Inferring the income level of individuals is important for public policy as it can inform the targeting of social protection programmes, among other policy areas. The data sets generated to this end are traditionally produced in a resource-intensive way and are often outdated shortly after the point of production.
In this study, Pulse Lab Jakarta analyzed mobility signatures in financial transaction records, namely in debit card transactions, to infer the income level of people in the capital and major metropolitan areas of a country. To build the classification model, the Lab used basic demographic information, such as gender and age, together with mobility signatures in debit card transactions to classify users into lower and higher economic level groups. The accuracy of the model can reach up to 78.29 %.
The study suggests that basic demographic information can be used together with data from financial transactions to understand the income level of a population in near real-time. To improve the model and obtain a more detailed classification of high-income individuals, other data sources like credit card transactions and location characteristics could be analyzed.