What a central bank actually does

The one price that is actually set 

The rate in the headline is not a rate any household or firm is offered anywhere. It is the price of the shortest dated and least risky money in the system: the balances commercial banks hold at the central bank and use to settle payments with one another at the end of each day. Those balances are called reserves, exactly one institution creates them, and a payment from a customer of one bank to a customer of another is finally settled by moving them. The central bank is the monopoly supplier of the thing the banking system has to settle in, and a monopoly supplier can name its price.

Naming it is done with two standing offers rather than with an instruction. A bank holding surplus reserves can leave them at the central bank overnight and earn a stated rate. A bank short of reserves can borrow them overnight, against collateral, at a second and higher stated rate. The two offers bracket the market, because no bank lends reserves to a rival for less than it can earn by doing nothing, and none pays a rival more than the central bank would charge it. The overnight market rate therefore trades inside the bracket, and the announced target is a point inside it.

The quantity of reserves is managed alongside the two rates so the market rate settles near the target rather than drifting to one edge. What matters is the boundary. The price of overnight money is set. Every other rate in the economy, including what a mortgage costs and what a government pays to borrow for a decade, is influenced rather than set, because each is agreed between parties reading the overnight rate and forming a view about where it goes next.

What it is told to achieve 

What a central bank is trying to achieve with that price is not the institution's own choice. Mandates are written into statute by legislatures, they differ by country, and they are amended over time. The recurring items are these.

• Price stability, almost always first, usually made operational as a stated target for the annual rate of consumer price inflation, published by the bank or set for it by the government.
• A second real economy objective in some statutes, most often employment or growth, either standing beside price stability as an equal or explicitly subordinate to it.
• Exchange rate stability, where a currency is fixed to another. The exchange rate becomes the objective and the domestic interest rate becomes the means of holding it.
• Financial stability, covering the soundness of banks and of the plumbing they settle through, which in many countries brings bank supervision into the same institution.
• Issuing the currency and operating the payment system, an obligation so basic it is rarely discussed until the day it fails.
• Acting as lender of last resort to solvent banks that cannot fund themselves in a panic, a function used rarely, always under pressure, and argued about afterwards on the grounds that rescue teaches banks to expect rescue.

Two kinds of independence are commonly distinguished and constantly confused. Goal independence is the power to choose the objective. Operational independence is the power to choose how to pursue an objective somebody else has set. Most modern central banks hold the second and not the first: the target is written into law or set by a finance ministry, and the committee is then left alone to decide what to do about it. That split puts an elected body in charge of the goal and a technical body in charge of the instrument, and it is defended on the argument that the temptation to hold rates low is political while the discipline to raise them is not.

A target is not an instrument 

The most useful distinction in reading anything a central bank publishes is between what it aims at and what it can actually move. A target is an outcome the institution is answerable for and cannot set directly: nobody at a central bank can typeset next year's inflation rate. An instrument is a variable it changes by announcement and sees change immediately, because it is one of the two parties to every transaction that variable governs: the rate on its own facilities, the quantity of reserves it supplies, the securities on its own balance sheet.

1. The instrument is set, on the day, with certainty. It is the only step in the sequence the central bank performs by itself.
2. Transmission follows, unevenly and with a delay. Bank funding costs, deposit and lending rates, yields at longer maturities, credit conditions, asset prices and the exchange rate each respond to different degrees and on different timetables, and every one of them is decided by somebody else.
3. The target moves last, if it moves. Prices in shops respond to demand, demand responds to borrowing and spending, and borrowing responds to rates, a chain long enough that the effect of one decision is conventionally described as arriving over quarters and years rather than weeks.

Those lags are why committees act on a forecast rather than on the data in front of them, and a forecast is an argument rather than an observation, which is why votes split. They are also why an unchanged rate is a decision rather than the absence of one. What matters to a borrower or a lender is the rate after inflation, conventionally called the real rate, and the simplest version of it is the policy rate less the rate of inflation. When inflation falls and the policy rate is left where it is, the real rate rises, and policy has become more restrictive without anybody voting for anything.

A second consequence is a counting rule, associated with the economist Jan Tinbergen: one instrument can be aimed at one target. A central bank that has committed to holding its currency at a fixed rate against another has already spent its instrument, because its policy rate has to track the anchor country's closely enough to keep the exchange rate where it has been promised to be, whatever domestic conditions look like. The same arithmetic reappears as a trilemma: a fixed exchange rate, free movement of capital and an independent domestic interest rate cannot all three hold at once, and every country ends up choosing two.

The tools, and what each is intended to affect 

The toolkit is considerably shorter than the commentary written about it.

• The policy rate and the standing facilities that enforce it. Intended to set the price of overnight money and, through it, the front end of the yield curve.
• Open market operations. Intended to hold the market rate at the announced target by adjusting the quantity of reserves rather than a rate.
• Reserve requirements, a proportion of deposits banks must hold rather than lend. Intended to affect how much credit the banking system creates. Some institutions leave the requirement untouched for decades, others move it several times a year as a front line tool.
• Purchases and sales of securities, which change the size and composition of the balance sheet. Intended to affect yields at maturities the overnight rate does not reach, and reached for most often when the policy rate is already near its floor.
• Operations in the foreign exchange market, buying or selling the domestic currency against holdings of another. Intended to affect the exchange rate directly, and the only item here aimed at a price the institution does not otherwise touch.
• Communication. Intended to affect the expected path of every rate above, and treated by most committees as a tool of the same standing as the others rather than as a report on one.

The fourth item is worth stating mechanically, because it is the one most often described loosely as printing money. When a central bank buys a government bond, it pays by crediting the seller's bank with newly created reserves. The bond arrives on the asset side of its balance sheet and the reserves sit on the liability side: an asset acquired, a liability issued against it, and the sheet grown at both ends. What the operation achieves is contested even among the people who authorise it. One account has it working by removing long dated bonds from the market and pushing their holders into other assets, another mainly as a signal that the policy rate will stay low for a long time, and the two imply different things about what happens as the holdings are run down.

Why the words are a tool 

By the time a decision is announced, most of it is already in the price. Interest rate markets do not trade today's rate, they trade a path: the sequence of rates expected over the coming months and years, assembled from every previous statement, projection and speech. What moves on a decision day is the difference between that path and the one implied by whatever the committee has just published. A change that was entirely expected can pass with very little movement, while a rate left untouched alongside a materially different statement can move a currency further than the change itself would have.

That is why the artefacts around a decision are drafted with as much care as the decision. A committee typically publishes a statement on the day, sometimes a vote count showing how divided it was, often projections, frequently a press conference, and minutes some weeks later that reconstruct the argument. Each is read for the shape of the path rather than the level.

What it does not control 

A central bank controls the price of overnight reserves outright and nothing else outright. Everything a policy rate is supposed to reach, it reaches through a decision taken by somebody else. Banks decide how much of a change to pass into their lending and deposit rates, and how quickly. Firms and households decide whether to borrow at the new rate. Holders of long dated government debt decide what yields to accept, and those holders are global, so a domestic policy rate can rise while long term borrowing costs stay pinned by demand from abroad. Governments decide fiscal policy, which can push demand in the opposite direction to the entire monetary effort, and frequently has.

A large share of what lands in an inflation number also arrives from outside the economy altogether. Energy and food prices, freight costs, exchange rate moves and changes to indirect taxes all pass into consumer prices, and none of them responds to an overnight rate. Committees therefore distinguish between shocks of that kind, which pass through and drop out of the annual comparison, and the persistent part of inflation, which is what the instrument is aimed at. That judgement is made in real time, on data describing the recent past that arrive late and are revised afterwards.

Where practitioners disagree 

Two arguments about central banking are genuinely open. The first is rules against discretion. One tradition holds that policy should follow a published rule mapping observable variables, such as inflation and the gap between output and its sustainable level, onto a rate, because a predictable institution is easier to plan around and harder to lean on politically. The other holds that no rule survives a novel shock, that the variables a rule needs are estimates carrying wide uncertainty, and that judgement is unavoidable. Most institutions have settled in the middle, publishing forecasts and describing how they tend to react without binding themselves, which satisfies neither camp.

The second concerns the reach of the institution. Independence was granted for a narrow purpose, and a large balance sheet, bank supervision and exchange rate operations are decisions with distributional consequences taken by officials insulated from elections. Defenders argue each function is inseparable from price and financial stability, and that the alternative is a politician setting rates in an election year. Critics argue that a mandate broad enough to cover everything cannot be held accountable for anything. The argument is unsettled, and it shapes how far a committee goes and how carefully it explains itself when it does.