Seventy Two Years Of Tightening, One Table

Every tightening campaign the Fed has run since 1954, dated from the policy record rather than the rate chart, and what followed each one.

One more thing about this campaign before the history. The Fed cut three times in the second half of 2025, which brought the effective funds rate down to about 3.63 percent by August 2026, and then it turned around. The last two tightening campaigns both began from the floor, at 0.08 percent and 0.12 percent. This one began from a rate the Fed had just spent a year lowering. It is not normalizing away from emergency settings. It is adding back to a stance it had only recently finished loosening.

The hard part is counting them

I expected this to be a lookup. It is not, and the reason changes every number below, so it goes first.

The Fed has not always tightened with the same tool. For the first forty years of this data there was no public target for the federal funds rate at all. The Fed moved the discount rate, and the market worked the rest out from what the funds rate actually did. The FOMC started setting an internal funds rate target in late 1982 and did not publish it. February 4, 1994 is the first time it announced a change in its stance on the day it made the decision, and the statement did not carry a numerical funds rate figure until July 1995.

My first attempt tried to dodge all of that by ignoring policy and looking only at the rate. Find the local troughs and peaks, call every big rise a cycle, let the data speak. It does not work, and finding out why was the most useful hour I spent on this.

The method needs a threshold, some amount the rate has to turn by before you call it a turn. Every threshold produces a different history. At 100 basis points the stretch from 1971 to 1982 breaks into seven separate cycles, which is not how anyone who lived through it would describe it. At 150, the 1994 campaign and the 1999 campaign merge into a single run of ninety months. At 200, the entire 1960s becomes one ninety seven month cycle. At 250, the 2015 campaign disappears from the record altogether.

There is no setting that reproduces the cycles the Fed itself would recognize, and that is the answer rather than a failure to tune it properly. A tightening cycle is a sequence of decisions. The rate is what happened as a result. You cannot recover the decisions from the consequence.

So the dates in the table are policy dates, and the rule for grouping them is this. A campaign runs from the first increase in the instrument to the last increase before the campaign breaks. It breaks on either of two tests: the instrument gives back more than forty percent of its rise, or it never beats its own high again within twelve months of the last increase.

Both tests are needed and each catches something the other misses. Depth alone merges the 1963 and 1967 campaigns, because the April 1967 cut gives back exactly a third of the rise and sits on top of any threshold near there. Time alone merges the 1987 campaign into the 1988 one, because the target did climb back above its 1987 high inside twelve months, after the October crash had already reversed more than half of it. Run both and nothing in the table rests on a point or two either way. No campaign that survives contains a cut deeper than eighteen percent of its rise, and the single campaign broken on depth had given back forty eight.

A pause does not end a campaign. The Fed held the discount rate still for eight months in the middle of the 1977 campaign and cut it a quarter point in the middle of the 1967 one, and went on to a higher peak both times. Nothing smaller than fifty basis points is called a campaign at all, which is what disposes of the lone quarter point increase of March 1997.

Three things in the table I want on the record rather than buried. Three campaigns contain a cut in the middle: 1967, 1983 and 1987. The two rows from the unpublished era rest on a reconstruction, and reconstructions disagree. The Board's own released records date the last increase of the 1988 campaign to May 4, 1989, while Thornton argues for May 17 and says so explicitly; the level is identical either way, so I have used the Board's date. And the modern dates are decision dates. The Fed's own open market operations table lists effective dates, which from 2008 onward fall one day later.

Every campaign since 1954

From and To are the monthly average effective federal funds rate in the month before the first increase and the month after the last one, because an increase on the sixteenth only moves part of its own month. Tool / rate gives the same campaign measured two ways, in basis points: how far the Fed moved its own instrument, then how far the market rate actually travelled. Wait is the months between the previous campaign's last increase and this one's first. What followed is the National Bureau of Economic Research business cycle peak, which is the last month of the expansion; NBER dates the contraction from the month after.

What the record says

Ten of the fifteen completed campaigns were followed by a recession. Five were not: 1963, 1983, 1987, 1994 and 2022. That is a worse record than the Fed would like and a better one than the permanent bears will tell you.

The 2015 campaign is the one I would argue about with you. It ended in December 2018 and the economy peaked in February 2020, which is a fourteen month gap and well inside the normal range. But that recession was a pandemic and I am not prepared to hang it on the Fed. Count it the other way if you prefer. It moves the record to nine and six and changes the shape of nothing.

The 1994 campaign is the one everyone cites as the soft landing and it earns it. Greenspan took the effective rate from 3.05 percent to 5.98 percent in fourteen months, roughly doubled it, and the expansion ran another six years.

Here is the part I did not expect. In three of the fifteen, the Fed's last increase came at or after the expansion had already peaked. In the campaign that ended in April 1974, the last increase came five months after the peak. In the one that ended in February 1980, one month after. And the last increase of the 1955 campaign landed on August 23, 1957, inside the peak month itself.

That is not incompetence, it is the job. The data that tells you a recession has started arrives long after it started. The NBER has announced business cycle peaks anywhere from four months afterward, which is the February 2020 peak called on June 8, 2020 and the fastest on record, to twelve months, which is the December 2007 peak called on December 1, 2008. Before 1978 there was no dating committee at all, so the November 1973 peak was never called in real time by anybody. It was dated years later, by which point the Fed had raised once more and then spent the rest of the decade regretting it.

Across all ten, counting those three as the negative numbers they are, the median wait from the last increase to the peak was seven and a half months. So the honest version of the question is not whether rates are too high today. It is whether anyone could tell for about a year.

The lever and the rate

The two change columns are there because they do not agree, and the way they stop disagreeing is the most interesting thing in the table.

In the discount rate era the funds rate routinely went much further than the tool that was supposed to be driving it. Across those seven campaigns it overshot the discount rate move by an average of 217 basis points. In the campaign that ran from September 1980 to May 1981, the Board raised the discount rate by 400 basis points and the funds rate rose 949. Once the FOMC was targeting the funds rate directly but not publishing it, the average gap fell to 54 basis points. In the five completed campaigns since 1994, every single one landed within seven basis points of its own announced target change, and the average miss was three.

Which means the sentence "the Fed raised rates by two and a half points" only became literally true somewhere in the 1990s. For the first half of this table it is shorthand for something looser: the Fed leaned, and the market moved further than the lean.

The 1987 campaign is the exception that runs the other way, and it is why that row looks so small. The target rose 144 basis points while the effective rate rose 38, because the funds rate was already trading well above target when the campaign started. It is also the clearest case in the table of a campaign stopped by a market event rather than by the economy or by the Fed deciding it was finished.

The stretch that keeps getting brought up

Look at the campaigns that ended in April 1974, February 1980 and May 1981 together. Three tightening campaigns inside a decade, each starting from a higher floor than the last, the third of them taking the effective rate to 19.10 percent.

The Wait column is where that story actually sits. The median gap between one campaign's last increase and the next one's first is thirty nine months. After the last increase of February 1980 the Fed was raising again seven months later. After the 1987 campaign it was back inside six. Those are the two fastest restarts in seventy years, and both followed a campaign that had been treated as finished.

The Dallas Fed published a piece on this by Lutz Kilian on February 17, 2026, and its closing line is blunt enough that I would rather quote it than summarize it:

What we can learn from the 1970s is that a well-intentioned policy of stimulating the economy by lowering interest rates has the potential of inadvertently reigniting inflation.

Consumer price inflation ran under 2 percent through the early 1960s. It reached 6.2 percent in January 1970, 12.3 percent in December 1974, and peaked at 14.8 percent in March 1980. It did not get there in one move. It got there in steps, and each step started from a floor higher than the last one, because every easing let inflation settle at a new level before the next campaign began.

Arthur Burns, who chaired the Fed from February 1970 to January 1978, did not think he was causing it. He attributed inflation to monopoly power, to external shocks, to a lack of fiscal discipline. Kilian makes the opposite case and makes it well. His argument is that the 1973 oil price surge was a symptom rather than a cause, that inflation was already climbing long before it, and that both the inflation and the oil price had the same monetary root.

I am not saying 2026 is 1972. The levels are nowhere close and the Fed of 1972 was operating without a published target, without a dating committee, and without most of the data it now takes for granted. What I am saying is that the shape of the mistake is specific and documented, and it is not the mistake of failing to raise rates. It is the mistake of stopping early, letting inflation settle somewhere above where it started, and having to begin again from there.

Where this one sits

On timing, it is unremarkable. Thirty eight months passed between the last increase of the 2022 campaign and the first of this one, against a median of thirty nine. The Fed waited about as long as it usually does.

The Fed's own projections say it is not finished. Sixteen of the eighteen participants who submitted forecasts in September expected at least one further increase before the end of 2026. The same projections put core PCE inflation at 3.4 percent for 2026 and headline at 3.7 percent, and do not show inflation back at 2 percent until 2029, which is the last year of the forecast horizon.

Read that again. The central bank is saying in its own published forecast that it will miss its target every year it has bothered to forecast. And the stretch is longer than it looks from there, because headline PCE inflation has not printed below 2 percent since early 2021. If 2029 turns out to be right, that is eight or nine years above target rather than three.

For scale, the closest historical comparison is looser than it sounds, because the 2 percent target itself only dates from January 2012 and the Fed steers by PCE while the long historical series is CPI. But the run is worth knowing anyway. Consumer price inflation last printed below 2 percent in January 1966. It did not print below 2 percent again until March 1986.

The record is the record

Fifteen completed campaigns in seventy two years. Ten recessions, five clean exits, a median campaign of 293 basis points over eighteen months, and a median wait of seven and a half months after the last increase before the expansion turns over, in the cases where it turns over at all.

There is no honest way to say which column this one ends up in, and anybody telling you otherwise is guessing with more confidence than the record supports. What the table does tell you is the shape of the distribution, and that the Fed has been in this exact position before, including the specific version where it cut first and had to turn around.

Sources. Effective federal funds rate, monthly, Federal Reserve Bank of St. Louis, series FEDFUNDS, from July 1954. Discount rate changes, Board of Governors historical dates and rates, series DISCOUNT. Intended federal funds rate target 1982 to 1993, Daniel L. Thornton, A New Federal Funds Rate Target Series: September 27, 1982 to December 31, 1993, Federal Reserve Bank of St. Louis Working Paper 2005-032, May 2005. Announced target changes from 1994, Federal Open Market Committee statements and the Board's open market operations tables. Business cycle peaks and announcement dates, National Bureau of Economic Research. September 2026 decision, vote and Summary of Economic Projections, Federal Open Market Committee, September 16, 2026. Employment figures, Bureau of Labor Statistics, Employment Situation, released October 2, 2026. Consumer price inflation, Bureau of Labor Statistics. 1970s analysis, Lutz Kilian, Lessons from the destabilization of inflation in the 1970s, Dallas Fed Economics, February 17, 2026. Chair tenures, Board of Governors.