Hawkish vs Dovish Monetary Policy: Reading a Stance

A hawkish stance leans against inflation and accepts weaker activity as the price. A dovish stance leans the other way, toward activity and employment, and takes on more inflation risk as the cost. The two differ in which risk the authority is currently more willing to run, not in the level of the rate. One rate can therefore be either stance, depending on the conditions it is held against.

Underneath that answer sits one idea, and the whole of what follows is built on it. A stance is not a number. A stance is a position taken relative to conditions, and a position needs two things to exist: something set, and something it is set against. The rate is only the first of those. Prices running fast or slow, activity above or below what an economy can sustain, jobs plentiful or scarce: those are the second, and they move on their own schedule without asking permission from anybody. Change either half and the position changes. Two economies can therefore sit on the identical rate and lean in opposite directions, and an authority can move its rate and end up leaning harder in the direction it just moved away from.

What is a hawkish stance, on its own?

A hawkish stance is a leaning, and the direction of the lean is against inflation. The fastest way to lose the idea is to define each stance as the absence of the other, so everything else can be set aside for a moment, including the word dovish. A hawkish stance treats the price reading as the first call on the lever, and treats weaker activity as an acceptable price for getting that reading down. That is a choice about which of two risks to run, made in conditions where running neither is not on offer.

Think about a household that has watched its monthly grocery bill climb for a year and finally decides to do something about it. The household cancels the second scooter, moves the wedding to a smaller hall and stops eating out. Nobody in that household thinks the cuts are pleasant. The cuts are made because the alternative looked worse, and everyone involved understands that the year ahead will be duller than the year behind. A hawkish stance carries exactly that flavour: the discomfort is not a side effect that somebody failed to notice, it is the recognised cost of the thing being attempted.

A hawkish leaning gives something up, and the cost is worth naming precisely rather than gesturing at. Borrowing gets dearer while it lasts, so a business deferring a new line of machinery keeps deferring it. A business that cannot fund the line does not staff it, so hiring slows. Activity comes in below what the economy could otherwise have produced. None of that is hidden or unintended. The lean is being taken in the knowledge that it costs something, and a reader who understands a hawkish stance as simply high rates has taken the number and dropped the reason.

What does a hawkish leaning look like in a decision and in the wording?

Two things carry a hawkish stance out into the world, and they are read very differently. The first is the decision itself. A decision is a number, and a number is not open to argument. The second is the wording published around the decision. Wording is prose, and prose is open to argument all day. A hawkish decision looks like a rise of twenty five or fifty basis pointsA hundredth of a percentage point, which is the unit rate moves get counted in. A quarter point move is twenty five of them, and a hundred of them make a full point.. A hold can carry the same leaning: the rate stays exactly where it was, and the price reading falls underneath it. The hold is the case worth sitting with. Nothing moved, and yet the leaning got firmer.

A hold is never neutral. While the authority sits on its hands, the conditions the rate is being held against keep moving. Standing still on a moving walkway is not standing still relative to the building. If prices are decelerating and the rate does not follow them down, the distance between the two is widening, and widening that distance is exactly what a rise would have done. The widening distance is the single most useful thing to carry out of the decision half of a stance reading, and it is invisible to anybody who only records whether the rate changed.

The wording half works differently. The wording has no arithmetic in it at all. Words such as watchful, vigilant and prepared to act further do not commit anybody to anything, and they are not measurements, but they are the only forward-looking material a reader gets between one decision and the next. Where the price reading is being described as stubborn rather than easing, and where the risks are described as tilted upward rather than balanced, a reader is being shown a leaning that has not yet turned into a decision. No reader can convert that into a number, and pretending otherwise is where most stance-reading errors begin.

What is a dovish stance, on its own?

A dovish stance is also a leaning, and the direction of this one is toward supporting activity. A dovish stance treats weak activity and thin employment as the thing most worth acting on, and accepts more inflation risk as the price of acting on it. That sentence has the same shape as the hawkish one, with the two risks swapped, because the two stances are not opposites in the way that hot and cold are opposites. The two stances are two answers to one question. Given that both risks are live and the lever can only lean one way at a time, which risk is the authority more willing to run this year?

Here is the household version. A small tailoring shop has had a thin six months. Orders are down, one of the two machines sits idle, and the owner is deciding whether to let the helper go. Instead the owner borrows a little, keeps the helper on the payroll and keeps the shutter open through the quiet stretch, knowing that money will be tighter later and that the borrowing has to be repaid out of a business that is not currently thriving. The owner has made a dovish choice in miniature: protect the activity now, accept the bill that arrives afterward, and be honest that a bill is arriving.

A dovish leaning gives something up too, and the cost follows from the same shape. Cheaper borrowing supports spending, and spending pressing against an economy that cannot immediately produce more is precisely the condition in which prices rise. So a dovish stance carries a real risk of a worse price reading later, and the people who bear that risk first are the ones with fixed incomes and savings rather than borrowings. None of that is a criticism of the stance. The risk is part of the stance itself, and an account that described a dovish leaning as simply the friendlier one would hide half of it.

Now that both are built, where do the two actually part?

Most of the work is already done, so with both stances standing on their own the contrast becomes short. Set them against four criteria and the whole difference reduces to one thing repeated four times. The only real difference between a hawkish stance and a dovish stance is which of two live risks the authority is currently more willing to run, and every other difference on the axis is a consequence of that one. Which risk is being run decides what is being protected. The protected side shapes what the wording sounds like. And what the wording sounds like names who gets helped first in the short run.

The last row deserves care, and it is the one that gets moralised. A hawkish leaning helps a saver first: the money in a deposit is worth more when the price reading falls, and the return on it is firmer while rates are held up. A dovish leaning helps a borrower first: repayment costs less, and the deferred machinery decision gets taken. Neither of those is a favour granted to a deserving group. Each is an unavoidable consequence of leaning the lever one way rather than the other, and the group that benefits changes with the direction of the lean rather than with anybody's opinion of them.

Why can one policy rate be hawkish in one year and dovish in another?

The section worth slowing down for is this one. Here the idea stops being a vocabulary lesson and starts being useful. A rate is a setting. A stance is that setting measured against what it is being set against. A stance is the relationship between the rate and the price reading, and a relationship needs both halves present. Neither figure on its own names a stance.

The everyday version is an air conditioner. A setting of 24 degrees says nothing about whether the machine is cooling the room or warming it. In Nagpur in May, 24 degrees is the machine working hard to pull the temperature down. In a hill town in January, the identical setting has the machine pushing the temperature up. Same number on the panel, opposite job, and the only thing that changed is the weather outside. The panel alone gives the setting. The panel held against the weather gives the machine's actual job.

The Republic of Sankhya is an invented country carried through these notes, and its invented policy rate is 6.00 per cent. The 6.00 per cent rate can be held against Sankhya year 8, where headline inflationThe price reading covering the whole basket of goods and services, before anything volatile is removed from it. How the basket is built and what the alternatives are is covered separately under inflation and prices. reads 3.00 per cent, output sits well below potential outputThe most an economy can keep producing using the workers, machines and know-how it already has, before prices start accelerating. Potential output is estimated rather than measured, and the estimates get revised. Covered separately under growth and the cycle. and unemployment reads 7.40 per cent, the highest anywhere on Sankhya's recorded path. Subtract: 6.00 less 3.00 leaves 3.00 percentage points of rate sitting above the price reading, in an economy with a great deal of slack in it. A rate sitting 3.00 points above the price reading is a lever leaning firmly against activity. Hawkish.

Now hold the identical 6.00 per cent against Sankhya year 6, where headline inflation reads 8.00 per cent, output is falling by 1.00 per cent and the output gapHow far actual output sits from the level an economy could keep up indefinitely, written as a percentage of that second figure. Covered separately under growth and the business cycle. reads minus 0.65 per cent, so output sits close to potential. Subtract again: 6.00 less 8.00 leaves minus 2.00 percentage points, meaning the rate sits two points below the pace at which prices are rising. Money is cheaper than the price reading. A rate sitting below the price reading is a lever tolerating the inflation rather than leaning against it. Dovish. The rate is 6.00 per cent in both readings, unchanged to the second decimal, and the stance is the opposite in each. Reading a rate without its conditions tells almost nothing. Sankhya year 6 is the stagflationA stretch in which output is falling while prices are rising quickly at the same time, so the two usual signals point in opposite directions. Covered separately under the business cycle. year in the invented path, which is exactly why it is the useful one to hold the rate against.

Can the rate move while the stance stays exactly where it was?

Yes, and it is more common than the reverse. If a stance is a relationship, then either half can move it, and there is no rule saying the two halves move by the same amount. When conditions move further than the rate does, the rate can fall while the leaning firms. The distance between the two has widened, even though the number on the announcement went down.

An escalator running downward, walked up, has the same shape. A step up is a real step, and yet if the escalator speeds up at the same moment the walker ends further from the top than before. Whether ground is being gained or lost has nothing to do with the step in isolation and everything to do with the step measured against the escalator. A rate cut announced into a price reading that is falling faster than the cut is exactly that step.

Work it on the invented Sankhya path. In year 7, inflation reads 6.00 per cent and output sits well below potential, and suppose the policy rate stood at 7.00 per cent, a setting invented in the same way every other Sankhya figure is. Rate less inflation gives 1.00, and with the allowance of 1.00 for output well below potential the lean reads 2.00, a hawkish reading. Move to year 8. The rate is cut by a full 100 basis points to 6.00 per cent. Any headline would report the cut as an easing. But inflation over the same stretch falls from 6.00 per cent to 3.00 per cent, a fall of 300 basis points, three times the size of the cut. Rate less inflation now gives 3.00, and with the same allowance of 1.00 the lean reads 4.00. Still hawkish, and firmer by two full points than before the cut.

Read that again slowly. The result is counterintuitive and it is correct. Inflation gave up three times as much ground as the rate did. The nominal rate fell 100 basis points and the real policy rate rose 200. An authority that cut has ended up leaning harder against activity than it was before it cut. Nothing improper has happened and nobody has made a mistake here; this is simply what it means for a stance to be a relationship rather than a level.

Readings taken off the panel and written down so they survive without it. The rate at 6.00 per cent against inflation of 3.00 per cent with output well below potential gives a lean of 4.00 and reads hawkish, the worked reading above. Leave the rate at 6.00 and move inflation alone to 8.00 per cent with output close to potential, and the lean turns to minus 2.00 and reads dovish: the same rate, a different stance. Now push the rate up to 9.00 per cent against inflation of 5.00 per cent with output close to potential, and the lean returns to 4.00 and reads hawkish again: three full points of rate away from the worked reading, and the identical stance. Reach for 4.00 per cent against inflation of 2.00 per cent. The lean reads 2.00 and the stance reads hawkish, at a rate lower than the 6.00 per cent that read dovish, which is the whole of the next section in one click.

What does each stance do to a borrower and to a saver?

A stance does not arrive in a household as a stance. A stance arrives as a number on a statement, and two households on the same street receive opposite versions of it. Those opposite versions are the cleanest way to see that no stance is simply good.

Take two invented households in Sankhya. The Bhosle household has retired, holds Rs 8,00,000/- in a fixed deposit and has no borrowing at all. The Kadre household is younger, holds Rs 24,00,000/- of borrowing taken for a house and a wedding, and has very little put aside. Rates then move up by 100 basis points, and the whole move is assumed to reach both households, an assumption made only to keep the sums simple. On Rs 8,00,000/- a move of 100 basis points is Rs 8,000/- more in a year. On Rs 24,00,000/- the identical move is Rs 24,000/- more in a year, going the other way. Same decision, same day, same country: one household is Rs 8,000/- better off and the other is Rs 24,000/- worse off. A single household holding both positions would be Rs 16,000/- worse off on the net.

The borrower and the saver sit on opposite sides of the identical move. No stance is simply good, and no stance is simply harmful. A hawkish leaning is a relief to the Bhosle household and a squeeze on the Kadre household. A dovish leaning reverses both. Anyone describing one stance as the responsible one and the other as the reckless one has quietly picked a household to speak for and not said which.

One honest caveat sits under all of that arithmetic. The assumption that the whole 100 basis points reaches both households is not how it works. How much of a policy move actually arrives in a deposit rate or a borrowing rate, and how long the transmission lagThe stretch of time between a policy rate changing and the change showing up in what households and businesses actually pay or receive. Covered separately under how a rate change reaches borrowers. runs before it does, is a separate subject covered elsewhere. Only the direction matters here: a stance that helps one of these two households is, in the same motion, costing the other.

Why does the label tell more than the level does?

A rate is a position. A stance is a position plus a direction. A reader who knows the stance can anticipate the shape of what is likely to follow. A reader who knows only the rate knows where things are and nothing whatever about where they are going. That is the whole of the label's advantage, and it is worth being precise that it is an advantage of kind rather than of accuracy. The rate is the more accurate of the two by a distance. The rate is simply answering a smaller question.

Each of the two is good for a different thing. For what borrowing costs today, the rate answers exactly and the stance does not. For whether the cost is likely to be higher or lower in a year, the rate is silent and the stance at least gestures. A business deciding whether to sign a five year lease now or wait two quarters is asking the second question, not the first, and no amount of precision about today's level helps it. The label is a compressed statement about the near future, offered by the only party that gets to decide the near future, and that is why people who follow this material read the wording as carefully as they read the number.

How much of a stance reading is interpretation rather than fact?

A great deal of it, and this is the honest limit of everything above. A rate is a number anybody can look up and everybody agrees on. A stance is inferred from wording and from a pattern of decisions. Reasonable readers therefore disagree about a stance far more often than they disagree about a rate. One of those things is measured. The other is read.

Ordinary life already shows how this works. Ten people read the same short letter from a relative and five of them think it sounded worried while five think it sounded fine. Nobody is being careless. Words carry a range, and a range is not a number, so the letter genuinely supports both readings. The same problem put in front of people who are paid to get it right, with a paragraph about whether risks are balanced or tilted, produces exactly the same spread of answers. Two analysts can read the identical statement, agree completely on every figure inside it, and file opposite conclusions about the leaning.

Three things make a stance reading softer than it looks. The wording is prose and carries a range rather than a value. The pattern of decisions has to be interpreted: when conditions are stable a hold means one thing, and when they are moving fast underneath it a hold means another. And the conditions themselves are estimates that get revised, so a reading taken today can be undermined next quarter by nothing more than a corrected statistic. None of that makes stance reading useless. A stance reading is a judgement, and a judgement should be carried with a note about how firmly it is held. A stance reading an analyst produces is a claim that may have to be revised, and a stance reading somebody else hands over is one they may have to revise too.

How does a lender actually put a stance reading to work?

Here is where the difference between a level and a direction earns its keep, in a decision somebody makes on a Tuesday. A lender funding a book of loans has to decide how to fund it: lock a cost for a fixed stretch, or leave it on a floating rateA rate that moves with a reference rate instead of staying fixed for the term. How lenders price and fund with these is covered under banking, not here. that moves as conditions move. The decision is not about where rates are today. Today's level is identical under either choice. The funding decision turns entirely on where rates go next, and a stance reading is built to gesture at exactly that.

A lender reading a firm hawkish leaning is being told that the cost of funding is more likely to rise than fall, and a lender reading a dovish leaning is being told the opposite, and the whole value of the reading sits in the word likely. Locking the cost while the leaning is hawkish buys protection from a rise. Leaving it floating while the leaning is dovish keeps the benefit of falls that have not happened yet. Neither is the correct answer. The reading can be wrong, the conditions can turn, and the leaning itself can be revised at the next meeting.

The same shape of question turns up everywhere once it is noticed. A business deciding whether to draw its borrowing now or in six months is asking it. A treasurer deciding how long to fix a deposit is asking it. In every case the question is about direction, the answer is a judgement rather than a fact, and the person answering it should know which of those two they are holding. A leaning can be read carefully and still be acted on badly, because a reading is an input to a funding decision rather than the decision itself.