Literature Review 3

M PRA
Munich Personal RePEc Archive
What Does Crypto-currency Look Like?

Gaining Insight into Bitcoin Phenomenon
Jamal Bouoiyour and Refk Selmi
CATT, University of Pau, France, ESC, University of Manouba,

Tunisia.
26. August 2014
Online at http://mpra.ub.uni-muenchen.de/58133/
MPRA Paper No. 58133, posted 2. September 2014 09:55 UTC
WHAT DOES CRYPTO-CURRENCY LOOK LIKE?
GAINING INSIGHT INTO BITCOIN PHENOMENON
JAMAL BOUOIYOUR
CATT, University of Pau, France.
E-mail: [email protected]
REFK SELMI
ESC, University of Manouba, Tunisia.
E-mail: [email protected]
Abstract: The present paper seeks to effectively address the following question: What
Bitcoin looks like? To do so, we regress Bitcoin price on a number of variables (Bitcoin
fundamentals recorded in the literature) by applying an ARDL Bounds Testing approach for
daily data covering the period from December 2010 to June 2014. Our findings highlight the
speculative nature of Bitcoin. We also provide insightful evidence that Bitcoin may be used
for economic reasons but there is any sign of being a safe haven or a long-term promise. By
considering the Chinese trading bankruptcy, the contribution of users interest stills sharply
dominant, highlighting the robustness of our results.
Keywords: Bitcoin; ARDL Bounds Testing method; innovative accounting approach; VEC
Granger causality test.
1
1. Introduction
Since its creation in 2009 by Satoshi Nakamoto-pseudonym, the Bitcoin has
experienced multiple peaks and successive ups and downs. Is it a safe haven or a speculative
trap? Is it a short-term hedge? Is it a poor long-term investment or a long-run promise? The
opinions about this nascent currency have drawn a substantial attention from investors,
advisers and market regulators. The fact that questions get frequently and heavily asked
indicates the very prime importance of this phenomenon.
Bitcoin is virtual money with zero intrinsic value issued by computer code in
electronic portfolios, which is not convertible into anything and not have the backing of any
Central Banks and any government. The value of a Bitcoin is neither a convertible tangible
asset (such as gold) nor a fiat currency (such as dollar). It is determined by the interplay of
supply and demand. This nascent crypto-currency fulfills various functions. It facilitates
business transactions from person to person worldwide without any intermediary, reduces
trade barriers and increases the productivity. Nevertheless, Bitcoin remains far from certain
because of its sizeable price volatility, the inelastic money supply coded by mathematic
formula and the lack of legal security. Bitcoin is a digital currency in a nascent stage closely
associated to multiple risks stemming from its extra volatility and its speculative nature.
Despite its sharp popularity, there still very few works analyzing Bitcoin phenomenon.
These researches seem insufficient to appropriately address the huge amount of questions
around it. For instance, the study of Kristoufek (2013) focuses only on assessing whether
Bitcoin is a speculative bubble by exploring the link between Bitcoin and users interest. In
addition, Glaser et al. (2014) have attempted to evaluate if Bitcoin is an asset or a currency.
Besides, Kristoufek (2014) has tried to investigate whether Bitcoin is more driven by
technical, financial or speculative factors by applying coherence wavelet. This technique
allows it to consider the interconnection between each two variables without considering the
possible interaction with other time series. In other words, this analysis is incomplete and may
lead to biased results. More accurately, wavelet coherence may not be considered usually as
perfect technique. On the one hand, it may lead to confuse outcomes since the occurrence of
noise cannot be heavily neglected, disrupting then the studied relationship (Ng and Chan,
2012). On the other hand, wavelet decomposition is generally applied to assess the periodicity
and the multiple signals that happen over time. Moreover, when we consider only two
variables in wavelet analysis, we generally fall on the problem of simple regression without
control variables. This highlights the inability of this technique to capture proper and accurate
outcomes since it may distort the estimate. In that context, Aguiar-Conraria and Soares (2011)
argue that the findings change intensely when we move from wavelet investigation with two
variables for conditional wavelet estimation (with more than two variables or by adding other
explanatory time series). This implies that the use of large-scale parameters of each two
variables as the case of Kristoufek (2014)s study may prompt inconclusive results in terms of
the interaction dynamic between Bitcoin price and its main drivers. This reinforces the need to
take into account the control variables to confirm the obtained findings.
Due to the complexity of this new digital currency, the Bitcoin phenomenon demands
a deeper investigation. Hence, the present paper attempts to address several questions in order
to elucidate readers information about Bitcoin: What this crypto-currency looks like? Is it a
2
safe haven or a speculative trap? Is it a business income? Is it a short-term hedge? Is it good
idea to invest in Bitcoin? Is it a long term promise?
To find better paths, our contribution to this debate is to check the robustness of the
previous results and to answer further questions by adding additional explanatory variables
and by carrying out convenient method that considers the interaction dynamic between
several variables and captures the shocks of own series with others. To this end, we regress
Bitcoin price on investors attractiveness, exchange-trade volume, monetary Bitcoin velocity,
estimated output volume, hash rate, gold price and Shangai market index. We apply an ARDL
Bounds Testing approach, innovation accounting by simulating variance decomposition and
impulse response function and VEC Granger causality test for daily data for the period
spanning between December 2010 and June 2014.
We show interesting outcomes: In the short-run, the investors attractiveness, the exchange-
trade ratio, the estimated output volume and the Shangai index affect positively and
significantly the Bitctoin price, while the monetary velocity, the hash rate and the gold price
have no influence. In the long-run, the speculative nature of Bitcoin, the output volume and
the Chinese stock market index have no significant effect on Bitcoin price, while the hash rate
explains significantly the dynamic of this new virtual currency. The influence of exchange-
trade ratio becomes less strong, whereas the effects of the monetary velocity and the gold
price still insignificant in the long term. These findings appear solid and unambiguous since
there is a very slight change when incorporating a dummy variable relative to the bankruptcy
of Chinese trading company. The inclusion of additional variables which have no great
influence on Bitcoin price development (oil price, Dow Jones index and a dummy variable
denoting the closing of Road Silk by FBI) has led to unstable estimates. Beyond the nuances
of short and long terms, this research confirms the speculative nature of Bitcoin and its partial
usefulness in economic reasons without forgetting the utmost importance of accounting for
Chinese stock market and the processing power of Bitcoin network when analyzing the
Bitcoin price dynamic. This new digital money seems far from being a safe haven and a long-
term promise.
The remainder of the article proceeds as follows: Section 2 presents a brief literature
survey. Section 3 describes our data and presents our methodological framework. Section 4
reports our main results and discusses them. Section 5 focuses on robustness check. Section 6
concludes and offers policy implications that may be fruitful for investors and regulators.
2. Brief literature survey

Bitcoin has engaged the attention of Medias and researchers, acknowledging the
complexity of this new digital currency. Some researchers considered Bitcoin as financial
instrument rather than currency or payment system. Others called it evil since it is not
controlled nor by central banks nor by governments. Some economists defined it as a
speculative trap because of its extreme volatile behavior (Buchholz et al. (2012), Kristoufek
(2013, 2014), Bouoiyour et al. (2014) and Ciaian et al. (2014)). Others showed that with the
absence of hedging instruments able to appropriately prevent Bitcoin volatility, this digital
money can behave as a speculative trap (Yermack, 2014). Consistently, Glouderman (2014)
3
argue that economists scoffed at Bitcoin as more of a financial experiment than a legitimate
payment system. Some economists denounced it as evil, because its value is not backed by
any government nor can it be used to make pretty things as can gold. Others show that with no
intrinsic value, Bitcoins rising price constituted a speculative bubble.
The study of Kristoufek (2014) attempts to determine whether Bitcoin is likely to be
safe haven, speculative bubble or transactions tool by analyzing the potential sources of
Bitcoin price fluctuations including supply-demand fundamentals, speculative and technical
drivers. Wavelet coherence has been carried out to investigate properly and effectively the
evolution of correlations between the considered variables at different time frequencies. The
obtained results reveal that the fundamental factors such as exchange-trade ratio play
substantial roles in the long-run (short frequencies). The Chinese index seems an important
source of Bitcoin price evolution, while the contribution of gold price dynamic appears minor
and sometimes unclear. He finds also that Bitcoin prices are mainly influenced by investors
interest and thus by the speculative behaviors of businesses. This interconnection is most
dominant at lower frequencies (higher time scale). Intuitively, the findings reveal that during
the explosive prices period, the investors attractiveness to this nascent currency drives this
currency price up, while it drives it down during rapid declines period.
Glaser et al. (2014) have tried to address what intentions are businesses and investors
following when moving their currencys usage from domestic ones into a crypto-currency like
Bitcoin. By applying an Autoregressive Conditional Heteroskedasticity model, they show that
the motivation of investors to Bitcoin and their intention to gather proper and additional
information about its development has a great effect on this crypto-currency exchange
volume, while the nexus between Bitcoin and users interest seems insignificant when
considering the volume within the Bitcoin system. These observed outcomes may be owing to
the fact that exchange users prefer usually to keep their Bitcoins in their exchange wallet to
avoid speculation and cyber-attacks without any intention to use them in economic reasons
(trade transactions, for example).
Bouoiyour et al. (2014) attempt to appropriately address whether Bitcoin is a business
income or risky investment. They use Granger causality to assess the relationship between
Bitcoin price and exchange-trade ratio to answer the first question and the link between
Bitcoin price and investors attractiveness to address the second one. These tests have been
carried out within a frequency domain framework (unconditional versus conditional causality)
by applying a Breitung and Candelons (2006) approach. Their results reveal that Bitcoin
price Granger-causes exchange-trade ratio in the medium- and long-run. Besides, the
investors attractiveness Granger-cause Bitcoin price in the short term. These relationships
change substantially when considering the Shangai index and the hash rate (i.e. conditional
causality), highlighting therefore the complexity of evaluating what exactly Bitcoin looks like.
To sum, the focal studied links seem bidirectional and cyclical. These cycles can be short,
medium or long depending to directional causality in question. Their research provides
insightful evidence by confirming the extremely speculative nature of Bitcoin without
neglecting its great usefulness in economic reasons. The conditional causality through the
consideration of the Shangai index and the hash rate appears valuable since it has succeeded
to reach solid findings connecting further Bitcoin to the speculative behavior of investors.
4
3. Data and methodology
The existing literature on Bitcoin price suggests different factors that may play
important roles in explaining its evolution including the Bitcoin attractiveness of investors,
the global macroeconomic and financial indicators and the technical drivers. To measure the
users attractiveness to Bitcoin, we follow Kristoufek (2013) by using daily Bitcoin views
from wikipedia as it allows us to capture the speculative behavior of investors. In order to
detect Bitcoin economy, we use two respective indicators which are exchange-trade ratio, the
monetary Bitcoins velocity determined by the Bitcoin days destroyed for given transactions
and the estimated output volume. Technical drivers have been also considered to explain the
dynamic of Bitcoin measured through the hash rate available at Blockchain. We consider also
the global macroeconomic and financial indicators following the studies of Ciaian et al.
(2014) and Kristoufek (2014) including the gold price and the Chinese or Shangai stock
market index. Before beginning our analysis, it seems highly important to give some details
about these considered variables:
- The Bitcoin price (BPI): As stated previously, the Bitcoin is new digital money that has
recently attracted Medias and a wide range of people. It is an alternative currency to the fiat
currencies including dollar, euro and yen, with several advantages like lower transactions fees
and transparent information about the trade transactions. It has also some drawbacks where
the most damageable are the lack of legal security, the extra volatility and the speculation
(Kristoufek, 2014).
- The investors attractiveness (TTR): To effectively determine the investors attractiveness to
Bitcoin, we can use daily Bitcoin views from Google 1 as it able to depict properly the
speculative character of Bitcoin users (Kristoufek, 2013). Likewise, Bouoiyour et al. (2014)
have chosen to use the number of times a key word search term in relation to this famous
crypto-currency is entered into the Google engine.
- The exchange-trade ratio (ETR): The trade transactions and exchanges expand the utility of
holding the currency that may prompt an increase in Bitcoin price. The exchange-trade ratio is
measured as a ratio between volumes on the currency exchange market and trade. It can be
considered as measure of transactions (Kristoufek, 2014), or to address whether Bitcoin is
business income (Bouoiyour et al. 2014).
- The monetary Bitcoin velocity (MBV): By definition, the velocity of money is
the frequency at which one unit of each currency is used to purchase tradable or non-tradable
products for a given period. Because of the sharply large daily fluctuations of Bitcoin, the
velocity of the economy of this new crypto-currency has stayed relatively stable.
- The estimated output volume (EOV): Basically, there is a negative relationship between the
estimated output volume and Bitcoin price, i.e. an increase in output volume leads to a drop in
Bitcoin price especially in the long-run (Kristoufek, 2014).
- The Hash rate (HASH): The emergence of the famous virtual money has provided new
approaches concerning Bitcoin payments. Hence, some new words have emerged such as the
hash rate. It may be considered as an indicator or measure of the processing power of the
1
The views from Google used here as indicator of users interest is determined via the frequency of the online
Google search queries related to new digital money generally and Bitcoin particularly. Piskorec et al. (2014)
highlight the great usefulness of this proxy to accurately describe the behavior of Bitcoin investors.
5
Bitcoin network. For security goal, Bitcoin network must make intensive mathematical
operations, leading to an increase in the hash rate itself heavily connected with an increase in
cost demands for hardware. This may affect widely Bitcoin purchasers and thus expands the
demand of this new currency and in turn their prices. Theoretically, the hash rate is associated
positively to Bitcoin price (Bouoiyour et al. 2014).
- The gold price (GP): Bitcoin does not have an underlying value derived from consumption
or production process such as the precious metals including gold. Arguably, Ciaian et al.
(2014) put in evidence that there is any sign of Bitcoin being a safe haven.
- The Chinese market index (SI): The Chinese market index is considered as the biggest player
in Bitcoin economy and then it may be a potential source of Bitcoin price volatility.
Kristoufek (2014) takes an important example that may confirm this evidence, which is the
development around Baidu that may be considered as a potential determinant of the Chinese
online shopping. The announcement that Baidu is accepting Bitcoin has influenced
substantially the price dynamic of this virtual currency. Arguably, Bouoiyour et al. (2014)
provides insightful evidence that Bitcoin is likely to be a speculative trap rather than business
income, but this is conditioning upon the performance of Chinese market.
During the period between 05/12/2010 and 14/06/2014, this study disentangles the
existence of long-run cointegration between the above mentioned variables by considering a
dummy variables denoting the bankruptcy of Chinese trading company (it amounts 1 from
02/2013 and 0 otherwise). All these data are extracted from Blockchain2 and quandl3. To
improve the precision power of results, we carry out two log-linear specifications. For the first
equation, we regress BPI on TTR, ETR, MBV, EOV, HASH, GP and SI. Likewise, the second
one explains Bitcoin price dynamic in function of the same determinants (economic, technical
and financial) by incorporating a dummy variable (DV) denoting the Chinese trading
bankruptcy in order to check the robustness of our results.
LBPI t a 0 a1 LTTRt 2 LETR a3 LMBVt a4 LEOVt a5 LHASH t 6 LGPt 7 LSI t t (1)

LBPI t 01LTTRt 2 LETR 3 LMBVt 4 LEOVt 5 LHASH t 6 LGPt 7 LSI t 8 DV t (2)
Where , are the error terms with normal distribution, zero mean and finite variance. The
letter L preceding the variable names indicates Log. Kristoufek (2013, 2014) and Bouoiyour
et al. (2014) assume that an increased users interest searching for information about Bitcoin
leads to an increase in Bitcoin prices. Then, we expect a1, 1 0 . The exchange-trade ratio
denotes the ratio between volumes on the currency exchange market and trade. Theoretically,
the price of the currency is positively associated to the use of transactions as it expands the
utility of holding the currency, increasing then Bitcoin price (Kristoufek, 2014). So, it is
expected that a2 , 2 0 . The monetary Bitcoin velocity is measured by taking the number of
Bitcoin in a transaction and multiplying it by the number of days where coins are already
spent. Greater is Bitcoin velocity, greater will be Bitcoin prices (Ciaian et al. 2014). We
expect a3 , 3 0 . An increase in the estimated output volume affects negatively Bitcoin price
2
https://blockchain.info/
3
http://www.quandl.com/
6
in the long term (Kristoufek, 2014). We expect therefore a4 , 4 0 . The hash rate is
associated positively to Bitcoin price. According to Bouoiyour et al. (2014), an increase in
Bitcoin price generates the intention of market participants to invest and to mine, leading to a
higher hash rate. We expect that a5 , 5 0 . Kristoufek (2014) reveals that Bitcoin is not
heavily interacted with gold price. Palombizio and Morris (2012) argue that gold price may be
considered as the main source of demand and cost pressures and then seems a contributor of
inflation development and thus affect positively Bitcoin price. We expect a6 , 6 0 . The
Chinese market index is considered as a substantial player in digital currencies and in
particular Bitcoin. According to Kristoufek (2014) and Ciaian et al. (2014), the Bitcoin price
is correlated with well Chinese performing economy. We expect thus that a7 , 7 0 . The
Chinese trading bankruptcy may affect considerably Bitcoin price since Chinese market is one
of the Biggest Bitcoin market. This event has led to a remarkable drop in the prices of Bitcoin
(Bouoiyour et al. 2014). Indeed, it is well expected that 8 0 .
3.1.The ARDL Bounds Testing Method

The ARDL bounds testing approach introduced by Pesaran and Shin (1999) allows us
to see whether there is a long-run relationship between a group of time-series, some of which
may be stationary at level, while others are not. This method has various advantages: First,
the time series are assumed to be endogenous. Second, it obviates the need to classify the time
series into I(0) or I(1) as Johansen cointegration. Third, it allows us to assess simultaneously
the short-run and the long-run coefficients associated to the variables under consideration.
This paper applies this technique to investigate the relationship between Bitcoin price
and the aforementioned determinants on the one hand (Equation 1) and by incorporating then
a dummy variable that denotes the bankruptcy of Chinese trading company on the other hand
(Equation 2) to check the robustness of our results. The ARDL representation of equations (1)
and (2) are formulated as follows:
n m l h v r s z
DLBPI t a 0 a1i DLBPI t 1 a 2i DLTTRt 1 a3i DLETRt 1 a 4i DLMBV t 1 a5i DLEOV t 1 a 6i DLHASH t 1 a 7i DLGPt 1 a8i DLSI t 1
i 1 i0 i0 i0 i 0 t 0 t 0 i0
b1 LBPI t 1 b2 LTTRt 1 b3 LETRt 1 b4 LMBV t 1 b5 LEOVt 1 b6 LHASH t 1 b7 LGPt 1 b8 LSI t 1 't (3)
n m l h v r s z
DLBPI t c0 c1i DLBPI t 1 c 2i DLTTR t 1 c3i DLETRt 1 c 4i DLMBV t 1 c5i DLEOV t 1 c6i DLHASH t 1 c7i DLGPt 1 c8i DLSI t 1
i 1 i0 i0 i 0 i0 t 0 t 0 i0
d1 LBPI t 1 d 2 LTTRt 1 d 3 LETRt 1 d 4 LMBV t 1 d 5 LEOVt 1 d 6 LHASH t 1 d 7 LGPt 1 d8 LSI t 1 d 9 DV 't (4)
Where D denotes the first difference operator; ' , ' are the usual white noise residuals. To
evaluate whether there is a cointegration or not depends upon the critical bounds tabulated by
Pesaran et al. (2001, pp.300). There is a cointegration among variables if calculated F-statistic
is more than upper critical bound. If the lower bound is superior to the computed F-statistic,
we accept the null hypothesis of no cointegration. Moreover, if the F-statistic seems between
lower and upper critical bounds, the cointegration outcomes are inconclusive. The stability of
ARDL approach is assessed by carrying out various diagnostic tests and the stability analysis.
The diagnostic tests include the adjustment R-squared, the standard error regression, Breush-
7
Godfrey-serial correlation and Ramsey Reset test. The stability of short-run and long-run
estimates is checked by applying the cumulative sum of recursive residuals, the cumulative
sum of squares of recursive residuals and the recursive coefficients.
3.2.The innovative accounting approach and VEC Granger causality

The majority of empirical studies on the nexus between macroeconomic variables use
the standard Granger causality test augmented with a lagged error correction term.
Nevertheless, this method may be ineffective since it is unable to properly detect the possible
effects of shocks. To resolve these limitations, we explore an innovative accounting approach
by simulating variance decomposition and impulse response function. The purpose here is to
assess whether Bitcoin seems a safe haven, risky investment, business income, speculative
trap or long-run promise. Using variance decomposition, we decompose forecast error
variance for Bitcoin price following a one standard deviation shock to investors
attractiveness, exchange-trade volume, monetary Bitcoin velocity, estimated output volume,
hash rate, gold price and Shangai market index. This technique enables to test the strength of
its impact on the series. The impulse response function captures the shock of the own series
(the focal variable) with others series in the studied specifications. In an effort to identify
whether there is a short-run causality between the variables in question, the Granger
causality/Block Exogeneity Wald tests based upon VEC model may be useful and, to some
extent, the most convenient. It determines if the lags of any time series does not Granger
cause any other variable in the system using LM-test. The null hypothesis is accepted or
rejected based on chi-squared test based on Wald criterion to properly capture the joint
significance of the restrictions under the null hypothesis already mentioned above.
4. Results and discussion

4.1.ARDL results
To determine the most potential driver of Bitcoin price dynamic and what this crypto-
currency looks like, we start by reporting the descriptive statistics (Table-1). We clearly show
a substantial data variability, highlighting the very prime need to use robust models. The
coefficient of kurtosis appears inferior to 3 for all variables (except LTTR, LETR, LMBV and
LEOV), indicating that the distribution is less flattened than normal distribution. The
Skewness coefficient is positive for all time series (except LETR and LGP), indicating that the
asymmetrical distribution is preferable. The Jarque- Bera test revealed high and significant
values, leading to reject the assumption of normality for the concerned variables.
8
Table-1: Summary of statistics
LBPI LTTR LETR LMBV LEOV LHASH LGP LSI

Mean 3.052919 1.574058 13.41844 15.01983 13.69757 10.83858 7.319273 7.744138
Median 2.507972 1.565531 13.32571 14.95729 13.68825 9.846016 7.357317 7.717494
Maximum 7.048386 4.804185 18.09288 18.97052 17.10051 18.45453 7.547765 8.022789
Minimum -1.480693 -1.033161 4.057230 11.58991 10.64887 4.528026 7.084017 7.568131
Std. Dev. 2.078718 0.918618 2.235922 1.019057 1.033003 3.263868 0.120834 0.114295
Skewness 0.203586 0.201630 -0.668879 0.116808 0.009475 0.687444 -0.243169 0.761047
Kurtosis 2.280162 3.326236 4.017153 3.887130 3.684876 2.922190 1.703855 2.590701
Jarque-Bera 21.23110 8.362903 87.78542 26.12393 14.57141 58.86658 59.57174 77.22019
Probability 0.000025 0.015276 0.000000 0.000002 0.000685 0.000000 0.000000 0.000000
Before proceeding ARDL estimation, we determine the degree of integration of

variables. To this end, we apply Dickey-Fuller (ADF) and Phillips-Perron (PP) tests. The
results are reported in Table-2. We notice that the variables are integrated either at level or at
first difference. Given this finding, the ARDL bounds testing approach can be carried out to
test the cointegration hypothesis among the considered variables. According to the ARDL
bounds testing approach, lag order of the variables is important for the model specification.
Hence, we determine the lag optimization based on lag-order selection using various
information criteria including Akaike Information Criterion (AIC), Schwarz information
criterion (SC), and Hannan-Quinn criterion (HQ). Since AIC has superior power properties
for sample data compared to any lag length criterion, we show that the optimum lag is 3
(Table-3). When considering the Chinese trading bankruptcy (DV), the selected lag order is
also 3.
Table-2: Results of ADF and PP Unit Tests

Variables ADF test PP test
Level First difference Level First difference
LBPI --- -15.8916*** --- -32.5107***
LTTR -5.8908** --- -15.5010*** ---
LETR -2.9074** --- -31.0877*** ---
LMBV -5.5649*** --- -25.8706*** ---
LEOV -3.7443** --- --- -72.5447***
LHASH --- -29.0159*** --- -13.7236***
LGP --- -26.9126*** --- -23.3523***
LSI --- -28.5842*** --- -18.5978***
Notes: ***, ** and * imply significance at the 1%, 5% and 10% level, respectively ; The numbers within parentheses for the
ADF and PP statistics represents the lag length of the dependent variable used to obtain white noise residuals ; The lag
lengths for the ADF and PP tests were selected using Akaike Information Criterion (AIC).
9
Table-3: Lag-order selection
Lag LogL LR FPE AIC SC HQ
(1) FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LSI)
0 795.3703 NA 0.006820 -2.149987 -2.048775 -2.110926
1 799.7037 8.463462 0.006758 -2.159183 -2.051645* -2.117680
2 802.3041 5.071735* 0.006728 -2.163598 -2.049734 -2.119654*
3 803.4872 2.304132 0.006725* -2.164103* -2.043913 -2.117718
4 803.6028 0.224915 0.006741 -2.161663 -2.035148 -2.112837
5 803.6350 0.062545 0.006759 -2.158993 -2.026152 -2.107726
6 803.9671 0.643943 0.006772 -2.157151 -2.017984 -2.103442
7 804.0653 0.190309 0.006789 -2.154663 -2.009171 -2.098513
8 804.9309 1.673839 0.006791 -2.154292 -2.002474 -2.095701
(2) FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LSI, DV)
0 781.6729 NA 0.007309 -2.080742 -1.974351* -2.039709*
1 782.5517 1.714736 0.007312 -2.080413 -1.967763 -2.036966
2 782.9059 0.690066 0.007325 -2.078656 -1.959747 -2.032795
3 785.3696 4.793244* 0.007295* -2.082638* -1.957472 -2.034364
4 785.3825 0.025151 0.007315 -2.079952 -1.948528 -2.029264
5 785.4114 0.056055 0.007334 -2.077310 -1.939627 -2.024208
6 785.4309 0.037764 0.007354 -2.074642 -1.930700 -2.019126
7 785.4515 0.039790 0.007374 -2.071977 -1.921777 -2.014047
8 785.6675 0.417417 0.007390 -2.069844 -1.913385 -2.009500
Notes: * indicates lag order selected by the criterion; LR: sequential modified LR test statistic (each test at 5% level); FPE:
Final prediction error; AIC: Akaike information criterion; SC: Schwarz information criterion; HQ: Hannan-Quinn
information criterion.
Using ARDL Bounds testing approach, we show interesting results (Table-4): The
impact of users interest to Bitcoin or investors attractiveness plays a significant role in
explaining Bitcoin price formation. Indeed, an increase by 10% in TTR expands the BTP by
about 2.01%. The exchange-trade ratio affects positively and significantly the price of
Bitcoin. An increase by 10% of ETR leads to an increase by 0.32% of BPI. Bitcoin velocity
and estimated output volume have no significant impact on Bitcoin price formation. The
influence of technical driver (HASH) seems positive and significant but minor. We notice that
an increase by 10% of HASH prompts an increase by 0.03% in the prices of Bitcoin. Gold
price has no influence on Bitcoin price, while Shangai market index contributes positively and
significantly to BPI, i.e. an increase by 10% of SI leads to an increase by 1.18% (Equation (1),
Table-3). When including the dummy variable denoting Chinese bankruptcy, the results still
stable in terms of signs and significance (Equation (2), Table-4). This implies their sharp
robustness.
In addition, we depict from Table-5 that the value of F-statistic exceeds the upper
bound at the 10% significance level either by considering or not DV, implying that there is
evidence of a long-run relationship among variables at this level of significance. These results
seem insufficient to capture accurately the evidence of long-term linkage because ARDL
bounds test is unable to detect structural breaks stemming in the time series under
consideration. Given its inability to account for nonlinearity, we believe that it is substantial
to apply the method of Gregory and Hansen (1996) to re-explore this nexus. This technique is
based on an unknown structural break in the focal variables with respect to Engle-Granger
residual. This test reinforces the fact that there is a long-run cointegration between Bitcoin
10
price and its drivers and highlights the great importance to consider structural breaks in the
interaction dynamic process of BPI as well as its main determinants (Equation (1), Table-6).
These outcomes do not change substantially when accounting for DV (Equation (2), Table-6).
The effect of the added dummy variable seems negative and statistically significant as
expected (Section-3).
Table-4: The ARDL Bounds Testing Analysis
Dependent variable: DLBPIt
(1) (2)
C 0.6078 3.4815
(1.0537) (1.1373)
DLBPIt-1 0.11687** 0.5641**
(2.96916) (3.0184)
DLBPIt-2 0.11154** 0.1557***
(2.95493) (3.8357)
DLBPIt-3 -0.0618 -0.0523
(-1.6440) (-1.5666)
DLTTRt-1 0.20127*** 0.4846*
(9.12259) (1.8352)
DLETRt-1 0.0329* 0.0825*
(1.6778) (1.6934)
DLMBVt-1 0.00134 0.0049
(0.2775) (0.2057)
DLEOVt-1 0.0030 0.0428
(0.37838) (1.9022)
DLHASHt-1 0.01192 0.0075
(0.4814) (0.4132)
DLGPt-1 0.17445 0.3248
(0.6631) (0.1847)
DLSIt-1 0.1182* 0.3516*
(1.9049) (2.2567)
LBPIt-1 -0.01014 0.1602***
(-1.0310) (3.2488)
LTTRt-1 0.0038 0.0336
(0.4752) (1.1308)
LETRt-1 0.0096* 0.0314
(1.8057) (0.8947)
LMBVt-1 0.0038 0.0344
(0.6587) (1.2216)
LEOVt-1 0.0034 0.0137
(0.5983) (0.4755)
LHASHt-1 0.0035* 0.0092*
(1.7380) (1.8607)
LGPt-1 -0.1189 -0.0555
(-1.3637) (-1.1431)
LSIt-1 0.02128 -1.0622
(0.4324) (-0.8250)
DV --- -0.0957*
(-1.8796)
Diagnostic tests
R-squared 0.4586 0.48
SE regression 0.8859 0.7241
Breush-Godfrey serial correlation 0.0955 [0.9089] 0.0133 [0.6214]
Ramsey Reset test 0.03503 [0.8516] 0.0217 [0.6528]
Notes: ***, ** and * imply significance at the 1%, 5% and 10% level, respectively; [.]: p-value.
11
Table -5: The ARDL Bounds Testing Analysis
Estimated model Optimal lag length F-statistic Prob.
(1) FBPI (LBPI/LTTR, LETR, LMBV, 3, 3,4, 1, 0, 0 4.7029* 0.0106
LEOV, LHASH, LGP, LSI)
(2) FBPI (LBPI/LTTR, LETR, LMBV, 3, 3,4, 1, 0, 0 4.2852* 0.0381
LEOV, LHASH, LGP, LSI, DV)
Significance level Critical values
Lower bounds I(0) Upper bounds I(1)
1% 6.84 7.84
5% 4.94 5.73
10% 4.04 4.78
Notes: ***, ** and * imply significance at the 1%, 5% and 10% levels, respectively; Critical values were obtained from
Pesaran et al. (2001).
Table-6: Gregory-Hansen Structural Break Cointegration Test

Estimated model (1) FBPI (LBPI/LTTR, LETR, LMBV, (2) FBPI (LBPI/LTTR, LETR, LMBV,
LEOV, LHASH, LGP, LSI) LEOV, LHASH, LGP, LSI, DV)
Structural break year 27/10/2013 28/10/2013
ADF-test -4.9861** -4.8743***
Prob.values 0.0029 0.0000
Significance level Critical values of the ADF test
1% -5.8652
5% -4.9271
10% -4.8135
Notes: ***, ** and * imply significance at the 1%, 5% and 10% level, respectively.
The diagnostic tests show that there is no evidence of serial correlation. The Ramsey
reset test statistic reveals the performance of the short-run model (Table-4). The CUSUM and
the CUSUM Squares test show the adequacy of the considered models at 5% level of
significance (Equations (1) and (2), Figure-2) and the stability of ARDL parameters
(Equations (1) and (2), Figure-3).
12
Figure-2: Plots of cumulative sum of recursive and of squares of recursive residuals
(1): FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LSI)
80 1.2
60
1.0
40
0.8
20
0.6
0
0.4
-20
0.2
-40
-60 0.0
-80 -0.2
I II III IV I II III IV IV I II III IV I II III IV
2011 2012 2011 2012
CUSUM 5% Significance CUSUM of Squares 5% Significance

(2) : FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LSI, DV)
80 1.2
60
1.0
40
0.8
20
0.6
0
0.4
-20
0.2
-40
-60 0.0
-80 -0.2
IV I II III IV I II III IV IV I II III IV I II III IV
2012 2013 2012 2013

Notes: The straight lines represent the critical bounds at 5% significance level.
13
Figure-3: Plots of cumulative sum of recursive coefficients
(1): FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LGP, LSI)
200 8 3 2 1
100 2
6 1
0
0 1
4 0
-100 0 -1
2 -1
-200 -1
-2
0 -2
-300 -2
-400 -2 -3 -3 -3
I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV
2011 2012 2011 2012 2011 2012 2011 2012 2011 2012
Recursi ve C(1) Esti mates Recursi ve C(2) Esti mates Recursi ve C(3) Esti mates Recursi ve C(4) Esti mates Recursi ve C(5) Esti mates
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
1 .6 2.0 4 20
.4 1.5 2
0 10
.2 1.0 0
-1 0
.0 0.5 -2
-2 -10
-.2 0.0 -4
-3 -.4 -0.5 -6 -20

2011 2012 2011 2012 2011 2012 2011 2012 2011 2012
Recursi ve C(6) Esti mates Recursi ve C(7) Esti mates Recursi ve C(8) Esti mates Recursi ve C(9) Esti mates Recursi ve C(10) Estimates
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
10 2 6 6 .4
.2
0
0 4 4
.0
-2
-10 2 2 -.2
-4
-.4
-20 0 0
-6 -.6
-30 -8 -2 -2 -.8
2011 2012 2011 2012 2011 2012 2011 2012 2011 2012
Recursi ve C(11) Estimates Recursi ve C(12) Estimates Recursi ve C(13) Estimates Recursi ve C(14) Estimates Recursi ve C(15) Estimates
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
1 6 20 60
0 10
4 40
-1 0
2 20
-2 -10
0 0
-3 -20
-4 -2 -30 -20
I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV
2011 2012 2011 2012 2011 2012 2011 2012
Recursi ve C(16) Estimates Recursi ve C(17) Estimates Recursi ve C(18) Estimates Recursi ve C(19) Estimates
2 S.E. 2 S.E. 2 S.E. 2 S.E.
(2) FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LGP, LSI, DV)
20 0.4 .05 .08 4
.00
15 .04
0.0 2
-.05
10 .00
-0.4 -.10 0
5 -.04
-.15
-0.8 -2
0 -.20 -.08
-5 -1.2 -.25 -.12 -4

M4 M5 M6 M7 M8 M9 M10 M11 M12 M4 M5 M6 M7 M8 M9 M10 M11 M12 M4 M5 M6 M7 M8 M9 M10 M11 M12 M4 M5 M6 M7 M8 M9 M10 M11 M12 M4 M5 M6 M7 M8 M9 M10 M11 M12
2013 2013 2013 2013 2013
Recursi ve C(1) Esti mates Recursi ve C(2) Esti mates Recursi ve C(3) Esti mates Recursi ve C(4) Esti mates Recursi ve C(5) Esti mates
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.8 .10 .20 .8 1.2
.6
.15 .6
0.8
.4 .05
.10 .4
.2 0.4
.05 .2
.0 .00
0.0
-.2 .00 .0
-.4 -.05 -.05 -.2 -0.4

2013 2013 2013 2013 2013
Recursi ve C(6) Esti mates Recursi ve C(7) Esti mates Recursi ve C(8) Esti mates Recursi ve C(9) Esti mates Recursi ve C(10) Estimates
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
4 .1 1 .1 .15
2 .10
.0 0 .0
0 .05
-.1 -1 -.1
-2 .00
-.2 -2 -.2
-4 -.05
-6 -.3 -3 -.3 -.10

2013 2013 2013 2013 2013
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.10 .4 .1 0.8 .3
.2
.3 0.4
.05 .0
.1
.2 0.0
.00 -.1 .0
.1 -0.4
-.1
-.05 -.2
.0 -0.8 -.2
-.10 -.1 -.3 -1.2 -.3

2013 2013 2013 2013 2013
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
Notes: The straight lines represent the critical bounds at 5% significance level.
14
From our results reported in Table-7, we clearly notice that Bitcoin price interacts
differently with its determinants depending to time periods. In the short-run, the users
interest, the exchange-trade ratio, the estimated output volume and the Shangai index affect
positively and significantly the BPI. However, the monetary velocity, the hash rate and the
gold price have no influence on this digital money. These outcomes change intensely in the
long-run, i.e. the speculation, the EOV and the Chinese stock market index which play
determinant roles in the short term become without significant influence on Bitcoin price
development in the long-run. The impact of ETR on BPI stills positive and significant, but
becomes much less important. The effects of MBV and GP on BPI remain insignificant,
whereas the hash rate plays a significant role in the long term. Furthermore, the value of ECT
is negative and statistically significant at 5 percent level, which is theoretically correct. It
amounts (-7.97E-06), implying that the deviation in the short-run is corrected by 0.0007%
towards the long-run equilibrium path. The R-squared value indicates that 44% of Bitcoin
price dynamic is explained by the explanatory variables under consideration. These findings
change slightly when moving from Equation (1) to Equation (2) that considers the Chinese
Bankruptcy. Both estimates show that investors attractiveness and the performance of
Chinese market index are the most influent variables on the dynamic of Bitcoin price.
4.2.Innovative accounting approach results

The results of the variance decomposition are reported in Table-8. We find that
69.17% percent of Bitcoin price is explained by its own innovative shocks (Upper Table-8).
The investors attractiveness (TTR) plays the major role in explaining the price dynamic of
Bitcoin (20.34%). The contribution of ETR appears minor, amounting 0.16%. Similarly for
Bitcoin monetary velocity, the estimated output volume and the hash rate with respective
percentages equal to 0.035%, 0.037% and 0.003%. Gold price explains 0.095% of BPI but we
should not forget to mention that the link between GP and BPI appears insignificant in the
above results. Additionally, the contribution of Chinese market index (SI) in explaining the
Bitcoin dynamic seems sharply intense (10.14%). When considering DV (Lower Table-8), the
contributions of speculation and Shangai market index on the evolution of BPI remain
dominant.
To be more effective in our analysis, we add the results of the impulse response
function. It traces the time path of the impacts of shocks of independent variable on the
dependent variables in a VAR system. The impulse response function allows us to show how
long independent variable reacts to shock stemming in the dependent variables. We can see
also the magnitude of the response of Bitcoin price to its own shock, those of investors
attractiveness, exchange-trade volume, monetary Bitcoin velocity, estimated output volume,
hash rate, gold price and Shangai market indices. Figure-4 worthy indicates that the response
in Bitcoin price owing to forecast error stemming in investors attractiveness (TTR) is positive
over time. The contributions of ETR, MBV, EOV, HASH and GP to Bitcoin price appear
negligible. Besides, the Bitcoin price reacts positively to the Chinese market index over all the
considered period. It is substantial to mention here that the innovative accounting approach
outcomes are very close with ARDL results. We always show that TTR and SI are dominant
in the explanation of Bitcoin price dynamic. There is no remarkable change in shocks
15
responses when accounting for Chinese bankruptcy (Lower graph, Figure-4). The reaction of
BPI to TTR and SI still positive, while the contributions of the rest of variables remain very
slight or negligible
Table-7: Short-run and long-run Analysis

Dependent variable: LBPIt
(1) (2)
Short-run
DLBPIt 0.1252*** 0.3722***
(3.1873) (7.6306)
DLTTRt 0.5269** 0.3107**
(2.8944) (3.2019)
DLETRt 0.1287*** 0.0954***
(7.0988) (5.4125)
DLMBVt 2.7411 -5.1072
(0.2189) (-1.3082)
DLEOVt 0.0798*** 0.1583***
(3.6287) (3.7943)
DLHASHt 0.0594 0.3040
(0.5379) (0.1569)
DLGPt -0.2415 -0.0238
(-0.9103) (-0.9867)
DLSIt 0.3802* 0.2272**
(1.6444) (2.9769)
ECTt -7.97E-06** -3.20E-06*
(-2.5130) (-1.7186)
Long-run
LBPIt 0.1328*** 0.2309***
(3.3635) (4.7347)
LTTRt 0.1434 0.0279
(0.5414) (1.2933)
LETRt 0.0180* 0.0222*
(1.7073) (1.9182)
LMBVt 0.0043 0.0287
(0.8892) (0.9623)
LEOVt 0.0073 -0.0030
(0.8993) (-0.0778)
LHASHt 0.0072* 0.0076*
(1.8478) (1.9784)
LGPt -0.0015 0.2140
(-0.1556) (0.8852)
LSIt 0.2157 0.3295
(0.1062) (0.2478)
DV --- -0.0812*
(-1.7697)
Diagnostic tests
R-squared 0.44 0.36
SE regression 0.7812 0.5376
Breush-Godfrey serial correlation 0.3987 [0.1125] 0.0862 [0.5034]
Ramsey Reset test 0.2419 [0.6038] 0.0129 [0.3185]
Notes : ***, ** and * imply significance at the 1%, 5% and 10% levels, respectively Diagnostic tests results are based on F-
statistic ; [.] : p-values.
16
Table-8: Variance Decomposition of Bitcoin price
Period S.E. LBPI LTTR LETR LMBV LEOV LHASH LGP LSI
(1) FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LGP, LSI)
1 0.089209 100.0000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
2 0.133356 69.62125 20.02477 0.099387 0.021195 0.048033 0.000927 0.002721 10.18171
3 0.173881 69.36913 20.14811 0.154151 0.041684 0.040414 0.008345 0.074429 10.16373
4 0.207915 69.31502 20.21095 0.143917 0.034885 0.040420 0.005948 0.079367 10.16948
5 0.237979 69.26216 20.26038 0.154534 0.037175 0.038559 0.004840 0.083554 10.15879
6 0.264822 69.22643 20.29075 0.160299 0.037687 0.038561 0.004506 0.087948 10.15380
7 0.289336 69.20724 20.31188 0.161535 0.037241 0.038131 0.003989 0.091187 10.14878
8 0.311935 69.19196 20.32765 0.163871 0.036489 0.037956 0.003689 0.093026 10.14535
9 0.333019 69.18027 20.33966 0.165645 0.035905 0.037888 0.003476 0.094519 10.14264
10 0.352847 69.17171 20.34903 0.166578 0.035233 0.037921 0.003293 0.095698 10.14054
(2) FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LGP, LSI, DV)
1 0.437211 100.0000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
2 0.531016 69.16401 20.07857 0.046293 0.192572 0.172621 0.216206 7.05E-05 10.12964
3 0.587408 68.89641 20.06423 0.074224 0.207786 0.157107 0.180322 0.175893 10.24402
4 0.653719 68.88240 20.05204 0.094030 0.169006 0.140286 0.155463 0.211353 10.29542
5 0.713412 68.85767 20.04848 0.091867 0.142428 0.156410 0.158901 0.212927 10.33130
6 0.765985 68.85128 20.04238 0.094067 0.123555 0.162226 0.144646 0.224575 10.35726
7 0.815668 68.84969 20.03788 0.097420 0.109980 0.162901 0.135923 0.233969 10.37223
8 0.862787 68.84846 20.03494 0.099140 0.098834 0.165991 0.130940 0.239833 10.38186
9 0.907295 68.84839 20.03210 0.100438 0.090140 0.169011 0.125686 0.244983 10.38925
10 0.949679 68.84880 20.02980 0.101707 0.083155 0.170850 0.121426 0.249415 10.39483
17
Figure-4: Impulse Response Functio
(1): FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LSI)
Response to Nonfactorized One Unit Innovations
Response of BP I to B P I Response of BP I to TTR Response of B P I to E TR Response of B P I to MB V Response of B P I to EOV Response of B P I to HA S H Response of BP I to GP Response of B PI to S I
1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of TTR to BP I Response of TTR to TTR Response of TTR to E TR Response of TTR to MB V Response of TTR to EOV Response of TTR to HA S H Response of TTR to GP Response of TTR to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of E TR to BP I Response of ETR to TTR Response of E TR to E TR Response of E TR to MB V Response of ETR to EOV Response of E TR to HA S H Response of E TR to GP Response of ETR to S I
4 4 4 4 4 4 4 4
2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of MB V to B P I Response of MB V to TTR Response of MB V to ETR Response of MB V to MB V Response of MB V to E OV Response of MB V to HA S H Response of MB V to GP Response of MB V to SI
2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2
-4 -4 -4 -4 -4 -4 -4 -4
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of E OV to B P I Response of EOV to TTR Response of E OV to ETR Response of EOV to MB V Response of EOV to E OV Response of EOV to HA S H Response of E OV to GP Response of EOV to SI
2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2
-4 -4 -4 -4 -4 -4 -4 -4
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of HA S H to BP I Response of HA SH to TTR Response of HA S H to ETR Response of HA S H to MB V Response of HA S H to EOV Response of HA S H to HA S H Response of HA S H to GP Response of HA SH to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of GP to B P I Response of GP to TTR Response of GP to E TR Response of GP to MB V Response of GP to EOV Response of GP to HA S H Response of GP to GP Response of GP to SI
2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2
-4 -4 -4 -4 -4 -4 -4 -4
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of SI to B P I Response of SI to TTR Response of SI to E TR Response of S I to MB V Response of SI to EOV Response of SI to HA S H Response of SI to GP Response of SI to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
(2): FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LSI, DV)
Response to Nonfactorized One S.D. Innovations
Response of B P I to B P I Response of B P I to TTR Response of B P I to E TR Response of B P I to MB V Response of B P I to E OV Response of B P I to HA S H Response of B P I to GP Response of B P I to S I
.6 .6 .6 .6 .6 .6 .6 .6
.4 .4 .4 .4 .4 .4 .4 .4
.2 .2 .2 .2 .2 .2 .2 .2
.0 .0 .0 .0 .0 .0 .0 .0
-.2 -.2 -.2 -.2 -.2 -.2 -.2 -.2

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of TTR to B P I Response of TTR to TTR Response of TTR to E TR Response of TTR to MB V Response of TTR to E OV Response of TTR to HA S H Response of TTR to GP Response of TTR to S I
.8 .8 .8 .8 .8 .8 .8 .8
.4 .4 .4 .4 .4 .4 .4 .4
.0 .0 .0 .0 .0 .0 .0 .0
-.4 -.4 -.4 -.4 -.4 -.4 -.4 -.4

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of E TR to B P I Response of E TR to TTR Response of E TR to E TR Response of E TR to MB V Response of E TR to E OV Response of E TR to HA S H Response of E TR to GP Response of E TR to S I
.01 .01 .01 .01 .01 .01 .01 .01
.00 .00 .00 .00 .00 .00 .00 .00
-.01 -.01 -.01 -.01 -.01 -.01 -.01 -.01

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of MB V to B P I Response of MB V to TTR Response of MB V to E TR Response of MB V to MB V Response of MB V to E OV Response of MB V to HA S H Response of MB V to GP Response of MB V to S I

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of E OV to B P I Response of E OV to TTR Response of E OV to E TR Response of E OV to MB V Response of E OV to E OV Response of E OV to HA S H Response of E OV to GP Response of E OV to S I

.6 .6 .6 .6 .6 .6 .6 .6
.4 .4 .4 .4 .4 .4 .4 .4
.2 .2 .2 .2 .2 .2 .2 .2
.0 .0 .0 .0 .0 .0 .0 .0
-.2 -.2 -.2 -.2 -.2 -.2 -.2 -.2

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of HA S H to B P I Response of HA S H to TTR Response of HA S H to E TR Response of HA S H to MB V Response of HA S H to E OV Response of HA S H to HA S H Response of HA S H to GP Response of HA S H to S I

.12 .12 .12 .12 .12 .12 .12 .12
.08 .08 .08 .08 .08 .08 .08 .08
.04 .04 .04 .04 .04 .04 .04 .04
.00 .00 .00 .00 .00 .00 .00 .00
-.04 -.04 -.04 -.04 -.04 -.04 -.04 -.04
-.08 -.08 -.08 -.08 -.08 -.08 -.08 -.08

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of GP to B P I Response of GP to TTR Response of GP to E TR Response of GP to MB V Response of GP to E OV Response of GP to HA S H Response of GP to GP Response of GP to S I

.015 .015 .015 .015 .015 .015 .015 .015
.010 .010 .010 .010 .010 .010 .010 .010
.005 .005 .005 .005 .005 .005 .005 .005
.000 .000 .000 .000 .000 .000 .000 .000
-.005 -.005 -.005 -.005 -.005 -.005 -.005 -.005

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of S I to B P I Response of S I to TTR Response of S I to E TR Response of S I to MB V Response of S I to E OV Response of S I to HA S H Response of S I to GP Response of S I to S I

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5

2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
18
Furthermore, we evaluate whether there is a causal relationship between the
explanatory variables in question and the Bitcoin price dynamic. Before testing the non-
causality hypothesis, we start by examining the residuals using the LM test for serial
independence against the alternative of AR(k)/MA(k), for k = 1, ...., 12. From the findings
reported in Table-9, the serial correlation may be removed at the maximum lag length which
is 3. The non-causality test findings are reported in Table-10. It is clearly notable that we can
reject the null hypothesis of no causality DLTTR to DLBPI, from DLETR to DLBPI and from
DLSI to DLBPI, while the reverse link is not supported confirming therefore the above
outcomes obtained through the ARDL Bounds Testing method and the innovation accounting
approach (variance decomposition and impulse responses). For the rest of variables, we
accept the null hypothesis of non-causality (except for the relationship that runs from DLBPI
to DLHASH and the link running from DLBPI to DLMBV). These results may very useful for
businesses, investors and regulators.
Table-9: VEC Residual Serial Correlation LM Tests

Null Hypothesis: no serial correlation at lag order h
Lags LM-Stat Prob
1 165.7815 0.0000
2 162.7223 0.0000
3 172.6073 0.0000
4 74.87208 0.1661
5 108.8017 0.0004
6 52.65505 0.8435
7 86.67175 0.0312
8 59.58174 0.6333
9 73.80962 0.1882
10 67.46570 0.3595
11 69.17378 0.3071
12 88.51908 0.0229
Table-10: VEC Granger Causality/Block Exogeneity Wald Tests

Dependent variable: DLBPI
Excluded Chi-sq df Prob
DLTTRDLBPI 4.4897 2 0.0474
DLBPIDLTTR 0.7034 2 0.7035
DLETRDLBPI 2.9722 2 0.0226
DLBPIDLETR 4.2470 2 0.1196
DLMBVDLBPI 0.9299 2 0.6281
DLBPIDLMBV 13.698 2 0.0011
DLEOVDLBPI 1.1004 2 0.5768
DLBPIDLEOV 1.9394 2 0.3792
DLHASHDLBPI 0.3544 2 0.8376
DLBPIDLHASH 6.2336 2 0.0443
DLGPDLBPI 1.0579 2 0.3574
DLBPIDLGP 1.0588 2 0.3572
DLSIDLBPI 3.5051 2 0.0733
DLBPIDLSI 1.4394 2 0.4869
19
5. Robustness
The above findings clearly indicate that the investors attractiveness, the exchange-
trade ratio, the estimated output volume and the Shangai index affect positively and
significantly the Bitcoin price, while the monetary velocity, the hash rate and the gold price
have no influence in the short term. However, the speculative nature of Bitcoin, the EOV and
the Chinese stock market index which play the major role in the short-run appear without
statistically significant impact on Bitcoin price in the long-run. The influence of ETR on BPI
becomes less strong, whereas the effects of MBV and GP on BPI remain statistically
insignificant in the majority of cases. The hash rate plays a significant role on explaining the
dynamic of this nascent virtual currency. To check properly and appropriately the robustness
of these evidences, we re-estimate the relationships between Bitcoin price and its
determinants by incorporating a dummy variable relative to the bankruptcy of Chinese trading
company, using the same methods (i.e. an ARDL Bounds Testing method, an innovation
accounting approach by simulating variance decomposition and impulse response function
and VEC Granger causality test). Comparing these results with those of Equation without
dummy variable, we put in evidence that the effects of TTR, ETR, MBV, EOV, HASH, GP and
SI are solid and unambiguous. Beyond the nuances of short and long terms, the present study
confirms the speculative nature of Bitcoin without neglecting its usefulness in economic
reasons and the importance of accounting for Chinese stock market and the processing power
of Bitcoin network. At this stage, we can consider it only as a risky investment, short-term
hedge and partially as business income. Nonetheless, this new crypto-currency seems far from
being a safe haven and a long-term promise.
To be more effective, we believe that the use of other combinations of variables by

adding other variables in Equations 3 and 4 (oil price 4, Dow Jones index5 and a dummy
variable denoting the closing of road silk by FBI 6) that may affect the Bitcoin price based on
few studies on the field (Ciaian et al. 2014, for example) may be fruitful. Nevertheless, the
obtained findings reveal that the effects of the additional time series are in the majority of
cases insignificant and more importantly the estimates become sharply unstable (see Figure
A-1, particularly). More details about outcomes are summarized in Table A-1, Table A-2,
Table A-3, Table A-4, Table A-5, Table A-6, Figure A-2 and Figure A-3.
4
Palombizio and Morris (2012) find that oil price (OP) is a potential factor that may affect intensely the inflation outcomes.
If the price of oil indicates great ups and downs (i.e. sizeable volatility), the Bitcoin depreciates. Besides, the exchange rate
may reflect inflationary pressures affecting positively the prices of this crypto-currency.
5
The relationship between Bitcoin price and the Dow Jones index (DJI) appears complex, since the two variables seem
sometimes correlated but not usually. After the announcement of American satellite TV provider that it would start accepting
Bitcoin as payment tool, the prices of this digital money increased approximately by $40 touching the level of $ 600, while
the Dow Jones Index was down by 300 points. A perfect example of how the Bitcoin and the American markets have been
initially unrelated. Nevertheless, the offshoots of Al-Qaeda over different cities in Iraq and the Obamas declaration (i.e.
America will not send the military in order to fight off the terrorist organizations) have affected Bitcoin price and
simultaneously Dow Jones index. Due to the sizeable connection between the turmoil and Bitcoins value, the price of
Bitcoin started dropping and as response the Dow Jones index started falling by 200 points. This implies that there is some
connection between both variables. For details, you can refer to: http://coinbrief.net/bitcoin-price-news-analysis/
6
The Road Silk is a roating-platform of drug on which transactions were through Bitcoin. Thus, its closing by FBI in
23/10/2013 (DV) may affect substantially the dynamic of Bitcoin price.
20
6. Concluding remarks and Policy implications
The present research attempts to reach clearer knowledge about the nascent crypto-
currency (Bitcoin) by effectively answering the following questions: What Bitcoin looks like?
Is it a safe haven or a speculative bubble? Is it a business income, a short-term hedge, a
risky investment or a long-term promise?
To this end, we have regressed Bitcoin price on investors attractiveness, exchange-
trade volume, monetary Bitcoin velocity, estimated output volume, hash rate, gold price and
Shangai market index using an ARDL Bounds Testing method, an innovation accounting
approach and VEC Granger causality test for daily data covering the period from December
2010 to June 2014. By doing so, we have checked the speculative nature of Bitcoin. We also
provide insightful evidence that Bitcoin may be used for economic reasons but there is any
sign of being a safe haven. By accounting for the Chinese trading bankruptcy, the contribution
of the speculative behavior of investors and the performance of Chinese stock market remain
dominant, while the role of Bitcoin as transactions tool dissipates in the long-run, highlighting
the robustness of our results. Intuitively, by using other combinations of variables by adding
other time series (including oil price, Dow Jones index and a dummy variable denoting the
closing of road silk by FBI) to confirm our findings, the estimates become remarkably
unstable. It is important to mention here that these last variables have no statistically
significant influence in the majority of cases in Bitcoin price development.
In a nutshell, Bitcoin behaves heavily as a speculative bubble, short-term hedge and
risky investment and partially as business income. This new digital money is far from being a
long-term promise, especially when considering that this virtual currency faces a great
challenge (in particular a structural economic problem) regarding its limited amount recording
21 million units in 2140, implying that the money supply would not expand after this date. If
this digital currency succeeds really to displace fiat currencies, it would exert great
deflationary pressures.
This goes without saying that these findings should be treated with caution. Nobody is,
up to now, able to estimate the true value of Bitcoin. The fact that the dynamic of the focal
digital money is uncertain even more sustains speculation. Without tackling the main causes,
the virtual currency seems highly correlated to the speculative behaviors of investors and
people who hold this money. Bitcoin is not issued by banking system and even less by
governments, but by a computing algorithm. Unfortunately, the majority of Bitcoin users have
not heavily acknowledged about mathematical programs, and it is of course unknown for
them how far it can go. The volatility of Bitcoin and the difficulty of processing power
network are likely to discourage investors and users of this money. Intuitively, China
represents the most active Bitcoin market in the world. The sizeable attention to this crypto-
currency in the Chinese media has drawn a huge number of investors. However, the attitude
of Chinese practitioners, advisers and regulators towards Bitcoin is ambiguous, yielding to
much more speculation. This may reinforce the evidence thereby Bitcoin is short-term hedge,
a risky investment. We cannot confirm if this currency may be considered as long-term
promise since the contribution of investors interest appears dominant among the different
estimations. This may support the conclusion of Bouoiyour et al. (2014) showing that it is
very difficult to reach clearer insights and one sided evidence into Bitcoin Phenomenon.
21
To conclude, we can say that our study has focused on one aspect of Bitcoin, i.e. its
main determinants (technical, economic and financial fundamentals). Now, this nascent
crypto-currency is primarily a protocol. In other words, it is a communication tool between
computers. It is a useful way to exchange money between different points in the world with
almost no cost. But beyond that, this language allows us to use this virtual currency in a
variety of situations. It can be used for example in inheritance while avoiding the problem of
time inconsistency or to issue shares and bonds. It can be used for alternative currencies,
smart property and other financial instruments. We are only at the beginning of a great
adventure and this remains in our opinion the most substantial.
22
References
Aguiar-Conraria, L. and Soares, M-J. (2011), The continuous wavelet transform: A
primer. NIPE working paper n16, University of Minho.
Bouoiyour, J., Selmi, R. and Tiwari, A-K. (2014), Is Bitcoin Business Income or
Speculative Bubble? Unconditional vs. Conditional Frequency Domain Analysis. Working
paper, CATT, University of Pau.
Breitung, J., and Candelon, B. (2006), Testing for short and long-run causality: a
frequency domain approach. Journal of Econometrics, 132, 363-378.
Buchholz, M., Delaney, J., Warren, J. and Parker, J. (2012), Bits and Bets,
Information, Price Volatility, and Demand for Bitcoin. Economics 312,
http://www.bitcointrading.com/pdf/bitsandbets.pdf
Ciaian, P., Rajcaniova, M. and Kancs, D. (2014), The Economics of BitCoin Price
Formation. http://arxiv.org/ftp/arxiv/papers/1405/1405.4498.pdf
Glaster, F., Kai, Z., Haferkorn, M., Weber, M. and Sieiring, M. (2014), Bitcoin -
asset or currency? Revealing usershidden intentions. Twenty Second European Conference
on Information Systems. http://ecis2014.eu/E-poster/files/0917-file1.pdf
Gregory, A.W. and Hansen, B.E. (1996), Residual based Tests for Co-integration in
Models with Regime Shifts. Journal of Econometrics, 70, 99-126.
Kristoufek, L. (2013), BitCoin meets Google Trends and Wikipedia: Quantifying the
relationship between phenomena of the Internet era. Scientific Reports 3 (3415), 1-7.
Kristoufek, L. (2014), What are the main drivers of the Bitcoin price? Evidence from
wavelet coherence analysis. http://arxiv.org/pdf/1406.0268.pdf
Glouderman, L. (2014), Bitcoins Uncertain Future in China. USCC Economic Issue
Brief n 4, May 12.
Ng, E.K. and Chan, J.C. (2012), Geophysical Applications of Partial Wavelet
Coherence and Multiple Wavelet Coherence. American Meteological Society, December.
DOI: 10.1175/JTECH-D-12-00056.1
Palombizio E. and Morris, I. (2012), Forecasting Exchange Rates using Leading
Economic Indicators. Open Access Scientific Reports 1(8), 1-6.
Pesaran, M. and Shin, Y. (1999), An Autoregressive Distributed Lag Modeling
Approach to Cointegration Analysis. S. Strom, (ed) Econometrics and Economic Theory in
the 20th Century, Cambridge University.
Pesaran, M.H., Y. Shin., and Smith R. (2001), Bounds testing approaches to the
analysis of level relationships. Journal of Applied Econometrics, 16, 289-326.
Piskorec, P., Antulov-Fantulin, N., Novak, P.K., Mozetic, I., Grcar, M., Vodenska, I.
and muc, T. (2014), News Cohesiveness: an Indicator of Systemic Risk in Financial
Markets. arXiv:1402.3483v1 [cs.SI], http://arxiv.org/pdf/1402.3483v1.pdf
Yermack, D. (2013), Is Bitcoin a Real Currency? An economic appraisal. NBER
Working Paper No. 19747. http://www.nber.org/papers/w19747
23
Appendices
Table-A.1: Lag-order selection (Equations with additional variables)
Lag LogL LR FPE AIC SC HQ
(1) : FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LOP, LDJI, LSI)
0 3678.627 NA* 2.36e-06* -10.11759* -10.04801* -10.09074*
1 3678.644 0.032814 2.37e-06 -10.11488 -10.03897 -10.08558
2 3678.673 0.057395 2.38e-06 -10.11220 -10.02997 -10.08046
3 3678.675 0.003638 2.38e-06 -10.10945 -10.02089 -10.07527
(2) : FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LOP, LDJI, LSI, DV)
0 782.4109 NA 0.006972 -2.128030 -2.058447 -2.101176
1 788.0603 11.11191 0.006883 -2.140856 -2.064947* -2.111560*
2 791.0228 5.818642 0.006846 -2.146270* -2.064035 -2.114533
3 792.0847 2.082820 0.006844* -2.146441 -2.05738 -2.112262
(3) : FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LOP, LDJI, LSI, DV)
0 163.4746 NA 0.004414 -2.585117 -2.544254 -2.569759
1 164.5226 20.77749 0.004348 -2.600201* -2.555252* -2.583308
2 164.5759 1.055509 0.004351 -2.599458 -2.550422 -2.581029*
3 164.6161 0.795628 0.004355* -2.598506 -2.545384 -2.578541
Notes: * indicates lag order selected by the criterion; LR: sequential modified LR test statistic (each test at 5% level); FPE:
Final prediction error; AIC: Akaike information criterion; SC: Schwarz information criterion; HQ: Hannan-Quinn
information criterion. DV: The closing of the Road Silk by FBI, which amounts 1 from 23/10/2013 and 0 otherwise.
24
Table A-2: The ARDL Bounds Testing Analysis (Equations with additional variables)
(1) (2) (3)
C -2.4325* -1.7262* -1.4941*

(-1.7278) (-2.5645) (-2.1939)
LBPIt-1 0.1185** 0.0376* 0.0288*
(3.0231) (2.0056) (1.6232)
LBPIt-2 --- 0.0394* ---
(2.2019)
LTTRt-1 0.1222** 0.2062* 0.0068*
(3.1537) (1.7683) (1.7044)
LETRt-1 0.1153** 0.0093* 0.0087*
(3.0589) (1.8553) (1.7147)
LMBVt-1 -0.1222 0.0010 0.0011
(-0.2482) (0.4548) (0.6971)
LEOVt-1 0.0030 0.0016 0.0021
(0.3763) (0.4187) (0.5425)
LHASHt-1 -0.0141 -0.0079 -0.0060
(-0.5719) (-0.6775) (-0.5051)
LGPt-1 0.1559 -0.0614 -0.1064
(0.5900) (-0.4894) (-0.8379)
LOPt-1 -0.1043 0.1004 0.0086
(-0.5383) (1.0901) (0.9297)
LDJIt-1 -0.1268 -0.1267 -0.0971
(-0.3857) (-0.8120) (-0.6185)
LSIt-1 0.1468* 0.1235* 0.1104*
(2.000) (1.9516) (1.8452)
LBPIt-1 0.0186* 0.0141** -0.0079
(1.6551) (2.6353) (-1.3922)
LTTRt-1 -0.0162 0.0043 -0.0064
(-1.5979) (1.0714) (-1.3244)
LETRt-1 0.0158* 0.0039* 0.0059*
(2.2800) (1.9519) (1.8516)
LMBVt-1 0.0032 -0.0027 -0.0037
(0.5693) (-0.9879) (-1.3088)
LEOVt-1 0.0026 0.0051* 0.0039
(0.4453) (1.7506) (1.3735)
LHASHt-1 0.0056* -0.0010 0.0081**
(1.8862) (-0.5489) (2.6473)
LGPt-1 -0.0534 -0.0011 -0.0143
(-0.9023) (-0.0405) (-0.4907)
LOPt-1 -0.0161 -0.0653 -0.0310
(-0.2627) (-0.2364) (-0.9948)
LDJIt-1 0.0355* 0.1008*** 0.1002***
(2.2728) (3.8895) (4.0147)
LSIt-1 0.0762 0.0104 -0.0186
(1.3060) (0.3766) (-0.5807)
DV --- -0.0163* ---
(-1.7604)
DV --- --- -0.0278*
(-2.4188)
R-squared 0.54 0.44 0.42
SE regression 0.8881 0.7923 0.7795
Breush-Godfrey serial correlation 0.6231 [0.4097] 0.0069 [0.9338] 0.0081 [0.4276]
Ramsey Reset test 0.2664 [0.6058] 0.0316 [0.9689] 0.0049 [0.6618]
Notes: ***, ** and * imply significance at the 1%, 5% and 10% level, respectively; [.]: p-value; DV: The closing of the
Road Silk by FBI, which amounts 1 from 23/10/2013 and 0 otherwise.
25
Table A-3: The ARDL Bounds Testing Analysis (Equations with additional variables)
Estimated model Optimal lag length F-statistic Prob.
(1): FBPI (LBPI/LTTR, LETR, LMBV, LEOV, 3, 3,4, 1, 0, 0, 0, 0 4.5711* 0.0659
LHASH, LGP, LOP, LDJI, LSI)
LHASH, LGP, LOP, LDJI, LSI, DV)
LHASH, LGP, LOP, LDJI, LSI, DV)
Significance level Critical values
Lower bounds I(0) Upper bounds I(1)
1% 6.84 7.84
5% 4.94 5.73
10% 4.04 4.78
Notes: ***, ** and * imply significance at the 1%, 5% and 10% levels, respectively; Critical values were obtained from
Pesaran et al. (2001); DV: The closing of the Road Silk by FBI, which amounts 1 from 23/10/2013 and 0 otherwise.
Table A-4: Gregory-Hansen Structural Break Cointegration Test

(Equations with additional variables)
Estimated model (1): FBPI (LBPI/LTTR, LETR, (2): FBPI (LBPI/LTTR, LETR, (3): FBPI (LBPI/LTTR, LETR, LMBV,
LMBV, LEOV, LHASH, LGP, LMBV, LEOV, LHASH, LGP, LEOV, LHASH, LGP, LOP, LDJI,
LOP, LDJI, LSI) LOP, LDJI, LSI, DV) LSI, DV)
Structural break 23/10/2013 26/2/2013 23/10/2013
year
ADF-test -5.9234*** -4.9782** -5.2139***
Prob.values 0.0015 0.0015 0.0004
Significance level Critical values of the ADF test
1% -5.8652
5% -4.9271
10% -4.8135
Notes: ***, ** and * imply significance at the 1%, 5% and 10% level, respectively; DV: The closing of the Road Silk by
FBI, which amounts 1 from 23/10/2013 and 0 otherwise.
26
Figure A-1: Plots of cumulative sum of recursive and of squares of recursive residuals
(Equations with additional variables)
(1): FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LOP, LDJI, LSI)
80 1.2
60
1.0
40
0.8
20
0.6
0
0.4
-20
0.2
-40
-60 0.0
-80 -0.2
100 200 300 400 500 600 700 100 200 300 400 500 600 700
(2): FBPI (LBPI/LTTR, LETR, LMBV, LEOV, LHASH, LGP, LOP, LDJI, LSI, DV)
80
1.2
60
1.0
40
0.8
20
0 0.6
-20 0.4
-40 0.2
-60
0.0
-80
100 200 300 400 500 600 700 -0.2
475 500 525 550 575 600 625 650 675 700 725
CUSUM 5% Significance
CUSUM of Squares 5% Significance
40 1.2
30 1.0
20
0.8
10
0.6
0
0.4
-10
0.2
-20
-30 0.0
-40 -0.2
625 650 675 700 725 625 650 675 700 725
Notes: The straight lines represent the critical bounds at 5% significance level; DV: The closing of the Road Silk by FBI,
which amounts 1 from 23/10/2013 and 0 otherwise.
27
Figure A-2: Plots of cumulative sum of recursive coefficients (Equations with additional
variables)
1,500 3 3 4 .6
2 2
1,000 2 .4
1 1
500 0 0 0 .2
-1 -1
0 -2 .0
-2 -2
-500 -3 -3 -4 -.2
100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700
Recursive C(1) Estimates Recursive C(2) Estimates Recursive C(3) Estimates Recursive C(4) Estimates Recursive C(5) Estimates
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.8 2 20 40 80
.6 15 30 60
0
.4 10 20 40
.2 -2 5 10 20
.0 0 0 0
-4
-.2 -5 -10 -20
-.4 -6 -10 -20 -40

100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
12 4 4 6 .6
8 .4
2 2 4
.2
4
0 0 2 .0
0
-.2
-2 -2 0
-4 -.4
-8 -4 -4 -2 -.6
100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
0.5 4 10 40 40
3 20
0.0 0
0 0
2
-0.5 -20 -40
1
-10 -40
-1.0 -80
0 -60
-1.5 -1 -20 -80 -120

100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700 100 200 300 400 500 600 700
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
40
20
-20
-40
-60
100 200 300 400 500 600 700
Recursive C(21) Estimates

2 S.E.
28
8 .3 .30 .1 .04
.25
.2
4 .0 .02
.20
.1
0 .15 -.1 .00
.0
.10
-4 -.2 -.02
-.1 .05
-8 -.2 .00 -.3 -.04

500 550 600 650 700 500 550 600 650 700 500 550 600 650 700 500 550 600 650 700 500 550 600 650 700
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.06 .02 .03 .08 1.0
.02
.04 .04
.01 0.5
.01
.02 .00
.00 .00 0.0
.00 -.04
-.01
-.01 -0.5
-.02 -.02 -.08
-.04 -.02 -.03 -.12 -1.0

500 550 600 650 700 500 550 600 650 700 500 550 600 650 700 500 550 600 650 700 500 550 600 650 700
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.8 1.5 1.5 .06 .03
.04
1.0 .02
.4 1.0
.02
0.5 .01
.0 0.5 .00
0.0 .00
-.02
-.4 0.0
-0.5 -.04 -.01
-.8 -1.0 -0.5 -.06 -.02

500 550 600 650 700 500 550 600 650 700 500 550 600 650 700 500 550 600 650 700 500 550 600 650 700
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.04 .02 .02 .04 .4
.01 .02
.02 .01 .2
.00 .00
.00 .00 .0
-.01 -.02
-.02 -.01 -.2

-.02 -.04
-.04 -.03 -.02 -.06 -.4

500 550 600 650 700 500 550 600 650 700 500 550 600 650 700 500 550 600 650 700 500 550 600 650 700
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.4 .6 .4 .12
.3
.4
.2 .08
.2
.2
.1 .0 .04
.0
.0
-.2 .00
-.2
-.1
-.2 -.4 -.4 -.04

500 550 600 650 700 500 550 600 650 700 500 550 600 650 700 500 550 600 650 700
Recursive C(21) Estimates Recursive C(22) Estimates Recursive C(23) Estimates Recursive C(24) Estimates
2 S.E. 2 S.E. 2 S.E. 2 S.E.
29
2 .25 .03 .03 .015
.02 .010
.20 .02
0
.01 .005
.15 .01
-2 .00 .000
.10 .00
-.01 -.005
-4
.05 -.01 -.02 -.010
-6 .00 -.02 -.03 -.015

650 675 700 725 650 675 700 725 650 675 700 725 650 675 700 725 650 675 700 725
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.03 .08 .8 .4 0.8
.02 .2 0.4
.04 .4
.01 .0 0.0
.00 .0
.00 -.2 -0.4
-.04 -.4
-.01 -.4 -0.8
-.02 -.08 -.8 -.6 -1.2

650 675 700 725 650 675 700 725 650 675 700 725 650 675 700 725 650 675 700 725
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
1.6 .04 .03 .03 .02
1.2 .02 .02

.02 .01
0.8 .01 .01

.00 .00
0.4 .00 .00
-.02 -.01
0.0 -.01 -.01
-0.4 -.04 -.02 -.02 -.02

650 675 700 725 650 675 700 725 650 675 700 725 650 675 700 725 650 675 700 725
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.02 .02 .2 .2 .5
.4
.01
.01 .1 .1
.3
.00
.00 .0 .0 .2
-.01
.1
-.01 -.1 -.1
-.02 .0
-.02 -.03 -.2 -.2 -.1

650 675 700 725 650 675 700 725 650 675 700 725 650 675 700 725 650 675 700 725
2 S.E. 2 S.E. 2 S.E. 2 S.E. 2 S.E.
.2 .16
.12
.1
.08
.0
.04
-.1
.00
-.2 -.04
650 675 700 725 650 675 700 725
Recursive C(21) Estimates Recursive C(22) Estimates

2 S.E. 2 S.E.
Notes: The straight lines represent the critical bounds at 5% significance level; DV: The closing of the Road Silk by FBI,
which amounts 1 from 23/10/2013 and 0 otherwise.
30
Table A-5: Short-run and long-run Analysis (Equations with additional variables)
(1) (2) (3)
Short-run
LBPIt 0.1270*** 0.0281* 0.0269**
(3.2270) (2.1537) (2.5852)
LTTRt 0.4305* 0.5702* 0.4787***
(2.0214) (2.1522) (4.1026)
LETRt 0.2157*** 0.0192*** 0.0172**
(8.4441) (7.3397) (2.6367)
LMBVt -2.2467 0.7897 0.4398*
(-0.1721) (0.2109) (1.7485)
LEOVt 0.4158* -0.4434 0.0172
(2.5803) (-0.2068) (0.3859)
LHASHt -0.0283 -0.0915 -0.0057
(-0.3214) (-0.7780) (-0.3802)
LGPt -3.4273 -0.0054 -0.0928
(-1.5320) (-0.3213) (-0.6674)
LOPt -2.4806 -0.7780 0.7488
(-1.5448) (-1.4343) (1.4354)
LDJIt 2.0697 0.8341 -0.0259
(0.5522) (0.6264) (-1.3648)
LSIt 0.3256* 0.4786** 0.4784***
(1.6625) (2.6372) (4.6666)
ECTt -0.0023** -0.0020* -0.0026**
(-2.8790) (-1.6791) (-2.5190)
Long-run
LBPIt 0.1340*** 0.1265*** 0.1275**
(3.3768) (3.2112) (3.2394)
LTTRt -0.0131 0.0016 -0.0529
(-1.3168) (0.1611) (-0.2708)
LETRt 0.0088* 0.0010* 0.0029*
(1.8163) (1.7842) (1.8604)
LMBVt 0.0001*** 0.0921 -0.0012
(8.8192) (0.9284) (-0.2067)
LEOVt 0.0043 0.0655 -0.0070
(0.5435) (1.0307) (-0.8598)
LHASHt 0.0077* 0.0029* 0.0053*
(1.9745) (1.8148) (1.8371)
LGPt 0.1518 0.1534 -0.1684
(0.5697) (0.5752) (-0.6232)
LOPt -0.0518 -0.0515 0.0019
(-0.2658) (-0.2642) (0.1915)
LDJIt 0.1420*** 0.1852* 0.2417***
(4.2680) (2.4937) (3.8358)
LSIt 0.4400 0.4406 0.4457
(1.5950) (1.5948) (1.5960)
DV --- -0.0569* ---
(-1.8245)
DV --- --- -0.0782**
(-2.2516)
R-squared 0.48 0.49 0.46
SE regression 0.8561 0.8934 0.8357
Breush-Godfrey serial correlation 0.4597 [0.1386] 0.0437 [0.6795] 0.0398 [0.5012]
Ramsey Reset test 0.2392 [0.5674] 0.0087 [0.9015] 0.0127 [0.8564]
Notes : ***, ** and * imply significance at the 1%, 5% and 10% levels, respectively Diagnostic tests results are based on F-
statistic ; [.] : p-values; DV: The closing of the Road Silk by FBI, which amounts 1 from 23/10/2013 and 0 otherwise.
31
Table A-6: Variance Decomposition of Bitcoin price (Equations with additional variables)
Period S.E. BPI TTR ETR MBV EOV HASH GP OP DJI SI
1 0.089236 100.0000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
2 0.133510 69.64294 20.10299 0.012666 0.014143 0.042821 0.002420 0.007915 0.000159 0.021659 10.15228
3 0.174247 69.31781 20.09368 0.084297 0.069088 0.082248 0.008574 0.004690 0.089813 0.132293 10.11750
4 0.208220 69.21861 20.07800 0.087726 0.063105 0.091891 0.006137 0.003851 0.130538 0.194279 10.12585
5 0.238292 69.13212 20.07648 0.093821 0.068997 0.098099 0.004751 0.004467 0.153696 0.242479 10.12509
6 0.265110 69.07429 20.07543 0.098891 0.069911 0.104294 0.004269 0.004888 0.171241 0.272138 10.12463
7 0.289584 69.04017 20.07283 0.102049 0.070048 0.107904 0.003690 0.005221 0.182453 0.292235 10.12339
8 0.312142 69.01439 20.07158 0.104564 0.069695 0.110543 0.003311 0.005473 0.190445 0.307239 10.12275
9 0.333190 68.99426 20.07075 0.106614 0.069345 0.112625 0.003047 0.005651 0.196888 0.318703 10.12211
10 0.352985 68.97904 20.06981 0.108108 0.068821 0.114341 0.002823 0.005788 0.201978 0.327628 10.12165
1 0.088898 100.0000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.0000000
2 0.133945 72.56927 20.13121 0.041758 8.8E-05 0.098224 0.027560 0.001589 0.000687 0.002292 17.127313
3 0.175764 72.08224 20.13425 0.148067 0.034699 0.244634 0.017965 0.081727 0.122574 0.031775 17.102061
4 0.208055 71.73926 20.10767 0.289199 0.034402 0.381936 0.029360 0.123798 0.144773 0.075313 17.074290
5 0.237772 71.19855 20.217509 0.322583 0.032966 0.647179 0.022938 0.127155 0.139636 0.215343 17.076146
6 0.263958 70.90378 20.290786 0.336065 0.046484 0.709422 0.019024 0.136528 0.172126 0.316877 17.068907
7 0.288247 70.70841 20.360593 0.333563 0.079187 0.730169 0.015955 0.137717 0.184304 0.375281 17.074816
8 0.310877 70.57716 20.401228 0.330260 0.120080 0.722513 0.013992 0.144631 0.194569 0.419226 17.076343
9 0.332613 70.42705 2 0.440570 0.343948 0.162169 0.723344 0.013478 0.146085 0.200372 0.461578 17.081402
10 0.353263 70.29720 20.481974 0.350348 0.201365 0.724066 0.012238 0.149376 0.210477 0.488857 17.084102
1 0.087395 100.0000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000 0.000000
2 0.130853 74.35845 25.00083 0.169084 0.063336 0.249291 0.056673 5.73E-05 0.015324 0.003965 10.08298
3 0.170888 74.07583 25.08213 0.210320 0.151004 0.260412 0.067889 0.071403 0.009058 0.013847 10.05810
4 0.200639 73.91041 25.06713 0.208223 0.140833 0.232576 0.149281 0.114483 0.080100 0.046427 10.05053
5 0.228146 73.36040 25.05225 0.334346 0.171296 0.384731 0.198527 0.116988 0.070455 0.209062 10.10193
6 0.251440 72.85983 25.05138 0.483718 0.211823 0.461448 0.248267 0.096316 0.075465 0.401673 10.11008
7 0.272403 72.41273 25.07048 0.585694 0.414078 0.473728 0.263102 0.097604 0.065593 0.506023 10.11096
8 0.292613 71.84532 25.11079 0.536605 0.866225 0.467039 0.267483 0.109727 0.058930 0.607852 10.13001
9 0.312471 71.23209 25.16030 0.483560 1.349822 0.463842 0.254317 0.124232 0.055452 0.733107 10.14327
10 0.332569 70.60522 25.19070 0.429863 1.850939 0.469308 0.239178 0.156563 0.053518 0.849822 10.15488
Notes: DV: The closing of the Road Silk by FBI, which amounts 1 from 23/10/2013 and 0 otherwise.
32
Figure A-3: Impulse Response Function
Response of BPI to BPI Response of BPI to TTR Response of BPI to ETR Response of BPI to MBV Response of BPI to EOV Response of BPI to HASH Response of BPI to GP Response of BPI to OP Response of BPI to DJI Response of BPI to SI
1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of TTR to BPI Response of TTR to TTR Response of TTR to ETR Response of TTR to MBV Response of TTR to EOV Response of TTR to HASH Response of TTR to GP Response of TTR to OP Response of TTR to DJI Response of TTR to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of ETR to BPI Response of ETR to TTR Response of ETR to ETR Response of ETR to MBV Response of ETR to EOV Response of ETR to HASH Response of ETR to GP Response of ETR to OP Response of ETR to DJI Response of ETR to SI
4 4 4 4 4 4 4 4 4 4
2 2 2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2 -2 -2
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of MBV to BPI Response of MBV to TTR Response of MBV to ETR Response of MBV to MBV Response of MBV to EOV Response of MBV to HASH Response of MBV to GP Response of MBV to OP Response of MBV to DJI Response of MBV to SI
4 4 4 4 4 4 4 4 4 4
2 2 2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2 -2 -2
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of EOV to BPI Response of EOV to TTR Response of EOV to ETR Response of EOV to MBV Response of EOV to EOV Response of EOV to HASH Response of EOV to GP Response of EOV to OP Response of EOV to DJI Response of EOV to SI
4 4 4 4 4 4 4 4 4 4
2 2 2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2 -2 -2
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of HASH to BPI Response of HASH to TTR Response of HASH to ETR Response of HASH to MBV Response of HASH to EOV Response of HASH to HASH Response of HASH to GP Response of HASH to OP Response of HASH to DJI Response of HASH to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of GP to BPI Response of GP to TTR Response of GP to ETR Response of GP to MBV Response of GP to EOV Response of GP to HASH Response of GP to GP Response of GP to OP Response of GP to DJI Response of GP to SI
2 2 2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2 -2 -2
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4
-6 -6 -6 -6 -6 -6 -6 -6 -6 -6
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of OP to BPI Response of OP to TTR Response of OP to ETR Response of OP to MBV Response of OP to EOV Response of OP to HASH Response of OP to GP Response of OP to OP Response of OP to DJI Response of OP to SI
1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of DJI to BPI Response of DJI to TTR Response of DJI to ETR Response of DJI to MBV Response of DJI to EOV Response of DJI to HASH Response of DJI to GP Response of DJI to OP Response of DJI to DJI Response of DJI to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of SI to BPI Response of SI to TTR Response of SI to ETR Response of SI to MBV Response of SI to EOV Response of SI to HASH Response of SI to GP Response of SI to OP Response of SI to DJI Response of SI to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
33
Response of BPI to BPI Response of BPI to TTR Response of BPI to ETR Response of BPI to MBV Response of BPI to EOV Response of BPI to HASH Response of BPI to GP Response of BPI to OP Response of BPI to DJI Response of BPI to SI
1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of TTR to BPI Response of TTR to TTR Response of TTR to ETR Response of TTR to MBV Response of TTR to EOV Response of TTR to HASH Response of TTR to GP Response of TTR to OP Response of TTR to DJI Response of TTR to SI
1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of ETR to BPI Response of ETR to TTR Response of ETR to ETR Response of ETR to MBV Response of ETR to EOV Response of ETR to HASH Response of ETR to GP Response of ETR to OP Response of ETR to DJI Response of ETR to SI
6 6 6 6 6 6 6 6 6 6
4 4 4 4 4 4 4 4 4 4
2 2 2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2 -2 -2
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of MBV to BPI Response of MBV to TTR Response of MBV to ETR Response of MBV to MBV Response of MBV to EOV Response of MBV to HASH Response of MBV to GP Response of MBV to OP Response of MBV to DJI Response of MBV to SI
4 4 4 4 4 4 4 4 4 4
0 0 0 0 0 0 0 0 0 0
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of EOV to BPI Response of EOV to TTR Response of EOV to ETR Response of EOV to MBV Response of EOV to EOV Response of EOV to HASH Response of EOV to GP Response of EOV to OP Response of EOV to DJI Response of EOV to SI
2 2 2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2 -2 -2
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of HASH to BPI Response of HASH to TTR Response of HASH to ETR Response of HASH to MBV Response of HASH to EOV Response of HASH to HASH Response of HASH to GP Response of HASH to OP Response of HASH to DJI Response of HASH to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of GP to BPI Response of GP to TTR Response of GP to ETR Response of GP to MBV Response of GP to EOV Response of GP to HASH Response of GP to GP Response of GP to OP Response of GP to DJI Response of GP to SI
1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of OP to BPI Response of OP to TTR Response of OP to ETR Response of OP to MBV Response of OP to EOV Response of OP to HASH Response of OP to GP Response of OP to OP Response of OP to DJI Response of OP to SI
4 4 4 4 4 4 4 4 4 4
0 0 0 0 0 0 0 0 0 0
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4
-8 -8 -8 -8 -8 -8 -8 -8 -8 -8
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of DJI to BPI Response of DJI to TTR Response of DJI to ETR Response of DJI to MBV Response of DJI to EOV Response of DJI to HASH Response of DJI to GP Response of DJI to OP Response of DJI to DJI Response of DJI to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of SI to BPI Response of SI to TTR Response of SI to ETR Response of SI to MBV Response of SI to EOV Response of SI to HASH Response of SI to GP Response of SI to OP Response of SI to DJI Response of SI to SI
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
34
Response of BPI to BPI Response of BPI to TTR Response of BPI to ETR Response of BPI to MBV Response of BPI to EOV Response of BPI to HASH Response of BPI to GP Response of BPI to OP Response of BPI to DJI Response of BPI to SI Response of BPI to DV2
2 2 2 2 2 2 2 2 2 2 2
1 1 1 1 1 1 1 1 1 1 1
0 0 0 0 0 0 0 0 0 0 0
-1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of TTRto BPI Response of TTRto TTR Response of TTRto ETR Response of TTRto MBV Response of TTRto EOV Response of TTRto HASH Response of TTRto GP Response of TTRto OP Response of TTRto DJI Response of TTRto SI Response of TTRto DV2
1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5
1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0
0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5
0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0
-0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of ETRto BPI Response of ETRto TTR Response of ETRto ETR Response of ETRto MBV Response of ETRto EOV Response of ETRto HASH Response of ETRto GP Response of ETRto OP Response of ETRto DJI Response of ETRto SI Response of ETRto DV2
8 8 8 8 8 8 8 8 8 8 8
4 4 4 4 4 4 4 4 4 4 4
0 0 0 0 0 0 0 0 0 0 0
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of MBV to BPI Response of MBVto TTR Response of MBVto ETR Response of MBV to MBV Response of MBV to EOV Response of MBVto HASH Response of MBV to GP Response of MBV to OP Response of MBV to DJI Response of MBV to SI Response of MBVto DV2
5 5 5 5 5 5 5 5 5 5 5
0 0 0 0 0 0 0 0 0 0 0
-5 -5 -5 -5 -5 -5 -5 -5 -5 -5 -5
-1 0 -1 0 -1 0 -1 0 -1 0 -1 0 -1 0 -1 0 -1 0 -1 0 -1 0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of EOV to BPI Response of EOV to TTR Response of EOV to ETR Response of EOV to MBV Response of EOV to EOV Response of EOV to HASH Response of EOV to GP Response of EOV to OP Response of EOV to DJI Response of EOV to SI Response of EOV to DV2
2 2 2 2 2 2 2 2 2 2 2
0 0 0 0 0 0 0 0 0 0 0
-2 -2 -2 -2 -2 -2 -2 -2 -2 -2 -2
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4
-6 -6 -6 -6 -6 -6 -6 -6 -6 -6 -6
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of HASHto BPI Response of HASHto TTR Response of HASHto ETR Response of HASHto MBV Response of HASHto EOV Response of HASHto HASH Response of HASHto GP Response of HASHto OP Response of HASHto DJI Response of HASHto SI Response of HASHto DV2
1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0
0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5
0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0
-0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of GP to BPI Response of GP to TTR Response of GP to ETR Response of GP to MBV Response of GP to EOV Response of GP to HASH Response of GP to GP Response of GP to OP Response of GP to DJI Response of GP to SI Response of GP to DV2
1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5
1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0
0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5
0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0
-0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of OP to BPI Response of OP to TTR Response of OP to ETR Response of OP to MBV Response of OP to EOV Response of OP to HASH Response of OP to GP Response of OP to OP Response of OP to DJI Response of OP to SI Response of OP to DV2
4 4 4 4 4 4 4 4 4 4 4
0 0 0 0 0 0 0 0 0 0 0
-4 -4 -4 -4 -4 -4 -4 -4 -4 -4 -4
-8 -8 -8 -8 -8 -8 -8 -8 -8 -8 -8
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of DJI to BPI Response of DJI to TTR Response of DJI to ETR Response of DJI to MBV Response of DJI to EOV Response of DJI to HASH Response of DJI to GP Response of DJI to OP Response of DJI to DJI Response of DJI to SI Response of DJI to DV2
1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0
0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5
0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of SI to BPI Response of SI to TTR Response of SI to ETR Response of SI to MBV Response of SI to EOV Response of SI to HASH Response of SI to GP Response of SI to OP Response of SI to DJI Response of SI to SI Response of SI to DV2
1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0
0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5
0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Response of DV2 to BPI Response of DV2 to TTR Response of DV2 to ETR Response of DV2 to MBV Response of DV2 to EOV Response of DV2 to HASH Response of DV2 to GP Response of DV2 to OP Response of DV2 to DJI Response of DV2 to SI Response of DV2 to DV2
1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5 1 .5
1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0
0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5 0 .5
0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0 0 .0
-0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5 -0 .5
2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10 2 4 6 8 10
Notes: DV: The closing of the Road Silk by FBI, which amounts 1 from 23/10/2013 and 0 otherwise.
35

Literature Review 3

Uploaded by

Copyright:

Available Formats

Literature Review 3

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Literature Review 3

Uploaded by

Copyright:

Available Formats

M PRA

Munich Personal RePEc Archive

What Does Crypto-currency Look Like?

Jamal Bouoiyour and Refk Selmi

CATT, University of Pau, France, ESC, University of Manouba,

26. August 2014

2. Brief literature survey

LBPI t a 0 a1 LTTRt 2 LETR a3 LMBVt a4 LEOVt a5 LHASH t 6 LGPt 7 LSI t t (1)

3.1.The ARDL Bounds Testing Method

3.2.The innovative accounting approach and VEC Granger causality

4. Results and discussion

LBPI LTTR LETR LMBV LEOV LHASH LGP LSI

Before proceeding ARDL estimation, we determine the degree of integration of

Table-2: Results of ADF and PP Unit Tests

Table-6: Gregory-Hansen Structural Break Cointegration Test

CUSUM 5% Significance CUSUM of Squares 5% Significance

CUSUM 5% Significance CUSUM of Squares 5% Significance

-3 -.4 -0.5 -6 -20

-5 -1.2 -.25 -.12 -4

.8 .10 .20 .8 1.2

-.4 -.05 -.05 -.2 -0.4

-6 -.3 -3 -.3 -.10

-.10 -.1 -.3 -1.2 -.3

4.2.Innovative accounting approach results

Table-7: Short-run and long-run Analysis

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0

0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

-.2 -.2 -.2 -.2 -.2 -.2 -.2 -.2

-.4 -.4 -.4 -.4 -.4 -.4 -.4 -.4

Response of E TR to B P I Response of E TR to TTR Response of E TR to E TR Response of E TR to MB V Response of E TR to E OV Response of E TR to HA S H Response of E TR to GP Response of E TR to S I

.01 .01 .01 .01 .01 .01 .01 .01

.00 .00 .00 .00 .00 .00 .00 .00

-.01 -.01 -.01 -.01 -.01 -.01 -.01 -.01

Response of MB V to B P I Response of MB V to TTR Response of MB V to E TR Response of MB V to MB V Response of MB V to E OV Response of MB V to HA S H Response of MB V to GP Response of MB V to S I

0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0

-0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5 -0.5

Response of E OV to B P I Response of E OV to TTR Response of E OV to E TR Response of E OV to MB V Response of E OV to E OV Response of E OV to HA S H Response of E OV to GP Response of E OV to S I

-.2 -.2 -.2 -.2 -.2 -.2 -.2 -.2

Response of HA S H to B P I Response of HA S H to TTR Response of HA S H to E TR Response of HA S H to MB V Response of HA S H to E OV Response of HA S H to HA S H Response of HA S H to GP Response of HA S H to S I

.08 .08 .08 .08 .08 .08 .08 .08

.04 .04 .04 .04 .04 .04 .04 .04

.00 .00 .00 .00 .00 .00 .00 .00

-.04 -.04 -.04 -.04 -.04 -.04 -.04 -.04

-.08 -.08 -.08 -.08 -.08 -.08 -.08 -.08

Response of GP to B P I Response of GP to TTR Response of GP to E TR Response of GP to MB V Response of GP to E OV Response of GP to HA S H Response of GP to GP Response of GP to S I

.010 .010 .010 .010 .010 .010 .010 .010

.005 .005 .005 .005 .005 .005 .005 .005