Paper: arXiv 2504.06028

Authors: SeungJae Hwang

Abstract

This paper examines the empirical failure of uncovered interest parity (UIP) and proposes a structural explanation based on a mean-reverting risk premium. We define a realized premium as the deviation between observed exchange rate returns and the interest rate differential, and demonstrate its strong mean-reverting behavior across multiple horizons. Motivated by this pattern, we model the risk premium using an Ornstein-Uhlenbeck (OU) process embedded within a stochastic differential equation for the exchange rate. Our model yields closed-form approximations for future exchange rate distributions, which we evaluate using coverage-based backtesting. Applied to USD/KRW data from 2010 to 2025, the model shows strong predictive performance at both short-term and long-term horizons, while underperforming at intermediate (3-month) horizons and showing conservative behavior in the tails of long-term forecasts. These results suggest that exchange rate deviations from UIP may reflect structured, forecastable dynamics rather than pure noise, and point to future modeling improvements via regime-switching or time-varying volatility.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced stochastic calculus (Ornstein-Uhlenbeck SDEs, derived closed-form approximations) and rigorous statistical estimation, demonstrating high mathematical density. It is empirically robust, featuring a detailed backtesting framework with coverage-based metrics, real financial data (USD/KRW 2010-2025), and quantitative performance evaluation across multiple horizons.

Research Flowchart

  flowchart TD
  A["Research Goal: Explain UIP failure<br>via Mean-Reverting Risk Premium"] --> B["Data: USD/KRW 2010-2025"]
  B --> C["Methodology: Calculate Realized Premium<br>UIP Deviation = FX Return - Interest Diff"]
  C --> D["Modeling: Fit Ornstein-Uhlenbeck OU Process<br>to Premium Dynamics"]
  D --> E["Computational Process: Derive Closed-Form<br>Future Exchange Rate Distributions"]
  E --> F["Validation: Coverage-Based Backtesting<br>across Short, Medium, Long Horizons"]
  F --> G["Key Outcomes: Strong predictive power at<br>short/long horizons; 3-month underperformance"]
  F --> H["Implication: Structured, forecastable<br>dynamics vs. pure noise"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes a mean‑reverting Ornstein‑Uhlenbeck model for the exchange‑rate risk premium and embeds it in a continuous‑time stochastic exchange‑rate equation.

Source sentence
a mean-reverting model of the exchange rate risk premium, motivated by persistent empirical deviations from the uncovered interest parity (UIP) condition.

Assumes 6

  • The risk premium follows a single‑regime Ornstein‑Uhlenbeck process. model
    Source sentence
    Modeling the premium as an Ornstein‑Uhlenbeck (OU) process embedded within a stochastic differential equation for the exchange rate
  • The risk premium is assumed to vanish in the long run (initial hypothesis). model
    Source sentence
    Starting from the hypothesis that the premium should vanish in the long run—allowing exchange rates to eventually revert to parity
  • Exchange‑rate and interest‑rate data are daily and can be split into a training (80%) and validation (20%) set for out‑of‑sample backtesting. data
    Source sentence
    The dataset is divided into a training set (80%) and a validation set (20%), with the latter used for out-of-sample evaluation.
  • Log‑exchange rates are approximately normally distributed, allowing log‑normal approximation for the spot rate. statistical
    Source sentence
    Since log St is approximately normal, we can approximate the probability density function (PDF) of St using the log‑normal distribution.
Show 2 more
  • Forecast confidence intervals are constructed using the approximate log‑normal distribution of future spot rates. model
    Source sentence
    Given the approximate log‑normality of St+h, confidence intervals for future spot rates are constructed as:
  • The single‑regime OU model cannot capture transitional dynamics around the 3‑month horizon. model
    Source sentence
    This highlights a potential limitation of the single‑regime OU model in capturing transitional dynamics.

Data

  • USD/KRW exchange rate and interest rates market-prices unknown
    January 2010 to April 2025
    Source sentence
    We use daily data on the USD/KRW exchange rate and interest rates from January 2010 to April 2025.

Costs

Not applicable The paper reports no trading or execution result.

Finds 5

  • The model shows strong empirical coverage at short‑term horizons (2 weeks and 1 month), with coverage rates slightly exceeding nominal confidence levels. performance
    Source sentence
    Coverage rates slightly exceed nominal levels, indicating a conservative but reliable forecast.
  • Coverage at the 6‑month horizon aligns closely with target confidence levels, indicating the mean‑reverting structure remains informative at medium‑to‑long horizons. performance
    Source sentence
    The model also performs well at the 6‑month horizon, with coverage rates closely aligned with target confidence levels.
  • The 3‑month horizon consistently underperforms, with empirical coverage below nominal levels, suggesting a transitional volatility zone not captured by the single‑regime OU model. negative
    Source sentence
    The 3‑month horizon consistently underperforms, with coverage falling below nominal levels across all intervals.
  • At the 1‑year horizon the model produces overly conservative tail forecasts, yielding near‑perfect coverage (≈99.98%–100%) that likely reflects over‑estimated long‑run uncertainty. performance
    Source sentence
    near‑perfect coverage, reaching 99.98% and 100%, respectively. While this might seem desirable, it more likely reflects an overestimation of long‑term uncertainty due to the OU process’s asymptotically stabilizing variance.
  • Overall, the mean‑reverting risk‑premium model captures key exchange‑rate dynamics, confirming that deviations from UIP are structured and forecastable rather than pure noise. theoretical
    Source sentence
    Our work contributes a tractable framework that connects short‑run fluctuations with long‑run equilibrium through a dynamic, mean‑reverting premium.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 36 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (2)

  1. Forward and spot exchange rates (1984)
  2. Stochastic Differential Equations: An Introduction with Applications (2003)

0 of 2 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.