Effects of digital agricultural extension on the adoption of smart farming technologies: Evidence from North China

Last registered on January 04, 2021

Pre-Trial

Trial Information

General Information

Title
Effects of digital agricultural extension on the adoption of smart farming technologies: Evidence from North China
RCT ID
AEARCTR-0006887
Initial registration date
December 26, 2020

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
January 04, 2021, 9:17 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Last updated
January 04, 2021, 10:27 AM EST

Last updated is the most recent time when changes to the trial's registration were published.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Affiliation
Huazhong Agricultural University

Other Primary Investigator(s)

PI Affiliation
Wageningen University & Research

Additional Trial Information

Status
On going
Start date
2020-11-26
End date
2022-12-31
Secondary IDs
Abstract
The registered study is a clustered randomized controlled experiment (RCT) examining the impacts of digital-based agricultural extension on farmer’s adoption of smart farming technologies in Inner Mongolia, China. We conduct this RCT with a total of 115 rural villages in two counties in Inner Mongolia, Northwest of China. The selected counties are typical semi-arid climate conditions, maize production is the main staple grain, and the local farming relies on underground water irrigation. Focusing on maize farmers, this study investigates farmers’ adoption of an additional data-driven technology on top of the drip irrigation system they currently use. We explore how a video-mediated agricultural extension program affects farmers’ knowledge about data-driven technologies and the resulting adoption behavior. Furthermore, we examine the adopters’ maize production as well as their water and chemical fertilizers usage; that is, the economic and environmental consequence of adopting data-driven technologies. This study aims to shed light on the effectiveness of digital agricultural extension in overcoming farmers’ knowledge barriers.
External Link(s)

Registration Citation

Citation
Xiong, Hang and Fan Li. 2021. "Effects of digital agricultural extension on the adoption of smart farming technologies: Evidence from North China." AEA RCT Registry. January 04. https://doi.org/10.1257/rct.6887
Sponsors & Partners

Sponsors

Experimental Details

Interventions

Intervention(s)
The smart WFI technology is an improved data-driven dripping irrigation technology. It optimizes the use of water and fertilization by integrating the plant characteristics, weather conditions as well as land parcels’ soil moisture and soil nutrient information. The technology integrates devices for the remote control of water flow and a smart irrigation application (smartphone APP) with an operating system and various data into a conventional dripping system. The use of smart WFI technology can be categorized into three levels:
Level 1: An agricultue APP
The agriculture APP is an integrated platform that provides geographical-based climatic-weather predictions, soil and local main grain planting information for farmers. Famers can install the application on their smartphone, and through the application, farmers will receive all relevant local farming information timely.
Level 2: APP + remote control devices
The smart irrigation APP also has the function of operating devices for controlling field irrigation automatically and remotely. The devices mainly include electromagnetic valves and a wireless gateway. They enable switching on/off the irrigation as well as fertilizer addition through the operation on the APP.
Level 3: APP + remote control devices + smart farming decision-making
Level 3 of the smart WFI technology is further developed from Level 2. It integrates the function of smart decision-making on top of the remote control system. It adds farmer and farmland specific data (which often requires the installation of corresponding sensors) and data-driven algorithms for providing smart decision-making on farming planning.

Three types of treatments are designed and are expected to be carried out to promote the adoption of smart WFI technologies.
Treatment 1: Field Demonstration
The field demonstration shows farmers the installation and operations of the smart WFI technology. Specifically, it includes the installation of both the APP and relevant devices (electromagnetic valve and wireless gateway, etc.) and the field operation of the full set of the smart WFI technology.
Treatment 2: Presentation-based training
Treatment 2 is conventional presentation-based training, which we plan to organize jointly with local agricultural extension agents. The training aims to provide the following information and knowledge regarding smart WFI technology: (a) general knowledge of the technology and the basic rationale of how it is designed and implemented in the field, (b) evidence showing the effect of the technology on yield and resources saving (particularly water and fertilizer saving); (c) a menu indicating the different levels of the technology and their associated costs.
Treatment 3: Video-based training
As an alternative to presentation-based training, we design video-based training. It uses a series of short videos, instead of PowerPoint presentations, presenting information and knowledge equivalent to that provided in presentation-based training. All videos will be presented to the participants once by enumerators. Afterward, they can watch through the APP on their own.

Based on the above three types of treatments, we develop the following three interventions. They will be implemented in two counties, Urad Front Banner and Otog Front Banner, in Inner Mongolia.
Intervention 1: Field demonstration + Presentation-based training
Intervention 1 consists of a field demonstration and a conventional presentation-based training session. Farmers assigned to this intervention group will be invited to visit a field demonstration site in the county. A training session with a PowerPoint presentation follows the field demonstration. The training is provided by researchers and the local agricultural extension agents jointly.
Intervention 2: Field demonstration + Video-based training
Intervention 2 consists of a field demonstration and a video-based training session. Specifically, farmers assigned to this intervention group will receive the same field demonstration as in Intervention 1. However, instead of receiving a presentation-based training program, farmers be invited to a video-based training session. A series of short videos, as training materials, will be provided to farmers through the APP. Farmers can watch the videos repeatedly on their own.
Intervention 3: Field demonstration + Presentation-based training + Video-based training
Intervention 3 consists of a field demonstration, a presentation-based training session and a video-based training session. That is, in addition to presentation-based training following a field demonstration, farmers assigned to this intervention group can also watch short videos through the APP.
Intervention Start Date
2021-01-10
Intervention End Date
2021-03-31

Primary Outcomes

Primary Outcomes (end points)
This program focuses on three sets of primary outcomes.
1. The first is farmers’ adoption of the smart WFI technology. Given that the smart WFI technology is of three levels, farmers’ technology adoption will be measured by three dummy variables, each reflecting the adoption status of a level. The three dummy variables are (a) whether or not a farmer installs the APP and regularly check farming-related information, (b) whether or not a farmer installs remote control devices and uses the APP to control its irrigation remotely, and (c) whether or not a farmer installs the full set of the smart WFI technology and control its irrigation following the farming plan that the APP advises.
2. The second is the effects of interventions on farmers’ agricultural production, specifically maize production. We will measure farmers’ various inputs (i.e., labor, capital) and their maize yields.
3. The third is the effects of interventions on farmers’ environmental outcomes. We will measure farmers’ water usage for irrigation and chemical fertilizer (i.e., nitrogen, phosphorus and potassium) application.
Primary Outcomes (explanation)
Additional outcomes can be constructed from each set of the main variables.
1. Farmers’ adoption intensity can be constructed using the three dummy variables reflecting their adoption status.
2. Farmers’ agricultural productivity can be constructed using main variables reflecting farmers’ inputs and maize yields. First, a farms’ partial factor productivity with a particular focus on land productivity (yield) and labor productivity. Second, a farmer’s total factor productivity (TFP) will be calculated using both the non-parametric data envelop analysis (DEA) approach and the parametric stochastic frontier analysis (SFA) approach.
3. Water and nutrient use efficiency (including nitrogen, phosphorus and potassium use efficiency) can be constructed using main variables reflecting farmers’ water usage and fertilizer application.

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
The experiment includes 116 villages (44 in one county and 72 in another). We sample 10 farm households per village, which produces a total of 1160 sample households. The sampled villages are assigned to four groups at random. Following a factorial experiment design, 29 villages receive conventional agricultural extension services, 29 villages receive digital agricultural extension services, 29 villages receive both, and 29 villages serve as control.
Experimental Design Details
Not available
Randomization Method
Randomization by a computer, in Stata.
Randomization Unit
The basic randomization unit is natural village. The number of households in a village varies from around 50 to over 500 and is on average around 200.
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
We plan to sample 116 clusters (i.e., rural villages) from the two survey counties in Inner Mongolia.
Sample size: planned number of observations
We plan to to have 1160 observations (i.e., farm households), sampling 10 from each village.
Sample size (or number of clusters) by treatment arms
Twenty-nine villages will be assigned to each of the four groups, i.e., Intervention 1, 2, 3 and control.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We estimate the total sample size of the experiment using the optimal design (OD). The significance level is set to be 0.05 (i.e., α=0.05). We plan to sample 10 farm households per cluster. We expected that our experiment will be able to yield a 0.2-0.4 standard deviation increase in primary outcome (e.g., farmers’ yields). The intra-cluster correlation (ICC) is assumed to be 0.02-0.05, relatively lower than those for other regions of China (e.g., North China Plain). This is a reasonable assumption the sampled villages and households in the study region (Inner Mongolia) are much more dispersedly located than other regions. Calculation result shows that around 100 clusters is needed to ensure the power to be higher than 0.8 (a commonly used threshold) with the lowest minimum detectable effect (MDE). Considering cases of losing clusters, we plan to include 116 clusters in the baseline survey.
IRB

Institutional Review Boards (IRBs)

IRB Name
Ethics Committee of Huazhong Agricultural University
IRB Approval Date
2020-09-14
IRB Approval Number
HZAUHU-2020-0001