FOMC Statement March 2026: framing analysis of an institutional monetary-policy release

Context

The Federal Open Market Committee publishes a short statement at

the close of each scheduled meeting. The statement is the formal

public record of whatever the Committee decided to do with the

federal funds rate and is the most widely-read monetary-policy

release in the world. Its tone, vocabulary, and structure are

all deliberately uniform across decades; changes in wording are

parsed word-by-word by markets.

This worked example analyses the

March 18, 2026 statement,

roughly three hundred words. Published a day after this writeup

was drafted, so it is genuinely a fresh document. Chosen as a

counterpoint to

the Altman "Intelligence Age" essay:

same paragraph-scale length, very different framing posture.

Where Altman is promotional, future-oriented, and unsourced, the

FOMC statement is analytical, present-oriented, and institutional.

What Frame Check saw

The structural measurements, from the detectors in `framing.py`

and `claim_analysis.py` (deterministic, no LLM):

• Voice: analytical. Zero first-person-plural. Zero

second-person. Zero imperatives. Twenty sentences, all in the

institutional register ("the Committee decided to...," "the

Committee is attentive to..."). The reader is not addressed at

all; the text positions itself as a neutral record.

• Analytical coverage: 3 of 5 perspectives detected. Risks,

trends, and uncertainty register as present. Causes and

stakeholders register as absent. Density is high where it is

present: risks at 12.3 mentions per 1,000 words, trends at

12.3, uncertainty at 6.2. Compare the Altman essay: risks at

3.6, trends at 2.7, uncertainty at 0.

• **Temporal orientation: present 70 percent, future 20 percent,

past 10 percent.** The statement is grounded in what the

Committee is doing now and what it plans to monitor going

forward; past tense appears only in describing recent data.

• Sourcing: 0 percent by the structural detector. This is a

correct measurement but an incomplete reading. The Fed is

itself the authoritative source for its own policy values. The

epistemic detector is calibrated against documents that cite

external evidence; a primary-source release from the body that

sets the rates does not need to cite anything. Naming this is

part of the tool's construct-honesty commitment: the detector

reports a measurement, not a judgment about what it should

have measured.

• Claims: one specific numeric extracted by the claim-analysis

layer. Several explicit policy values appear in the text (target

range 3-1/2 to 3-3/4 percent; 2 percent longer-run inflation

objective; 1/4 percentage point dissent) but the extractor is

tuned for continuous values in prose and did not isolate all

of them as claims. This is a real limit worth naming; see

"What the method missed" below.

Frame detections

The frame-library matcher suggests one entry:

• FVS-009 Risk Frame (active).

Triggered by substantive risk density (12.3 mentions per 1,000

words) paired with uncertainty acknowledged. The library entry

distinguishes an active risk frame (what could go wrong, what

is vulnerable, what depends on holding assumptions) from a

nominal one (risk as a single sentence pivoted past). The

FOMC statement is the active kind: it names the risks

("somewhat elevated inflation," "implications of developments

in the Middle East," "the risks to both sides of its dual

mandate"), names the balance the Committee is trying to strike

("maximum employment and inflation at the rate of 2 percent"),

and closes by describing how the Committee will reassess

(labor-market conditions, inflation pressures, inflation

expectations, financial and international developments).

The contrast with the Altman worked example is the teaching

point of this pair. Altman's essay also registers "risks" as

covered by the same detector, and it triggers

FVS-002 Fluency Quality Illusion

instead of FVS-009. Same keyword category, different frame

assignment, different substance. The detector output by itself

does not tell a reader which is which; the detector plus the

library entry plus the reader's eye on the text does.

What the method missed

Honest naming of the specific limits Frame Check hit on this

document:

• Multiple policy values in prose, one extracted as a claim.

The claim extractor scans for structured numeric patterns in

sentences; it caught the "2 percent" longer-run inflation

objective but did not isolate the "3-1/2 to 3-3/4 percent"

target range or the "1/4 percentage point" dissent as

individual claims. All three are material; a more aggressive

claim-extraction pass would catch them. This is a known gap

that shows up especially in institutional writing where values

are embedded in formal constructions ("decided to maintain the

target range... at X to Y percent") that do not match the

patterns the extractor is tuned for.

• Source-network coverage for policy values. The Source

Network is designed for empirical measurements where the

provider is external authority (SEC EDGAR for company financials,

FRED for macro data, Wolfram Alpha for reference facts). A Fed

policy value IS the reference fact; the Fed's own data pages

would be the target. FRED does publish "Federal Funds Target

Range - Upper Limit" and "Federal Funds Target Range - Lower

Limit" as series; routing the extracted claim there would

return "verified" trivially because the Fed sets what it

reports. That verification is not false, but it is not

informative in the way a Source Network verdict usually is.

The honest naming: verification is most useful against

independent primary sources, which monetary policy values

mostly are not.

• Stakeholder analysis. The detector reports stakeholders

as absent from the statement. The statement talks about "the

economy," "the Committee," and "the dual mandate" but does

not name workers, borrowers, savers, holders of different

assets, or any specific group that the rate decision affects

differently. The absence is real and worth naming. It is also

characteristic of the institutional register: the FOMC

statement is written to be read as a single voice speaking

to "the markets" as an abstraction, not a policy text that

engages with distributional consequences. Reading the

statement against that gap is work the tool will not do for

the reader.

Why this example is worth publishing

Because the structural detector fires "risks covered" on both

this document and the Altman essay, but the two documents are

doing radically different things with that coverage. The Altman

essay uses the risk keyword once, then pivots. The FOMC

statement organises itself around risks. Same flag, different

frame assignment (FVS-002 vs FVS-009), different reader takeaway.

This pair, read together, is the single sharpest case for

construct honesty that Frame Check can make: a structural

detector that returns the same category label does not mean two

documents are doing the same thing. The tool surfaces surface

patterns; the library entries name which shape of the pattern

was detected; the reader closes the loop. That is the whole

method. A worked-example archive that holds only documents

Frame Check treats favourably is a selection-biased archive;

this entry exists partly to anchor the archive against that

bias.

Citation

Lucic, L. (2026). *FOMC Statement March 2026: framing analysis

of an institutional monetary-policy release*. Frame Check

Worked Examples.

frame.clarethium.com/corpus/worked-examples/fomc-statement-march-2026/

Licensed CC-BY-4.0. The FOMC statement is a U.S. federal

government work in the public domain; this analysis does not

alter that status.