Authors: David Folch*, Florida State University, Matthew Laird, Florida State University
Topics: Population Geography, Quantitative Methods
Keywords: poverty, clustering, microdata, demographics
Session Type: Paper
Start / End Time: 3:05 PM / 4:45 PM
Room: Forum Room, Omni, West
Presentation File: No File Uploaded
The U.S. official poverty measure is simply a function of household composition, income and a food budget developed in the 1960s. In contrast, more and more countries are adopting multidimensional measures. As measures of poverty are integral for the accurate targeting of anti-poverty programs and the benchmarks by which we measure their success, considerable effort should be taken to align them with our modern understanding of poverty.
To capture the multidimensional nature of poverty, we use data from the Public Use Microdata Sample (PUMS) to combine individuals’ income, health, education, employment, community, etc. into a single measurement framework. Specifically, we employ a self-organizing map (SOM) to identify underlying trends between these contributing factors and provide a fine scale classification for each individual. We then partition the SOM to generate multidimensional clusters, which provides information on the shared characteristics of individuals across the nation.
The contribution of this study is threefold. First, it improves the mathematical grounding of poverty measurement by identifying the poor through statistical patterns rather than whether a person falls below an arbitrary threshold. Second, this study adds much needed nuance to our measure of poverty in two important ways: (1) it allows for the development of multiple, concurrent definitions of poverty that reflect how individuals experience deprivation in different ways, and (2) it highlights the interaction between the determinants of poverty. Finally, this study highlights how poverty varies across the U.S. by providing counts on each poverty type in each location.