There is a lack of high-resolution data on economic activity and related development outcomes in contexts where it is too difficult or too costly to conduct traditional surveys. The intent of this project is to develop a set of algorithms and techniques to predict several important aspects of individual and community economic behavior, including: wealth, income and expenditures, the geographic distribution of poverty and extreme poverty, and changes in these indicators over time. While research in this area has so far been limited, this approach has the potential to transform observational measurement strategies available to policymakers, businesses and social scientists.
Project Vision and Strategy
Under the leadership of Joshua Blumenstock, the team will develop innovative methods for estimating a range of development indicators continuously across time and space, by combining traditional face-to-face and phone-based surveys with large-scale data from telecommunications networks. These techniques will be deployed and calibrated in Afghanistan, a uniquely challenging “real world” context that is a priority focus of US and international policymakers.
In 2016, the baseline survey for the study was completed. The team is currently in the process of data cleaning and conducting an initial analysis of survey data. Bi-weekly phone surveys continue, with 9 rounds now complete and an attrition management plan is in place. Preparations are currently being made for delivery of small cash transfers.
- Professor Joshua Blumenstock, School of Information, University of Washington