Economists built a new statistical tool to measure what rich and poor households actually spend — with messy real-world data

A team of researchers developed a method to estimate spending patterns across income levels when data is incomplete, contradictory, or biased — a common problem in household survey work. The technique combines two existing statistical approaches in a way that lets economists measure real heterogeneity in how different income groups spend money on different goods.

Household expenditure surveys are one of the primary data sources governments and development agencies use to set policy — what to subsidize, who to tax, where poverty concentrates. Until now, the standard statistical methods for analyzing that data either couldn't handle the real messiness of what surveys actually contain, or they produced estimates that were mathematically inconsistent. This paper doesn't change the world tomorrow, but it removes a technical barrier that has forced researchers to either use worse methods or exclude real data from analysis. The applied example uses UK spending data to show the method works in practice, which matters because spending elasticity estimates directly inform tax and subsidy policy.