Reconciling Random Walks and Predictability: A Dual- Component Model of Exchange Rate Dynamics

IMF Working Papers, December 13, 2024

Source details

Canonical URL: Reconciling Random Walks and Predictability: A Dual- Component Model of Exchange Rate Dynamics

Addresses whether exchange rates follow a random walk or exhibit predictable patterns.
Demonstrates that exchange rates can possess a unit root while maintaining substantial predictability over certain horizons.
Proposes a dual-component model combining:
- A stochastic trend representing the slowly moving equilibrium exchange rate.
- A stationary cyclical component capturing temporary deviations.

Dual-component framework reconciles long-term random walk behavior with medium-term predictability.
Role of components:
- Stochastic trend: slow-moving equilibrium exchange rate; imparts unit root behavior.
- Stationary cyclical component: temporary deviations that generate nonzero expected changes and a strong level–change relationship.
Without the stationary component, expected exchange rate changes would be zero.
If the stochastic trend evolves too quickly, the relationship between level and expected future changes would break down.
Extension: builds on Bacchetta and van Wincoop (2021) by adding a stochastic trend to generate both stationary and stochastic-trend behavior.

Expected exchange rate changes are not zero.
Expected exchange rate changes are highly persistent.
Strong relationship between exchange rate levels and expected future changes.
Forecast accuracy follows an inverted U-shaped pattern: accuracy peaks at intermediate horizons.
Multi-year exchange rate changes are increasing multiples of one-year changes.

Data: 2000–2024 for nine inflation-targeting countries with freely floating exchange rates.
Empirical findings:
- Strong empirical support for the model’s predictions.
- The model consistently outperforms the random walk benchmark in out-of-sample tests.