Back to History Current Version

Participatory agricultural technology development and dissemination: An experimental evidence from rural China

Last registered on January 05, 2021

Pre-Trial

Trial Information

General Information

Title
Participatory agricultural technology development and dissemination: An experimental evidence from rural China
RCT ID
AEARCTR-0006792
Initial registration date
January 05, 2021

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 05, 2021, 6:57 AM EST

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

Locations

Primary Investigator

Affiliation
Wageningen University

Other Primary Investigator(s)

PI Affiliation
Nanjing Agricultural University
PI Affiliation
Wageningen University

Additional Trial Information

Status
On going
Start date
2018-03-01
End date
2021-07-01
Secondary IDs
Abstract
The registered study is a randomized controlled trial (RCT) implemented to assess the impact of a participatory agricultural extension program that aims to promote sustainable farming in rural China. We conduct this RCT in 135 villages, which are mainly grain farming villages in the North China Plain. Using detailed panel data, we explore if the program improves rural smallholders’ agricultural and environmental outcomes.
External Link(s)

Registration Citation

Citation
Feng, Shuyi, Fan Li and Maarten Voors. 2021. "Participatory agricultural technology development and dissemination: An experimental evidence from rural China." AEA RCT Registry. January 05. https://doi.org/10.1257/rct.6792-1.0
Sponsors & Partners

Sponsors

Experimental Details

Interventions

Intervention(s)
We conduct a randomized controlled experiment with a total of 135 villages in the rural North China Plain. We implemented two types of extension programs as the main intervention arms. Intervention one is the participatory extension services, which integrates farmer field schools, field demonstrations, and case-to-case counseling to promote sustainable agricultural production among rural smallholders. This intervention is a modification of the China Agricultural University's science-and-technology-backyard [STB] design. The STB extension service is taking an active participatory approach to re-develop the local farming technologies (or farming practices) with the consultation from local stakeholders, including local smallholders, machinery service providers, cooperatives, inputs providers at the village. The re-developed farming technology is named by local extensionists as “double-high” technologies, which stands for high yield and high nutrient use efficiency. The technology redevelopment mainly derives from three primary concerns: (a) Local physical context adjustment, which enables extension services are locally relevant; (b) Local social context adjustment, which enables the recommended improved farming practices and technologies are socioeconomically feasible; (c) A local farmer-oriented technology dissemination approach, in which farmer field school, field demonstration, and case-to-case counseling are main activities implemented during the intervention.

Besides the STB extension services, treatment two is an extension program through text message (an SMS intervention), in which the content of SMS is equivalent to the STB extension service; however, the information was disseminated through the SMS. Rural smallholders were invited to receive the recommended farming practices (or the double-high technologies) through timely text messages.
Intervention Start Date
2018-09-15
Intervention End Date
2019-12-30

Primary Outcomes

Primary Outcomes (end points)
The outcomes of our study, following the research hypotheses, can be divided into three levels: (a) the intermediate outcomes, which are mainly the outcomes that used to explain smallholder’s final behaviors; (b) the behavioral outcomes, which are mainly about smallholder’s actual farming practices; and (c) smallholder’s agricultural production, environmental and socioeconomic outcomes. The structure of these outcomes is presented as shown in Figure 1 (in the attached project document).
Primary Outcomes (explanation)
The primary outcomes that we are interested are smallholders' agricultural and environmental outcomes.
(1) Smallholder's agricultural outcomes include smallholder's agricultural productivity outcomes, both partial factor productivity, such as land productivity, labor productivity, and their total factor productivity (TFP). We want to assess if STB extension services (and combined with nudging through text messages) could significantly increase rural households’ agricultural productivity in the NCP. Agricultural productivity will be measured with two sub-groups of indicators, partial factor measures and total factor measures of agricultural productivity.
(a) Partial factor productivity. Yield is the most conventional indicator in measuring agricultural productivity. We use smallholder’s self-reported (both maize and wheat) yields as the primary outcome. To avoid the potential reported bias, we conduct two rounds (both in 2019 and 2020 endline survey) of small-scale field measurement of yields to compare with the self-reported yields. A series of other inputs will also be measured to calculate other partial agricultural productivity, including total labor input and machinery input. All these indicators will be calculated to reflect each factors productivity.
(b) Total factor productivity (TFP). TFP can comprehensively reflect the efficiency of the whole agricultural production process. We will use several different approaches to calculate the TFP at both household and village level. Specifically, we will use Cobb-Douglas function to measure the TFP. (Detailed calculation can be found in Appendix 1: Productivity Measurements). Further revision about the PAP will be updated once we have a better concrete strategy in measuring the TFP growth at the village level.
(c) Agricultural Technical Efficiency is another group of indicators that commonly used in measuring efficiency of agricultural production. To measure technical efficiency, we will use the Data Envelopment Analysis (DEA) method—a nonparametric methods—to calculate the technology efficiency. Different from previous two measures in calculating agricultural productivity, DEA is a non-parametric method and also taking account of all inputs in calculating efficiency. It will provide use three indicators in measuring agricultural efficiency, including total efficiency (TE), pure technical efficiency (PTE) and scale efficiency (SE).

(2) Smallholder's farming environmental outcomes. We want to assess if receiving STB extension services (treatment 1) and combined with nudging through text message (treatment 2) can reduce pollution from agricultural production and improve community environment. Specifically, we examine if treatment 1 and 2 could reduce smallholders’ use of chemical fertilizer, pesticide and herbicide usage. We consider environmental sustainability in agriculture production as the balance of nutrient application and plant absorption. This is particularly the case with nitrogen (N) use efficiency (NUE). Therefore, the first group of environmental sustainability outcomes will focus on nutrient inputs in the field. We will measure the input of nitrogen (N), phosphorus (P) and potassium (K), and calculate nitrogen use efficiency (NUE), phosphorus use efficiency (PUE) and potassium use efficiency (KUE).

(3) Smallholder's socioeconomic outcomes. We want to study the effect of receiving STB extension services on improving smallholders’ socioeconomic development outcomes. We specifically focus on two sub-groups of outcomes. First, if receiving STB extension services increase rural households’ economic welfare. These indicators include (1) household total income, agricultural income and profit from grain production. Second, if receiving STB extension services increase household off-farm employment, increased migration. Third, if receiving STB extension services increase smallholders’ social capital. These indicators include smallholders’ interpersonal trust, confidence in public institutions. To measure the individual household social trust, we adopt the same measurements from world value survey (WVS 2016).

Secondary Outcomes

Secondary Outcomes (end points)
The secondary outcomes we are interested includes two levels outcomes:
(1) Intermediate outcomes that we have presented in the causal chain analysis and the research hypothesis test. It incluldes smallholder's knowledge of sustainable farming, their awareness of consequences of unsustainable farming practices, smallholder's perceptive elasticity of chemcial inputs and their preceived behavioral control and their behavioral attitude.

(2) Smallholder's farming behavvioral changes. We focus on types of farming behaviors: First, smallholder's farming practices that enhance their agricultural production (productivity enhancement behaviors). The field practices that could enhance smallholder’s agricultural production can be summarized in six specific segments, including land preparation, seeding, fertilization, irrigation, plant protection and harvest. Second, we focus on smallholder's framing practices to improve their environmental sustainability. The field practices that could enhance smallholder’s environmental sustainability mainly focuses on the chemical inputs—chemical fertilizer use, and the pesticide and herbicide use.
Secondary Outcomes (explanation)
Measuring smallholders’ intermediate outcomes
1. Knowledge of (sustainable) farming (both production enhanced production and environmentally sustainable production).
To measure smallholder’s farming knowledge, we developed a standard farming knowledge test jointly with agronomists from China Agricultural University. The test consist seven sets of specific farming knowledge include seeding, land tillage, plant root management, irrigation, fertilization, plant protection, and harvest. In total the standard test has 20 specific questions associated with all different segments of farming practices. This farming knowledge will be used to measure smallholder’s knowledge about how to improve its agricultural production and ensure a sustainable resource use and environmental conservation. Specifically, we will only measure the main decision maker within the household. We present the standard test in the Appendix Table A1.

2. Awareness of consequences of unsustainable farming. Measuring smallholder’s awareness of consequences of unsustainable farming practices, we specifically focus on smallholder’s use of chemical inputs and irrigation. First, the non-point source pollution caused by inappropriate use of chemical fertilizer (e.g., synthetic fertilizer, carbamide, phosphate fertilizer) and pesticide and herbicide are the main sources of pollution. How smallholders perceive about such a chemical pollution impacts the environmental condition and if smallholders are aware of these pollutions. We measure smallholders’ awareness of consequences of chemical fertilizer and pesticide and herbicide with the following set questions (Bamberg and Schmidt, 2003; Stern et al., 1999): (Extremely disagree [1] ~ Extremely agree [7]).
Second, water scarcity is another most pressing environmental problem in the North China Plain. Irrigation is one of the main farming practices consuming a large amount of fresh clean water. We measure smallholder’s awareness of water scarcity with the following set of questions (Extremely disagree [1] ~ Extremely agree [7], Bamberg and Schmidt, 2003; Stern et al., 1999).

3. Perceived elasticity of input factors. To measure smallholders’ perceived elasticity of different input factors on agricultural output, we mainly focus on two primary input—chemical input and smallholder’s labor input. we developed the following two groups of questions with base fertilization/top-dressing fertilization, plant protection (pesticide and herbicide use) and labor input.
"Based on your largest piece of land, how much additional wheat do you think it can produce per mu if you apply an additional 5kg of synthetic fertilizer per mu in the base fertilization? ( )kg per unit of land"
"Based on your largest piece of land and the current amount of synthetic fertilizer you used for wheat production, if you use 5kg less synthetic fertilizer per mu in the base fertilization, how much wheat do you think you might lose per mu? ( )kg per unit of land"

4. Attitudes and perceived behavioral control on sustainable farming. To measure smallholder’s attitudes towards the sustainable farming practices introduced by the STB and SMS interventions, two specific sets of sustainable farming practices are measured by using the following set of questions.
a. “Do you think that adopting {the recommended farming practices} is a right choice to do.
b. Do you believe that adopting {the recommended farming practices} is good towards the environment?
c. Do you support further disseminate these { the recommended farming practices} to others?
d. Do you believe that your farm is a good case to apply these { the recommended farming practices}?
e. Do you think that you and your family can afford any technic risks that caused by {the recommended farming practices}?
f. Do you think that you could solve any practical problems when adopt these {the recommended farming practices}?”

Measuring smallholder’s behavioral outcomes
We first measure smallholder's farming practices to enhance agricultural production. The field practices that could enhance smallholder’s agricultural production can be summarized in six specific segments, including land preparation, seeding, fertilization, irrigation, plant protection and harvest.

We then measure smallholder's farming practices to improve environmental sustainability. The field practices that could enhance smallholder’s environmental sustainability mainly focuses on the chemical inputs—chemical fertilizer use, and the pesticide and herbicide use.

[Detailed survey instruments are presented in the attached support document]

Experimental Design

Experimental Design
The experiment is a factorial experimental design with two proposed interventions, thus we will have four treatment arms. Group 1 will only receive treatment 1 (the STB extension service program); Group 2 will receive only SMS treatment, and Group 3 will receive both treatment 1 and treatment 2; and the Group 4 will be the pure control group. We randomly assigned all 135 villages into four equal sized groups. There are 34 villages received only the STB extension services during 2018-2019; and 34 villages received the SMS treatment during the same period; and 34 villages received both the STB extension services and the SMS messages, and the remaining 33 villages receive no any treatment.
Experimental Design Details
We randomly assinged 135 villages into four groups of design within each block (county), which means villages within each block (county) will be divided with a same percentage (25 percent) division. We further balance three core covariates in getting the results. These three covariates are distance to the county seat (v2_1), village total seeding area (v5_7) and reported previous year maize yield per hectare (v5_7_d). We run the randomization 500 times to receive the best quality randomized results, and set the initiate seed with 101101 (it could be any number, just for the replication). We use the blocked randomization to ensure a balanced treatment and control groups within each county, and it can also serve us in estimation to improve the estimation efficiency.
Randomization Method
We used stata command “randomize” randomly assigned villages within each block (county) into four equal sized groups.
Randomization Unit
It is a village-level clustered randomized controlled experiment. Village in rural China's context is the basic unit of a community. In Hebei province, the rural village on average containes about 100 to 500 households per village.
Was the treatment clustered?
Yes

Experiment Characteristics

Sample size: planned number of clusters
In total, we have 135 clusters (rural villages) to participate in our experimental study.
Sample size: planned number of observations
Within each cluster, we randomly selected 16 rural households to participate in our survey. Field enumerators were asked to comply with the following criteria in selecting the interviewee within the household. First, the field survey can only be implemented with adults above 18 years old and not vulnerable groups (e.g. children below 16 years old, elderly, disabled, or mentally handicapped persons). Second, the interview should be mainly conducted with a household member who is in charge of farming decisions. If the household head is the main family member is doing (also making decisions of) farming, the household head should be interviewed as the first priority. If the household head was different from the main farming member within the family, then we interview the household member who is mainly in charge of the farming work.
Sample size (or number of clusters) by treatment arms
All 135 clusters (villages) were randomly and evenly assigned to 4 different groups. There are 34 villages (524 surveyed rural households) were assigned to received treatment 1, 34 villages (535 rural households) were assigned to receive treatment 2, 34 villages (about 538 rural households) were assigned to receive treatment 1 and treatment 2, and last, there were 33 villages (about 522 rural households) assigned as the pure control group, and these 33 villages did not receive any treatment.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
We use the following parameters for the power calculation. We set the experiment with minimum power at 0.8, and a 0.05 significance level (a standard desired level). The number of units per cluster is set up as 16 households. To set the basic parameters, such as mean yield, standard variation, and intra-cluster correlation (ICC) for power calculation, we used two previous survey data for estimation. However, given the fact that the randomization was conducted after the baseline survey, we, therefore, use the actual baseline survey data and the ex-post parameters to recalculate the minimum detectable effects (MDE). With N=135 clusters, about 67 villages (clusters) per comparison group. We calculate that with the total sample of 135 villages, we will be able to detect a minimum 0.145 standard deviation increase of wheat yield (relative to the control group), which is about 120.57 kg per hectare.
Supporting Documents and Materials

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

Request Information
IRB

Institutional Review Boards (IRBs)

IRB Name
Wageningen University
IRB Approval Date
2020-09-29
IRB Approval Number
09215846

Post-Trial

Post Trial Information

Study Withdrawal

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

Request Information

Intervention

Is the intervention completed?
No
Data Collection Complete
Data Publication

Data Publication

Is public data available?
No

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials