Year of Publication



Martin School of Public Policy and Administration

Executive Summary

Poverty, while an intuitive concept, proves harder to specify and measure, especially over time. Foster, Greer and Thorebecke (1984) introduced an axiomatically sound, and commonly used poverty index Pα that is later developed by Foster (2009) to include time spent in poverty. I use their dual-cutoff spells approach to determine the occurrence, depth, and severity of poverty in a longitudinal dataset.

Using the Panel Survey of Income Dynamics over the timeframe 1990-2010, high-risk populations are explored further, especially African Americans, unmarried females, and those without a high school diploma. I explore the proportions of poverty incidence to find head of households without a diploma comprise a decreasing share over time (true of both the general public and the poor). The average years of education has steadily risen for both the poor and nonpoor.

Of those in poverty, African Americans and single females are disproportionally represented as are individuals who have children. These predictors generally correspond to deep poverty, although duration alters the portrait of severity. Single females have lower duration adjusted percentage shortfalls (gap) and severity (gap2 ) than their counterparts. A very small portion of the poor population is in poverty more than 90% of the time. When duration is considered, having children corresponds to the deepest, chronic poverty, followed by African American.

My results are limited mostly by the measurement of resources and cost of living adjustments. Recommendations include changing the source(s) of income and leveraging the nongovernmental/private industry ability to target programs to subpopulations in the most need.