Key takeaways

• The depressed overall hires rate is a key driver of new labor market weakness for young college graduates, as it makes it harder for them to break into the labor market. This is true across industries, not just in those that disproportionately employ young college graduates—suggesting the culprit is not a structural change in the economy like AI but a labor market in which employers are hiring less and workers are holding onto the jobs they have.
• The information sector—posited to be more AI-exposed—has experienced recent job losses but employs only 2.3% of young college graduates.
• High-tech industries, which employ about 1 in 10 college workers, expanded at a historically rapid pace in the early 2020s but have shown signs of softening over the last three years.

The early 2020s labor market for young college graduates was strong. But, as we showed in this series’ first blog post, the Class of 2026 is graduating college into a labor market that has notably weakened in the past two years. A growing share of young college graduates are looking for jobs, but their employment rates have not kept pace—meaning unemployment is rising faster for young graduates than for the overall workforce. While their outcomes remain better than those of their noncollege counterparts, the uptick in unemployment has been a rising concern.

In this blog post, we delve deeper into the industries where young college graduates are likely to work,1 examining whether it has been relatively more difficult to secure employment in these fields as the labor market has weakened. Our analysis first examines employment changes, then turns to labor market flows, including hires and separations rates. We also scour the data for signs of contraction in the tech sector that may disproportionately affect the prospects for young college graduates as they enter the labor market.

In the third blog post in the series, we will examine the occupations where young college graduates work with particular attention to occupations that may have grown or shrunk, as well as to those most exposed to AI.

Young college graduates work in industries with strong growth in this business cycle

Over half of young college graduates work in private education and health services, professional and business services, or public-sector jobs. Figure A displays the share of employment in each industrial sector or sector grouping for young college graduates ages 22 to 27, all college graduates, and young workers without a four-year college degree. Industries in the figure appear in order of the share of young college graduates they employ, from largest to smallest.

The types of jobs where young college graduates work look similar to those of college graduates generally. Young workers without a college degree (i.e., noncollege) are far more likely to work in trade, transportation, and utilities; mining, construction, and manufacturing; or leisure and hospitality. All groups of workers are least likely to work in the information sector, closely watched for signs of AI-induced displacement.

Figure A

Figure B shows the change in employment in each sector between 2019 and 2026 and between 2023 and 2026, arranged in the same order as Figure A for comparison. The two fastest growing sectors since the last business cycle peak occurred in the two largest sectors for young college graduates: private education and health services and professional and business services.

Since the rollout of ChatGPT, many have looked at industries and occupations likely to be exposed to AI to see whether this has led to weaker job growth. Among the most closely watched of these industries is the information sector, which has seen an 8.5% employment decline since 2023.2 While these losses are notable—and especially relevant to understanding AI’s fingerprints on the labor market for young college-educated workers—it cannot be overemphasized just how small this sector is in the overall economy. Less than 2% of overall employment is in the information sector, including only 2.3% of young college graduates. Further, the sector saw a rapid employment expansion between 2019 and 2023—the employment loss between 2019 and today is just 2.0%.

Figure B

High-tech sectors have relatively more college graduates but haven’t experienced large AI-induced employment losses

In recent years, the Census Bureau has created an experimental data series on high-tech industries to better understand business dynamics. These include both manufacturing and service sector components of high tech.3 We translate their NAICS classification for high-tech industries into Census Industry Classifications used by the Current Population Survey (CPS) to determine the likelihood of young college graduates working in these sectors.

In the 2026 economy, about 5.6% of workers were employed in what the Census considers high-tech industries. Just about 1 in 10 (9.9%) of the college-educated workforce works in the tech industry. Young college graduates are similarly represented: 10.3% work in tech.

Overall, the Current Establishment Survey tells us that tech industry employment tracked changes in overall private employment in the prior business cycle (between 2007 and 2019) but expanded sharply in the early 2020s and has softened a bit in the last three years. Since 2023, the tech sector has fallen by 0.7%. While overall employment using CPS does show modest growth, neither shows large swings that suggest a large impact for young college graduates.4

Weak hires may be the biggest culprit to labor market weakness for young college graduates

The Job Openings and Labor Turnover Survey (JOLTS) can shed light on the question of whether entry-level workers—with or without college degrees—are facing a harder labor market to break into. While JOLTS doesn’t include demographic characteristics, it presents jobs openings as well as rates of hiring, layoffs, quits, and other separations. Today’s economy has substantially less churn than during the recovery from the pandemic, when millions of workers reentered the labor market after mass layoffs—many quit soon after as they searched for, and generally found, better opportunities.

Figure C shows the hires and separation rates. The lighter colors represent the monthly seasonally adjusted data for each series while the darker colors represent a 12-month moving average that provides a better overall picture of recent trends, smoothing out some data volatility. Over the last five years, the hires rate has steadily fallen and now sits at levels last seen in 2013 and 2014, when the labor market was still struggling to recover from the Great Recession.

The total separations rate includes quits, layoffs and discharges, and other separations.5 As with the hires rate, the separations rate has been declining over the last few years and now sits about where it was in 2014. Much of this is due to reductions in quits. Quits are higher when workers feel confident that they will find better job opportunities. Right now, workers are sitting tight, more so than any point in the past 10 years. Taken together, there is simply less churn in the labor market. But reduced churn is not inherently bad. If the frantic labor market of the early 2020s led to many workers and employers finding satisfactory matches, it could make sense that the following years would see less churn than normal. But for young workers looking to enter the job market, a reduction in hiring can make it harder to find a job.

Figure C

Table 1 breaks down the change in the hires and separations rate over the last three years, again using 12-month moving averages to smooth some volatility in the data. The industries are listed in order of the share of young college graduates they employ, similarly to Figure A. Overall, the hires rate fell 0.8 percentage points and the separations rate fell 0.6 percentage points between 2023 and 2026. The industries where young college graduates are more likely to work saw smaller reductions in both hires and quits than the overall. Industries where young workers without a college degree are more often found—over a quarter are in trade, transportation, and utilities—saw greater losses. Finally, leisure and hospitality, where young noncollege are more than twice as likely to work as young college graduates, saw the largest declines in hiring.

It does not appear that the industries where young college graduates tend to work are experiencing more weakness than other industries. Job gains are just as strong, if not stronger, and hiring hasn’t fallen as far in other industries. In short, there does not seem to be any profound structural change in the economy affecting the industry composition of employment—AI or anything else—that would easily explain the softening of the labor market for young college graduates in recent years. What it does appear to be is a harder labor market for young workers to break into when employers are less likely to hire and workers are more likely to sit tight in the job they have.

1. Throughout this blog post, we define young college graduates as people between the ages of 22 and 27 with only a four-year college degree. Unlike similar analyses of young workers, we do not exclude young college graduates that are currently enrolled in school, but the results here are robust either way. Unless otherwise noted, data for 2026 represent a 12-month average from April 2025 through March 2026 for the most up to date and reliable estimates, which removes seasonality and increases sample sizes.

2. ChatGPT was first introduced in November 2022 but took several months for more widespread usage.

3. High tech industries: Computer and Peripheral Equipment Manufacturing, Communications Equipment Manufacturing, Semiconductor and Other Electronic Component Manufacturing, Navigational, Measuring, Electromedical, and Control Instruments Manufacturing, Aerospace Product and Parts Manufacturing, Software Publishers, Data Processing, Hosting, and Related Services, Other Information Services, Architectural, Engineering, and Related Services, Computer Systems Design and Related Services, and Scientific Research and Development Services.

4.  It’s not unusual for the CPS and CES to display small differences in employment levels or trends considering nontrivial differences in their methodologies. 

5. Other separations include separations due to retirement, death, disability, and transfers to other locations of the same firm.

Since the second Trump administration swept into office in January 2025, it has undertaken a range of damaging and destabilizing actions that have weakened the economy, undermined workers, hurt businesses and consumers, and threatened core elements of our democracy. While Trump has targeted numerous Democratic-led states and cities, the Washington, D.C., region has faced acute and prolonged harms since day one. From the first set of executive actions signed on Inauguration Day, the Trump administration has attacked people and businesses in the capital region repeatedly and intensely. These initial actions announced the president’s dubious claims of authority to fire large segments of the federal workforce, eliminate long-standing federal agencies and programs, and begin a campaign of illegal and inhumane mass deportations.  

The Trump administration’s damaging actions have been enabled and abetted by Republican members of Congress. Their passage of H.R. 1, the bill that the White House has referred to as the “One Big Beautiful Bill Act” (OBBBA), amplifies the administration’s mass deportation agenda and shreds critical health care and food supports for lower-income families to finance tax cuts for the wealthy. This funding bill will only cause more pain in the years ahead for Washington, D.C.-area households and throughout the country.

Congress also passed a federal spending bill that constrained the District of Columbia’s ability to spend its own tax revenue (Koma 2025) and a resolution that may force the district to adopt local tax code changes that match the OBBBA, whether the city wants to or not—changes that will jeopardize hundreds of millions of dollars for city programs (D.C. Fiscal Policy Institute 2026).

In this report, we assess the early indicators of the damage of Trump’s actions and their effects on the Washington, D.C., regional economy, with particular attention to effects on workers and the labor market. We focus on this region due to its prominence as an early target of the Trump administration, in part due to its large federal workforce. Additionally, the district’s unique status as a non-state means that its leaders have far less legal authority to resist Trump’s interference than other target areas do.

Throughout this report, unless otherwise indicated, the data describe economic conditions for the Washington, D.C., metropolitan statistical area (MSA), which includes the District of Columbia, four nearby counties in Maryland, six cities and 11 counties in northern Virginia, and one county in West Virginia. We also refer to this region as the DMV (Washington, D.C.; Maryland; and Virginia). While we do not yet have the requisite data to fully and precisely document all the effects of the administration’s actions, we can see clear signals that the regional economy is already struggling, with more severe impacts likely to register in the data soon.

We then explore some of the factors that make other regions particularly vulnerable to significant economic harm from the Trump administration’s agenda. These include counties with large concentrations of federal workers, areas where federal transfer income (such as Medicaid and Social Security) makes up a significant portion of the region’s economic base, and places with significant immigrant populations. Though Trump has largely targeted prominent, Democratic-led areas, many of the regions most susceptible to the harmful economic consequences of the administration’s actions are rural counties, frequently represented in Congress by Republicans.

Trump’s actions in Washington, D.C., have led to reduced employment and rising unemployment

The clearest sign of the harm that the Trump administration’s actions have done to the Washington, D.C., regional economy is the substantial drop in the region’s employment rate. Based on EPI analysis of Current Population Survey data from the Bureau of Labor Statistics, from December 2024 to December 2025, the share of the regional working-age population with a job fell by 3.2 percentage points.1 As shown in Table 1, this compares with a decline of just 0.4 percentage points for the country over the same period. Among prime-age workers (those ages 25–54), the share employed in the DMV fell by 2.7 percentage points, compared with a decline of just 0.1 percentage points for the country overall.

This dramatic drop in regional employment is a direct result of the Trump administration’s relentless attacks on federal government workers, cuts to federal programs and agencies, and their cascading effects on connected regional industries. Prior to Trump’s taking office, federal employees made up 11.2% of the metro area’s total workforce (BLS-CES-SAE 2025).2 Between the end of 2024 and 2025, federal employment in the DMV region fell by more than 53,800 jobs (-14.2%) (BLS-CES-SAE 2026).3 These losses reverberated through the regional economy as affected households pulled back on spending, and many may have even opted to move, as data show the DMV region had the largest increase in home sale listings of any major metro last year (Brookings Institution 2026).

These significant cuts to federal employment, though highly damaging on their own, are only the first layer of the administration’s harm on the regional labor market. The DMV has a non-federal workforce of over three million people (BLS-CES-SAE 2026), many of whom work at firms that consult with, contract with, are funded by, or are otherwise connected to the government.4 The Trump administration has terminated thousands of grants to scientific research institutions (Kozlov, Tollefson, and Garisto 2026) and frozen or delayed funding for tens of thousands of nonprofit organizations, causing those targeted to limit operations or lay off staff (Tomasko et al. 2025). These cuts have also shrunk the funding pool for nonprofit groups, causing budget challenges even for those not previously receiving federal funding, as they must compete with groups previously funded through federal programs that are now scrambling to fill gaps with private support (Barrett 2025). The administration has also moved to cancel contracts with any company that maintains a commitment to DEI standards (Singh 2026). Although these cuts affect organizations everywhere, the DMV is disproportionately vulnerable to the economic harms of attacks on this sector as it has one of the highest concentrations of nonprofits in the country (Friesenhahn 2025). This is evident in the region’s slight dip (-0.3%) in private-sector employment from December 2024 to December 2025, a change from the consistent, albeit slowing, growth that had marked the years following the COVID-19 pandemic. At the national level, private-sector employment experienced slow but still positive change (0.5%) over the same period (BLS-CES-SAE 2026).5

The widespread impact of the administration’s actions can be seen in the breadth of employment declines across racial, ethnic, gender, and age groups in the region. As shown in Table 1, the employment rate fell by at least 2 percentage points for every demographic category of workers in the DMV. Notably, young workers under age 25 (-4.3 percentage points), workers age 55 and older (-3.3 percentage points), men (-3.5 percentage points), and Black workers (-5.9 percentage points) all experienced drops in their employment rates larger than the regional average. For older workers, the above-average decline likely reflects, at least in part, the firings and retirements of many federal employees, including many who had been near retirement age and opted into the so-called “Fork in the Road” deferred resignation program. For young workers, the administration’s funding and programmatic cuts directly reduced many traditional Beltway early-career opportunities (internships, fellowships), while weakness in the broader regional economy simultaneously forced area employers to pull back on entry-level positions.

Still, not all groups have been equally affected by Trump’s actions. As Table 1 shows, Black workers in the DMV region have suffered the largest employment declines, with the share employed falling by 5.9 percentage points in 2025. This is nearly triple the employment drop experienced by white workers (2.0 percentage points) in the region and, notably, more than seven times the employment drop of Black workers throughout the country overall (0.8 percentage points). Again, this is a direct consequence of the administration’s attacks on the federal workforce. Black workers have long tended to make up a larger share of the public sector than they do in the private sector—both in the DMV and across the country. This is because the public sector has historically been a pathway to the middle class for workers of color who face labor market discrimination in the private sector (Maye and Marvin 2025).

Trump’s massive cuts to federal employment have also rapidly undone what had been considerable progress in shrinking the regional Black-white employment gap. Figure A shows the employment rate of DMV workers, overall and by race/ethnicity, since the end of 2018. The rapid drop in the Black employment rate since the start of President Trump’s second term is striking, bringing the regional Black employment rate back down to its pandemic-era low. It is also notable that before that drop began, Black workers in the region were employed at essentially the same rate as their white counterparts—the only time in the last two decades when that occurred. These losses in employment will exacerbate existing racial and gender inequity across wages, poverty, and unemployment (Markoff and Zielinski 2026; Zielinski 2025; Busette and Elizondo 2022).

Recent increases in the DMV’s overall unemployment rate underscore the damage Trump is doing to the region. The non-seasonally adjusted unemployment rate jumped more than a full percentage point, from 3.1% in January 2025 to 4.4% in January 2026—more than four times the increase in the national figure. (Importantly, this increase understates the weakening of the area labor market, as the BLS estimates the DMV labor force shrank by 3% over the same period—meaning that many workers who would have been counted as unemployed simply left the area labor force.) For comparison, the national non-seasonally adjusted unemployment rate increased by less than half a percentage point, moving from 4.4% in January 2025 to 4.7% in January 2026 (BLS-LAUS 2026).

These numbers do not capture the full extent of the economic downturn in the DMV area, nor can they give us precise insight into where the pain has been most acutely felt. The administration’s violent deportation agenda, for example, will lead to a drop in immigrant and U.S.-born Hispanic workers’ employment, but resulting changes in Hispanic employment rates may be muted by the corresponding shrinking of the overall Hispanic population (Zipperer 2025). In other words, while the overall Hispanic population in the U.S. may fall dramatically in coming years, the ratio of remaining employed workers to remaining total population may stay somewhat consistent. This will mask the true scale of the economic and social harm being done to immigrant communities in the DMV and across the country.

It is also difficult to fully quantify how the deployment and continued presence of National Guard troops, violent immigration actions, and other authoritarian, fear-inducing tactics have impacted D.C.-area businesses, workers, and families, particularly in neighborhoods with predominately Black and Latino populations. Early data show regional declines in tourism, consumer spending, and foot traffic; harder to capture are the emotional and long-term economic consequences (Montgomery 2025; Hadden Loh and Haskins 2025; Sachs and Cocco 2025). Other recent analyses estimate similar economic harms in cities where targeted federal immigration enforcement actions have been aggressively deployed (Rosenthal and Sojourner 2026). A full accounting of the Trump administration’s harms on the Washington, D.C., region will take years to document.

Other localities should brace for similar consequences

Some of the Trump administration’s actions and their acute consequences are unique to the DMV, a function of the region’s high concentration of federal employees and government contractors, as well as the District of Columbia’s lack of statehood and full constitutional rights. However, the anti-government attacks the administration has unleashed on DMV-area households, workers, and businesses will have cascading consequences for communities throughout the country. The effects of the administration’s authoritarian attacks on the civil service, democratic institutions, and immigrants (Human Rights Watch 2026) that first registered across the DMV should be viewed as a preview of the consequences that will be felt in other regions. While no locality will be spared, regions particularly at risk include those with large shares of government workers (especially federal workers, but state and local government workers too), localities in which federal funding and social safety net programs make up a large portion of total area income, and those with large immigrant populations.

Trump’s attacks on the federal workforce will harm communities that rely on their employment

The day Trump returned to power in January 2025, he began attacking the federal workforce, first by moving to reclassify tens of thousands of federal employees to make it easier to fire and replace them with political loyalists (EPI 2026c), and then by stripping more than one million federal workers of their collective bargaining rights (EPI 2025a). The Trump White House subsequently worked feverishly to slash federal employment, attempting large and chaotic reductions in force, shuttering entire agencies, and coercing tens of thousands of staff to resign, among many other attacks (Poydock 2025). As of March 2026, the administration’s actions have reduced nationwide federal government employment by over 350,000 (11.7%) since January 2025 (Gould 2026).

Though federal workers make up a sizeable share of the DMV’s workforce, over 80% of federal workers live outside the region (Partnership for Public Service 2024). For instance, in Alaska, Hawaii, and New Mexico—states that are home to large swaths of federal and Native land, military bases, and federal research institutions—federal workers make up at least 4.5% of total employment (EPI 2025c). Within states, federal workers tend to be concentrated in specific localities. For instance, in Apache County, Arizona, which is largely made up of the Navajo Nation and the White Mountain Apache Reservations, lands that extend beyond county lines, the federal government employs 12% of the county’s workers, more than double the next most significant county for federal worker employment in the state (EPI 2025c). There are 22 U.S. counties, spread across the South, Midwest, and West Census regions, where federal workers comprise at least 10% of the county’s workforce (see Table 2).

In these counties and elsewhere, federal workers are the backbone of the regional economy, both through the essential services they provide and through their contributions to the local economy. Trump’s attacks simultaneously threaten federal workers’ livelihoods and the economic health of communities in which these workers’ spending on goods and services makes up a large share of economic activity in the region. In Apache County, Arizona, civilian government workers’ earnings comprise 11.7% of total economic activity in the county (see Table 3)—roughly the same as their share of overall county employment. However, in some counties, federal employees’ earnings are a disproportionate share of the regional economic base. For instance, in Leavenworth County, Kansas, where federal employees make up 10.0% of employment (Leavenworth has a large federal prison), federal civilian earnings comprise 22.1% of total income in the county.

The effects from lost federal jobs and income in these regions could be devastating. Some of these communities are places that have already faced historic disinvestment and in which there are few local employment opportunities that can match the quality of federal government jobs. These jobs are historically stable, good quality, union jobs that offer a pathway to the middle class, particularly for workers without a college education.6

Regions highly dependent on federal revenue will also suffer from a reduction in services and a loss of income

Beyond the harm to localities from reductions in the federal workforce, localities that are particularly reliant on federal government revenue and services will bear the consequences of Trump’s actions most acutely, though no locality will be spared from harm. For example, the Trump administration has announced or considered $23 billion in cuts to federal clean energy projects in nearly every state (CATF 2025) and $8 billion in cuts to colleges and universities that will impact every state’s economy (Bedekovics and Ragland 2025). Trump’s 2025 budget bill also made massive cuts to federal safety net programs that millions of low-income households rely on in order to finance tax cuts for the wealthiest households and corporations.

Funds from federal programs such as SNAP, Medicaid, and other social programs not only help struggling families make ends meet, they also comprise a significant share of a locality’s “economic base,” the amount of money circulating in that region, as shown by sociologist Robert Manduca in a recent working paper (2025). Indeed, an often-overlooked benefit of Medicaid coverage is its role as a source of income for low-income households (money they would have had to spend on medical care in the absence of Medicaid). For the bottom 20% of households in the U.S., Medicaid comprised 70% of their total money income, based on recent data from the Congressional Budget Office (Bivens, Wething, and Morrissey 2025). In fact, government transfers such as Social Security, Medicare, and Medicaid collectively made up 40% of the economic base of U.S. regions in 2022 (Manduca 2025). Substantial cuts to government social programs that support low-income households could reduce the economic base of these localities, at a scale equivalent, in many cases, to the loss of entire private industries in those areas.

Without deliberate intervention by state lawmakers to offset lost federal revenues, localities in every state face dire economic losses, but states particularly reliant on government transfers will suffer most. For instance, take Clay County, West Virginia, which is represented in Congress by Rep. Carol Miller (R-WV01), who voted in support of Trump’s budget bill (Miller 2025). Clay County’s poverty rate is more than double the national rate, and its per capita income is half the national amount (U.S. Census 2024a). Of the 10 U.S. counties that rely most on each of the largest federal social insurance programs (Medicare, Medicaid, SNAP, and Social Security) as a share of their economic base, Clay is the only county in the country to show up three times (see Table 4). Federal government transfers in the form of Medicare, SNAP, and Social Security payments comprise 57% of Clay County’s economic base, 20 times the share comprised by the earnings of every private industry in the county combined. Alaska, Arizona, Florida, Georgia, Kentucky, Tennessee, and West Virginia all have at least three counties that are ranked in the top 10 in the country for their reliance on a given social safety net program as a share of the county’s economic base (see Table 4).

Localities that have significant shares of federal workers and rely heavily on federal government transfers may face particularly significant consequences as a result of Trump’s attacks on the federal workforce and the Republican budget bill’s cuts to essential social safety net programs. For example, in Rio Arriba County, New Mexico, and Apache County, Arizona, federal government workers make up 16.1% and 12.0% of all workers in the county, respectively (EPI 2025b). At the same time, both counties are ranked in the top-10 counties most reliant on federal government transfers—Apache is #2 for Medicaid, and Rio Arriba is #10 for SNAP. In Apache County, federal government transfers account for three-quarters (76.9%) of the county’s economic base, and the earnings of federal government civilian workers account for 11.7%—the Navajo Nation Tribal Government is the county’s largest employer (NACOG 2023). Meanwhile, private earnings account for a mere 2.8% of the county’s economy. In Apache, Trump’s cuts to both the federal workforce and federal government programs mean that the federal government may be unable to fulfill its legal obligations to tribal communities (Brown 2025) that have faced decades of disinvestment and depressed economic outcomes resulting from historic land theft and forced assimilation. Apache County’s poverty rate of 31.2% (AZ Economics 2026) is nearly triple the national rate of 11.1% in 2023 (Shrider 2024).

Trump’s anti-immigrant crackdown and deportation agenda hurt localities with large immigrant populations

Trump has launched a campaign of terror against immigrant communities, communities of color, and those who stand with them. Last summer, Trump federalized local police and deployed thousands of federal troops to diverse cities with large immigrant populations (Kim 2025). Though Washington, D.C., may have experienced the most visible federal troop presence, a function of the district’s lack of statehood and the president’s unchecked authority to mobilize the National Guard there (Dallas 2025), Los Angeles was the first city Trump targeted after public opposition to aggressive immigration raids (Kim 2025). It was soon followed by Washington, D.C.; Memphis, Tennessee; Portland, Oregon; New Orleans, Louisiana; Minneapolis, Minnesota; and Portland, Maine.

These attacks are characteristic of an authoritarian playbook that includes forcing the leaders of diverse, opposition-led communities to bend to the strongman government’s will (McManus, Benson, and Herman 2024). Minneapolis, home to a large immigrant population, was subjected to an unprecedented immigration crackdown that drew widespread protests (Boone 2026). During “Operation Metro Surge,” as it was called, federal immigration enforcement officials made 4,000 arrests and killed two U.S. citizens. Though the true toll of this violent operation may never be fully quantified, initial economic data show clear cause for concern. A recent analysis estimated that Trump’s immigration crackdown has led to a 2.9% decline in consumer spending in Minnesota over a single month—the equivalent of the state’s economy losing $626 million (Rosenthal and Sojourner 2026). Relative to overall consumer spending, the food and accommodation sector (which employs a large share of immigrant workers) saw the most significant decline in January 2026—3.8% or a $46 million reduction in economic activity. Researchers also estimated that nearly 3% of workers in the Minneapolis-Saint Paul region were unable to work during the occupation, resulting in a loss of over $100 million in wages (Sojourner and Rosenthal 2026).

Trump’s deportation agenda will continue to destabilize local communities and result in job losses for immigrant and U.S.-born residents alike (Zipperer 2025). Though immigrants live in counties across the U.S., coastal urban areas tend to have the largest shares of foreign-born residents. Counties with the largest foreign-born populations include Miami-Dade, Florida; Queens, New York; Aleutians, Alaska; and Hudson, New Jersey (see Table 5). Counties with relatively large shares of immigrants may see particularly acute harms from aggressive immigration enforcement.

Communities face overlapping economic threats from attacks on federal workers, the social safety net, and immigrants, but state and local lawmakers can resist them.

The Trump administration’s attacks on the federal workforce, the social safety net, and immigrant communities are designed to exacerbate economic precarity in many communities that are already struggling (Bivens 2026). The implementation of Trump’s authoritarian agenda in the DMV region may be the first, clearest, and in some cases most direct manifestation of its harms, but other localities across the country—particularly those with large federal workforces, those that are heavily dependent on federal revenue and those with sizeable immigrant populations—are far from immune, and many will suffer as much, if not more, from this agenda.

While state and local leaders cannot stop federal attacks, they do have the power to resist Trump’s agenda by improving state labor standards (EPI 2026b), advancing protections for immigrant workers (Díaz and Whitaker 2026), investing in the public-sector workforce (Bivens and Shierholz 2026), and using progressive tax policies (Austin and Davis 2025) to stabilize funding for critical social programs and other investments that workers, families, and communities need.

Notes

1. Throughout this report, unless explicitly noted, the source for all employment rate data is the authors’ analysis of Current Population Survey data (EPI 2026a). We compare an average of calendar year 2025 with calendar year 2024 in order to have adequate sample sizes for the noted demographic groups.

2. Employment level by industry and sector data come from the authors’ analysis of the Bureau of Labor Statistics’ Current Employment Statistics (CES) State and Metro Area (SAE) data.

3. These numbers are calculated using monthly totals rather than annual averages. A quarterly comparison of 2025Q4 to 2024Q4 finds roughly the same results—employment fell by 52,600 jobs (13.9%). The quarterly analysis omits October in both years to maintain an apples-to-apples comparison, accounting for missing data due to the government shutdown that began in October 2025 and the subsequent lapse in Bureau of Labor Statistics funding.

4. The non-federal workforce includes private sector workers as well as state and local government employees.

5. These numbers are calculated using monthly totals rather than annual averages. Quarterly comparisons of 2025 Q4 to 2024 Q4 produce similar results—private sector employment fell by 0.1% in the DMV and grew by 0.7% nationally. The quarterly analysis follows the methodology outlined in note 2.

6. On average, federal workers with advanced degrees typically earn less in wages and total compensation than their private-sector counterparts. Federal workers without an advanced degree typically earn more than their private-sector counterparts and have access to retirement benefits that have become less common in the private sector (CBO 2024).

References

Austin, Sarah, and Carl Davis. 2025. The Wealth Proceeds Tax: A Simple Way for States to Tax the Wealthy. Institute on Taxation and Economic Policy, October 2025.

AZ Economics. 2026 “Apache County, Arizona.” U.S. Economic Research. Accessed April 2026.

Barrett, William P. 2025. “America’s Top 100 Charities: A Year of Pain After Trump Cuts.” Forbes, December 12, 2025.

Bedekovics, Gréta, and Will Ragland. 2025. Mapping Federal Funding Cuts to U.S. Colleges and Universities. Center for American Progress, July 2025.

Bivens, Josh. 2026. The Trump Administration’s Macroeconomic Agenda Harms Affordability and Raises Inequality. Economic Policy Institute, February 2026.

Bivens, Josh, and Heidi Shierholz. 2026. “You Can’t Starve the Public Sector to Excellence.” Working Economics Blog (Economic Policy Institute), February 27, 2026.

Bivens, Josh, Hilary Wething, and Monique Morrissey. 2025. Cutting Medicaid to Pay for Low Taxes on the Rich Is a Terrible Trade for American Families. Economic Policy Institute, February 2025.

Boone, Rebecca. 2026. “A Timeline of Trump’s Immigration Crackdown in Minnesota.” PBS Newshour, February 13, 2026.

Brookings Institution. 2026. “Active Residential For-Sale Listings,” DMV Monitor. Last updated February 18, 2026.

Brown, Alex. 2025. “For Indian Country, Federal Cuts Decimate Core Tribal Programs.” Stateline, March 4, 2025.

Bureau of Labor Statistics, Current Employment Statistics State and Metro Area (BLS-CES-SAE). Various years. Public data series accessed through the CES State and Metro Area Databases and through series reports. Accessed April 2026.

Bureau of Labor Statistics, Local Area Unemployment Statistics (BLS-LAUS). Various years. Data from the LAUS are available through the LAUS database and through series reports. Accessed April 2026.

Busette, Camille, and Samantha Elizondo. 2022. “Economic Disparities in the Washington, D.C. Metro Region Provide Opportunities for Policy Action.” Commentary, Brookings Institution, April 27, 2022.

Clean Air Task Force (CATF). 2025. “The High Cost of Retreat: Impacts of Department of Energy Project Cuts.” Clean Air Task Force, November 21, 2025.

Congressional Budget Office (CBO). 2024. Comparing the Compensation of Federal and Private-Sector Employees in 2022. Congressional Budget Office, April 2024.

Dallas, Kelsey. 2025. “The President’s Power to Deploy Troops Domestically: An Explainer.” SCOTUSblog, October 28, 2025.

D.C. Fiscal Policy Institute. 2026. “Congressional Interference Will Cost D.C. Nearly $700 Million in Local Revenue and Jeopardize Efforts to Reduce Child Poverty.” D.C. Fiscal Policy Institute, February 4, 2026.

Díaz, Marisa, and Mimi Whitaker. 2026. How States and Localities Can Strengthen Workplace Protections for Immigrant Workers. National Employment Law Project, January 2026.

Economic Policy Institute (EPI). 2025a. “Executive Order on ‘Exclusions from Federal Labor-Management Relations Programs.’” Federal Policy Watch (Economic Policy Institute), December 17, 2025.

Economic Policy Institute (EPI). 2025b. How Many Federal Employees Live in Your State? Economic Policy Institute.

Economic Policy Institute (EPI). 2025c. “New Resource Calculates How Many Federal Workers Live in Every State, County, and Congressional District” Economic Policy Institute (press release). March 3, 2025.

Economic Policy Institute (EPI). 2026a. Current Population Survey Extracts, Version 2026.3.11, https://microdata.epi.org.

Economic Policy Institute (EPI). 2026b. Holding the Line: State Solutions to the U.S. Worker Rights Crisis. Economic Policy Institute.

Economic Policy Institute (EPI). 2026c. “OPM Finalizes Regulation Enabling Firing Federal Employees for Political Reasons.” Federal Policy Watch (Economic Policy Institute), March 4, 2026.

Friesenhahn, Erik. 2025. “Nonprofit Organizations: State and Regional Employment Trends.” Monthly Labor Review (U.S. Bureau of Labor Statistics), March 2025. https://doi.org/10.21916/mlr.2025.6.

Gould, Elise. 2026. “Attacks on the federal workforce continue (down 18k jobs in March).” Bluesky, @elisegould.bluesky.social, April 3, 2026, 9:01 a.m.

Hadden Loh, Tracy, and Glencora Haskins. 2025. Consumer Spending and Visitor Demand in the Washington, D.C. Region Are Dropping. Brookings Institution, December 2025.

Human Rights Watch. 2026. “Sliding Towards Authoritarianism?” January 2026.

Kim, Juliana. 2025. “Trump Says National Guard Will Soon Go to New Orleans. Here’s the Latest.” NPR, December 3, 2025.

Koma, Alex. 2025. “Here’s How D.C. Solved the Billion-Dollar Budget Problem Congress Created.” WAMU, October 22, 2025.

Kozlov, Max, Jeff Tollefson, and Dan Garisto. 2026. “U.S. Science After a Year of Trump.” Nature 649 (January): 812–815.

Lynch, Teresa M., and Robert Manduca. 2024. “Beyond Local and Traded: Evidence for a Third Industry Market Area Type and Implications for Regional Economic Development.” Economic Development Quarterly 38, no. 3: 183–194, July 2024. 

Manduca, Robert. 2025. Financial and Transfer Income as Components of the Regional Economic Base. Washington Center for Equitable Growth, June 2025.

Markoff, Shira, and Connor Zielinski. 2026. Chronic Racial Inequality Holds Back Workers and Equitable Economic Growth. D.C. Fiscal Policy Institute, March 2026.

Maye, Adewale A., and Stevie Marvin. 2025. “Trump Attacks on Federal Agencies Have Steep Implications for Black Workers.” Working Economics Blog (Economic Policy Institute), April 10, 2025.

McManus, Allison, Robert Benson, and Dan Herman. 2024 “The Dangers of Project 2025: Global Lessons in Authoritarianism.” Center for American Progress, October 2024.

Miller, Carol. 2025. “Miller Votes to Send the One, Big, Beautiful Bill to President Trump’s Desk” (press release). Office of Congresswoman Carol Miller, West Virginia’s First District, July 3, 2025.

Montgomery, Mimi. 2025. “Trump Crackdown Is Affecting D.C.’s Image and Tourism Numbers.” Axios, August 29, 2025.

Northern Arizona Council of Governments (NACOG). 2023. “Business, Jobs, and Industry Highlights for Apache County.” Northern Arizona Council of Governments, November 20, 2023.

Partnership for Public Service. 2024. Beyond the Capital: The Federal Workforce Outside the D.C. Area. March 2024.

Poydock, Margaret. 2025. “How Trump Has Dismantled the Federal Workforce in His First 100 Days.” Working Economics Blog (Economic Policy Institute), May 23, 2025.

Rosenthal, Aaron, and Aaron Sojourner. 2026. The Economic Impact of Operation Metro Surge in January 2026: A Synthetic Difference-in-Differences Analysis. North Star Policy Action, February 2026.

Sachs, Andrea, and Federica Cocco. 2025. “D.C. Tourism Was Already Struggling. Then the National Guard Arrived.” Washington Post, August 29, 2025.

Shrider, Emily A. 2024. Poverty in the United States: 2023. United States Census Bureau, Report Number P60-283, September 2024.

Singh, Kanishka. 2026. “Trump Signs Executive Order Asking Federal Contractors to Eliminate DEI.” Reuters, March 26, 2026.

Sojourner, Aaron, and Aaron Rosenthal. 2026. Impact of DHS Agent Surge on Minneapolis-Saint Paul Metro Area Labor Outcomes. North Star Policy Action, February 2026.

Tomasko, Laura, Hannah Martin, Katie Fallon, Mirae Kim, Lewis Faulk, and Elizabeth T. Boris. 2025. How Government Funding Disruptions Affected Nonprofits in Early 2025: Nationally Representative Findings from the Nonprofit Trends and Impacts Study. Urban Institute, October 2025.

U.S. Census Bureau. 2024a. “American Community Survey 5-Year Estimates: Retrieved from Census Reporter Profile Page for Clay County, WV.” Accessed April 14, 2026.

U.S. Census Bureau. 2024b. “U.S. Foreign-Born Population: 2018–2022 American Community Survey, 5 Year-Estimates (Table B05006).” Accessed April 14, 2026.

Zielinski, Connor. 2025. “Inequality Remained Extreme in 2024 as D.C. Backslid on Poverty.” DCFPI Blog (D.C. Fiscal Policy Institute), September 15, 2025.

Zipperer, Ben. 2025. Trump’s Deportation Agenda Will Destroy Millions of Jobs: Both Immigrants and U.S.-Born Workers Would Suffer Lob losses, Particularly in Construction and Child Care. Economic Policy Institute, July 2025.

Key takeaways:

• The growing revenue of college sports and the heightened attention on the experience of college athletes suggest that college athletics is far from the amateur endeavor it might have started as decades ago.
• Recent policy changes have allowed some college athletes to receive compensation, whether in the form of name, image, and likeness (NIL) rights or revenue sharing. However, not all college athletes have the right to be compensated.
• The NCAA has backed the SCORE Act, which would jeopardize college athlete compensation by prohibiting them from being classified as employees in the first place.
• Policymakers should consider proposals that strengthen rights for college athletes, including granting them employee status under federal labor laws.

Introduction

It has long been argued that college athletes should not receive compensation to maintain the “amateurism” of college sports. However, the growing revenue generated from college sports and heightened attention on the experience of college athletes suggest that college athletics is far from an amateur endeavor.

Only recently have college athletes been granted the right to be compensated for name, image, and likeness (NIL) rights. This decision came into effect after years of antitrust lawsuits against the National Collegiate Athletic Association’s (NCAA) compensation rules. These lawsuits culminated in the Supreme Court decision in NCAA v. Alston, as well as a growing number of states enacting their own compensation laws for college athletes. The recent House v. NCAA settlement allows Division I schools—those with the largest and most economically lucrative athletic programs—to share revenue with college athletes, and further expands opportunities for college athletes to receive compensation.

As a result of these policy changes and a growing movement among college athletes to demand fair compensation for their performance, federal policymakers have put forward proposals to address college athlete compensation. In this blog post, we examine these proposals and their impacts on college athletes and their labor/employment status.

A brief history of college athlete compensation

Despite claims of “amateurism” in college sports, the experience of college athletes showcases a reality in which athletics is prioritized over academics. For example, while the NCAA puts limits on how many hours college athletes can engage in athletic-related activities during playing season, many coaches create expectations for students to exceed these limits, with some athletes exceeding over 40 hours per week. News coverage has reported that coaches have issued fines to athletes who miss practices. Many college athletes are also required to travel for their games, forcing them to miss classes. If college athletes fail to meet these expectations, they may be cut from the team, which could jeopardize future scholarships and other academic opportunities.

Simply put, some college athletes are expected to perform a physical regimen that more closely resembles professional sports than amateur endeavors on top of their academic coursework. The athletic commitment is demanding enough to be its own job, yet college athletes are performing them without any meaningful compensation in return.

In recent years, there have been several policy changes related to college athlete compensation. In 2019, California became the first state to pass a law that granted college athletes NIL rights. The NCAA permitted NIL compensation in 2021 and since then, more than 30 states have enacted laws related to college athlete compensation, with remaining states deferring to NCAA rules to regulate such compensation.

A primary driver of the NCAA’s change of rules regarding NIL compensation was the 2021 Supreme Court decision in NCAA v. Alston. The unanimous decision upheld a lower court decision that found the NCAA’s rules restricting certain educational benefits for college athletes violated federal antitrust laws. In a concurring opinion, Justice Brett Kavanaugh questioned “whether the NCAA’s remaining compensation rules can pass muster under ordinary rule of reason scrutiny” and suggested collective bargaining as an avenue for college athletes to receive a fairer share of the revenue that they generate for their schools. Soon after the NCAA v. Alston decision, the National Labor Relations Board (NLRB) General Counsel Jennifer Abruzzo issued a memorandum taking the position that college athletes are employees under the National Labor Relations Act.

In response to this memo, men’s basketball players at Dartmouth College filed for a union election petition at the NLRB; however, the petition was withdrawn shortly after the 2024 presidential election. In January 2025, Acting General Counsel William Cowen rescinded Abruzzo’s memorandum, leaving college athletes’ employee status in limbo.

The House v. NCAA settlement, which allowed Division I schools to share revenue directly with college athletes, was another turning point in the college athlete compensation landscape. The majority of states with college athlete compensation laws have considered legislation to modify their statues to reflect the terms of the House settlement, but not all have done so.

Who are college athletes?

The National Collegiate Athletic Association is the governing body for college athletics in the United States, overseeing sports programs for 557,000 college athletes at more than 1,100 colleges. It organizes institutions into three divisions based on size, athletic scope, and financial resources. Division I schools are the largest, with the most extensive athletic programs and highest scholarship limits. Approximately 37% of college athletes compete for Division I schools. Division II schools offer fewer scholarships and financial resources, while Division III has the greatest share of college athletes (38%), but offers no athletic scholarships.

During the 2024–2025 school year, the college athlete population was 57% male and 43% female. These young men and women are diverse: 61% are white, 16% are Black, 7% are Hispanic or Latino, 7% report more than two races, and 2% are Asian. Breaking down demographics by race and gender, we find that white males make up the largest group at 32%, followed by white females at 28%, Black males at 12%, and Black females at 4%. The remaining athletes fall into other demographic categories. If we focus on men’s basketball and men’s football athletes at the highest revenue-earning,1 there are 11,504 total athletes, 32% of whom are white and 48% of whom are Black, with the remaining athletes falling into an “other” race category.

Figure A

In terms of geography, college athletes tend to be from the most populous states. According to estimates using NCAA data and population data from Census, most student-athletes are from California, Texas, Florida, New York, and Pennsylvania (in descending order). On a per capita basis, it is Georgia, North Carolina, and Michigan (in descending order) that produce the highest rates of college athletes. This is likely due to having several large state universities with strong athletic programs and an impressive high school sports infrastructure. NCAA-affiliated institutions are also concentrated in the populous states, but especially among states in the Northeast. The states with the most NCAA schools are Pennsylvania (96), New York (93), California (59), Texas (53), and Massachusetts (51).

Current policy landscape

As mentioned above, many states have enacted laws that grant college athletes NIL rights. In the wake of the House v. NCAA settlement, there have been calls for federal policymakers to pass legislation addressing college athlete compensation.

One of the most prominent pieces of federal legislation is the Student Compensation and Opportunity through Rights and Endorsements (SCORE) Act. Backed by the NCAA, this bill would prohibit college athletes from being classified as employees, denying basic labor rights to over half a million young people. The bill creates a federal standard for NIL rights. In doing so, the SCORE Act preempts state legislation concerning college athlete compensation, creating a ceiling rather than a floor for setting standards around college athlete compensation. Further, the SCORE Act limits the types of NIL deals athletes can enter, places caps on NIL payments, and restricts athletes’ abilities to transfer and play at new schools. Finally, the bill would grant the NCAA broad antitrust immunity by authorizing them to limit revenue sharing and education-related benefits to athletes.

On April 3, 2026, President Trump issued an executive order on college athletics. Similar to the SCORE Act, the order directs the NCAA to tighten rules on transfers, eligibility, and NIL compensation, threatening noncompliant schools with the loss of federal funding. It does not, however, address whether college athletes are employees (an earlier executive order from Trump directed the Department of Labor and National Labor Relations Board to clarify employee status of college athletes). Multiple lawyers have argued the latest executive order would not survive a legal challenge. The NCAA president nonetheless praised it, and both the administration and conference commissioners are using the order to push Congress to pass the SCORE Act.

The Student Athlete Fairness & Enforcement (SAFE) Act is another proposal that seeks to codify a federal standard for NIL rights. However, unlike the SCORE Act, the SAFE Act establishes strong health and safety protections for college athletes, allows flexibility for transfers, and places penalties on bad actor agents, among other reforms. Furthermore, the bill does not address college athletes’ employee status or shield the NCAA from antitrust liability.

By far the most effective policy solution for college athletes to be fairly compensated is to grant them the right to form unions and bargain collectively. Legislation like the College Athlete Right to Organize Act  would classify college athletes as employees, granting them the right to form unions and bargaining collectively under the National Labor Relations Act. The bill would also amend the NLRA to define public colleges—in addition to private colleges—as an employer in the context of intercollegiate sports so that all college athletes have the right to organize and collectively bargain.

Below we evaluate whom these proposals impact and estimate how much revenue the college sports industry generates under current compensation policies.

College athlete demographics versus college attendee demographics

College sports are frequently presented as disproportionately Black, but the data show a slightly different story. Black college athletes make up roughly 16% (89,000) of all college athletes compared with 13% (3.31 million) of the total college student population, not significantly different from the NCAA share. Hispanics are drastically underrepresented in the NCAA, accounting for only 7% of college athletes, despite representing over 20% of total college enrollment. In fact, it is white college athletes, and white male athletes in particular, who are disproportionately represented in college athletics: While 61% of college athletes are white and 32% are white males, only 48% of all college students are white and only 19.1% are white males. Notably, it is Black female athletes who are left out of NCAA college athletics at the highest rates. While they account for 8.3% of total college enrollment (2.14 million), they are only 4.5% of total college athletes in the NCAA (25,000).

Figure B

How much do collegiate sports make?

By far, the most economically lucrative division in the NCAA is Division I sports, which includes 37% of total athletes but generates 96% of total revenue across the three divisions, according to the NCAA. According to the Knight-Newhouse College Athletics Database (an authoritative source on college athletics finances and a better representation of self-generated revenue), Division I schools generated $14.6 billion during the 2024 fiscal year. For context, of the five major professional sports leagues in the United States, only the NFL generated more revenue than Division I schools did during the same time period. The NFL, MLB, NBA, NHL, and MLS generated $22.2 billion, $12.8 billion, $12.3 billion, $6.6 billion, and $2.2 billion, respectively, in fiscal year 2024. The primary revenue sources for NCAA Division I are media rights (27%), donor contributions (22%), ticket sales (15%), and institutional support (14%). NCAA Division I revenue has grown 115% (in 2024$) since 2015.

Figure C

Due to the House v. NCAA settlement, schools gained the ability to share revenue directly with athletes beginning in the 2025–2026 school year, adding to any third-party NIL earnings athletes may receive. Though official figures for both revenue sharing and NIL deals are unavailable, schools are currently capped at $20.5 million under the revenue-sharing agreement. Not every university joined the new revenue-sharing arrangement, but every Power 4 school did (the 68 universities in the four highest revenue-generating conferences). Under the generous assumption that all Power 4 schools share the full $20.5 million with their athletes, this would amount to approximately $1.394 billion in athlete earnings, or about 15.1% of total revenue across these conferences. For comparison, coaches at the same set of schools receive $2.3 billion in compensation or 19% of total expenditure. However, if implemented as intended, the revenue-sharing agreement would be a step-up for revenue-generating athletes. Prior to House v. NCAA, the most Power 4 schools could provide the athletes was $2 to 4 million dollars in athletic scholarship money.

Conclusion

Despite the growing revenue that athletes are generating for college sports, many college athletes are not being compensated for their work. Recent policy changes have allowed some college athletes to receive compensation, whether in the form of NIL rights or revenue sharing. However, the reality is that not all college athletes have the opportunity to be compensated. Federal policy proposals, such as the SCORE Act, would further jeopardize college athlete compensation by prohibiting them from being classified as employees in the first place. It is bad policy to deny any worker basic labor rights. Policymakers should consider proposals that strengthen rights for college athletes, including granting them employee status under federal labor laws.

Acknowledgments

The authors thank the Notre Dame Student Policy Network (SPN) for their contributions to the background research for this blog post. The authors would like to thank Billy Bonnist and Liesl Erhardt for leading the SPN team, which included Sarah Francis, Evan Fitzpatrick, Ciara Gilligan, Anvita Jaipura, Owen Murphy, and Caroline Streicker.

1. Defined as the Football Bowl Subdivision (FBS) autonomy schools or schools in the Power 4 (formerly Power 5) conferences. It is worth acknowledging that other sports also produce significant revenue, including women’s basketball, softball, men’s baseball, and women’s volleyball.

Amid increased child labor violations, youth work permit systems have been under attack in some states

In recent years, child labor violations have been on the rise across the country. At the same time, lawmakers in many states have proposed bills to reverse long-standing state child labor standards that prohibit employers from exposing youth under 18 to hazardous jobs or overly long work hours that interfere with their health and well-being. Youth work permits—which many states have historically required—have been a repeated target of this coordinated, industry-backed campaign to weaken child labor laws. Such permits typically require employers to outline the potential hours and work duties for a minor worker, as well as parental approval and verification that the minor is attending school.

Since 2021, lawmakers in at least nine states have proposed weakening or eliminating youth work permit systems, and four have enacted such legislation (Alabama, Arkansas, Iowa, and West Virginia). Most recently, in 2025, Alaska Governor Mike Dunleavy encouraged the legislature to pass a bill that would have eliminated the requirement that minors receive individual authorization to work (and replaced it with a general authorization for employers to hire minors). And in West Virginia, lawmakers successfully eliminated youth work permits for 14- and 15-year-olds and replaced them with age certificates following a two-year push by the right-wing think tank Foundation for Government Accountability (FGA). FGA has played a leading role in efforts to eliminate youth work permits in Arkansas, Iowa, Missouri, and Wisconsin.

New research explains how and why youth work permits are so effective

Proponents of eliminating youth work permits have often argued that work permits are not necessary, are overly burdensome for employers, or that they infringe on parents’ right to decide whether, where, and how long their child should work. In reality, work permits are a proven, effective policy for ensuring that young teens can enter the workforce safely by making sure employers are aware of child labor laws and that parents are fully informed about the conditions of a proposed job.

A year ago, we reported on research providing new quantitative evidence that work permits help prevent federal child labor violations. Using comprehensive data from the U.S. Department of Labor’s Wage and Hour Division from 2008 to 2020, researchers at the University of Maryland and Nanyang Technological University, Singapore, found that states requiring employment certificates saw 13.3% fewer violation cases and 31.8% fewer minors involved in these violations.1 States with work permits also saw 34.9% lower civil penalties per minor, indicating reduced severity of violations that do occur.

New findings from the same research team now reveal two key mechanisms that explain how work permits provide this protection: 1) work permits create a documentary paper trail that increases employers’ accountability and aids government enforcers, and 2) work permits improve compliance with state and federal standards by increasing employers’ awareness of child labor laws. According to the new analysis, requiring verification of parental consent for a minor to work and providing education to employers about hours restrictions are the main features that make work permits effective. State lawmakers can use the new findings to strengthen and modernize their youth work permit systems, using strategies proven to reduce violations and protect youth well-being.

Work permits create legal accountability and enable effective enforcement

Researchers found that work permits enable more effective enforcement of child labor standards by creating a record that employers were informed of child labor standards, therefore making it harder for employers to claim ignorance if violations occur. By analyzing publicly available federal court records, researchers found that in states with work permit mandates, 91% of child labor cases were classified as “willful” or “repeated” violations. These more serious classifications carry higher penalties. In contrast, only 33% of cases in states without work permit mandates received these classifications.

This finding implies that when employers hiring teens must complete a work permit that documents the minor’s age, obtains parental consent, and acknowledges legal requirements, they are made aware of child labor standards and can fully comply with state and federal laws. Work permits also enhance the investigatory capacity of federal enforcement agencies by providing basic documentation about youth employment that investigators can scrutinize when they suspect violations of federal law, as well as bolstering their ability to take effective action if an employer violates the law despite having been informed.

Work permits enhance awareness of specific child labor standards

Second, researchers found that work permits enhance awareness and monitoring of employers’ compliance with federal and state laws—but only for standards that are explicitly mentioned in the permitting process. Analyzing all relevant Department of Labor news releases detailing specific violations (118 in total) from 2020 through 2025, researchers found that work permits reduce precisely the types of violations that the permit forms explicitly warn employers are prohibited under federal law. As Figure A highlights, states with work permits showed: 1) fewer hours violations—minors working beyond federally permitted hours (e.g., federal law limits 14–15 year-olds to 18 hours per week during school weeks); 2) fewer age-limit violations—employment of children below minimum working age (typically 14 for nonagricultural work); and 3) fewer recordkeeping violations—failure to maintain required documentation such as age verification. On the other hand, hazardous occupation violations remained similar across both types of states. Analysis of employment certificate forms from all 38 states with mandates reveals why: While 100% mention age requirements and 60% mention work hours restrictions, most forms do not enumerate the specific hazardous occupations prohibited under federal law.

Figure A

Parental consent and hour limits provide strongest protection

The researchers also examined which features make permits most effective. Figure B highlights the findings. Examining employment certificate forms from 37 of the 38 states that require them (Mississippi’s form was not publicly available), researchers found that parental consent requirements had the strongest protective effect, reducing violations by 13.9% and case severity by 38.7% (measured by civil penalties assessed). Work hours documentation—in which the certificate must record the minor’s planned work schedule (typically completed by employers, though responsibilities vary by state)—also proved effective, reducing the number of minors involved in violations by 24.0%. In contrast, more passive requirements, such as employer signatures and job description requirements, showed lesser independent effects. This suggests that active oversight mechanisms, particularly parental involvement and explicit requirements to record work schedules to show compliance with legal guidelines on hours of work, drive the protective benefits. That these two features are particularly impactful provides further evidence that requiring employer documentation on work permits and verifying parental consent make work permits effective.

Figure B

Work permits prevent violations. State lawmakers should strengthen, not eliminate them.

Understanding how work permits prevent violations points to ways for states to further increase their effectiveness. The researchers identified four best practices lawmakers should consider:

• Strengthen parental consent requirements: Youth work permit applications should require parental signature and include a process for parents to revoke their consent in the future. 
• Strengthen requirements to outline the specific duties of the potential job: Youth work permit applications should require employers to document specific duties of the potential job and include the minor’s planned work schedule. 
• Include information about hazardous occupation restrictions on permit forms: Youth work permit applications should include a list of prohibited jobs for minors under state and federal law and affirm the employer’s commitment not to employ a minor for hazardous tasks and occupations.
• Clearly state hour limits: Youth work permit applications should include permitted daily and weekly hours and prohibitions on overnight work under both state and federal law. These forms should also clearly state that, where there are discrepancies between state and federal law, the more protective law applies.

These new insights into how youth work permits function reinforce the researchers’ original conclusion: Youth work permits are a proven method for reducing child labor violations. And they show that the permitting process can be a highly effective vehicle for educating employers, teen workers, and parents about legal rights and protections. States with existing work permit systems can strengthen them to enhance their protective effects—as Illinois, Michigan, and Washington have done—and states that do not have work permit requirements should take immediate steps to implement or reinstate them.

1. Throughout this report, the terms “work permits” and “employment certificates” are used interchangeably. Age certificates are distinct and more limited; they typically verify age but do not include the same safeguards.

It might be tempting to think that this preliminary downward revision means that the U.S. economy was much weaker than originally reported. But most of the slower job growth in 2024 was the result of smaller working-age population growth due to reduced immigration and the aging of the workforce—it was not due to degraded labor force participation or opportunities for prime-age workers in the U.S. labor market. In fact, research shows that there were about 600,000 to 900,000 fewer net immigrants between 2023 and 2024. Smaller population growth requires smaller increases in the number of jobs to maintain employment rates.

A clear way to see how the labor market stayed strong in 2024 in the face of lower job growth is to look at prime-age labor force participation, ages 25 through 54. This uses data from the Current Population Survey—which is unaffected by potential payroll revisions— and focusing on prime-age workers limits the role of the aging workforce from affecting recent trends in the data. If rates of labor force participation stay stable even as the number of jobs falls, then this implies that a reduction in labor supply has been well-absorbed by the labor market and has not translated into fewer job opportunities for the workers remaining in the U.S. economy.

Figure A shows that the prime-age labor force participation rate was higher in every month of 2024 compared with 2023. So even as payroll growth may have slowed, labor force participation grew in mid-2024 and stayed at least as high through the beginning of 2025.

Figure A

Labor force participation has stayed relatively flat in 2025, and in general most labor market indicators have remained pretty solid in 2025 compared with the longer historic record. The prime-age employment-to-population ratio is still close to its pre-pandemic level. Unemployment did increase in 2024 and 2025 but the rate is still low by historical standards.

At the same time, some indicators suggest a labor market that is softening as 2025 moves on. There has been a marked decline in payroll employment growth—averaging only 29,000 jobs per month since May. Nominal wages are still rising faster than inflation, but the pace of private-sector real wage growth is half as fast as it was three months ago. Layoffs remain low and regular state unemployment insurance claims aren’t rising, but federal unemployment insurance claims are about twice as high as they were last year at this time.

We also see troubling signs of weakness in the unemployment rates for specific demographic groups. For example, the labor market for Black workers has deteriorated in 2025. Figure B shows that Black unemployment held relatively steady in 2024, but over the last three months Black unemployment rose to 7.5%, its highest in nearly three years. The labor market experience of Black workers has often been considered a bellwether, as Black workers not only experience much worse outcomes in a downturn but also may experience that downturn first.

Figure B

Another group seeing troubling trends are young workers, those ages 16 to 24 (shown in Figure C). Although the unemployment rate for workers above the age of 25 has stayed relatively flat over the last year, the unemployment rate for young workers has been steadily rising—hitting 10.5% in August, the highest it has been in 3.5 years. Higher unemployment among young workers is also consistent with a softer hires rate, making it harder for new entrants to break into the labor market.

Figure C

Black and young workers’ labor market outcomes are more volatile because of smaller sample sizes, but the recent deterioration is hard to ignore. Because many rates in the household survey did not noticeably deteriorate in 2024, slower payroll growth in that year seemed to be driven simply by slower population growth. In 2025, however, there are now preliminary signals that job growth may be slowing enough to reflect deteriorating employment prospects for at least some groups of workers. If the latest jobs report is released on Friday (which may be delayed given the possibility of a government shutdown), we’ll track employment and unemployment rates from the household survey to see if the labor market is continuing to weaken.

Key findings:

• Young workers—those 16–24 years old—have experienced historically strong real wage growth (9.1%) since February 2020, exceeding the wage growth for workers ages 25 and older (5.4%).
• Wages for young workers have also grown faster than the prices of rent and college tuition since February 2020.
• A smaller share of young adults is unemployed, underemployed, or “idled”—neither employed nor enrolled in further education—than their averages over the prior three decades.
• However, recent Trump administration policy actions could be devastating for young adults trying to get a foothold in the labor market as they enter the workforce following graduation.

Young workers have experienced a strong labor market coming out of the pandemic recession, with better job opportunities and faster wage growth than they experienced in much of the prior four decades. However, the Trump administration’s recent attacks on the federal workforce, higher education, and registered apprenticeships—as well as imposing extreme tariffs—threaten to reverse these gains. In this first post in a series on young adults, we examine their labor market prospects as they graduate from high school and college this spring and discuss how policy changes might impact their prospects.1

Young adults have experienced historically fast wage growth in this business cycle

Figure A shows that real (inflation-adjusted) wage growth has been strong for workers of all ages in the pandemic recovery, but young workers have experienced faster wage growth (9.1%) than workers ages 25 and older (5.4%). Compared with the previous four business cycles, real wage growth in this recovery has been extraordinarily fast for young workers. It was not only significantly above zero for the first time in the early stages of a recovery, but also 14.4 percentage points faster than the recovery following the Great Recession of 2008.

Though the difference is not as stark, it is worth noting that workers ages 25 and older have also experienced faster wage growth this business cycle than in prior business cycles. This is no accident—all age groups have seen strong wage growth during the pandemic recovery because of intentional policy decisions.

After the huge job losses in March and April 2020 (specifically in industries most likely to employ young workers), policymakers passed large fiscal recovery packages that spurred rapid rehiring efforts, which gave workers leverage to secure higher wages and better working conditions. Further, pandemic relief efforts like expanded unemployment insurance coverage and economic impact payments gave these workers the economic security to be more selective than normal when job hunting.

Figure A

Wages for young workers have grown faster than rent and college tuition

We also compare nominal wage growth with the growth in prices of goods and services that stress the budgets of young adults. In addition to exceeding overall inflation, nominal wage growth for young workers (40.3%) has significantly outpaced growth in the cost of rent (27.4%) and college tuition (8.6%) since February 2020 (Figure B).

While rent and college tuition are unaffordable for many young adults and their families, the rate of growth since 2020 suggests they have not become any more unaffordable over this period. This is a crucially under-recognized achievement: strong labor markets directly made both college attendance and the cost of rent more affordable for young workers in recent years.

Figure B

Young adults are less likely to experience unemployment and underemployment now than in the past

Labor market outcomes for young workers have also been more promising recently than on average over the prior three decades. Figure C below shows the unemployment rate, underemployment rate, and “idling” rate in March 2025 and the average in the July 1990 to March 2024 period.2

The unemployment rate for young workers now is 9.2% compared with 12.1% on average between 1990 and 2024. The underemployment rate is also lower today compared with the prior three decades. The underemployment rate is the share of the labor force that either 1) is unemployed, 2) is working part time but wants and is available to work full time (an “involuntary” part timer), or 3) wants and is available to work and has looked for work in the last year but has given up actively seeking work in the last four weeks (“marginally attached” worker).

Another important measure of opportunities for young adults is what we call the idled rate—the share of young adults who are neither employed nor enrolled in school. This idled rate is useful because it is often hard to judge whether higher employment of young people is unambiguously good. For example, in some states that have weakened child labor laws, 16- and 17-year-olds are now at higher risk of facing exploitative conditions like subminimum wages, safety hazards, or long hours that interfere with high school completion. But it is almost always unambiguously bad when young people lack opportunities to either work or be enrolled in school—and this is what the idled rate measures. The share idled in 2025 is lower than on average between 1990 and 2024.

Figure C

Recent Trump policy decisions could harm the economic futures of young adults

These promising outcomes for young workers are a tremendous policy achievement. The decision to use large fiscal relief and recovery packages to heal the labor market as quickly as possible after the pandemic recession has paid off enormously for the nation’s young workers. However, recent actions by the Trump administration threaten to reverse these gains.

Specifically, attacks on higher education in the form of cuts to university funding and uncertainty around student loans will make college less attainable for all, but in particular for young women and Black and Hispanic people. This will only further exacerbate significant gaps in education, wages, and lifetime earnings across race/ethnicity and gender.

The administration has also attacked federal initiatives to expand access to registered apprenticeships, another accredited system through which young workers advance their careers and reach higher earnings. Data show that union registered apprenticeships are increasingly important pathways to living-wage skilled trades careers for young and low-income people, women, and Black and Hispanic workers, illustrating the disproportionate harm that these attacks will have.

Additionally, the current administration has launched large-scale attacks on the federal workforce, supported drastic cuts to Medicaid, pursued mass deportations, and imposed extreme tariffs.

Cuts to the federal workforce mean that young people interested in public service careers will have fewer opportunities. Further, major staffing cuts to the Department of Labor and National Institute for Occupational Safety and Health will weaken the enforcement of laws that keep workers safe and investigate wage violations (which disproportionately harm young workers, women, people of color, and immigrants). Meanwhile, Medicaid cuts would directly harm young people who are lower income and rely on Medicaid to meet their health care needs, including the high share of women who depend on Medicaid for their pregnancy.

Those cuts—combined with current immigration and tariffs policies—could lead to an economic recession. Young workers are often hurt more in a recession due to the “last hired, first fired” phenomenon and their lack of a significant foothold in the labor market. Graduating into a recession can set them back for years to come, depending on the depth and duration of the recession. To prevent a recession and remove the threat of undoing gains made over the pandemic recovery, these policy actions must be halted immediately.

In the next blog post in this series, we will delve deeper into the wages of high school and college graduates, respectively, and discuss gaps that exist across race and ethnicity and gender.

1. We define young workers as 16–24 years old. For all of our data in this post we use 12-month moving averages. For example, March 2025 data represent the average of April 2024 to March 2025. Smoothing over 12 months allows for sufficient sample sizes for analysis and to account for seasonal fluctuations throughout the year. Data for young workers, in particular, may vary by season given changes in their schedule between the academic year and the summer.

2. The idling data are available starting in 1984, but we use July 1990 as the start date to capture only full business cycles.

Findings: Between 2019 and 2024, there has been a notable reversal of long-term trends in wage growth. Low-wage workers experienced historically fast real wage growth (adjusted for inflation) and the strongest wage growth compared with workers at all other parts in the wage distribution. The hourly wage for the lowest-paid workers at the bottom 10% grew a tremendous 15.3% over this period. The wage growth at the lower end of the wage distribution was exceptional, significantly faster than their average growth in the prior 40 years and faster than higher-wage workers over the same five-year period. Wage growth for low-wage workers also far exceeded the 2.1% wage loss that characterized the five years following the start of the last pre-COVID business cycle (2007–2012).

Faster wage growth at lower-wage levels has resulted in a compression of wages (or a narrowing of the wage distribution among the bottom 90% of wage earners). In addition, Black and Hispanic workers, young workers, and workers with lower levels of educational attainment experienced relatively fast wage growth over the last five years. Nevertheless, because pay at the bottom of the distribution started at such a low point in 2019, low-wage workers today continue to suffer from wages that are grossly inadequate to sustain families, and significant wage gaps exist at all points in the distribution across demographic groups.

Implications: Policymakers responded to the pandemic recession with actions that made a real difference in people’s lives: Wages grew for those who needed it most. Thoughtful policymaking going forward can help ensure that lower-wage workers continue to see improvements in their standard of living.

Recommendations: Policymakers can choose to prioritize a strong labor market that continues to deliver these gains for lower-wage and historically marginalized demographic groups. Unfortunately, recent actions from the Trump administration will put downward pressure on wage growth and raise the risk of a recession. Policymakers should:

• reverse the general assault on the public sector, restoring federal employment levels and federal payments
• reject cuts to benefit programs like Medicaid and SNAP, critical parts of the safety net for low-wage workers and their families
• prevent an escalation of deportations, which will lower employment and wages across the economy
• oppose across-the-board tariffs that lower real wages without delivering key industrial benefits that a more strategic trade policy approach could realize

Introduction

The current business cycle is a notable reversal of historic trends that increased wage inequality in the United States labor market. Between 1979 and 2019, lower-wage workers experienced only a few short years of strong wage growth in real (inflation-adjusted) wages, and their wage growth over that period significantly lagged behind the wage growth of higher-wage workers. But, between 2019 and 2024, workers in the bottom of the wage distribution have seen fast wage growth compared with their historical norm and with higher-wage workers. This stronger relative growth for lower-wage workers has led to a compression, or a narrowing, of the gap between hourly wages near the bottom and the top of the wage distribution.

Policy choices in the wake of the pandemic and the strong labor market have made these unusually fast gains possible. Historically disadvantaged groups—such as Black and Hispanic workers, young workers, and workers with less than a college degree—have experienced particularly strong wage growth in recent years. Of course, this recent growth has only just begun to narrow these large wage gaps, and the nation’s lowest-paid workers still receive wages that are inadequate to meet most families’ basic needs. Policymakers need to strengthen labor standards so that workers can lock in the gains and continue to build on them, even in weaker labor markets.

Wage growth continued to be strongest for low-wage workers between 2019 and 2024

In this report, we analyze the wage distribution at deciles from the 10th to the 90th percentile of the wage distribution, using Current Population Survey (CPS) Outgoing Rotation Group microdata (EPI 2025a). In our analysis, we employ a new methodology to better smooth our wage deciles by using information from nearby percentiles, described in detail in the appendix. Notably, the labor market story the data tells is unaffected by these changes in methodology. Regardless of method, it is important to note that our estimates of the 90th percentile wage do not fully capture the earnings of those at the very top of the wage distribution, which is better captured with other data sets, discussed briefly later on.1

Our analysis focuses on changes in real wages between 2019 and 2024, as well as historical comparisons of real wage changes between 1979 and 2019. Figure A displays wage growth at each decile of the wage distribution. Between 2019 and 2024, hourly wage growth was strongest at the bottom of the wage distribution. The 10th-percentile real hourly wage grew 15.3% over this five-year period.

Figure A

To be clear, these are real (inflation-adjusted) wage changes. Overall inflation grew 21.3%, or about 3.9% annually, between 2019 and 2024.2 Even with this historically fast inflation, particularly in the immediate aftermath of the pandemic recession, low-end wages grew substantially faster than price growth. Nominal wages (i.e., not inflation adjusted) for these lower-wage workers rose 39.8% cumulatively since 2019.

Lower-end wages grew faster than any other group within the bottom 90% of earnings. In fact, across the wage distribution, we see the pace of wage growth declining for each successive wage group until we reach the highest wage groups.3 Compared with the 15.3% wage growth at the 10th percentile, growth was less than half as fast at the median or 90th percentile.

The bounceback low- and middle-wage workers experienced was stronger than in any business cycle since 1979

Figure B shows just how exceptional this recovery has been in achieving strong wage growth for low-wage workers. The figure presents real changes in the 10th-percentile wage and the 50th-percentile wage five years from the prior peak in each business cycle since 1979. Wage growth at the 10th percentile was positive in only one other five-year recovery cycle (1989–1994), and even compared with then, the current 10th percentile wage growth is seven times as fast.

Middle-wage workers experienced slower gains in the recent business cycle compared with low-wage workers. However, the slower middle-wage growth over the last five years was still significantly faster than that found in the four prior business cycles, more than twice as fast as the next fastest growth rate across business cycles.

Figure B

Wage compression in the most recent period contrasts sharply with the prior 40 years

This wage compression is in stark contrast with the experience of workers in the prior four decades.4 Figure C displays wage growth between 2019 and 2024 compared with wage growth between 1979 and 2019. This time we report annualized wage changes—which allow for comparison across periods that span different numbers of years (e.g., a five-year span versus a 40-year span).

Figure C

The differences in wage growth between these periods are striking because the recent pattern is so contrary to historical trends. Whereas in the most recent period, wage growth was stronger among each successive lower-wage group starting with 70th percentile workers on down, the opposite pattern occurs in the earlier forty-year period. Each successive higher-wage group displays wage growth at least as fast as the previous one. In the most recent period, middle-wage workers’ growth was not far behind growth for the highest-wage workers—about four-fifths as fast—but in the 1979-2019 period, their wage growth was less than half as fast. The difference is even more extreme for the lowest-wage workers: only an average of 0.3% growth over the 40-year period versus nearly 2.9% annualized growth over the past five years. All wage groups experienced annualized wage growth faster in the most recent period as between 1979 and 2019 and much faster among roughly the bottom 60% of the wage distribution. Had the median wage grown over the last 44 years at the 1.1% rate it did from 2019–2024, it would be over $31 per hour today rather than its current value of just under $25 per hour.

While the strongest growth in the recent period swamps the slower growth in the prior period, the overall trend since 1979 is still one of rising inequality since the prior period is nearly 10 times as long as the most recent period being analyzed. (See EPI’s new State of Working America data library for the full complement of data years and trends (EPI 2025f).)

Wage compression over the last five years narrowed the 90–10 wage ratio

Another way to analyze wage inequality or quantify the extent of wage compression across periods is to analyze the wage ratios between different points in the wage distribution. The 90–50 and 90–10 wage ratios are measured by the 90th percentile wage divided by the 50th or 10th percentile wage, respectively. For instance, the 50th and 90th percentile wages were $24.87 and $62.75, respectively.5 Therefore, the 90–50 wage ratio is 2.5, which means that higher-wage workers are paid about two-and-a-half times as much as middle-wage workers on an hourly basis. The average annual change in the wage ratios is determined by comparing these ratios over time and dividing by the number of years in each period.

Figure D displays changes in the 90–50 and 90–10 wage ratios over the 1979–2019 and 2019–2024 periods. Over the 1979 to 2019 period, we see inequality ticking up, with a widening between high-wage workers and both middle- and low-wage workers through the growth in both the 90–10 and 90–50 wage ratios.

Between 2019 and 2024, there was a slight increase in the 90–50 wage ratio because wages at the middle grew just a bit slower than wages at the top. On the other hand, much faster low-end wage growth has shrunk the 90–10 wage ratio considerably over the last five years. While this wage compression is welcome news, it still does not reverse the decades-long growth in equality when measured over the entire 1979 to 2024 period.

Figure D

The very top continues to amass larger shares of the overall pie

Changes at the very top of the wage distribution cannot be accurately measured using the Current Population Survey, given the censoring of high-end wages and the possibility that high earners do not accurately report their income, but Social Security Administration (SSA) data reveal what’s happening within the top 10%, 5%, 1%, and even 0.1% of the annual earnings distribution.6 Between 2019 and 2023 (the latest data available), average annual earnings also showed signs of wage compression: The bottom 90% grew by 5.0%, and the top 5% and top 1% grew by 2.3% and 0.6%, respectively. But, like the long-term pattern we observed in hourly wages, the five-year compression of wages pales in comparison with the four-plus-decade increase in inequality. Between 1979 and 2023, the average earnings of the bottom 90% grew 43.7%, while the top 5% grew 135.4% and the top 0.1% jumped 353.9% (Gould and Kandra 2024).

Faster wage growth for low-wage workers was driven by policy decisions and a tight labor market

The fast growth over the last five years, particularly for low-wage workers, didn’t happen by chance: It was largely the result of intentional policy decisions that addressed the pandemic and subsequent recession at the scale of the problem. Wage growth in the most recent five-year period far exceeded that of the last four business cycles, including notably, the downright losses following the start of the last pre-COVID business cycle (2007–2012). In the aftermath of the Great Recession, policymakers learned a lesson about the pitfalls of austerity: The pursuit of austerity led to a slow and prolonged economic recovery.

Several large spending bills were passed in the first year of the pandemic, which provided enhanced and expanded unemployment insurance, economic impact payments, aid to states and localities, child tax credits, and temporary protection from eviction among other measures (Gould and Shierholz 2022). These actions provided relief to workers and their families to help them weather the recession. These measures also fed the surge in employment, which gave low-wage workers better job opportunities and leverage to see strong wage growth.

Though ticking up slightly in 2024, the unemployment rate remained low, averaging 4.0% over the year, and the share of the population ages 25–54 with a job—the prime-working-age employment to population ratio (EPOP)—remained high in 2024 at 80.7%, surpassing even the pre-pandemic peak in 2019 of 80.0%. 

This tightening labor market further bolstered workers’ leverage. Low unemployment means that workers are relatively scarce, which requires employers to work harder to attract and retain workers and lessens their discretion to discriminate without facing a profitability penalty. In low-unemployment labor markets, lower-wage and historically marginalized workers experience better labor market outcomes and faster wage growth (Bivens and Zipperer 2018; Wilson and Darity 2022).

In addition, the sudden loss of millions of low-wage jobs at the start of the pandemic, followed by the extraordinarily fast employment recovery, meant that the frictions that tie workers to particular jobs—that is, the barriers that would normally keep workers from searching for better employment opportunities—were not constraining workers looking for work in this period. This “severed monopsony” in a time of furious rehiring reduced the normal drag on wage growth imposed by these frictions (Bivens 2023). High numbers of low-wage workers quit and found better jobs, increasing churn in the low-wage labor market (Autor, Dube, and McGrew 2023). This phenomenon increased low-wage workers’ leverage, which further contributed to faster wage growth. Employers, and particularly employers of low-wage workers, simply had to work harder to attract and retain the workers they wanted.

Higher minimum wages can lock in the gains made by low-wage workers

The minimum wage is an essential labor standard that establishes a baseline for earnings, strengthens the negotiating power of low-wage workers, and helps reduce gender, racial, and ethnic wage disparities. Robust labor standards—like the minimum wage—complement tight labor markets by accelerating wage growth for lower-wage workers. Higher minimum wages help sustain these gains, providing stability for low-income workers during both economic downturns and periods of growth.

Despite the federal minimum wage remaining stagnant at $7.25 per hour since 2009, since then more than half of U.S. states have implemented increases (EPI 2025d). By analyzing wage growth trends across states that have and have not raised their minimum wage, we can assess whether these increases have contributed to wage growth for lower-wage workers.

Before 2019, state minimum wage increases did more to bolster wages at the bottom

Between 2016 and 2017, wage growth for workers at the 10th percentile was twice as fast in states that raised their minimum wage compared with those that did not (Gould 2017). For women at the 10th percentile, wage growth was 2.5 times higher in states with minimum wage increases than in those without, contributing to a reduction in the gender wage gap among low-wage workers.

From 2013 to 2019, leading up to the pre-pandemic economic peak, low-end wages grew by 17.6% in states that increased their minimum wage at least once, almost twice as high as the 9.3% in states that kept it unchanged (Gould 2020). However, the gap in wage growth was smaller during the 2017–2019 period due to a tightening labor market. As previously mentioned, when the unemployment rate is low, the minimum wage has less of an impact, as employers must already offer higher wages to attract and retain workers, which reduces the number of employees directly affected by minimum wage increases. While these increases may not bite in good times, higher state minimum wages can help workers lock in those gains when the labor market softens and lower-wage workers lose the leverage they may have had in the tighter labor market.

A tight labor market and state minimum wage increases worked in tandem to generate immense low-end wage growth

In the most recent period between 2019 and 2024, 29 states and the District of Columbia raised their minimum wage through legislation, referendum, or indexing. To analyze the relationship between these state-level increase and low-end wage growth, we group all 50 states (plus D.C.) into three categories, as shown in Figure E: states with no minimum wage increase, states with a small increase, and states with a large increase.7

Figure E

In the last five years, nearly every state with a higher minimum wage than the federal minimum of $7.25 experienced an increase in their minimum wage.8 The average nominal increase (not adjusted for inflation) in the minimum wage between 2019 and 2024 among states with any increase was 37.8%. Even with relatively fast inflation of about 21% over this period, average minimum wage increases outpaced inflation.

Figure F compares real wage increases at the 10th percentile, overall and for women, across these three groups of states.9 The key result is clear: Regardless of minimum wage changes, low-wage workers experienced extremely fast wage growth in all states. Even in states with no minimum wage increase, low-wage workers experienced an 11.7% wage increase between 2019 and 2024. Low-end wages grew at incrementally faster rates in states with small and large minimum wage changes compared with states without any change in their minimum wage, 13.7% and 14.8% compared with 11.7%, although these differences were not statistically significant at conventional levels of significance.10

For lower-wage women, specifically, the findings are stronger. The 10th percentile wage for women grew 10.8% in states with no minimum wage change, but low-end wage growth in states with large minimum wage increases was far faster at 15.6%, an extra 4.8 percentage points of wage growth due to the minimum wage. In this case, the wage growth in states with large minimum wage changes was statistically larger than wage increases in states without a minimum wage change.11

Figure F

Research shows that a tightening labor market alone leads to stronger wage growth among lower-wage workers (Bivens and Zipperer 2018). The 10th-percentile wage grew across all states regardless of changes in state minimum wages because of enhanced relief measures and reduced frictions that boosted low-wage workers’ leverage. It is also important to note that low-wage workers in states with minimum wage increases saw even faster growth than low-wage workers in states without such increases.

Minimum wage increases are crucial to lock in low-wage workers’ gains and build on them

It is essential that we increase the federal minimum wage in order to secure the real wage gains for low-wage workers over the last five years. Unfortunately, Congress has failed to increase the federal minimum wage in the last 15 years, and it is now at its lowest value in real terms in 68 years (Cooper, Hickey, and Zipperer 2022; Zipperer 2024).

Many states and localities have continued to increase their minimum wages in response to federal inaction. On January 1, 2025, 21 states increased their minimum wage, benefiting more than 9.2 million workers (Hickey 2024). Among those affected, 20.4% are in families with incomes below the poverty line, while nearly half (48.5%) have incomes below twice the poverty line (Hickey 2024).

Low-wage workers experienced vital gains due to the tight labor market and legislative measures enacted early in the pandemic recovery. To secure these gains and have protection from weaker labor markets, these workers need strong labor standards such as a higher minimum wage.

Despite historic wage growth, low-wage workers continue to suffer from grossly inadequate wages

Although tight labor markets and, to some extent, minimum wages have bolstered wages at the low end of the wage distribution, millions still work for grossly inadequate wages. Federal policy action is needed to aid working families across the United States in making ends meet.

In 2024, the 10th-percentile hourly wage was $14.26. While this represents a significant improvement from 2019, it is still far from sufficient to make ends meet: Even if that 10th-percentile worker worked full time, their annual pay would be only $29,661. In states that saw increases in the minimum wage between 2019 and 2024, the average 10th-percentile hourly wage was $15.24 in 2024, more than 18% higher than in states that saw no minimum wage increase ($12.85).12

Even with 15.3% wage growth since 2019, it is still difficult—if not impossible—for a 10th-percentile worker to make ends meet. According to EPI’s Family Budget Calculator, whether a worker is making $12.85 an hour or $15.24 an hour, they are still not earning enough to attain a modest, yet adequate, standard of living (a basic family budget for a single individual with no children) in any county or metro area in the United States (EPI 2025b). In fact, there is nowhere in the country where a minimum-wage worker is paid enough to meet the requirements of their one-person local family budget on their wages alone (deCourcy and Gould 2025; Gould, Mokhiber, and deCourcy 2024).

Wage compression meant faster growth for historically marginalized workers

Long-standing discrimination and occupational segregation have led women and Black and Hispanic workers to be disproportionately overrepresented in the low-wage workforce (Bahn and Cumming 2020; Wilson and Darity 2022). Young workers and workers with lower levels of educational attainment also face higher unemployment and lower wages than their more experienced or more educated counterparts.

Table 1 provides wage levels at the middle of the wage distribution—the 50th percentile—for select demographic groups in 1979, 2019, and 2024. We use this to examine how wages within groups have changed in recent times (the last five years) compared with the prior 40 years, with wage changes calculated on an annualized basis for better comparability.

Wage growth has been fastest for Black workers, young workers, and less educated workers

Historically disadvantaged demographic groups experienced far faster wage growth over the last five years compared with the prior 40 years. Although women experienced significant wage growth between 1979 and 2019 due to the increase in labor market opportunities, their wage growth was even greater in the most recent period. Men and white workers’ wages grew in line with overall gains, while Black workers saw the greatest boost in wage growth.

Middle-wage Black men saw the biggest boost in wage growth compared with the earlier period. After increasing at an annualized rate of only 0.1% between 1979 and 2019, Black men’s wages increased at an annualized rate of 1.7% between 2019 and 2024. Black women saw the fastest wage growth over the past five years (1.8%), after experiencing only moderately paced growth in the earlier period (0.7%).

Table 1

Middle-wage young workers — aged 16-24— and workers with lower levels of educational attainment—with less than a bachelor’s degree—also experienced dramatically faster wage growth between 2019 and 2024 than in the 40 years prior. Both groups saw extremely sluggish growth between 1979 and 2019 (0.2% and 0.1%, respectively), while young workers saw 20 times faster annualized growth (2.6%) and less educated workers saw 10 times faster annualized growth (1.1%) in the most recent period.

Key wage gaps narrowed but still remain large

The wage levels by race/ethnicity and by race/ethnicity and gender in Table 1 can be used to calculate the wage gaps between groups for 1979, 2019, and 2024. For example, the Black-white wage gap is calculated by subtracting the median Black wage ($21.40) from the median white wage ($27.28) and then dividing this number by the white wage. In 2024, this equates to Black workers being paid 21.6% less than white workers. We go one step further and calculate an annualized percentage point change in these racial and gender wage gaps over the most recent period and the prior 40 years (Figure G).

After widening between 1979 and 2019, the wage gaps between Black and white workers overall and between Black and white men, specifically, narrowed over the past five years. Notably, the gap between Black and white men closed at an average annual rate of 0.7 percentage points between 2019 and 2024, making a significant dent in the gap between these groups. The gap between Hispanic and white men also narrowed during this period, but at a slower rate (0.4 percentage points). Black and Hispanic women each experienced significant narrowing compared with white men in the last five years (0.6 and 0.7 percentage points, respectively), much faster than compared with the prior 40 years.

Although these data show promising signs for racial and gender wage equality, significant progress is still needed. In 2024, middle-wage Black workers are still being paid $5.88 less per hour than their white counterparts, while the gap for middle-wage Hispanic workers is even larger ($6.94). For full-time workers, this gap translates to more than $12,200 lower pay for Black workers and $14,400 lower pay for Hispanic workers than white workers. These gaps are even larger between White men and Black and Hispanic women. White men at the median are paid $9.09 and $10.36 more than 50th percentile Black and Hispanic women, which translates to annual earnings gaps of more than $18,900 and $21,500, respectively.

Figure G

Policy matters

The source of much of the remarkable wage growth and compression over the last five years was a very tight labor market: high aggregate demand and “severed monopsony” (Bivens 2023), breaking the frictions that tie workers, especially low-wage workers, to certain jobs. An expansion of child tax credits, unemployment insurance benefits, food assistance, and direct payments all contributed to the economic recovery from the COVID-19 pandemic and made the labor market the strongest it has been in a generation (Gould and Shierholz 2022).

The Trump administration, however, is currently taking concrete and alarming steps to reverse these accomplishments. Its general assault on the public sector, by terminating the employment of tens of thousands of federal government workers, will directly reduce incomes and demand throughout the country (EPI 2025c). The administration’s elimination of grants and contracts will further weaken the labor market, shrinking universities, charities, and possibly even the entire domestic semiconductor industry (García 2025; Mickle and Swanson 2025). All these efforts will weaken the demand for employment and, therefore, reduce wage growth.

An escalation of deportations will reduce employment of foreign-born and U.S.-born workers alike, and the resulting decrease in demand due to lower incomes is likely to put downward pressure on average hourly wage rates (East et al. 2023). SNAP cuts and Medicaid reductions would eviscerate an already threadbare safety net for low-income families and reduce workers’ fallback position and bargaining power for higher wages (Bergh 2025; Bivens, Wething, and Morrissey 2025). In addition, the Trump administration’s broad-based tariffs will raise prices and reduce real wages in every state (Hersh and Bivens 2025) without providing any benefits that might accrue from a more strategic approach that targeted protection at the specific sectors that need it.

Even after two months there is not a single development or pronouncement from the Trump administration or Congress that is consistent with broad-based wage growth. Instead of coasting on a historically strong economy, the Trump administration seems willing to trash it, and the only question is this: How deep of a hole will they dig for the vast majority of workers who depend on a strong labor market to make ends meet?

Appendix

Calculating wage percentiles

In this report, we calculate real hourly wage percentiles (10th–90th) to analyze changes to wages across the wage distribution. One challenge that researchers face when calculating percentiles is that wages reported in survey data tend to be bunched at round values. For example, in 2019, about 80% of hourly workers paid between $20 and $20.99 reported an hourly wage of exactly $20.00. This bunching arises because people are more likely to be paid at certain round wages (Dube, Manning, and Naidu 2020); survey respondents are more likely to respond to questions about pay with round values; and beginning in 2023, the Census Bureau (Census) began rounding wage and earnings survey responses (Census 2022).

Bunching in survey data is problematic for estimating year-to-year wage growth at specific percentiles because it can increase variability or noise in wage growth estimates. Even in an environment where wages are growing, bunching can cause the median wage, for example, to stay constant for two years at $20 per hour and then in the third year suddenly jump to $21 per hour. In that hypothetical example, median wage growth would be 0% between year 1 and year 2, and 5% between year 2 and year 3, instead of the average growth of 2.5% each year.

To reduce bunching-related problems in measuring wage levels and wage growth, this report adopts the averaged percentile smoothing method recommended by Tedeschi (2024). Specifically, we calculate a given percentile using a weighted average of 9 neighboring percentiles in the distribution. For example, the median wage (50th percentile) is calculated using the weighted average of the 46th through 54th percentiles with weights 1/25, 2/25, 3/25, 4/25, 5/25, 4/25, 3/25, 2/25, 1/25, respectively. This smooths out wage clumps in the wage distribution with an unbiased estimate of the value of a percentile. The averaged percentile method is the method of choice in EPI’s new State of Working American data library (EPI 2025f).

Appendix Figure A shows real wage growth from 2019 to 2024 at each wage percentile using the smoothed averaged percentile method and an unsmoothed percentile. When no smoothing method is applied, wage growth is extremely volatile. On the other hand, averaging the percentiles before calculating wage growth greatly reduces the observed volatility.

Appendix Figure A

Researchers have also used other methods to smooth percentiles estimated from bunched wage and income data. Autor, Dube, and McGrew (2023) use locally weighted regressions to smooth wage percentiles from the CPS. The Bureau of Labor Statistics smooths median usual weekly earnings from the CPS by linearly interpolating between $50 weekly earnings bins (BLS 2025). Similarly, in the past, researchers at EPI have estimated wage percentiles using linear interpolation between $0.25 hourly earnings bins and have estimated average wages within quintiles (Gould and deCourcy 2023; Gould and deCourcy 2024). This report and the new State of Working America data library uses the averaged percentile method described above (EPI 2025f).

An additional complication with CPS wage data is that high values of earnings are censored or “top coded” by the Census. For example, in 2022, about 6% of all wage earners and 13% of non-hourly wage earners had wage values that were top coded at the Census maximum of about $2,885. Beginning in April 2023, Census gradually transitioned to a dynamic top-code regime, adjusting the threshold every month to censor the top 3% of non-hourly and hourly wage earners.

Prior to the new dynamic top-code regime, we impute weekly earnings to those non-hourly workers by fitting year- and gender-specific Pareto distributions to weekly earnings above the 80th percentile and assigning those top-coded with the implied mean values; in 2022, those means were $4,803 and $5,903 for women and men.

Since April 2023, we impute wages for those top-coded by using the Census-provided mean above the top code; for those workers in this time period still subject to the static top code of $2,885, we impute their weekly wages by assigning the implied mean above $2,885 in a given month, using the sample of workers with dynamically assigned top codes and means. Given these new weekly earnings, we calculate hourly wages for non-hourly workers using usual hours worked at the main job.

In order to avoid how the imputation choices for high values of earnings may affect high values of hourly wages, when calculating the averaged 90th percentile in this report, we average only the 89th, 90th, and 91st percentiles, using weights 1/4, 1/2, 1/4, respectively, instead of averaging nine percentiles as we do for the other deciles.

Appendix Figure B shows the 90th percentile 2019–2024 wage growth using our preferred 89th–91st averaged percentile, as well as the 86th–94th averaged percentile, and the unsmoothed 90th percentile. Whether and how one smooths the 90th percentile has a noticeable effect on estimated wage growth because of the clumpiness of the wage distribution and perhaps because of the change in Census-provided top-code thresholds. Our preferred 3-bin 89th—91st percentile method suggests slightly faster growth at the 90th percentile than the 9-bin method and relatively similar growth to the unsmoothed method.

Appendix Figure B

A focus of this report is describing changes to wages of people in the middle of the wage distribution. Appendix Figure C reports the median wage of the smoothed averaged percentile method, the unsmoothed classic method of calculating percentiles, the quintile averaging method (the average of all people between 40th–60th percentiles) used in Gould and deCourcy 2023, and the binned linear interpolation method used in earlier EPI reports. Just as with the 90th percentile, the exact choice of method yields slightly different results. Our preferred averaged method suggests the 50th percentile grew by 5.8%, slightly below the 6.9% growth rate of the 90th percentile, but the range of estimates across methods in Appendix Figures B and C suggests that the growth rates for the two wage percentiles were broadly similar.

Appendix Figure C

Notes

1. Further, as we discuss in the appendix, wage changes at the top of the wage distribution are difficult to measure because of changes in how the Census Bureau censors high-end earnings.

2. Here and throughout this report, we measure inflation using an extended version of the Chained CPI-U, following the recently updated methodology the Census Bureau uses for its historical income series. Given this change, wage levels and trends are not directly comparable to earlier EPI reports, but differences between the points in the wage distribution—measured inequality—are unaffected. Specifically, in this report we use the annual values of the extended Chained CPI-U from version 0.19.0 of the R realtalk package (EPI 2025e).

3. In the appendix, we show that because of reported wage bunching and censoring, there is some uncertainty about the exact growth rate of the 90th percentile, with plausible estimates ranging from about 5.4% to 7.2%, all consistent with our conclusion in this report that lower-end wages grew much faster than those at the top between 2019 and 2024.

4. These findings are consistent with our State of Working America wage reports from prior years, as well as other research. See for instance, Gould and deCourcy 2024 and Autor, Dube, and McGrew 2023.

5. For all wage levels, please visit EPI’s State of Working America data library (EPI 2025f).

6. Wages in the CPS are censored—or hidden—for very high earners because of confidentiality concerns.

7. States with small increases saw nominal increases of less than or equal to 33.3% over the 2019–2024 period, while states with large increases saw nominal increases of more than 33.3% over the same period. We chose this nominal increase threshold because 33.3% is the unweighted median increase, which makes the number of states in each group relatively similar.

8. West Virginia is the one exception; the state’s minimum is higher than the federal, but its last increase was in 2015 (EPI 2025d).

9. In our state-level analysis, we exclude workers whose wages were allocated or imputed by the Census Bureau. The wage allocation model does not include the state (Census 2021), which can artificially mute or flatten differences in wages between states. 

10. In case there’s any confusion, the 10th percentile nationally is not just a weighted average of 10th percentiles in states with and without state minimum wage increases, which is why both growth rates can be lower than the overall 10th percentile reported earlier in this report. Here we calculate the 2024 employment-weighted average of the state-specific changes in the 10th percentile wage.

11. Among women, low-end wage growth in large change states was statistically significantly larger at the 5% level than in no change states, (using a one-tailed t-test, p = 0.032). For all workers, the difference was less precise (p = 0.075). For the 10th percentile for men (not shown), there was essentially no difference in wage growth between the state groups.

12. EPI analysis of Current Population Survey Outgoing Rotation Group microdata (EPI 2025a). The 10th-percentile wage in each state group is a weighted average of the states’ 10th-percentile wages. We exclude workers whose wages were allocated or imputed in these calculations.

References

Autor, David, Arindrajit Dube, and Annie McGrew. 2023. “The Unexpected Compression: Competition at Work in the Low Wage Labor Market.” National Bureau of Economic Research Working Paper no. 31010, March 2023. 

Bahn, Kate, and Carmen Sanchez Cumming. 2020. “Four Graphs on U.S. Occupational Segregation by Race, Ethnicity, Gender.” Washington Center for Equitable Growth, July 2020.

Bergh, Katie. 2025. Millions of Low-Income Households Would Lose Food Aid Under Proposed House Republican SNAP Cuts. Center on Budget and Policy Priorities, February 2025.

Bivens, Josh, Hilary Wething, and Monique Morrissey. 2025. Cutting Medicaid to Pay for Low Taxes on the Rich Is a Terrible Trade for American Families. Economic Policy Institute, February 2025.

Bivens, Josh. 2023. “Learning the Right Lessons from Recent Inflation: Why Macroeconomic Overheating Was the Problem and Austerity Is the Wrong Cure.” American Prospect, January 10, 2023. 

Bivens, Josh, and Ben Zipperer. 2018. The Importance of Locking in Full Employment for the Long Haul. Economic Policy Institute, August 2018. 

Bureau of Labor Statistics (BLS). 2025. “Handbook of Methods, Current Population Survey: Concepts.” Retrieved February 26, 2025.

Census Bureau. 2021. “Imputation of Unreported Data Items” (web page). Accessed February 2024. 

Census Bureau. 2022. “Improving Disclosure Avoidance Procedures for the Current Population Survey Public Use File.” Retrieved March 13, 2025.

Cooper, David, Sebastian Martinez Hickey, and Ben Zipperer. 2022. “The Value of the Federal Minimum Wage Is at Its Lowest Point in 66 Years.” Working Economics Blog (Economic Policy Institute), July 14, 2022.

DeCourcy, Katherine, and Elise Gould. 2025. “EPI’s Updated Family Budget Calculator Shows that States Like Virginia Need a Higher Minimum Wage.” Working Economics Blog (Economic Policy Institute), February 6, 2025.

Dube, Arindrajit, Alan Manning, and Suresh Naidu. 2020. “Monopsony and Employer Mis-Optimization Explain Why Wages Bunch at Round Numbers.” National Bureau of Economic Research Working Paper no. 24991, Revised January 2020. 

East, Chloe N., Annie L. Hines, Philip Luck, Hani Mansour, and Andrea Velásquez. 2023. “The Labor Market Effects of Immigration Enforcement.” Journal of Labor Economics 41, no. 4: 957–996.

Economic Policy Institute (EPI). 2025a. Current Population Survey Extracts, Version 1.0.61, https://microdata.epi.org. 

Economic Policy Institute (EPI). 2025b. Family Budget Calculator. Last updated January 2025.

Economic Policy Institute (EPI). 2025c. How Many Federal Employees Live in Your State? Last updated January 2025.

Economic Policy Institute (EPI). 2025d. Minimum Wage Tracker. Last updated January 1, 2025.

Economic Policy Institute (EPI). 2025e. realtalk R package, Version 0.19.0, https://economic.github.io/realtalk/.

Economic Policy Institute (EPI). 2025f. State of Working America Data Library. Last updated January 2025.

García, Uriel J. 2025. “Texas Refugee Aid Group Sues to Unfreeze $36 Million in Federal Funds.” Texas Tribune, March 3, 2025.

Gould, Elise. 2017. The State of American Wages 2016: Lower Unemployment Finally Helps Working People Make Up Some Lost Ground on Wages. Economic Policy Institute, March 2017.

Gould, Elise. 2020. State of Working America Wages 2019: A Story of Slow, Uneven, and Unequal Wage Growth over the Last 40 Years. Economic Policy Institute, February 2020.

Gould, Elise, and Katherine deCourcy. 2023. Low-Wage Workers Have Seen Historically Fast Real Wage Growth in the Pandemic Business Cycle: Policy Investments Translate into Better Opportunities for the Lowest-Paid Workers. Economic Policy Institute, March 2023.

Gould, Elise, and Katherine deCourcy. 2024. Fastest Wage Growth over the Last Four Years Among Historically Disadvantaged Groups: Low-Wage Workers’ Wages Surged After Decades of Slow Growth. Economic Policy Institute, March 2024.

Gould, Elise, and Jori Kandra. 2024. “Wage Inequality Fell in 2023 Amid a Strong Labor Market, Bucking Long-Term Trends: But Top 1% Wages Have Skyrocketed 182% Since 1979 While Bottom 90% Wages Have Seen Just 44% Growth.” Working Economics Blog (Economic Policy Institute), December 11, 2024.

Gould, Elise, Zane Mokhiber, and Katherine deCourcy. 2024. What Constitutes a Living Wage? A Guide to Using EPI’s Family Budget Calculator. Economic Policy Institute, January 2024.

Gould, Elise, and Heidi Shierholz. 2022. “The Economy Is Recovering Fast. But We Need to Ensure It Works for Everyone.” CNN Business Perspectives, March 3, 2022. 

Hersh, Adam S., and Josh Bivens. 2025. Tariffs—Everything You Need to Know but Were Afraid to Ask (fact sheet). Economic Policy Institute, February 10, 2025.

Hickey, Sebastian Martinez. 2024. “Over 9.2 Million Workers Will Get a Raise on January 1 from 21 States Raising Their Minimum Wages.” Working Economics Blog (Economic Policy Institute), December 17, 2024.

Mickle, Tripp, and Ana Swanson. 2025. “Trump’s Call to Scrap ‘Horrible’ Chip Program Spreads Panic.” New York Times, March 10, 2025.

Tedeschi, Ernie. 2024. “Introducing the Low-Wage Index: A Compositionally-Adjusted Look at Low-Wage Workers Since 1979.” Briefing Book, July 15, 2024.

Wilson, Valerie, and William Darity Jr. 2022. Understanding Black-White Disparities in Labor Market Outcomes Requires Models That Account for Persistent Discrimination and Unequal Bargaining Power. Economic Policy Institute, March 2022. 

Zipperer, Ben. 2024. “Real Value of the Minimum Wage (Adjusted for Inflation)” (web page). Accessed February 2025.

Good afternoon, Chair Cherry and members of the Senate Labor Committee. My name is Jennifer Sherer, and I’m the Acting Deputy Director, State Policy and Research, at the Economic Policy Institute or EPI. EPI is a nonprofit, nonpartisan think tank created in 1986 to research the economic status of working America and propose public policies that protect and improve economic conditions of low- and middle-wage workers. 

Thank you for the opportunity to testify today in support of Senate Bills 963, 964, and 965.

SB 965 increases protections for minors in the workplace while also expanding their right to be compensated when employers violate the law. SB 965 closes important loopholes in current laws, prohibiting employers from assigning minors to work on overnight shifts between midnight and 5 am or in jobs deemed hazardous. It also makes important changes to deter violations by increasing the cost to employers of violating the law. However, laws to strengthen child labor protections are only effective if they are adequately monitored and enforced. Fortunately, SB 964 addresses this need by updating Michigan’s work permit system to require that minors and employers register with the Department of Labor and Economic Opportunity. This streamlined registration system will enable LEO to more proactively monitor and prevent employment situations that may be hazardous or appropriate for minors.

I want to acknowledge how important it is that Michigan is taking steps to update and modernize its child labor laws in the midst of a growing child labor crisis in the U.S. Violations of federal child labor law are on the rise across the country, increasing 88 percent since just 2019. Over the past two years, the US DOL has issued judgments on dozens of cases involving hundreds of children illegally employed in meatpacking plants, in warehouses, on construction sites, at fast food franchises, and other workplaces in states across the country. Last summer alone, three teens were killed on the job while employed in violation of the law, and another teen was killed less than two months ago while doing hazardous work for a contractor at a municipal airport.

At the same time, we have seen some industry groups wage a coordinated attack on child labor protections in states across the country. Thirty states have introduced bills to weaken child labor protections since 2021, and eight states enacted legislation rolling back child labor laws in 2024 alone. Industry-backed proposals have included lifting restrictions on hazardous work, extending the number of hours per day or per week that employers can schedule youth to work (including longer/later hours or even overnight shifts during the school year), eliminating work permit requirements that facilitate compliance with the law, introducing subminimum wages for youth, and other changes.

In the face of these troubling developments, it’s become more important than ever for state policymakers to review current child labor laws and take action to strengthen them. In 2024, 20 states, including Michigan, introduced bills to strengthen child labor protections, and eight enacted them.

Proactive changes being proposed across the country include increasing penalties to deter violations, allowing victims of child labor violations to sue for damages or receive enhanced workers’ compensation benefits, restricting work hours or employment in hazardous occupations, educating youth on their workplace rights, and strengthening youth work permits. Each of these changes are helpful on their own, but they are more effective when implemented together.

Applying this lesson, we urge you to pass the three proposed bills before you and would also encourage you to consider amending SB 964 to add a requirement that as part of the registration process, all minors be informed of their workplace rights and employers hiring minors complete basic training to ensure full understanding of and compliance with the state’s child labor standards.

At a time when child labor violations are on the rise and child labor protections have come under attack in some states, we applaud Michigan’s commitment to strengthening and modernizing its child labor law. Changes proposed in the three bills before you will help solidify Michigan’s role as a leader in standing on the side of young workers and their families by setting standards that protect the rights of children to start their work lives under safe, age-appropriate conditions that safeguard their education and development.

Thank you.

By some measures, support for labor unions among the U.S. public is at its highest level in over half a century. In this report, we investigate this trend. Looking at multiple surveys over nearly 50 years, we find that U.S. workers today are much less likely to oppose union representation in their workplaces. Although there is evidence of greater support for unionization among workers, the most remarkable change is the much larger share of workers who report being unsure about whether they would vote for union representation. We call this trend “the rise of the ‘union curious.’” To better understand what today’s workers think about unions and where their ambivalence lies, we analyze the data from a recent, original survey of workers in five low-wage industries. Although many of the predictors of union support today mirror those from the past, we uncover a large generational divide that was not apparent even a few years ago. Workers 30 and under are far more likely than older workers to report both support for and uncertainty about unionization. The ranks of the “union curious”—workers who are open to, but uncertain about, the possibility that unionization can improve their lives—are large and growing. They are a pivotal group that will help determine whether the current increase in union interest results in sustained gains for working people. Whether the ambivalent ultimately become unionists will depend on whether unions are able to reach, educate, and organize these workers in traditional and perhaps new ways.

Introduction

U.S. workers are increasingly interested in unions, and in response, unions have stepped up their organizing activities. The number of strikes and strike threats has also risen, with over 450,000 workers striking in 2023. Several recent collective bargaining agreements secured large wage increases in industries ranging from parcel delivery to auto manufacturing, health care, and education, along with breakthroughs on nonwage issues. Is this upsurge a temporary, post-COVID-19 flash? And will it be smothered by employer resistance or lead to a sustained process of rebuilding worker power?

To answer these questions, we draw on recent survey data that we collected under the auspices of the Worker Empowerment Research Network (WERN) in 2022, and we compare the results of this survey with similar ones conducted recently and in prior decades. We first investigate the U.S. public’s attitudes toward labor unions, putting our new estimates in the context of longer-term trends in public opinion. We find that outright opposition to unionization has dramatically declined, with some suggestions that support may have increased.

However, we also find that an increasingly large share of workers are uncertain in their attitudes about unions and unionization. This ambivalence matters. For a union to organize, negotiate a contract, and strike successfully, a large majority of the relevant workplace must be actively supportive (McAlevey 2016). Most union organizers won’t “go public” with their campaign until at least 70% of the relevant workforce has signed a union interest card. We argue that these high levels of uncertainty—shown, for instance, in many workers responding that they “don’t know” whether they would vote for a union at their workplace—might arise from workers being unsure about whether unions can deliver the gains in wages and working conditions that they promise. These high rates of uncertainty are also consistent with the idea that, as unionization has declined, fewer and fewer workers have direct (or indirect, through family) experience with how labor unions and collective bargaining work. Ignoring the workers who express an ambivalent stance risks overstating the strength of union support while also masking the challenges and opportunities in the organizing environment.

We further investigate the correlates of workers’ support for, opposition to, and uncertainty about labor unions. We find that younger workers are both more supportive and more uncertain about unions compared with older workers. This is an especially important finding given young workers’ activism in recent organizing efforts at firms such as Amazon and Starbucks and nonprofits such as universities. In many areas of U.S. politics and policy, millennials and members of Gen Z have distinct attitudes (Grumbach and Hill 2022). These generations appear to be especially active in social movements and are turning out to vote at higher rates than members of earlier generations did in their 20s and 30s (though still at low rates overall). As we detail later in this report, harnessing the energy of young workers is a key challenge and opportunity for the labor movement.

Union approval

To put the data from our current survey in context, it is helpful to review prior polls that asked workers about their views of unions.1 For over 60 years, Gallup regularly asked about “approval” of labor unions, the longest such consistent series available. Figure A displays the percentage of respondents reporting that they approve of unions as well as those who don’t have an opinion in these historical Gallup polls (the remainder disapproved). For reference, we also include a line that shows the long-term decline in union density.2 

For nearly the entire time Gallup has asked this question, clear majorities declared their approval of labor unions. More noteworthy is the marked and sustained increase in union approval since 2010, with support in 2022 higher than at any time since 1965. Figure A also displays another important but understudied quantity: the percentage of respondents who were unsure about whether they approve or disapprove of labor unions, an issue we will take up in some detail. In the Gallup data, this proportion hovers around 10%, but it declines to about 5% in recent surveys, perhaps reflecting increased union activism and a corresponding uptick in media coverage of labor and labor issues in recent years.

Figure A

Encouraging as these estimates are, there are some concerns with the Gallup approach. The headline number masks important differences in union approval among various groups. For example, the Gallup numbers include both union members and nonunion members. Union members are (unsurprisingly) much more favorable to unions than nonmembers; Democrats and independents are more favorable than Republicans. But more importantly, as Figure A shows, union approval is disconnected from union membership. Indeed, the sustained increase in union approval since 2010 contrasts with the decrease in union membership rates from 12.4% in 2008 to 10.1% in 2022. It is, therefore, difficult to know exactly how to interpret trends in union approval.

Recent events suggest that workers supporting unions are putting their views into action. The National Labor Relations Board has certified and supervised more union organizing campaign elections each year since 2019, and the percentage of unions winning these elections increased each year, reaching 76% in 2022. The Cornell ILR Labor Tracker reported that the number of strikes and number of workers involved in strikes increased in each of the past three years (Ritchie, Kallas, and Iyer 2023). Some observers described the collectively bargained wage increases and benefit improvements as the “hot labor summer” (Dean 2023) and the year of the “great reset” (Greenhouse 2023). Yet, despite all this organizing activity and higher settlements in collective bargaining, the percentage of workers organized into unions has not increased. We will return to this important observation as we discuss the key lessons we take away from these developments to date.

Would you vote for a union?

If union approval is hard to interpret, are there other ways we could learn about whether nonunion workers want to be in a union? The Gallup question asks about unions in general. We turn to a regularly used survey question that instead asks workers whether they would vote for union representation at their own workplace. This is a helpful way to ask about union support rather than mere abstract approval, especially when we are concerned with building worker voice.

We found six prior national surveys that asked whether workers would vote for union representation at their jobs.3 The selected surveys include all those for which we could find the data for calculating the proportion of employed nonunion workers answering “yes,” “no,” or some variation of “don’t know.” The first dates back to the Quality of Employment (QoE) Survey in 1977 (Kochan 1979; Quinn and Staines 1977). Similar versions of the union vote question were asked of different populations using different survey modes in 1994 (Freeman and Rogers 2006), twice in 2018 (Hertel-Fernandez 2020; Hertel-Fernandez, Kimball, and Kochan 2020), 2022 (Ahlquist, Grumbach, and Thai 2023), and 2023 (Diaz-Linhart et al. 2023). Table 1 summarizes important differences across these surveys, including the survey mode and target population.

Table 1

Figure B displays the estimated level of union support across these surveys. In contrast to past research, which tends to ignore the “don’t know” (DK) responses, we explicitly call them out here.

Figure B

Figure B uncovers three important facts. First, across surveys, willingness to vote for a union is substantially lower than “union approval” as reported in the Gallup data. It appears that Americans approve of unions in general, but when nonunion workers are asked whether they personally want unionization at their workplace, there is more uncertainty and weaker support. 

Second, regardless of differences in the survey mode or target population, Figure B suggests that there has been large and sustained erosion of opposition to unionization among nonunion workers. In the last few years, outright opposition to unionization is a distinctly minority position in contrast to the late 1970s, when large majorities of the nonunionized stated that they would not vote for a union. In the late 1970s, opponents outnumbered supporters by nearly 2.4 to 1 if one were to look only at those expressing an opinion for or against unionization. In the 2022 Worker Empowerment Research Network survey, we see 1.4 union supporters for every worker opposed—a remarkable change in support.

Third, and most intriguing, the data also suggest that much of the opposition to unionization may have been replaced by the rise of the “union curious”—those expressing uncertainty or ambivalence—as opposed to an increase in outright support. In three of the four most recent surveys, “don’t know” is the most frequently chosen response by a substantial margin. We investigate this trend in further detail in the next section.

Workers’ uncertainty about unions

How should we interpret this increase in uncertainty about whether or not to unionize?

One possibility is respondent reluctance to answer definitively on a sensitive topic. Respondents may choose “don’t know” when in fact they lean in a particular direction.4 The evidence suggests that this is not the case with the “union curious” in these surveys. The 2017 Worker Voice Survey (Kochan et al. 2019) asked the union vote question without allowing for a “DK” option. In that survey only 3% of respondents refused to answer, suggesting that reluctance is not a major concern. When respondents were forced to choose, the Worker Voice Survey found a roughly even balance between union supporters and opponents (48% versus 52%).

Two of the surveys in Table 1 asked “DK” respondents for more information to better understand their responses. In both cases, most “DK” respondents are, in fact, legitimately uncertain about what unions do or what effects they might have on their jobs. Specifically, in the 2022 WERN survey, we probed a subset of the respondents who answered “don’t know” to the union vote question, asking whether they “leaned” toward voting for or against unionization. Over 80% of those we asked said they were actually uncertain. Similarly, Diaz-Linhart et al. (2023) asked those responding “don’t know” to provide a reason for that response and found that 70% of the “DK” responders indicated they didn’t know enough about unions or how they might affect their job to make a definitive “yes” or “no” response.

In a third survey, which is not included in Table 1, the think tank American Compass asked U.S. workers if they would vote for a union in their workplace and allowed respondents to answer “lean toward” or “lean against” unionization. They also find that opposition to unionization is the minority position and that “undecided” is the modal response among lower income workers.5 Thus the “union curious” appear to be a group that holds potential for building worker voice if they can be persuaded that unions can effectively address their concerns.

The apparent rise of the “union curious” poses important questions as we seek to understand the recent wave of union activism and the prospects for union renewal. But we need to be cautious about overstating the case. Recent national surveys tend to be administered over the internet rather than by telephone or in person. There is some suggestive evidence that survey mode may affect the frequency of “don’t know” responses (Ansolabehere and Schaffner 2014; De Leeuw and Hox 2014). It is possible that a visible response option for “don’t know” in an online survey produces greater rates of “don’t know” responses than when respondents are interacting with a survey administrator by telephone or in person. However, the 2018 Worker Organization Study (WOS) was administered by the National Opinion Research Center (NORC), which goes to greater lengths to track down and interview respondents who are online less often. The WOS survey nevertheless found “union curious” respondents at about the same rate as the other recent surveys using purely web-based samples. The prevalence of the “DK” responses in the recent surveys is underappreciated and sufficiently important to warrant further exploration.

Who are the union supporters and the ‘union curious’? Evidence from the 2022 WERN Workers’ Survey

Who are the union supporters? How do they differ from the “union curious” and those opposed to unions in their workplaces? We turn to an original survey of U.S. workers that we fielded in 2022 under the auspices of the Worker Empowerment Research Network (WERN). This survey is noteworthy because it concentrates on frontline workers in select industries in the post-COVID-19 period, whereas nearly all other similar recent surveys use broad national samples. Our survey sample allows us to drill down into five specific industries known for low wages, scheduling instability, and intense recent unionization drives: health care, hospitality, retail, telecommunications, and warehousing.6 The WERN sample also includes a variety of modules on job satisfaction, voice at work, and major problems at work (wage theft, harassment, scheduling instability, etc.). The WERN survey is, therefore, particularly well suited to understanding the “union curious.”

As a first cut, we look at responses to the union vote question across several potentially relevant demographic groups to see whether union supporters and the “union curious” tend to come from particular industries or parts of the population.

Our first key finding in the Worker Empowerment Research Network sample echoes similar recent surveys (Diaz-Linhart et al. 2023; Kochan et al. 2019): Younger workers are significantly more pro-unionization than older workers.7 Figure C makes this plain. In the WERN sample, over 40% of nonunion workers 30 and under are outright supporters of unionization, whereas only 32% of older nonunion workers hold the same view; union opposition is much stronger among older workers than among younger ones. But the “union curious”—the “DK” responders—comprise the largest share of both age groups. Over 45% of younger workers are “union curious.” Taken together, the WERN data show that younger workers are less opposed to unionization than older workers, with this age gap explained by higher levels of union support and more “union curiosity” among younger workers. This presents an interesting challenge for worker advocates: Younger workers are likely to be the most receptive to pro-union messages, but for unionization drives to succeed, both younger and older workers will need to be engaged. Younger workers will need to play important roles in persuading their older co-workers. 

Figure C

Figure D builds on the importance of age, again looking at nonunion workers in the WERN sample. The figure displays the proportions of union supporters and the “union curious” by gender, industry, and race, comparing younger (blue dots) with older (open diamonds) workers within each demographic subgroup. Across all these demographic subgroups, union support exceeds opposition. The age gap in union support (distance between the dot and diamond) is largest in telecommunications and health care. Looking at different racial groups, we see relatively high levels of union support among Black workers, whereas we see larger age gaps in union support among white workers.

But the “union curious” are a critical group. Among women of both age groups, the “union curious” are the largest group. They are an outright majority of young workers in retail and hospitality. The age gap in union uncertainty is largest among warehousing workers and Hispanic workers.

Figure D

We further explored the determinants of support for unionization using the tools of regression analysis.8 In this way, we can describe how different variables relate to the likelihood that a worker will support unionization (or oppose it or report uncertainty), holding other variables fixed. Regression models can be complicated, so we present graphical displays that describe how key variables relate to responses on the union-vote survey question in our survey sample. The shaded regions present a measure of our statistical uncertainty (95% confidence intervals). 

First, we confirm a common, repeated finding using our data: Bad jobs increase the likelihood of voting for the union and reduce both uncertainty and resistance to unionization. In our case, we asked respondents whether they had experienced a series of problems on the job in the last six months, ranging from underpayment to schedule instability to harassment. We count the number of problems a worker reported experiencing as a summary indicator. The left column of Figure E displays the relationships between problems at work and the likelihood of supporting, opposing, or being uncertain about unionization. A worker who experienced four different problems at work has double the expected probability of voting for a union compared with a worker who reported no major problems at work (60% versus 30%), holding other attributes fixed. A worker reporting four problems at work also had about 10 percentage points lower probability of being uncertain about unionization and was about one-third as likely to oppose unionization as a worker reporting no recent workplace problems. All of these relationships are quite large relative to statistical uncertainty.

Figure E

Second, a lack of voice on the job also increases support for unionization and reduces both uncertainty and opposition. To establish this, we asked respondents whether they had “the right amount,” “some,” or “no” say over a series of work conditions, including pay, benefits, scheduling, promotions, and safety. We constructed an index across all these domains to summarize the extent to which workers lacked the desired amount of voice on the job, something we call the “voice gap index.” The right column of Figure E displays the relationship between the voice gap and support for unionization. A worker with a job where she lacks voice across all domains is nearly 20 percentage points more likely to support unionization and 13 percentage points less likely to oppose it than a worker who reports “the right amount” of voice. Uncertainty about unionization also declines as the voice gap increases.

Finally, we asked workers in our sample about their beliefs about unions and expectations about how unionization might change their working conditions. By examining these responses, we get a sense of what drives the “union curious” and where more information and experience with unions might tip uncertainty into support.

We asked respondents how much they agreed or disagreed with a variety of statements about union in general. Figure F displays the summary of the outcomes. Reassuringly, large majorities “agreed” or “strongly agreed” that unions help low-paid workers, improve corporate social responsibility (CSR), and provide more voice at work. Nevertheless, about a quarter of respondents said they were “neutral” in their beliefs on these topics, something we take to be connected to uncertainty or ambivalence. Smaller majorities agreed that unions give “people like me a voice in how laws and policies are made” and that unions “protect incompetent workers,” with roughly a third of respondents “neutral” on those topics. More concerning for unions, large pluralities are “neutral” about whether unions reduce racial discrimination or are “controlled by the Democratic party,” pointing to important ways that broader social conflict and polarization connect with uncertainty about unions.

Figure F

We also asked workers to report how they think unionization at their own workplace would affect working conditions and relationships. We display responses in Figure G. Consistent with past work and the responses above, majorities of workers expect that unionization would improve wages and benefits “a little” or “a lot.” But around the subject of job security, training, and workplace safety, large proportions of workers (between 41% and 51%) state that they don’t think unionization would have any effect at all. Similarly, large majorities or pluralities don’t think unionization would have any effect on workplace relationships, including relationships with co-workers. All of this suggests that there is a large population of workers with a limited understanding of what successful unions require and what they can deliver.

Figure G

Organizing the ‘union curious’

The central conclusion and the implication we draw from the increase in survey takers’ responses of “don’t know” to questions about how they would vote in a union election is this—unions have a major educational challenge to overcome if the current period will be one that produces a sustained revival and rebuilding of union density in America. Moreover, the challenge and the opportunity are greatest among young, lower-income workers. These are workers who are both more likely to vote “yes” and to vote “don’t know” or “unsure” in recent surveys.

But pursuing this targeted opportunity will require investment of substantially more resources than unions currently expend on organizing. This is not a new critique; others have stressed this point. Given the rise of the “union curious,” a clear need exists for information and education about what unions do and how to form them; this information is likely to be especially important for younger workers, even those who express general approval for unions.

It also may require new approaches to organizing and recruiting union members. The difficulties of gaining 50% majorities and then negotiating first contracts when faced with fierce and determined employer opposition cannot be overstated. The model of organizing one location at a time (e.g., one retail establishment of a giant firm; one plant of a large, multisite manufacturing firm; one warehouse of a giant distribution firm) is an equally daunting and, to date, largely unachievable task. 

A number of alternative approaches to organizing have been suggested, and some are now underway. Agreements with companies to be neutral (often called neutrality agreements) in organizing campaigns, such as the one recently agreed to by Microsoft, are major steps forward. So too is the recent announcement by Starbucks and Starbucks Workers United that they have reached a framework agreement for negotiating contracts covering all the stores in the company that have organized or will organize. Labor advocates in California, New York, Minnesota, and several cities have been successful in establishing industrywide labor standards boards for fast food, health care, and nail salons and have combined these with efforts to recruit workers at the establishment level. The United Auto Workers is attempting to organize the full set of nonunion automakers. Workers at Alphabet have organized a union that speaks out on issues of concern to workers in the company without seeking the majority status needed to gain rights to collective bargaining.

Absent more alternative organizing strategies like these or others, it is difficult to see how the current upsurge in worker support for and curiosity about unions will be enough to generate a sustained process of rebuilding union density and membership across the economy. And there are substantial numbers of “union curious” workers awaiting more information, contact, and innovative organizing efforts that can address the pent-up demand for a greater voice at work. Today, the pressing question facing the labor movement and labor scholars is this—how can the curious be turned into actual union members?

About the authors

John S. Ahlquist is professor of political economy at UC San Diego’s School of Global Policy and Strategy. Current research projects include “decent jobs” in the modern economy, the puzzle of employer-sponsored worker mutual aid funds, and union membership at the county and commuting zone levels. His book, In the Interest of Others (with Margaret Levi), studies how some labor unions are able to provoke actions that transcend members’ private interests, building an expanded “community of fate.”

Jacob M. Grumbach is associate professor at the Goldman School of Public Policy at UC Berkeley. He was previously associate professor of political science at the University of Washington and a postdoctoral fellow at the Center for the Study of Democratic Politics at Princeton. Grumbach studies the political economy of the United States, with interests in democracy, public policy, racial and economic inequality, American federalism, and statistical methods. His book, Laboratories Against Democracy, investigates the nationalization of state politics over recent decades.

Thomas A. Kochan is the George Maverick Bunker Professor Emeritus at the MIT Sloan School of Management and a faculty member in the MIT Institute for Work and Employment Research. His recent work calls attention to the need for a new social contract at work, one that anticipates and engages current and future technological changes in ways that build a more inclusive economy and broadly shared prosperity. His most recent book is Shaping the Future of Work: A Handbook for Action and a New Social Contract. 

Acknowledgments

Support for this research was provided to the Worker Empowerment Research Network from the WorkRise program of the Urban Institute and the Ford, Hewlett, and Annie E. Casey Foundations.

Notes

1. There are earlier surveys of public opinion data about unions (Jarley and Kuruvilla 1994; Panagopoulos and Francia 2008).

2. Union density is the percentage of the employed workforce who are members of labor unions.

3. An additional survey, the 2017 Worker Voice Survey, found that 48% of respondents who answered the question would vote for a union at their workplace and 52% would vote against one. However, this survey did not include a “don’t know” option (3% of those completing the survey refused to answer the union vote question). We, therefore, do not include it as a direct comparison to the other surveys that did provide “don’t know” options. We note that the roughly equal balance of “yes” and “no” responses to the union vote question in the 2017 survey is consistent with the other surveys fielded in the late 2010s, as shown in Table 1.

4. U.S. public opinion surveys asking about political party identification regularly allow respondents to “lean” one way or another.

5. We do not include the American Compass survey in Figure B, as American Compass does not break out findings by union membership status, and their data are otherwise unavailable.

6. Survey respondents were recruited from the Qualtrics online panel with quotas for each of the targeted industries, age, race, and gender, ensuring broad representation for workers of color as well as younger workers across these industries. We then weighted the sample to match the national demographics of workers in each of the five industries based on gender, age, and race. The survey was in the field from June 21 to August 22, 2022, and received 2,561 completed responses.

7. See Ahlquist, Grumbach, and Thai (2023) for a detailed analysis using the WERN data of how younger and older workers differ in their experiences at work.

8. All of the results below represent predicted probabilities generated from multinomial logistic regression models that include age, gender, race, industry, college education, and Republican partisan identification as the other covariates.

References

Ahlquist, John S., Jacob M. Grumbach, and Eric Thai. 2023. Voice on the Job for Younger Workers. UC San Diego School of Global Policy and Strategy, July 2023.

American Compass. 2023. Labor Market Not Yet Working for Workers: New Data on Job Quality and Worker Views on Unions. October 2023.

Ansolabehere, Stephen, and Brian F. Schaffner. 2014. “Does Survey Mode Still Matter? Findings from a 2010 Multi-Mode Comparison.” Political Analysis 22, no. 3: 285–303. doi:10.1093/pan/mpt025.

Dean, Sam. 2023. “Corporate Greed, Low Unemployment, Housing Crisis: That’s the Recipe for Hot Labor Summer,” Los Angeles Times, August 11, 2023.

De Leeuw, Edith D., and Joop J. Hox. 2014. “Survey Mode and Mode Effects.” In Improving Survey Methods, edited by Uwe Engel, Ben Jann, Peter Lynn, Annette Scherpenzeel, and Patrick Sturgis. New York: Routledge.

Diaz-Linhart, Yaminette, Arrow Minster, Dongwoo Park, Duanyi Yang, and Thomas Kochan. 2023. “Bridging the Gap: Measuring the Impact of Worker Voice on Job-Related Outcomes.” MIT Working Paper.

Freeman, Richard B., and Joel Rogers. 2006. What Workers Want. Ithaca, N.Y.: ILR Press.

Gallup. 2024. “Do You Approve or Disapprove of Labor Unions?” [table], Labor Unions (web page). Accessed May 31, 2024.

Greenhouse, Steven. 2023. Labor’s Great Reset. The Century Foundation, November 9, 2023.

Grumbach, Jacob M., and Charlotte Hill. 2022. “Rock the Registration: Same Day Registration Increases Turnout of Young Voters.” Journal of Politics 84, no. 1: 405–417.

Hertel-Fernandez, Alexander. 2020. Power and Politics in the U.S. Workplace. Economic Policy Institute, October 2020.

Hertel-Fernandez, Alexander, William Kimball, and Thomas Kochan. 2020. “What Forms of Representation Do American Workers Want? Implications for Theory, Policy, and Practice.” ILR Review 75, no. 2: 267–294. doi.org/10.1177/0019793920959049.

Jarley, Paul, and Sarosh Kuruvilla. 1994. “American Trade Unions and Public Approval: Can Unions Please All of the People All of the Time?” Journal of Labor Research 15, no. 2: 97–116.

Kochan, Thomas A. 1979. “How American Workers View Labor Unions.” Monthly Labor Review 102, no. 4: 23–31.

Kochan, Thomas A., Duanyi Yang, William T. Kimball, and Erin L. Kelly. 2019. “Worker Voice in America: Is There a Gap Between What Workers Expect and What They Experience?” ILR Review 72, no. 1: 3–38. doi:10.1177/0019793918806250.

McAlevey, Jane F. 2016. No Shortcuts: Organizing for Power in the New Gilded Age. New York: Oxford University Press.

Organisation for Economic Co-operation and Development (OECD) and Amsterdam Institute for Advanced Labour Studies (AIAS). 2021. “Institutional Characteristics of Trade Unions, Wage Setting, State Intervention and Social Pacts” (web page). Accessed May 31, 2024.

Panagopoulos, Costas, and Peter L. Francia. 2008. “The Polls-Trends: Labor Unions in the United States.” Public Opinion Quarterly 72, no. 1: 134–159.

Quinn, Robert P., and Graham Staines. 1977. “Quality of Employment Survey, 1977: Cross-Section.” doi: https://doi.org/10.3886/ICPSR07689.v1.

Ritchie, Kathryn, Johnnie Kallas, and Deepa Kylasam Iyer. 2023. Labor Action Tracker Annual Report 2023. ILR School, Cornell University and LER School, University of Illinois, Urbana-Champaign.

Key findings:

• In the pandemic recovery, young high school graduates have experienced a much faster rebound in job prospects and stronger wage growth than any recovery in recent history.
  • The unemployment rate for young high school graduates—defined as workers ages 18 to 21—recovered in two years in the pandemic recovery compared with almost 9.5 years following the Great Recession of 2008–09. Meanwhile, the underemployment rate recovered more than five times faster in the pandemic recovery than the aftermath of the Great Recession.
  • Young high school graduates experienced 9.4% real (inflation-adjusted) wage growth between February 2020 and March 2024.
• Gaps in labor market outcomes across race and ethnicity and gender persist even among high school graduates who have the same basic level of education and little variation in professional experience.
  • The unemployment and underemployment rates of Black, Hispanic, and AAPI young high school graduates are much higher than their white counterparts.
  • On average, Black workers are paid 93.2% of what white workers are paid per hour, while women are paid 87.6% compared with their male counterparts.

As with young college graduates, young high school graduates are experiencing a much stronger labor market today than before the pandemic and at any point since 2000. The fast economic recovery from the pandemic shock is a direct result of the aggressive fiscal policy response that matched the scale of the problem—in stark contrast to policy responses following previous recessions.

In this blog post, we start by examining employment and enrollment outcomes for young high school graduates, defined as workers ages 18 to 21. We then analyze their short- and long-run trends in unemployment, underemployment, and wages, looking at those with only a high school degree and who are not enrolled in further schooling.1

To most accurately capture the choices that young high school graduates are making, we include all young people between the ages of 18 and 21 who have less than a bachelor’s degree (including those with some college) in our initial sample. We group this population into four categories: “employed only” and not enrolled in further schooling, “enrolled only” and not employed, employed and enrolled, or “idled” (not enrolled and not employed, which includes the unemployed). Among these young high school graduates, most are either “employed only” and not enrolled in further schooling (32.7%) or “enrolled only” and not employed (31.1%). Since 1989, the share of young high school graduates who are “employed only” has fallen 11.8 percentage points, while the share of those who are “enrolled only” has risen 10.1 percentage points. As of March 2024, the share of young high school graduates who are employed and enrolled (21.5%) and idled (14.7%) remains roughly similar from 1989 (21.0% and 13.5%, respectively).

Next, we narrow the sample of young high school graduates to those who are not currently enrolled in further schooling to assess the labor market for those making work their primary activity. The unemployment and underemployment rates for young high school graduates rose sharply during the Great Recession and the pandemic (Figure A). The unemployment rate reflects the share of the young high school graduate population who are jobless yet have reported that they are actively seeking work. Unemployment rates for high school graduates recovered nearly five times faster after the pandemic (two years, February 2020 to February 2022) compared with the Great Recession of 2008–09 (9.5 years, December 2007 to May 2017).

Figure A also shows that the underemployment rate for young high school graduates recovered in just under two years after the onset of the pandemic (February 2020 to December 2021) but took more than 10.5 years after the Great Recession (December 2007 to July 2018). This rate includes the unemployed plus “involuntary” part-timers (those who work part time but want full-time work) and “marginally attached” workers (those who want a job and have looked for work in the last year but have given up actively seeking work in the last four weeks and therefore are not captured in the official unemployment rate). Notably, the underemployment rate of young high school graduates has never returned to its low point in the tight labor market in 2000.

Figure A

Despite the fast recovery in the pandemic recession, gaps still remain between Black, Hispanic, and Asian American and Pacific Islander (AAPI) high school graduates and their white counterparts, as shown in Figure B. If labor markets were well-functioning and free of discrimination, we would expect little disparity in the unemployment rates of young high school graduates given that they have the same basic level of education and are in the same labor market position. However, in the 36 months ending in March 2024, Black (18.8%), AAPI (12.8%) and Hispanic (12.0%) graduates have much higher unemployment rates than white graduates (9.5%). This suggests that other factors such as discrimination or unequal access to job networks and opportunities play a role. Young Black high school graduates’ underemployment rates are significantly higher than any other group, which demonstrates that a much larger share of Black graduates is either discouraged from the job search or is working part time when they would rather work full time.

Figure B

Young high school graduates experienced real (inflation-adjusted) wage growth for the first time this early in a recovery compared with the prior three business cycles (Figure C). Even after accounting for unusually fast inflation, real wage growth between February 2020 and March 2024 was 9.4% for young high school graduates. By comparison, young high school graduates faced stark wage losses of 10.6% after the Great Recession, 4.3% between 2001–2005, and 5.5% between 1990–1994. This means that real wage growth in the pandemic business cycle was a tremendous 20.0 percentage points faster than in the aftermath of the Great Recession, 13.7 percentage points faster than between 2001–2005, and 14.9 percentage points faster than between 1990–1994. 

Figure C

Despite this positive wage growth, Black-white and gender wage gaps remain significant even among young high school graduates who have equal levels of education and very little variation in work experience. Compared with young college graduates, young high school graduates experience smaller wage gaps across race and ethnicity and gender in part due to crucial policies such as the minimum wage. Figure D shows that Black workers are paid $1.15 less per hour, on average, than white workers and women are paid $2.22 less per hour than their male counterparts. This translates to Black workers being paid, on average, 93.2% of average white workers’ wages and women being paid, on average, 87.6% of their male counterpart’s wages.

Figure D

These findings underscore the immense role that large fiscal relief and recovery packages, including expanded unemployment insurance coverage and aid to state and local governments, had in healing the labor market after the pandemic recession. For the first time in a recession over the past 35 years, young workers were not left behind by policy. However, policymakers must prioritize sustained full employment, increase the federal minimum wage, strengthen and enforce labor standards, and make it easier for workers to come together and form unions to ensure that these gains are fully realized and that racial and gender wage gaps are addressed.

Note

1. Notes about our data sample: Because we are examining such a small subset of the population, we pool 12 or 36 months of data to increase the sample size and mitigate some of the volatility in the series. Unless otherwise specified, when looking at “overall” trends in the data, we pool 12 months of data to create a pooled moving average, which also has the added advantage of removing any seasonal effects. We use 36-month pooled data to look at trends by gender and race/ethnicity, since breaking the population down by demographics reduces sample size and data reliability.