Content uploaded by Ahmed Hamed Elsayed
Author content
All content in this area was uploaded by Ahmed Hamed Elsayed on Feb 02, 2022
Content may be subject to copyright.
1
Risk Transmissions between Bitcoin and Traditional Financial Assets
during the COVID-19 Era: The Role of Global Uncertainties
Ahmed H. Elsayed
*
Department of Economics and Finance, Durham University, the United Kingdom
Department of Economics, Faculty of Commerce, Zagazig University, Egypt
Email: ahmed.elsayed@durham.ac.uk
Giray Gozgor
Faculty of Political Sciences, Istanbul Medeniyet University, Turkey
Email: giray.gozgor@medeniyet.edu.tr
Chi Keung Marco Lau
School of Business, Teesside University, the United Kingdom
Email: C.Lau@tees.ac.uk
Abstract
This paper examines return and volatility connectedness between Bitcoin, traditional financial
assets (Crude Oil, Gold, Stocks, Bonds, and the United States Dollar-USD), and major global
uncertainty measures (the Economic Policy Uncertainty-EPU, the Twitter-based Economic
Uncertainty-TEU, and the Volatility Index-VIX) from April 29, 2013, to June 30, 2020. To this
end, the Time-Varying Parameter Vector Autoregression (TVP-VAR) model, dynamic
connectedness approaches, and network analyses are used. The results indicate that total
spillover indices reached unprecedented levels during COVID-19 and have remained high
since then. The evidence also confirms the high return and volatility spillovers across markets
during the COVID-19 era. Regarding the return spillovers, Gold is the centre of the system and
demonstrates the safe heaven properties. Bitcoin is a net transmitter of volatility spillovers to
other markets, particularly during the COVID-19 period. Furthermore, the causality-in-
variance Lagrange Multiplier (LM) and the Fourier LM tests' results confirm a unidirectional
volatility transmission from Bitcoin to Gold, Stocks, Bonds, the VIX and Crude Oil.
Interestingly the EPU is the only global factor that causing higher volatility in Bitcoin. Several
potential implications of the results are also discussed.
Keywords: Return connectedness; Volatility connectedness; Bitcoin; Financial assets; Global
uncertainty measures
JEL Codes: C32, C50, G10, G11
*
Corresponding author: ahmed.elsayed@durham.ac.uk
2
1. Introduction
Bitcoin is becoming more integrated into the global financial system every year. The market
capitalisation of Bitcoin has exceeded the threshold of $1T in March 2021, a year after the
starting of the global COVID-19 pandemic. Although the market capitalisation of Bitcoin
decreased around $600B in July 2021, Bitcoin still has a dominant role in the global
cryptocurrency market, which has a market cap of around $1.3T. It is also important to note
that the dominance of Bitcoin has been steadily reducing with the introduction of Altcoins
(Elsayed et al., 2021; Ji et al., 2019; Shi et al., 2020; Yi et al., 2018), but Bitcoin still has almost
three-folds higher market cap than the runner cryptocurrency (Ethereum).
On the other hand, tremendous price volatility in Bitcoin has been observed during the
COVID-19 era. On March 11, 2020, the World Health Organization (WHO) announced that
the COVID-19 had been a global pandemic and the closing price of Bitcoin was $7,911. A day
later, the price of Bitcoin plunged to $4,970. However, Bitcoin's price had experienced a
significant upward trend throughout the year, and its closing price has first exceeded $60K on
March 13, 2021 (Coindesk.com, 2021). However, it decreased to the level of around $30K in
July 2021. Accordingly, various research has been performed to analyse Bitcoin's returns and
price volatility determinants during the COVID-19 crisis (see, e.g., Goodell and Goutte, 2021a;
Jiang et al., 2021).
Empirical literature examines different characteristics of cryptocurrencies. For
example, Caporale et al. (2018), Tiwari et al. (2018), and Urquhart (2016) investigate the
inefficiency of cryptocurrency markets. Corbet et al. (2018) find the significance of speculative
bubbles. Conlon et al. (2020), Damianov and Elsayed (2020), Liu et al. (2020), and Urquhart
and Zhang (2019) show the significant hedging and safe haven features. Bouri et al. (2019)
observe the significant herding feature of cryptocurrencies. Several studies also examine the
drivers and the patterns of price volatility (see, e.g., Caporale and Zekokh, 2019; Elsayed et al.,
3
2021; Katsiampa, 2019; Shi et al., 2020). Some studies analyse investors' attention to
cryptocurrencies (Dastgir et al., 2019; Urquhart, 2018).
In light of these developments, this paper analyses return and price volatility
connectedness between Bitcoin and traditional financial assets (Crude Oil, Gold, Stocks,
Bonds, and the USD). We also include the role of global uncertainty measures (the EPU, the
TEU, and the VIX) to address the catalyser impact of global uncertainty during the COVID-19
pandemic on the relationship between Bitcoin and traditional financial markets. This issue also
links our empirical exercises to the previous empirical papers, which use the global uncertainty
measures as the driver of the returns and the price volatility in Bitcoin and other cryptocurrency
markets (see, e.g., the VIX in Bouri et al., 2017, the EPU in Demir et al., 2018, and the TEU
in Wu et al., 2021). Furthermore, Fang et al. (2019) extend the results of Demir et al. (2018)
and find that the global EPU measure is a driving factor of the returns and the price volatility
of Bitcoin. Wu et al. (2019) demonstrate that Bitcoin has more hedging capacity than Gold
during times of economic policy uncertainty shocks. Gozgor et al. (2019) observe the
significant hedging feature of Bitcoin against trade policy uncertainty shocks in the United
States. Following the spirit of Demir et al. (2018), Cheng and Yen (2020) show that the Chinese
EPU has a significant capacity to predict Bitcoin returns, and the impact is negative. In a further
study, Yen and Cheng (2021) demonstrate that the Chinese EPU can successfully predict the
price volatility of Bitcoin. However, Wang et al. (2019) find an insignificant impact of the EPU
on the price volatility of Bitcoin. Colon et al. (2021) also observe that cryptocurrencies have a
weak hedging capacity against economic policy uncertainty shocks, especially during
optimistic economic expectations.
Indeed, global uncertainty has a significant role in Bitcoin as monetary and fiscal policy
implications have weakened the trust in traditional financial assets during the Global Financial
Crisis in 2008. This issue has also led to the introduction of Bitcoin in 2008 as a decentralised
4
alternative payment and investment asset (Aysan et al., 2019). Particularly, during periods with
higher economic policy uncertainty, as observed during the COVID-19 pandemic, Bitcoin can
be an effective alternative to traditional assets. It can hedge risks against uncertainty shocks
(Goodell, 2020). Nevertheless, the impact of economic policy uncertainty shocks on
cryptocurrency markets during the COVID-19 crisis is still limited. For instance, Goodell and
Goutte (2021a) show that the COVID-19 was positively related to the Bitcoin prices in April
2020. So et al. (2021) demonstrated that financial market connectedness in Hong Kong
financial markets increases substantially during the 2020 COVID-19 outbreak. Wu et al. (2021)
find that economic policy uncertainty during the COVID-19 pandemic has a limited effect on
the returns of major cryptocurrencies. Given that our sample covers the data from April 29,
2013, to June 30 (capturing the COVID-19 crisis), we control the role of global uncertainty
measures.
Following the previous papers, we also include traditional financial assets (Crude Oil,
Gold, Stocks, Bonds, and the USD) to address safe haven properties of these assets during
periods of higher uncertainty (see, e.g., Ji et al., 2020 for commodities; Bredin et al., 2015 for
Gold; Ashraf, 2020 for the stock markets; Ranaldo and Söderlind, 2010 for the USD; Flavin et
al., 2014 for the bonds; Goodell and Goutte, 2021b; Conlon et al., 2020 for cryptocurrencies).
†
Unlike previous literature, our paper utilises several econometric methods (time-
varying parameter vector autoregression model, causality-in-variance tests, dynamic
connectedness approach, and network analyses) to examine the relationships among Bitcoin,
traditional financial assets (Gold, Crude Oil, Stocks, Bonds, and the USD), and major global
uncertainty measures (the EPU, the TEU, and the VIX), including the COVID-19 era. The
results show that total spillover indices have reached unprecedented levels during the COVID-
†
See Corbet et al. (2019) for a review of the empirical literature in cryptocurrency markets.
5
19 period and remain high since then, confirming high return and volatility spillovers across
markets during the pandemic. Bitcoin is a net receiver of return spillovers from other markets;
however, it transmits volatility spillovers to other markets. Therefore, Bitcoin plays a major
role in volatility transmission during the COVID-19 period. Simultaneously, economic policy
uncertainty is the only global uncertainty factor to increase volatility in Bitcoin. The directions
of these relationships are robust to different econometric techniques. To the best of our
knowledge, our study provides the first evidence in the empirical literature to show the pivotal
role of Bitcoin in global financial markets and global uncertainty, including the COVID-19 era.
Indeed, the EPU is the significant driver of the boom and bust cycle (price volatility) in Bitcoin.
The remaining structure of the paper is designed as follows. Section 2 explains the
description of the data and the econometric methods. Section 3 discusses the empirical findings
with their potential implications. Section 4 concludes.
2. Data and Econometric Methodology
2.1. Data Sources and Descriptive Statistics
We consider the data for Bitcoin (closing price) and traditional financial assets: Stocks (S&P
500 index), Bonds (S&P 500 bond index), the United States Dollar (broad exchange rate), Gold
(spot prices) and Crude Oil (West Texas Intermediate-WTI spot prices). These variables are
calculated as the first logarithmic difference between two consecutive observations. Following
the usual practice, the Volatility Index (VIX), the United States (US) Economic Policy
Uncertainty (EPU) index, and Twitter-based Economic Uncertainty (TEU) index are calculated
in logarithmic form. All data series are collected from the Thomson Reuters DataStream,
except for the Economic Policy Uncertainty, and the Twitter-based Economic Uncertainty
6
indices are collected from Baker et al. (2016).
‡
We focus on the daily sample from April 29,
2013, to June 30, 2020, and the starting date is related to the data availability.
2.2. Econometric Methodology: TVP-VAR Dynamic Connectedness Approach
To answer the search questions, we utilise the Time-Varying Parameter Vector Autoregressive
(TVP-VAR) model developed by Koop and Korobilis (2014) in conjunction with the dynamic
connectedness approach introduced by Diebold and Yilmaz (2012 and 2014). Including the
TVP-VAR model overcomes Diebold and Yilmaz's connectedness approach's limitations and
significantly improves the estimation technique in several ways. Firstly, the TVP-VAR model
provides a more accurate measurement of connectedness since the rolling window approach
overestimates the connectedness measures by generating the "built-in persistence" and hence
does not accurately capture the downward move in connectedness indices on time. Secondly,
there is no need to arbitrarily set a fixed rolling window to capture the dynamics of the
connectedness across variables as the VAR parameters are allowed to vary over time. Thirdly,
no losses of observations as a result of the rolling window estimation approach. Finally, the
TVP-VAR-based connectedness technique is not sensitive to outliers (Antonakakis et al., 2018
and 2019; Koop and Korobilis, 2014, Korobilis and Yilmaz, 2018).
A stationary TVP-VAR model of order one can be written as:
is an vector of variables under consideration, while is an vector of
the disturbance errors with time-varying variance-covariance matrix (). is an
dimensional time-varying coefficient matrix, whereas is the vectorisation of
which is an dimensional vector. Finally, is an vector of error terms
with covariance matrix of an dimensional matrix. Consequently, based on the
‡
They are downloaded from the website (https://www.policyuncertainty.com).
7
Wold representation theorem, the TVP-VAR model could be transferred into its Moving
Average representation (VMA) in the form of:
With is an dimensional matrix. Following this, the TVP-VMA estimates are
used to calculate the Generalised Impulse Response Functions (GIRF) and Generalised
Forecast Error Variance Decomposition (GFEVD) of Koop et al. (1996) and Pesaran and Shin
(1998) that form the basis of the dynamic connectedness approach. Accordingly, the H-step-
ahead GFEVD function could be written as follows:
Where represents the response of all variables to a shock in variable . By
construction,
and
. On that basis, the Total Connectedness
Index (TCI) is calculated as the ratio of the sum of all off-diagonal elements to the total
variance. It, therefore, represents the average contribution of volatility spillovers across all
variables to the total forecast error variance and hence is calculated as follows:
The above representation of the generalised variance decomposition matrix is helpful
as it allows the estimation of directional spillover indices among variables. The total directional
connectedness from others is defined as the spillovers received by variablefrom all other
variables, , which is measured as:
Likewise, the total directional connectedness to others demonstrates the informational
outflow transmitted from variableto all other variables, which is given by:
8
Consequently, the net directional connectedness, the net spillovers transmitted/received
by variable , is measured as the difference between total directional spillovers "to" and "from"
variable :
The calculation of net spillover indices is very valuable as it allows us to determine
whether a given variable is a net transmitter or receiver of shocks.
where positive (negative) values of the net pairwise connectedness index imply that the
variable is a net transmitter (receiver) of shocks to (from) other variables. At last, the directional
connectedness network is constructed based on the Net Pairwise Spillover (NPS)
connectedness between and variables as follows:
where nodes represent components of a generalised variance decomposition matrix
(variables), and edges demonstrate the direction and strength of the pairwise connectedness
among each pair. Notably, the linkages between nodes are directed given that
that is
the volatility spillover transmitted from variable to variable is not necessarily equivalent to
those received by from .
3. Empirical Findings
3.1 Preliminary Results
Figure 1 presents the time series plot of raw data for the spot Bitcoin prices, the S&P 500 Stock
Index, the S&P 500 Bond Index, the USD Broad Exchange Rate, Gold, the WTI Oil Price, the
VIX, the US Economic Policy Uncertainty (US_EPU) index, and Twitter-based economic
uncertainty (TEU) index from April 29, 2013, to June 30, 2020. We can see extreme
movements starting from January 1, 2020. The gold price reached its historical high while the
9
WTI Crude Oil price reached its lowest point (see the highlighted region in Figure 1). We also
notice that all uncertainty measures climbed up significantly during the first wave of the
COVID-19 pandemic.
<<Figure 1 appropriately here>>
<<Figure 2 appropriately here>>
Figure 2 portrays the transformed series under consideration, including returns on
Bitcoin and other asset classes (i.e., the USD, Bonds, Stocks, Gold, and Crude Oil Returns),
the VIX, the US EPU index, and Twitter-based economic uncertainty (TEU) index in log form.
Table 1 shows the descriptive statistics for daily variables. We observe the stationarity
property of variables in interest following the results of Augmented Dickey-Fuller (ADF) and
Phillips-Perron (PP) unit root tests (see Table 1).
<<Table 1 appropriately here>>
3.2 Return Spillovers
Table 2 presents the findings of the TVP-VAR model. As can be seen, the overall level of
connectedness is 21.4%, which means that 21.4% of the forecast error variance comes from
the connectedness between the underlying variables. Table 2 also summarises the empirical
results of the directional and pairwise return spillovers between Bitcoin, financial markets, and
global uncertainty measures. These results are based on the generalised forecast error variance
decomposition (GFEVD) obtained from the TVP-VAR model of order one and 10-step ahead
forecasts. The lag length is selected by the Bayesian information criterion (BIC). It is observed
that the financial market volatility (VIX) contributes most of the spillover to the stock market
(i.e., 32.23%), while its spillover to the Bitcoin market is minimal, with only 0.378%. The
impact of the US EPU on financial markets is not significant. It contributes the most to the
Crude Oil market (0.259%) while contributing only 0.09% to the Bitcoin market. Similarly,
the impact of Twitter Based Economic Uncertainty (TEU) on financial markets is muted. It
10
transmits spillover mostly to the stock market (0.306%) while contributing only 0.023% to the
Bitcoin market.
It is worthy to note that the stock market is the main net transmitter of information in
the financial markets, amounting to 11.67%. In contrast, the bond market is a primary net
receiver of spillover (-5.56%).
<<Table 2 appropriately here>>
Figure 3 summarises the directional connectedness network of pairwise return
spillovers, and it shows the average pairwise directional return spillovers among all possible
pairs of variables in the model. A node's colour implies whether a variable is a net
transmitter/receiver of return spillovers. The red colour indicates a net transmitter, while the
green colour shows a net receiver, respectively. Furthermore, the thickness and arrows' colour
represents the average return spillovers magnitude and strength between each pair. Red
indicates strong, navy shows moderate, and green refers to weak return spillovers. The result
highlights the central role of the VIX and stock markets. They are the net transmitter with
significant magnitude. The pairwise relationship between Gold and USD is interesting as they
are significantly affecting each other. Moreover, the same pairwise relationship is found
between the US EPU and the TEU, where the latter has a greater impact on the former.
<<Figure 3 appropriately here>>
Figure 4 portrays the time-varying behaviour of the total connectedness index across
the Bitcoin, financial markets and major global uncertainty measures from April 29, 2013, to
June 30, 2020. The total connectedness index is time-varying and very responsive to economic
and financial turbulences. In line with the preliminary analysis, the total connectedness index
peaked when the COVID-19 pandemic started early 2020.
<<Figure 4 appropriately here>>
11
In addition, Figure 5 portraits the dynamic net directional return spillover indices for
each variable throughout the full sample period. Noteworthy, the net directional return
spillovers are time-varying and behaviour differently over different periods. Positive (negative)
values indicate that the variable is a net transmitter (receiver) of return spillover to (from) all
other variables. As can be seen, Stock and Bitcoin markets are the main net return transmitters
to others, while Bond, Crude Oil, Gold, and the USD markets are the net return receivers.
<<Figure 5 appropriately here>>
Figure 6 provides the dynamic connectedness of pairwise return spillovers, and the
graph represents the pairwise directional return spillovers time-varying behaviour between
Bitcoin and each of the variables under consideration. Positive (negative) values indicate that
Bitcoin is transmitting (receiving) return spillover to (from) the other variable. Bitcoin is
receiving returns spillover from the TEU and the EPU before the COVID-19 pandemic.
However, it is transmitting returns spillover to the TEU and the EPU during the second wave
of the COVID-19 pandemic, while Bitcoin received returns spillover from the VIX for most of
the time. Regarding other financial assets, we observe that Bitcoin is the returns transmitter of
spillovers to all financial assets during the COVID-19 pandemic, except the stock market,
where Bitcoin is the returns receiver. Interestingly, the directional spillover changed after the
first wave of the COVID-19 pandemic (for example, Gold and Bond markets).
<<Figure 6 appropriately here>>
3.3 Volatility Spillovers and Causality in-variance Tests
Now, let us turn our attention to the analysis of volatility spillovers among Bitcoin, financial
assets and uncertainty measures. Table 3 reports the volatility connectedness where the total
connectedness index accounts for 31.7% of the total forecast error variance. It is worth noting
that the results for the VIX as a transmitter are the same as that of returns connectedness.
Regarding the impact of the EPU on the USD is the largest volatility receiver, while Crude Oil
12
is the least impacted market. For the TEU, the results are the same as that of the returns spillover
analysis. Finally, similar to the returns spillover analysis, the stock market is the major net
transmitter in the system (105.61%), followed by VIX (43.87%). In contrast, the bond market
is the main net receiver of the spillover (-53.13%), followed by the oil market (-42.38%).
<<Table 3 appropriately here>>
Figures 7 to 10 presents the spillover patterns for volatility spillover among variables.
These diagrams show the average pairwise directional volatility spillovers among all possible
pairs of variables in the model. We observe that the stock market is in the centrality of the
system. It transmitted volatility to all financial assets, including Bitcoin. The biggest receiver
of the volatility from the stock market is the bond market. The VIX transmitted the biggest
volatility to the stock market.
<<Figure 7 appropriately here>>
Figure 8 shows the total volatility spillover index with time-varying behaviour of the
total volatility connectedness across the Bitcoin, financial markets and major global uncertainty
measures. Interestingly, the system has the highest volatility spillover during the first wave of
the COVID-19 pandemic.
<<Figure 8 appropriately here>>
Figure 9 displays the net directional volatility spillovers time-varying behaviour for
each variable under consideration. It's worth noting that the stock and the VIX are the net
contributors of volatility to other variables. Crude Oil and Bitcoin are also the net volatility
transmitters during the COVID-19 pandemic. Bond, Gold, and Dollar are the net receiver of
volatility during the COVID-19 pandemic. It is interesting to note that the directional spillover
pattern of Bitcoin and Crude Oil is similar. They changed from net volatility receiver before
the COVID-19 pandemic to net volatility transmitter during the first wave of the COVID-19
pandemic. This behaviour of changing directional spillover was observed for the USD, but it
13
changed from net volatility transmitter before the COVID-19 pandemic to net volatility
receiver during the first wave of the COVID-19 pandemic
<<Figure 9 appropriately here>>
Finally, Figure 10 indicates the time-varying behaviour of the pairwise directional
volatility spillover between Bitcoin and each variable under consideration. We first observe
that Bitcoin transmits volatility to all volatility indices and Gold during COVID-19 pandemics
while it also transits volatility to the stock market in the second wave. At the same time, it
receives volatility from the VIX and stock market before the COVID-19 pandemic. For bonds
and the USD, Bitcoin is the receiver of volatility spillover during the COVID-19 pandemic. At
the same time, we observed that Bitcoin was the net volatility transmitter before the COVID-
19 pandemic.
<<Figure 10 appropriately here>>
As a robustness test, we conducted a causality-in-variance Lagrange Multiplier (LM)
test of Hafner and Herwartz (2006) and the Fourier LM test introduced by Li and Enders (2018)
to investigate the causality direction in-variance between Bitcoin, traditional financial assets,
and major global uncertainty measures. Results are reported in Table 4, where the LM tests
show a unidirectional risk transmission from Bitcoin to other markets, including Gold, Stock
market, Bonds, the VIX, and the Crude Oil markets. Simultaneously, the EPU is the only global
factor that causing higher volatility in Bitcoin.
<<Table 4 appropriately here>>
3.4 Discussion and Implications of the Findings
The findings in this paper have several implications on portfolio allocations and diversification
benefits. Specifically, if there is a well-diversified portfolio, there will be a decline in the return
and the volatility spillovers among assets and uncertainty measures. Therefore, it can be
suggested that efficient portfolio allocations can hedge the market risks related to uncertainty
14
shocks. However, if there will be a significant increase in the returns and volatility spillovers
among financial markets and uncertainty measures, the spillover effects reduce diversification
benefits. In this case, uncertainty shocks can create systematic risks, which are not subject to
hedging features. Therefore, our findings on the Bitcoin-Traditional Financial Assets-
Uncertainty Measures system have important implications for portfolio diversification benefits
and effective risk management.
We observe that total spillover indices have reached unprecedented levels during the
COVID-19 period. This evidence is in line with the hypothesis provided by previous papers
(e.g., So et al., 2020) that there is a significant connectedness during the period of high-level
uncertainty, such as observed during the COVID-19 era. Our results confirm this hypothesis
since the high level and persistence return and volatility spillovers are observed across Bitcoin
and financial markets during the COVID-19 era. This evidence implies decreasing
diversification benefits, especially during the COVID-19 crisis.
On the other hand, Gold is the centre of the system regarding the return spillover and
demonstrates the "safe heaven" properties against uncertainty shocks. This evidence aligns
with previous findings on the pre-COVID-19 period (e.g., Wu et al., 2019) and the COVID-19
era (see, e.g., Ji et al., 2020). There is a significant return spillover from the USD and bond
markets to Gold Markets. This evidence can be related to the Fed's monetary policy at the start
of the pandemic in the United States in early 2020. When Fed decided to decrease policy
interest rates with the start of the COVID-19 in the United States, this decision decreased the
real value of the USD and Bond returns due to the inflation expectations. The decline in the
USD and Bond returns causes investors to invest in the Gold market. Therefore, we observe
that Gold is the dominant asset to alternate portfolio allocation. In addition, the VIX has
significant return spillovers to stock returns implying that uncertainty shocks decrease the stock
returns. Although there is a significant interaction between the EPU and the TEU indices, they
15
seem to have an insignificant role in determining the returns of Bitcoin and traditional financial
assets. Here, the only significant return connectedness observed between the VIX and the S&P
500 stock returns.
Regarding the volatility spillover analyses, Bitcoin is a net transmitter of volatility
spillovers to other markets, particularly during COVID-19. This evidence is consistent with the
findings in Goodell and Goutte (2021a). Also, there is a significant price volatility
connectedness between the VIX and the S&P 500 stocks. Further results from the causality-in-
variance LM and the Fourier LM tests indicate a unidirectional volatility transmission from
Bitcoin to other markets, including Gold, Stocks, Bonds, VIX, and Crude Oil. This evidence
shows that Bitcoin is the centre of volatility spillovers to financial markets. This evidence is in
line with previous papers (e.g., Goodell and Goutte, 2021b; Yi et al., 2021). Bitcoin price
volatility also increases the volatility of the VIX. Therefore, Bitcoin is the leading asset to
provide diversification benefits against volatility shocks (see Corbet et al., 2019 and the related
references therein).
Interestingly, the EPU is the only factor that causing higher volatility in Bitcoin. This
evidence means that the EPU shocks increase the price volatility of Bitcoin, and this transmits
to Gold, Stocks, Bonds, and Crude Oil markets. Other studies have shown the catalyser effect
of the EPU on the price volatility in Bitcoin (see, e.g., Fang et al., 2019; Wang et al., 2020;
Yen and Cheng, 2021), and our results are in line with the results of these papers.
4. Conclusion
This paper analyses the return connectedness and the price volatility connectedness between
Bitcoin, traditional financial assets (Crude Oil, Gold, Stocks, Bonds, and the USD), and major
global uncertainty measures (the EPU, the TEU, and the VIX) from April 29, 2013, to June 30,
2020. For this purpose, we utilise the time-varying parameter vector autoregression model,
16
dynamic connectedness approach, and network analyses. Furthermore, we applied the
causality-in-variance LM and the Fourier LM tests to detect the causality direction in-variance.
We observe that total spillover indices reached unprecedented levels during the COVID-19
period and remain high. Since then, we confirmed the high return and volatility spillovers
across markets during the COVID-19 pandemic. During the COVID-19 era, Bitcoin has been
a net receiver of return spillovers from other markets, and it has been a net transmitter of
volatility spillovers to other markets. However, Bitcoin has been weakly connected over the
full sample period. Therefore, we concluded that Bitcoin plays a major role in the volatility
transmission during the COVID-19 period.
On the other hand, the LM-type tests show a unidirectional risk transmission from
Bitcoin to other markets, including Gold, Stock market, Bonds, the VIX, and to some extent,
to the Crude Oil market. Simultaneously, the EPU is the only global factor that causing higher
volatility in Bitcoin. Therefore, we observed that the EPU is positively related to the price
volatility of Bitcoin. In short, we concluded that Bitcoin has a significant price volatility
transmission to traditional financial markets during the COVID-19 period, and its price
volatility has been driven by economic policy uncertainty. Policymakers should realise that
their decisions can significantly increase the price volatility of Bitcoin. Investors and traders
should also realise that the high price volatility in Bitcoin can transmit to traditional financial
markets (most significantly to stock and bond markets).
Future studies on this subject can focus on other cryptocurrencies to examine their
relationships between traditional financial markets. At this stage, future papers can include
emerging and global exchange rates. Another research plan is that to focus on new measures
of global uncertainty. For instance, we have limited knowledge of the effects of Twitter-based
uncertainty measures on the relationship between cryptocurrencies and traditional financial
markets. These research topics can be examined by utilising new econometric techniques.
17
References
Antonakakis, N., Gabauer, D., & Gupta, R. (2019). International Monetary Policy Spillovers:
Evidence from A Time-Varying Parameter Vector Autoregression. International
Review of Financial Analysis, 65, 101382.
Antonakakis, N., Gabauer, D., Gupta, R., & Plakandaras, V. (2018). Dynamic Connectedness
of Uncertainty Across Developed Economies: A Time-Varying Approach. Economics
Letters, 166, 63–75.
Ashraf, B. N. (2020). Stock markets' Reaction to COVID-19: Cases or Fatalities? Research in
International Business and Finance, 54, 101249.
Aysan, A. F., Demir, E., Gozgor, G., & Lau, C. K. M. (2019). Effects of the Geopolitical Risks
on Bitcoin Returns and Volatility. Research in International Business and Finance, 47,
511–518.
Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring Economic Policy Uncertainty. The
Quarterly Journal of Economics, 131 (4), 1593–1636.
Bouri, E., Gupta, R., & Roubaud, D. (2019). Herding Behaviour in Cryptocurrencies. Finance
Research Letters, 29, 216–221.
Bouri, E., Gupta, R., Tiwari, A. K., & Roubaud, D. (2017). Does Bitcoin Hedge Global
Uncertainty? Evidence from Wavelet-based Quantile-in-quantile Regressions. Finance
Research Letters, 23, 87–95.
Bredin, D., Conlon, T., & Potì, V. (2015). Does Gold Glitter in the Long-run? Gold as a Hedge
and Safe Haven across Time and Investment Horizon. International Review of
Financial Analysis, 41, 320–328.
Caporale, G. M., & Zekokh, T. (2019). Modelling Volatility of Cryptocurrencies Using
Markov-Switching GARCH Models. Research in International Business and Finance,
48, 143–155.
18
Caporale, G. M., Gil-Alana, L., & Plastun, A. (2018). Persistence in the Cryptocurrency
Market. Research in International Business and Finance, 46, 141–148.
Cheng, H. P., & Yen, K. C. (2020). The Relationship between the Economic Policy Uncertainty
and the Cryptocurrency Market. Finance Research Letters, 35, 101308.
Coindesk (2021). https://www.coindesk.com/price/bitcoin.
Colon, F., Kim, C., Kim, H., & Kim, W. (2021). The Effect of Political and Economic
Uncertainty on the Cryptocurrency Market. Finance Research Letters, 39, 101621.
Conlon, T., Corbet, S., & McGee, R. J. (2020). Are Cryptocurrencies a Safe Haven for Equity
Markets? An International Perspective from the COVID-19 Pandemic. Research in
International Business and Finance, 54, 101248.
Corbet, S., Lucey, B., & Yarovaya, L. (2018). Datestamping the Bitcoin and Ethereum
Bubbles. Finance Research Letters, 26, 81–88.
Corbet, S., Lucey, B., Urquhart, A., & Yarovaya, L. (2019). Cryptocurrencies as a Financial
Asset: A Systematic Analysis. International Review of Financial Analysis, 62, 182–
199.
Damianov, D., & Elsayed, A. (2020). Does Bitcoin Add Value to Global Industry Portfolios?
Economics Letters, 191, 108935.
Dastgir, S., Demir, E., Downing, G., Gozgor, G., & Lau, C. K. M. (2019). The Causal
Relationship between Bitcoin Attention and Bitcoin Returns: Evidence from the
Copula-based Granger Causality Test. Finance Research Letters, 28, 160–164.
Demir, E., Gozgor, G., Lau, C. K. M., & Vigne, S. A. (2018). Does Economic Policy
Uncertainty Predict the Bitcoin Returns? An Empirical Investigation. Finance Research
Letters, 26, 145–149.
19
Diebold, F. X., & Yilmaz, K. (2012). Better to Give Than to Receive: Predictive Directional
Measurement of Volatility Spillovers. International Journal of Forecasting, 28 (1), 57–
66.
Diebold, F. X., & Yılmaz, K. (2014). On The Network Topology of Variance Decompositions:
Measuring the Connectedness of Financial Firms. Journal of Econometrics, 182 (1),
119–134.
Elsayed, A. H., Gozgor, G., & Lau, C. K. M. (2021). Causality and Dynamic Spillovers among
Cryptocurrencies and Currency Markets. International Journal of Finance and
Economics, forthcoming, https://doi.org/10.1002/ijfe.2257
Fang, L., Bouri, E., Gupta, R., & Roubaud, D. (2019). Does Global Economic Uncertainty
Matter for the Volatility and Hedging Effectiveness of Bitcoin? International Review
of Financial Analysis, 61, 29–36.
Flavin, T. J., Morley, C. E., & Panopoulou, E. (2014). Identifying Safe Haven Assets for Equity
Investors through an Analysis of the Stability of Shock Transmission. Journal of
International Financial Markets, Institutions and Money, 33, 137–154.
Goodell, J. W. (2020). COVID-19 and Finance: Agendas for Future Research. Finance
Research Letters, 35, 101512.
Goodell, J. W., & Goutte, S. (2021a). Co-movement of COVID-19 and Bitcoin: Evidence from
Wavelet Coherence Analysis. Finance Research Letters, 38, 101625.
Goodell, J. W., & Goutte, S. (2021b). Diversifying Equity with Cryptocurrencies during
COVID-19. International Review of Financial Analysis, 76, 101781.
Gozgor, G., Tiwari, A. K., Demir, E., & Akron, S. (2019). The Relationship between Bitcoin
Returns and Trade Policy Uncertainty. Finance Research Letters, 29, 75–82.
Hafner, C. M., & Herwartz, H. (2006). A Lagrange Multiplier Test for Causality in
Variance. Economics Letters, 93 (1), 137–141.
20
Ji, Q., Bouri, E., Lau, C. K. M., & Roubaud, D. (2019). Dynamic Connectedness and
Integration in Cryptocurrency Markets. International Review of Financial Analysis, 63,
257–272.
Ji, Q., Zhang, D., & Zhao, Y. (2020). Searching for Safe-haven assets during the COVID-19
Pandemic. International Review of Financial Analysis, 71, 101526.
Jiang, Y., Wu, L., Tian, G., & Nie, H. (2021). Do Cryptocurrencies Hedge against EPU and
the Equity Market Volatility during COVID-19? – New Evidence from Quantile
Coherency Analysis. Journal of International Financial Markets, Institutions and
Money, 72, 101324.
Katsiampa, P. (2019). An Empirical Investigation of Volatility Dynamics in the
Cryptocurrency Market. Research in International Business and Finance, 50, 322–335.
Koop, G., & Korobilis, D. (2014). A New Index of Financial Conditions. European Economic
Review, 71, 101–116.
Koop, G., Pesaran, M. H., & Potter, S. M. (1996). Impulse Response Analysis in Nonlinear
Multivariate Models. Journal of Econometrics, 74 (1), 119–147.
Korobilis, D., & Yilmaz, K. (2018). Measuring Dynamic Connectedness with Large Bayesian
VAR Models. Available at SSSRN, No. 3099725.
Li, J., & Enders, W. (2018). Flexible Fourier form for Volatility Breaks. Studies in Nonlinear
Dynamics and Econometrics, 22 (1), 20160039.
Liu, W., Semeyutin, A., Lau, C. K. M., & Gozgor, G. (2020). Forecasting Value-at-Risk of
Cryptocurrencies with RiskMetrics Type Models. Research in International Business
and Finance, 54, 101259.
Pesaran, M. H. & Shin, Y. (1998). Generalised Impulse Response Analysis in Linear
Multivariate Models. Economics Letters, 58 (1), 17–29.
21
Ranaldo, A., & Söderlind, P. (2010). Safe Haven Currencies. Review of Finance, 14 (3), 385–
407.
Shi, Y., Tiwari, A. K., Gozgor, G., & Lu, Z. (2020). Correlations among Cryptocurrencies:
Evidence from Multivariate Factor Stochastic Volatility Model. Research in
International Business and Finance, 53, 101231.
So, M. K., Chu, A. M., & Chan, T. W. (2021). Impacts of the COVID-19 Pandemic on
Financial Market Connectedness. Finance Research Letters, 38, 101864.
Tiwari, A. K., Jana, R. K., Das, D., & Roubaud, D. (2018). Informational Efficiency of
Bitcoin—An Extension. Economics Letters, 163, 106–109.
Urquhart, A. (2016). The Inefficiency of Bitcoin. Economics Letters, 148, 80–82.
Urquhart, A. (2018). What Causes the Attention of Bitcoin? Economics Letters, 166, 40–44.
Urquhart, A., & Zhang, H. (2019). Is Bitcoin a Hedge or Safe Haven for Currencies? An
Intraday Analysis. International Review of Financial Analysis, 63, 49–57.
Wang, P., Li, X., Shen, D., & Zhang, W. (2020). How Does Economic Policy Uncertainty
Affect the Bitcoin Market? Research in International Business and Finance, 53,
101234.
Wu, S., Tong, M., Yang, Z., & Derbali, A. (2019). Does Gold or Bitcoin Hedge Economic
Policy Uncertainty? Finance Research Letters, 31, 171–178.
Wu, W., Tiwari, A. K., Gozgor, G., & Leping, H. (2021). Does Economic Policy Uncertainty
Affect Cryptocurrency Markets? Evidence from Twitter-Based Uncertainty Measures.
Research in International Business and Finance, 58, 101478.
Yen, K.C., & Cheng, H.P. (2021). Economic Policy Uncertainty and Cryptocurrency Volatility.
Finance Research Letters, 38, 101428.
22
Yi, S., Xu, Z., & Wang, G.J. (2018). Volatility Connectedness in the Cryptocurrency Market:
Is Bitcoin a Dominant Cryptocurrency? International Review of Financial Analysis, 60,
98–114.
23
0
5,000
10,000
15,000
20,000
2013 2014 2015 2016 2017 2018 2019 2020
Bitcoin
2,000
3,000
4,000
5,000
6,000
7,000
2013 2014 2015 2016 2017 2018 2019 2020
Stock
320
360
400
440
480
520
2013 2014 2015 2016 2017 2018 2019 2020
Bond
90
100
110
120
130
2013 2014 2015 2016 2017 2018 2019 2020
U.S. Dollar
500
600
700
800
900
1,000
2013 2014 2015 2016 2017 2018 2019 2020
Gold
-40
0
40
80
120
2013 2014 2015 2016 2017 2018 2019 2020
WTI_Oil
0
20
40
60
80
100
2013 2014 2015 2016 2017 2018 2019 2020
VIX
0
200
400
600
800
1,000
2013 2014 2015 2016 2017 2018 2019 2020
US_EPU
0
1,000
2,000
3,000
2013 2014 2015 2016 2017 2018 2019 2020
TEU
Figure 1. Daily Time Evolution of Bitcoin, Traditional Financial Assets and Major Global Uncertainty Measures
Notes: These figures portrait the variation of the daily data series for the Bitcoin price, the S&P 500 stock index, the S&P bond index, the USD Broad Exchange Rate,
Gold price, the WTI Crude Oil price (WTI_Oil), the VIX, the US economic policy uncertainty index (US_EPU), and the Twitter-based economic uncertainty index
(TEU) from April 29, 2013, to June 30, 2020.
24
-.6
-.4
-.2
.0
.2
.4
.6
2013 2014 2015 2016 2017 2018 2019 2020
Bitcoin
-.15
-.10
-.05
.00
.05
.10
2013 2014 2015 2016 2017 2018 2019 2020
Stock
-.03
-.02
-.01
.00
.01
.02
.03
2013 2014 2015 2016 2017 2018 2019 2020
Bond
-.03
-.02
-.01
.00
.01
.02
2013 2014 2015 2016 2017 2018 2019 2020
U.S. Doll ar
-.08
-.04
.00
.04
.08
2013 2014 2015 2016 2017 2018 2019 2020
Gold
-.4
-.2
.0
.2
.4
.6
2013 2014 2015 2016 2017 2018 2019 2020
WTI_Oil
2.0
2.5
3.0
3.5
4.0
4.5
2013 2014 2015 2016 2017 2018 2019 2020
VIX
1
2
3
4
5
6
7
2013 2014 2015 2016 2017 2018 2019 2020
US_EPU
2
4
6
8
10
2013 2014 2015 2016 2017 2018 2019 2020
TEU
Figure 2. Daily Returns of Bitcoin, Traditional Financial Assets and Major Global Uncertainty Measures
Notes: These figures portrait the returns of the daily data series for the Bitcoin price, the S&P 500 stock index, the S&P bond index, the USD Broad Exchange Rate,
Gold price, the WTI Crude Oil price (WTI_Oil), the VIX, the US economic policy uncertainty index (US_EPU), and the Twitter-based economic uncertainty index
(TEU) from April 29, 2013, to June 30, 2020.
25
Figure 3. Directional Connectedness Network of Pairwise Return Spillovers
Notes: This diagram shows the average pairwise directional return spillovers among all possible pairs of variables in the model. A node's colour implies whether a variable is
a net transmitter/receiver of return spillovers. The red colour indicates a net transmitter, while the green colour shows a net receiver, respectively. Furthermore, the thickness
and arrows' colour represents the average return spillover magnitude and strength between each pair. Red indicates strong, navy shows moderate, and green refers to weak
return spillovers.
26
15
20
25
30
35
40
45
50
55
2013 2014 2015 2016 2017 2018 2019 2020
Figure 4. Total Return Spillover Index
Notes: This graph portrait the time-varying behaviour of the total connectedness index across the Bitcoin, financial markets and major global uncertainty measures. It is based
on the generalised forecast-error variance decomposition (GFEVD) obtained from estimating a TVP-VAR model of order one and 10-step ahead forecasts from April 29, 2013,
to June 30, 2020.
27
-3
-2
-1
0
1
2
2013 2014 2015 2016 2017 2018 2019 2020
NET BITCOIN
-2
0
2
4
6
8
2013 2014 2015 2016 2017 2018 2019 2020
NET STOCK
-3
-2
-1
0
1
2
2013 2014 2015 2016 2017 2018 2019 2020
NET BOND
-4
-3
-2
-1
0
1
2
2013 2014 2015 2016 2017 2018 2019 2020
NET DOLLAR
-6
-4
-2
0
2
2013 2014 2015 2016 2017 2018 2019 2020
NET GOLD
-2
-1
0
1
2
2013 2014 2015 2016 2017 2018 2019 2020
NET OIL
-2
0
2
4
6
8
2013 2014 2015 2016 2017 2018 2019 2020
NET VIX
-6
-4
-2
0
2
4
2013 2014 2015 2016 2017 2018 2019 2020
NET US_EPU
-8
-6
-4
-2
0
2
4
2013 2014 2015 2016 2017 2018 2019 2020
NET TEU
Figure 5. Dynamic Net Directional Return Spillover Indices
Notes: This graph represents the net directional return spillovers time-varying behaviour for each variable under consideration. Positive (negative) values indicate that the
variable is a net transmitter (receiver) of return spillover to (from) all other variables. Indices are estimated based on the generalised forecast-error variance decomposition
(GFEVD) obtained from estimating a TVP-VAR model of order one and 10-step ahead forecasts.
28
-.3
-.2
-.1
.0
.1
.2
.3
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-TEU
-.6
-.4
-.2
.0
.2
.4
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-US_EPU
-2.0
-1.5
-1.0
-0.5
0.0
0.5
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-VIX
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-BOND
-1.00
-0.75
-0.50
-0.25
0.00
0.25
0.50
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-U.S. DOLLAR
-0.4
0.0
0.4
0.8
1.2
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-GOLD
-.6
-.4
-.2
.0
.2
.4
.6
.8
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-OIL
-.8
-.6
-.4
-.2
.0
.2
.4
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-STOCK
Figure 6. Dynamic Connectedness of Pairwise Return Spillovers
Notes: This graph represents the pairwise directional return spillovers time-varying behaviour between Bitcoin and each variable under consideration. Positive (negative) values
indicate that Bitcoin is transmitting (receiving) return spillover to (from) the other variable. Indices are estimated based on the generalised forecast-error variance decomposition
(GFEVD) obtained from estimating a TVP-VAR model of order one and 10-step ahead forecasts.
29
Figure 7. Directional Connectedness Network of Pairwise Volatility Spillovers
Notes: This diagram shows the average pairwise directional volatility spillovers among all possible pairs of variables in the model. A node's colour implies whether a variable
is a net transmitter/receiver of volatility spillovers. The red colour indicates a net transmitter, while the green colour shows a net receiver, respectively. Furthermore, the
thickness and arrows' colour represents each pair's average volatility spillovers magnitude and strength. Red indicates strong, navy shows moderate, and green refers to weak
volatility spillovers.
30
10
20
30
40
50
60
70
80
90
100
2013 2014 2015 2016 2017 2018 2019 2020
Figure 8. Total Volatility Spillover Index
Notes: This graph shows the time-varying behaviour of the total volatility connectedness across the Bitcoin, financial markets and major global uncertainty measures. It is based
on the generalised forecast-error variance decomposition (GFEVD) obtained from estimating a TVP-VAR model of order one and 10-step ahead forecasts from April 29, 2013,
to June 30, 2020.
31
-8
-4
0
4
8
2013 2014 2015 2016 2017 2018 2019 2020
NET BITCOIN
-4
0
4
8
12
16
2013 2014 2015 2016 2017 2018 2019 2020
NET STOCK
-8
-4
0
4
8
2013 2014 2015 2016 2017 2018 2019 2020
NET BOND
-8
-4
0
4
8
2013 2014 2015 2016 2017 2018 2019 2020
NET DOLLAR
-8
-4
0
4
8
2013 2014 2015 2016 2017 2018 2019 2020
NET GOLD
-8
-4
0
4
8
12
2013 2014 2015 2016 2017 2018 2019 2020
NET OIL
-10
-5
0
5
10
15
2013 2014 2015 2016 2017 2018 2019 2020
NET VIX
-12
-8
-4
0
4
2013 2014 2015 2016 2017 2018 2019 2020
NET US_EP U
-12
-8
-4
0
4
2013 2014 2015 2016 2017 2018 2019 2020
NET TEU
Figure 9. Dynamic Net Directional Volatility Spillover Indices
Notes: This graph displays the net directional volatility spillovers time-varying behaviour for each variable under consideration. Positive (negative) values indicate that the
variable is a net transmitter (receiver) of volatility spillover to (from) all other variables. Indices are estimated based on the generalised forecast-error variance decomposition
(GFEVD) obtained from estimating a TVP-VAR model of order one and 10-step ahead forecasts.
32
-1
0
1
2
3
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-TEU
-2
-1
0
1
2
3
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-US_EPU
-2
-1
0
1
2
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-VIX
-1.5
-1.0
-0.5
0.0
0.5
1.0
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-BOND
-1.5
-1.0
-0.5
0.0
0.5
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-DOLLAR
-2
-1
0
1
2
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-GOLD
-1.6
-1.2
-0.8
-0.4
0.0
0.4
0.8
1.2
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-OIL
-1.6
-1.2
-0.8
-0.4
0.0
0.4
0.8
1.2
2013 2014 2015 2016 2017 2018 2019 2020
BITCOIN-STOCK
Figure 10. Dynamic Connectedness of Pairwise Volatility Spillovers
Notes: These graphs portrait the time-varying behaviour of the pairwise directional volatility spillover between Bitcoin and each of the variables under consideration. Positive
(negative) values indicate that Bitcoin is transmitting (receiving) volatility spillover to (from) the other variable. Indices are estimated based on the generalised forecast-error
variance decomposition (GFEVD) obtained from estimating a TVP-VAR model of order one and 10-step ahead forecasts.
33
Table 1
A Summary of Descriptive Statistics
Bitcoin
Stock
Bond
U.S. Dollar
Gold
WTI-Oil
VIX
US_EPU
TEU
Mean
0.0022
0.0004
0.0002
0.0001
0.0001
-0.0001
2.721
4.381
4.831
Std. Dev.
0.0505
0.0108
0.0028
0.0031
0.0094
0.0324
0.3156
0.6125
0.7675
Max.
0.5207
0.0897
0.0206
0.0191
0.0560
0.4258
4.4151
6.6941
7.9791
Min.
-0.4647
-0.1276
-0.0284
-0.0208
-0.0660
-0.2813
2.2126
1.1999
2.3371
Skewness
-0.035
-1.085
-1.506
0.188
0.086
1.426
1.697
0.528
0.319
Kurtosis
13.494
24.867
19.368
5.009
5.092
41.738
4.297
1.661
0.06
J-B
14181***
48519***
29919***
1965***
2021***
136298***
2335***
301***
32***
ADF
-44.86***
-13.31***
-19.53***
-41.72***
-45.21***
-16.05***
-5.31***
-5.09***
-3.93***
PP
-45.25***
-51.48***
-36.77***
-41.73***
-45.17***
-47.41***
-4.86***
-32.95***
-16.83***
Q(20)
23.987***
258.570***
141.036***
14.952
16.518*
112.827***
13210.218***
4750.260***
11276.557***
Q2(20)
1.764
510.707***
1001.681***
54.830***
83.718***
204.285***
13370.530***
8453.823***
11993.894***
ARCH(20)
56.032***
427.549***
256.992***
107.417***
106.676***
461.034***
82.36***
331.923***
413.959***
Notes: This table reports descriptive statistics for the daily data series. Daily Bitcoin, financial markets, USD, Gold and Crude
Oil returns are calculated as the first logarithmic difference between two consecutive observations. Following the usual
practice, the VIX, the U.S. economic uncertainty index, and Twitter-based economic uncertainty index are calculated in log
form. All data series are collected from Thomson Reuters DataStream except for the Twitter-based economic uncertainty
index collected from the Economic Policy Uncertainty website (https://www.policyuncertainty.com). J-B is the Jarque–Bera
test for Normality. ADF and PP denote the empirical statistics of the Augmented Dickey-Fuller and Phillips-Perron unit root
tests, respectively. Q(20) and Q2(20) are the Ljung–Box statistics for serial correlation in raw series and squared residuals.
ARCH (20) testing Engle's ARCH effects up to 20 lags. Finally, ***, **, * indicate significance at 1%, 5%, and 10% levels.
34
Table 2
Returns Connectedness
Bitcoin
Stock
Bond
U.S. $
Gold
WTI_Oil
VIX
US_EPU
TEU
FROM
Bitcoin
97.612
1.036
0.275
0.256
0.168
0.161
0.378
0.09
0.023
2.388
Stock
0.643
59.691
0.451
3.736
0.162
2.597
32.231
0.183
0.306
40.309
Bond
0.182
4.726
80.315
3.658
6.723
0.573
3.488
0.124
0.209
19.685
U.S. $
0.072
5.73
4.303
73.703
10.36
2.719
2.722
0.114
0.279
26.297
Gold
0.193
0.221
7.417
11.155
79.471
0.261
0.961
0.124
0.197
20.529
WTI_Oil
0.142
3.981
0.246
3.072
0.351
88.9
2.792
0.259
0.258
11.1
VIX
0.502
31.285
0.757
2.047
0.795
1.647
62.582
0.027
0.358
37.418
US_EPU
0.026
2.564
0.194
0.157
0.147
0.229
5.685
80.341
10.656
19.659
TEU
0.051
2.435
0.483
0.197
0.184
0.142
4.973
7.084
84.451
15.549
TO others
1.811
51.978
14.126
24.279
18.89
8.33
53.229
8.005
12.286
TCI =
21.44%
Net spillovers
-0.577
11.669
-5.559
-2.019
-1.639
-2.771
15.811
-11.654
-3.263
Notes: This table summarises the empirical results of the total, directional and pairwise return spillovers between
Bitcoin, financial markets and global uncertainty measures. These results are based on the generalised forecast
error variance decomposition (GFEVD) obtained from a TVP-VAR model of order one and 10-step ahead
forecasts. The lag length is selected by the Bayesian information criterion (BIC). 'TO others' signifies directional
spillovers correspond to the off-diagonal column sums, i.e., spillovers from variable i to all variables j. 'FROM'
represents the off-diagonal row sums of directional spillovers, i.e., spillovers from all variables j to the variable
i. Net spillovers are simply the "TO others" minus "FROM others". Finally, TCI, the total spillover index,
demonstrates that 21.4% of the forecast error variance comes from spillovers.
35
Table 3
Volatility Connectedness
Bitcoin
Stock
Bond
U.S. $
Gold
WTI_Oil
VIX
US_EPU
TEU
FROM
Bitcoin
86.155
7.348
1.535
0.582
0.472
0.032
3.668
0.173
0.033
13.845
Stock
5.18
62.561
0.875
1.874
0.077
2.562
26.656
0.117
0.098
37.439
Bond
4.172
48.176
25.072
5.31
0.137
3.867
13.096
0.144
0.026
74.928
U.S. $
1.542
19.666
6.249
63.482
1.795
1.837
5.014
0.381
0.033
36.518
Gold
1.671
14.936
0.26
4.56
70.939
0.621
5.966
1.001
0.044
29.061
WTI_Oil
1.32
28.19
12.262
2.72
0.031
46.921
8.48
0.003
0.073
53.079
VIX
1.275
22.442
0.299
0.957
0.496
1.234
73.023
0.047
0.226
26.977
US_EPU
0.114
1.88
0.144
0.252
0.463
0.357
4.925
91.404
0.46
8.596
TEU
0.121
0.407
0.171
0.079
0.06
0.191
3.039
1
94.932
5.068
TO others
15.396
143.047
21.795
16.335
3.532
10.702
70.845
2.867
0.994
TCI =
31.72%
Net spillovers
1.551
105.607
-53.134
-20.184
-25.529
-42.377
43.868
-5.729
-4.074
Notes: This Table summarises the empirical results of the total, directional and pairwise volatility spillovers
between Bitcoin, financial markets and global uncertainty measures. These results are based on the generalised
forecast error variance decomposition (GFEVD) obtained from a TVP-VAR model of order one and 10-step
ahead forecasts. The lag length is selected by the Bayesian information criterion (BIC). 'TO others' signifies
directional spillovers correspond to the off-diagonal column sums, i.e., spillovers from variable i to all variables
j. 'FROM' represents the off-diagonal row sums of directional spillovers, i.e., spillovers from all variables j to the
variable i. Net spillovers are simply the "TO others" minus "FROM others". Finally, TCI, the total spillover index,
demonstrates that 31.7% of the forecast error variance comes from spillovers.
36
Table 4
Causality in-variance Tests
p-value
n
p-value
p-value
n
p-value
Stocks
8.715**
0.012
3
8.974**
0.011
3.232
0.198
3
1.813
0.403
Bonds
8.444**
0.014
3
13.287***
0.001
2.771
0.250
3
2.048
0.358
USD
1.897
0.387
3
2.784
0.248
2.569
0.276
3
0.648
0.723
Gold
8.291**
0.015
3
10.771***
0.004
1.398
0.497
3
1.384
0.500
WTI_Oil
5.478*
0.064
2
7.950**
0.018
1.699
0.427
3
1.035
0.595
VIX
7.399**
0.024
1
8.838**
0.012
0.668
0.715
3
0.458
0.795
US_EPU
0.671
0.714
0
0.671
0.714
11.205***
0.003
3
9.967***
0.006
TEU
1.011
0.603
3
1.534
0.464
0.951
0.621
3
0.735
0.692
Notes: This table is based on the Lagrange multiplier (LM) volatility spillover test developed by Hafner and Herwartz
(2006) and the Fourier LM test introduced by Li and Enders (2018). is the statistic from a Lagrange multiplier
(LM) test for testing the null hypothesis of no-volatility spillover from asset j to asset i whereas denotes the
test statistic from a Fourier LM test that accounts for structural breaks. The maximum number of Fourier frequency
(n) is set to 3, where the optimal frequency is determined by Akaike Information Criterion (AIC). Finally, ***, **,
and * denotes 1%, 5%, and 10% statistical significance levels, respectively.
