Decomposing Carry Trade Returns in G10 FX

1. Carry Is Not What You Think

The textbook carry trade is simple: borrow in a low-interest-rate currency, invest in a high-interest-rate currency, and earn the differential. The uncovered interest rate parity (UIP) hypothesis predicts that this differential should be offset by currency depreciation — the high-rate currency should weaken by exactly the amount of the rate advantage. Decades of empirical evidence show that UIP fails: high-rate currencies depreciate less than predicted, and sometimes appreciate. This “forward premium puzzle” is the basis of the carry trade’s profitability.

But the total return of a carry position has three components, not one. The interest rate differential is the static, predictable component. The spot return — the change in the exchange rate — is the large, volatile component that determines whether the trade is profitable in any given period. The roll yield, arising from the shape of the forward curve and the mechanics of rolling futures positions, is the often-overlooked third component. Understanding which component dominates, and when, is essential for building robust carry strategies.

2. Decomposition Framework

For a long position in currency pair i funded in the base currency, the total return over period t decomposes as:

where r_carry is the annualised interest rate differential (observable ex ante), r_spot is the change in the spot exchange rate (stochastic), and r_roll is the roll yield from the forward-spot basis (partially predictable). We compute this decomposition for all 28 G10 cross pairs on a monthly basis from 2015 to 2024.

import numpy as np

def decompose_carry(spot_t0, spot_t1, fwd_t0, ir_domestic, ir_foreign, dt=1/12):
    """
    Decompose carry trade return into three components.
    All rates annualised; dt is the holding period in years.
    """
    # Interest rate differential (carry component)
    r_carry = (ir_foreign - ir_domestic) * dt

    # Spot return
    r_spot = (spot_t1 - spot_t0) / spot_t0

    # Roll yield (forward-spot basis)
    r_roll = (fwd_t0 - spot_t0) / spot_t0 - r_carry

    r_total = r_carry + r_spot + r_roll
    return {'total': r_total, 'carry': r_carry,
            'spot': r_spot, 'roll': r_roll}

3. Variance Decomposition

Table 1: Variance decomposition of equal-weighted G10 carry portfolio, 2015–2024. The interest rate differential is the most predictable component (Sharpe 2.63) but contributes only 3% of return variance. Spot moves dominate.

This is the central finding: the carry trade’s return volatility is almost entirely driven by spot currency movements, not by the interest rate differential that motivates the trade. The rate differential is highly predictable (Sharpe 2.63) but tiny in magnitude. The spot component is enormous in magnitude but nearly unpredictable (Sharpe 0.16). The carry trader is essentially making a bet on spot currency movements, with the interest rate differential as a small tailwind.

4. Rate Cycle Dependence

The relative importance of each component varies across monetary policy regimes. During the 2015–2019 low-rate era, the interest rate differential was compressed (mean 1.2% annualised for the top-quintile carry pair), and the spot component dominated entirely. During the 2022–2024 hiking cycle, the differential widened to 4.8% for the top pair (USD vs. JPY), making the carry component more meaningful — but the spot component also became more volatile, with the yen’s sharp moves in both directions overwhelming the rate advantage.

Table 2: Carry performance by rate regime.

5. Conditional Carry Strategies

If carry returns are dominated by the spot component, and the spot component is driven by risk appetite, then conditioning the carry trade on risk measures should improve performance. We test three conditioning variables: the VIX (high VIX → reduce carry exposure), the TED spread (high TED → reduce), and the slope of the US yield curve (inverted → reduce). The VIX filter provides the largest improvement, raising the Sharpe from 0.45 to 0.68 by avoiding the worst carry drawdowns — which invariably coincide with risk-off spikes.

6. Implications for Strategy Design

Carry strategies should not be thought of as “collecting interest rate differentials.” They are fundamentally bets on currency spot movements with a small interest rate sweetener. The implication for position sizing is that carry positions should be sized based on the volatility of the spot component (typically 8–12% annualised for G10 pairs), not on the volatility of the carry return alone. Traders who size carry positions based on the interest rate differential alone will systematically underestimate their true risk exposure.

7. Conclusion

The carry trade’s return variance is 89% spot, 8% roll, and 3% interest rate differential. The interest rate component is highly predictable but economically small. Carry is effectively a bet on risk appetite with a positive expected tailwind from the forward premium puzzle. Conditional strategies that reduce exposure during high-volatility regimes improve the Sharpe ratio by approximately 50%, confirming that the primary risk in carry is the spot component’s sensitivity to global risk sentiment.