Mapping Disconnection Among American Youth

Photo by fizkes/Adobe Stock

At any point in time, about one out of every seven young people (aged 18 to 24) across the United States neither works nor participates in education or training. These disconnected youth (sometimes called opportunity youth) often lack a path to economic success.^[1] The costs of disconnection are high, falling both on the young people and on society as a whole (Belfield, Levin, and Rosen, 2012).

Youth Disconnection Varies Across Individuals and Across Communities

Rates of disconnection vary by race/ethnicity, veteran status, and disability status (Wenger and Bonds, 2025; Lewis et al., 2024). Young men who experience school suspensions face increased risk of later disconnection, as do young women who experience early pregnancy (Bonds and Wenger, 2025). Community factors also matter, perhaps especially for the disconnection rates of young men. Areas where prime-aged men (aged 25–54) without college degrees are less likely to work or search for work also typically have high rates of youth disconnection (Wenger and Bonds, 2025).

Research across multiple disciplines shows that living in a poor household or a poor community has long-lasting, negative effects on children.^[2] Lack of resources affects many aspects of children's lives, including school quality, health outcomes, recreational opportunities, and, later in life, employment options. Poorer communities may suffer from geographic isolation, high levels of pollution, and high levels of violence; each of these also harms young people.

Research also shows that low levels of social capital and weak social networks are harmful to children and young adults.^[3] Social capital or social networks include resources in the community that can be accessed through relationships, such as information about job opportunities. While poorer communities often have weaker networks, the relationship is not one-to-one. This suggests that community factors may help to explain the overall levels of youth disconnection, even after accounting for the direct effects of poverty. If this is the case, specific community-level policies could improve outcomes for disconnected youth. We explore this idea below.

Access to college may also help to explain disconnection rates. For example, Acton, Cortes, and Morales (2024) found that access to a public two-year college explains student enrollment but that different students respond in different ways: Relatively advantaged students substitute the two-year college for a four-year school, whereas Black, Hispanic, and less advantaged students substitute the two-year college for nonenrollment. This suggests that communities with access to a college can be expected to have lower rates of youth disconnection.

The existing literature often examines individual children or households, modeling their outcomes as a function of both their own resources and those of their communities. In this report, we take a somewhat different approach by focusing on the relationship between the youth disconnection rate within a community and that community's resources, geography, and social networks.

Our primary data source is the American Community Survey (ACS), which we supplement with measures of persistent poverty, geography, and social networks. We define communities geographically at the county level, and we address the implications of this choice in the appendix. We identify counties that experience persistent poverty over at least three decades. We also consider the larger geographic area where each county is located. Finally, we assess each community's social networks using a measure based on social media data that captures the extent of connections among people from different socioeconomic backgrounds within a specific community. We refer to this as a measure of economic connectivity. The appendix includes additional detail on each of these measures.

These findings will interest policymakers or practitioners working at the state or local level to improve educational and employment opportunities for young people, as well as community leaders seeking to build stronger relationships within their communities to further economic development. Researchers working in related areas, as well as members of the general public who wish to better understand youth disconnection, may also be interested in our findings.

Smaller Communities, Those with Persistent Poverty, and Those with Weak Economic Connectivity Have the Highest Rates of Youth Disconnection

Figure 1 shows how the average rate of disconnection varies across different types of communities. The dotted vertical line shows the average county-level rate of disconnection across the United States.^[4]

Figure 1 shows that some types of communities have much higher rates of disconnection than others. Communities with small populations (less than 100,000) have higher rates of disconnection than those with larger populations, and communities in the Appalachian and Ozark regions, the rural Deep South, the rural Southwest, and on tribal lands have substantially higher rates of disconnection than other communities (each county in the United States falls into one of the eight geography types listed in Figure 1). Communities that are persistently poor have rates of disconnection that are more than 40 percent above the average rate. Finally, communities with low rates of economic connectivity—counties that fall in the bottom 25 percent for measured online ties between groups with high and low socioeconomic status (SES)—have rates of disconnection that are more than 30 percent above the average rate, whereas counties in the top 25 percent for economic connectivity have disconnection rates that are substantially lower than is typical. Counties lacking a community college have higher disconnection rates, but the difference between counties with and without community colleges is small (about 1 percentage point).

Of course, these measures are correlated. For example, persistent poverty occurs frequently among counties in the rural Deep South, and the persistently poor counties in this area also tend to have small populations. Economic connectivity is closely related to persistent poverty: Although persistent poverty is relatively uncommon—affecting about 13 percent of counties and about 6 percent of the U.S. population—counties with persistent poverty typically have quite low levels of economic ties between people on different parts of the socioeconomic spectrum.

To better understand the relationships between the factors shown in Figure 1, we estimated a regression model. The results indicate that four distinct measures—community size, community type, poverty status, and economic connectivity—help to explain the differences in youth disconnection rates across communities, and the relationships generally are similar to those shown in Figure 1.^[5] Notably, each of these measures separately contributes to youth disconnection.

To demonstrate how these measures explain youth disconnection, Figure 2 plots the expected levels of youth disconnection in two example communities. One community is small, experiences persistent poverty, and has low levels of ties across the socioeconomic spectrum. The second community has a much larger population (over 250,000), does not experience persistent poverty, and has a high level of ties across the socioeconomic spectrum. The model predicts the smaller, poorer community's youth disconnection rate to be about 70 percent higher than that of the larger, better-resourced community.

Mentoring and Other Community Programs Could Reduce Youth Disconnection

Youth disconnection costs individuals and communities. Prior research (cited in the opening paragraphs of this report) has found higher rates of disconnection among some groups and in some places (rural areas and areas where fewer men participate in the economy). Key risk factors at the community level include relatively small populations, areas with persistent poverty, isolated rural areas, and areas with weak economic ties. Although these factors are correlated, each adds independently to the risk of community-level youth disconnection. Previous research (again, cited in the opening paragraphs of this report) has documented the negative effects of poverty and place, and researchers also understand that social capital shapes outcomes. Measures of economic connectivity, however, have only recently become available. Evidence (here and elsewhere) shows that these measures predict young people's outcomes.

These findings suggest a specific policy approach: Programs that increase connections across socioeconomic groups within communities have the potential to lower rates of youth disconnection. Such programs could include mentoring programs focused on youth, as well as any community-based programs, activities, or clubs that serve to increase ties across the socioeconomic spectrum. Connecting young people to opportunities is a recognized goal in youth mentoring programs. But our results indicate that other community-based programs—including neighborhood associations, volunteer or service programs within the community, continuing education classes, cultural programs, faith-based groups, and leadership development programs—may also serve to decrease rates of youth disconnection. Our results provide an additional reason for supporting such programs, as well as a new measure of their potential impact within communities.

Appendix

This appendix includes complete regression results (in Table A.1), as well as additional details about the data used in our analyses. We do not interpret this model as causal; in other words, our model does not identify and separate all the specific causes of disconnection. However, the model does help in understanding the relationships between our different measures.

Data Sources and Descriptions of Key Variables

The ACS provides annual measures of demographic and economic characteristics at the household and individual levels. We used the 5-year ACS dataset covering 2019–2023; these data include about 5 percent of the population in total (some 3.3 million observations per year); with weights, the results are representative of the U.S. population. We used these data to calculate a measure of youth disconnection, based on the number of young people who are not in school, not working, and not enrolled in a training program, as discussed below.

County-level data on persistent poverty and typology (such as data on the rural Deep South and the Appalachian and Ozark regions) come from Benzow et al. (2023).^[6] A persistently poor county is one in which at least 20 percent of the population has lived in poverty for at least three decades after adjusting for students (whose presence can otherwise indicate a misleadingly high level of poverty). Benzow et al. (2023) discuss the origins of this measure. Using shared historical characteristics, Benzow et al. (2023) classified each county in the United States as part of one of the following regions: Appalachian/Ozark, Rural Deep South, Rural Southwest, Other Rural, Tribal, Urban High Black Share, Urban High Hispanic Share, and Urban High White or Asian American Pacific Islander share. We included these measures in our models.

Data on economic connectivity come from Opportunity Insights, and these data include estimated measures of connectivity across the socioeconomic spectrum based on social media data; see Chetty et al. (2022a, 2022b). This measure, based on Facebook data among adults aged 25 to 44 who actively use the platform, is calculated by doubling the share of high-SES friends found among low-SES individuals and then taking the average at the county level. High and low SES are defined as above and below, respectively, the median for a composite SES measure. Counties whose residents have more ties across the socioeconomic spectrum have higher levels of economic connectivity. This measure is not available in a repeated panel; rather, it is a static measure calculated as of 2018.

The ACS data are reported at the Public Use Microdata Area (PUMA) level, whereas the other sources of data are reported at the county level. PUMAs are geographic areas defined for the purpose of reporting and tabulating U.S. Census data; PUMAs do not overlap, and they cover the United States in its entirety. Each PUMA contains at least 100,000 people. To merge these data from different sources, we first matched PUMAs to counties based on the allocation of the population using a crosswalk available at Missouri Census Data Center (undated). Our final dataset is at the county level. We have data on 3,008 counties.

Tests of Alternative Measures

We explored other county-level measures. For example, we tested a measure of youth economic connectivity (also using data from Opportunity Insights) and measures of marginally connected youth (we used ACS data to identify young people who worked few hours or worked for very low wages). The model performed less well (in terms of prediction) when we included these alternative measures rather than the ones shown in Table A.1.

We also tested measures of pre-kindergarten enrollment, access to broadband, and community distress (also from Benzow et al., 2023), as well as measures of the number of community establishments. Community establishment data are described in Rupasingha, Goetz, and Freshwater (2006); these data are periodically updated and are available at Northeast Regional Center for Rural Development (undated). The community distress measure, in particular, is strongly correlated with youth disconnection, but in each case we found that the measures in Figure 1 fit the data better than these potential alternatives.

We explored several measures of access to postsecondary education; we formed these from National Center for Education Statistics data on postsecondary institutions (the Integrated Postsecondary Education Data System [IPEDS]; see National Center for Education Statistics, undated). We used these data to form indicators of postsecondary institutions in each county. These measures surely contain some error; as noted by Acton, Cortes, and Morales (2024), community colleges may have a single administrative unit but multiple locations. Aggregating these measures to the county level should help with, but we do not expect it to solve, this problem. We tested indicators for the presence of any postsecondary institution, a two-year public college ("community college"), or any institution classified as two-year or less. We used the IPEDS "sector" variable to make these distinctions. All three measures are correlated with county-level disconnection rates. We used the community college indicator because of Acton, Cortes, and Morales's (2024) findings indicating that community colleges appear to affect students' decisions on the extensive margin (i.e., those with less access to community colleges are less likely to attend any college). We do not view these results as causal; college locations reflect endogenous factors.

Limitations Include Measurement Challenges and Some Missing Data

This exploratory work has limitations. First, matching youth disconnection rates (measured in the ACS) to other community measures (which exist at the county level) introduces some measurement error. In sparsely populated areas, many counties make up a single PUMA, whereas in densely populated areas, a single county contains many PUMAs. Although this merge was necessary because we lacked reliable youth disconnection measures at the county level, it means that the youth disconnection measures are somewhat inaccurate both in the least- and the most-populated counties. We could have conducted the analysis at the PUMA level, but then the measures that exist at the county level (persistent poverty, typology, and economic connectivity) would require merging to the PUMA level. This approach would also introduce some measurement error. No guidance exists as to which method introduces smaller errors. We intend to explore the relative effects of these two approaches in future work. Finally, our data include information on the 50 states and the District of Columbia but not the U.S. territories.

A few counties do not appear in the typology data from Benzow et al. (2023) or in the data on economic connectivity from Chetty et al. (2016). These counties tend to have very small populations. Their exclusion decreases the number of counties in our dataset by less than 4 percent and decreases the population included in our dataset by less than 0.2 percent. After accounting for the missing information, we have complete data on 3,008 counties. A few counties merged or changed names during recent decades, and we coded these counties consistently to include them in our data.

In this report, we do not explore the individual characteristics or experiences that lead to disconnection. For example, we know that some experiences (suspensions for young men, early pregnancies for young women) predict later disconnection (see Bonds and Wenger, 2025).

Acknowledgments

We are grateful to Ben Master and Melanie Zaber of RAND for their careful reviews. We are grateful to James Torr and Babitha Balan for their assistance with editing and the publication process.

Notes

1. This report is the third in a series of publications describing disconnected youth, their circumstances, and their experiences, following Wenger and Bonds, 2025; and Bonds and Wenger, 2025. Return to content ⤴
2. See, among many, Wilson, 1987; Sampson, Morenoff, and Gannon-Rowley, 2002; Leventhal and Brooks-Gunn, 2000; Chetty, Hendren, and Katz, 2016; and Ludwig et al., 2012. Return to content ⤴
3. Coleman (1988) is a seminal piece in this area but also see, among many, Gowdy and Spencer (2021) and Duncan and Brooks-Gunn (1997). For measuring social capital at the community level, see Chetty et al., 2022a; and Chetty et al., 2022b. Pierre Bourdieu developed the idea of social capital; in his view, social capital is one mechanism through which individuals or families maintain power (Bourdieu, 1986). Robert Putnam presented a somewhat contrasting view, in which social capital (or at least one form of social capital, which he referred to as bridging social capital) is a community or public resource (Putnam, 2000). Therefore, a decrease in social capital would be expected to lead to negative outcomes across the community. Our definition in the main text hews more closely to Putnam's ideas than to Bourdieu's, and our results are broadly consistent with Putnam's view and generally accord with the findings of Chetty et al. (2022b). Return to content ⤴
4. The average county-level rate of youth disconnection is 16.4 percent; this unweighted average is slightly higher that the individual-level average of about 15 percent (Wenger and Bonds, 2025).Return to content ⤴
5. The differences between some of the typologies are smaller and statistically insignificant in the regression results; see the appendix. Return to content ⤴
6. Benzow et al. (2023) draw on data from the Economic Innovation Group's Distressed Communities Index (Economic Innovation Group, undated). However, findings and opinions expressed in this report are solely those of the authors and not necessarily those of the Economic Innovation Group. The Economic Innovation Group does not guarantee the accuracy or reliability of, or necessarily agree with, the information provided in this report. Return to content ⤴

References

• Acton, Riley K., Kalena Cortes, and Camila Morales, "Distance to Opportunity: Higher Education Deserts and College Enrollment Choices," National Bureau of Economic Research, NBER Working Paper 33085, October 2024. As of April 4, 2026: https://www.nber.org/papers/w33085
• Belfield, Clive R., Henry M. Levin, and Rachel Rosen, The Economic Value of Opportunity Youth, Corporation for National and Community Service, January 2012.
• Benzow, August, Kenan Fikri, Joanne Kim, and Daniel Newman, Advancing Economic Development in Persistent-Poverty Communities, Economic Innovation Group, 2023.
• Bonds, Stephanie, and Jennie W. Wenger, What Predicts Disconnection Among American Youth? RAND Corporation, PE-A4207-2, October 2025. As of February 10, 2026: https://www.rand.org/pubs/perspectives/PEA4207-2.html
• Bourdieu, Pierre, "The Forms of Capital," in J. Richardson, ed., Handbook of Theory and Research for the Sociology of Education, Greenwood, 1986.
• Chetty, Raj, Nathaniel Hendren, and Lawrence F. Katz, "The Effects of Exposure to Better Neighborhoods on Children: New Evidence from the Moving to Opportunity Experiment," American Economic Review, Vol. 106, No. 4, 2016.
• Chetty, Raj, Matthew O. Jackson, Theresa Kuchler, Johannes Stroebel, Nathaniel Hendren, Robert B. Fluegge, Sara Gong, Federico Gonzalez, Armelle Grondin, Matthew Jacob, et al., "Social Capital I: Measurement and Associations with Economic Mobility," Nature, Vol. 608, 2022a.
• Chetty, Raj, Matthew O. Jackson, Theresa Kuchler, Johannes Stroebel, Nathaniel Hendren, Robert B. Fluegge, Sara Gong, Federico Gonzalez, Armelle Grondin, Matthew Jacob, et al., "Social Capital II: Determinants of Economic Connectedness," Nature, Vol. 608, 2022b.
• Coleman, James S., "Social Capital in the Creation of Human Capital," American Journal of Sociology, Vol. 94, supplement, 1988.
• Duncan, Gregory J., and Jeanne Brooks-Gunn, eds., Neighborhood Poverty: Context and Consequences for Children, Russell Sage Foundation, 1997.
• Economic Innovation Group, "Distressed Community Index," dataset, undated.
• Gowdy, Grace, and Renee Spencer, "It's Who You Know That Matters: Identifying Which Type of Informal Mentor Is Most Likely to Promote Economic Mobility for Vulnerable Youth," Journal of Primary Prevention, Vol. 42, 2021.
• Leventhal, Tama, and Jeanne Brooks-Gunn, "The Neighborhoods They Live in: The Effects of Neighborhood Residence on Child and Adolescent Outcomes," Psychological Bulletin, Vol. 126, No. 2, 2000.
• Lewis, Kristen, Alex Powers, Cara Wohnsigl, Kate Harvey, and Tara Shawa, Broad Recovery, Persistent Inequity: Youth Disconnection in America, Social Science Research Council, 2024.
• Ludwig, Jens, Greg J. Duncan, Lisa A. Gennetian, Lawrence F. Katz, Ronald C. Kessler, Jeffrey R. Kling, and Lisa Sanbonmatsu, "Neighborhood Effects on the Long-Term Well-Being of Low-Income Adults," Science, Vol. 337, No. 6101, 2012.
• Missouri Census Data Center, homepage, undated. As of May 5, 2026: https://mcdc.missouri.edu/
• National Center for Education Statistics, "Integrated Postsecondary Education Data System," webpage, undated, As of May 5, 2026: https://nces.ed.gov/ipeds
• Northeast Regional Center for Rural Development, "County-Level Measure of Social Capital," webpage, undated. As of April 23, 2026: https://nercrd.psu.edu/data-resources/county-level-measure-of-social-capital/
• Papke, Leslie E., and Jeffrey M. Wooldridge, "Econometric Methods for Fractional Response Variables with an Application to 401(k) Plan Participation Rates," Journal of Applied Econometrics, Vol. 11, No. 6, 1996.
• Putnam, Robert, Bowling Alone: The Collapse and Revival of American Community, Simon and Schuster, 2000.
• Rupasingha, Anil, Stephan J. Goetz, and David Freshwater, "The Production of Social Capital in US Counties," Journal of Socio-Economics, Vol. 35, 2006, with updates.
• Sampson, Robert J., Jeffrey D. Morenoff, and Thomas Gannon-Rowley, "Assessing 'Neighborhood Effects': Social Processes and New Directions in Research," Annual Review of Sociology, Vol. 28, No. 1, 2002.
• U.S. Census Bureau, "American Community Survey 5-Year Data (2009–2024)," webpage, last updated January 29, 2026. As of May 5, 2026: https://www.census.gov/data/developers/data-sets/acs-5year.html
• Wenger, Jennie W., and Stephanie Bonds, Understanding Disconnection Among American Youth, RAND Corporation, PE-A4207-1, October 2025. As of February 10, 2026: https://www.rand.org/pubs/perspectives/PEA4207-1.html
• Wilson, William Julius, The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy, University of Chicago Press, 1987.

Topics

• Community Health and Well-Being
• Employment and Unemployment
• Life Skills
• School-to-Work Transitions
• United States
• Young Adults