Poverty is traditionally estimated from national household surveys which are (i) expensive (~USD 1M), (ii) labour intensive and (iii) time-consuming (~12 to 20 months to complete). Consequently, poverty maps are generally available at low frequency (every 5 to 10+ years) and only representative at coarse granularity. For instance, between 2002 and 2011, among the 155 countries for which the World Bank monitors poverty data, 57 have only one or zero poverty data point available (Serajuddin, 2015).​

On the other hand, non-traditional data sources such as mobile phone data (Blumenstock et al., 2015), satellite data (Jean et al., 2016, Engstrom et al., 2014) and social media data have demonstrated great potential for estimating poverty at low cost, more frequently and at high resolution. A few studies have also explored the combination of different data sources (in particular, mobile phone and satellite data) for poverty mapping, showing improved accuracy using complementary data sources (Njuguna et al., 2017, Steele et al., 2017 and Pokhriyal et al., 2017).​ However, there are still significant technical (bias, accuracy), practical (training data availability, data access) and ethical (individual and group privacy) challenges preventing the successful operationalization of these methods.

The goal of this workshop is to pool knowledge, data and resources on how to best use satellite, mobile phone, social media, and other non-traditional data to map poverty in near real time, at high spatial resolution, and with demographic disaggregation – and to do this in a manner that’s of operational value to National Statistics Offices, UN agencies and other relevant entities.