How to read the Fed’s dot plot and Summary of Economic Projections

Which meetings have projections

The FOMC meets eight times a year. On the Fed’s 2026 meeting calendar, four meetings carry an asterisk, which the calendar defines as a meeting associated with a Summary of Economic Projections: 17–18 March, 16–17 June, 15–16 September and 8–9 December. The January, April, July and 27–28 October meetings have no asterisk, so no new projections come with them. The October meeting produces a statement but no new dots; the next SEP is due with the 8–9 December decision.

The September 2026 projections were marked for release at 2:00 p.m. ET on 16 September, the same time as the statement. The September round also adds a year to the horizon: the June 2026 tables ran to 2028, while the September tables run to 2029.

What the dots are, and what they are not

Each participant, the Board governors and the Reserve Bank presidents, submits a projection for the economy under his or her own view of appropriate monetary policy. The Fed defines that as the policy path each participant judges most likely to meet his or her own reading of the maximum-employment and price-stability mandate. A dot in Figure 2 is one participant’s view of the appropriate midpoint of the federal funds target range at the end of a calendar year, or over the longer run, rounded to the nearest 1/8 percentage point.

Three things follow. The dots are individual views, not a committee forecast, and the SEP does not say which dot belongs to whom. They are not a commitment: they describe the policy each person thinks appropriate given his or her own economic forecast, and they change when the forecasts change. And the participant count is not the voting count. Eighteen participants submitted projections in September 2026, while the statement was approved by a 12–0 vote of the Committee’s voting members.

The Fed also prints its own reminder of how wide forecast misses have been. Table 2 of the September SEP gives average historical errors for short-term interest rates of ±0.5 percentage point for 2026, ±1.7 for 2027, ±2.3 for 2028 and ±2.8 for 2029, measured as the root mean squared error of fall projections for 2006 to 2025 made by various private and government forecasters.

Median, central tendency and range

Table 1 condenses every variable into three summaries, and the notes to the table define each one.

• Median: the middle projection when all of them are sorted. With an even number of projections it is the average of the two middle ones, so a median can fall between two dot levels.
• Central tendency: the spread after dropping the three highest and three lowest projections.
• Range: every projection, lowest to highest.

A worked example: counting the end-2026 dots

The accessible version of the September 2026 SEP prints Figure 2 as a table of how many participants placed a dot at each level, so the dots can be counted rather than read off a chart. For end-2026 there are 18 dots at three levels. The target range set on 16 September is 3.75–4.00%, so the current midpoint is 3.875% ((3.75 + 4.00) ÷ 2). Each 25 bp move shifts the midpoint by 0.25.

Sorted from lowest to highest, dots 1–2 sit at 3.875%, dots 3–14 at 4.125% and dots 15–18 at 4.375%. With 18 dots the median is the average of the 9th and 10th, both 4.125%, which Table 1 prints as 4.1. Two meetings remained in 2026 when the projections were made: 27–28 October and 8–9 December. A dot at 4.375% is two 25 bp steps above the current midpoint, so it implies a further 50 bp by year-end, at one meeting or across both.

The median path against the current range

Applying the same arithmetic to every horizon turns the median dots into a count of 25 bp steps from today’s 3.875% midpoint. The medians below are taken from the dot counts, since Table 1 rounds to one decimal place; for the longer run Table 1 prints 3.2 while the 9th and 10th dots both sit at 3.25%.

Read plainly: the median participant in September saw one more 25 bp increase by end-2026, no change during 2027, and a return to today’s midpoint by end-2028. The longer-run median of 3.25% is 62.5 bp below the current midpoint, which is two and a half steps, a reminder that a median is a statistic and not a path any one participant drew. The middle of the distribution also hides its width: for end-2027 the dots ran from 3.125% to 4.375%, a 125 bp spread.

What changed between June and September

Every SEP prints the previous round’s numbers on the line below, which makes the change easy to read. In June 2026 the range was 3.50–3.75% (FRED shows it from 11 December 2025 until 16 September 2026), a 3.625% midpoint. The June median for end-2026 was 3.8 in Table 1; from the dots it was 3.75%, the average of the 9th dot at 3.625% and the 10th at 3.875%. That median sat half a step above the June midpoint, an example of an even-count median landing between two levels. Eight of the 18 June participants had their end-2026 dot at the then-current midpoint.

From June to September the Table 1 medians for the funds rate rose from 3.8 to 4.1 for end-2026, from 3.6 to 4.1 for end-2027, from 3.4 to 3.9 for end-2028 and from 3.1 to 3.2 for the longer run. Over the same rounds the median 2026 PCE inflation projection went from 3.6 to 3.7 and core PCE from 3.3 to 3.4, while the median 2026 unemployment projection fell from 4.3 to 4.1. One September participant did not submit projections for 2028 and 2029, which is why those horizons have 17 dots.

How to check the numbers yourself

Open the FOMC meeting calendar on federalreserve.gov and look for the asterisks; each starred meeting links to its projection materials in PDF and an accessible HTML version. The HTML version prints Figure 2 as a count table, which is the easiest way to count dots exactly. Table 1 gives the medians, central tendencies and ranges, with the previous round on the line below each variable. The policy statement on the same page gives the decision and the vote. For the range in force on any day, FRED’s DFEDTARU and DFEDTARL series give the upper and lower limits.

Our central bank tracker at /central-banks follows the Fed’s range and the next decision dates, and /rates shows the Treasury curve that these expectations feed into. For the split of the 10-year yield into real yields and breakevens, see the separate explainer in our library.