Advancing Economic Development in Persistent-Poverty Communities


Below we provide an essential overview of the concepts and methods underlying this data interactive, but we encourage readers to read the report, “Advancing Economic Development in Persistent-Poverty Communities,” for a full exploration of the project’s methods and analysis.

Defining persistent poverty

NOTE: EIG’s definition of persistent poverty differs from that used by federal agencies and the Economic Development Administration and should not be relied upon to determine program eligibility.

Cutoff and time periods used

“Persistent-poverty areas” have been part of the country’s policy lexicon since at least the 1980s. The American Recovery and Reinvestment Act of 2009 contained what has become known as the “10-20-30” provision, which required 10 percent of funds from specific development programs to go to persistently poor counties, defined as those that have had a poverty rate of 20 percent or higher for at least 30 years. In this report, we employ a definition of persistent poverty that follows this precedent for counties and census tracts. Unlike the approach taken by some federal agencies, the dataset created for this report does not round poverty rates or take into account margins of errors but maintains the 20 percent cutoff.


Poverty rates for 1990, 2000, 2010, and 2019 are used to cover a 30-year time period. Decennial Census data is used for 1990 and 2000 and American Community Survey 5-year estimates for 2010 and 2019 (2006-2010 and 2015-2019, respectively). Importantly, geographies are allowed to fall below 20 percent in either 2000 or 2010 and still qualify as persistently poor so long as they had a poverty rate of 20 percent or higher in the start and end time periods. This choice allowed communities that experienced a temporary but unsustained reduction in poverty to still qualify as persistently poor since their escape was not durable.

Longitudinal dataset

Census tract boundaries change with every decennial Census. Consequently, data that used 1990 and 2000 tract boundaries had to be reallocated to 2010 tract boundaries in order to compare the same geographies over time. Interpolation weights from the Longitudinal Tract Database (LTDB) were used to accomplish this.

Student poverty

College campuses and tracts adjacent to them (in addition to a handful of counties with large college towns) pose a special challenge for defining persistent poverty. The Census Bureau does not include students living in dormitories (or anyone living in group quarters, for that matter, which includes prisons and military bases) in the population for whom poverty status is determined, but students who live off campus are considered in calculating local poverty rates. Students are typically lower income than the general population, and their presence can inflate the poverty rate of otherwise well-off areas.

Although no federal agency attempts to control for the effect of large student populations on local poverty rates, this analysis pilots a methodology designed to correct for it and bolster the integrity of the persistent-poverty universe. We employed a two-step process to exclude areas with large student populations from being considered persistently poor:

  1. Areas where the student share of the population [1] exceeds 80 percent were excluded entirely, regardless of poverty status. This only affected census tracts and almost exclusively led to the removal of college campuses.
  2. Areas where the student share of the population falls between 20 and 80 percent were given an adjusted poverty rate for 2019 that excluded students. [2] As with the standard poverty rate, areas had to have a non-student poverty rate of 20 percent or higher to qualify as persistently poor. (No adjustment was made for areas where the student share of the population fell below 20 percent).

Persistent-poverty tract groups

To construct contiguous groups of persistent-poverty census tracts, we identified all persistent-poverty tracts that shared a border with at least one other persistent-poverty tract. We set a rule that at least four persistent poverty tracts must be contiguous to form a group. Although there is no clear guideline that dictates where the minimum number of persistently poor tracts that constitute a group must be set, we ultimately opted for this cutoff because it established a higher-need geography compared to smaller groups (the more persistent poverty tracts that are clustered together, the higher their average poverty rates, for example) while still including nearly four-fifths of all persistent-poverty tracts. Adjacent high-poverty tracts were added to the persistent poverty cores, as were any single tracts completely surrounded by them, regardless of their poverty rate.

These persistent-poverty tract groups (PPTGs) were formed to create an optimal geography that balances the narrow targeting offered by census tracts with the reality that economic development-related interventions—capacity building, fostering entrepreneurship, supporting anchor institutions, increasing access to capital, upgrading workforce skills, or investing in infrastructure, to name just a few—function at greater geographic scales.


The research establishes that persistent poverty communities clearly sort along lines of race, region, and rurality, reflecting the different histories of a people in a place. A typology was created based on the racial and ethnic composition of an area and a mix of other geographic criteria to aid in understanding both the variety of persistent poverty communities in the United States today, and what unites them.

Development assessment

The development assessment scores communities across 14 distinct metrics to identify strengths and weaknesses in the local economic foundations. Metrics were selected to capture critical inputs to development, such as human and social capital or infrastructure; the level of economic development or local economic performance; and affordability and inclusivity, or the extent to which lower-income individuals are participating in the economic development of an area. Six are considered people-based (i.e., descriptive of the local population), and eight are considered place-based (i.e., descriptive of the local economy, jurisdiction, or built environment). Each location’s performance on individual metrics is converted into a percentile based on its rank among its peers.


[1] Sourced from ACS table B14001: School Enrollment by Level of School for the Population 3 Years and Over: Share made up of undergraduate and graduate students.

[2] This adjusted poverty rate subtracts undergraduate and graduate students from the three-and-older poverty universe population and from that universe’s population living in poverty. Sourced from ACS table B14006: Poverty Status in the Past 12 Months by School Enrollment by Level of School for the Population 3 Years and Over. Tables can be cross-referenced, and students can be removed from the full poverty universe to produce similar results.

This report was prepared by the Economic Innovation Group using Federal funds under award ED21HDQ3120059 from the Economic Development Administration, U.S. Department of Commerce. The statements, findings, conclusions, and recommendations are those of the authors and do not necessarily reflect the views of the Economic Development Administration or the U.S. Department of Commerce.