A plain-English guide to the SEP and the dot plot.

The Federal Reserve's Summary of Economic Projections, or SEP, lands four times a year, alongside the March, June, September, and December FOMC statements. The SEP is the single document that maps individual Committee members' quantitative views of the future onto a public sheet of paper. The dot plot is the chart inside the SEP that plots each participant's projection of the federal funds rate at year-end for the next three years and in the longer run. It is the most over-read document the Committee publishes, and the most miscommunicated. This is the explainer on what it is, what it is not, and why today's OIS curve repricing makes the 16-17 June SEP the most-watched piece of paper of the year.

What's actually in the SEP

Four projections per participant, for four future endpoints each. Eighteen voting and non-voting Committee members file submissions; nineteen during transitions. The variables projected are:

• Real GDP growth. Year-end annualised, four values per participant: 2026, 2027, 2028, and "longer run".
• Unemployment rate. Fourth-quarter average, same four years.
• Headline PCE inflation. Fourth-quarter annualised year-on-year, same four years.
• Core PCE inflation. Same.
• Federal funds rate (the dot plot). Year-end level, same four years.

The published SEP reports the central tendency (the range excluding the three highest and three lowest projections) and the full range, along with the median and the prior projection for comparison. The submissions are anonymous in the published document, although a participant's individual views often leak through their public speeches between SEPs.

How the dot plot is read

Median, central tendency, and range

The three published summary measures answer different questions and they should not be conflated.

• Median. The dot that has at least half the dots at or above it. It is the headline number every market participant quotes. It is also the most easily moved by a single defection; a swing of two participants between adjacent 25-basis-point buckets can shift the median by a full quarter point.
• Central tendency. The range of the submissions excluding the three highest and three lowest. This is closer to a robust summary of the Committee's "middle" and less sensitive to outliers. A widening central tendency in successive SEPs is a real signal of disagreement.
• Range. The full minimum-to-maximum. Useful mostly for spotting the hawkish and dovish tails. The published range is asymmetric in interesting ways under most regimes; the dispersion of dovish dots tends to be wider than hawkish in soft-landing cycles, narrower in hard-landing ones.

What the dot plot is not

1. A commitment. The submissions are individual projections under each participant's modal view of the economy, not a vote on a particular policy path. The next meeting can produce a different decision from the median; it routinely does.
2. A consensus. The median is the middle of the distribution, not the centre of agreement. In a bimodal distribution (some participants thinking cuts, some thinking hikes, with a thin middle), the median can land at a level that almost no participant is actually projecting. The 2024 episode of this is the canonical case study.
3. The Committee's reaction function. The dots are conditional on each participant's modal economic forecast. If the realised data diverges from those forecasts, the path will move; the dots tell you nothing about how the Committee weights a 50 basis-point inflation surprise versus a 100k payroll surprise.
4. The Chair's view alone. The Chair gets one vote and one dot, like every other participant. The Chair's framing in the press conference reveals which way the Chair's dot is positioned, but it does not change the median.

The longer-run dot

The fourth column of the dot plot, labelled "longer run", is the Committee's estimate of the steady-state nominal federal funds rate consistent with maximum employment and 2 percent inflation. It is the published version of r-star, the neutral real rate, plus 2 percent. Through 2024-2025 the median longer-run dot drifted up from 2.50 percent to 3.00 percent, reflecting a Committee that increasingly believes the neutral rate is structurally higher than the pre-pandemic estimate. That drift is the most important durable signal in the SEP, and it gets very little press attention because it moves slowly.

For FX, the longer-run dot reads through to the implied long-end real yield differential against G10, which is the cleanest input to the dollar smile's right-tail rate-differential channel.

How the SEP connects to the OIS curve

The market-implied OIS path through year-end and the median dot for year-end almost always disagree, sometimes by 50 basis points or more. The persistent gap exists because the OIS path is risk-neutral (incorporates the term premium and the probability distribution of outcomes), while the dot is the participant's modal forecast. When the OIS gap to the median widens, the risk-neutral pricing is doing more work; when it narrows, the market is increasingly comfortable with the Committee's modal path.

The post-NFP move we cover in today's analysis has narrowed the OIS-to-median gap dramatically, with OIS now within 25 basis points of where we expect the median to land at the 16-17 June SEP. That is unusually close.

What to read at Warsh's first SEP

Three things worth reading carefully on June 17:

1. The 2026 median. Where the median lands relative to the current OIS pricing. A median above the OIS path implies a hawkish surprise; below, a dovish one.
2. The dispersion. Has the central tendency widened or narrowed since March? A widening signals the 8-4 April-meeting fracture is persisting; a narrowing would be a meaningful institutional repair.
3. The longer-run dot. Any drift higher reinforces the structurally-higher-neutral-rate narrative; a drift lower reverses it. This is the dot with the most durable FX implications.
4. Warsh's press conference framing. The published SEP is the medians; the Chair's framing is the Chair's interpretation. Reading the two together is the complete picture.
5. The trimmed-mean comment. If Warsh explicitly references the trimmed-mean PCE framing, the Committee's communication strategy on inflation is shifting. That is the single most actionable item for cross-asset positioning.

What the SEP does not tell you

1. The next meeting's decision. The SEP projects year-end levels, not meeting-by-meeting actions. Reading the SEP into the next meeting requires assuming a smooth path, which is a strong assumption.
2. The framework. The SEP does not publish participants' implicit policy rules. Two participants can have the same dot from different reaction functions, which means they will react differently to the next data print.
3. The forecast revisions. The SEP shows the current projection compared to the prior SEP. It does not show whether the revision is driven by data surprises or by a framework change. The minutes (released three weeks later) fill in this gap.

Related reading

• The companion analysis piece on today's repricing: The OIS curve repriced. From cuts to hike risk in five sessions.
• The OIS framework the SEP gets read against: A plain-English guide to OIS
• The dollar-smile channel through which longer-run dots transmit: A plain-English guide to the dollar smile
• End-of-week post-mortem on the data-not-dots thread: End of week: the thesis missed