travelling distance. Due to difficulty in estimating transportation cost and travelling distance, our analysis in this work is limited to top-line impact only. Our analysis does not account for users who switched to a new buyer for travelling reason and benefit from the saving in transportation cost. Consequently, the results reported here are underestimation of the actual impact of the intervention.

Overall, we presented proof of concept for information crowd-sourcing platform as a low cost option to improve access to market information.