What moves a currency pair

An exchange rate is a relative price 

A share price can be discussed on its own terms, because the company behind it is a single object. An exchange rate cannot. It is a ratio between two currencies, so it is a statement about one economy measured against another, and it can be repriced by a change on either side of the comparison. That single fact is the reason currency analysis is harder than it looks, and it is the source of most of the mistakes made in describing it.

The practical form of the problem is this. When a pair rises, three explanations fit the observation equally well: the base currency was repriced upward, the quote currency was repriced downward, or both moved in the same direction and one moved further. The pair itself contains no information about which. The conventional way to separate them is to look at the same currency against several counterparts at once, which is set out at the end of this guide.

Interest rate expectations 

The most closely watched input into any exchange rate is the expected path of the two central banks' policy rates. The reasoning is direct: money held in a currency earns that currency's rate of return, so the difference between two currencies' expected returns is the most immediate comparison the market can make between them. That difference is the interest rate differential, and a very large share of day-to-day currency movement is a repricing of it.

The word expected is doing the work. Rates that a central bank has already set are known to everyone and are already reflected in the price. What moves a currency is a change in what the market expects the bank to do next, which is why a rate decision that lands exactly where it was expected can leave a rate almost unchanged while an unchanged decision accompanied by an unexpected description of the outlook can move it substantially. Market pricing of the expected path is observable in short-dated interest rate markets, and it is that pricing, rather than the current policy rate, that practitioners compare between two currencies.

Longer-dated government bond yields are read the same way and for the same reason. The gap between two countries' ten-year yields is a comparison of the returns available on capital held in each currency over a longer horizon, and several pairs, USD/JPY in particular, have historically been discussed almost entirely in those terms.

Inflation, growth and the data that feed them 

Inflation matters to a currency mostly through the rate channel rather than directly, because inflation is what a central bank is mandated to respond to. An inflation release is therefore read as information about what the bank is likely to do, and the same release can be read very differently depending on where the bank has said its attention currently sits.

Growth and labour market data enter through the same channel and through a second one. Stronger activity is conventionally read as raising the probability of tighter policy, and it also affects the attractiveness of the economy to foreign capital independently of the policy rate. The most closely followed releases are the monthly labour market reports, the inflation prints, the purchasing managers' surveys and the quarterly national accounts, and their publishers and calendars are covered in the events guides.

There is a long-run channel that operates on a completely different timescale. Purchasing power parity holds that the same basket of goods should eventually cost the same in two currencies once converted, so persistent inflation differences should show up as a currency trend over years. The empirical record supports it weakly over very long horizons and not at all over short ones, which makes it a description of a tendency rather than a tool for reading a chart.

External balances and capital flows 

A country that sells more abroad than it buys generates foreign currency that has to be converted into its own, and a country in the opposite position generates the reverse flow. That is the trade channel, and it is genuine but slow: trade flows are a small fraction of daily currency turnover and they express themselves over quarters rather than sessions.

Capital flows are far larger and far faster. Foreign purchases of a country's bonds or equities require its currency, and a reversal requires the opposite transaction. This is the channel through which a large equity market move or a sovereign credit event reaches a currency, and it is why an exchange rate can move sharply on news that has no obvious monetary content at all.

Risk appetite 

The market habitually sorts currencies along an axis that has nothing to do with the economies behind them: whether capital moves into them or out of them when confidence falls. A small group, historically the Japanese yen, the Swiss franc and the US dollar, has tended to be bought during periods of stress. Currencies of commodity-exporting and emerging economies have tended to be sold in the same periods.

This pattern is an observed regularity across many episodes and not a mechanism, and its regularity is exactly what makes it fragile. It has broken down repeatedly, most obviously when the source of the stress was the safe-haven economy itself. Treating it as a rule about how a pair will behave in the next episode extends the observation past what it supports.

Direct policy action 

Some currencies move because an authority moves them. Intervention, in which a central bank buys or sells its own currency in the market, is used to slow or reverse a move a bank considers disorderly, and it can produce the largest single-session moves in the market because it arrives unannounced and in size. Verbal intervention, conventionally called jawboning, is the cheaper version: an official statement about the level of the currency, made in the expectation that the statement alone will move it.

At the far end of the same spectrum, a currency held at a fixed rate against another is not really responding to any of the drivers above. Its rate is an administered number, and everything the market would otherwise express through it is expressed instead in the domestic policy rate, in reserves, or in the eventual moment the arrangement is changed.

Separating the two legs of a move 

Because a pair cannot say which of its two currencies moved, practitioners read the same currency against several counterparts and compare. A currency that has moved in the same direction against every counterpart has been repriced itself. A currency that has moved against one counterpart and not against the others has not: the counterpart has. Trade-weighted indices exist for exactly this purpose, and the US Dollar Index is the most widely quoted of them.

The two cases produce an identical chart of EUR/USD and have nothing else in common. Everything that would follow from one reading, which other pairs moved and which did not, which economy the information was about, and which release calendar the next piece of information sits on, differs completely between them.

Where the framework stops working 

Everything above describes channels, not predictions, and the distance between the two is large. Three limits are worth stating plainly, because they are the reason experienced practitioners disagree about currencies far more than the tidiness of the framework suggests.

1. The mapping from a data release to a direction is unreliable. A stronger than expected number is followed by a stronger currency often enough for the convention to exist and rarely enough that it is not a rule. What was already priced in is unobservable, and it is the variable that decides the outcome.
2. The dominant driver rotates. There have been years in which currencies traded almost purely on rate differentials, years in which they traded on risk appetite regardless of rates, and years in which one political process dominated everything else. Which regime is in force is usually clear afterwards and contested at the time.
3. Positioning cuts across all of it. When a large share of the market already holds the same view, the price has already moved, and the response to news that confirms the view can be smaller than the response to news that contradicts it. This is the most common explanation offered for a currency that moves the opposite way to its own data.