Are Americans still looking? Revisiting long-term unemployment

Last year, we explored the trajectory of long-term unemployment—the status of those who have been seeking employment for 27 weeks (about six months) or more. At that time, long-term unemployment had completed a remarkable recovery from highs following the COVID-19 recession, but it had started to climb, a fact that was unusual relative to other business cycles.

Since then, the U.S. economy has experienced a number of significant events, but the labor market remained stable and has even shown signs of strengthening.¹ That makes it a good time to ask, Where is long-term unemployment now?

Long-term unemployment in 2026

Figure 1 shows the monthly long-term unemployment rate—the share of the U.S. labor force experiencing long-term unemployment—from the beginning of 2018 to September 2026. The long-term unemployment rate had returned to its pre-pandemic level around mid-2024. However, the rate crept up until it reached 1.1 percent at the end of summer 2025, the time of our earlier analysis. It has held steady since then, fluctuating between 1.0 and 1.2 percent since August 2025. This September, the long-term unemployment rate again came in at 1.1 percent.

The long-term unemployment rates of the past year are moderate compared with spikes in this measure that have happened during economic downturns. Instead, what stands out about the current episode is that long-term unemployment has climbed as a share of overall unemployment. This is called the long-term unemployed share. It reflects the prevalence of long-term unemployment among workers who are unemployed generally.

Figure 2 shows that this share has been climbing since early 2023, even as the long-term unemployment rate and the overall unemployment rate have leveled off. Historically, increases in the long-term unemployed share have followed recessions, but the current rise has occurred outside of a downturn. Figure 2 shows this for the last four recessions. Even over a longer time frame, the current period is an exception.

Who have seen their long-term unemployed shares rise?

To better understand the increase in the long-term unemployed share, we take a closer look at who is experiencing long-term unemployment today and compare them with workers in the same situation in the past.

We define groups of workers at different points in time. Today’s group consists of individuals who reported being unemployed for at least six months when interviewed.² We compare them with workers reporting long-term unemployment after the Great Recession, at a time when long-term unemployment also stood at roughly 1.1 percent.³ We label this the post−Great Recession group. We also compare today’s long-term unemployed workers with those from a two-year period just before the pandemic.⁴

Looking at these groups gives us snapshots at three points: a time when the long-term unemployment rate was similar to today’s, though trending down; the point just before the COVID recession; and the most recent 12 months.

Within each period, we focus on groups defined by age and education. These have been the subject of much discussion over the past year, as economists consider whether demand for labor is affecting the groups differently.

We first assess whether the prevalence of long-term unemployment differs across groups. Figure 3 shows the long-term unemployed shares for age-and-education groups at our three points in time.

Some clear patterns emerge. First, long-term unemployment is less common among younger unemployed workers (gold lines). The long-term unemployed share for the youngest group is below that for the two older groups for every education group (panels) and in every time period (dots). The opposite is nearly true for the oldest group: Their long-term unemployed shares (teal lines) are greater than the shares for younger groups in most periods.

We also see that long-term unemployment as a share of all unemployment is more common now than before the pandemic for most groups. This increase is largest for midcareer workers with four-plus years of college. Their long-term unemployed share now stands at 29 percent, compared with 22 percent in the 2018−19 period. However, all midcareer workers have seen their long-term unemployed share rise since 2018−19, and young workers have seen this share rise or move sideways. For most age-and-education groups, then, long-term unemployment is a larger share of unemployment now than it was just prior to 2020.

Older workers with fewer than four years of college are the exception to this increase. They are less likely to be reporting long-term unemployment now than in the pre-pandemic period. This decline is 5 percentage points for the group with the least formal education and 6 percentage points for the group with some college education.

Taking a longer view, today’s long-term unemployed shares are comparable to or lower than at our post–Great Recession comparison point. However, midcareer workers with four years of college or more are again the exception here. Their long-term unemployed share is now 6 percentage points above where it stood a decade ago.

Who are the long-term unemployed workers?

Has the sharp rise in long-term unemployment for midcareer workers with four years of college or more changed the face of long-term unemployment generally? To assess this, we ask whether any of the groups in Figure 3 are overrepresented among long-term unemployed workers. We call this “excess long-term unemployment,” and we compute it as the gap between a group’s share of long-term unemployed workers versus their share of employed workers. For instance, if a group makes up 20 percent of the employed workers and 25 percent of long-term unemployed workers, then their excess long-term unemployment is 5 percentage points.⁵

Figure 4 graphs excess long-term unemployment for the same groups and time periods explored above. Bars above zero indicate that an age-and-education group is overrepresented among long-term unemployed workers: That is, their share among all long-term unemployed workers is higher than their share among all employed workers for the time period indicated. Bars below zero indicate that a group is underrepresented.

Among older workers, the right panel of Figure 4 shows that long-term unemployment shares are roughly balanced with employment shares for all education groups in all three periods. Among younger and midcareer workers, the story is different. Workers with a high school education or less are significantly overrepresented among long-term unemployed workers, while those with four years of college or more are underrepresented.

Representation gaps are largest among midcareer workers. In the post-COVID period, the share of workers with a high school education or less among all long-term unemployed workers exceeds their share of employment by over 8 percentage points. Workers with four years of college or more are underrepresented by about 5 percentage points, for a gap of 13 percentage points in the current period. For young workers, representation gaps are smaller but still large—about 10 percentage points today. By contrast, workers with some college are roughly equally represented among workers who are long-term unemployed and employed in each age group.

Finally, Figure 4 shows these gaps have shifted. Notably, the underrepresentation of workers with four years of college or more has declined over time. The representation gap between midcareer workers with the most and least education is 4 percentage points narrower today than following the Great Recession.

A role for industry?

While workers with a four-year college education remain underrepresented among long-term unemployed workers, our analysis shows this advantage has eroded somewhat over time. A reasonable question is whether shifting tasks toward artificial intelligence may be contributing to more long-term unemployment for the most educated workers. To look for evidence of this, we reviewed excess long-term unemployment by sector in separate analysis.⁶ Sectors exposed to AI are not systematically associated with more long-term unemployment in the current period compared with a decade ago, after accounting for overall sector size.

What do changing patterns say about long-term unemployment today?

The period since 2023 stands out historically as the only time long-term unemployment grew as a share of all unemployment without a preceding recession. These long-term unemployed shares are now higher for most groups of workers defined by age and education than they were prior to the pandemic but about on par with a comparison period following the Great Recession.

The change is more acute for the most-educated midcareer workers. This group has seen its long-term unemployed share rise 7 percentage points since the pandemic. It is similarly elevated relative to a decade ago. The most-educated workers remain underrepresented among long-term unemployed workers despite these shifts, but their advantage relative to other workers has narrowed.

These patterns suggest that some of today’s long-term unemployment among highly educated workers is part of a long-run trend. On the other hand, the marked rise in the long-term unemployed share of midcareer workers may be due to their particular position in their careers at this point in time. They are too young to retire, they are too old to change direction or take a short-term job, and the trend is against them.

The picture is different for less educated, older workers. Their long-term unemployed share has fallen over time, both going into the pandemic and again today. After facing high rates of long-term unemployment following the Great Recession—related to delayed retirements after asset values fell—today this group is likely retiring rather than engaging in an extended search for new work.

Aidan Bryant provided research assistance for this article.

Endnotes

1 The view that the labor market is stable or strengthening has been shared recently by Minneapolis Fed economists Amanda Michaud and Simon Mongey in Conversations with the Fed; remarks by New York Fed President John Williams at the University of Buffalo; and San Francisco Fed economist Thomas Mertens in SF FedViews.

2 Specifically, these are respondents in the Current Population Survey who report that they have been unemployed for 27 weeks or more at the time of their first interview between October 2025 and September 2026.

3 We choose a point for comparison when the long-term unemployment rate was at a similar level, to abstract from the ways that a higher or lower level of long-term unemployment might change the composition of this group. This corresponds to the 12-month period from January 2016 to December 2016.

4 This period spans January 2018 through December 2019.

5 Excess long-term unemployment is computed within our three time periods, based on a group’s shares of all employed or all long-term unemployed in that period. The equation for the excess share is as follows:

where ea indicates an age-and-education group, t indicates time period, LTU denotes long-term unemployed, E denotes employed, and N is the number of workers in the categories indicated by super- and subscripts in each period t.

6 Comparing sectors of employment over long time periods is difficult due to changing classification schemes as work changes over time. We aggregate to a stable set of sectors (two-digit industries), but changes in assignment of workers within those sectors may mean that some differences across time are due to shifting classification of work. AI use is also widespread, and its functionality can be applied in many sectors (Broady 2026). Hence sectoral comparisons may understate the impact of AI.