The Monetary System, Part 3: Rates & Inflation (2026)

Framework evergreen; figures dated. Data as of August 2026. How a central bank works and how QE transmits are durable and read that way here. Every figure is a dated benchmark from the source named beneath its table — re-source before reuse. Balance-sheet-to-GDP ratios are approximate and flagged; QE programme sizes are headline totals. Informational only, not investment advice; AI-assisted and human-reviewed (see Section 14). Update cadence: annual.

The first guide in this series explained why governments run deficits; the second, how they borrow. This one covers the institution that sits above both — the central bank. Since 2008, central banks have moved from quiet rate-setters to the most powerful actors in the economy, printing trillions to buy bonds and, in 2022, slamming the brakes as inflation returned for the first time in forty years. This guide explains what a central bank actually controls, who runs it, and each of its powers in turn — from setting interest rates to quantitative easing — alongside the record of rates and balance sheets across the major economies since 1995. It is the free primer; for the company- and project-level data these policy regimes ultimately move, that is what Metal Pilot is for.

The headline: a central bank sets the price of money and, when that runs out of room, changes the quantity of it — and inflation is the scoreboard that tells it whether it went too far.

This is Part 3 of “The Monetary System” — a three-part series on how government money works. The parts: Part 1 · Spending & Debt , Part 2 · Bonds & Yields , and Part 3 · Interest Rates & Inflation (this guide).

TL;DR & Key Takeaways

• A central bank controls one thing directly: the short-term policy rate. Everything else — long yields, mortgages, the currency — it influences through expectations.
• Different banks answer to different masters. The Fed alone carries a jobs-and-prices dual mandate; the ECB, Bank of England and Bank of Japan put price stability first — and the ECB is the outlier, setting one interest rate for 21 countries that still run their own budgets.
• When rates hit zero, the toolkit changes. Quantitative easing (QE) — creating reserves to buy bonds — became the main tool after 2008, once rates could fall no further.
• The balance sheets exploded. The Fed’s assets went from ~6% of GDP in 2007 to ~35% by 2022; the Bank of Japan’s exceed 125%. This is the visible fingerprint of QE.
• 2022 flipped the script. When inflation returned, the major central banks delivered the fastest rate hikes in forty years and began unwinding QE — while China, facing weak demand, eased instead, a divergence that shapes commodity and asset markets.
• What they own is mostly bonds — and, increasingly, gold. A balance sheet is dominated by government bonds (from QE) or, for reserve-accumulators like China, foreign currency. Gold is the one reserve that is no government’s liability — the US holds the most at 8,133 tonnes, while Poland, China and India lead the record central-bank buying.

New to the topic? Read straight through from Section 1. Here for the payoff? Jump to the checklist in Section 11. In a hurry? Table 1 maps every power to its section, and every term is defined in the Vocabulary (Section 13).

1. What a central bank actually controls

A central bank is the bank for the banks — and the monopoly issuer of a currency. Its job, in most advanced economies, is a mandate: keep inflation low and stable (usually a 2% target), and — for some, like the US Federal Reserve — also support maximum employment. It pursues that mandate with one direct lever and a great deal of influence.

The direct lever is the policy rate (the federal funds rate for the Fed, the deposit rate for the ECB, the Bank Rate for the Bank of England): the interest rate at which banks lend to each other overnight. By moving it, the central bank sets the price of money at the very short end. Everything else — the rates on mortgages, corporate loans, and the government bond yields from the previous guide — is influenced indirectly, because those long rates are built partly from the expected path of the policy rate (recall the yield’s largest component in the bond guide).

That indirect power runs through expectations. When a central bank signals that rates will stay low for years — forward guidance — long yields fall today, before a single rate has moved. Much of modern central banking is the management of expectations, not just the setting of a rate.

The final, contested pillar is independence. Most central banks are legally insulated from day-to-day politics, so they can raise rates into an unpopular slowdown or resist pressure to fund deficits directly. That independence is the foundation of their credibility on inflation — and it is exactly what comes under strain when debts are high and governments want cheap financing, the tension running through this whole series.

1.1 What a central bank can do

Behind that one headline lever sits a handful of distinct powers — things a central bank, and only a central bank, is able to do. It sets the short-term interest rate; it issues the currency and creates the economy’s base money; it can buy financial assets on a vast scale (quantitative easing); it stands behind the banking system as the lender of last resort; and it holds and manages the nation’s foreign-exchange reserves. Most of the time only the first is in active use — the rest are held back for stress. Each has its own section later in the guide, so the table below is the map.

Table 1. What a central bank can do

The durable powers a central bank holds; the remaining sections take each in turn. Bank supervision sits alongside these in many countries and, for the euro area, at the ECB (Table 3).

2. Who runs a central bank

Section 1 covered what a central bank does; this section covers who decides it — who writes the mandate, who staffs the bank and for how long, and how the euro area splits power between the ECB and its member states. Governance is not a footnote: the term lengths and appointment rules below are the machinery that makes independence real or leaves it exposed.

2.1 Mandates: what each bank is told to do

A central bank does not choose its own objective — the political principal sets it in law. The mandates then differ in one important way. The US Federal Reserve is unusual in carrying an explicit dual mandate: maximum employment and stable prices, given equal statutory standing. The European Central Bank, the Bank of England and the Bank of Japan instead put price stability first, treating growth and employment as secondary — pursued only “without prejudice” to the inflation goal — and most converge on a ~2% target (Table 2). The outlier is China’s People’s Bank, which is not independent: it operates under the State Council with several objectives at once — currency stability, growth, employment, the exchange rate — and takes direction from the government.

2.2 Appointments and terms

Independence is built less by the words of a statute than by the design of appointments — who names the leadership and how long they serve without needing reappointment. A long, staggered or non-renewable term insulates a policymaker from the political cycle; a short, renewable one leaves them wanting another term from the government that grants it.

The Fed’s governors are nominated by the President and confirmed by the Senate to single 14-year terms, staggered so one expires every two years — longer than any presidential term, so no president can remake the Board (the Chair serves a renewable 4-year term). The ECB’s Executive Board is the mirror image for the same purpose: eight-year, non-renewable terms, appointed by the European Council, non-renewability the deliberate feature that removes any incentive to court reappointment. The others fall between — the Bank of England’s Governor serves a single eight-year term, the Bank of Japan’s a renewable five-year term (the shortest of the majors), and the People’s Bank of China’s governor no protected term at all, consistent with an arm of the state (Table 2).

Rates are set not by the leader alone but by a committee, one member one vote. The Fed’s Federal Open Market Committee (FOMC) carries 12 votes — the seven Governors, the New York Fed president, and four of the other eleven regional presidents on yearly rotation — while all 19 policymakers debate. The ECB’s Governing Council seats 27 (the six-member Executive Board plus 21 national governors), with voting rights capped at 21 and rotating. The Bank of England’s MPC and the Bank of Japan’s Policy Board have nine members each. The chair casts just one vote, yet the role outweighs the arithmetic: the chair sets the agenda, builds consensus, and is the bank’s public voice, whose press conferences move markets more than the vote count — and almost never ends up on the losing side.

Table 2. Central-bank mandates, decision-makers and terms

Sources: Federal Reserve , ECB , Bank of England , Bank of Japan , Deutsche Bundesbank , Banque de France , National Bank of Poland , Bank of Canada and the Reserve Bank of Australia ; ECB appointments via the European Council . Germany and France set no national rate — the Bundesbank and Banque de France sit inside the Eurosystem, and the euro-area rate is decided by the ECB Governing Council. Structural governance facts; mandates paraphrased from each bank’s statute.

2.3 The ECB: one bank, many countries

Every bank above answers to one nation. The ECB is the exception: it runs monetary policy for the 21 countries that use the euro — Bulgaria became the 21st on 1 January 2026 — which makes its structure, and its limits, unlike any single-country central bank’s.

Before the euro, each member ran its own central bank, currency and interest rate: the Bundesbank, the Banque de France, the Banca d’Italia. The euro replaced all of that with one currency and one interest rate for the bloc. The national central banks did not disappear — with the ECB they form the Eurosystem — but they no longer set rates or create money. The single rate is decided by the ECB’s Governing Council (the six Executive Board members plus the 21 national governors); the national banks then implement it, distribute banknotes, hold reserves and gather statistics. Sovereignty over the money moved to the centre; the national banks became its branches.

What did not move to the centre matters just as much. The ECB controls the euro area’s money; it does not control its budgets. Taxing, spending and government borrowing stay fully national — there is no euro-area treasury and no shared national debt. The result is a monetary union without a fiscal union, and that split defines what the ECB can and cannot do.

Table 3. The euro area: what the ECB runs vs. what stays national

Source: ECB — tasks of the Eurosystem and the division of competences under the EU treaties. The “capital key” is each country’s share of ECB capital, set by its population and GDP.

Three constraints follow, none of which binds a single-country bank. First, one size must fit all: a single rate is set for economies rarely at the same point in the cycle — too tight for a member in recession, too loose for one booming. That mismatch drove the euro’s first two decades, when a low rate suited to a sluggish Germany helped inflate periphery booms that later broke. Second, the ECB is barred from financing governments directly: under the EU treaties’ ban on monetary financing it may not lend to a state or buy its bonds at issue, so its QE is bought in the secondary market and rationed across countries by the capital key. Third, because rates are shared but debts are national, a scare about one government’s solvency can splinter euro-area yields apart — fragmentation — even with the policy rate unchanged; preventing that produced the ECB’s most famous interventions, the 2012 “whatever it takes” pledge and the anti-fragmentation backstops since (Section 10).

The contrast is the sharpest way to hold it: the Fed sets one rate for one treasury under one government; the ECB sets one rate for 21 treasuries under none. The non-euro EU members kept their own central banks and rates precisely to avoid that trade-off; the euro’s members traded it for a shared currency, and the tension between shared money and national budgets is the defining feature of central banking in Europe.

3. Interest rates

Section 1 called the policy rate the one lever a central bank controls directly. This section opens that lever up: the different kinds of interest rate and how they connect, how a change in the policy rate spreads through the economy and why that makes it the most-watched number in finance, how often and on what evidence the decision is actually made, and where the rate has sat in every year since 2000.

3.1 The types of interest rates

“Interest” is the price of borrowing money, quoted as a percent per year. But there is no single interest rate — there is a chain of them, and the central bank sets only the one at the very start.

At the base is the policy rate (also called the target rate): the rate the central bank administers directly — the Fed’s federal funds target range, the ECB’s deposit rate, the Bank of England’s Bank Rate. It anchors the overnight interbank rate, what banks charge each other to borrow reserves overnight, which the central bank steers to sit at its target. Alongside it is the discount rate — the rate at which the central bank itself lends directly to banks as a backstop, set a little above the policy rate.

The overnight rate is worth pausing on, because it is the foundation the whole structure rests on. Banks hold reserves — balances at the central bank used to settle payments between them — and at the close of each day some banks run a surplus while others fall short. The short banks borrow the surplus overnight, and the interest they pay is the overnight interbank rate. Being the shortest and safest rate in the system, it is the one the central bank actually targets: pin down the overnight rate, and every longer rate is built on top of it through expectations of where it will go next (the yield curve). This is why a rate decision is expressed as a target for this rate and nothing else.

The rate that actually trades is a published benchmark in its own right — the effective federal funds rate in the US, €STR in the euro area, SONIA in the UK — and these reference rates (which replaced the discredited LIBOR) price trillions of dollars of loans and derivatives quoted as “the overnight rate plus a margin.” The central bank keeps the market rate on target by managing the supply of reserves: historically through open-market operations that add or drain them, and since 2008 — with reserves abundant — mainly by paying banks a set rate of interest on reserves, which puts a floor under the market. So when a headline says the Fed “raised rates by a quarter-point,” what literally moved is this target for the overnight rate; everything else below follows from it.

From that base, everything else is priced off the policy rate by the market, not set by the central bank. The prime rate, banks’ benchmark for their most creditworthy customers, sits roughly three points above the fed funds rate and drives credit-card and business-loan rates. Government bond yields across maturities form the yield curve, built from today’s policy rate plus the market’s expected path of future rates (the subject of the bond guide). And the rates households actually pay — mortgages, car loans, credit cards — are priced off those yields and the prime rate.

Two distinctions run through all of it. First, a nominal rate is the quoted figure; the real rate is that minus expected inflation, and it is the real rate that decides whether money is genuinely cheap — a 5% rate against 6% inflation is negative in real terms. When a central bank deliberately holds the real rate negative for years to erode the real value of government debt, that is financial repression. Second, economists reason about the neutral rate (r*), the policy rate that neither stimulates nor restrains the economy: a central bank is “loose” below it and “restrictive” above it. Moves are measured in basis points — one basis point is 0.01%, so a “quarter-point” cut is 25 basis points.

Table 4. The main types of interest rate

Conceptual reference, not measured data. The central bank sets only the first rate directly; the rest are priced off it by the market.

3.2 How rates move the economy

A central bank moves one overnight rate, and the effect radiates across every borrowing and saving decision in the economy — which is why the policy rate is the single most important number in finance. Raise it, and credit becomes dearer and saving more rewarding, cooling demand and, in time, inflation; cut it, and the reverse. That transmission runs through several channels at once (Table 5): the cost of borrowing, housing, business investment, the incentive to save, asset prices, and the exchange rate.

Two features make this power both formidable and dangerous. Its reach is enormous — one decision reprices mortgages, corporate investment, the government’s own interest bill and the currency together. And it works with long and variable lags: a rate change takes roughly twelve to eighteen months to feed fully through to inflation. The central bank is therefore always steering toward where it thinks the economy will be, not where it is — which is why rates are moved pre-emptively, and why over- or under-tightening is obvious only in hindsight.

Table 5. How a rate change spreads through the economy

Conceptual framework, not measured data. All six channels act together and with a lag of roughly 12–18 months.

3.3 When rates are decided

Rate decisions are not made continuously but at scheduled meetings. The Fed’s FOMC meets eight times a year, about every six weeks; each meeting ends with a statement, and since 2019 every one is followed by a press conference. Four times a year — March, June, September and December — it also publishes the Summary of Economic Projections, including the “dot plot,” each policymaker’s anonymous forecast for the rate. The ECB, the Bank of England and the Bank of Japan all run a similar eight-meetings-a-year rhythm with their own projection rounds. Between meetings the rate normally stays put; only a real emergency brings an inter-meeting move — the Fed’s emergency cut to zero in March 2020 is the textbook case.

By the day it lands, the decision itself is rarely a surprise. Central banks telegraph their intentions through speeches and guidance (the expectations channel of Section 1), so the market has usually priced the move in advance; what actually moves markets is the surprise — a decision, or a projection, that differs from what was expected.

Table 6. When the major central banks decide

Sources: Federal Reserve , ECB , Bank of England , Bank of Japan , National Bank of Poland , Bank of Canada and the Reserve Bank of Australia . Germany and France set no separate schedule — the euro-area rate is decided by the ECB Governing Council. Scheduled meetings; emergency inter-meeting decisions can occur.

3.4 What the central bank watches

The mandate tells the Fed what to target — maximum employment and ~2% inflation — so above all it watches the data that measure those two things. On inflation, its preferred gauge is the core PCE price index from the U.S. Bureau of Economic Analysis (BEA), released monthly, alongside the more timely CPI. (How these gauges are built and who publishes them is covered in Part 2 of this series .) On employment, the marquee release is the monthly Employment Situation report — “the jobs report” — whose nonfarm payrolls and unemployment rate anchor the labour read, with JOLTS job openings, jobless claims and the quarterly Employment Cost Index filling in demand and wages. Growth comes from GDP and the monthly ISM business surveys, and — because expectations can be self-fulfilling — it tracks survey- and market-based measures of expected inflation. Table 7 lists the dashboard; no single number decides a meeting — the committee weighs the whole of it against its dual mandate.

Table 7. Key data the Fed watches

Sources: BEA , BLS , ISM and University of Michigan . The Fed’s ~2% target is defined on PCE inflation.

3.5 Policy rates since 2000

Put the Fed’s policy rate on a timeline and the whole period since 2000 reads as one series: two collapses to near zero — after the 2008 crisis and again in the 2020 pandemic — and two climbs back, the last of them the fastest tightening in forty years, before the easing to 3.50–3.75% that held through 2026. Table 8 gives the year-by-year record, charted in Figure 1; why each turn happened — the crises, the inflation, the balance sheet moving alongside — is the story of Section 8.

Figure 1. The US policy rate since 2000: two collapses to zero and two climbs back.

Source: year-end levels from Table 8 (upper bound of the target range from December 2008). Every year 2000–2025 is a bar; a repeated identical value is labelled once, on the first bar of the run.

Table 8. US federal funds target rate by year, 2000–2025

Source: Federal Reserve FOMC decisions, via FRED (target range, upper bound from December 2008). Year-end level; single target before 2008, a 0.25-point range since. The rate was held at 3.50–3.75% through mid-2026.

4. Currency & money creation

If interest rates are the price of money, this section is about the money itself — the central bank’s second great power. The one institution that can create money from nothing, it nonetheless does not create most of the money in the economy — and seeing who does is the key to the whole subject.

4.1 The layers of money

Money comes in layers. At the core is base money (the “monetary base,” or M0): physical cash — banknotes and coins — plus the electronic reserves commercial banks hold at the central bank. Only the central bank can create base money. Around it sits broad money (M2): the money households and firms actually spend, which is overwhelmingly bank deposits — the balances in current and savings accounts. Base money is the small foundation; broad money is many times larger.

Table 9. The layers of money

Conceptual reference; M0 and M2 are standard central-bank monetary aggregates. The exact composition and the ratio between the layers vary by country and over time.

4.2 What bank reserves are

The reserves folded into base money are the part of the money system the public never sees, because only banks can hold them. Money sits in two tiers: households and firms hold deposits at commercial banks, while those commercial banks in turn hold reserves — electronic balances in their own accounts at the central bank. Reserves are the banks’ own money and, alongside physical cash, the only form of central-bank money there is; you and I can hold the cash, but never the reserves.

Their core job is settlement. When you pay someone who banks elsewhere, the two banks square up by shifting reserves between their central-bank accounts — so reserves are the final means of payment between banks, the plumbing beneath every transfer. Banks also hold them as a liquidity buffer, and, where the rules demand it, to meet a reserve requirement.

Two things make reserves unusual. Only the central bank can create or destroy them — electronically, at will — and they never leave the central-banking system: reserves move from one bank’s account to another but are never handed out to the public, because a bank lends by creating a new deposit, not by paying away its reserves. That is why reserves are the hinge of so much of this guide — the overnight rate is the price of borrowing them (Section 3), quantitative easing creates them on a vast scale (Section 5), and the interest a central bank pays on them is now its main lever for keeping market rates on target (Section 3).

4.3 Who actually creates money

Here is the counter-intuitive part: most money is created by commercial banks, not the central bank. When a bank makes a loan, it does not hand over someone else’s deposit — it simply credits the borrower’s account with a new deposit, creating money in the act of lending. (The old textbook image of banks “multiplying up” a fixed base of reserves is misleading; in practice lending creates the deposits, and the central bank supplies whatever reserves the system then needs to settle.) The central bank shapes this indirectly, by setting the price of money through the policy rate (Section 3) — which makes borrowing more or less attractive — and directly, by creating reserves itself, which is exactly what quantitative easing does (Section 5).

4.4 Cash, reserves and “printing money”

When people say a central bank “prints money,” they rarely mean literal banknotes — physical cash is a shrinking share of the total and is supplied simply on demand. What they mean is the central bank creating reserves electronically, with a keystroke, to buy assets or lend. The profit it earns from issuing money — the gap between the near-zero cost of creating it and the interest the assets it buys pay — is called seigniorage, and it is handed to the treasury. In nominal terms the central bank can create money without limit; the discipline that stops it is the inflation mandate, because money created faster than the economy can supply goods ends in rising prices. That link is loose in the short run — the reserve creation of the 2010s did not stoke consumer inflation, while the mix that returned inflation in 2020–22 did (Section 5) — but over the long run, currencies are debased by creating too much money.

5. Quantitative easing

Because it is the tool that changed everything after 2008, QE deserves its own section.

What it is. In quantitative easing, the central bank creates new bank reserves — electronically, “printing money” in the loose sense — and uses them to buy financial assets, overwhelmingly government bonds, from banks and investors. Once the policy rate reaches its floor — roughly zero, the zero lower bound, below which savers would sooner hold cash than accept a negative return — it cannot cut further, so instead of changing the price of money it changes the quantity: it floods the banking system with reserves and takes bonds out of the market. (A few central banks — the ECB, the Bank of Japan — pushed their policy rate slightly below zero in the 2010s, but the floor is close.)

Why it works — two channels. First, the portfolio-balance channel: by buying long-dated bonds, the central bank pushes their price up and their yield down (the seesaw again), which lowers borrowing costs across the economy and nudges investors toward riskier assets like equities and corporate debt. Second, the signalling channel: a commitment to buy bonds tells the market the central bank is serious about keeping policy loose, reinforcing forward guidance. Both work by pushing down long yields and loosening financial conditions when the policy rate can’t fall any further.

The transmission to inflation. QE is meant to raise spending and investment by making money cheap and plentiful — and, ultimately, to lift inflation back toward the 2% target when it is running too low (the fear through the 2010s was deflation, not inflation). For a decade, QE seemed to boost asset prices far more than consumer prices. Then 2020–21 happened: unprecedented QE plus massive fiscal transfers plus supply shocks combined, and inflation returned with force (Section 8). Whether QE “caused” the 2022 inflation is fiercely debated, but the episode ended the assumption that QE was a free lunch.

The reverse operation is quantitative tightening (QT): letting bonds mature without replacing them, which drains reserves and shrinks the balance sheet, tightening policy without touching the rate. The major central banks began QT in 2022 to unwind the pandemic expansion (Section 8).

The scale, across the major central banks, was enormous.

Table 10. Major quantitative-easing programmes

Sources: Federal Reserve , ECB , Bank of England . Headline programme totals; the Bank of Japan pioneered QE from 2001 (its balance sheet, the largest relative to GDP of any major central bank, is in Section 8).

5.1 Short- and long-term consequences

In the short term, QE does what it is designed to do: long-term yields fall, borrowing gets cheaper, and financial conditions loosen. Asset prices rise — bonds, equities and housing all re-rate as the discount rate drops and investors are pushed toward risk — and the currency tends to weaken as domestic yields fall relative to abroad. Bank reserves swell. The effect on the real economy is slower and more muted, which is why a decade of QE lifted markets far more visibly than it lifted growth or wages.

Over the long term, the consequences are more contested. The balance sheet becomes enormous and awkward to unwind, so QT is slow and can jolt markets. Years of rising asset prices flatter the asset-owning wealthy, feeding concerns about inequality. The central bank ends up holding a large share of the government’s debt, blurring the line between monetary and fiscal policy and inviting the charge that it is quietly financing deficits. Investors come to expect support in every downturn — the so-called “central-bank put.” And when rates finally rise, the central bank takes losses on the low-yielding bonds it bought, handing a bill back to the taxpayer. None of this makes QE a mistake, but it is why it is no longer seen as the free lunch it once appeared to be.

6. Lender of last resort

A central bank’s fourth power appears only in emergencies, but it is the one that defines the modern institution: it is the lender of last resort. When panic hits the financial system and depositors and lenders pull their money at once, even a solvent bank can collapse simply because it cannot raise cash fast enough. The central bank stops the run by doing what no private lender will — lending freely, against good collateral, to any sound institution that needs it. The nineteenth-century editor Walter Bagehot reduced the rule to a line still quoted today: in a crisis, lend freely, against good collateral, at a penalty rate.

The routine version is the discount window (Section 3), the standing facility through which banks can always borrow from the central bank. In a real crisis the central bank goes much further, standing up special lending facilities and, across borders, opening swap lines. A swap line is a standing agreement between two central banks to exchange currencies: in a global dollar shortage the Fed lends dollars to, say, the ECB against euros, the ECB passes them to euro-area banks that cannot create dollars, and at maturity the swap reverses at the same rate — so neither bank bears exchange-rate risk. Because banks all over the world borrow and lend in dollars they cannot print, the Fed’s swap lines effectively make it the world’s lender of last resort in dollars — a decisive backstop in both 2008 and 2020.

Some rescues, though, need the taxpayer rather than the central bank. Recapitalising a failed bank, or guaranteeing depositors and creditors, spends public money, and that is the Treasury’s job, not the central bank’s — the US TARP in 2008, the euro area’s ESM. This is where the monetary and fiscal arms visibly act as one: in the depths of 2008 the Fed cut rates to zero, launched QE and opened swap lines while the Treasury stood up TARP to recapitalise the banks (the fuller crisis sequence is in Section 10).

Table 11. Crisis tools: who wields them

Conceptual framework, not measured data. The central bank supplies liquidity; only the Treasury can commit public money.

Each rescue, however, leaves the central bank more powerful and more entangled with the state’s finances — which is exactly why its independence (Section 1) is watched so closely, and why every crisis reopens the question of where monetary policy ends and fiscal policy begins.

7. Foreign-exchange reserves

The central bank’s last power looks outward, to the currency’s value against other currencies. Every central bank holds foreign-exchange reserves — a stockpile of foreign currencies (mostly US dollars), together with gold and claims on the IMF — as a national war chest. The reserves let a country pay for imports and service its foreign debt even when private markets seize up, and they are the ammunition for acting on the exchange rate.

That action is FX intervention: buying or selling currency in the market to move its price. To prop up a falling currency, the central bank sells reserves and buys its own money; to hold a currency down — often to keep exports cheap — it buys foreign currency and sells its own, accumulating reserves in the process. In many countries the Treasury sets exchange-rate policy and the central bank executes it, so this power straddles the monetary–fiscal line.

There is a hard asymmetry here. A central bank can create its own currency without limit, but its foreign reserves are finite — so defending a peg by selling reserves works only until they run out, which is how many currency crises end. The largest reserve holders are the emerging economies, above all China, which built the world’s biggest stockpile partly to manage the yuan; how those reserves and the dollar’s central role interact is the subject of the companion series, The Dollar Order .

8. Rates and balance sheets, 1995–2025

Put the two variables — the policy rate and the balance sheet — on one timeline and the last thirty years split cleanly into three acts. (Section 3 gave the policy rate’s full year-by-year path in Table 8; this section pairs it with the balance sheet and explains why each turn happened.)

Act one: the Great Moderation (1995–2007). Inflation was low and stable, and central banks managed the economy with the policy rate alone. The Fed funds rate sat around 5–6.5% in normal times; the ECB and Bank of England similar. Balance sheets were small and boring — the Fed’s assets around 6% of GDP. Central banking was, by later standards, quiet.

Act two: the zero-rate decade (2008–2020). The financial crisis drove policy rates to the zero lower bound, where they stayed for the best part of a decade, and forced the pivot to QE. Balance sheets ballooned — though by very different amounts around the world (Table 12).

Table 12. Central-bank balance sheets at their peak, ~% of GDP

Sources: Federal Reserve (H.4.1), ECB , Bank of Japan , Bank of England , Bank of Canada , Reserve Bank of Australia and BIS ; approximate and rounded, at each bank’s post-2008/2020 peak. Japan is the outlier, its balance sheet larger than the whole economy. Germany and France share the euro-area Eurosystem balance sheet (the ~60% is of euro-area GDP, not national); China’s reflects foreign-exchange reserves rather than QE, and Poland ran only limited asset purchases.

The Fed’s expansion is the clearest single picture of the QE era.

Figure 2. The Fed’s balance sheet exploded after 2008.

Source: Federal Reserve (H.4.1), total assets as a share of GDP; approximate and rounded.

Act three: the tightening (2021–2025). After two decades near or below target, inflation broke out in 2022 (the QE-plus-fiscal-plus-supply-shock mix of Section 5). (How inflation is measured and the full 2000–2025 price record are covered in Part 2 of this series ; the point here is the policy response.) That response was the fastest tightening in forty years: the Fed took its policy rate from near zero to above 5% in 2022–23, the ECB and Bank of England close behind, and all three began quantitative tightening — letting bonds roll off to shrink the balance sheets QE had built (the Fed’s assets easing from ~35% back toward ~24% of GDP by 2024).

The one divergence is China. While the West tightened hard, the People’s Bank of China eased, its economy flirting with falling prices as a property downturn dragged on demand. That split — tightening West, easing East — is a defining macro feature of the mid-2020s, and it pulls hard on currencies, capital flows and commodities.

9. What a central bank owns

The last section measured the balance sheet by its size; this one is about its contents. Everything a central bank does leaves a mark on the asset side: the bonds it buys under QE (Section 5), the foreign currency and gold it keeps in reserve (Section 7), and the loans it makes to banks (Section 6) are all assets it holds — funded by the reserves and cash it issues on the other side of the ledger. A handful of durable classes make up almost all of it, and their mix says a lot about what a given central bank is for.

9.1 The assets on the balance sheet

Table 13. The main assets a central bank holds

Conceptual reference, not measured data. Composition varies by country and shifts with policy; the “largest for” column names where each class dominates.

Two things sort the major banks. A domestic-QE bank holds mostly its own government’s bonds — the Fed, the Bank of Japan, the Bank of England and the Eurosystem all fit this mould, their balance sheets a mirror of the bonds bought after 2008 and 2020. A reserve-accumulator holds mostly foreign assets: the People’s Bank of China is the extreme case, its balance sheet built on the world’s largest foreign-exchange pile rather than on QE (which is why its ~30% of GDP in Table 12 is “mostly FX, not QE”). The one asset almost all of them share is gold — but in wildly uneven amounts.

Table 14. What sits on each major central bank’s balance sheet

Sources: Federal Reserve (H.4.1 / SOMA), Bank of Japan , Deutsche Bundesbank , Bank of England , Banque de France , National Bank of Poland , Bank of Canada , Reserve Bank of Australia ; gold from the World Gold Council . Composition shares approximate, mid-2026; the Fed’s QT ended in December 2025. Gold in tonnes — Canada sold its last bullion in 2016, the only G7 nation to hold none.

9.2 Gold: the largest official holders

Gold is the one asset in Table 14 that is no government’s liability — it cannot be printed, defaulted on, or frozen by an issuer, which is exactly why central banks keep it (Section 7). Who holds it splits the world in two. The advanced economies that built their reserves under the gold standard still keep most of their reserves in gold — over 70% for the US, Germany, Italy and France — and rarely trade it. The Asian reserve giants hold very little relative to their means: China and Japan carry gold at a single-digit share of far larger reserves. A third group — the emerging-market buyers — has been closing that gap, lifting official gold demand above 1,000 tonnes a year across 2022–2024, the strongest run on record, as they diversify away from the dollar (the theme of the companion series, The Dollar Order ). Poland was the single largest buyer in the first half of 2026, pushing the National Bank of Poland — one of the banks in Table 14 — into the global top ten.

Table 15. The ten largest official gold holders

Source: World Gold Council , from IMF International Financial Statistics; holdings mid-2026, reserve shares approximate. National central banks only — the IMF holds ~2,814 t and the ECB itself ~507 t as supranational bodies, and Turkey (~614 t) and the Netherlands (~612 t) sit just outside the ten.

Figure 3. The United States holds more official gold than the next two holders combined.

Source: gold holdings from Table 15 (World Gold Council ).

The pattern beneath the ranking is what matters for markets: the advanced economies sit still on legacy hoards while the emerging markets buy, and that steady official bid has been a structural support under the gold price. For the gold miners geared to that price, Metal Pilot is where the company-level data begins.

10. What triggers intervention

Central banks and treasuries do not act at random; a few triggers reliably bring them off the sidelines, and the past two decades supply the case studies.

A financial panic is the classic trigger: the 2008 banking collapse pulled in both the central bank and the Treasury at once, the crisis response detailed in Section 6. A sovereign-debt scare triggered the euro-area response of 2012 — the ECB’s “whatever it takes” and the creation of the ESM rescue fund. A collapse in demand triggered the largest intervention of all in 2020, when the pandemic brought zero rates, trillions in QE and vast fiscal packages together within weeks. And inflation itself is the trigger in reverse: the 2021–23 surge forced the sharpest tightening in a generation.

Table 16. Major interventions, 2008–2022

Sources: Federal Reserve , ECB , US Treasury . Illustrative selection of landmark interventions.

The pattern: the bigger the shock, the more the monetary and fiscal arms act as one — the entanglement, and its cost to central-bank independence, is the thread running through Section 6 and the whole series.

11. How to read a central bank

The ten sections above explain what a central bank controls and each lever in turn. This section turns them into a repeatable read: six questions to ask of any central bank at any moment, each answered by a section you have already met, and each carrying a signal for the assets a monetary regime prices. It is a framework for orientation, not advice, and no single answer decides anything on its own.

First, which mandate is it under? A dual-mandate bank (the Fed) reacts to a jobs slowdown in a way a price-stability-first bank (the ECB, Bank of England, Bank of Japan) will not, and the People’s Bank of China answers to the State Council (Section 2). Second, where is the policy rate relative to neutral? Policy is loose below the neutral rate (r*) and restrictive above it, so its distance from neutral matters more than its level (Section 3). Third, is the rate the tool, or the balance sheet? At the zero lower bound the action moves to the quantity of money — QT restrains even while the headline rate holds (Sections 4–5). Fourth, read the real rate, not the nominal: a rising nominal rate can still be a falling real one, and a real rate held negative is financial repression (Section 3). Fifth, which way is the balance sheet moving — QE loosening conditions, QT tightening them (Section 8)? And sixth, West or East? The mid-2020s split — the West tightening, China easing — is the live divergence pulling on currencies and commodities (Section 8).

Table 17. Reading a central bank: six questions and their tells

Conceptual framework, not measured data — a way to orient, not a rule to trade on. Each read points back to the section that explains it.

Why this matters beyond the mechanics: because real interest rates — the policy rate net of inflation — ultimately decide the return on cash, bonds and real assets alike, monetary regimes are the backdrop against which commodities and the equities tied to them are priced — the gold that tends to shine when real rates fall, the industrial metals geared to growth. How those regimes map onto commodity exposure is the subject of the macro-regime guide ; this guide supplies the policy engine underneath it. Take the six reads above to the companies and commodities a regime actually prices — that is what Metal Pilot is built for.

12. Summary

A central bank controls one thing directly — the short-term policy rate — and everything else through expectations, under a mandate to keep inflation near 2%. Who wields that lever differs by country — the Fed under a jobs-and-prices dual mandate, most others under price stability first, and the ECB uniquely setting one rate for 21 euro members that still run their own budgets — and it is a committee, not a chair alone, that decides. That policy rate sits atop a whole chain of borrowing costs, is set at eight meetings a year against a dashboard of inflation and jobs data, and since 2000 has twice collapsed to zero and twice climbed back, last easing to 3.50–3.75% by the end of 2025. When the crisis of 2008 drove rates to zero, it reached past the price of money to its quantity, launching quantitative easing on a scale that took the Fed’s balance sheet from ~6% to ~35% of GDP and the Bank of Japan’s beyond 125%. What sits on those balance sheets is mostly government bonds — or, for reserve-accumulators like China, foreign currency — alongside gold, the one reserve that is no one’s liability, of which the US holds the most while the emerging markets have lately been the keenest buyers. For a decade QE seemed to lift asset prices more than consumer prices; then 2020–22 combined QE, fiscal transfers and supply shocks, and inflation returned with a vengeance — ~8% in the US, ~9% in the UK and euro area — forcing the fastest rate hikes in forty years and a pivot to quantitative tightening, even as China moved the opposite way toward deflation. The triggers are consistent — panics, debt scares, demand collapses, and inflation itself — and each one binds the monetary and fiscal arms more tightly together. This closes the three-part arc: the fiscal guide showed the debt, the bond guide showed how it’s financed, and this one shows the policy engine that sets the rates deciding it all. When you are ready to move from these macro regimes to the companies and commodities they price, that is what Metal Pilot is built for.

Next → the companion series The Dollar Order , on the international monetary system — the dollar, trade and de-dollarization.

13. Vocabulary

14. Sources, methodology & disclaimer

14.1 Sources, methodology & data vintage

Central-bank balance-sheet-to-GDP ratios are approximate, from the Federal Reserve (H.4.1), ECB , Bank of Japan , Bank of England , Bank of Canada , Reserve Bank of Australia and BIS ; euro-area figures are Eurosystem-wide (shared by Germany, France and the other members). QE programme sizes are headline totals from each central bank. Policy-rate levels are described from central-bank records. The mandates, decision-makers and term lengths in Section 2 cover nine economies and are structural facts drawn from each institution’s governing statute and official pages (Federal Reserve , ECB , Bank of England , Bank of Japan , Deutsche Bundesbank , Banque de France , National Bank of Poland , Bank of Canada and the Reserve Bank of Australia ) plus the European Council ; Germany and France are represented by their national central banks within the Eurosystem, whose rate is set by the ECB, and the euro area comprised 21 members after Bulgaria adopted the euro on 1 January 2026. The year-by-year US policy-rate history (Table 8, and every year in Figure 1) is from Federal Reserve FOMC decisions via FRED , shown as the year-end target level (a single figure before December 2008, a 0.25-point range since); it stood at 3.50–3.75% through mid-2026. The indicators the Fed watches (Table 7) are official US statistical releases from the BEA, BLS and ISM at the stated frequencies. The rate types and transmission channels (Tables 4–5) and the decision cadences (Table 6) are conceptual and structural references. The layers of money (Table 9) use the standard M0/M2 monetary aggregates, whose exact composition varies by country. The crisis tools in Table 11 are a conceptual framework, not measured data. The asset-composition figures in Section 9 (Tables 13–14) are approximate and structural — dominant asset classes and rounded balance-sheet shares from each central bank’s own disclosures (Federal Reserve H.4.1/SOMA, Bank of Japan , Deutsche Bundesbank , Bank of England , Banque de France , National Bank of Poland , Bank of Canada and the Reserve Bank of Australia ); the Fed’s QT ended in December 2025. The official gold holdings (Table 15 and Figure 3) are from the World Gold Council , compiled from IMF International Financial Statistics, at mid-2026, with the reserve-share percentages approximate; the ranking lists national central banks only, with the IMF and ECB noted separately as supranational holders. Data as of August 2026; policy and prices move continuously, so re-source before reuse.

14.2 Disclaimer & disclosure

This guide is for informational purposes only and is not investment advice, a recommendation, or a solicitation to buy or sell any security. Figures are estimates as of the stated date and are revised; readers should verify against the primary sources before relying on any number. Nothing here should be taken as a forecast of interest rates, inflation, or any market outcome. This article was prepared with AI assistance: figures were gathered from the named public sources and human-reviewed, but readers should independently confirm before acting. The author holds no position that the content is designed to promote.