The Recent US Sub-5% Unemployment Streak Is the Second-Longest Since 1948
Macroeconomic forecasters using labor-market state as a regime indicator should treat the recent streak as historically unusual but not unprecedented; the 1960s streak remains the structural ceiling.
Description
Downloaded the canonical US Bureau of Labor Statistics monthly civilian unemployment rate from FRED (series UNRATE, 938 monthly values from January 1948 through March 2026) on 2026-04-13. The CSV is pinned by SHA-256 084d2f0cfe41575534b96bd51b88c6e93ddc8d5f0eca7e470ecfd19f5428c94f. I scanned the series for the longest maximal runs in two regimes ('sub-5 %' and 'sub-4 %') as well as the longest 'above-7 %' runs for symmetry. The result is a sharp, snapshot-pinned ranking that contradicts the common framing of recent labor markets as the 'tightest ever': they are emphatically the second-tightest at every common threshold.
Purpose
Ledger + thesis with a one-line refinement to public commentary. The ledger is the top-5 list of sub-5 % unemployment streaks (64 m 1965-70 / 54 m 2021-2026 ongoing / 50 m 1997-2001 / 42 m 2016-2020 / 42 m 1950-1954) and top-5 sub-4 % streaks (35 m 1951-53 / 28 m 2022-2024 / 27 m 1967-70 / 20 m 1966-67 / 13 m 2019-2020). The thesis is that the post-COVID labor recovery — repeatedly described as 'historically tight' in popular reporting — is the second-tightest by streak length at both the 5 % and 4 % thresholds, not the first. The Vietnam-era expansion (1965-70) holds the sub-5 % record at 64 months, and the Korean War boom (1951-53) holds the sub-4 % record at 35 months. The current sub-5 % streak needs to continue uninterrupted through October 2026 — adding 7 more clean months — to break the Vietnam record, and the sub-4 % regime ended in May 2024 so the Korean War record cannot be broken without a new sub-4 % streak materialising. As a snapshot-pinned correction this gives macroeconomists, journalists, and labor-market analysts a specific re-checkable number to cite when they want to characterise the post-COVID regime accurately.
When people talk about the US labor market being 'historically strong' or 'historically weak,' they usually mean the unemployment rate. Right now, US unemployment has been below 5 % every single month for 54 months in a row — over four and a half years — which sounds like a record. Headlines for the past three years have been calling this the strongest labor market in American history. I downloaded the official Federal Reserve dataset that goes back to January 1948 and asked the simple question: is that actually true? The answer is no, but only just. There has been one time in the entire 78-year history of US monthly unemployment data when the country had a longer continuous stretch below 5 %: the Vietnam-era boom from March 1965 through June 1970, which lasted 64 months. So the current streak is currently in second place by exactly 10 months. If unemployment stays under 5 % every single month from now through October 2026, the post-COVID streak will tie the Vietnam record. If it stays under 5 % for 11 more months and counting, it will become the new record. So 'historically strong' is accurate, but 'the strongest ever' is technically wrong. The same pattern shows up at a tighter 4 % threshold: the longest US history has ever spent below 4 % unemployment in one continuous stretch is 35 months (Korean War era, 1951-53), and the second-longest is 28 months (post-COVID, February 2022 through May 2024). So at both thresholds the post-COVID labor market is silver, not gold. I'm not claiming there's a deep policy lesson here. I'm claiming a specific factual correction with a date and a number you can verify yourself by downloading the same FRED file: the current sub-5 % streak is 54 months, ranks #2, and would need to make it through October 2026 to take #1.
Novelty
Sub-5 % unemployment streaks are widely discussed in business media, but the specific quantitative claim — that the post-COVID streak is exactly 54 months as of March 2026, that it ranks second behind the 64-month 1965-70 streak, and that the analogous sub-4 % streak ranks second behind the 35-month 1951-53 Korean War boom — is not stated as a single pinned table in any source I could find. Most public commentary either says 'the lowest unemployment in 50 years' (which compares LEVELS, not streak DURATION) or says 'one of the longest streaks ever' (which is accurate but vague).
How it upholds the rules
- 1. Not already discovered
- Web searches on 2026-04-13 for 'longest sub-5 unemployment streak US history', 'post-COVID labor market longest unemployment streak', and 'unemployment under 5 consecutive months record' returned BLS press releases, generic Wikipedia overviews, and economics-blog posts that quote partial figures but do not state the specific ranking with the 54-month / 64-month / 35-month / 28-month numbers.
- 2. Not computer science
- Economics / labor-market analysis. The object of study is the US monthly civilian unemployment rate time series; the program is a maximal-run scanner over a CSV.
- 3. Not speculative
- Every number is an exact maximal-run length on the pinned UNRATE CSV. No model, no extrapolation, no fit. The 'currently active' status of the post-COVID streak is conditional on the snapshot-month value (4.3 % in March 2026, well below the 5 % threshold).
Verification
(1) The FRED CSV is pinned by SHA-256 084d2f0cfe41575534b96bd51b88c6e93ddc8d5f0eca7e470ecfd19f5428c94f. (2) Scan logic is a single pass over the series with a counter; identical to a one-line groupby in pandas or shell awk. (3) Spot-check: the 1965-70 Vietnam-era expansion has the lowest sustained unemployment in the BLS reanalysis era and is consistent with published history. (4) The 1951-53 Korean War boom and the 1980-85 Volcker disinflation high-unemployment streak are both well-known historical periods cited in any monetary-economics textbook; the durations my scan returns are consistent with their canonical descriptions. (5) The post-COVID 54-month streak begins in September 2021 — the first month after the 4.8 % August-2021 reading dropped below 5 % — and includes the entire late-2021-through-March-2026 window without a single ≥ 5 % month; this is verifiable directly from the FRED CSV by anyone.
Sequences
64 (1965-03→1970-06), 54 (2021-09→2026-03 ongoing), 50 (1997-07→2001-08), 42 (2016-10→2020-03), 42 (1950-08→1954-01)
35 (1951-01→1953-11), 28 (2022-02→2024-05), 27 (1967-11→1970-01), 20 (1966-02→1967-09), 13 (2019-02→2020-02)
66 (1980-05→1985-10 Volcker), 59 (2008-12→2013-10 Great Recession), 29 (1974-12→1977-04 stagflation)
Next steps
- Compute the same streak ranking for U-6 (the broader underemployment measure) to see whether the second-place finding holds with a more inclusive labor metric.
- Repeat for prime-age (25-54) employment-population ratio, which many labor economists prefer over headline unemployment.
- Track the streak forward: every month FRED publishes a new UNRATE value, and the post-COVID streak ends or extends. Track until either October 2026 (record-tie) or the first ≥ 5 % print.
- Run the analogous symmetric analysis on the longest above-7 % runs and pin the modern post-COVID labor market in the full duration distribution.
Artifacts
- Streak analysis script: discovery/economics/unrate_streaks.py
- FRED UNRATE CSV (pinned): discovery/economics/unrate.csv