not-registered Login to view full entry:

Lab-in-the-field and Weather Forecast Service Experiment with Coffee Farmers

INDIND -23 -1862

Last modified on December 19th, 2025 at 10:42 am

Basic Information

Abstract
PxD operates the Coffee Krishi Taranga (CKT) platform in collaboration with the Coffee Board of India to provide a voice-based advisory service for coffee farmers through a two-way Interactive Voice Response (IVR) system. We partnered with the Climate Forecast Applications Network (CFAN) to develop highly localized, 5-day rainfall forecasts tailored to the contexts of coffee farmers in Karnataka.

Farmers face significant productivity risks from weather variability, and these risks are amplified by climate change. We conducted an experiment with coffee farmers registered with CKT in Karnataka to examine how short-to-medium-range rainfall forecasts could support farmers’ decision-making, conditional on their ability to accurately interpret, trust, and act on forecasts. The study examined: how farmers form beliefs about weather and forecast accuracy as they observe repeated forecasts and outcomes; whether light-touch informational treatments improve farmers’ understanding of probabilities; and how climate vulnerability influences farmers’ beliefs about weather and forecast accuracy. Using lab-in-the-field (LIF) and real-world IVR experiments, we exposed farmers to weather forecasts and light-touch informational treatments.

While farmers had a high demand for forecast services, their trust in forecasts decreased after they received erroneous forecast predictions, which led to a decrease in the frequency of farmers’ use of the service. Accuracy in initial forecast delivery mitigated this effect, which highlights the importance of early successes for building long-term trust in a new service feature. When climate change was made salient, farmers were more likely to use forecasts and were more tolerant of forecast errors, which underscores the value of forecasts in climate adaptation.
Status
Completed
Start date
Q3 Jul 2023
Experiment Location
India / Karnataka, India
Partner Organization
Coffee Board of India, CFAN
Agricultural season
_Multiple seasons

Research Design

Experiment type
Other
Sample frame / target population
Smallholder coffee farmers in Karnataka
Sample size
27,000
Outcome type
Beliefs or perceptions, Service engagement
Mode of data collection
In-person survey, Phone survey, PxD administrative data
Research question(s)
1. How do farmers form beliefs about weather, based on forecast information and forecast outcomes?
2. How do farmers form beliefs about forecast accuracy, as they observe repeated forecasts and realizations?
3. Can light-touch informational treatments train farmers to correctly interpret probabilities, and do probabilistic-type forecasts influence these interpretations?
4. Does a sense of vulnerability to climate change affect weather beliefs?
5. How much are farmers willing to pay for a weather forecast service?

Research theme
Agricultural management advice, Message framing, Service design, Weather information
Research Design
We conducted LIF and real-world IVR experiments to expose farmers to weather forecasts and light-touch informational treatments. We first conducted the LIF experiment before rolling out the IVR forecast service experiment. Phase 0 of the forecast experiment included farmers who had participated in the LIF and had opted to receive forecasts, thereby creating overlap between the two study samples. We combined data from both components in some analyses, to link farmers’ responses and behaviors in the LIF with their subsequent engagement with the forecast service.
- Lab-in-the-field experiment
  - Sample frame and selection criteria:
    We randomly selected 21 gram panchayats (GPs) across two blocks in Chikmagalur and Kodagu districts in Karnataka. Within these GPs, we randomly sampled 1,212 farmers from the rosters, maintained by the Coffee Board of India, of small- and medium-holder coffee farmers and from existing users of CKT.
  - Randomization protocol with clustering or stratification:
    We conducted randomization on the spot at the GP level. We stratified farmers by GP and assigned them to one of three groups:
    (1) climate change salience treatment (T1),
    (2) climate change salience and probability training treatment (T1 + T2), or
    (3) control group (C).
  - Intervention details:
    T1: We showed farmers a short video on the effects of climate change on coffee cultivation in Karnataka, India.
    T1 + T2: We showed farmers a video that combined the climate change content with an explanation of probability concepts.
    The control group did not receive any video information before participating in the experimental games.
  - Data collection and measurement methods:
    The experiment involved two experimental games designed to assess how farmers interpreted, updated, and acted upon probabilistic rainfall forecasts. The first, a “market-choice” game, measured how accurately farmers interpreted probabilistic forecasts. The second, an “agricultural decision-making” game, measured how farmers used probabilistic forecasts to update their beliefs and make agricultural decisions. In addition to game outcomes, we collected in-person data on farmer characteristics, farm characteristics, risk preferences, and understanding of probabilities.
- IVR Weather Service Experiment via CKT: We included a total of 27,120 farmers across 21 forecast grid cells in 11 blocks in Karnataka in the study sample. We conducted randomization at the village level and stratified it by forecast grid cell. Farmers received an onboarding call explaining the service, and then, according to their treatment status, five-day cumulative rainfall forecasts via voice calls every five days. We made up to three call attempts per scheduled forecast.
- - Phase 0: We first offered the forecast service in April 2024 to farmers who had participated in the LIF experiment and had opted to receive forecasts. We assigned villages with at least five farmers to one of two experimental arms that received: (1) probabilistic forecasts only, and (2) probabilistic forecasts with additional forecast interpretation voice calls.
  - Phase 1: We assigned farmers in five blocks—Somwarpet, Mudigere, Sakleshpura, Belur, and Alur—to Phase 1 and gave them access to the forecast service, starting in July 2024. We assigned villages to one of four arms that received: (1) probabilistic forecasts only; (2) probabilistic forecasts with forecast interpretation voice calls; (3) deterministic forecasts only; and (4, the control group) standard advisory voice calls only.
  - Phase 2: We assigned farmers in six additional blocks—Chikmagalur, Koppa, Narasimharajapura, Madikeri, Arkalgud, and Sringeri—to Phase 2 and gave them access to the forecast service, starting in August 2024. We assigned villages to one of five arms that received: (1) probabilistic forecasts only; (2) probabilistic forecasts with a recommended action or advisory; (3) deterministic forecasts only; (4) deterministic forecasts with a recommended action or advisory; and (5, the control group) standard advisory voice calls only.
For further information see working paper: Surendra, Cole, and Harigaya (2025)

Results

Results
The experiment includes four parts: LIF, Phase 0, Phase 1, and Phase 2. The key experimental results below are mapped to these components, with the components in brackets for clarity in mapping.

Willingness to pay (WTP): In an incentive-compatible elicitation using the Becker-DeGroot-Marschak (BDM) mechanism in the LIF experiment, farmers’ average WTP for an 8-month subscription to the forecast service was INR 204.4 (USD 2.42) or INR 25.55 (USD 0.30) per month. [LIF]

WTP and engagement with forecast service: Farmers’ WTP was positively correlated with their engagement with the forecast service. Farmers who answered more than 50% of the forecast calls had a 6% higher WTP for the service (significant at the 5% level). [LIF, Phase 0]

Accurate weather expectations: Weather expectations were accurate for more farmers in the forecast group (14.9% more farmers) relative to the control group (significant at the 10% level). [Phase 2]

The effect of incorrect forecasts: Farmers were less likely to answer forecast calls following incorrect forecasts. Engagement decreased by 4.5% after a false alarm (significant at the 1% level) and by 3.4% after a missed event (significant at the 5% level), which demonstrates a discouraging effect. Early forecast successes (i.e., no incorrect forecasts in the first five calls) reduced the discouraging effect of false alarms by 4.6% (significant at the 1% level). The reduction in engagement was more pronounced for farmers who were more risk-averse, i.e., those who grew the more weather-sensitive Arabica variety or lacked working irrigation facilities, which suggests that higher stakes amplified discouragement. In blocks with high historical rainfall variability (above the median from 2000 to 2022; serving as a proxy for climate-change exposure), engagement declined by 6.2% following a false alarm, compared to an 11.1% decline in other blocks (significant at the 1% level). [Phase 0, Phase 1, Phase 2]

Climate-change salience treatment effect: The climate change salience treatment increased the forecast service take-up rate by 3%, more than six months after administration (significant at the 5% level). [LIF, Phase 0]

Forecast interpretation treatment effect: The forecast interpretation treatment had muted effects. Intended as a behavioral nudge to remind farmers about forecast uncertainty and probability interpretation, it led to a 3% reduction in overall engagement (significant at the 1% level), which decreased the likelihood of farmers answering both forecast calls and standard advisory calls, likely due to call fatigue. [LIF, Phase 0]

You have successfully submitted or edited this experiment.