Hawkish and dovish language

Two things are published at once 

A policy meeting produces a number and a document, released together, and they do different work. The number sets the rate that applies from that day. The document describes how the committee is reading the economy and, in most cases, what would have to change for the rate to move again. By the time the number is published it is rarely news: policy rates move in small steps at scheduled meetings, committees signal their intentions before they act, and prices already carry an expectation of the outcome. The document is where the new information usually sits, because a committee's assessment can shift between meetings even when its rate does not.

This is why a decision and the movement around it can look unrelated. An unchanged rate accompanied by a materially altered assessment is a change in policy in every sense that reaches a price, while a widely anticipated cut accompanied by an unchanged assessment can pass with very little movement. The rate is the state of policy today. The language is the committee's account of where policy is heading, and most of what trades is priced against the second rather than the first.

What hawkish and dovish describe 

Hawkish and dovish are shorthand for the direction of a policy stance. A hawkish stance leans toward tighter policy, meaning higher rates or rates held high for longer, and the concern it expresses is usually inflation. A dovish stance leans toward looser policy, meaning lower rates or rates held low for longer, and the concern it expresses is usually growth, employment or stress in the financial system. Both are old market slang rather than official vocabulary. No central bank describes itself with either word: the label is applied by the people reading the document, not by the people who wrote it.

Two properties of the words are easy to miss and matter more than the definitions. The first is that they are comparative rather than absolute. A statement is hawkish relative to something, normally the previous statement or what participants had assumed the committee would say, so a committee that cuts its rate can still deliver a statement read as hawkish, if it cuts by less than had been assumed. The second is that neither word carries an evaluation. Hawkish is not good and dovish is not bad. They are directions on a single axis, and that axis is the expected path of the policy rate.

Why the wording can matter more than the rate 

The mechanism runs through expectations. A currency, a government bond and an index are priced against a whole expected sequence of future policy rates rather than against the rate that applies today, because the money committed through them is committed over months and years rather than overnight. That expected sequence is what practitioners mean by the path. A meeting can leave the current rate untouched and still move the path, and the path is what most of the pricing is attached to.

An expectation that is already held is already in the price. What moves a price is the difference between what was expected and what arrives, and that is as true of language as it is of numbers. A statement confirming what participants had already assumed adds nothing to the path and tends to move very little, however severe its tone reads to somebody encountering it for the first time.

Reading the two cases together shows why the announced rate is often the least informative part of a release. It is identical in both, while the expected rate a year out differs between them by half a percentage point, and every instrument priced off that expectation is repriced accordingly. It also explains why the size of a reaction is a poor guide to the severity of the language: the same words landing on a market that had already assumed them move very little, and landing on one that had assumed the opposite, move a great deal.

What a central bank publishes around a decision 

The language arrives across several documents rather than in one, on different schedules, and they are conventionally read as a set. Which documents exist differs by institution, and the differences are large enough that a reader moving from one central bank to another cannot assume the same furniture is there.

• The decision itself: the policy rate, together with any change to the other instruments a committee operates, such as the size of its holdings of government bonds.
• The statement, a short prepared text agreed by the committee and released at the moment of the decision. It is the most closely read document of the set, because it is written in the knowledge that it will be compared with the previous one.
• Economic projections, published by some institutions on a quarterly cycle: the committee's own forecasts for growth, inflation and, at some central banks, the policy rate itself.
• A press conference, usually shortly after the release, in which prepared remarks are followed by unscripted answers to questions.
• The vote, where it is published, including who dissented and in which direction they dissented.
• The minutes, called the accounts at some institutions, published weeks after the meeting. They record the discussion and the spread of views inside the committee in more detail than a statement can carry.
• Speeches, testimony and interviews given by individual committee members between meetings. These are not committee documents and carry no collective authority.

Forward guidance sits across several of these rather than in any one of them. It is the practice of describing the future conduct of policy explicitly, either tied to conditions or tied to a date, and its stated purpose is to make the path more predictable than a sequence of unexplained decisions would be. It also creates the sensitivity this lesson is about: once a committee has committed itself in language, an alteration to that language becomes an event in its own right, and the removal of a phrase can be read as a change of intent even when every other line is unchanged.

The set is thinner where policy is not set locally. A central bank operating a currency peg has tied its own rate to the rate of the currency it is pegged to, so its announcements tend to be short, procedural, and released shortly after the anchor central bank has moved. There is less interpretive language to read because there is less local discretion being exercised, and the document that carries the information is the one published by the anchor. The regional institutions and the mechanics of a peg are the subject of later lessons in this module.

How the language is compared 

The convention is literal. The new statement is placed against the previous one and the differences are read word by word, which is why terminals and news wires publish a marked up version within seconds of a release. Small alterations are conventionally treated as deliberate, on the reasoning that a text agreed line by line by a committee does not change by accident. Dropping a single adjective describing patience, changing a verb from will to may, moving the description of risks from balanced to tilted in one direction, changing many members to some members in the minutes: each is read as a marker of a shift in the balance of opinion.

The convention has limits its own practitioners state openly. A phrase can be removed because it had become factually inaccurate rather than because any intent changed, and language drafted in the knowledge that it will be compared this way is a negotiated object rather than a transparent one. Some desks now score tone with text models instead of by reading, counting hawkish and dovish terms against a fixed dictionary, and that approach is contested for a reason its users acknowledge: it is consistent and fast, and it has no way of recognising a sentence that means the opposite of the words it contains.

Why the same words are read differently 

Two participants can read one statement and disagree completely without either of them misreading it, because the reading is never of the document alone. It is of the document against a prior expectation, and expectations differ. A statement that is hawkish against the previous meeting's language can be dovish against what participants had come to assume in the weeks since, and prices respond to the second comparison rather than the first. Positioning compounds this. Where a large share of participants already hold the same view, an ambiguous statement can move a price sharply as those positions are adjusted, and an emphatic one can move it very little because there is little left to adjust.

The documents can also disagree with each other. A prepared statement is a committee text, and the press conference that follows it is one person answering questions that were not agreed in advance, so a market that has already read the statement one way can be turned by a single unscripted answer. The minutes, weeks later, can describe a discussion whose balance is different again, by which point the data the committee was arguing about has been superseded. Compressing a document covering inflation, employment, financial stability and the size of a balance sheet onto a single hawkish to dovish axis discards most of it, which is a limitation of the shorthand rather than a failure of the person using it.

One consequence is mechanical rather than interpretive. A scheduled release concentrates a great deal of repricing into a very short window, so quoted spreads widen around it and the price at which an instruction executes can sit some distance from the price on screen when it was sent. Volatility around scheduled events is treated in this curriculum as a risk topic rather than as a schedule of opportunity, and it has a lesson of its own later in this module.

Where practitioners disagree 

The first disagreement is about which document carries the information. One tradition treats the prepared statement as authoritative, on the reasoning that it is the only text the whole committee has agreed. Another treats the press conference as more informative, on the reasoning that an unscripted answer reveals what a drafted sentence was constructed to leave unsaid. Both describe the same event, and the argument does not resolve, because the two documents carry different kinds of information rather than different amounts of one kind.

The second concerns the minutes. Because they are published weeks after the meeting, one view holds that they are history by the time they arrive and that anything material in them has already been said in speeches since. Another holds that the distribution of views inside a committee is visible nowhere else, and that it is the closest available description of what would have to change for the next decision to differ. The third disagreement is about the shorthand itself, and is newer: whether one word for a whole assessment is worth keeping. Its defenders answer that a market needs a shared vocabulary that fits in a headline, and that the compression is what makes the vocabulary usable rather than what makes it wrong.