The Quarterly Review of Economics and Finance 81 (2021) 370–375

Contents lists available at ScienceDirect

The Quarterly Review of Economics and Finance

journal homepage: www.elsevier.com/locate/qref

COVID-19 pandemic and the safe haven property of Bitcoin

Ibrahim D. Raheem
Independent Researcher, Nigeria

a r t i c l e i n f o a b s t r a c t

Article history: The commentaries in the literature point out that cryptocurrencies, particularly Bitcoin, provide safe
Received 18 June 2020 haven features to investors. The advent of COVID-19 pandemic offers a perfect opportunity to test this
Received in revised form 8 May 2021 hypothesis. This study tries to validate this claim by examining the safe haven prowess of Bitcoin against
Accepted 21 June 2021
measures of uncertainty (VIX, EPU, and Oil Shock). We further make a comparison between pre-and
Available online 24 June 2021
post-COVID-19 analyses. Results confirm that prior to COVID-19, Bitcoin was able to maintain its widely
acknowledged characteristics. However, the post COVID-19 announcement upturned the tides previously
Keywords:
identified.
Bitcoin
© 2021 Board of Trustees of the University of Illinois. Published by Elsevier Inc. All rights reserved.
Safe haven
COVID-19
Uncertainty

JEL classifications:
E60
G1
I10

1. Introduction tially, this index jumped from 160 points in January 2020 to 400
by the end of April1 . Similarly, the measure of risk averseness of
The safe haven property of cryptocurrencies, with more empha- investors, proxied by VIX, has also skyrocketed, reaching 83 as at
sis on Bitcoin, has been widely documented in the literature (see March 20202 .
Corbet, Lucey, Urquhart, & Yarovaya, 2019; Ramona, Christina, & Expectedly, the adverse consequences of the sudden increase in
Raluca, 2019; for literature survey). The literature concludes that the level of uncertainty and risk averseness caused by the emer-
Bitcoin could be used as a hedging or diversification tool during gence of COVID-19 have been felt in the global financial markets.
shocks and exogenous turbulent periods. Three major reasons have China and other Asian stock markets were the first causalities of
been adduced to this stance: (i) it is immune to government power COVID, before it spread to the rest of the world. Relatedly, virtu-
and control; (ii) it is uncorrelated with the dynamics of formal ally all the traditional and new financial assets have experienced
financial institutions and traditional assets; and (iii) the trading downturn in their performance (Shehzad, Xiaoxing, & Kazouz,
dynamics is purely conducted with the advent of technology and 2020). The commodity markets are not exempted from this pan-
innovation techniques (Bouri, Molnar, Azzi, Roubaud, & Hagfors, demic’s shenanigan. For instance, the recorded oil price in April
2017; Smales, 2019; and Isah & Raheem, 2019). The attendant 2020 was sub-zero (i.e. negative), just as other oil related stocks
consequences of COVID-19 on the world economy are unprece- (EOG Resources). The collapse in the oil price, in addition to oil
dented and likened to the Great Depression (Bouri, Demirer, Gupta, price wars has triggered the free fall in the global stock prices. His-
& Pierdzioch, 2020). There has been an excessive increase in the torical data shows that other commodities (with the exceptions of
volatility of, and jumps in, the financial markets, which are per- gold, natural gas, coffee and rice) have recorded negative returns,
ceived to be in response to either bad news related to COVID-19 since the advent of the pandemic (Salisu, Akanni, & Raheem, 2020).
or government’s policy responses (Baker, Bloom, Davis, Kost et al., A strand of the literature has shown that countries that are sensi-
2020; Baker, Bloom, Davis, & Terry, 2020). This has influenced tive to commodity prices movements tend to commove with the
the rapid increase in the news-based measure of economic policy US markets conditions, i.e. tranquil and turbulent periods, (Aloui,
uncertainty (EPU) index of Baker, Bloom, and Davis (2016). Essen- Aissa, & Nguyen, 2011). The continuous spread of the coronavirus

1
Values surpassing 100 imply above average uncertainty level.
2
E-mail address: i raheem@ymail.com Prior to the pandemic, this index has hovered between 12 to 15 points.

https://doi.org/10.1016/j.qref.2021.06.004
1062-9769/© 2021 Board of Trustees of the University of Illinois. Published by Elsevier Inc. All rights reserved.
I.D. Raheem The Quarterly Review of Economics and Finance 81 (2021) 370–375

has instilled fear and panic among investors and other market par- examine the relationship between Bitcoin and other conventional
ticipants. As such, they have started to resort to exploring avenues assets. Results showed there is a lagged relationship, especially
to diversify their portfolios. Guided by the literature, one of the when Bitcoin is in bear market condition. Using Cross-quantilogram
often sought-after approaches is the design of portfolios to include approach, Bouri, Shahzad, and Roubaud (2020) tested the safe
cryptocurrencies. haven and hedging hypothesis of cryptocurrencies and US equity
Since their emergence, cryptocurrencies have proven to be market.
a reliable provider of safe-haven and hedge features against Bouri, Shahzad, Roubaud, Kristoufek, and Lucey (2020) stud-
uncertainties and shocks. Moreover, studies have found that cryp- ied how new assets (Bitcoin, gold, and commodities) provide safe
tocurrencies gain value during periods of market turbulence. Since havens for selected stock market indices. For a more detailed
the Great Depression, the global economy has not witnessed the review, readers are referred to Corbet et al. (2019) and Ramona
magnitude of shock posed by COVID-193 . Similarly, no previously et al. (2019) for a literature survey.
identified epidemic has affected the global economy as much as Interestingly, the linkage between Bitcoin and COVID-19 is bur-
COVID-19 (Ma, Rogers, & Zhou, 2020). The hedging and/or safe- geoning. This is largely connected to the recency in the pandemic,
haven characteristics of cryptocurrencies have largely remained which was first reported in January, 2020. Kristoufes (2020) stud-
unexplored during extreme bear market conditions. As such, it is ies the quantile correlation between Bitcoin and S&P500 and VIX.
yet to be validated whether cryptocurrencies would retain such The author concludes that recent data do not support the claim
features during COVID-19 pandemic era. The novelty and severity of safe haven feature of Bitcoin. Conlon and McGee (2020) show
of this disease suggest that the best time to examine the safe haven that Bitcoin does not provide cover against turbulence in traditional
property of Bitcoin is now. markets. Using wavelet methods, Sharif, Aloui, and Yarovaya (2020)
The objective of this study is to examine the predictive prowess examine the connectedness between the spread of COVID-19, oil
of Bitcoin in providing safe haven to COVID-19 induced shocks occa- shock and economic policy uncertainty. Grobys (2020) presents
sioned by increase in uncertainty (VIX and EPU) and commodity results supporting a high correlation between Bitcoin and US stock,
(oil) price collapse. The motivation for the consideration of these with the former providing a poor hedging option.
variables is due to the interconnection between financial and com- An interesting strand of the literature has upheld the widely cel-
modity markets, which has been enhanced by the activities of the ebrated features of cryptocurrencies during the pandemic period.
financial investors. Hence, this serves as a new channel through For instance, Corbet, Larkin, and Lucey (2020) confirmed the diver-
which market participants’ sentiment spillover these two markets sification benefits of Bitcoin in the Chinese stock markets. Conlon,
(Bouri, Demirer et al., 2020). Corbet, and McGee (2020) examined the safe haven hypothesis
We offer four contributions to the literature. First, we built a pre- of Bitcoin, Ethereum and Tether from the perspective of interna-
dictive model of Bitcoin returns that accounts for the influence of tional equity index investors. Results confirm that both Bitcoin
shocks attributed to the COVID-19 pandemic. Second, we consider and Ethereum were unable to validate the hypothesis, while the
multiple shocks (VIX, EPU, and oil prices). This is in sharp contrast reverse is the case for Tether. Goodwell and Goutte (2020) exam-
to previous studies that focused on a particular shock. Third, we ine Bitcoin’s reaction to, and co-movement with, COVID-19 using
conduct a pre- and post-COVID-19 announcement analyses. The wavelet coherence approach. The study concludes that there is a
need for this comparison is partly attributed to the stance that the positive relationship between COVID-19 and Bitcoin, thus estab-
existing market condition has an important role to play on the effec- lishing a safe-haven hypothesis of Bitcoin. Mariana, Ekaputra, and
tiveness of the safe-haven property of financial assets (Iqbal, 2017). Husondo (2020) found that Bitcoin and Ethereum have a negative
Fourth, we conduct a more recent analysis, spanning 5 months post- correlation with stock returns during the pandemic. Naeem, Bouri,
COVID announcement. This study’s dataset is limited to the US for Peng, Shahzad, and Vo (2021) concluded that COVID-19 affected
two reasons: (i) there is a high spillover and contagion effects of the efficiency of major cryptocurrencies, with Bitcoin and Etherum
the US markets on the rest of the world (Syriopoulos, Makram, & being the worst hit. Results further show that Ethereum provides a
Boubaker, 2017); and (ii) the country is the epicentre of the pan- better safe-haven, as compared to Bitcoin. Dutta, Das, Hana, and Vo
demic as it accounts for at least one-fifth of the indicators of the (2020) using time-varying correlation, through DCC-GARCH model,
disease (number of death, hospitalisation rate, total infection rate found that Bitcoin is a diversifier, and not a safe-haven asset.
and the number of new cases).
Highlighting the results, we show that Bitcoin is not able to
effectively provide safe haven feature during the pandemic period. 3. Methodology and data
Following this introductory section, the rest of the paper is orga-
nized as follows: a short literature review is presented in section 2. The safe haven hypothesis makes two fundamental assump-
Section 3 presents data and methodology. Section 4 discusses the tions: (i) investments in financial assets (Bitcoin, in this case)
empirical results. Section 5 gives the concluding remarks. should, at least, preserve its prices during bear market or turbu-
lent episodes; (ii) there is a positive correlation between measures
2. Succinct literature review of uncertainty and returns on financial assets. Financial models are
highly susceptible to various forms of endogeneity issues (caused
Studies examining the safe haven property of cryptocurrency by reverse causality, omitted variables bias); heteroscedasticity and
is huge and has been adequately reviewed in the literature. Dif- persistence (Salisu, Raheem, & Vo, 2021). Westerlund and Narayan
ferent methods(and models) have been specified(and analysed) to (2012 and 2015) showed that these problems could be rectified by
examine the safe haven hypothesis. For instance, Bouri, Gupta, Lauc, specifying the equation:
Roubaudd, and Wang (2019) examined the relationship between
Bitcoin and global financial stress using a copula-based approach
rt = ∝ + ϑUNC t−1 + t (1)
to dependence and causality in the quantiles. Ji, Bouri, Gupta,
and Roubaud (2019) used a directed acylic graph approach to
Where rt is the return of Bitcoin prices, UNC is the measures of
uncertainty. In this study, we used variants of uncertainty (EPU,
3
The potency of this pandemic is largely attributed to the lockdown policies VIX, and oil price shock) while t is the error term. The null hypoth-
instituted in order to tame the spread of the virus (Bouri, Shahzad et al., 2020). esis of no predictability in Eq. (1) is that ϑ = 0, which is estimated

371
I.D. Raheem The Quarterly Review of Economics and Finance 81 (2021) 370–375

Table 1
Descriptive Statistics.

Pre-COVID Post-COVID

BITC. VIX EPU WTI BITC VIX EPU WTI

Mean −0.003 15.123 100.34 0.075 −0.008 32.704 314.85 −0.0053

SD 0.038 2.781 48.657 1.385 0.047 17.831 207.60 0.1199
SK 0.882 0.940 1.397 1.768 −2.15 0.763 0.213 −0.8570
KT 6.417 3.272 6.086 12.67 20.86 2.904 1.646 11.8865

Auto K=2 1.234 0.868 3.265 10.08a 3.0873 200.65a 3.231 1.385
K=4 0.885 2.654 4.213 11.60a 13.124b 379.11a 7.865 5.498
K=6 0.820 3.269 6.265 11.61a 16.697b 531.18a 16.270 7.232

Hetero K=2 1.973 6.569a 17.582a 1.637 0.0619 134.567a 2.1222 5.8283a
K=4 1.193 4.665a 15.707a 0.8298 0.3171 85.0841a 1.2664 3.0300b
K=6 0.878 2.256b 10.585a 0.5373 0.3183 58.2011a 1.1448 2.181b

Source: Authors’ computation.

Note: SD, SK and KT are the associated statistics for standard deviation, skewness and kurtosis, respectively. Ljung-Box test Q-statistics and ARCH-LM test F-statistics are
used to test for autocorrelation and heteroscedasticity, respectively. K is the number of lag. a, and b are the level of statistical significance at 1% and 5%, in that order.

using ordinary least squares. To resolve the potential bias and the
adjusted forecasted error. The sample average f̂ t+k is expressed as:
suspected persistence effect, equation one is re-specified as thus:
MSE 1 − (MSE 2 − adj.). Each term is computed as:


5  2
rt = ∝ +
adj
ϑi UNC t−1 +  (UNC t − 0 UC t−1 ) + t (2) MSE 1 = P −1 CC t+k − ĈC 1t+k
i=1
 2
Thus,
adj
ϑi
is derived from
adj
= ϑ −  (1 − 0 ), where 0 is the
ϑi MSE 2 = P −1 CC t+k − ĈC 2t+k ; and
degree of persistence in UNC t , which Lewellen (2004) described as
bias adjusted OLS that corrects for persistence. Endogeneity effect  2
is corrected by the term  (UNC t − 0 UC t−1 ) resulting from the cor- Adj. = P −1 ĈC 1t+k − ĈC 2t+k
relation between UNCt (i.e. the predictor) and the error term. The
choice of 5 lags is to capture the days-of-the-week dynamics in Where P is the number of the predictors employed to com-
the estimation process4 . The problem posed by heteroscedasticity pute the averages. The forecast performances of both the restricted
could be resolved by pre-weighting all the data by 1/ˆ ε and then and unrestricted model must be ascertained. Hence, a model is
estimate the model using the OLS approach (Westerlund & Narayan,
specified where f̂t+k is estimated against a constant and the result-
2012, 2015).
ing t-statistic for a zero coefficient is used to draw inference. A
The baseline model is the historical average. We test whether
situation where the t-statistics > +1.286(+1.464), for a one-side,
the inclusion of various measures of uncertainty will enhance the
0.10test(0.05test) leads to the rejection of the null hypothesis.
predictive model, relative to the baseline model. The uncertainty
Understandably, a myriad of models have been explored by
augmented model is termed the unrestricted model, while the
existing studies to examine the safe haven property of Bitcoin
baseline model is referred to as the restricted model. In line with
(Bouri et al., 2017; Conlon & McGee, 2020; Smales, 2019). The com-
the extant literature, we use three methods of forecast evaluation:
parative advantage of our model over others is due to the ability of
Campbell and Thompson (2008, hereafter CT test), Clark and West
the former to adequately capture some inherent properties of the
(2007, hereafter CW test) and Theil’s U- Statistics. Theil U-statistic
predictor (endogeneity, persistence and conditional heteroscedas-
less than unity (i.e.1) suggests that the predictive accuracy of the
ticity). More recent studies, such as Salisu, Akanni et al. (2020);
unrestricted model outweighs that of the restricted model. CW
Salisu, Raheem, & Eigbiremolen, 2020) have used similar model. In
examines the forecast performance of these two nested models.
addition, our model is predictive in nature, which thus allows for
The beauty of CW is to examine the statistical significance of the
out-of-sample estimation. Whereas, previous models had mainly
difference between the two errors of the models (i.e. unrestricted
rely on in-sample analyses.
and restricted). The CW test is computed as follows:
We use daily dataset from 01-08-2019 to 30-05-2020. This
 2  2  2  sample period is dichotomized into pre-COVID (01/08/2019-
f̂ t+k = rt+k − r̂ 1t,t+k − rt+k − r̂ 2t,t+k − r̂ 1t,t+k − r̂ 2t,t+k 30/12/2019) and post-COVID (01/01/2020-31/05/2020) announce-
ments. As a result of data constraint (i.e. limited data availability),
(4) we use 75-25 data split for the in- and out-of-sample forecast eval-
  uations, respectively. The three out-of-sample forecast horizons
Where k is the forecast period; rt+k − r̂ 1t,t+k and considered are: 10-day, 20-day and 30-day ahead forecast. Bitcoin
  data is sourced from CoinDesk (https://www.coindesk.com/price).
rt+k − r̂ 2t,t+k are the forecasted errors due to the restricted Baker et al. (2016) is the source of EPU data; VIX and oil prices are
 
collected from St. Louise Fred databank.
and unrestricted models, respectively. r̂ 1t,t+k − r̂ 2t,t+k is the

4. Empirical results

4
Table 1 reveals the descriptive statistics. The Table indicates
Studies have found that financial returns tends to exhibit days-of-the-week
effect (Salisu, Raheem, & Vo, 2021). The return series is regressed against dummy
that the pricing dynamics of bitcoin was negatively impacted as
variables constructed for the five days of the week, which would help obtain days- a result of the pandemic. This is due to the relatively higher return
of-the week adjusted returns. (Salisu, Raheem, & Vo, 2021). for the periods prior to the pandemic’s announcement, compared

372
I.D. Raheem The Quarterly Review of Economics and Finance 81 (2021) 370–375

Table 2
Unit Root Tests.

ADF Test NL Test

Bitcoin VIX EPU WTI Bitcoin VIX EPU WTI

Pre-COVID Announcement
−8.8089 a −8.7096 a −14.1421a −10.1384 a −9.561 a −5.4842 a −11.0776 a −13.245 a
I(0) I(1) I(1) I(0) I(0) I(0) I(1) I(0)
Post-COVID Announcement
−10.9371a −4.3894 a −17.2788a −12.1845a −12.053a −19.0114I(1) −15.8577I(1) −15.252a
I(0) I(1) I(1) I(0) I(0) I(1)

Source: Authors’ computation.

Note: a implies 1% level of statistical significance, while I(0) and I(1) implies level and first difference stationary.
ADF = Augmented Dickey Fuller and NL is the Narayan and Liu (2015) test.

Table 3
Persistence and Endogeneity Model.

Pre-COVID Post-COVID

Endogeneity Endogeneity
Persistence Persistence
VIX WTI EPU VIX WTI EPU

0.104 a 13.826a 0.0002 a −0.4688 −0.0652a −0.238 −0.002 −40.325

Source: Authors’ computation.

Note: a implies 1% level of statistical significance.

Table 4 found that Bitcoin provides a weak (almost non-existence) haven

Results of Predictability Test.
for the oil markets. Our results about oil shock are also similar to
VIX EPU WTI the conclusion of Selmi et al. (2018) that the safe haven hypothesis
Pre-COVID-19 Announcement of Bitcoin does not hold all the time.
0.0654*** 0.0598** 0.0050* Table 5 reports the out-of-sample forecast evaluations. The deci-
(0.0013) (0.0198) (0.0015) sion criteria for the Theil-U statistics is that values less than unity
[5.659] [3.255] [2.536] imply the unrestricted (i.e. uncertainty augmented) models are
Post-COVID-19 Announcement
the preferred models relative to the restricted (i.e. historical aver-
0.00256** 0.0364** 0.0002
(0.0085) (0.0103) (0.0016) age) model. For the pre-announcement era, our results support the
[2.2153] [1.569] [2.556] preference for the unrestricted model. These results hold for the
Note: *** and ** & * imply the rejection of the null hypothesis of no predictability at different measures of uncertainty and forecasting horizons. The
1%, 5% & 10 % levels of significance, respectively. Values in parentheses - () denote significance of the CW test further strengthens the results.
standard errors while those reported in square brackets – [] are for t-statistics. Focusing on the post-announcement period, both measures of
forecasting evaluation were unable to confirm the safe haven prop-
erty of Bitcoin. It then suggests that bitcoin can only shield investors
to the post announcement era. Unsurprisingly, there is higher
of potential loss if the severity of the shock is mild. Since the advent
level of uncertainty (as proxied by VIX and EPU) for the pan-
of Bitcoin, the global economy has not witnessed shock of similar
demic era. Statistics confirm weak potency of autocorrelation and
magnitude to those exhibited by COVID-19. This might partly be
heteroscedasticity for Bitcoin for period before the pandemic. How-
an attributable reason for the previously impressive performance
ever, the exact opposite suffices for the pandemic period. The two
of Bitcoin as reported in the literature. Relatively newer studies
unit root tests explored in this study are the Augmented Dickey-
have also supported our results. For instance, many studies have
Fuller and Narayan and Liu (2015). Table 2 provides results of the
shown a positive and high correlation between Bitcoin and stock
unit root tests, which confirms Bitcoin is level stationary, while VIX
returns (Kristoufes, 2020; Conlon and McGee, 2020; and Grobys,
and EPU are first difference stationary, over the competing periods.
2020). If Bitcoin was truly a safe haven asset, it is expected that it
There is evidence of persistence and endogeneity in the variables
should have a negative correlation with other competing financial
of interest, as shown in Table 3. Persistence is calculated based on
assets.
AR(1).
Results of the predictability test is presented in Table 4. It can be
deduced from the Table that Bitcoin is able to provide safe haven 5. Conclusion
against some exogenous shocks during the period of market unrest,
especially for EPU and VIX. The prowess of the safe haven feature Bitcoin has emerged as the new financial instrument that
is higher for pre-COVID announcement. Largely, these results have promises to provide safe haven for investors. Since its emergence
been confirmed by previous studies (e.g. Mariana et al., 2020; Selmi, and prior to COVID-19 pandemic, there has not been any significant
Mensi, Hammoudeh, & Bouoiyour, 2018). This finding is also sim- shock to the financial market, hence the inability to empirically
ilar to the conclusion of Hood and Malik (2013) that safe haven validate the safe haven behaviour of Bitcoin in extreme market
effectiveness of financial assets is lower during crisis period. Inter- period(s). This study specified a predictive model for the out-of-
estingly, a strand of the literature has shown that the significance sample forecast of Bitcoin’s safe haven property.
of the safe haven characteristics of Bitcoin is shock-specific (Selmi We show that prior to COVID-19, Bitcoin was able to main-
et al., 2018; Shahzad, Bouri, Roubaud, & Kristoufek, 2020). For tain its widely acknowledged characteristics. However, the post
instance, Bouri et al. (2017) and Dutta et al. (2020) concluded that COVID-19 announcement upturned the tides and showed that
Bitcoin has a higher hedging and diversification prowess against the safe haven hypothesis of Bitcoin fizzles out. Hence, it is safe
oil markets related shock. Relating this stance to our results, we to conclude that the safe haven property of Bitcoin is only lim-

373
I.D. Raheem The Quarterly Review of Economics and Finance 81 (2021) 370–375

Table 5
Results of Out-of-Sample Forecast Evaluation.

VIX EPU WTI

H = 10 H = 20 H = 30 H = 10 H = 20 H = 30 H = 10 H = 20 H = 30

Theil 0.997 0.993 0.995 0.994 0.994 1.006 0.985 0.976 1.004
Pre-Announcement
CW 3.727 3.254 1.652 2.984 2.287 1.285 3.174 4.612 1.259
Theil 1.265 1.005 1.018 1.009 1.039 1.049 0.992 1.368 1.227
Post-Announcement
CW 3.823 4.528 3.054 4.101 2.592 1.425 3.215 2.979 3.721

Note: Theil is the Theil U statistics.

ited to calm and tranquil financial market conditions. We also Conlon, T., & McGee, R. (2020). Safe haven or risky hazard? Bitcoin during the
show that the prowess of the safe haven nature is sensitive to COVID-19 bear market. Finance Research Letters, https://doi.org/10.1016/j.frl.
2020.101607
the type of shock, as Bitcoin was able to provide strong cover Conlon, T., Corbet, S., & McGee, R. (2020). Are cryptocurrencies a safe haven for
against EPU and VIX shocks; the same cannot be said about oil equity markets? An international perspective from the COVID-19 pandemic.
shock. Research in International Business and Finance, 54, Article 101248 https://doi.
org/10.1016/j.ribaf.2020.101248
An obvious policy implication of the study is that market par- Corbet, S., Larkin, C., & Lucey, B. (2020). The contagion effects of the COVID-19
ticipants would be better off by designing their portfolios to align pandemic: Evidence from gold and cryptocurrencies. Finance Research Letters,
more towards Bitcoin during tranquil period. In addition, the https://doi.org/10.1016/j.frl.2020.101554
Corbet, S., Lucey, B., Urquhart, A., & Yarovaya, L. (2019). Cryptocurrencies as a
non-oil market investors should consider building portfolios that
financial asset: A systematic analysis. International Review of Financial Analysis,
would seek to include more presence of Bitcoin, especially when 62, 182–199.
the market is in unsettled state. However, oil market investors Dutta, A., Das, D., Hana, R. K., & Vo, X. V. (2020). COVID-19 and oil market crash:
Revisiting the safe haven property of gold and Bitcoin. Resources Policy, 65
should consider other assets, aside Bitcoin, in their investment
https://doi.org/10.1016/j.resourpol.2020.101816
strategies to offset the down-risk of the market in the pandemic Goodwell, J. W., & Goutte, S. (2020). Co-movement of COVID-19 and Bitcoin:
era (i.e. when the market is in turmoil state). We would like Evidence from wavelet coherence analysis. Finance Research Letters, Article
to urge readers to interpret our results with caution given the 101625 https://doi.org/10.1016/j.frl.2020.101625
Grobys, K. (2020). When bitcoin has the flu: On Bitcoin’s performance to hedge
small sample size and the infancy in the conception of cryptocur- equity risk in the early wake of the COVID-19 outbreak. Applied Economic
rencies as compared to the traditional assets. There has been Letters, https://doi.org/10.1080/13504851.2020.1784380
increased in the pace of Bitcoin’s volatility in the pandemic era, as Hood, M., & Malik, F. (2013). Is gold the best hedge and a safe haven under
changing stock market volatility? Review of Financial Economics, 22,
compared to pre-pandemic (as indicated in Table 1). This might 47–52.
imply that the safe haven characteristics is dynamic and time- Iqbal, J. (2017). Does gold hedge stock market, inflation and exchange rate risks?
varying. As such, this provides an opportunity for future studies to An econometric investigation. International Review of Economics & Finance, 48,
1–17.
account for the time-varying nature of the safe haven properties of Isah, K., & Raheem, I. D. (2019). The hidden predictive power of cryptocurrencies
Bitcoin. and QE: Evidence from US stock market. Physica A, 536(121032), 1–10.
Ji, Q., Bouri, E., Gupta, R., & Roubaud, D. (2019). Network causality structures
among Bitcoin and other financial assets: A directed acyclic graph approach.
Quarterly Review of Economics and Finance, 70, 203–213.
Declaration of Competing Interest Kristoufes. (2020). Grandpa, grandpa, tell me the one about Bitcoin being a safe
haven: Evidence from the COVID-19 pandemics Available at. https://ideas.repec.
There is no conflict of interest between the authors. org/p/arx/papers/2004.00047.html
Lewellen, J. (2004). Predicting returns with financial ratios. Journal of Financial
Economics, 74, 209–235.
Ma, C., Rogers, J., & Zhou, S. (2020). Global economic and financial effects of 21st
References century pandemics and epidemics Available at. https://papers.ssrn.com/sol3/
papers.cfm?abstract id=3565646
Mariana, C. D., Ekaputra, I. A., & Husondo, Z. A. (2020). Are Bitcoin and Ethereum
Aloui, R., Aissa, B., & Nguyen, D. K. (2011). Global financial crisis, extreme
safe-havens for stocks during the COVID-19 pandemic? Finance Research
interdependences, and contagion effects: The role of economic structure?
Letter., https://doi.org/10.1016/j.frl.2020.101798
Journal of Banking & Finance, 35, 130–141.
Naeem, M. A., Bouri, E., Peng, Z., Shahzad, S. J. H., & Vo, X. V. (2021). Asymmetric
Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring economic policy uncertainty.
efficiency of cryptocurrencies during COVID19. Physica A: Statistical mechanics
Quarterly Journal of Economics, 131(4), 1593–1636.
and its applications (vol. 565(C) Elsevier.
Baker, S. R., Bloom, N., Davis, S. J., Kost, K., Sammon, M., & Viratyosinm, T. (2020).
Narayan, P. K., & Liu, R. (2015). A unit root model for trending time-series energy
The unprecedented stock market reaction to COVID-19. In NBER working paper
variables. Energy Economics, 50, 391–402.
series number NBER working paper No. 26945.
Ramona, O., Christina, M. A., & Raluca, S. (2019). BITCOIN in the scientific literature
Baker, S. R., Bloom, N., Davis, S. J., & Terry, S. J. (2020). COVID-induced economic
– A bibliometric study. Studies in Business and Economics, 14(3), 160–174.
uncertainty Available on the link. http://www.policyuncertainty.com/media/
Salisu, A. A., Akanni, L., & Raheem, I. D. (2020). The COVID-19 global fear index and
COVID-Induced%20.pdf
the predictability of commodity price returns. Journal of Behavioral and
Bouri, E., Molnar, P., Azzi, G., Roubaud, D., & Hagfors, L. I. (2017). On the hedge and
Experimental Finance, 27, Article 100383 https://doi.org/10.1016/j.jbef.2020.
safe haven properties of Bitcoin: Is it really more than a diversifier? Finance
100383
Research Letters, 20, 192–198.
Salisu, A. A., Raheem, I. D., & Eigbiremolen, G. (2020). The behaviour of US stocks to
Bouri, E., Gupta, R., Lauc, C. K. M., Roubaudd, D., & Wang, S. (2019). Bitcoin and
financial and health risks. International Journal of Finance and Economics,
global financial stress: A copula-based approach to dependence and causality
https://doi.org/10.1002/ijfe.2390
in the quantiles. Quarterly Review of Economics and Finance, 69, 297–307.
Salisu, A. A., Raheem, I. D., & Vo, X. V. (2021). Assessing the safe haven property of
Bouri, E., Demirer, R., Gupta, R., & Pierdzioch, C. (2020). Infectious diseases, market
the gold market during COVID-19 pandemic. International Review of Financial
uncertainty and oil market volatility. Energies, 13, 4090. https://doi.org/10.
Analysis, 74, Article 101666 https://doi.org/10.1016/j.irfa.2021.101666
3390/en13164090
Selmi, R., Mensi, W., Hammoudeh, S., & Bouoiyour, J. (2018). Is Bitcoin a hedge, a
Bouri, E., Shahzad, S. J. W., & Roubaud, D. (2020). Cryptocurrencies as hedges and
safe haven or a diversifier for oil price movements? A comparison with gold.
safe-havens for US equity sectors. Quarterly Review of Economics and Finance,
Energy Economics, 74, 787–801.
75, 294–307.
Shahzad, S. J. H., Bouri, E., Roubaud, D., & Kristoufek, L. (2020). Safe haven, hedge
Bouri, E., Shahzad, S. J. W., Roubaud, D., Kristoufek, L., & Lucey, B. (2020). Bitcoin,
and diversification for G7 stock markets: Gold versus bitcoin. Economic
gold, and commodities as safe havens for stocks: New insight through wavelet
Modelling, 87, 212–224.
analysis. Quarterly Review of Economics and Finance, 77, 156–164.
Sharif, A., Aloui, C., & Yarovaya, L. (2020). COVID-19 pandemic, oil prices, stock
Campbell, J. Y., & Thompson, S. B. (2008). Predicting excess stock returns out of
market, geopolitical risk and policy uncertainty nexus in the US economy:
sample: Can anything beat the historical average? The Review of Financial
Fresh evidence from the wavelet-based approach. International Review of
Studies, 21(4), 1509–1531.
Financial, https://doi.org/10.1016/j.irfa.2020.101496
Clark, T. E., & West, K. D. (2007). Approximately normal tests for equal predictive
accuracy in nested models. Journal of Econometrics, 138, 291–311.

374
I.D. Raheem The Quarterly Review of Economics and Finance 81 (2021) 370–375

Shehzad, K., Xiaoxing, L., & Kazouz, H. (2020). Covid-19’s disasters are perilous Westerlund, J., & Narayan, P. K. (2012). Does the choice of estimator matter when
than global financial crisis: A rumor or fact? Finance Research Letters, https:// forecasting returns? Journal of Banking and Finance, 36, 2632–2640.
doi.org/10.1016/j.frl.2020.101669 Westerlund, J., & Narayan, P. K. (2015). Testing for predictability in conditionally
Smales, L. A. (2019). Bitcoin as a safe haven: Is it even worth considering? Finance hetoroscedasticity stock returns. Journal of Financial Econometrics, 13, 342–375.
Research Letters, 63, 322–330.
Syriopoulos, T., Makram, B., & Boubaker, A. (2017). Stock market volatility
spillovers and portfolio hedging: BRICS and the financial crisis. International
Review of Financial Analysis, 39, 7–18.

375