FX Carry and the Forward Rate Bias

If interest-rate parity worked the way a textbook says it should, currency carry trades would not exist. A currency paying a higher interest rate should be expected to depreciate by roughly that same amount, so a trader borrowing the low-yield currency and holding the high-yield one would earn the extra interest but lose it right back on the exchange rate. In practice, for decades, that has not been what happens.

The theory versus the data

Uncovered interest parity (UIP) says the expected change in the spot exchange rate should offset the interest-rate differential exactly, so no currency is a systematically better place to park cash than another once expected currency moves are included. Covered interest parity, a close cousin, is close to mechanically true because arbitrage in the forward market enforces it. UIP has no such enforcement mechanism — it is a bet on average investor expectations, not a no-arbitrage condition — and the data has rejected it consistently since the 1980s. Currencies with higher short rates have tended to appreciate slightly on average, or depreciate by far less than parity predicts, the opposite of what the textbook implies. This gap between prediction and reality is the forward rate bias, or the forward premium puzzle.

Why the trade exists

The carry trade is the direct exploitation of this bias: borrow in a currency with a low policy rate, convert the proceeds into a currency with a higher policy rate, invest it there, and collect the rate differential. If UIP held, the expected profit from this would be zero. Because it does not hold, the strategy has earned a positive average return over long samples across many currency pairs, resembling an insurance premium: carry traders are effectively short volatility, collecting steady income most of the time in exchange for occasional violent losses when funding currencies suddenly strengthen and high-yield currencies sell off together — a carry crash. The 2024 yen carry unwind was a textbook instance of this pattern.

Drag the parameters above and notice how a mean-reverting path can drift steadily in one direction for a long stretch before snapping back — carry returns behave similarly: calm accumulation followed by an abrupt reversal, not a smooth random walk.

Worked example

A trader funds in Japanese yen at a 0.1% annual rate and holds Australian dollars yielding 4.1%. Ignoring exchange-rate moves entirely, the carry earned on $10 million notional over one year is:

$$(0.041 - 0.001) \times 10{,}000{,}000 = 400{,}000$$

That is $400,000 of pure interest-differential income, before any currency move. If AUD/JPY is flat over the year, the trader pockets it. If UIP held exactly, the yen would be expected to appreciate against the Australian dollar by close to 4% over the year, wiping out most of that $400,000. Historically the yen has appreciated by far less, which is why the position has had a positive expected return — until a shock year, when AUD can fall 10-15% against JPY in weeks and erase years of collected carry at once.

What this means in practice

Systematic FX carry strategies hold baskets of several high-yield currencies against several low-yield funding currencies rather than a single pair, which diversifies away some of the pair-specific risk but does nothing to protect against the crash risk, since carry crashes tend to hit most high-yield currencies simultaneously. Risk managers size carry books assuming fat left tails, not the smooth returns the average-return numbers suggest.