When the Dot Plot Sits Above the Current Rate, Holding Steady Is Not Neutral

A central bank meeting that changes nothing is usually filed under "no news." That filing is wrong more often than not. The Federal Open Market Committee has now left the target range for the federal funds rate at 3-1/2 to 3-3/4 percent at consecutive meetings. The headline outcome was identical both times. The information content was not.

The reason has little to do with the rate number and everything to do with what surrounds it: a projection median sitting above the rate actually in force, a voting record that split three ways against a hold, and a communications calendar that meters when any of it can be discussed. Reading a decision means reading the distance between those three things.

Where Policy Actually Stands

Start with the settings, because everything downstream is measured against them. The Committee held the range at its June 16-17 meeting on a unanimous 12-0 vote, and held it again on 28-29 July on a 9-3 vote. The three July dissents did not favour a cut. They preferred to raise the range by a quarter point at that meeting.

The implementation note accompanying the July decision set the interest rate on reserve balances at 3.65 percent effective July 30, the primary credit rate at 3.75 percent, and overnight reverse repurchase operations at an offering rate of 3.50 percent with a per-counterparty limit of $160 billion per day. Standing overnight repurchase agreement operations were set at 3.75 percent. The effective federal funds rate has printed at 3.63 percent through mid-August, near the middle of the range rather than drifting toward either administered boundary. The directive also instructs the desk to roll over at auction all principal payments from Treasury holdings and reinvest agency principal into Treasury bills.

The projection sits above the setting

The June 2026 Summary of Economic Projections put the median federal funds rate at the end of 2026 at 3.8 percent, with 3.6 percent for 2027, 3.4 percent for 2028, and 3.1 percent as the longer-run median. The current target range is 3.50 to 3.75 percent. The year-end median therefore sits above the setting currently in force, not below it. Alongside it, the median 2026 projections were 3.6 percent for PCE inflation, 3.3 percent for core PCE inflation, 4.3 percent for the unemployment rate, and 2.2 percent for real GDP growth.

The July statement framed the inflation problem in supply terms, noting that inflation remains elevated relative to the 2 percent goal, "in part reflecting supply shocks that have driven price increases in certain sectors, including energy," and citing elevated uncertainty owing in part to conflict in the Middle East. On activity it described solid expansion, strong productivity growth and capital investment, job gains keeping pace with the workforce, and little change in unemployment.

The most recent Consumer Price Index release, published 12 August for July data, showed the all-items index up 0.1 percent for the month on a seasonally adjusted basis after falling 0.4 percent in June, and up 3.4 percent over twelve months. The index less food and energy rose 0.2 percent for the month and 2.5 percent over twelve months. Headline running above core is the fingerprint of an energy-led impulse rather than a broad demand-led one — the distinction a supply-shock framing turns on, and the distinction that splits a committee.

How the Price Gets Set Before the Vote

The claim that a decision is "already priced" is often stated as folklore. It is not. It is a specific instrument with specific settlement mechanics, and understanding those mechanics is what separates reading a probability from repeating one.

The instrument that carries the expectation

Thirty-day federal funds futures are quoted as 100 minus the implied average effective federal funds rate for the contract month. A contract priced at 97.4475 implies an average rate of 2.5525 percent for that month. Because the contract settles against the average of the daily effective rate across the calendar month, a month containing no scheduled meeting acts as an anchor: the implied average for that month can be treated as both the ending rate of the prior month and the starting rate of the next.

Months that do contain a meeting require unpacking. If N is the number of days before the meeting and M the number after, the implied starting rate is recovered from the observed average by removing the post-meeting portion — the published methodology expresses the starting rate as the month's average less the M/(M+N) weighted ending rate, divided by N/(M+N). The expected change is then the ending rate less the starting rate, divided into 25 basis point units. The integer part gives the number of quarter-point steps; the decimal remainder is read directly as the probability split between the two adjacent outcomes. An expected 2.9 steps is published as a 90 percent probability of the larger outcome and 10 percent of the smaller. Probabilities across successive meetings are chained by multiplication.

Two consequences follow. The published probability is an arithmetic decomposition of a single price, not a poll of forecasters. And because settlement is a monthly average, a meeting late in a month contributes only a few days of the post-decision rate to that contract, which pushes the informative signal into the following month.

The calendar that meters the information

Expectations do not form continuously. They form inside a schedule that the institution publishes in advance, and the schedule itself is a tradeable fact.

• Eight scheduled meetings a year. The 2026 dates are 27-28 January, 17-18 March, 28-29 April, 16-17 June, 28-29 July, 15-16 September, 27-28 October, and 8-9 December.
• Projections at four of the eight. March, June, September and December carry a Summary of Economic Projections. The other four do not, which mechanically reduces the amount of new information a non-projection meeting can deliver.
• A communications blackout. Under the Committee's published policy on external communications, the blackout "will begin at 12:00 a.m. Eastern Time the second Saturday before a meeting and end at 11:59 p.m. Eastern Time the day after a meeting." Reserve bank calendars list the 2026 windows, including 5-17 September, 17-29 October and 28 November to 10 December.
• Minutes three weeks later. Minutes of regularly scheduled meetings are released three weeks after the date of the policy decision, which makes the reasoning behind a decision public well after the price reaction to it.

The blackout is the underrated item on that list. It creates a two-week interval in which incoming data arrives but official interpretation of it does not, so releases landing inside the window price without institutional commentary.

What is left to surprise

If the level is priced, the residual information is dispersion. The count and direction of dissents shows how wide the internal distribution is: three votes for a quarter-point increase is a materially different signal from three votes for a cut, though both produce the same 9-3 headline. The spread of individual projections shows whether a median is a consensus or a midpoint between two camps. And revisions between meetings show which variable is doing the work.

Three Cycles That Show the Full Range

The relationship between announcement and price reaction is not fixed. It has taken at least three distinct forms in the modern record, each documented in the institution's own research.

February 1994 — the announcement was itself the news

Before 1994 the Committee did not announce policy changes immediately. The 4 February 1994 release broke that practice, stating that the action was announced immediately "so as to avoid any misunderstanding of the Committee's purposes, given the fact that this is the first firming of reserve market conditions by the Committee since early 1989." Federal Reserve research on that episode records futures priced at roughly 111 basis points of tightening over the coming year against roughly 300 basis points delivered. The ten-year Treasury yield rose 14 basis points on the announcement day and roughly 200 basis points over the following nine months. With no post-meeting statements, no timely minutes and no forward guidance, the surprise had nowhere to go but into price after the fact.

June 2004 — the increase that moved yields the wrong way

A decade later, with post-meeting statements, balance-of-risks language and explicit signalling in place, the same research records that the ten-year Treasury yield decreased 8 basis points on the day the Committee raised rates. Measures of interest rate uncertainty narrowed rather than widened. A rate increase produced a yield decline because the increase had already been absorbed and the accompanying language resolved a question the market cared about more than the step itself.

May to September 2013 — repricing with no policy action at all

The cleanest case is the one with no decision in it. Federal Reserve staff research measuring the episode from 2 May to 5 September 2013 records a 137 basis point rise in the ten-year Treasury yield. No purchase pace was reduced during that window. The actual change came on 18 December 2013, when the Committee announced that beginning in January it would add to agency mortgage-backed securities holdings at $35 billion per month rather than $40 billion, and to longer-term Treasury holdings at $40 billion per month rather than $45 billion — a $10 billion monthly reduction, arriving more than three months after the repricing had largely finished.

Three relationships: a decision that was the news, a decision already spent, and a repricing that finished before any decision existed. The current configuration matches none exactly, which is the honest starting point.

The Conditions That Invert This

The framing above has specific conditions attached, and several of them are live.

• Funding stress makes the corridor the story. The effective rate at 3.63 percent sits comfortably inside the range. If it drifted toward the upper boundary, where the standing repo facility and primary credit rate both sit at 3.75 percent, the operative question would shift to reserve scarcity — and the bill-purchase instruction would move prices with the target range unchanged.
• A projection is not a commitment. A 3.8 percent year-end median is the middle of a distribution of individual views submitted at one meeting. It carries no undertaking. Treating the gap between 3.8 percent and the current range as a scheduled action is a category error, and one that has been expensive in prior cycles.
• Dissents rotate. Voting composition changes on an annual cycle, so a 9-3 split under one composition does not translate mechanically into the same split under the next. Comparing dissent counts across year boundaries compares unlike things.
• Supply-driven inflation weakens the whole mechanism. If the impulse is genuinely energy-led — headline at 3.4 percent against core at 2.5 percent is consistent with that — policy expectations are downstream of a variable the Committee does not control. The marginal information then arrives from energy markets, not the meeting calendar, and futures-implied probabilities can be stale within hours of a supply headline.
• The futures decomposition has blind spots. The output assumes 25 basis point units and a distribution concentrated on two adjacent outcomes. An intermeeting action, a non-standard increment, or a change delivered through the balance sheet would not appear in it beforehand.
• Thin conditions exaggerate everything. A repricing that looks decisive against late-summer liquidity can look considerably smaller against September volumes.

What to Watch Next Week

• Minutes of the 28-29 July meeting. On the standard three-week schedule these fall in the week of 17 August. The item worth reading is not the decision but the characterisation of the argument for a quarter-point increase — how many participants found it persuasive without voting for it.
• The Jackson Hole symposium, 27-29 August 2026. The 2026 theme is "Financial Innovation: Implications for Payments and Policy." A payments theme is not a rate-path theme, which lowers the odds the event resolves the September question. Treat rate-path content as incidental.
• The data sequence into September. JOLTS for July on 1 September, the Employment Situation for August on 4 September, PPI for August on 10 September, and CPI for August on 11 September.
• The blackout overlay. The September window runs 5-17 September. Payrolls land just before it opens; PPI and CPI both land inside it. Two of the three most consequential releases will therefore be priced without any official interpretation available.
• September is a projection meeting. The 15-16 September meeting carries a Summary of Economic Projections. Whether the year-end median moves away from 3.8 percent, and in which direction, is a larger piece of information than the rate decision it accompanies.

Concrete Framework — What to Track, in Order

A repeatable routine for any scheduled policy decision, stated as steps rather than conclusions.

1. Record the priced path before reading commentary. Note the implied path from federal funds futures for the next two meetings and write the number down. Doing this first stops the commentary from becoming the memory of what was priced.
2. Locate the meeting in the calendar. Establish whether it carries a Summary of Economic Projections. Four of the eight 2026 meetings do — March, June, September and December. A non-projection meeting has structurally less to deliver, so a large move on such a day deserves more scrutiny, not less.
3. Mark the blackout window and map data against it. The window opens at 12:00 a.m. Eastern on the second Saturday before the meeting and closes at 11:59 p.m. Eastern the day after. List which releases fall inside. Those are the ones that will price without official interpretation.
4. Separate the three surprise channels. Attribute the move to one of: a level surprise against the priced path; a change in the internal distribution, meaning dissents or the dot spread; or a change in the stated diagnosis. Picking the wrong channel is how a one-day move becomes a cycle view.
5. Check the corridor, not just the range. Compare the effective rate against IORB and the ON RRP offering rate. Persistent drift toward either administered boundary is a liquidity signal that operates independently of the target range.
6. Read the minutes three weeks later against the price reaction. The gap between what they reveal and what the market did on the day is the cleanest available measure of how much was genuinely priced in advance.
7. Set a falsification condition in advance. Write the outcome that would show the reading was wrong — a median revision in the unexpected direction, a dissent flipping sides, headline and core converging. A reading with none is not a reading.

The durable point is narrower than the folklore suggests. Markets do not always price decisions in advance, and they do not always fail to. What decides which regime applies is how much the institution has already communicated, and how much of the outcome depends on variables outside its control. The configuration in place through mid-2026 — a projection median above the current setting, a split vote, and an inflation impulse the statement itself attributes partly to supply shocks — does not map cleanly onto 1994, 2004 or 2013. That is a reason to watch the dispersion rather than the headline, and a reason to hold the conclusion loosely.

This article is analysis of publicly available policy documents and economic data. It is not investment or financial advice.