Study reveals significant discrepancies in common approaches to measuring poverty

Commonly used methods for measuring poverty can lead to very different conclusions about who actually lives in poverty, according to a new study led by Stanford University. Based on household surveys in sub-Saharan Africa, the first analysis of its kind, published in Proceedings of the National Academy of Sciences, highlights the importance of accurately defining and measuring poverty.

Its findings could help inform how governments, nonprofits and international development agencies allocate resources and evaluate the effectiveness of poverty reduction policies around the world.

“They say you can’t manage what you don’t measure,” said the study’s lead author, Eric Lambin, the George and Setsuko Ishiyama Senior Professor at the Stanford Doerr School of Sustainability and a principal investigator at the Stanford Woods Institute for the Environment.

“In our study, we find that how one chooses to measure poverty can completely change the extent to which programs and policies are managed and reach vulnerable populations,” said the study’s lead author, Christine Pu, who holds a Ph.D. student in environmental engineering at the Stanford Doerr School of Sustainability.

Compare definitions

Governments around the world want to support households living in poverty, but it is not always easy to determine which households need help. For example, two American families of the same size could be classified as poor – and eligible for public support programs like food assistance and subsidized utilities – because their annual income is below the federal poverty line of $31,200 .

In reality, families may have dramatically different overall costs or assets. For example, one may own their house and two cars, while the other may rent their house and rely on public transportation.

The study examined four widely used approaches to measuring poverty. Each measure is based on different priorities ranging from declared assets, such as household appliances, to self-defined well-being milestones, such as the ability to send children to school. Working with colleagues from Ethiopia, Ghana and Uganda, the Stanford researchers surveyed 16,150 households.

Surprisingly, the research found virtually no agreement on how these approaches classified households by their poverty status. The lack of agreement persisted even among households ranked in the poorest 20% in terms of poverty.

Even after controlling for geographic variability, the study found low correlations between measurement approaches, indicating that the discrepancies were not simply due to regional differences. The differences in relative rankings were not small either. On average, households’ poverty rankings differed by 25 percentage points. In other words, a household ranked in the 25th percentile by one measure may be classified as the poorest household or as the median household by another measure.

“Organizations that adopt a measurement approach without thinking about how it fits their understanding of poverty are, at best, rolling the dice to create household classifications that work in accordance with their mission and goals,” write the researchers. “At worst, these organizations adopt methodologies that may be completely inappropriate for their poverty reduction goals.”

Choose wisely

A striking example of this conceptual misalignment is the wealth index of the U.S. government’s Demographic and Health Survey (DHS) program. The index was designed to explain disparities in health outcomes. However, it is widely used to represent the poverty status of a household.

This application can lead to counterintuitive classifications of households. For example, while most rural development scholars would view livestock ownership as a sign of family wealth, the DHS index lowers households’ rankings for each additional unit of livestock they own, because rural households that own Livestock generally have limited access to health services.

Given the widespread influence of the DHS wealth index, this measurement problem spreads and amplifies across many applications and decision-making processes. The problem is not unique to the DHS wealth index, but is emblematic of a broader problem inherent in many indexes and measurement tools.

Overall, the results suggest that the choice of measurement approach can lead to very different conclusions about who is eligible for poverty reduction programs and policies, and what outcomes these efforts achieve. The authors argue that organizations should carefully consider their definition of poverty and select measurement approaches that match their specific goals.