The purpose of this study is to revisit the relationship between US dollar exchange rates and crude oil prices in the post-2000 era, when both oil prices and USD exchange rates experienced record-high volatility. Specifically, I aim to answer: (1) do oil prices drive US dollar exchange rates or vice versa, (2) are oil prices and US dollar exchange rates integrated in the long-run, and (3) are the conflicting results found in previous studies on the relationship between the two due to the use of different frequencies of data (e.g. daily, weekly, and monthly). To answer these questions, I applied time series procedures, in particular Granger causality tests based on vector auto-regressive and vector error correction models, on daily, weekly, and monthly oil prices and US dollar exchange rates against a variety of foreign currencies (including both oil-importing and oil-exporting countries) from January 2000 to September 2017. Results from the analyses suggest that the relationship between oil prices and exchange rates does differ at daily, weekly, and monthly frequencies. In daily frequency, AUD/USD, CAD/USD, EUR/USD, and NOK/USD are cointegrated with oil prices while no long-run relationship has been found between JPY/USD, MXN/USD, and GBP/USD with oil prices. The pattern of cointegrated and non-cointegrated variables remained the same in weekly period. However, in monthly period, NOK/USD is not cointegrated with oil prices highlighting the effect of different time frequencies on the relationship between US dollar exchange rates and oil prices. The result of this study provides evidence that in all cointegrated cases, when deviations to the long-run equilibrium occur, the US dollar exchange rates corrected its deviation by adjusting back the oil prices to their equilibrium. Furthermore, the type of causality varies in different time frequencies. In daily and monthly frequencies, unidirectional, bidirectional and no causality have been detected while in weekly frequency, all short-run relationships are independent from each other. Therefore, as opposed to daily and monthly frequencies, there is no possibility of predictability in weekly frequency.
展开▼
