(no title)

This chain of blocks, which contains all transaction history, is constantly sent to network participants to inform them of the new operations. In this sense, Nakamoto compared the digital currency to a stream of digital signatures. The ordinary least squares method can be used in the VAR to consistently estimate the coefficients. The optimal choice of the lags to be added to the regression mitigates the existence of the serial correlation in the residuals. The findings may show to policymakers and monetary authorities that society is afraid about losing their purchasing power in period of crisis, so people are looking for buying safe haven assets, and Bitcoin might be one of them. Understanding these interests is fundamental to investors, who want to diversify their portfolio, and to governments, which can establish alternatives to avoid having their currencies depreciated against Bitcoin. The lag order selection criteria results in two lags, according to Akaike Information , Hannan-Quinn , Schwarz and Final Prediction Error . Based on this, two lags were selected for parameterization of the Johansen cointegration test , as shown in Table 2.
This research is based on previous studies that used the same methodology and similar variables of attractiveness. The result of the VEC model and the significance of the coefficients demonstrate that the increase in Bitcoin interest, as measured by the number of searches for the keyword bitcoin , is followed by an increase of Bitcoin price. The bidirectional relationship exists and demonstrates that price Granger-causes the behavior of lnbtc and lncrash, intensifying the understanding that there is a speculative driver in Bitcoin’s transactions. The coefficients for the variable Δlnbtct-1 in the equation Δlnpricet are positive for all currencies and are significant at the level of 5% for six of twelve.
Bitcoins priced in different sovereign currencies follow global price behavior and are quickly adjusted by changing interest in currencies around the world and by crisis events. There are authors who report that they find no consistent evidence regarding the causal relationship between macroeconomic variables and the Bitcoin price. When including demand and attractiveness variables in their model, Ciaian et al. concluded that there was no significant statistical relevance of macroeconomic factors such as the Dow Jones index and oil prices and suggested speculation was the primary driver of price. Polasik et al. concluded that the correlation between Bitcoin returns and the fluctuations of sovereign currencies was weak and statistically insignificant. Al-Khazali et al. argued via a GARCH model that Bitcoin is weakly related to macro-developments due to low predictability for Bitcoin return and volatility after macroeconomic news surprises. According to Al-Khazali et al., the cryptocurrency acts more like a risky asset than a safe haven instrument. When analyzing the regression of the dependent term Δlnbtct, the independent variable Δlnpreçot-1 is significant at the 5% level. It is inferred, therefore, that a 1% increase in the Bitcoin price is followed in the following period by a weekly increase of around 0.92% of searches for Bitcoin. With these results, it is possible to establish a bidirectional dynamic between lnbtc and lnprice. It is suggestive that the intensification of the interest of the population in Bitcoin influences positively the value of the currency and the reverse also seems to be coherent, that is, the increase of the price intensifies the number of searches.
The terms searched in the tool were bitcoin and bitcoin crash covering the whole world and all categories. The result of these two surveys generated two curves with weekly values that will represent the btc and crash variables. Peaks in Google Trends searches for the term bitcoin crash as shown in Graph 1 is a graphical representation of negative news events that have had an intense and negative impact on Bitcoin’s price. Details of these events will attempt to demonstrate that bitcoin pricing seems to be highly sensitive to such sudden events. Ciaian demonstrated that the increase in the number of available bitcoins was related to a decrease in its price, while the increase in the number of addresses accompanied an increase in price. Considering that the amount of currency offered by the Bitcoin platform is finite and known, Buchholz et al. stated that fluctuations in the Bitcoin price occurred mostly because of shocks in the demand curve. In addition to the factors highlighted above, there are others that measure the size of the Bitcoin market and cause a direct shock to the curve. Such examples include the volume variables of daily transactions and transfers by network users. The equilibrium point of the supply and demand curve determines the Bitcoin price in a brokerage firm. However, what is peculiar about this digital currency is that the supply curve is known and pre-determined since there is a definitive limit on the quantity of virtual money offered in the market.

The curve obtained is described in Graph 2, which is about the impact of crises on Bitcoin pricing. Based on the weekly return calculation of this curve, we selected the five largest positive returns for determining the crisis dummy variable. The purpose is to analyze whether, during the five biggest positive changes caused by the increase in the number of searches for crisis news, the Bitcoin price also increased. If the database week corresponded to one of those times of greatest variation, the dummy crisis for that week was equal to 1, otherwise the value was zero. The Bitcoincharts platform is also a quantitative analysis tool that provides the Bitcoin price. However, it details the data by date, by sovereign currency, by brokerage and by volume; therefore, it is possible to have greater detail of the behavior of the price in different regions and even to analyze the spread between different countries.

Bitcoin Price Table, 2010

Some authors have verified in their research that macro-financial variables do not have a statistically significant influence on Bitcoin pricing in the long term (Bouri et al. 2017; Chao et al. 2019; Ciaian et al. 2016a; Polasik et al. 2015). The price of gold, much compared to Bitcoin, also does not seem to be related to Bitcoin pricing (Bouoiyour and Selmi 2015; Kristoufek 2015). However, in the short term, economic factors seem to have a significant impact, as in the U.S. dollar quotation (Dyhrberg 2016; Zhu et al. 2017) and in the Chinese market represented by the Shanghai index (Bouoiyour and Selmi 2015; Kristoufek 2015). The sum of the alphas allows to infer that α1 e α2 contribute proportionally with 59% and 41%, respectively, for long-term dynamics. Zhu et al. is one of the most recent studies about the impact of macroeconomic-financial factors on Bitcoin pricing. The author used some of the variables that affect gold pricing to identify those that have the same effect on Bitcoin pricing.

• Chao et al. say that TBML is seriously concerned by emerging markets and developing economies in a way that regulations and methods to monitor and fight against it have been created.
• Therefore, the local price is denominated in the respective sovereign currency, based on the negotiations between users of these currencies.
• According to Graph 3, a long-term relationship between lnprice and lnbtc curves was observed, with the aspect of cointegration between them.
• It is suggestive that the intensification of the interest of the population in Bitcoin influences positively the value of the currency and the reverse also seems to be coherent, that is, the increase of the price intensifies the number of searches.

Once the daily BCX curve was obtained, the average daily price for weekly data aggregation was computed. The average daily price for 1 week, therefore, represents each observation of the price variable in the overall analysis of the survey. The increasing realization of Bitcoin transactions tends to stimulate its adoption by other economic agents, boosting the demand for bitcoins. Ciaian et al. noted that the size of the bitcoin economy’s impact on demand tends to grow over time. The expectation is that the more frequent the use of money, the greater the demand and, consequently, the higher the price for bitcoins . Polasik et al. cited e-commerce as a major driver of payment systems that do not involve banking institutions and, in this sense, payment service providers aid in the development and adoption of virtual currencies.

Bitcoin Price Chart, 2010

According to Graph 3, a long-term relationship between lnprice and lnbtc curves was observed, with the aspect of cointegration between them. The choice to use the VEC model is due to the need to understand if there is an influence of the attractiveness factor in the price and if the price also explains the factors of attractiveness in the short and long term. In this way, the model allows the construction of equations in which the price is dependent variable while, in the other equation, it is independent, plus price information lagged. The insertion of the error correction term allows understanding the long term dynamics, which would not be possible in a process of differentiation of the variables. Table 7 presents the coefficients of the cointegration matrix β1 and β2 of the error correction term by sovereign currency. The biggest change in the number of online news stories about an economic crisis—a 99% increase from one week to the next—occurred during the week of June 28, 2015. According to G1 , Greece had failed to pay part of its indebtedness to the International Monetary Fund .

The study defined Bitcoin as an investment asset rather than as a currency, because of its sensitivity to variations in macroeconomic indices. The study also noted that there was evidence of Granger causality in relation to gold price and dollar index factors as applied to the dependent variable Bitcoin price. In VAR analysis, therefore, n variables are established to compose the model, which will contemplate n equations so that each variable is dependent on one of the equations and independent on the others. Each equation has as independent variable lags of the dependent variable itself and lags of the other variables, plus an error term. The objective of this model is to understand how past data influences the values of the dependent variable in the present. The initial hypothesis of the research is that attractiveness factors influence the Bitcoin price at both global and local levels, updating previous studies of attractiveness pricing.
The vector Ɛt refers to independent, random and uncorrelated disturbances (Ɛt ~ i.i.d. (0; In)). Searches on electronic media for information about what Bitcoin is and how it works may be a variable that explains demand increases for the coin and, consequently, its price. Some authors sought to estimate a relationship between the search history of the term Bitcoin on platforms such as Google (Kristoufek 2013; Buchholz et al. 2012; Bouoiyour and Selmi et al. 2015; Polasik et al. 2015; Nasir et al. 2019), Wikipedia , Twitter and online forums (Kim et al. 2017). Based on this behavior, Dyhrberg said that bitcoin could be used as a hedging product for the dollar exposure in the short term and as an additional instrument for market analysts to protect against specific risks. It should be noted that the dollar quotation against other currencies was negatively correlated with the Bitcoin price, not only in the short term but also in the long run, according to Van Wijk and Zhu et al. .

Vector Autoregressive Model

In this sense, there is no consensus among scholars about using of the term currency when referring to Bitcoin. Some relevant aspects of Bitcoin differ from traditional fiduciary currencies that will be analyzed. On the other hand, a sudden increase of lncrasht-1 generates a positive error which, when multiplied by α2, generates a decrease of Δ lnprice. The histogram of the residuals of the model shows a concentration of the near zero observations with progressive reduction of the frequency along the tails. In order to verify the existence of serial correlation in the residuals of the model, the tests of Portmanteau and Breusch & Godfrey were applied. The test results showed that the null hypothesis of no serial correlation cannot be rejected at the significance level of 5%. For stability analysis of the model, the eigen values were obtained and they are contained within the unit circle, confirming the stability of the model. Based on the case of series with a unit root, if each element of a vector of time series Xt, stationary only after the first differentiation, generates by linear combination βXt a stationary process with finite variance, they are said to be cointegrated.

The independent variable Δlncrasht− 1, which identifies the moments in which there was some negative event for the virtual currency community, resulted in the equation Δlnpricet, a negative and significant coefficient, in both specifications. This is an intuitive result since negative events tend to be accompanied by an increase in market mistrust and, consequently, a fall in price. This variable is fundamental, while allowing the model to identify the moments in which there is an intense fall of the price and better adjusting the curve of the model to the valleys. It is estimated that a positive variation of 1% in the number of searches for bitcoin crash is followed in the following period by a decrease of 0.06% in the price when only short term is analyzed. It is anticipated that the hypotheses and a feedback effect between endogenous variables will be confirmed.

Although the attractiveness variable, represented by quantification of searches and use of the term bitcoin in certain relevant sites, was of great value for predicting the price of the currency for some authors, it is limited by the horizon of long-term analysis. Ciaian et al. , when analyzing a database with a higher data history between 2009 and 2015, indicated that online searches were better predictors of punctual returns in the early years of bitcoin. With the consolidation of the currency, we can see a reduction in the relevance of this prediction. Hayes believed that searches for the term bitcoin would lessen with the spread of knowledge about the currency and make the variable unsatisfactory for inclusion in predictive models. Bouoiyour and Selmi’s analysis also did not find evidence of the impact of Google searches on price in the long run. However, legal issues may compromise Bitcoin’s role as a medium of exchange since sovereign governments has authority to prohibit its adoption by their populations and emphasize negative aspects such as cyberattacks and virtual crimes—all characteristics that are cited by an analysis by Ciaian et al. . While investigating fraudulent activities at the MtGox brokerage firm aimed at leveraging the Bitcoin price, Gandal et al. highlighted threats to the Bitcoin network, such as Ponzi schemes, theft of Bitcoin brain wallets, and malware.

Therefore, variations in the factors that determine and directly impact the demand curve enable the high volatility of this currency over time. In this sense, research seeks to use the variables that directly influence demand to predict currency pricing. The analysis of VECM results, summarized in Table 5, shows that the coefficient of the independent variable Δlnbtct-1, in the regression Δlnpreçot, is positive, equal to 0.07 and significant only to level of 10%. In this sense, it is inferred that a 1% increase in Google searches for the term bitcoin may be accompanied in the following period by a weekly increase of 0.07% of the current price of the digital currency. It is interesting to note that most published studies give important prominence in their analyses to attractiveness factors, such as the variable number of searches over time using the term bitcoin in Google Web Search.

Because it is a virtually mined coin and with peculiar characteristics, there is a certain unfamiliarity with its modus operandi, even to those who use in their day-to-day interactions with the Internet. Bitcoin it is not simple to understand since this is a new technology based on encryption and codifications that are more technically familiar to information technology professionals. Virtual money use has increased as a medium of exchange in the e-commerce environment where major brands such as Microsoft and Subway have offered it as a payment method in online purchases. The speed and low cost of transferring Bitcoin, the anonymity of the transference, and the transparency of transactions recorded in the blockchain are positive aspects that promote adoption of Bitcoin as cash. The transactions recorded and confirmed are inserted into a block that becomes part of the blockchain, through a process known as mining.
Another striking difference concerns to divisibility since the coin can be denominated beyond two decimal places . Yermack stated that the market can be disconcerted about the use of multiple decimal places, hindering price comparisons by the consumer. At the 5% confidence level, the feedback effect is confirmed and states that Δlnprice Granger-causes Δlnbtc and Δlncrash. The converse is also true, that is, Δlnbtc and Δlncrash Granger-cause Δlnprice, as presented in Table 6. The cajorls function for extracting VEC model regressions is then applied based on the determination of amount of error correction terms. In all of them the transformation of the natural logarithm was applied to minimize problems of heteroscedasticity and to make the model estimators less sensitive to unequal estimates . In order to denote this change, the prefix ln will be appended to the variable names .
Also, cryptocurrencies could be illegally used to facilitate Trade-based Money Laundering schemes and it can be justified by the easy way the digital coins are transferred. Chao et al. say that TBML is seriously concerned by emerging markets and developing economies in a way that regulations and methods to monitor and fight against it have been created. The volume variable, according to Bouoiyour and Selmi , impacts Bitcoin pricing in the short term. Balcilar et al. emphasized that the variable can predict returns, except in up- or down-market periods. Therefore, under normal market conditions, investors have transacted volume as a prediction tool; in contrast, during stress scenarios, an association between the variable and price returns is not identified.

Factors that make the asset extremely volatile to information and market variables include the absence of a centralized institution that controls and guarantees the value of Bitcoin and the understanding that its price is based on the belief that the virtual currency will continue its upward trajectory. It seems that there are yet opportunities to get benefits from Bitcoin volatilities and its market inefficiencies (Bouri et al. 2018). It is important to highlight that this inefficiency is getting weaker over time since liquidity seems to have a positive effect on the informational efficiency of Bitcoin prices . Engle and Granger sought to check the cointegration between the variables by obtaining an equation that represents the long-term relationship between them, using a least squares methodology. Once the relationship has been established, the residues are extracted and the Dickey-Fuller test is applied to verify their stationarity. Concerning the unit of account function, Ciaian et al. highlighted the high volatility of Bitcoin pricing as costly from the point of view of the virtual re-mark of goods and services prices denominated in Bitcoin monetary units. This function is the main differentiating factor between Bitcoin and sovereign currencies.

The lack of regulation is also an unfavorable criterion, since it eliminates judicial settlements of disputes and makes it difficult to obtain reimbursement from operations prejudiced against cryptocoins. In November 2017, the Central Bank of Brazil – Bacen said that does not regulate or supervise virtual currencies even though it monitors related discussions in international forums. In addition, the bank emphasized the imponderable risks of this type of investment to the market, including the loss of all invested capital. He has worked for Google, NASA, and consulted for governments around the world on data pipelines and data analysis. Disappointed by the lack of clear resources on the impacts of inflation on economic indicators, Ian believes this website serves as a valuable public tool. Based on Engle and Granger , the cointegration is characteristic of a series vector Xt, with the same order of integration d, whose linear combination results in a process with integration order d minus b, according to Eq. Bitcoin emerged at a time of massive expansion of the Internet, search engines, and social networks.

Read more about Dragonchain exchange here. In its turn, the Johansen uses VAR model and the matrix of its coefficients to determine its rank and, in this way, estimates the cointegration vectors. VAR is an autoregressive model composed of p-lags and a vector with n endogenous variables. The methodology differs from univariate models by analyzing the trajectory caused by a structural shock on endogenous variables. According to Zhu et al. , the influence of the USDI was negative, possibly because a valuation of the U.S. dollar currency against other currencies is also applicable to the virtual currency Bitcoin. Therefore, it was inferred that at the moment of U.S. dollar appreciation, there would be a devaluation of the Bitcoin price denominated in dollars. In the second half of 2014, for example, there was a continuous increase in the USDI caused by the resumption of the U.S. economy and, at the same time, there was a significant drop in the Bitcoin price. In addition, investment in virtual currencies can generate interest income, including through available platforms, such as BitPass, that offer interest payments to customers who leave their bitcoins stored for a certain period of time. This study seeks to advance knowledge of how Bitcoin prices are set by the market, to identify the relevant variables affecting the Bitcoin market, and to provide a technical reference for the investors that believes in its appreciation over time and invests in this asset. The findings are also relevant for policymakers and monetary authorities in order to understand why people are seeing increasingly their interests to trade or hold Bitcoin.

Understanding these interests is fundamental to create alternatives to avoid governments having their currencies depreciated against Bitcoin. The generalization of VAR model, order p, with the addition of exogenous variables is given by Eq. Being ɸ the coefficient vector n x g and D is the matrix containing exogenous variables. This entire technological and cryptographic framework already makes Bitcoin different from sovereign currencies, primarily because of its ability to be cited as a representation of digital value and its virtual decentralization.

These combined attractiveness factors define the interest of the world’s population in the asset, as measured by the number of Google searches for the terms bitcoin and bitcoin crash between December 2012 and February 2018. The procedure applied to BCX can be replicated to local prices specified by each sovereign currency. The objective is to check if prices traded in different currencies are also influenced by the structure of the global variables previously established. Only price observations are altered, which will be denominated in each respective currency. The expectation is that world events consistently impact the price at local brokerages. Bitcoin.com is a platform that aims to help Bitcoin stakeholders by offering news, brokerage, and quantitative analysis tools.