US CPI Re-Acceleration Episodes Since 1948

Since 1948, US headline inflation has fallen below 3% year-over-year and then re-accelerated meaningfully four times. The 2024–2026 episode now in progress meets the same entry conditions as those four prior cases. Whether it follows the same trajectory remains an open empirical question.

This page documents each of the four prior re-acceleration episodes, aligned on the month its CPI year-over-year reading reached its trough, alongside the current episode whose trough was reached in April 2025 (2.33% year-over-year, the lowest reading since February 2021). The dataset, the methodology used to identify each episode, and the BLS series underlying every datapoint are downloadable below.

What the Data Shows

The chart at the top of this page is a heatmap. Each of the five rows represents one disinflation cycle. The horizontal axis is the number of months relative to that episode’s trough, from twelve months before to forty-eight months after. The color of each monthly cell encodes the year-over-year CPI level on a divergent scale: blue tones for below-target inflation (or outright deflation, as in 1949), near-white in the neighborhood of the 3% editorial threshold, and warm tones above. The bottom row, covering the current 2024–2026 episode, stops at the most recently published BLS observation, March 2026, eleven months after the April 2025 trough; cells beyond +11 months are shown as a pale-gray “no data yet” zone. The interactive version on this page allows toggling between three readings of the same dataset: the raw YoY CPI level, the gap to the 2% PCE Fed target, and the month-over-month pace of change. This connection is taken up in copper’s role in imported inflation. The mechanism is taken apart in the most common misreadings of inflation.

Three features of the historical record are visible on inspection. First, all four prior re-accelerations exceeded their preceding trough by at least two percentage points, with three of the four exceeding it by more than four points. Second, the time elapsed from trough to subsequent peak varies considerably across episodes: 20 months in the post-war 1948–51 case, 27 months in the 1970–74 case, 39 months in both the 1976–80 and 2000–05 cases. Third, the height reached at the subsequent peak varies by close to a factor of three, from 4.74% in 2005 to 14.59% in March 1980.

The current episode (bottom row) currently sits 0.96 percentage points above its trough at month +11. By comparison, at the same +11 month mark the prior episodes were positioned at: +4.55pp above trough in the 1948–51 case (July 1950), +2.79pp in the 1970–74 case (July 1973), +1.67pp in the 1976–80 case (November 1977), and +0.82pp in the 2000–05 case (May 2003). The current trajectory therefore sits between the 1970–74 and 2000–05 paths at this comparable horizon — slower than the post-war and first oil shock cases, faster than the second oil shock and commodities-boom cases. March 2026’s 0.86 percentage point monthly jump in the year-over-year rate is, however, larger than the equivalent single-month move at the +11 month mark in three of the four prior cases. Only the 1948–51 episode shows a steeper move at the same horizon (+1.73pp in July 1950), reflecting the mechanical rebound from the brief 1949 deflation.

The Four Prior Episodes, Plus the Current One

Sources: BLS Consumer Price Index for All Urban Consumers (CPIAUCSL series), monthly, seasonally adjusted, retrieved from FRED on May 10, 2026. Eco3min calculation for trough and peak identification. October 2025 missing in the source series due to the federal government shutdown of October–November 2025; the gap does not affect the trough date (April 2025) or the latest reading (March 2026).

What Happened Between Trough and Re-acceleration in Each Case

1948–51: Post-war price unwinding meets the Korean War

The post-war episode begins with the unwinding of wartime price controls and pent-up demand, which generated the 10.24% headline reading of January 1948. By August 1949, deflation had set in (−2.99% year-over-year), driven by the post-war contraction and the dollar revaluation effects of the 1949 sterling devaluation. The Korean War, which began in June 1950, triggered a procurement and commodity surge that pushed CPI back to 9.60% by April 1951. The proximate driver of the re-acceleration was a national-security shock that reactivated wartime-style commodity demand within an economy whose price-control machinery had been only partially dismantled.

1970–74: From the Phillips-curve high to the first oil shock

The first oil shock episode begins after the 1970 recession, during which CPI fell from a Phillips-curve-era peak of 6.42% (February 1970) to a Nixon-administration trough of 2.95% (August 1972), helped by the Phase I and Phase II wage-and-price controls of 1971–72. The October 1973 OPEC oil embargo, in combination with the post-Bretton-Woods dollar depreciation that had begun in August 1971, produced a commodity-driven inflation surge that lifted CPI to 12.20% by November 1974. The proximate trigger was an external supply shock striking an economy in which inflation expectations had not been re-anchored after the 1960s.

1976–80: From the post-1974 disinflation to the second oil shock

The second oil shock episode begins from a much higher trough than the others (5.04%, December 1976), reflecting the partial nature of the post-1974 disinflation. The empirical detail sits in our complete guide to inflation. The 1979 Iranian Revolution and the resulting second oil shock pushed CPI to its all-time post-1948 high of 14.59% in March 1980, which proved to be the immediate prelude to the Volcker disinflation that began in October 1979 and culminated in the 1981–82 recession. As in 1973, the proximate driver was an external supply shock; the larger amplitude reflected the absence of any consolidation of inflation expectations during the four-year window between the two shocks.

2000–05: From the dot-com disinflation to the China-driven commodities cycle

The commodities-boom episode begins after the 2001 recession, during which CPI fell from 3.76% (March 2000) to 1.07% (June 2002). The structural drivers of inflation regimes examines this with the series in hand. The subsequent re-acceleration to 4.74% by September 2005 was driven by the China-led commodities super-cycle, in which crude-oil prices rose from approximately $26 per barrel (West Texas Intermediate, 2002 annual average) to approximately $66 per barrel by September 2005 — the same month headline CPI peaked. The Federal Reserve’s policy stance, which kept the federal funds rate at 1.0% from June 2003 through mid-2004 before a measured-pace tightening cycle, has been examined in subsequent literature as a contributing factor to housing-market and broader-asset-price dynamics, but the direct mechanism for headline CPI in this period was the energy-price channel.

2024–26: From the post-pandemic disinflation to a partial energy-driven rebound

The current episode trough was reached in April 2025 at 2.33%, the lowest year-over-year headline reading in the United States since February 2021. The Federal Reserve held the federal funds rate at 3.50% to 3.75% through its March 17–18 and April 2026 meetings, after a measured easing cycle that ran through late 2024 and early 2025. The March 2026 reading at 3.29% — the highest year-over-year headline print since April 2024 (3.36%) — was driven primarily by an 18.9% year-over-year jump in retail gasoline prices, with monthly gasoline up 21.2% in March alone, the largest single-month increase since the BLS series began in 1967. The proximate cause was the late-February 2026 onset of the US–Iran conflict and the resulting disruption to Middle Eastern oil supply. A formal ceasefire was announced in April 2026, but the economic standoff has continued and a structural risk premium remains incorporated in the energy component of the March CPI print.

Three Observations Across the Five Cases

Three statistical features are common across the four prior episodes and are descriptively present in the current one. They do not constitute a forecast.

First, every prior re-acceleration was triggered by a specific external shock, not by a generalized warming of price pressures. The Korean War in 1950, the OPEC embargo in 1973, the Iranian Revolution in 1979, and the China commodities cycle starting in 2003 each provided a discrete identifiable mechanism by which a disinflation in progress was reversed. The 2024–2026 episode currently has a candidate analogue in the US–Iran conflict that erupted in late February 2026 and the gasoline pass-through visible in the March 2026 print. Whether that channel is sufficient to drive a multi-year re-acceleration analogous to the 1973 or 1979 cases is the empirical question this dataset frames but cannot answer.

Second, the speed of the re-acceleration after the trough was generally not the most informative variable; the direction was. Across the four prior episodes, the slope between trough month and trough+12 ranged widely, from +0.88 percentage points in June 2003 to +5.10 percentage points in August 1950. What was common was the absence of a sustained period below 3% after the trough was reached. The current episode has now spent eleven months above its April 2025 trough and has just printed its highest reading since the trough (3.29%). It has not yet returned to or below the trough level.

Third, the re-acceleration phase ended in three of the four cases with a recession that compressed inflation back down. The 1953–54 recession ended the Korean War price surge; the 1973–75 recession was concurrent with the first oil shock and ended its CPI trajectory; the 1980 and 1981–82 recessions ended the second oil shock; the 2007–2009 recession ended the 2000–05 episode (although the 2005 peak itself was followed by partial moderation before the GFC). Our macro-financial-regimes pillar maps out its implications. The relationship between re-acceleration and subsequent recession is not causal in the data — recessions occurred for many reasons — but it is a stylized fact of the historical record that re-accelerations have not, on the four prior observations, been resolved by a soft landing alone.

Counter-Arguments and Limitations

First, the structural conditions of 2026 differ from those of the 1970s. The Yale Budget Lab note What Are the Macroeconomic Implications of Recent Turmoil in Oil Markets? (March 2026) reports that the oil intensity of US GDP — the number of barrels needed to produce a dollar of inflation-adjusted output — has declined by more than 50% since 1973, with the post-2011 fracking expansion further offsetting the negative effects of oil shocks via positive effects on US producers. The Crestwood Advisors May 2026 Economic Update notes two further distinguishing structural conditions: long-term inflation expectations that remain anchored on most measures, and a productivity-supported earnings backdrop in the corporate sector. Each of these conditions weakens the mechanical chain through which a 1973-style or 1979-style trajectory would unfold. None of them, however, removes the chain entirely: the same Yale Budget Lab note finds that oil shocks still produce statistically detectable negative responses in real GDP and positive responses in core PCE prices, even in the post-2011 fracking era.

Second, four prior observations is a small sample for any quantitative inference. The dataset frames an empirical question about the current episode’s trajectory; it does not support probability statements. The choice of trough threshold (3.0%) is editorial: a 2.5% threshold would exclude the 1976–80 case (trough 5.04%); a 3.5% threshold would include the 1986–90 case but compress the historical signal. The five-row heatmap presented here is one of several legitimate ways to organize the post-war US inflation record. The downloadable dataset includes the two excluded cases (1986–90 and 2009–11) so readers can construct alternative cuts.

Third, the 2024–26 path could resolve as a one-off energy-driven spike rather than a re-acceleration. The March 2026 print is a single observation. The headline rate’s 0.86 percentage point single-month jump is unusual in size at this horizon, but its persistence depends on whether the gasoline pass-through and the Iran-conflict risk premium dissipate over the next two to four months or compound. The April 2026 BLS print, scheduled for release in mid-May, and the May 2026 print, scheduled for mid-June, will provide the first independent tests. Each print will be incorporated into the dataset on this page within 48 hours of publication.

Fourth, the choice of CPI-U over alternative inflation measures is consequential. The PCE deflator (the Fed’s preferred measure) has run approximately 0.3 to 0.5 percentage points below CPI-U over the post-1980 period. The same five-row heatmap exercise on PCE would shift each trough date by zero to three months and lower each trough level by approximately 0.3 percentage points; it would not change the count of episodes or the qualitative pattern. CPI-U is used here because it has the longer continuous monthly series, complete from 1947, and because its current reading is the one most-cited in market commentary on the 2024–26 episode.

Common Misinterpretations

Reading the chart as a forecast that the current episode will reach 14.59% (or any other prior peak). The chart documents what happened in four prior cases. It does not assign probabilities to outcomes for the current case. The four prior cases differ from each other by close to a factor of three in their post-trough peaks (4.74% to 14.59%) and by close to a factor of two in their durations (20 to 39 months), which is itself evidence that the historical record does not support tight inference about the current case.

Inferring from the four-case pattern that recession is necessary to end a re-acceleration. Three of the four prior episodes ended in or coincided with recessions, but the relationship is not deterministic. The 1948–51 episode ended with the 1953–54 recession after the post-Korean War demobilization; absent the recession, the trajectory cannot be extrapolated. The 2000–05 episode is the closest case to a trajectory that did not require a recession to compress the re-acceleration: CPI peaked at 4.74% in September 2005, was still at 4.18% in June 2006, then moderated sharply to 2.01% by September 2006 as the energy-price impulse reversed, before the financial-crisis dynamics took over in 2007–08. A modern soft-landing scenario for the 2024–26 episode is not foreclosed by the historical record; it is, however, less common than its alternative.

Treating the 1986–90 and 2009–11 episodes as full counter-examples. Both episodes meet the entry filters used in this analysis but are excluded from the chart for transparently documented editorial reasons (length and structural anomaly, respectively). Their inclusion would weaken neither of the page’s two specific descriptive claims: that the current episode meets the same entry conditions as the cases shown, and that no re-acceleration episode in the post-1948 record has resolved without a meaningful price overshoot relative to its trough.

Assuming the current trajectory will tightly track one of the four historical lines. The five rows share an x-axis (months since trough) and an identical color encoding for year-over-year CPI, but the underlying drivers of each episode differ enough that close path-tracking should not be expected. The exercise here is to position the current episode within the cluster of historical re-accelerations, not to extrapolate from a single most-similar prior case.

Methodology and Sources

Frequently Asked Questions

Why does this page focus on the four prior episodes rather than other historical inflation surges?

The four episodes are the cases since 1948 in which year-over-year CPI fell below 3% (a level commonly cited as proximate to the Federal Reserve’s 2% PCE objective in headline-CPI terms) and then re-accelerated meaningfully. Other historical inflation surges — for example the 1979–80 peak itself, or the 2021–22 pandemic re-opening — are documented in the dataset as the “peak before” reference points but do not constitute re-acceleration episodes in the sense used here. The framing is empirical, not theoretical: it asks how often the specific pattern of “fell below 3%, then re-accelerated above 3% within a few years” has occurred, and what the prior cases looked like.

Is the current episode confirmed as a re-acceleration?

It meets the entry conditions used in this analysis (trough below 3%, prior peak above 5.5%, and a subsequent reading above the trough by at least 0.5pp). It does not yet meet the full historical analogue threshold of +2.0pp above trough, which would require an additional rise of approximately one percentage point from the March 2026 reading. The April and May 2026 BLS prints, scheduled for release in mid-May and mid-June respectively, will be the first independent tests of whether the March 2026 jump persists or reverses. The dataset on this page will be updated within 48 hours of each new release.

What does “aligned on the trough” mean, and why is it the right alignment?

Each row’s horizontal axis is the number of months from that episode’s CPI year-over-year trough, with the trough at month zero. This alignment makes the slopes of the post-trough trajectories directly comparable across episodes that occurred decades apart and at different absolute levels of CPI. It is the alignment used in academic work on inflation cycles, where trough-aligned event studies allow the recovery dynamics of episodes with very different absolute levels to be visually compared. Calendar-year alignment would obscure the comparison; the same shape would appear at different horizontal positions on the chart, which would prevent the side-by-side reading the heatmap is designed for.

Could the episode-identification rule be tightened or loosened to produce a different count?

Yes. A trough threshold of 2.5% would exclude the 1976–80 case (trough 5.04%) and reduce the count to three. A threshold of 3.5% would include the 1986–90 case (trough 1.19%) and one or two additional smaller episodes, raising the count to five or six. The choice of 3.0% is editorial: it is round, broadly understood as “near the Fed’s headline-CPI implicit target,” and produces a count of cases (four prior plus one current) that is the maximum supporting a clean heatmap visualization without sacrificing distinctiveness. The downloadable dataset contains the data needed to apply alternative thresholds.

What is the role of monetary policy across these episodes?

The Federal Reserve’s policy stance differs across the four prior cases. In 1948–51 and 1970–74, the Federal Reserve was operating under different mandates and frameworks (the 1951 Treasury-Fed Accord post-dates the post-war episode). In 1976–80, the pre-Volcker FOMC was widely judged in subsequent literature to have under-tightened relative to the inflation backdrop; the 1979 transition to Volcker is itself part of the second-oil-shock episode’s trajectory. In 2000–05, the Greenspan FOMC held the federal funds rate at 1.0% from June 2003 through mid-2004 before a measured-pace tightening cycle that brought it to 5.25% by June 2006. As of the April 28–29, 2026 FOMC meeting, the Powell Fed has held the federal funds rate at 3.50%–3.75%, with the post-meeting statement removing the easing bias and citing the energy shock as a continued source of inflation risk. The historical record does not isolate a single policy response that has consistently followed each trough; the dataset on this page provides the inflation series only, not the contemporaneous policy series.

How will this page be updated?

The dataset will be refreshed within 48 hours of each BLS CPI release. The heatmap will be regenerated whenever the current row extends or whenever a methodological refinement is made (in which case the change will be flagged at the top of this page). The historical four-case rows are stable and will not be revised, except in the unlikely event of a BLS revision to data more than two years old.

Why is October 2025 missing from the BLS series?

The 43-day federal government shutdown of October–November 2025 prevented BLS from collecting and publishing the October 2025 CPI data on its normal schedule. BLS announced in December 2025 that the October 2025 print would not be reconstructed and that the year-over-year and month-over-month chains would resume with the November 2025 release. The gap is documented in the BLS technical notes for that period and does not affect the trough date (April 2025) or the latest reading (March 2026) used here.

Conclusion

The 2024–2026 US inflation episode currently in progress meets the same entry conditions as four prior cases between 1948 and 2024. Each of the prior four was triggered by an identifiable external shock — a war, two oil shocks, a commodity-driven super-cycle. Each ended, in the historical record as observed, with a meaningful price overshoot relative to its trough and, in three of four cases, a subsequent recession that compressed the rate back down. The amplitudes and durations differ across the prior cases by close to a factor of three, which is itself a reason to be cautious about tight inference.

What can be said with confidence on the current data is that the April 2025 trough has been definitively passed and that the March 2026 reading at 3.29% is the highest year-over-year print since April 2024. Whether the current episode resolves as a fifth historical re-acceleration in line with the prior cases, as a one-off energy-driven spike that compresses again over the next quarters, or as a fundamentally novel pattern not anticipated by the historical record, will depend on the April and May 2026 BLS prints and on the trajectory of oil prices through the rest of 2026. The dataset on this page exists to make those tests legible as they unfold.

The data and analysis presented on this page are provided for informational and educational purposes only. They do not constitute investment advice or a recommendation to take any specific action. Eco3min is registered as a non-prescriptive financial information publisher.