BPS Explains Indonesia's Poorest Definition, Regional Differences

Indonesia's Central Statistics Agency (BPS) has clarified its methodology for calculating 'deciles,' which categorize the socio-economic welfare of the population into ten levels, emphasizing that regional characteristics are taken into account. Consequently, the living standards of the poorest in Jakarta (decile 1) may not necessarily be the same as those of the poorest in Papua. Setia Pramana, Director of Methodology for Statistics and Data Science at BPS, stated that deciles are established to 'rank' the level of public welfare. For the ranking criteria, household expenditure is primarily used due to the difficulty in obtaining income data. 'Intuitively, we have a ranking system. Deciles are a ranking. What needs to be ranked? There must be something to be ranked,' he said during a discussion event on 'Methodology for Determining Deciles and Measuring Poverty in Indonesia' at Politeknik Statistika STIS, East Jakarta, on Friday (11/9/2026). 'One of the things ranked is expenditure. Why not income? Because we don't have income data, so we use expenditure,' he explained. In this context, he mentioned that household expenditure figures utilize data on individual, household, and settlement characteristics from the National Socio-Economic Survey (Susenas) as training data in statistical modeling. From this, BPS builds mathematical equation models to predict the expenditure values of other families. 'From the data we obtained, we use information or characteristics, in statistical terms or model language, an equation that we can use to say, 'Oh, someone with this specific characteristic has expenditure that might be the same.' That's why it's considered an estimate,' he added. Only then are the estimated per capita expenditures used to compile the order or ranking of all families in stages. 'From the prediction, we then use from the approximately 90 million plus [households] that do not have expenditure information, we use predictor variables, indicators, such as household characteristics, housing conditions, and assets, etc., to estimate their per capita expenditure,' Setia explained. Despite this, Setia emphasized that the calculation and modeling of deciles are not directly averaged nationally but are adjusted according to the geographical conditions and socio-economic characteristics in each region. 'Regional characteristics will definitely vary between regions. In Jakarta, we are now living in our bubble. We might feel poor in our area. Papua could be much poorer, for example,' he stated. Therefore, BPS conducts specific modeling for each region. This means the welfare level of decile 1 in one region can be different from that in another. For instance, the expenditure level of residents in decile 1 could differ from that of residents in decile 1 from the Papua region, as exemplified by Setia. 'It's just the ranking. We rank at the regency/city level. Then, from the regency ranking, we move to the provincial ranking, and up to the national ranking. That's how it is,' he explained. Regarding the potential for prediction errors in decile ranking, Setia admitted that academically and statistically, such errors can occur, but BPS is continuously making improvements. 'When we talk about data, about models, there are indeed errors. The first error comes from the data itself. Whatever data we provide, whether the model we use is good or not, if the data does not match the field conditions, it's wrong,' he asserted.