Using Tech to Help African Farmers Collect Index Insurance Payouts

Easy-to-use app helps fill “data voids” by cleaning and cross-verifying farmer drought reports

Aug 05 2022 | By Bernadette Young
Helping African Farmers Collect Index Insurance Payouts

New York, NY—August 5, 2022—Natural disasters such as floods, wildfires, and droughts affect farmers around the world. In Africa, in order to reduce risks and loss of income for smallholder farms, local governments have turned to index insurance, a type of insurance that pays out benefits according to a predetermined index. For example, if seasonal rainfall is high and hits the preset level, a payout is triggered. In theory, this system works well but the reality is that the climate data, like drought severity and timing, are so complex and the farmers’ reports so hard to collect that they create a “data void”--the index insurance payouts aren’t always as fair and timely as they should be.

Researchers from Columbia Engineering’s Department of Computer Science, together with colleagues from Columbia Climate School’s International Research Institute for Climate and Society’s Financial Instruments Sector Team (FIST), worked with farmers in Ethiopia, Zambia, and Senegal to create tools that accurately collect data and systems that are scalable and can easily be used by any organization or government.

One of the tools developed to help with this task is Reptile, an easy-to-use app that helps fill the “data voids” by cleaning and cross-verifying farmer drought reports. Created by Computer Science Associate Professor Eugene Wu and third-year PhD student Zachary Huang, Reptile provides a new way to visualize survey data collected from thousands of villages throughout a country, cross-check the data with sensor data (rain gauges, satellite feeds), and identify data errors that lead to inconsistencies.

Huang presented their NSF-funded research on June 15 at the ACM SIGMOD/PODS International Conference on Management of Data, the leading forum for data management researchers.

Existing data cleaning tools work by identifying incorrect data and repairing the dataset. However, the FIST team, led by Daniel Osgood, wanted a way to find the regions where the rainfall data disagrees with farmer experiences so that they could decide what interventions are most appropriate. With Reptile, they can visualize the survey data at the national level and cross-check it with external climate and rainfall data. When they spot an inconsistency, they can incrementally “zoom in” on the specific region, then district, then village that is most responsible. Then they can decide whether to re-interview those farmers, visit the villages to collect new data, or take another course of action. The data from Reptile has helped insure millions of farmers in Africa and other parts of the world.

The group, along with Computer Science Assistant Professor Lydia Chilton, is working on an “open insurance tool kit” so any government or organization can deploy fair and transparent disaster insurance to help even more farmers around the world.

“We are excited and grateful to work with Dan Osgood’s FIST team because the data problems they face are completely different from the ones computer science researchers traditionally focus on,” said Wu. “It’s amazing to see our research directly impact millions of lives in a positive way.”

Media contact:

Holly Evarts, Director of Strategic Communications and Media Relations
347-453-7408 (c) | 212-854-3206 (o) | [email protected]

About the Study:

Journal:

The study is titled “Reptile: Aggregation-level Explanations for Hierarchical Data”

Authors are: Zezhou Huang, Eugene Wu

The study was supported by NSF 1845638, 1740305, 2008295, 2106197, 2103794, Columbia SIRS.

The authors declare no financial or other conflicts of interest.