Alleviating poverty with data

Time has come to use Big Data on a massive scale

600 million people worldwide now risk falling into poverty according to a recent Oxfam study and three decades of development progress could be lost in some places in sub-Saharan Africa, the Middle East and North Africa.

Can this “stepping back” be an opportunity somehow to accelerate the fight against poverty ? Can governments seize this momentum to promote evidence-based public policy by using Big Data on a massive scale to improve health, education and welfare services for the poorest ?

I’m a data enthusiast, not a data expert. In the field of data science and “big data” for international development, Dr Joshua Blumenstock, Director of the Data-intensive development lab is one of the world's leading scholars applying machine learning, statistics and large-scale analytics to transform our approaches to development. A believer in fighting poverty with data who also reminds us don't forget people in the use of big data for development.

In Morocco, joblessness and a dependency on government welfare is exacerbated as in many countries. Paradoxically, obligatory containment has made the poor even more visible and vulnerable.

Current data is incomplete and outdated but estimates show that some 9 million Moroccans are considered poor or at risk of poverty. There could be up to 2,5 million informal workers and 5 million “Not in Education, Employment or Training” (NEETs). Without universal social security protection they rely on common solidarity, their resourcefulness and governmental support. 

Morocco’s proactive response to the pandemic did generate a new wave of public trust, hence opening a unique window of opportunity to unleash the power of Big Data to better target those most in need. The point here is that governments must use those tools not just in times of crisis but in the long run. We need just a bit of investment and foresight as Jeffrey D. Sachs pointed out back in 2015 when writing about data for developement.

“Why should the financial services industry, where mere dollars are at stake, be using more advanced technologies than the aid industry, where human life is at stake”

S. Mullainathan, in the New York Times - Satellite images can pinpoint poverty where surveys can't (2016)

Satellite photographs for instance can provide “nightlights” data. Studies have shown correlations between nightlight luminosity and economic productivity. With higher resolution, it is even possible to differentiate between poor and ultrapoor regions. Morocco’s new Mohammed VI-B Satellite put into orbit in 2018 can actually gather such data to create real-time “poverty maps”.

Data from the internet and social media can provide useful measures for economic activity, although those methods are less applicable for remote and rural areas with limited access to internet infrastructure and smartphones. This is why conventional data sets will remain essential, as well as to cross-check and validate data collected.

Other more sophisticated techniques can also infer socioeconomic conditions from mobile phone usage with machine learning algorithms. According to Dr Joshua Blumenstock, aggregating these individual metrics can provide measures as accurate as classic household surveys. 

Big data used to complement survey-based indicators could tremendously improve national statistics. Eligible recipients would be identified with less costs and humain constraints, with almost real-time estimates of poverty. In the end this can ensure that resources get to those most vulnerable. In turn, program monitoring and evaluation could take a whole new dimension. 

By pairing data scientists with classic face-to-face or classic household survey and “on the ground” people (local communities, public authorities, development experts, civil society and private actors including startups), we can develop social dashboards to monitor most vulnerable groups and event start introducing so-called “nudges”.

Take the question of cash transfers. In their paper on "Conditional cash transfers for education in Morocco", researchers Najy Benhassine, Pascaline Dupas, and Esther Duflo (Nobel Prize winner in Economics in 2019), concluded that a relatively cheap “nudge” (in the form of unconditional but labeled transfers) can improve school attendance. 

Gathering nationwide Big Data on education, health, welfare, as well as employment, could dramatically and permanently improve government's strategies. Data is a public good. It now must rise up on policy agendas.

When data can help policy-makers design better policies at such a large-scale, there is no excuse for looking away or not trying.

-------------------------------

Some useful ressources

> Google and Facebook are funding an African Master’s of Machine Intelligence: a one-year intensive programme that has launched its third cohort this year (see www.nexteinstein.org/).

> Data Science Africa, a conference conceived by African researchers.

> World Bank Open Learning Campus : Big data in action for development