How COVID-19 Is Changing the World: A Statistical Perspective Volume 3
The granularity of poverty statistics—the scale or level of detail in the data—can have a significant impact on the effectiveness of public policy, particularly for targeting areas that need immediate intervention. Conventionally, poverty statistics are compiled using data from surveys of household income, expenditure, or living standards. However, sample sizes of these surveys are rarely large enough to provide reliable estimates of poverty at granular levels and increasing sample sizes can be costly.
In partnership with national statistics offices in the Philippines and Thailand and other development partners, ADB statisticians applied computer vision techniques and machine-learning algorithms to demonstrate increased granularity of poverty estimation using artificial intelligence. They produced poverty maps by training a computer vision algorithm to spot specific features from daytime satellite images to predict the level of economic activity. Since satellite imagery is available for granular areas, this method can produce poverty maps at granular levels too.
A COVID-19 emergency food program in the Philippines used this approach to deliver essential supplies to households most in need when the country went into lockdown to control the spread of the virus. The program used satellite-based poverty maps to target the most vulnerable households and maximize resources, illustrating the benefits of using high-quality data generated by this novel approach.
The report provides a snapshot of how COVID-19 is affecting different aspects of public and private life. It is published by the Committee for the Coordination of Statistical Activities.
- Economic Impact
- Environmental Impact
- Social Impact
- Regional Impact
- Statistical Impact