ABSTRACT
We examine the relationship between oil prices, foreign exchange (FX) swaps and local interbank offered rates in the six Gulf Cooperation Council (GCC) countries. We also investigate the potential hedging and diversification benefits from adding oil positions to portfolios containing GCC FX swaps or interest rate positions. Our findings confirm that oil predicts, and in some cases causes, movements in the various GCC FX swaps and interbank offered rates. We also find that the Saudi FX swap market has the highest volatility spillover from the oil market compared to other markets in the region. Furthermore, our analysis shows a significant change in liquidity conditions in the GCC FX swap markets following a sudden shift in oil prices. Lastly, we document the presence of significant risk reduction benefits from adding oil exposure to portfolios of GCC FX swaps or interest rates with risk going down by at least half in the case of the GCC FX swaps.
Disclosure Statement
No potential conflict of interest was reported by the author(s).
Notes
1. As of the time of writing this study, the Central Bank of Bahrain priced its FX swap facility far from the prevailing market rates with the implicit intention of receiving USD liquidity rather than injecting it through such facility.
2. We discuss the current literature about the relationship between oil and financial markets in the GCC in the literature review below.
3. We focus our analysis on FX swaps and not FX spots because most of the GCC currencies are pegged to the USD and fluctuate within a relatively tight range.
4. As discussed earlier, USD is the second leg in all FX swaps involving GCC currencies.
5. The adjustment is done through dividing the FX swap price by the number of days applicable on that day and then multiplied by 92 days.
6. For instance, a 1 pip increase in the price of FX swap from 1 to 2 pips might sound like a 50% increase, whereas it is a marginal increase in the price of FX swap and implied interest rates especially considering the tenor used in our study (three months).
7. The conclusions are the same using the one-year FX swaps and interbank offered rates.
8. For instance, interbank offered rates are published in Kuwait and Qatar, however no derivatives or commercial activities are indexed against such rates.
9. The results of the study with regards to FX swaps and interbank offered rates remain unchanged when considering observations during the Covid-19 pandemic only. This is likely a result of the liquidity injection and stabilizing efforts by the various central banks.
10. The index was adopted by several similar studies (Awartani, Maghyereh, and Shiab Citation2013: Maghyereh, Awartani, and Bouri Citation2016; McMillan and Speight Citation2010) and is also called the directional connectedness measure. A full explanation of the measure is available from Diebold and Yilmaz (Citation2012, Citation2015) and Maghyereh, Awartani, and Bouri (Citation2016).
11. Following the convention in the market and the literature we use the highest bid and the lowest ask to calculate the spread.
12. We calculate MEC using a one-month moving average. We use 20 (5) trading days to calculate the one-month (one-week) variance.
13. Following Bai and Perron (Citation2003), we allow up to five breaks and use a 15% trimming, which means that each segment has at least 15% of the data. We also allow error distributions to differ across segments and account for serial correlation using lagged variables. All five breaks are significant at 5%.
14. Full results of the model are available with the authors and can be provided upon request.
15. We arrive at the same conclusion when applying the Lagrange multiplier unit root test with two structural breaks proposed by Lee and Strazicich (Citation2003).
16. The models are run with the individual FX swap market as a dependent variable while all other GCC markets as well as the lagged value of the dependent variable are included as explanatory variables. We use LSDV instead of the generalized method of moments (GMM) because the former provides more consistent estimates in models with large time series and small cross-sectional data (Mensi et al. Citation2017). The results are available from the authors and can be provided upon request.