Hawkish and dovish describe relative preferences or assessments in monetary policy. A hawkish message favours maintaining or increasing restraint; a dovish one favours easing it. Specify the comparison: the previous statement, another policymaker or expectations before the meeting. The labels do not supply a currency direction or an order to trade.

What do hawkish and dovish actually mean?

A hawkish assessment can reflect concern about persistent inflation and a preference for tighter policy. It need not mean a rate rise today: delaying anticipated cuts can also be a more hawkish message. Distinguish the policy change from its level. A cut can reduce restraint while leaving policy restrictive relative to economic conditions.

A dovish assessment can favour less restrictive policy in response to weaker activity or employment. It need not promise an immediate cut or disregard inflation. Policy affects credit costs and broader financial conditions, while its effects on employment and inflation are neither direct nor immediate. Read the bank’s mandate and the conditions attached to its outlook.

How do you read the tone in a bank's statement?

Read policy language in context. Persistent inflation or discussion of further tightening can indicate a more restrictive assessment, while concerns about growth can indicate pressure for easing. Data dependence, by itself, is neither hawkish nor dovish: future decisions can move in either direction as data change.

Compare changed phrases with the preceding statement and the accompanying projections, keeping conditional language intact. A press conference can add information, but no verified rule establishes that it always produces the largest move.

Same median, different dispersion

Five people in a fictional bank assess the appropriate rate at the end of the same future period. Two groups can have median 4% with differently distributed assessments.

Scroll sideways to see the full diagram

Projection median and rangeGroup A: 3%, 3%, 4%, 5%, 5%. Group B: 3.5%, 4%, 4%, 4%, 4.5%. Both medians are 4%, but ranges are 2 and 1 percentage points.MEDIAN DOES NOT SHOW RANGETwo fictional groups of 5 participantsA: 3, 3, 4, 5, 5 (%)Range A: 5 − 3 = 2 ppB: 3.5, 4, 4, 4, 4.5 (%)Range B: 4.5 − 3.5 = 1 pp33.544.55Both middle assessments = 4%One dot = one participant assessment

Assessments, groups and bank are fictional; all dots have the same instrument and horizon. Each participant supplies one equally weighted assessment. This is not a vote, bank promise or probability distribution. The documentation Fed: FOMC projection notes explains that a rate dot represents an individual assessment of appropriate policy over a specified horizon; a projection is not a certain future outcome.

Range is highest minus lowest assessment; it is not a probability interval for the future rate.

Record horizon and range alongside the median. Greater dispersion alone neither makes a whole group more hawkish nor determines a currency direction. For actual projections, also read the explanation of each dot.

Separate the decision, message and surprise

The decision records what changed now; the message describes the outlook and conditions for later actions. A surprise compares an announced value with an earlier expectation for the same instrument and date. Survey forecasts and measures derived from financial prices use different methods. Record their source before publication rather than substituting one after seeing the currency move.

A dovish hike combines a rise with a less restrictive message about the future; a hawkish cut combines a cut with a more restrictive assessment relative to the reference point. Neither specifies a currency outcome. Checking expectations lets you describe the information more precisely, but does not make the reaction predictable. Apply the same discipline to CPI inflation releases.

Three hypothetical decisions and their reference points

Use a fictional bank with one rate, initially 4.00% in each independent scenario. All six-month expectations refer to the same future date before and after publication; they are model assumptions, not actual market prices.

Hypothetical rates and expectation changes
ScenarioExpected / announced rate nowDecision surpriseSix-month expectation: before / after
Rise, softer future message4.25% / 4.25%0 bp4.75% / 4.25% (−50 bp)
Smaller cut than expected3.50% / 3.75%+25 bp3.25% / 3.75% (+50 bp)
No change, less expected easing4.00% / 4.00%0 bp3.50% / 3.75% (+25 bp)

The first row has a +25 bp actual change but no decision surprise; its assumed future-rate expectation falls 50 bp. The second has a −25 bp actual change, although a −50 bp cut was expected: the announced rate is 25 bp above that expectation. In the third, expectations change with no actual rate move. One basis point is 0.01 percentage points: 4.00% to 4.25% is +25 bp, whereas its relative percentage increase is 6.25%. Do not add a decision surprise to a future revision as a supposed currency return. See the impact of Fed decisions on forex and the ECB decision and its impact on the euro for the respective institutions.

Conditions also matter. Suppose a fictional rule cuts 4.00% to 3.75% only when a specified inflation measure is strictly below 3%. At 2.9% it cuts 25 bp; at 3.0% or 3.1% it does not. This threshold exercise is not a quotation, an actual bank rule or a forecast: real decisions use broader assessments. Read the action, condition, measure and horizon together.

What to do tomorrow

  1. Record expectations before the decision. Save the source, timestamp, instrument and horizon. A Fed target range and an ECB deposit rate are different measures; if using a range midpoint or upper bound, apply that choice consistently.
  2. Compare the bank's two most recent statements sentence by sentence. Open the bank's official site, find the statement from the last and the previous meeting, and highlight every sentence that changed. Those differences — words about inflation and further moves that were added or removed — show you which way the tone is shifting.
  3. Record phrases with their conditions. Note the action, inflation or employment assessment and future horizon. Data dependence allows different responses as information changes; it is not a fixed dovish code.
  4. Separate observation from interpretation. Record the statement, later conference and minutes as different information sets. Choose an observation window before comparing a pair and yields, note simultaneous news and the other currency’s outlook. A small journal is material to review, not evidence of a trading advantage.