Reading a Central Bank Decision · Lesson 2 of 6
Reading a diff: what changed between two statements, and what did not
One concept: reading a statement-to-statement change
Why it matters
The newest FOMC brief on Kitalpha does something a statement alone cannot: beside each sentence it cites from the statement of 2026-09-16, it cites the matching sentence from the statement before, 10 such pairs in all. The previous statement described economic activity this way: "Economic activity is expanding at a solid pace despite elevated uncertainty that owes, in part, to the conflict in the Middle East.". The current one says: "Economic activity is expanding at a solid pace.". A reader sees two quotes side by side and asks what the change means. The lesson’s answer is that the pair is a record of wording, and reading it well means saying exactly what differs and then stopping there.
The concept
A statement-to-statement diff is the oldest tool in central-bank reading, and the simplest: put the new statement beside the old one and mark what changed. Committees write in a fixed structure and reuse language on purpose, so a changed phrase is a deliberate act and a kept phrase is one too. Kitalpha’s brief does the marking as citation. Each sentence it quotes from the current statement is an attributed span with its source and date; where the previous statement had a counterpart, that sentence is cited as well, from its own dated source, and the brief’s prose sets the two side by side.
Two disciplines make the reading honest. The first concerns changed wording. When an assessment moves from “remains elevated” to “has eased somewhat”, the pair records that the committee’s described view of inflation softened between the meetings. That is all it records. It is not a rate decision, which sits in its own sentence; it is not a statement that inflation reached a target; and it is not a forecast of the next meeting. A changed adjective is a change in what the committee chose to say. Turning it into a prediction adds something the pair does not contain.
The second discipline concerns unchanged wording, which is where readers most often over-read. A sentence repeated word for word records that the committee kept a form of words. It does not say that the economy stood still, because the data between two meetings never stand still; it does not say the committee had no view, because repeating an assessment is stating it again. Kept language is a choice, and the diff shows the choice, no more.
The brief adds one thing a hand-made diff lacks: every claim in its prose is checked against the spans it cites before publication, and the result is a groundedness figure on the brief page, the share of evaluated claims the checker found supported. A brief is published only above a stated threshold. That figure belongs to the brief’s own reliability, not to the committee, and it is why the pairs can be trusted to be quotes rather than paraphrases.
Not every brief has a diff. A policy statement has a predecessor whose wording it revises; a weekly data release, such as the petroleum report, states figures and has no statement language to pair. The table below sets three pairs from the newest brief side by side. Read each as two quotes, find what differs, and notice that nothing in the table says what the committee does next.
| Item | Detail |
|---|---|
| Activity — previous statement | Economic activity is expanding at a solid pace despite elevated uncertainty that owes, in part, to the conflict in the Middle East. |
| Activity — current statement | Economic activity is expanding at a solid pace. |
| Investment — previous statement | Productivity growth and capital investment are strong. |
| Investment — current statement | Productivity growth is strong, and capital investment is robust. |
| Decision — previous statement | The Committee decided to maintain the target range for the federal funds rate at 3-1/2 to 3-3/4 percent, in support of the Federal Reserve's dual mandate. |
| Decision — current statement | The Committee decided to raise the target range for the federal funds rate by 1/4 percentage point to 3-3/4 to 4 percent, in support of the Federal Reserve's dual mandate. |
Worked example
Take the newest FOMC brief. The steps below count what it cites: spans from the current statement, spans from the previous statement, the current spans left unpaired, the claims the checker evaluated, and the share found supported. Every figure comes from the brief’s own record.
Record: FOMC Raises Federal Funds Rate to 3-3/4 to 4 Percent in Unanimous 12-0 Vote · as of · Federal Open Market Committee statement
- Attributed spans the brief cites from the current statement 10
- Spans it cites from the previous statement, for comparison 10 Each previous-statement span is paired in the brief's prose with its current counterpart.
- Current spans left over once each previous span is paired 0
- Claims the brief's checker evaluated 20
- Share of evaluated claims the checker found supported by a cited span, in percent 100.0% A published brief must clear a threshold on this measure; the figure is the checker's, cited on the brief page.
What to read off the steps. The second line is the size of the diff: how many previous-statement sentences the brief holds against their current counterparts. The third is what the current statement carries that had no counterpart to pair, such as the vote or new guidance. The last line is the brief’s own quality figure, the checker’s share of supported claims, which is what licenses reading its pairs as verbatim. None of these counts says what changed in substance; that reading is done pair by pair, in the table above, by comparing two quotes.
Faded example
Now a brief of a different kind: the WPSR brief for the week ending 2026-09-11, on a weekly data release with no previous statement to pair. Its span count and claim count are given. Complete the last step: its groundedness as a percentage.
Second record: EIA Weekly Petroleum Status Report — Week Ending September 11, 2026 · as of
- Attributed spans the WPSR brief cites from its source report0
- Claims its checker evaluated7
- % Tolerance ±0.1 %
Reveal the answer and the explanation
100.0% — Multiply the brief's groundedness figure by 100. It is the same measure on a different instrument: the weekly petroleum report has no previous-statement pairing, since it is a data release rather than a policy statement, but every claim in its brief is still checked against a cited span before publication.
Stored on this device only; not graded.
Retrieval check
Mark your confidence before each answer. Every option carries an explanation; read the ones you rejected too.
-
1. A diff shows the phrase 'inflation remains elevated' in the previous statement and 'inflation has eased somewhat' in the current one. Factually, the pair records:
Choose your confidence first. -
2. The brief's pairs show a sentence repeated word for word from the previous statement. That repetition records:
Choose your confidence first. -
3. Using the newest FOMC brief, subtract the number of previous-statement spans it cites from the number of current-statement spans it cites. Enter a whole number.
Source record: FOMC Raises Federal Funds Rate to 3-3/4 to 4 Percent in Unanimous 12-0 Vote (as of 2026-09-16)
Tolerance ±0Choose your confidence first. -
4. Using the newest FOMC brief's verdict, multiply its groundedness figure by 100. Enter the share of evaluated claims found supported, in percent to one decimal.
Source record: FOMC Raises Federal Funds Rate to 3-3/4 to 4 Percent in Unanimous 12-0 Vote (as of 2026-09-16)
Tolerance ±0.1 %Choose your confidence first.