Do Jumps and Co-jumps Improve Volatility Forecasting of Oil and Currency Markets?

• Author: Jawadi, Fredj

1. INTRODUCTION

   Since the eighties, the seminal paper of Hamilton (1983) has highlighted the important effect of an oil price shock on the economy. Hamilton (1983) has been the origin of serval studies (Reboredo, 2012) showing that an oil price shock might induce both increases in inflation, unemployment, and trade deficits and a decrease in investment, yielding thus rising uncertainty and triggering an economic recession. Furthermore, as the international oil market trade is conducted in the U.S. dollar, the first channel showing the transmission of oil price shocks on the real economy is the U.S. exchange rate. Accordingly, the linkage between oil price and the U.S. dollar exchange rate is obvious and does matter for investors, hedgers, and policymakers. Indeed, a depreciation of the U.S. dollar raises the purchasing power for oil importing countries and decreases it for oil exporting countries. However, an appreciation of the U.S. currency might induce an oil price increase for oil importing countries, yielding a further oil demand shock that might affect the economies of the oil exporting countries as has occurred since the slowdown of the Chinese economy (the top oil consumer in the world) since the aftermath of the recent global financial crisis.

   Interestingly, during the last two decades, the evolution of oil prices and U.S. currency has been characterized by extreme movements. Indeed, from 2001 to 2008 the U.S. dollar fell against main currencies, such as 63.3% against the Euro, and showed similar behavior against most currencies defining the DXY index. However, since 2008, this trend has been reversed, as the U.S. dollar has gained more than 20% of its value in 2018 compared to 2008. In the same time, oil prices have exhibited great volatility, jumping from USD 27 in 2001 to 142 USD in 2008 before falling to 49 USD in 2009, then rising again and exceeding 120 USD in 2011, before another fall to 40 USD in 2018.

   To understand better these oil and USD changes and their bilateral effects, it would be interesting to investigate how oil exchange rate volatilities are related. This question appears relevant for investors in identifying investment and speculation opportunities, for hedgers regarding currency hedging, and for policymakers in terms of pricing and currency risk management. Accordingly, we study in this paper the relationship between oil and exchange rate volatilities while focusing on these extreme movements of oil price and the USD and testing whether these extreme movements might be useful in forecasting the volatilities of both markets. More specifically, we aim to analyze whether these extreme movements occur simultaneously in both markets, thereby yielding co-jumps, and also whether these jumps and co-jumps provide helpful information in improving forecasting volatilities of both oil prices and the U.S. dollar.

   From a theoretical point of view, five main channels can explain the oil-U.S. dollar relationship. First, through the oil demand and supply channel, when the U.S. dollar appreciates, oil becomes more expensive for countries whose currencies are not pegged to the U.S. dollar. This yields a cut in the real income of these oil-consuming countries, dampening their oil demand, and thereby decreasing oil prices. Thus, a negative relationship is expected between oil prices and the U.S. dollar, and the causality runs from the U.S. dollar to oil prices. The second channel is related to petrodollar recycling, which was earlier documented by Golub (1983) and Krugman (1983a, 1983b), suggesting reverse causality (from oil to the U.S. dollar). Indeed, as noted by Coudert and Mignon (2016), most members of the Organization of the Petroleum Exporting Countries (OPEC) hold their wealth from oil production in U.S. dollars. Thus, an oil price increase implies a further increase in demand for U.S. dollar assets, as these petrodollars boost the USD exchange rate. Third, similar to the second channel and with reference to the literature on the long-run real exchange rate (Clark and McDonald, 1998), oil might affect U.S. currency because changes in oil prices might drive important variations in trade through the Balassa-Samuelson effect. (1) A fourth channel has been observed through the recent financialization of commodities and, in particular, the development of commodity futures since mid-2005, which has yielded an on-going arbitrage between financial assets, foreign exchange (forex) currencies, and commodity contracts (Domanski and Heath, 2007; Greely and Currie, 2008). (2) All things being equal, a fall in U.S. financial asset prices would push investors to prefer commodities, yielding an oil price increase and therefore, this financialization channel also illustrates a negative oil-dollar relationship from the dollar to oil prices. Finally, a fifth indirect channel through U.S. monetary policy might explain the oil-dollar relationship. For example, when the Fed increases the U.S. interest rate, it might lead to lower investment, a higher dollar (through the uncovered interest parity theory), yielding a slowdown in the world economy with lower oil demand, which implies a lower oil price and also supports a negative relationship between oil prices and the U.S. dollar.

   From an empirical viewpoint, the related literature does not provide a unanimous conclusion regarding this oil-dollar relationship. Indeed, Hamilton (1983) showed that oil prices affect U.S. macroeconomic variables. Amano and Van Norden (1995) also found oil prices as the main driving factor for the long-term evolution of U.S., German, and Japanese exchange rates. Sadorsky (2000), Zhang et al. (2008), Reboredo et al. (2014), and Coudert and Mignon (2016) found a negative relationship between oil and the dollar, while De Truchis and Keddad (2016) found no significant relationship between the dollar and oil in the long run. Huang and Guo (2007) found a weak relationship between the U.S. dollar and oil prices, while Golub (1983), Krugman (1983a), and Bloomberg and Harris (1995) found a positive relationship. Ding and Vo (2012) found a bidirectional transmission between oil and the U.S. currency markets, while more recently, Ferraro et al. (2015) showed that a significant relationship between oil prices and the U.S. dollar is more pronounced in the very short-term. Overall, there is no empirical consensus on the co-evolution of oil prices and the U.S. dollar exchange rate and this heterogeneous result can be explained in many different ways: i. an unstable correlation between oil prices and the U.S. dollar since the 2000s, with this relationship alternating between positive (1975-1979, 1986-1989, and 1996-2002) and negative (1976, 1980-1981, 1991-1993, and 2007-2008); ii. the difference between oil-U.S. dollar relationship in the short and the long terms; and, iii. the presence of jumps and co-jumps in both markets implying more complexity. (3)

   Further, in the related literature, this relationship has always been investigated using low-frequency data, without distinguishing between continuous and discontinuous price changes, while the availability of high-frequency data for both markets might help to improve characterization of this relationship and improve forecasting of oil and the U.S. dollar volatility, through taking their abrupt movements into account. (4) This is particularly interesting, as the West Texas Intermediate spot price moved from USD25/barrel in 2000 to USD145 in July 2008 and USD43 in June 2017, with various jumps over the last period. The U.S. dollar exchange rate has also fluctuated significantly over the last years, yielding further evidence of jumps and co-jumps in the oil and U.S. exchange rate markets.

   More particularly, the issue ofjumps in the oil market was not extensively discussed in the literature except in some recent studies (Askari and Krichene, 2008; Chevallier and Lepo, 2013; Sevi, 2015; Da Fonseca and Ignatieva, 2019). Moreover, these studies focused mostly on the factors impacting oil price jumps. As for jumps in the U.S. dollar market, we also noted only a few studies (El Ouadghiri and Uctum, 2016; Kilic, 2017). Regarding the relationship between jumps across oil and the U.S. dollar markets, Jawadi et al. (2016) investigated the relationships between jumps in the oil price and in the Euro/U.S. dollar and they found further evidence of volatility spillover effect through jumps from the exchange rate to oil price. Li et al. (2017) investigated the conditional spillover effect between oil price jumps and exchange rate on a weekly basis. However, these studies do not allow an analysis of the co-jumps between these two markets, which correspond to a simultaneous occurrence of jumps in both markets.

   Unlike the few above studies, we focus on jumps, co-jumps, and even multiple jumps between oil and the U.S. dollar exchange rate markets. Co-jumps are simultaneous jumps or abrupt price changes that occur for different assets that simultaneously react to common news. Further, co-jumps might be associated with common information on economic fundamentals (Das and Uppal, 2004; Bollerslev et al., 2008).

   Accordingly, using intraday data for the West Texas Intermediate (WTI) oil index, the U.S. dollar effective exchange rate, and its main components--the Euro (EUR), Japanese Yen (JPY), British Pound (GBP), Canadian Dollar (CAD), Swiss Franc (CHF), and Swedish Krona (SEK), our paper provides an original intraday study of the oil-U.S. dollar relationship over the period September 1, 2014 to April 30, 2018. (5) Thus, we conduct an original empirical investigation of the instantaneous intraday linkage between oil prices and the U.S. exchange rates that makes several contributions to the related literature. First, we compute intraday jumps for both markets using intraday jump tests proposed by Andersen et al. (2007a) and Lee and Mykland (2008), which enable us to instantaneously identify extreme movement in...