A central bank forecast is a statement of what its policymakers expect at the moment they publish it. The Federal Open Market Committee sets out its members' expectations for growth, inflation, unemployment and the federal funds rate four times a year in a document called the Summary of Economic Projections (SEP), and the middle value across all the individual projections, known as the median projection, is the figure most often quoted as "the Fed's forecast".

Because a fresh SEP appears each quarter, the same target year picks up a new median projection at every publication date, and that median can move a considerable distance between the first projection made for a year and the last one made before the year closes. Setting that sequence of projections beside the economic data that eventually arrived is one way to weigh how much a single projection, taken on its own, was worth believing at the time it was made.

What a shifting projection tells you, and what it does not

A projection that moves between one SEP and the next is not automatically a sign that policymakers got something wrong. It can just as easily mean the data itself moved: inflation or employment figures that arrived partway through the year gave the Committee new information, and the median projection was revised to reflect it. Looking at the whole run of projections for a year, rather than any single one in isolation, shows whether the forecast held steady, drifted early, or shifted sharply close to the year's end.

None of this turns a projection into a guarantee. A median projection is a snapshot of what the Committee expected on the date it was published, built from the figures available on that date, and those underlying figures can themselves be revised later, which is worth keeping in mind when reading any single point on the record.

A projection made in one quarter is not the last word on it. The Federal Open Market Committee publishes a fresh Summary of Economic Projections (SEP) four times a year, and the median forecast for any given target year, say 2026, appears again in each one, revised in light of whatever has happened since the last publication.

How a single year is tracked

The scorecard takes one target year and follows its median projection across every SEP where that year appears, from the earliest publication that included it through to the last one before the year arrived. Each point is dated to the meeting that produced it, because a projection made eight quarters out and one made two quarters out are not answering the same question even though they carry the same label.

What it is set against

Once the target year has passed, the actual outcome, drawn from the same published data series the FOMC itself uses, either GDP growth, unemployment, or inflation depending on which projection is being checked, is plotted alongside the run of forecasts. That lets you see not just whether the median forecast was right in the end, but at what point in its life it started moving toward the right answer, or away from it.

The tool draws on the Federal Reserve's own projection tables, retrieved through the FRED API. This product uses the FRED® API but is not endorsed or certified by the Federal Reserve Bank of St. Louis.

A projection for a given year rarely arrives fully formed. The Federal Open Market Committee publishes an updated median four times a year in its Summary of Economic Projections (SEP), the table showing where each participant expects growth, unemployment and inflation to land, and the median for a distant year often sits still for a few publications before it starts to move.

Early movement versus late movement

Sometimes the median for a target year shifts well before the year itself begins, as incoming data changes the Committee's read on the economy. Other times it barely moves until the year is close at hand, then adjusts sharply as the picture sharpens. Setting each SEP publication date against the eventual outcome shows which pattern held for a given year.

The final projection against the outcome

The last SEP published before a target year closes is the Committee's most informed guess for that year, made with more of the year's data already in hand than any earlier edition had. Comparing that final median to the published outcome shows how far even the most current forecast landed from what actually happened, which is a different question from how far the earliest forecast missed.

None of this tells a reader what the Committee will project next, and a projection is not a commitment. It shows how one institution's own stated expectations moved against its own later data, using figures the Federal Reserve itself published.

Four times a year, each member of the Federal Open Market Committee writes down a number for where they expect growth, unemployment, inflation and the policy rate to land over the next few years. The median of those individual numbers is what gets published as the table you see on this page, in a document called the Summary of Economic Projections (SEP). It is released by the Federal Reserve Board alongside the post-meeting materials for the relevant FOMC meeting.

Finding the tables yourself

The SEP sits on the Federal Reserve Board's own site, in the section covering monetary policy releases, filed under the date of the meeting it accompanies. Each edition carries its own release date printed on the document, which is the date to cite if you are quoting a figure from it. Reading the table alongside the FOMC statement from the same day shows how the Committee's stated reasoning and its members' numbers line up, or where they do not.

What to check when a new one is published

• Whether the median for the year you are tracking moved from the previous edition, and by how much.
• Whether the central tendency (the narrower band excluding the highest and lowest projections) widened or narrowed, which says something about how much members agree.
• Whether the individual projections, shown as a chart in the SEP itself, cluster tightly around the median or spread out.
• Whether the actual outcome for an earlier year, once official data is revised, still matches what was recorded against it. Statistical agencies revise figures after first release, so a comparison drawn today can shift later even though the projection itself does not change.

The mechanics of how an FOMC meeting produces both the statement and the projections are set out in What Happens on an FOMC Decision Day, and the question of who is actually in the room writing these numbers down is answered in The Federal Reserve: Who Decides What.