When Central Banks Buy Currency: How Sterilisation Works

Start with the plumbing

Imagine a central bank decides the local currency has moved too far, too fast. It steps into the foreign exchange market and sells its own currency, buying foreign currency instead. Simple enough on the surface. But every purchase has a tail that wags: paying for that foreign currency means creating domestic money, and that newly created money lands as reserves in the banking system.

Reserves are the balances banks hold at the central bank. They are the safest asset in the market, because the central bank carries virtually no counterparty risk, and unlike other high-quality liquid assets they do not need to be sold to raise cash — they are already cash, in the only form banks can settle payments with each other. So when a central bank buys foreign currency without doing anything else, the system wakes up with more reserves than before.

That is where the word "sterilise" enters. It sounds medical. In practice it is bookkeeping with a policy motive.

What sterilisation actually does

Sterilisation means the central bank offsets the reserve effect of its currency purchase. If buying foreign currency injected domestic reserves, the central bank then absorbs an equivalent amount, for example by issuing central bank bills or running reverse repurchase agreements — operations in which it takes money back in and hands out a liability instead.

The point is not to undo the currency trade. The point is to stop that trade from quietly rewriting domestic monetary conditions. Without sterilisation, a large foreign currency purchase is functionally very close to an interest rate cut: more reserves, easier money, lower short-term rates. With sterilisation, the central bank keeps the quantity of reserves roughly where it was, so it can act in the currency market without necessarily shifting its domestic policy stance.

Exchange rate policy and monetary policy are not the same dial. Sterilisation is the attempt to turn one without turning the other.

The trade-off hiding in plain sight

There is a catch, and it is not a small one. A central bank can set the quantity of reserves in the system, or it can set the price of reserves, but it cannot pin down both at once. If banks are to be willing to hold a larger pile of reserves, something has to give: all else equal, the average cost of holding them — roughly, what a bank gives up by keeping reserves rather than lending them out overnight — has to fall.

That is why the mechanics of reserve supply matter, and why central banks have been rethinking their operating frameworks. A system that fixes the quantity of reserves leaves the price to be discovered by demand. A system that fixes the price supplies whatever quantity banks want at that price. In practice, most central banks blend the two, and the labels often hide more than they reveal. Australia, Canada and Sweden run frameworks that look different on paper yet end up in a similar place: a moderately large quantity of reserves paired with a low but still positive cost of holding them.

Why some interventions move markets and others do not

Here is the uncomfortable part, and it deserves to be stated plainly rather than dressed up. Sterilised intervention — the kind that leaves the domestic money base unchanged — has a long record of disappointing results. Research on Switzerland's experience after 2009 is blunt about it: operations that did not expand the monetary base failed to prevent the Swiss franc from appreciating, and the central bank carried foreign exchange losses for its trouble. The interventions that did move the currency were the ones that let the monetary base expand.

The intuition is not exotic. If you buy a currency and simultaneously neutralise the money you created, you have changed the composition of portfolios without changing the supply of money. Currency values, over any sustained period, tend to follow money and fundamentals. Sterilised operations can nudge, signal and smooth. They rarely reverse a tide.

That does not make them pointless. Research using an event-study approach — looking at clusters of intervention days rather than isolated ones — found that intervention episodes were often successful at slowing or reversing a currency's direction over horizons of up to about two weeks when operations were persistent, larger in size and sometimes coordinated with other central banks. The same body of work is candid that this says nothing reliable about longer horizons. Intervention is a short-horizon tool. Treating it as anything more is where people get hurt.

Why central banks bother anyway

Because disorderly markets are expensive. In emerging economies especially, the motive has shifted since the global financial crisis toward financial stability. Foreign currency debt has grown, foreign investors hold large slices of local currency government bonds, and a sharp depreciation can tighten domestic financial conditions through risk limits at global banks that supply credit. In that setting, a central bank is not so much defending a number as defending market functioning — keeping the foreign exchange market liquid enough that firms with genuine hedging needs can close positions at sane prices.

Timely intervention has been found to reduce measures of foreign exchange market illiquidity. That is a narrower, more defensible claim than "the central bank can set the exchange rate."

How reserves get spent without looking spent

Notice a pattern in how central banks behave: they prefer instruments that economise on reserves. Foreign exchange swaps, forwards, options and repos can supply foreign currency or hedges without an outright drawdown of the reserve stock. Some operations settle in domestic currency, reducing pressure to buy foreign currency without actually handing any over. Others, like Chile's and Colombia's preannounced daily purchase programmes, are explicitly rule-based, signalling that reserves will move predictably and that no exchange rate target is being chased.

Mexico's experience is a useful caution here, and it is nobody's mistake — it is a genuinely hard problem. A rule-based auction scheme introduced at the end of 2014 was discontinued in early 2016 amid pressure linked to proxy hedging and signs that traders were positioning against the mechanism. Rules buy credibility; they can also hand the market a predictable target. That tension between rules and discretion has no clean answer.

What this means in practice

If you follow markets, the useful things to watch are less dramatic than headlines suggest. Watch whether reserves are rising because of valuation effects or actual purchases — the two are often confused, and in some periods valuation accounted for roughly half of reported changes in aggregate emerging market reserves. Watch whether the central bank is sterilising, because that tells you whether the currency operation is spilling into domestic interest rates. Watch derivatives positions and forward books, because intervention increasingly happens there rather than in spot. Watch the size of the reserve buffer relative to short-term external debt and the banking system.

And keep two limits in mind. First, jurisdiction matters enormously: frameworks, instruments and legal mandates differ widely, and what one central bank can do, another simply cannot. Second, nobody, including the central banks, knows the right level of reserves or the exact point where intervention stops working. The honest summary is that sterilised intervention can buy time and calm. It cannot buy a different economy.