The impact of a flood forecasting tool on socioeconomic outcomes in Bihar, India

Decision-Maker’s Challenge

In South Asia, 576 million people face flood risks, and these risks are worsening with climate change. In North Bihar, India, three-quarters of the population lives under the recurring threat of flood devastation. Introducing better early warning systems has the potential to mitigate the impact of floods by enabling households to prepare and adapt.

In 2018, Google introduced a flood warning system on Android devices, but few people own the smartphones needed to receive these life-saving messages. To tackle this issue, in 2021, Google, in collaboration with Yuganter, a Bihar-based NGO, launched a community-based flood Early Warning System (EWS) in the Ganges river basin across Bihar. Dr. Rohini Pande (Henry J. Heinz II Professor of Economics and Director of the Economic Growth Center at Yale University) and Dr. Maulik Jagnani (Assistant Professor of Economics at The Fletcher School of Law and Diplomacy at Tufts University) – both affiliates at Inclusion Economics India Centre  (IEIC) (formerly EPoD India) at the Institute for Financial Management and Research (IFMR) have undertaken a multi-year experimental evaluation to measure the impact of the early-warning system.

IDinsight’s DataDelta team supported IEIC with collecting endline data for the evaluation at the end of the 2023 floods season. We surveyed households on various outcomes, including the accessibility and effectiveness of flood early warning systems (EWS), preemptive adaptation measures taken to mitigate flood impacts, physical health outcomes during the flood season, economic well-being during and after the floods, and post-flood adaptation measures.

Impact Opportunity

This research, led by Dr. Jagnani and Dr. Pande, with data collection support from IDinsight at the end of the 2023 floods season, aims to inform policy and program execution to enhance flood management in Bihar, ultimately contributing to the state’s improved preparedness for future floods. Our role focused on providing high-quality data to inform recommendations by Inclusion Economics India Centre (IEIC). 

Our Approach

IDinsight’s DataDelta team managed a large-scale survey of 5,582 households across 12 districts. We hired and trained 151 surveyors, leveraging custom technology to optimize field operations and ensure high-quality data collection.

Our SurveyStream software streamlined the survey process, integrating with SurveyCTO to enhance the monitoring and management of the field team. Additionally, we employed SurveySparrow, an innovative tool that optimizes surveyor assignments and minimizes travel costs. Previously, field supervisors manually assigned targets; now, the software automatically creates efficient routes, saving supervisors’ time and reducing surveyors’ travel.

Despite the diverse terrain—which often required traveling long distances, crossing rivers by boat, and navigating difficult landscapes—we maintained rigorous standards. Surveyors received comprehensive training and detailed procedural guides, and their submissions underwent continuous data quality checks. Ultimately, we completed surveys with approximately 79% of the total sample respondents, demonstrating our team’s adaptability and technological innovation.

The Results

Note: This evaluation was led by IEIC-affiliated researchers; IDinsight collected endline data in 2023 only.

Early results from the ongoing evaluation found that the early warning systems increased accessibility to flood warnings, leading to improvements in physical health. Furthermore, households with access to community-disseminated flood alerts who faced the most severe flooding had a score that was 0.16 standard deviations higher on indices measuring economic well-being compared to households that were similarly affected by flooding without access to volunteers.

Alert accessibility:

The early warning system demonstrated significant improvements in alert delivery and accuracy. Households in areas with volunteers gain access to more accurate, timely, and informative alerts: In 2022, households in volunteer communities received 2.40 more alerts (a 294% increase), were 26% more likely to receive any alerts, 45% more likely to receive alerts before water reached their area, and 56% more likely to say they trust the alerts completely. In 2023, treatment communities received 3.52 more alerts on average (a 503% increase), were 71% more likely to receive any alert, 83% more likely to receive an alert before water reached their area, and 91% more likely to say they trust alerts completely.

Health impacts:

Households receiving the most alerts exhibited markedly improved proactive behaviors and physical health. Quantitatively, these households scored higher on indices measuring adaptive preparedness and health resilience. Overall, treatment households incurred roughly INR 6,890 less on illness treatment (a 35.51% decrease) and INR 1,730 less on injury treatment (a 8.92% decrease).

“Researchers estimate that for every $1 spent on the program, the intervention would save households 12.14 to 36.43 USD in medical costs during a severe flooding season.”

Economic well-being:

With better preparation and improved health outcomes, households with access to community-disseminated flood alerts who faced the most severe flooding were also less likely to engage in reactive adaptive responses after flooding. Specifically, households in volunteer communities were less likely to have to take out money from savings, mortgage assets, reduce consumption of essential items, and take on new loans compared to control households.These early results suggest that access to early warning systems were highly effective at reducing household losses from floods.