Does Bitcoin hedge crude oil implied volatility and structural shocks? A comparison with gold, commodity and the US Dollar

Journal Pre-proof

Does Bitcoin hedge crude oil implied volatility and structural shocks? A comparison with gold, commodity and the US Dollar Debojyoti Das , Corlise Liesl Le Roux , R.K. Jana , Anupam Dutta PII: DOI: Reference:

S1544-6123(19)30672-5 https://doi.org/10.1016/j.frl.2019.101335 FRL 101335

To appear in:

Finance Research Letters

Received date: Revised date: Accepted date:

3 July 2019 1 October 2019 20 October 2019

Please cite this article as: Debojyoti Das , Corlise Liesl Le Roux , R.K. Jana , Anupam Dutta , Does Bitcoin hedge crude oil implied volatility and structural shocks? A comparison with gold, commodity and the US Dollar, Finance Research Letters (2019), doi: https://doi.org/10.1016/j.frl.2019.101335

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Inc.

Highlights      

We examine the hedging and safe-haven properties of Bitcoin against crude oil. We compare the performance of Bitcoin with gold, commodity and the US Dollar. The Dollar is a better alternative to hedge oil implied volatility (OVX) than Bitcoin. Bitcoin outperforms gold and commodity in hedging OVX. Bitcoin is a better hedge to demand-side oil shocks in comparison to Dollar. Bitcoin is not the superior asset over others to hedge oil-related uncertainties.

1

Does Bitcoin hedge crude oil implied volatility and structural shocks? A comparison with gold, commodity and the US Dollar Debojyoti Das1, Corlise Liesl Le Roux2, R. K. Jana3*, Anupam Dutta4 1

Woxsen School of Business, Hyderabad, Telangana, India American University in the Emirates, Dubai, United Arab Emirates 3 Indian Institute of Management Raipur, Chhattisgarh, India 4 University of Vaasa, Vaasa, Finland 2

*Corresponding author

Does Bitcoin hedge crude oil implied volatility and structural shocks? A comparison with gold, commodity and the US Dollar Abstract In this article, we examine the hedging and safe-haven properties of Bitcoin against crude oil implied volatility (OVX) and structural shocks using a dummy variable GARCH and quantile regression model. In addition, we also compare the hedging and safe-haven performance of Bitcoin with gold, commodity and US Dollar. We conclude that Bitcoin is not the superior asset over others to hedge oilrelated uncertainties. Besides, hedging capacity of different assets is conditional upon the nature of oil risks and market situation. Thus, investors may prefer different investment instruments to hedge downside risks in different economic situations and market states. JEL classifications: C22; G15 Keywords: Bitcoin; Gold; Commodity; Hedge; Safe-haven; Crude oil; Quantile regression.

The resilience of Bitcoin during the phases of financial turmoil has indicated its hedging and safe-haven potentials against global economic adversities, thereby it is often regarded as the “digital gold” (Selmi et al., 2018). Bouri et al., (2017a) argue that the past incidents of rising Bitcoin prices during uncertain economic states (such as during the European Sovereign Debt Crisis (2010-13) and Cypriot Banking Crisis (2012-12)) were the prime reason for its prominence as a safe haven avenue. Thus, the hedging and safe-haven capabilities of Bitcoin with reference to diversified streams of uncertainty shocks have become a subject-matter of interest for the academicians and practitioners alike. 1 Bouri et al., (2017a) investigate the hedging capabilities of Bitcoin against the global uncertainties proxied by the option implied Volatility Indexes (VIXs) of 14 developed and developing markets. The findings of their study qualify Bitcoin to be an effective hedging instrument. Similarly, several other studies empirically observe the hedging capacities of Bitcoin with respect to Economic Policy Uncertainty (EPU) (Demir et al., 2018; Wang et al., 2018; Wu et al., 2019). Their studies 1

The readers may refer Corbet et al., (2019) for a broad coverage of studies concerning the financial asset feature of cryptocurrencies. 2

conclude that Bitcoin can be used as a safe-haven and diversifier asset for portfolio perspective. Likewise, Bouri et al. (2018) inspect the sensitivity of Bitcoin to global financial stress and document that Bitcoin is a safe-haven asset to financially turbulent conditions. Similarly, Aysan et al. (2019) inspect the association of Geopolitical Risk (GPR) and Bitcoin and conclude in favour of Bitcoin as a hedging tool. Besides, Urquhart and Zhang (2019) document that Bitcoin can act as a hedge for the fiat currencies. Most of the studies argue in favour of Bitcoin to hedge risk of various forms, however, some studies also report contradictory findings. For instance, Bouri et al., (2017b) evaluate the safe haven and hedging properties of Bitcoin on energy and non-energy commodity markets. They conclude that the hedging and save-haven properties were no longer present in Bitcoin after the price crash in December 2013. Bitcoin exhibited only limited diversification benefits for the non-energy commodity indexes. Correspondingly, Bouri et al., (2017c) examine the hedging

potential of Bitcoin against world stock indexes, bonds, commodities and the US Dollar. Their results show that Bitcoin provides poor hedging opportunities, except for a safe haven asset in the Asian (Japanese and Asia Pacific) stock index, during extreme down movements. Moreover, Smales (2018) critiques Bitcoin on several grounds such as volatility, illiquidity and transaction cost. The author suggests that the save-haven dimension of Bitcoin must be considered prudently until the market reaches the stage of maturity. Additionally, Klein et al. (2018) pose that Bitcoin is less effective than the traditional safe-haven asset such as gold. Further, Bouri et al., (2019) also find that Bitcoin lacks in stable hedging capabilities as a portfolio component. Dorfleitner and Lung (2018) investigate the diversification benefits of Bitcoin along with seven other cryptocurrencies using a mean-variance framework. Their results indicate that seven of the eight cryptocurrencies provided diversification benefits in bullish market periods only. Shahzad et al., (2019) report that Bitcoin did not show any strong safe-haven properties for the world stock markets under consideration; however, some weak safe-haven properties were identified in their study. Kang et al., (2019) focus upon the degree of co-movement between Bitcoin and gold futures in order to determine the hedging properties and diversification opportunities between Bitcoin and gold. They find the evidence of intense co-movements during the crisis periods, thereby limiting the safe-haven feature of Bitcoin. Of lately, Gozgor et al. (2019) find a negative impact of Trade Policy Uncertainty (TPU) on Bitcoin. Though it is true that Bitcoin is not associated with variables of the real economy, nevertheless, some relational dynamics between crude oil and Bitcoin may be postulated. Oil prices may have an impact over Bitcoin on several counts. Energy is a crucial requirement for mining Bitcoin. The scale of power consumption for companies that run the software of bitcoin aggregates to 22 terawatt-hours (TWh), which is close to the annual energy demand of Ireland. 2 Thus, the increasing oil prices is expected to steer up the cost of Bitcoin production (Das and Dutta, 2019) and hence may influence the price fundamentals (Bouri et al., 2017b). van Wijk, (2013) and Ciaian et al., (2016) argue that potential changes in oil price signal the shift in general price level, which could appreciate (or depreciate) the value of Bitcoin. Further, the inflationary pressures led by oil price rise could depreciate the fiat currencies. Thus, a currency which is decoupled from the real economy, such as Bitcoin could be used to settle the cross-border payments. Consequently, the price of Bitcoin could 2

The Economist, “Why bitcoin uses so much energy”, https://www.economist.com/the-economistexplains/2018/07/09/why-bitcoin-uses-so-much-energy, accessed February 24, 2019, 20:28 hours (Indian Standard Time). 3

rise corresponding to the demand. Additionally, we must recognize the fact that the rise in oil price due to demand or supply side pressures endow distinct economic implications (Kilian, 2009). The upsurge in oil price due to higher demand of oil is referred as the demand shock. Whereas, supply side shocks indicate the dearth of oil availability or increased cost of oil extraction leading to higher oil price. The recent studies have shown that these shocks differently and significantly impact the traditional assets such as stocks, precious metals and fiat currencies (Atems et al., 2015; Clements et al., 2019; Uddin et al., 2018). Now to what extent the postulated oil linkages can influence the price of Bitcoin is a question that is intriguing, hence worth examining. Selmi et al., (2018) examine the hedging potential of Bitcoin against oil price movements and conclude that Bitcoin can be used as a hedging instrument. Nevertheless, as discussed above, disentangling the oil price movements into shocks is essential to understand the nature of price movements. Thus, it is imperative to understand the nature of shock against which Bitcoin could be an effective safe-haven and hedge. We further extend the scope of our work to test the hedging capabilities of Bitcoin with respect to option implied crude oil volatility index (OVX). The recent literature has clearly established the adverse impacts of OVX on stocks, currencies and precious metals (Dogah and Premaratne, 2018; Dutta, 2018; Singh et al., 2018). Thus, we can offer additional insights concerning the suitability of Bitcoin in hedging OVX. We find that Bitcoin hedges the demand-side shocks better that supply-side shocks. Besides, US Dollar is better hedge to OVX and commodity is a better hedge to supply and risk shocks. Overall, we find that though Bitcoin offers certain hedging and safe-haven benefits, however, it is not the superior hedging instrument over the other alternative assets. Thus, investors may use these different assets to hedge downside risks based on the economic situation and market state. The rest of the article is structured as follows: Section 2 describes the data. Section 3 presents the estimation strategy. Section 4 presents the results and Section 5 concludes the paper. 2. Data Our sample includes prices of Bitcoin, gold, commodity, US Dollar and the OVX. In order to disentangle the oil price shocks using the structural vector autoregressive (SVAR) framework of Ready (2018), we use three more variables: (a) stock index of oil and gas producing firms (MSCI All Country World Index, Energy Index), (b) CBOE Volatility Index (VIX) and (c) to proxy for the oil price changes, we use the 1 month returns on NYMEX -Light Sweet Oil contracts. The period of our study span over 20 th July, 2010 to 20th June, 2019 enveloping 2292 daily observations. The dataset is extracted from the Bloomberg. The Bitcoin, gold and commodity price indexes are denominated in US Dollar. For a representing general commodity, we use the Bloomberg commodity index (BCOM). Whereas, to proxy for the US Dollar we have used the Bloomberg Dollar Spot Index (BBDXY), this index tracks the performance of 10 leading global currencies against the US Dollar. This diverse set of composition currencies are important for trading and liquidity perspective and hence is an improved measure of US Dollar (Bloomberg Inc.). We consider the logarithmic return of the of the variables Bitcoin, gold, commodity, US Dollar and the OVX. The structural crude oil shocks, i.e., Demand Shock (DS), Supply Shock (SS) and Risk Shock (RS) are derived from the SVAR procedure.3 The summary statistics of the variables under consideration is exhibited in Table 1. As we can observe the returns of Bitcoin is highest as compared to gold, commodity and Dollar. Nevertheless, the standard deviation of Bitcoin returns is also high as compared to the other assets. The skewness coefficient is negative for Bitcoin, gold and commodity indicating the more frequent negative returns values. The kurtosis coefficient The shock decomposition SVAR algorithm recently suggested by Ready (2018) is discussed in the estimation strategy section. 3

4

shows that the returns distribution has excess kurtosis than the Gaussian distribution. The Jarque-Bera test coefficient shows that the distribution of the variables departs from the normality assumptions. The Phillips Perron (PP) and Augmented Dickey-Fuller (ADF) test results prove the stationarity condition of the variables. Table 1. Descriptive statistics Bitcoin Gold Commodity Dollar OVX DS SS RS Mean 0.509 0.007 -0.021 0.006 0.007 -0.001 -0.001 0.000 Median 0.202 0.022 -0.008 0.008 -0.394 -0.002 0.026 -0.010 Maximum 51.704 4.581 3.702 1.832 32.779 0.707 4.685 8.641 Minimum -60.094 -9.512 -4.783 -1.760 -21.911 -0.829 -6.521 -7.012 Std. Dev. 6.626 0.955 0.839 0.386 4.705 0.152 1.138 1.520 Skewness -0.320 -0.647 -0.240 0.014 1.139 0.076 -0.318 0.234 Kurtosis 17.073 9.993 5.217 4.708 9.010 5.421 5.736 5.771 Jarque-Bera 18945.020 4827.353 491.121 278.423 3942.750 561.711 753.079 753.955 ADF -47.906 -48.143 -49.134 -49.153 -49.253 -42.294 -52.505 -47.596 PP -48.098 -48.139 -49.122 -49.159 -49.647 -44.271 -52.366 -47.605 Observations 2291 2291 2291 2291 2291 2291 2291 2291 Note: Table 1 reports the summary statistics of the variables under consideration. The level data is converted into returns by taking first logged difference. Similarly, log differences are also estimated for OVX, for determining the changes in OVX. The DS, SS and RS are computed using the SVAR framework suggested by Ready (2018). The Jarque-Bera test confirm that the time series under consideration depart from normal distribution assumptions. The ADF and PP test results confirm that the stationary process is followed by the variables and thus we may proceed for a regression-based analysis.

3. Estimation Strategy 3.1. Structural oil shock decomposition Ready (2018)⁠ proposes a new technique to decompose the changes in oil price into risk shock (RS), supply shock (SS), and demand shock (DS), which bases upon the tradable assets prices. The DS is calculated as the contemporaneous regression residuals of global oil producing firm index returns regressed upon the innovations of the VIX. The SS is approximated as the residue of contemporaneous changes in oil price that is orthogonal upon the demand shocks and innovations in VIX (risk shock). In this process, the DS and SS are estimated such that the entire variation in oil price changes is explained by them. We follow Ready’s (2018)⁠ shock decomposition procedure in this study. It is assumed that the three shocks are orthogonal to each other, which may be expressed as: [

]

[

]

[

]

(1)

where the oil price change is denoted as , is the global oil producing firm index returns and signifies innovation in the VIX. The recognized shocks are mapped to A as observable variables. (2)

In order to impose orthogonality, the following condition is fulfilled:

5

∑(

)

[

]

(3)

The covariance matrix of the observable term is denoted by ∑ The volatilities of the identified shocks are represented as , and . This is the normalization procedure of the standard orthogonalization in order to define the structural shocks in the framework of structural vector autoregression. Additionally, the shocks are constrained to represent the total oil price changes. 3.2. Dummy variable GARCH model This study utilizes the dummy variable GARCH (1,1) model (Baur and Lucey 2010; Wu et al., 2019). Using the maximum likelihood method, the following model is estimated: (

)

(

)

(

)

(4) (5)

where the dependent variable denote the returns for Bitcoin, gold, commodity and Dollar. represent changes in OVX and the structural crude oil shocks. The dummy variable of 90% quantile is indicated by ( ), which essentially indicates the situation where the OVX or oil shocks are greater than the 90% quantile. The dummy variable assumes the value of 1 when OVX and oil shocks are above 90% quantile and 0 otherwise. Similarly, ( ) and ( ) takes the dummies when OVX and oil shocks are higher than 95% and 99% quantiles respectively. We follow the hedge and safe-haven definition propounded by Baur and Lucey (2010). It must be noted that Baur and Lucey (2010) model the relationship between gold (dependent variable) and stocks (independent variable). Thus, if the contemporaneous regression coefficient is (i.e. ) zero or negative, then gold can be regarded as a hedge against the stocks. Further, they define lower quantiles of the stocks (5%, 2.5% and 1% quantile) to assess the safe-haven feature of gold against stocks. The lower quantiles signify a bearish stock market state, thereby, gold is a safe-haven if it is negatively associated with stocks during extreme downside stock market events. Accordingly, investors may switch the investments from stocks to gold to optimize the portfolio risk. In our case, we must consider that the rise in oil price or volatility is economically undesirable, any asset whose value is positively related to such undesirable events could be used as a hedging tool. Thus, Bitcoin can be regarded as a hedge if the contemporaneous relationship with oil shocks and OVX is positive or at least zero. Moreover, the economic condition is adverse (and somewhat favourable) if the structural oil shocks and OVX are high (low). Hence, contrary to Baur and Lucey (2010), we test the safe-haven hypothesis considering extreme upper tail quantiles (90%, 95% and 99% quantile). The underlying reason for considering higher quantiles is to test the resilience of Bitcoin against extreme events of oil shocks and volatility. Our approach is similar to Wu et al., (2019), likewise we define the instruments (Bitcoin, gold, commodity and Dollar) as a hedge if , and not a hedge otherwise. Similarly, the instruments would serve as safe-haven at 90%, 95% and 99% quantile if ∑ >0, where [k= 1, 2, 3]. 3.3. QR model with dummy variables Let us denote the returns on the instruments as L and changes in the OVX and oil shocks as M. Then, with the cumulative distribution function () ( ), considering L as a real random variable, the th conditional quantile of L given M=m can be written as: |

where

( )

( )

|

( )

{

|

()

is the coefficient for m. So, 6

}

( )

(6)

̂( ()

(

(

∑

)

(

)

(7)

)) I(.) denotes the indicator function. Thus, we specify the following

quantile regression (QR) model to examine whether the concerned instruments act as a safehaven or hedge against OVX and oil shocks at different quantiles: |

( )

( )

( )

(

)

( )

(

)

( )

(

)

(8)

where the definition of the dependent and independent variables is similar to Equation (4). 4. Results and Discussion 4.1. GARCH model results Table 2 exhibits the results for the GARCH model specification that compare the hedging capacities of Bitcoin, gold, commodity and the US Dollar at average conditions. The Panel A reports the results of Bitcoin and we observe that against OVX the >0, however it is statistically insignificant suggesting Bitcoin to be a weak hedge against OVX. We observe that Bitcoin is a safe-haven for OVX at the extreme of 99% quantile i.e. ∑ >0 and ∑ statistically significant at 1% level, a weak safe-haven at 90% quantile ( >0, statistically insignificant) and not a safe-haven at 95% quantile ( ∑ <0, statistically significant). Against DS, Bitcoin is a weak hedge ( >0, statistically insignificant) and a weak safe-haven at quantile 90% and 99% (∑ >0, [k= 1, 3], statistically insignificant). However, it is not a safe-haven for DS at 95% quantile (∑ <0, statistically insignificant). In the case of SS and RS, Bitcoin is a weak hedge ( >0, statistically insignificant). Additionally, a weak safe-haven at quantile 95% for SS and at quantile 95% and 99% for RS. Thus, our results are partially consistent with Selmi et al., (2018) as the hedging and safehaven potential of Bitcoin appear weak in most of the cases. Nonetheless, we further argue that the decoupling hypothesis of Bitcoin (i.e. Bitcoin is not integrated with the real economy) appear unconvincing. For gold, results stated in Panel B imply that gold is not a hedge against OVX ( <0, statistically significant). Nevertheless, gold performs as a strong safe-haven (∑ >0, [k= 1, 2, 3], statistically significant at quantile 95% and 99%), consistent with the findings of Baur and Lucey (2010). Regarding the oil shocks we find that gold is weakly exposed to DS, however, acts as a strong hedge in respect to SS and RS. Besides, gold also acts as a safehaven, except at 90% quantile of DS and 99% quantile of RS. Our findings are strongly consistent with Uddin et al., (2018) in respect of SS, however, inconsistent in the case of RS. We believe that our results are justifiable since it is well-established that rising oil prices are the precursors of inflationary economic conditions (Kang et al., 2017) and gold is often a hedge against inflation (Iqbal, 2017). Panel C report the case of commodity and we find that commodity is neither a safehaven nor a hedge against OVX except for 95% quantile. Yahya et al., (2019) argue that the correlations between oil and agricultural commodities have become more stronger onset 2006. Consequently, the higher correlations limit the hedging capabilities of commodities against the oil price movements. Similar results are also reported in respect of different commodity sectors (energy, petrochemicals, oils and fats, softs and non-ferrous metals) by Jiang et al., (2019). Thus, commodity also appears a weak safe-haven for the structural crude oil shocks with few exceptions. Nonetheless, it is a strong hedge for RS and SS, which is inconsistent with Yahya et al., (2019) and Jiang et al., (2019). The underlying reason could be the fact that as we mention RS stand for the innovations in the VIX, which essentially represent the stock market risk. Hammoudeh et al., (2014) assert that instrument such as commodity futures is capable of hedging the downside stock market risks. The SS could be positively associated since the increasing oil prices could drive up the prices of the 7

commodities by virtue of rising inflation. Thus, oil and commodity prices may positively comove simultaneously. Finally, Panel D shows that the US Dollar is a hedge towards OVX, however a weak safe-haven at 90% and 95% quantile and a strong exposure to risk at 99% quantile. Though the hedging potential of the US Dollar against the OVX is surprising, however, there could be a plausible economic explanation. The OVX is a forward-looking indicator of crude oil volatility. Since, the crude oil is denominated in terms of the US Dollar, the oil-importing countries may exhibit a precautionary tendency to hold the US Dollar. Thus, the value of the US Dollar may rise corresponding to OVX. Regarding the oil shocks, the US Dollar is a weak hedge against DS, weakly risk-exposed to SS and strongly-exposed to RS. Furthermore, the US Dollar fail to act as a safe-haven in most of the cases with limited exceptions. This result is consistent with Lizardo and Mollick (2010). The authors argue that substantial quantum of oil procurements by the US leads to excess flow of the US Dollars to the foreign exchange markets. Hence, abundance of the US Dollar in foreign exchanges impose a downward pressure in the US Dollar value. Thus, the US Dollar may fail to serve as a hedge or safehaven against the oil shocks. Overall, we find that in the case of OVX, the US Dollar is a strong hedge ( >0, statistically significant), Bitcoin is a weak hedge ( >0, statistically insignificant), gold and commodity is strongly exposed to risk ( <0, statistically significant). Regarding, the safehaven property at 90% quantile Dollar outperforms all other asset classes. At 95% quantile level gold emerges as a better safe-haven. Lastly, at 99% quantile both gold and Bitcoin serves as a safe-haven asset. Table 2. GARCH (1,1) model estimation results Hedge

Parameters Panel A: Bitcoin OVX DS SS RS Panel B: Gold OVX DS SS RS Panel C: Commodity OVX DS SS RS

Safe-Haven ∑

∑

Variance Equation ∑

0.247*** (0.090) 0.220** (0.094) 0.260*** (0.093) 0.241*** (0.092)

0.0210 (0.018) 1.040 (0.723) 0.0276 (0.073) 0.0297 (0.079)

0.461 (0.292) 0.491 (0.329) -0.097 (0.303) -0.079 (0.460)

-1.583*** (0.488) -1.113 (0.573) 0.125 (0.651) 0.076 (0.543)

2.625*** (0.631) 0.659 (0.895) -1.177 (1.311) 0.897 (0.942)

0.422*** (0.032) 0.435*** (0.031) 0.458*** (0.030) 0.468*** (0.031)

0.128*** (0.006) 0.126*** (0.005) 0.125*** (0.005) 0.126*** (0.005)

0.875*** (0.003) 0.876*** (0.003) 0.876*** (0.003) 0.875*** (0.004)

-0.006 (0.019) 0.003 (0.019) 0.001 (0.019) 0.007 (0.019)

-0.018*** (0.005) -0.024 (0.149) 0.039*** (0.015) 0.100*** (0.016)

0.038 (0.069) -0.032 (0.089) 0.039 (0.087) 0.037 (0.086)

0.179** (0.090) 0.221* (0.122) 0.039 (0.122) 0.022 (0.115)

0.278* (0.142) 0.146 (0.293) 0.252 (0.207) -0.352 (0.222)

0.008*** (0.002) 0.008*** (0.002) 0.008*** (0.002) 0.009*** (0.002)

0.039*** (0.003) 0.038*** (0.002) 0.039*** (0.002) 0.041*** (0.002)

0.953*** (0.003) 0.954*** (0.003) 0.953*** (0.003) 0.950*** (0.003)

-0.016 (0.017) -0.021 (0.017) -0.028* (0.015) -0.023

-0.048*** (0.004) 0.205 (0.135) 0.230*** (0.012) 0.453***

-0.118* (0.062) -0.093 (0.086) 0.048 (0.065) -0.093

0.101 (0.078) 0.099 (0.121) 0.006 (0.093) 0.127

-0.214** (0.104) 0.352 (0.251) 0.083 (0.145) -0.078

0.014*** (0.004) 0.006*** (0.001) 0.004*** (0.001) 0.011***

0.056*** (0.007) 0.047*** (0.006) 0.036*** (0.005) 0.053***

0.922*** (0.011) 0.944*** (0.007) 0.957*** (0.006) 0.921***

8

Panel D: US Dollar OVX DS SS RS

(0.014)

(0.013)

(0.068)

(0.087)

(0.176)

(0.003)

(0.008)

(0.012)

-0.001 (0.008) 0.008 (0.008) 0.006 (0.008) 0.009 (0.008)

0.010*** (0.003) 0.056 (0.063) -0.001 (0.006) -0.140*** (0.007)

0.057 (0.036) -0.011 (0.033) 0.027 (0.038) 0.003 (0.031)

0.044 (0.045) -0.014 (0.047) -0.052 (0.050) -0.041 (0.045)

-0.171** (0.068) -0.072 (0.081) -0.092 (0.062) -0.021 (0.058)

0.001** (0.001) 0.001** (0.002) 0.002** (0.003) 0.006** (0.003)

0.033*** (0.005) 0.036*** (0.005) 0.036*** (0.006) 0.034*** (0.006)

0.963*** (0.006) 0.961*** (0.006) 0.960*** (0.006) 0.962*** (0.006)

Note: Table 2 reports the estimation results of the GARCH (1,1) model specified in Eq. (4) as: ( ) ( ) ( ) , . The equations are estimated for each of the dependent variables (Bitcoin, Gold, Commodity and US Dollar) separately. The independent variables are OVX and structural crude oil shocks (DS, SS and RS). *, ** and *** denote statistical significance at 1%, 5% and 10% level, respectively. The standard errors are indicated in parentheses.

4.2. Quantile regression model results Table 3 exhibits the results for the QR model with dummy variables for Bitcoin. We report the values of seven regression quantiles. Quantiles 5, 10 and 25 designate bearish market states and quantile 50 represent normal market state and quantiles 75, 90 and 95 are representatives of bullish market states. In the case of OVX, Bitcoin is a weak hedge at the bearish market states and a weakly exposed to risk in the normal and bullish market states. Regarding the safe-haven condition, Bitcoin is a weak safe-haven at bullish market states in the case of (∑ >0, [k= 1, 2]), whereas somewhat exposed to risk at ∑ . Further, Bitcoin emerges to be a better hedge and safe-haven with respect to DS as compared to SS. In the case of RS, Bitcoin serves as hedge as well as safe-haven with exception at (∑ >0, [k= 1, 3]). As we mention above that the RS is essentially the innovations in the VIX, thus our findings are consistent with Bouri et al. (2017a). We also plot the hedge and safe-haven coefficient for Bitcoin in Figure 1.4 Table 4 reports the results for gold and we observe that gold cannot be used as a hedge against OVX. Nonetheless, it has some safe-haven utility mainly in the upper tail of the distribution. Against DS, gold does exhibit some hedging capacity in lower quantiles of 5 and 10 and middle quantile of 50. Risk exposure is evident in rest of the cases. Gold acts as a safe-haven mainly at the bullish market state with exception to ∑ . SS on the other hand, could be hedged by gold and it also serves certain safe-haven benefits particularly in the upper quantiles i.e. bullish state. Further, gold may also be used as a hedge against RS, since the results depict a strong hedging property in the lower to middle quantiles of 5 to 50. Additionally, it may also be used as a safe haven during the bullish market state. Strong exposure to risk may be observed in the bearish market state. Table 5 presents the results for commodity and we find that it is strongly exposed to risk and hence fails to hedge for OVX. Besides it also offers limited safe-haven opportunities in the upper quantiles of ∑ and lower to middle quantiles of ∑ . In addition, commodities can strongly hedge SS and RS. Whereas, the DS can be hedged only up to the middle quantiles. Investments in commodity could yield safe-haven benefit mostly when the market is bullish. Nonetheless, we observe risk exposure mainly in the lower quantiles i.e. at the bearish market state. Table 6 shows the results for Dollar and we note that it can strongly hedge OVX and it also offers some safe-haven benefits against OVX. Besides, it can also hedge DS up to the higher quantiles and could provide marginal safe-haven benefit at the upper quantiles. 4

We do not plot the coefficients for gold, commodity and the US Dollar for parsimony. 9

Furthermore, it is weak hedge in respect of SS, however, it exhibits some safe-haven properties in the extreme market conditions. Lastly, we find that Dollar is strongly riskexposed to RS and it is not a suitable instrument to hedge in this case. However, in the bullish market state it could act as a safe-haven asset. Overall, we find that Dollar is a better hedge against OVX than Bitcoin. Nevertheless, Bitcoin outperforms gold and commodity in hedging OVX. However, gold could serve as a better safe-haven against OVX as compared to Bitcoin. Bitcoin on the other hand, could serve as a better hedge to DS in comparison to Dollar. The hedging capacity of commodity is better in the case of SS and RS in comparison to other asset classes. In respect of structural crude oil shocks Dollar appears to be a better safe-haven than Bitcoin. Table 3. Quantile regression model estimation results for Bitcoin q05

q10

q25

q50

q75

q90

q95

∑

0.138 (0.124) -0.630

0.040 (0.097) -1.748

0.010 (0.031) -0.134

-0.013 (0.013) -0.128

-0.030 (0.029) -0.093

-0.078 (0.094) 1.478

-0.091 (0.148) 1.787

∑

(1.956) -8.975

(1.464) 2.840

(0.674) -0.404

(0.322) 0.133

(0.738) -0.120

(1.392) 0.175

(3.182) 1.346

∑

(5.907) -4.537

(4.750) -7.006

(0.804) -4.977***

(0.312) -0.963

(0.778) 2.633

(2.175) -1.112

(3.316) -4.516

(7.548) -8.668*** (0.547)

(6.395) -4.970*** (0.464)

(1.863) -1.289*** (0.107)

(1.503) 0.234*** (0.071)

(2.579) 2.441*** (0.171)

(4.801) 6.350*** (0.394)

(4.62) 10.210*** (0.466)

q05

q10

q25

q50

q75

q90

q95

11.240** (5.362) -2.619 (3.330) 1.467 (3.444) -4.563 (3.344) -8.839*** (0.257)

0.740 (3.911) 1.752 (2.551) 0.080 (2.178) -4.173 (3.475) -5.193*** (0.304)

1.898** (0.827) 0.366 (0.514) -1.107*** (0.384) 1.150 (2.411) -1.331*** (0.128)

0.421 (0.363) 0.0630 (0.236) -0.630* (0.345) 1.327 (0.979) 0.232*** (0.071)

-1.582 (1.283) 0.117 (0.562) -0.886 (0.751) 2.660** (1.059) 2.418*** (0.146)

0.297 (2.942) -1.253 (1.569) -1.680 (1.298) 0.699 (1.496) 6.880*** (0.352)

-8.870 (6.747) -0.780 (2.300) 0.846 (3.006) -0.917 (3.248) 10.360*** (0.746)

q05

q10

q25

q50

q75

q90

q95

∑

-0.392 (0.340) -2.570

0.010 (0.193) -3.090

0.008 (0.077) -0.439

0.074 (0.047) -0.176

0.191 (0.147) -0.433

-0.055 (0.248) -1.166

-0.223 (0.437) 1.037

∑

(2.735) 3.896

(2.697) 2.651

(0.510) 0.325

(0.438) -0.044

(0.687) -0.968

(2.100) -1.542

(3.414) -2.626

∑

(6.673) 5.605

(3.259) 2.087

(0.756) 0.007

(0.369) -0.425

(0.643) -1.018

(2.818) -1.884

(3.513) -0.931

(12.030) -8.860*** (0.481)

(8.080) -5.127*** (0.276)

(1.334) -1.325*** (0.073)

(0.403) 0.236*** (0.086)

(0.773) 2.598*** (0.163)

(3.464) 6.747*** (0.367)

(3.254) 10.580*** (0.813)

q05

q10

q25

q50

q75

q90

q95

0.631 (0.562)

0.532 (0.448)

0.240* (0.144)

0.118** (0.049)

0.147 (0.207)

-0.342 (0.275)

0.283 (0.593)

Panel A: OVX

Panel B: DS

∑

∑

∑

Panel C: SS

Panel D: RS

10

∑

-1.622

-0.943

-0.301

-0.413***

-0.409

-0.898

1.003

∑

(2.423) 1.289

(1.318) 0.960

(0.278) 0.322

(0.109) 0.469*

(0.988) 0.429

(1.554) 3.912

(4.235) 2.500

∑

(3.376) 1.191

(2.044) -0.784

(0.572) -0.943

(0.253) -0.302

(1.225) -1.149

(2.716) 1.156

(4.555) -3.317

(2.931) -8.711*** (0.344)

(2.269) -5.051*** (0.339)

(1.574) -1.366*** (0.106)

(0.727) 0.241*** (0.041)

(2.356) 2.393*** (0.148)

(8.290) 6.465*** (0.223) as: | ( )

(7.014) 10.410*** (0.646)

Note: Table 3 reports the estimation results of the QR model specified in Eq. (8) ( ) ) ) ) . The dependent variable is Bitcoin. The independent ( ) ( ( ) ( ( ) ( variables are OVX and structural crude oil shocks (DS, SS and RS). *, ** and *** denote statistical significance at 1%, 5% and 10% level, respectively. The standard errors are indicated in parentheses.

Table 4. Quantile regression model estimation results for gold q05

q10

q25

q50

q75

q90

q95

∑

-0.029 (0.023) -0.813**

-0.035*** (0.013) -0.175

-0.019** (0.008) -0.009

-0.014*** (0.005) 0.094

-0.013 (0.009) 0.232*

-0.008 (0.014) 0.256

-0.008 (0.014) 0.143

∑

(0.376) 0.718

(0.384) 0.177

(0.112) -0.011

(0.088) 0.211

(0.138) 0.107

(0.188) -0.182

(0.443) -0.267

∑

(0.539) -2.754

(0.452) -0.807

(0.293) 0.207

(0.182) 0.318

(0.182) 0.534

(0.166) 0.944

(0.650) 2.175**

(3.019) -1.440*** (0.074)

(1.514) -1.015*** (0.029)

(0.862) -0.456*** (0.019)

(0.344) -0.011 (0.019)

(0.583) 0.464*** (0.025)

(1.104) 1.026*** (0.045)

(1.078) 1.465*** (0.053)

Panel A: OVX

q05

q10

q25

q50

q75

q90

q95

∑

1.236* (0.714) -0.122

0.349 (0.483) 0.018

-0.213 (0.249) -0.004

0.047 (0.182) -0.111

-0.360 (0.256) -0.046

-0.786** (0.336) -0.009

-0.120 (0.822) -0.001

∑

(0.251) -0.375

(0.243) -0.155

(0.135) 0.130

(0.117) 0.189

(0.137) 0.372**

(0.205) 0.744**

(0.202) 0.118

∑

(0.296) 0.403

(0.231) 0.681

(0.125) 0.364

(0.188) 0.321

(0.178) 0.336

(0.353) -0.180

(0.710) 0.325

(0.993) -1.490*** (0.067)

(1.029) -1.046*** (0.040)

(0.259) -0.485*** (0.025)

(0.249) 0.026 (0.024)

(0.264) 0.495*** (0.024)

(0.756) 0.990*** (0.050)

(1.132) 1.450*** (0.069)

q05

q10

q25

q50

q75

q90

q95

∑

0.160*** (0.036) -0.298

0.138*** (0.031) -0.242

0.049** (0.021) -0.060

0.050** (0.024) 0.126

0.018 (0.020) 0.190

0.0121 (0.030) 0.225

0.025 (0.028) 0.189

∑

(0.311) 0.087

(0.230) -0.227

(0.102) -0.023

(0.152) -0.075

(0.122) -0.034

(0.246) 0.227

(0.229) 0.113

∑

(0.515) -0.330

(0.292) -0.363

(0.099) 0.123

(0.129) 0.157

(0.105) 0.776

(0.275) 0.522

(0.479) 1.006

(0.459) -1.447*** (0.063)

(0.431) -0.975*** (0.035)

(0.235) -0.461*** (0.031)

(0.376) 0.015 (0.016)

(0.577) 0.475*** (0.019)

(0.634) 1.023*** (0.045)

(0.704) 1.418*** (0.035)

q05

q10

q25

q50

q75

q90

q95

∑

0.403*** (0.105) -0.670***

0.303*** (0.056) -0.630***

0.166*** (0.039) -0.043

0.116*** (0.037) -0.010

0.020 (0.035) 0.156*

-0.018 (0.068) 0.356*

-0.039 (0.082) 0.347

∑

(0.240) -0.479**

(0.163) -0.152

(0.147) -0.254

(0.069) 0.004

(0.084) 0.279**

(0.207) 0.081

(0.265) -0.004

Panel B: DS

Panel C: SS

Panel D: RS

11

∑

(0.238) -2.055**

(0.310) -1.136

(0.258) -0.452

(0.152) -0.356

(0.132) 0.279

(0.968) -1.376*** (0.054)

(1.255) -0.935*** (0.043)

(0.405) -0.460*** (0.025)

(0.503) 0.005 (0.017)

(0.481) 0.467*** (0.025)

(0.209) 0.415 (0.519) 1.003*** (0.051) as: | ( )

(0.341) 0.693 (0.485) 1.433*** (0.064)

Note: Table 4 reports the estimation results of the QR model specified in Eq. (8) ( ) ) ) ) . The dependent variable is Gold. The independent variables ( ) ( ( ) ( ( ) ( are OVX and structural crude oil shocks (DS, SS and RS). *, ** and *** denote statistical significance at 1%, 5% and 10% level, respectively. The standard errors are indicated in parentheses.

Table 5. Quantile regression model estimation results for commodity q05

q10

q25

q50

q75

q90

q95

∑

-0.078*** (0.009) -0.862**

-0.069*** (0.006) -0.445*

-0.063*** (0.007) -0.282*

-0.052*** (0.006) -0.224*

-0.045*** (0.006) -0.063

-0.048*** (0.007) 0.008

-0.034** (0.014) 0.520*

∑

(0.349) 0.794**

(0.253) 0.235

(0.151) 0.171

(0.119) 0.149

(0.110) -0.102

(0.351) 0.353

(0.311) -0.331

∑

(0.327) -1.322

(0.274) -0.105

(0.181) -0.126

(0.149) -0.203

(0.137) 0.056

(0.467) -0.106

(0.337) -0.324

(0.847) -1.144*** (0.047)

(0.880) -0.892*** (0.033)

(0.395) -0.482*** (0.017)

(0.281) -0.034* (0.018)

(0.317) 0.438*** (0.010)

(0.689) 0.908*** (0.035)

(0.606) 1.270*** (0.050)

Panel A: OVX

q05

q10

q25

q50

q75

q90

q95

∑

0.411 (0.389) -0.074

0.099 (0.352) -0.081

0.095 (0.203) -0.194*

0.002 (0.152) 0.0640

-0.030 (0.219) 0.102

-0.231 (0.281) 0.068

-0.240 (0.338) -0.062

∑

(0.087) -0.317

(0.119) -0.237

(0.102) 0.009

(0.187) -0.068

(0.099) 0.133

(0.158) 0.677**

(0.283) 0.804***

∑

(0.396) 0.430

(0.291) 0.663

(0.150) 0.450**

(0.240) 0.318

(0.208) 0.360

(0.273) 0.069

(0.282) -0.013

(0.506) -1.317*** (0.039)

(0.423) -0.982*** (0.029)

(0.229) -0.498*** (0.023)

(0.330) -0.009 (0.014)

(0.353) 0.448*** (0.030)

(0.303) 0.933*** (0.048)

(0.241) 1.278*** (0.054)

Panel B: DS

q05

q10

q25

q50

q75

q90

q95

∑

0.267*** (0.0390) -0.162

0.244*** (0.0237) -0.142

0.263*** (0.0223) -0.0502

0.228*** (0.0161) -0.0201

0.218*** (0.0160) -0.00337

0.208*** (0.0277) 0.269

0.217*** (0.0324) 0.125

∑

(0.275) -0.296

(0.119) -0.230

(0.105) 0.0452

(0.0737) 0.0300

(0.125) 0.0976

(0.169) -0.0933

(0.183) -0.0160

∑

(0.304) -0.768

(0.319) -0.695

(0.0954) -0.0778

(0.117) 0.0641

(0.182) 0.342

(0.214) 0.141

(0.205) 0.766

(0.864) -1.167*** (0.0392)

(0.776) -0.877*** (0.0272)

(0.533) -0.452*** (0.0189)

(0.218) -0.0161 (0.0199)

(0.355) 0.417*** (0.0197)

(0.556) 0.844*** (0.0441)

(0.718) 1.157*** (0.0559)

Panel C: SS

q05

q10

q25

q50

q75

q90

q95

∑

0.613*** (0.025) -0.575**

0.533*** (0.026) -0.216

0.464*** (0.015) -0.119

0.430*** (0.017) -0.033

0.394*** (0.023) 0.105

0.423*** (0.028) -0.011

0.424*** (0.051) -0.022

∑

(0.269) -0.159

(0.150) -0.209

(0.087) 0.0314

(0.079) 0.092

(0.126) 0.330*

(0.084) 0.598**

(0.124) 0.590***

∑

(0.358) 0.009

(0.178) -0.080

(0.120) -0.246

(0.154) -0.128

(0.189) -0.289

(0.242) -0.570

(0.151) -0.497

(0.287)

(0.218)

(0.205)

(0.190)

(0.188)

(0.393)

(0.683)

Panel D: RS

12

-0.987*** (0.028)

-0.783*** (0.025)

-0.425*** (0.016)

-0.027* (0.016)

0.351*** (0.017)

0.759*** (0.020) as: | ( )

1.016*** (0.058)

Note: Table 5 reports the estimation results of the QR model specified in Eq. (8) ( ) ) ) ) . The dependent variable is commodity. The independent ( ) ( ( ) ( ( ) ( variables are OVX and structural crude oil shocks (DS, SS and RS). *, ** and *** denote statistical significance at 1%, 5% and 10% level, respectively. The standard errors are indicated in parentheses.

Table 6. Quantile regression model estimation results for the US Dollar q05

q10

q25

q50

q75

q90

q95

∑

0.0135 (0.009) 0.124*

0.012** (0.005) 0.051

0.014*** (0.004) 0.023

0.014*** (0.003) 0.051

0.010*** (0.004) 0.042

0.009** (0.004) 0.089

0.014** (0.006) 0.0661

∑

(0.073) -0.038

(0.061) -0.019

(0.063) -0.008

(0.035) 0.007

(0.035) 0.096

(0.072) 0.283***

(0.159) 0.218

∑

(0.088) -0.092

(0.085) -0.074

(0.096) -0.192

(0.055) -0.042

(0.137) -0.103

(0.107) -0.011

(0.218) 0.053

(0.220) -0.621*** (0.016)

(0.163) -0.450*** (0.009)

(0.172) -0.214*** (0.010)

(0.167) 0.006 (0.009)

(0.189) 0.211*** (0.010)

(0.285) 0.450*** (0.014)

(0.423) 0.605*** (0.016)

Panel A: OVX

q05

q10

q25

q50

q75

q90

q95

∑

0.081 (0.257) -0.086

0.068 (0.153) -0.107**

0.116 (0.111) -0.0617

0.140** (0.069) -0.009

0.101 (0.103) 0.055

-0.018 (0.162) 0.145**

-0.152 (0.188) 0.131**

∑

(0.110) 0.031

(0.047) -0.025

(0.067) -0.082

(0.039) -0.044

(0.068) -0.065

(0.067) -0.006

(0.062) 0.147

∑

(0.131) 0.085

(0.072) 0.041

(0.112) 0.003

(0.059) -0.065

(0.074) -0.113

(0.092) -0.085

(0.097) -0.209

(0.141) -0.603*** (0.023)

(0.097) -0.425*** (0.012)

(0.167) -0.210*** (0.009)

(0.071) 0.0113** (0.004)

(0.085) 0.225*** (0.009)

(0.218) 0.454*** (0.020)

(0.237) 0.620*** (0.018)

q05

q10

q25

q50

q75

q90

q95

∑

0.006 (0.023) -0.023

-0.002 (0.010) -0.041

0.007 (0.007) -0.042

0.004 (0.006) 0.012

0.000 (0.011) 0.072

-0.006 (0.011) 0.122

-0.029* (0.016) 0.183***

∑

(0.087) -0.106

(0.065) 0.003

(0.053) -0.067

(0.046) -0.030

(0.054) -0.075

(0.080) -0.111

(0.060) 0.247

∑

(0.148) -0.686**

(0.116) -0.691*

(0.089) 0.073

(0.071) 0.132

(0.081) 0.117

(0.332) 0.0830

(0.284) 0.258

(0.337) -0.604*** (0.029)

(0.399) -0.438*** (0.018)

(0.105) -0.210*** (0.009)

(0.177) 0.008 (0.006)

(0.163) 0.222*** (0.009)

(0.605) 0.460*** (0.016)

(0.663) 0.608*** (0.017)

q05

q10

q25

q50

q75

q90

q95

∑

-0.099*** (0.020) -0.186

-0.099*** (0.015) -0.188***

-0.112*** (0.012) -0.101

-0.125*** (0.010) -0.028

-0.160*** (0.011) 0.083***

-0.186*** (0.015) 0.099*

-0.188*** (0.018) 0.091

∑

(0.164) -0.035

(0.072) 0.025

(0.079) -0.044

(0.031) -0.032

(0.029) -0.016

(0.060) 0.175*

(0.160) 0.148

∑

(0.165) -0.414

(0.082) -0.487

(0.072) -0.050

(0.044) -0.029

(0.045) 0.291*

(0.091) 0.362

(0.163) 0.492*

(0.288) -0.528*** (0.016)

(0.363) -0.392*** (0.014)

(0.265) -0.191*** (0.008)

(0.154) 0.009 (0.008)

(0.155) 0.205*** (0.008)

Panel B: DS

Panel C: SS

Panel D: RS

Note: Table 6 reports the estimation results of the QR model specified in Eq. (8) 13

(0.254) 0.416*** (0.013) as: | ( )

(0.257) 0.554*** (0.016) ( )

) ) ) . The dependent variable is the US Dollar. The independent ( ) ( ( ) ( ( ) ( variables are OVX and structural crude oil shocks (DS, SS and RS). *, ** and *** denote statistical significance at 1%, 5% and 10% level, respectively. The standard errors are indicated in parentheses.

(a) Oil volatility index

(b) Demand shock

(c) Supply shock

(d) Risk shock

Figure 1. The hedge and safe-haven coefficients of Bitcoin against OVX, DS, SS and RS Note: The figure exhibits the estimates of OLS and coefficients of quantile regression. The predictor variables indicated on the vertical axis. The broken black and the dotted horizontal lines represent the coefficient of OLS at its 95% confidence intervals respectively. The shaded areas around the quantile regression coefficient plotline represent the confidence interval at 95%.

5. Conclusion We investigate whether Bitcoin acts as a safe-haven or hedge with respect to OVX and structural crude oil shocks. Additionally, we also compare the hedging performance of Bitcoin against other alternative assets. We report some interesting findings. Firstly, the US Dollar is a better alternative to hedge OVX than Bitcoin. Secondly, Bitcoin outperforms gold and commodity in hedging OVX. Thirdly, gold acts as a better safe-haven against OVX as compared to Bitcoin. Fourthly, Bitcoin is a better hedge to DS in comparison to the US Dollar. Fifthly, the hedging capacity of commodity is better in the case of SS and RS in comparison to other asset classes. Lastly, regarding the crude oil shocks the US Dollar appears to be a better safe-haven than Bitcoin. Moreover, the hedging and safe-haven property of different assets are market-state varying. Thus, we find mixed results and conclude that Bitcoin is not the superior asset over others to hedge oil-related uncertainties. We believe that our study offers certain policy related implications to the investors. Our results indicate that a single asset cannot hedge downside risks in different economic 14

situations and market states. Thus, an investor should carefully evaluate the hedging potential of different assets at different market states. Moreover, we must envisage the fact that though we advocate some hedging potentials of Bitcoin, nevertheless some inherent threats are also associated with its use. For instance, the price of Bitcoin is highly volatile, thus, the basic characteristic of a fiat currency such as the store of value may not be present in the case of Bitcoin. Bitcoin may lose its value entirely if any undesirable event take place (such as Cryptocurrency market crash). Another shortcoming of Bitcoin is related to its liquidity as compared to traditional assets such as gold. Additionally, higher information asymmetry exists in the case of Bitcoin than gold. Furthermore, the digital frauds and internet scandals are prevalent in the Bitcoin ecosystem (Selmi et al., 2018). Therefore, we believe that inclusion of Bitcoin should be carefully considered in the portfolio only when the expected benefits outweighs its potential risks. References Atems, B., Kapper, D., Lam, E., 2015. Do exchange rates respond asymmetrically to shocks in the crude oil market? Energy Econ. 49, 227–238. Aysan, A.F., Demir, E., Gozgor, G., Lau, C.K.M., 2019. Effects of the geopolitical risks on Bitcoin returns and volatility. Res. Int. Bus. Financ. 47, 511–518. Baur, D.G., Lucey, B.M., 2010. Is gold a hedge or a safe haven? An analysis of stocks, bonds and gold. Financ. Rev. 45, 217–229. Bouri, E., Gupta, R., Lau, C.K.M., Roubaud, D., Wang, S., 2018. Bitcoin and global financial stress: A copulabased approach to dependence and causality in the quantiles. Q. Rev. Econ. Financ. 69, 297–307. Bouri, E., Gupta, R., Tiwari, A.K., Roubaud, D., 2017a. Does Bitcoin hedge global uncertainty? Evidence from wavelet-based quantile-in-quantile regressions. Financ. Res. Lett. 1–9. https://doi.org/10.1016/j.frl.2017.02.009 Bouri, E., Jalkh, N., Molnár, P., Roubaud, D., 2017b. Bitcoin for energy commodities before and after the December 2013 crash: diversifier, hedge or safe haven? Appl. Econ. 49, 5063–5073. Bouri, E., Lucey, B., Roubaud, D., 2019. Cryptocurrencies and the downside risk in equity investments. Financ. Res. Lett. Bouri, E., Molnár, P., Azzi, G., Roubaud, D., Hagfors, L.I., 2017c. On the hedge and safe haven properties of Bitcoin: Is it really more than a diversifier? Financ. Res. Lett. 20, 192–198. Ciaian, P., Rajcaniova, M., Kancs, d’Artis, 2016. The economics of BitCoin price formation. Appl. Econ. 48, 1799–1815. Clements, A., Shield, C., Thiele, S., 2019. Which oil shocks really matter in equity markets? Energy Econ. 81, 134–141. Corbet, S., Lucey, B., Urquhart, A., Yarovaya, L., 2019. Cryptocurrencies as a financial asset: A systematic analysis. Int. Rev. Financ. Anal. 62, 182–199. Das, D., Dutta, A., 2019. Bitcoin’s energy consumption: Is it the Achilles heel to miner’s revenue? Econ. Lett. 108530. Demir, E., Gozgor, G., Lau, C.K.M., Vigne, S.A., 2018. Does economic policy uncertainty predict the Bitcoin returns? An empirical investigation. Financ. Res. Lett. 26, 145–149. Dogah, K.E., Premaratne, G., 2018. Sectoral exposure of financial markets to oil risk factors in BRICS countries. Energy Econ. 76, 228–256. Dorfleitner, G., Lung, C., 2018. Cryptocurrencies from the perspective of euro investors: a re-examination of diversification benefits and a new day-of-the-week effect. J. Asset Manag. 19, 472–494. Dutta, A., 2018. Impacts of oil volatility shocks on metal markets: A research note. Resour. Policy 55, 9–19. Dutta, P., Noor, M.H., Dutta, A., 2017. Impact of oil volatility shocks on global emerging market stock returns. Int. J. Manag. Financ. 13, 578–591. Gozgor, G., Tiwari, A.K., Demir, E., Akron, S., 2019. The Relationship between Bitcoin Returns and Trade Policy Uncertainty. Financ. Res. Lett. Hammoudeh, S., Nguyen, D.K., Reboredo, J.C., Wen, X., 2014. Dependence of stock and commodity futures markets in China: Implications for portfolio investment. Emerg. Mark. Rev. 21, 183–200. Iqbal, J., 2017. Does gold hedge stock market, inflation and exchange rate risks? An econometric investigation. Int. Rev. Econ. Financ. 48, 1–17. Jiang, Y., Jiang, C., Nie, H., Mo, B., 2019. The time-varying linkages between global oil market and China’s commodity sectors: Evidence from DCC-GJR-GARCH analyses. Energy 166, 577–586. 15

Kang, S.H., McIver, R.P., Hernandez, J.A., 2019. Co-movements between Bitcoin and Gold: A wavelet coherence analysis. Phys. A Stat. Mech. its Appl. 120888. Kang, W., Ratti, R.A., Vespignani, J.L., 2017. Oil price shocks and policy uncertainty: New evidence on the effects of US and non-US oil production. Energy Econ. 66, 536–546. Kang, W., Ratti, R.A., Yoon, K.H., 2015. Time-varying effect of oil market shocks on the stock market. J. Bank. Financ. 61, S150–S163. Kilian, L., 2009. Not all oil price shocks are alike: Disentangling demand and supply shocks in the crude oil market. Am. Econ. Rev. 99, 1053–1069. Klein, T., Thu, H.P., Walther, T., 2018. Bitcoin is not the New Gold–A comparison of volatility, correlation, and portfolio performance. Int. Rev. Financ. Anal. 59, 105–116. Lizardo, R.A., Mollick, A. V, 2010. Oil price fluctuations and US dollar exchange rates. Energy Econ. 32, 399– 408. Ready, R.C., 2018. Oil prices and the stock market. Rev. Financ. 22, 155–176. Selmi, R., Mensi, W., Hammoudeh, S., Bouoiyour, J., 2018. Is Bitcoin a hedge, a safe haven or a diversifier for oil price movements? A comparison with gold. Energy Econ. 74, 787–801. Shahzad, S.J.H., Bouri, E., Roubaud, D., Kristoufek, L., Lucey, B., 2019. Is Bitcoin a better safe-haven investment than gold and commodities? Int. Rev. Financ. Anal. 63, 322–330. Singh, V.K., Nishant, S., Kumar, P., 2018. Dynamic and directional network connectedness of crude oil and currencies: Evidence from implied volatility. Energy Econ. 76, 48–63. Smales, L.A., 2018. Bitcoin as a safe haven: Is it even worth considering? Financ. Res. Lett. Uddin, G.S., Rahman, M.L., Shahzad, S.J.H., Rehman, M.U., 2018. Supply and demand driven oil price changes and their non-linear impact on precious metal returns: A Markov regime switching approach. Energy Econ. 73, 108–121. van Wijk, D., 2013. What can be expected from the BitCoin (No. 345986), Erasmus Universiteit Rotterdam. Wang, G.-J., Xie, C., Wen, D., Zhao, L., 2018. When Bitcoin meets economic policy uncertainty (EPU): Measuring risk spillover effect from EPU to Bitcoin. Financ. Res. Lett. Wang, P., Zhang, W., Li, X., Shen, D., 2019. Is cryptocurrency a hedge or a safe haven for international indices? A comprehensive and dynamic perspective. Financ. Res. Lett. 31, 1–18. Wu, S., Tong, M., Yang, Z., Derbali, A., 2019. Does gold or Bitcoin hedge economic policy uncertainty? Financ. Res. Lett. 31, 171–178. Xiao, J., Zhou, M., Wen, Fengming, Wen, Fenghua, 2018. Asymmetric impacts of oil price uncertainty on Chinese stock returns under different market conditions: Evidence from oil volatility index. Energy Econ. 74, 777–786. Yahya, M., Oglend, A., Dahl, R.E., 2019. Temporal and spectral dependence between crude oil and agricultural commodities: A wavelet-based copula approach. Energy Econ. 80, 277–296.

16