Mathematicians have used machine learning to develop a new model for measuring poverty across countries that refutes old notions of a fixed “poverty line.” The results of the work are reported by the journal Nature Communications.
A study by researchers at the University of Aston suggests that conventional views of poverty are outdated. The fact is that they pay too much attention to subjective ideas about basic human needs. Experts fail to grasp the complexity of how people use their income.
In their new study, scientists say their new model, which uses computer algorithms to synthesize massive amounts of spending and economic data, could help governments worldwide predict future poverty levels and plan measures to mitigate the problem.
“No one has ever used machine learning to decode multidimensional poverty before,” said lead researcher Dr. Amit Chattopadhyay of the University of Aston’s College of Engineering and Physical Sciences. “It completely changes the way people should look at poverty.”
The established poverty measures aim to define a monetary threshold below which a person or household is defined as “poor.” The origins of these definitions are in the 19th and early 20th centuries.
The World Bank currently sets the international poverty line at $ 1.90 a day, with about 10% of the world’s population – about 700 million people – living on less.
In the new study, researchers analyzed data from India over 30 years, dividing spending into three broad categories: “staple foods,” such as cereals, “other foods,” including meat, and “non-food items,” which cover other costs such as housing, and travel costs. The model applies to any country.
By recognizing the interdependencies between the three categories – an increase in spending in one area usually means a decrease in spending in another – allows for a more holistic measure of poverty that can be adapted to individual country circumstances. Researchers combined income, asset, and commodity datasets from the World Bank and other sources to create a mathematical model that was able to not only accurately predict past poverty levels in both India and the United States but also predict future levels based on certain economic assumptions.
Considering the elasticity of supply and demand in the market, the model revises the number of people traditionally considered “poor” into a more practical “middle class.” It can be scaled to reflect conditions in a country’s sub-regions or even scaled down to a single city or area depending on the data available.
“The current perception of poverty is very subjective because ‘poverty’ will mean different things in different countries and regions,” added Dr. Chattopadhyay. “With this model, we finally have a multidimensional poverty index that reflects the real experience of people, wherever they live and is largely independent of the social class to which they are believed to belong.”