Covered interest parity: a forecasting approach to estimate the neutral band

Summary

Focus

In international finance, the concept of covered interest parity entails that the return of two assets, each denominated in a different currency, should be equal. Investors monitor deviations from parity because, if these are large enough, they can make a riskless profit through carry trades. Policymakers can determine if liquidity is low in the foreign exchange market when deviations are large. But how large? To answer this question, investors and policymakers estimate a neutral band around deviations from covered interest parity.

Contribution

Deviations from covered interest parity are no longer short-lived, and economists have found the main reason to be changes in the regulation of risk management. I propose a new methodology for obtaining estimates of the neutral band that deals with these new features of the covered interest parity. Investors and policymakers can use this approach in risk management decisions and liquidity monitoring. I assess the sterling-US dollar (a large global market) and Mexican peso-US dollar (the largest market for a key emerging market currency) pairs over 2000–21.

Findings

The risk management regulations in place after the Great Financial Crisis require risk managers to determine the minimum profits and maximum losses required for investors before they engage in trading. The predicted values from volatility models replicate how carry trades work. Higher volatility is related to stress events and to lower liquidity. The time-varying volatility models allow the estimates of the band to widen in periods of financial stress, when it is difficult to price currencies or loans.

Abstract

The neutral band is the interval where deviations from covered interest parity (CIP) are not considered profitable arbitrage opportunities. After the great financial crisis, deviations from CIP are no longer short-lived, exposing some limitations of the previous approaches to estimate the neutral band. In this paper, I argue that the one-step-ahead forecast distribution of deviations from CIP, with a time-varying variance component, provides an intuitive estimate of the neutral band. I use data for the Pound Sterling-US Dollar cross from 2000 to 2021, and find that a stochastic volatility model outperforms several alternative models in terms of fit and forecasting capability. The model estimates neutral band that are intuitive and consistent with market dynamics, widening during financial stress periods and consistent with no arbitrage. The results are maintained when I use data from the Mexican Peso-US Dollar cross.

JEL Classification: C52, C58, F31, F37, G15, G17

Keywords: covered interest parity; carry trade; stochastic volatility; predictive distribution