Every three years, Indonesia holds a national household survey (repeated cross section) that collects data on wealth of household characteristics and expenditure. The survey is called the Susenas. The gathered expenditure data form the basis of welfare statistics, including those on poverty. The expenditure data however is costly to collect. The full expenditure module requires households to answer around 600 questions on their expenditure, a time consuming task, conducted by goverment workers.
In the Econometric Game 2014, the participants had the objective to predict poverty statistics using variables that are easier to collect. There variables were for example background variables for a household and aggregated consumption data per household. By creating and correctly estimating a fitting model for this, it is in theoreticalley possible to create proverty statistics in a less expensive and faster way. With such a model less money is needed for collecting and more money is available for fighting poverty.
1. University of Copenhagen
2. University of Bristol
3. Aarhus University and University of Amsterdam