NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
The Value of Advice: Evidence from the Adoption of Agricultural Practices
Initial registration date
April 01, 2015
April 01, 2015 11:02 AM EDT
Other Primary Investigator(s)
Harvard Kennedy School
Additional Trial Information
Attempts to explain the astonishing differences in agricultural productivity around the world typically focus on farm size, risk aversion, and credit constraints, with an emphasis on how they might serve to limit technology adoption. This paper takes a different tack: can managerial practices explain this variation in productivity? A randomized evaluation of the introduction of a mobile phone-based agricultural consulting service, Avaaj Otalo (AO), to farmers in Gujarat, India, reveals the following. Demand for agricultural advice is substantial and farmers offered the service turn less often to traditional sources of agricultural advice. Management practices change as well: farmers invest more in recommended agricultural inputs resulting in dramatic increases in yield for cumin (26.3%), and improvements in cotton yield (3.5%) for a sub-group that received frequent reminders to use the service. Peers of treated farmers also change their information sources and cropping decisions, albeit to a much smaller extent. Farmers appear willing to follow advice without understanding why it is correct: we do not observe gains in agricultural knowledge. We estimate that each dollar invested in AO generates a return of $10. These findings highlight the importance of managerial practices in facilitating technology adoption in agriculture.
Cole, Shawn and A. Fernando. 2015. "The Value of Advice: Evidence from the Adoption of Agricultural Practices." AEA RCT Registry. April 01.
All study respondents first participated in a household paper survey, which covered questions on household characteristics, farm plot characteristics, social networks and agricultural practices. Following this, treatment farmers were provided access to a toll-free agricultural helpline. Farmers received weekly voice push messages through this service, and could also use this helpline to record agricultural questions, listen and respond to agricultural questions of other farmers and to share their agricultural experiences. The farmer did not have to bear any cost for listening to the weekly messages or calling in to the AO line.
Additionally, a smaller subset of farmers received complementary traditional agricultural extension services, which includes field visits and demonstrations by agronomists. Following the start of the intervention, all respondents participated in monthly phone surveys and two more paper surveys – mid-trial (in July 2012) and towards the end of the study (July 2013). During the 2012 survey, farmers were asked an additional set of questions on basic household and individual level questions in order to broadly compare the quality, response rate and accuracy of responses to paper and phone surveys.
Half of the participants were randomly administered these questions in the paper survey, and the other half were asked the same questions on the phone. In the final paper survey, respondents also participated in a “willingness to pay” game. During the game, 75% of the respondents, selected at random, were asked if they would like to purchase AO at a series of different price points (in decreasing order, ranging from Rs. 500 to Rs. 0) Following this, they were given a pre-assigned scratch card where their randomized price is listed (ranging from Rs. 40 to Rs. 240). If the scratch card price was lower than their demand then the respondent could buy the product at the discounted price. The other 25% played a “take it or leave it” game where they were simply asked if they are willing to purchase the service at X price where X is randomized between Rs. 40 and Rs. 240.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Impact on agricultural knowledge, adoption of recommended agricultural inputs, crop yields, expenditure on inputs, agricultural income, and trust in mobile-phone based agricultural information.
Primary Outcomes (explanation)
We use the paper-based household surveys as a means of testing the impact of the intervention over the course of the study. Prior to the start of the intervention, farmers participated in a household survey which covered major topics including key agricultural and demographic variables, and also tested their knowledge about basic agricultural information. Following the intervention which provided farmers with relevant agricultural information and answered any questions they might have, farmers were again tested on the same agricultural questions to see if the intervention had impacted their agricultural knowledge. Similarly, the paper surveys collected information on agricultural inputs used to gauge if there was a change in agricultural practices following the intervention. Farmers were also asked to identify which sources of agricultural information they rely on the most, and how much trust they place in these sources. We also collect information on crop yield, revenue and annual income to see if the use of mobile-phone based agricultural information leads to better agricultural outcomes for farmers.
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
With the help of our partner organization – Development Support Center (DSC), we identified 1200 farmers that grow cotton and own a mobile phone. All respondents were over 18 years of age and the agricultural head of household i.e. the person who makes all the agricultural decisions. These 1200 farmers were then randomly assigned into groups of 400 farmers each. 400 treatment farmers received access to the toll-free agricultural hotline and also received in-person traditional agricultural extension. Another 400 received only access to mobile-based agricultural information. Finally, 400 farmers served as the pure control group and did not receive mobile-phone based information or traditional extension.
Experimental Design Details
randomization from a computer in an office
Was the treatment clustered?
Sample size: planned number of clusters
Sample size: planned number of observations
1200 farmers who are above 18 years of age, grow cotton and own a mobile phone
Sample size (or number of clusters) by treatment arms
400 control farmers who do not have access to the toll-free agricultural helpline, 400 farmers who have access to the agricultural helpline and 400 farmers who have access to AO and also receive in-person physical extension.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Main Outcome Variable MDE As % of Mean S.D
Measure of Knowledge 0.2 0.31 1.07
Expenditure on Fertilizers per Acre in rupees 461.4 0.16 2329.31
Expenditure on Pesticides per Acre 185.8 0.15 938.02
Effect on Information Sources 0.1 0.47 0.36
Impact on Yield per Acre 46.1 0.10 232.9
INSTITUTIONAL REVIEW BOARDS (IRBs)
Harvard University- Committee On the Use of Human Subjects in Research
IRB Approval Date
IRB Approval Number
Post Trial Information
Is the intervention completed?
Is data collection complete?