Comparing Crypto and Fiat Currencies - An Event Study of Bitcoin and U.S. Dollar Volatility

(1)

University of Amsterdam, Amsterdam Business School

Master in International Finance

Master Thesis

Comparing Crypto and Fiat Currencies - An Event Study of

Bitcoin and U.S. Dollar Volatility

Candidate

Supervisor

Giovannini Francesco

Razvan Vlahu

Student Number

11662506

(2)

2

Abstract

This paper starts analysing the features of Bitcoin in terms of its strengths and weaknesses as a medium of exchange, unit of account and store of value. These features are analysed comparing them to the ones which have made fiat currency a standard for modern economy. Among all the weaknesses of Bitcoin, volatility has been chosen as element of analysis for the second part of this paper. The analysis has been performed with an event study, focusing on the effects of relevant events on the returns Bitcoin and the U.S. Dollar Index (USDX) presented over time. The results of the analysis showed that over each of the event windows, Bitcoin presented higher abnormal returns than the USDX, meaning that the volatility of the cryptocurrency is higher than the one of the U.S. Dollar not only in absolute terms, but also in terms of reaction to market shocks. The results obtained in this paper suggest that due to its features, with special regards to volatility, Bitcoin is not able to compete with fiat currency and impose itself as a new standard for the economic system.

(3)

3

Introduction

Economic growth and the development of trade guided the evolution from a barter-based economic system to the basis of the modern world, where the medium of exchange is mainly represented by fiat currency. The actual equilibrium characterized by the usage of fiat currency has been lasting for decades, if we only focus on the Bretton Woods agreement which based the economy on a commodity-based currency, but the reality is that fiat currency saw its origin centuries ago. The main feature of fiat currency is the total absence of any intrinsic value and a stability which is backed only by the reputation the currency has among market agents. The actual economic system is based on the fiat currency standard and the mechanism of money supply multiplication through deposit and reserves of the banks. Despite this system is embedded in our realty, the innovation has affected the currency world as well. In the last decades an increasing number of virtual currencies, the security of which is based on cryptography (i.e. cryptocurrencies), has been appearing in the market. Bitcoin represents the first one being introduced, the most common one and the most relevant in terms of market value.

In the public and scientific community there is still an ongoing debate on whether Bitcoin can be considered a currency and especially if there is a real possibility that it will substitute fiat currency as the standard which the economic system is based on. The discussion about the real nature of Bitcoin is based on whether it fulfils or not the three criteria used in scientific literature to define a currency: medium of exchange, unit of account and store of value. Beside the ongoing discussion in literature which sees two opposite positions, there is not a clear position even in terms of regulation, with different authorities all over the world treating and defining Bitcoin in different ways.

In order to shed light on this debate, the first part of this paper is represented by a literature review of the two opposite positions, presenting and analysing the features of Bitcoin in terms of their contribution to the three functions a currency should present. These features will be confronted with the ones of fiat currency, which allow it to represent an efficient medium of exchange, unit of account and store of value. The results of this first part consist of highlighting

(5)

5

strengths and weaknesses of Bitcoin as a currency, in terms of its features relatively to the ones characterizing fiat currency.

Among the features of Bitcoin, particular attention will be paid to volatility, due to its effects on each one of the three functions a currency should present. The second part of the paper will analyse the volatility of Bitcoins comparing it with the volatility of U.S. Dollar. The volatility comparison will be not performed in absolute terms, but through an event study analysis, which will be focused on measuring the effects of significant events on both Bitcoin and USD values.

The remaining structure of the paper will be divided as follows: chapter one will present a comparison between fiat currency and Bitcoin features in terms of their impact on the currency functions of medium of exchange, unit of account and store of value. Chapter two will present the event study analysis on Bitcoin price and the value of the U.S. Dollar Index (USDX), used as a proxy for the US currency value. The event study is based on the definition of significant events, time windows associated to the events over which estimating normal returns and observing actual returns and finally measuring abnormal returns as consequences of the events. The final part will present conclusions based on the literary review about Bitcoin features from chapter one and on the event study performed in chapter two.

(6)

6

1. Evolution of currency

Private property of the resources and utility maximization-oriented agents represent part of the prerequisites of an economy based on the division of labour and, hence, on continuous market transactions. Establishing a transaction-based economy creates the need for a standard of value, which can be used as a unit of measurement. This role can be taken by a numeraire, without any need for its physical existence (Schumpeter, 1970). This convention allows the evaluation process of a transaction, without solving the issue related to which equivalent should be demanded in exchange for a delivery of goods or services. Evaluating transactions using a common unit of measure allows calculation of the debt positions agents have towards their counterparties. The potential lack of interest the seller could have with regards to goods or services from the buyer, either in terms of delivery timing or in absolute terms, makes a promissory note as the most accurate instrument to conclude the transaction. A promissory note represents an entitlement to resources which could be asserted at will on any later date. This instrument creates flexibility for the counterparties of a transaction and its utility increases when it could be evenly used to settle transaction with a third party. These promissory notes can be considered the basic form of money, defined as an evidence that goods or services have been exchanged, while an equivalent has not been received yet, but can be demanded at any time (Macleod, 1855). This definition of money highlights its nature of credit, which opens to credit risk, mainly in terms of solvency, a measure which decreases the higher the number of promissory notes issued is. Credit risk is a consequence of a low reciprocal trust: if this does not represent a real issue in bilateral relations which both the parties wish to maintain, problems arise in many-sided market relations. The issue embedded in this kind of relation is the risk of free-riding behaviours by some agents, more likely to appear when transactions happen with multiple parties, outside the boundaries of any long-lasting bilateral relationship. The presence of multiple counterparties, however, could allow the circulation of the promissory notes and, consequently, ease the transactions flow. These instruments generally represent claims to unit of products or real economic services, the particularity of which could make the circulation of the notes more difficult. A solution to this problem is represented by promissory notes which claim rights on widely

(7)

7

known, standardized goods, the value of which is widely recognized by the market participants. Although this solution enhances the acceptancy of the promissory notes and provides an easy to calculate value of them, there is still a solvency problem in the final moment of exchanging the promissory note for the underlying asset. What finally defines the utility of the money, is not the value of the asset which backs this form of promissory note, but the reputation of the issuer. Accepting a promissory note circulating as money in a market means that a claim against one debtor has become a claim against the market society as a whole: each member of the market is willing to deliver goods or services in exchange for this promissory note (Spahn, 2001).

The development of a monetary economy represented an historical innovation: money allowed economic interaction in an environment characterized by imperfect information about values and lack of confidence about counterparties’ solvency. The supply of money has always been guaranteed by banks, the aim of which was maintaining a high level of trust in the medium of exchange they were providing. The increasing demand for trustworthiness, at first, forced the banks to maintain assets to back the issued money notes. Through history, the value of currencies has been linked to gold, crude oil and other goods, characterized by scarcity and widely acceptable by market agents in case they wanted to exchange their money. Time has shown to bankers, governments and market agents how the acceptance of money as a medium of exchange depends mainly on expectations with regards to its future acceptance by market agents (Spahn, 2001). This finding created the need for controlling money expected value in terms of other goods, a task which could not be solved simply by backing money.

The system resulting from the development broadly presented is composed of a central bank which controls the money supply based on its set target value in terms of other goods (i.e. target inflation). The independence from political influences and a complete divestment of any commercial banking interests, allows the central bank to carry out its task of providing currencies with price stability by controlling inflation. Central banks represent also the lender of last resort which consists of its role in providing funds to the economy when commercial banks cannot cover a supply shortage. Its independence, the role as lender of last resort and

(8)

8

the inflation target give the central bank a high level of trust among the market agents, which in the end represents the factor backing the issued currency value.

1.1 Fiat currency

Fiat money represents a currency that has no intrinsic value and that has been set as the standard for debt repayment by governmental authorities (i.e. legal tender). Fiat currency value is solely based on the issuer’s creditworthiness in the market and this differentiates it from both commodity and representative money, which have a high intrinsic value. Commodity currency is defined as money created from precious metals (i.e. gold or silver), while representative money consists of a claim on a commodity that can be redeemed in exchange for money.

Fiat money first appeared around 1000 AD in China, as an answer to the lack of metals previously used as currency. The recent history of the spread of fiat currency throughout the world, showed how it is the answer to declining physical reserves of the commodities backing representative money. The decrease in gold reserves and a supply limited by the intrinsic scarcity of the metal, caused the collapse of the Bretton Woods agreement which meant the end of the possibility to convert dollars in gold. After the US choice with regards to dollar conversion, most of the countries adopted fiat currencies, establishing a system of exchanges among them (Spahn, 2001).

Monetary theory defines currencies as having three main functions: working as a medium of exchange, providing a unit of account and representing a store of value (Mankiw, 2007). Being established as legal tender, fiat currency is commonly accepted in the markets, with a direct positive impact on its usage as a medium of exchange and unit of account. Transactions are eased by a commonly accepted currency, which allows to avoid physical goods exchange, when not needed. A legally recognized value of the currency, at the same time, provides more signalling power to prices of goods and services exchanged (Yermack, 2014).

Fiat currency ability to represent an effective medium of exchange is strongly related to one of its main features: its loss of value over time in terms of the panels of goods which composes the consumer price indexes (Putnam & Norland, 2017). This loss of value over time is what is

(9)

9

defined as inflation. Exhibit 1 shows the volatility of the USD as the increase of value in the United States Consumer Price Index from 2008 to 2018. Inflation is what drives holders to use fiat currency instead of hoarding it, starting a process of economic growth driven by consumption and investments.

Its effects on economic growth forces monetary and political authorities to align their policies, in order to reach a target level of inflation. Direct consequence is that prices of goods and services included in the pricing indexes increase at a relatively steady rate (i.e. inflation rate). Exhibit 2 compares volatility of fiat currency (expressed by the movement in the United States Consumer Price Index) and the one of gold and silver, two metals used in the past as commodity currencies or as backing commodities for representative money (both expressed by the movement in the indexes of gold and silver against USD) in the period 2008-2018. The lower volatility of fiat currency is justified by the action undertaken by authorities in order to achieve the inflation target, while the high volatility of commodities is in general explained by the economic cycles and their scarcity (Yermack, 2014).

Volatility is a feature affecting all the three functions of a currency mentioned above. The effectiveness as a medium of exchange is higher for stable and less volatile currencies. Having a more stable currency means an easier to predict future value of them, which indeed makes more convenient to exchange them for goods and services without the risk of losses driven by big decreases in value (Putnam & Norland, 2017). Being an effective store of value means that a currency should be able to maintain its economic value over time, especially in the short term (Yermack, 2014). Based on this definition, due to potential losses of economic value generated by volatility, a currency with a lower level of volatility would be more efficient as a store of value. The high volatility of a currency also affects its effectiveness as a unit of account: borrowing commodity money or issuing commodity-backed assets in periods of hyperinflation of these currency, would create high level of credit default (Putnam & Norland, 2017).

A last, but very important, characteristic of a fiat currency-based system is related to the role as lender of last resort the central banks have. Central banks do not just move the monetary

(10)

10

supply to achieve inflation and economic growth objectives. During crises, central banks provide liquidity to financial institutions and, indirectly, to the private investors.

1.2 Bitcoin

Fiat currency represents the actual standard, the use of which is spread through all the economic systems. Nonetheless, in the last decade an important innovation has appeared in the financial markets: cryptocurrencies. This kind of currency is a typical example of virtual currency for the protection of which cryptography is used. The increasing interest of investors in virtual currencies, pushed authorities to make clarifications on this topic. The ECB proposed the following definition of virtual currencies “a virtual currency is a type of unregulated, digital money, which is issued and usually controlled by its developers, and used and accepted among the members of a specific virtual community” (ECB, 2012). The EBA position on the same topic is that virtual currencies “are defined as a digital representation of value that is neither issued by a central bank or public authority nor necessarily attached to a [conventional currency], but is used by natural or legal persons as a means of exchange and can be transferred, stored or traded electronically.’’ (EBA, 2014). The definitions provided highlight the features of virtual currencies in order to distinguish them from electronic money. Electronic money is defined by its link with traditional money: it presents the same unit of account, legal foundation and regulation.

The most common cryptocurrency is Bitcoin, which has been created using a scheme proposed by Satoshi Nakamoto (Nakamoto, 2008) for a peer to peer electronic cash system. Bitcoin was the first open source virtual currency, managed by an open source software algorithm which uses the internet network to both create Bitcoin as well as to record and verify transactions. According to this algorithm, the creation of new Bitcoins is the award for the users who are able to solve pre-specified mathematical problems (i.e. mining process), which becomes more difficult and less frequent over time. High relevance has the mechanism controlling the level of Bitcoin supply: there is no entity which has the power of managing the quantity supplied, exactly the opposite of what happens with regards to fiat currency. Fixed amounts of Bitcoins are issued based on a publicly known, ex ante defined and decreasing rate, until the maximum amount will be reached (Yermack, 2014).

(11)

11

Bitcoins do not have a physical representation, they could be only stored electronically and the possibility to spend them on goods and services is conditioned to the acceptance by the supplier of this payment method. Interactions among users of Bitcoin happen anonymously and directly, without the intermediation of any third party. The only ways to obtain Bitcoins are the following: as a payment for a service provided or a good sold, by exchanging them for fiat currencies, or through the mining process (Plassaras, 2013).

Bitcoin’s circulation first took place among users from the computer world, until it started being traded on the Japanese based online platform Mt.Gox. Half of the circulation of Bitcoins in the early stage appeared on the Silk Road marketplace, an internet portal used for the illegal sale of narcotics. After the Silk Road marketplace had been closed by the US authorities, the popularity of Bitcoins started increasing. Trade of Bitcoins increased on Mt.Gox as well as on other platforms, while other cryptocurrencies started appearing on the market. At first Bitcoins appealed mainly to two categories of users: technology enthusiasts and libertarians. The former group of supporters of the first cryptocurrency was moved by the idea that the online migration of transactions would have increased Bitcoin’s value due to an increase in demand. The latter group based its enthusiasm about the currency mainly to its lack of connection with any government or central authority, the trust on which was decreasing in the period immediately after the 2007 US crisis (Yermack, 2014).

1.2.1 Bitcoin as a currency

Although Bitcoin was presented by Nakamoto as the solution to introduce a peer to peer cash transaction system, this currency has so far registered a limited use for trading (Chiu & T., 2017). This empirical finding goes along with the discussion in literature about whether Bitcoin fulfils the three functions of a currency: medium of exchange, unit of account, store of value. One side of the literature describes Bitcoin not as a currency, but as a vehicle for speculative investments, due to the fact it does not fulfil the main functions of a currency (Verde, 2013; Hanley, 2014; Yermack, 2014; Williams, 2014). An opposite line of research highlights the strong potential of Bitcoin as a global currency for the future (e.g. Plassaras, 2013; Satran, 2013; Luther and White, 2014; Folkinshteyn, Lennon, & Reilly, 2015). In the following an analysis of the literature about whether Bitcoin could be considered a currency or not will be

(12)

12

presented, highlighting its strenghts and weaknesses in terms of the three functions of a currency.

1.2.1.1 Medium of Exchange

Currencies have been introduced mainly to ease transactions by taking over the barter. The capability to act as a medium of exchange represents, indeed, a crucial feature for a currency. Literature identified several features of Bitcoin which influence its ability to serve as a medium of exchange. The features can be summarized as the following: transaction costs, anonymity and transparency, legal tender, fixed costs, network externalities, dispute resolution and credit market.

Transaction costs

The absence of intermediaries for transactions concluded using Bitcoins, represents a strong advantage for this currency. The fees embedded in Bitcoin-based transactions are meant to cover merely the costs of maintaining the systems, without any royalty due to third parties. In case of fiat currencies-based transactions, the fees are meant to cover the costs related to authorities providing several services, such as validation and security of the transactions (EBA, 2014). The average fee charged on Bitcoin transfers lays between 0 and 1% of the transaction value, while the same fee varies between 2 and 5% for fiat currency-based online transactions (Folkinshteyn, Lennon, & Reilly, 2015). Based on the results presented above, it is possible to conclude that the lower transaction costs could be considered as an advantage of Bitcoin as a currency.

Anonymity and transparency

Bitcoin had its first appeal among criminal and illegal activities because of its characteristics which make its use advantageous for this kind of activities. The main features in this direction are that Bitcoin payment transactions do not require any personal identity information and that this currency eases international transfers. The protection of customer privacy helped Bitcoin to spread as a medium of exchange in gambling as well, with an increasing number of Bitcoin-based casinos appearing on the internet. The Bitcoin system has sometimes been

(13)

13

referred as pseudonymous, because of all transactions’ history and Bitcoin creation being preserved in a public ledger (i.e. Blockchain). Being all the history of transactions reported on the ledger, a privacy break to a single transaction can expose the entire system to anonymity breach. Discovering a private information related to a user, as the address used to buy Bitcoins, could allow to track the entire history of this user’s transactions through the Blockchain (Bohme, Christin, Edelman, & Moore, 2015). Anonymity of users and transparency of transactions could be considered two advantages in terms of usability of Bitcoins; at the same time, anonymity should be considered as a risk factor, in terms of its suitability to be used as a mean of exchange in illegal activities.

Legal tender

Fiat currencies are legal tender, which means that everyone is obliged to accept them as a medium of exchange. At the opposite, a currency which is not legal tender, as it is the case for Bitcoins, has a limited power as a medium of exchange: its acceptance merely depends on voluntary adoption by market participants (EBA, 2014). Being not legal tender should be considered as a limit for Bitcoins, given the possibility for the counterparty of a transaction to refuse it as medium of exchange.

Fixed costs

Introducing a new currency in a system, requires relevant initial investments from both the sell side and the buy side of a transaction. Costs embedded in the adoption of Bitcoin as a medium of exchange are in general related to obtaining knowledge about the system, with particular regards to the adoption of the required technology. The Bitcoin system is based on a dedicated software, the usage and understanding of which requires IT knowledge. This knowledge requirement could represent a potential constraining factor for a wider adoption of the currency, due to information asymmetry issues (Velde, 2013). It is possible to conclude that high fixed costs, caused by technical knowledge requirements, could be considered a limiting factor for Bitcoins to become an effective medium of exchange, especially in a short-term horizon.

(14)

14

Network externalities

Bitcoins, as well as many other products and services related to the tech world, present the typical issues of network externalities: the higher the number of existing users, the bigger is the incentive for new users to take part in the network (Gowrisankaran & Stavins, 1999). From a consumer’s perspective, it is less convenient to buy Bitcoins when the currency is accepted by a little number of goods and services providers. On the other hand, if only a few consumers have Bitcoins, there is a lower incentive for businesses to invest in the technology needed to process Bitcoin payments for their transactions. Network externality issues demonstrate that an important challenge for Bitcoins to become a global currency is represented by the need for actors to be incentivized to use these currencies for their transactions. Exhibit 3 shows the number of daily transactions in Bitcoins registered since the introduction of the currency. The trend shows an exponential increase of Bitcoin transactions until 2017, while 2018 shows an important decrease, with an actual number of daily transactions in the range of 0,1 million. The recent continuously upward part of the trend could be interpreted as an increase in the usage of Bitcoin as medium of exchange and a potential resolution of the network externalities issue. The actual decreasing trend could be interpreted as a decrease in the level of trust in the currency, the consequence of which is an increase in the network externalities issue. Despite Bitcoin is a recently introduced currency, the downward trend showed in Exhibit 3, could be considered a signal that the issue of network externalities is a limit for this currency to become an effective medium of exchange, especially in a short-term horizon.

Dispute resolution

The absence of any centralized authority governing the Bitcoin transaction system, means that once a transaction is concluded, it is almost irreversible. Correcting a transaction is possible only with the agreement of both parties involved. The consequence is a high level of risk in terms of protection against human errors and frauds which could occur during a transaction. Without a centralized authority with the power of regulating Bitcoins related disputes, the usage of the currency as a medium of exchange could be limited, especially among risk averse users. The dispute resolution issue can be considered as a factor limiting the effectiveness of Bitcoin as a medium of exchange.

(15)

15

Credit market

Each Bitcoin is unique and cannot be duplicated, while fiat currency amount could be increased through loans due to the fractional-reserve nature of the banking system. This feature of Bitcoins could be considered a limit to the expansion of Bitcoins (Hanley, 2014). Another relevant aspect of the credit market issue is represented by the absence of any Bitcoin-denominated loan or credit card, with a consequent impossibility to conclude purchases on credit (Yermack, 2014). In a modern society where the credit transactions rate is very high and even increasing, the limits to a credit market based on Bitcoins, could be considered as an important factor affecting the effectiveness of this currency as a medium of exchange.

1.2.1.2 Unit of account

The unit of account function of a currency is defined by its ability to measure the relative value of goods and services exchanged in the transactions throughout the economy. The effectiveness of Bitcoins as a unit of account will be analysed based on two characteristics which differ substantially from fiat currencies: divisibility and price volatility.

Divisibility

A peculiar feature of Bitcoin is represented by its infinite divisibility. Quotes in Bitcoin could potentially be represented in four or even more decimals. This feature represents indeed a positive aspect for Bitcoin in terms of potential to serve as a unit of account: higher divisibility allows the currency to express a valuation for transactions of any kind and size. These high degree of divisibility, however, represents a double-edged sword because differentials in terms of fourth or even higher decimal point level, could be misleading. Comparing relative values of a good or service in terms of others could be confusing, mainly due to the actual fiat currencies-based system, which do not present more than two decimals for consumer transactions prices (Yermack, 2014). Based on this analysis, it is possible to conclude that divisibility could increase the effectiveness of Bitcoins as a unity of account in the long term,

(16)

16

while maybe it can have a short-term downside due to the strong difference compared to the actual fiat currencies-based system.

Price volatility

Exhibit 4 shows the high level of Bitcoin prices volatility since its introduction in the market. Frequent changes in prices have direct consequences on both consumers and producers. Confusion about the real level of relative values for goods and services could be spread throughout consumers, while producers are affected both directly and indirectly by high levels of volatility. High volatility of a unit of account means a need for reviewing prices frequently which can result in goods and services under or overpriced. In case of under-priced goods and services, a supplier experiences a decrease in returns, while overpriced goods and services affect returns through a loss of competitiveness against competitors expressing prices in terms of stable currencies. The issues for suppliers are exacerbated when inputs are paid by fiat currency, while output prices are expressed in Bitcoins (Yermack, 2014). Technical innovations could solve the issue of volatility, especially for consumers transactions and a typical example is represented by instantaneous exchange facilities, which enable sellers to accept a payment in Bitcoin, without really receiving Bitcoins. This system works through a third party which acts as intermediary between the payment completed by the consumer in Bitcoins and the collection of fiat currency by the seller. The Bitcoin volatility risk in this case is completely transferred to the intermediary, who exchanges Bitcoins for fiat currency at the market price, charging a fee for bearing the risk embedded in the operation (Luther & White, 2014). It is possible to conclude that the high volatility level of Bitcoin prices represents a limit to its effectiveness as a unit of account, due to the consequent uncertainty about relative values of goods and services in the market. Financial service innovation could be seen as a way to solve this issue in the long term, but in the short-term volatility represents an important issue for Bitcoin to become a market standard.

1.2.1.3 Store of value

In order to represent an effective store of value, a currency value should be stable over time. Stability is the feature that allows using money for exchanges in different points of time. Two

(17)

17

peculiar aspects of Bitcoin will be used to analyse its effectiveness as a store of value: the non-inflationary nature of its supply and the issues related to cyber security.

Non-inflationary supply

As presented above, an important characteristic which make fiat currency as global standard is represented by its inflationary nature, meaning that its value decreases over time in terms of goods and services. In contrast, Bitcoins can be considered a protection against inflation being completely neutral to any government interference. Moreover, Bitcoin presents a maximum amount of supply (21 million units) which cannot be increased. This feature represents a potential for deflationary pressure in case the demand for this asset will not decrease (Meiklejohn, et al., 2013). Under the hypothesis of non-decreasing demand, holding Bitcoins means potential profits for the investors in the terms of capital gains. As well as the inflationary nature of fiat currency is an incentive for agents to spend the money, the expected upward trend of Bitcoin’s price represents an incentive for actors to hoard the currency. It can be concluded that the deflationary nature of Bitcoin could maybe increase their attractiveness as an investment, but it also reduces its effectiveness as a currency.

Cyber security

One of the main concerns related to the effectiveness of a currency as a store of value is, by definition, its security in terms of protection from thefts. Bitcoin is a virtual currency, the security of which lies in the cryptography system which define it as a cryptocurrency. Not having any physical representation, Bitcoin cannot be stored in any of the traditional ways. Bitcoins can be only stored online in accounts known as virtual wallets, the security of which represents the real issue. Bitcoin exchanges have represented targets for cyber-attacks and thefts, the most important of which caused the collapse of the largest Bitcoin exchange, MtGox, in 2014, with a loss of 850 thousand units of Bitcoins (Bohme, Christin, Edelman, & Moore, 2015). The effectiveness of Bitcoin as a store of value is limited by its security issue, which consequently reduce the potential for this currency to become a global standard.

(18)

18

Table 1 summarizes the effects of the analysed Bitcoin characteristics on its effectiveness as a medium of exchange, unit of account and store of value, which are the functions introduced to define a currency.

Table 1Bitcoin's strenghts and weaknesses in terms of the typical functions of a currency

Function of a currency Bitcoin strengths Bitcoin weaknesses Medium of Exchange Lower transaction costs Does not represent a legal tender Anonymity and Transparency Fixed costs necessary for adoption Network externalities issues Absence of dispute resolution system Absence of credit

Unit of Account Divisibility Low explanatory power for relative prices Price volatility Store of Value Non-inflationary

supply

Deflationary tendency Cyber-security issues

(19)

19

2. Bitcoin and USD volatility comparison

As presented in the analysis provided in the previous chapter, among the features of a currency, volatility impacts all the three functions a currency should have. In this chapter a comparison between Bitcoins and fiat currencies volatility will be presented. The analysis will not be focused on the volatility in terms of absolute return (i.e. the degree of change in the two currencies prices over time) but on the dimension of price changes after relevant events. The price changes will be measured comparing the observed data with the equilibrium prices estimated from data collected before the relevant event. This analysis will be performed using an event study methodology, the classic model of which is composed by the following main steps (Henderson, 1990):

a) Defining the events considered significant for the purposes of the analysis

b) Determining the event dates and the time line for the analysis (i.e. estimation and event windows)

c) Characterizing the normal returns, meaning the returns which could be estimated by a theoretical model, based on the data observed in the estimation window

d) Calculating the abnormal returns over the event window, defined as the difference between the observed returns and the normal returns estimated for the same period using the theoretical obtained from the estimation window data

The volatility comparison will be performed between Bitcoin price and the U.S. Dollar Index (i.e. USDX). The USDX measures the US currency ‘s value relative to its most important trading partner and, for the purposes of this work, can be considered an efficient proxy for the price of U.S. Dollar. The statistics of abnormal returns in the event windows for USDX and Bitcoin will be compared, in order to determine which of the two currencies presents the higher level of volatility. Volatility in this case is not measured in absolute terms, but the analysis is focused on significant events impact on the currency returns, in terms of abnormal returns dimension, as defined above.

(20)

20

2.1 Significant events

The definition of significant events, as those events which could potentially affect the return of the currency, is a relevant step in the event-based methodology. The choice of the events raises a relevant problem in terms of definition of the event timing. In the event study methodology, two periods of time can be distinguished: the estimation window and the event window. Over the estimation window, theoretical equilibrium models for normal returns are obtained from a regression of observed data. These models are used to estimate normal expected returns over the event window. The event window is an interval which contains the event date, the length of which depends on estimation about the market reaction to the significant event. Abnormal returns are calculated during the event window as the difference between the observed return and the expected normal return (Henderson, 1990).

In this paper, different groups of significant events have been chosen for the volatility analysis of Bitcoin and USDX, due to the different nature of the currencies. The significant events used in the event study of USDX volatility are the following:

1. Terrorist attack on 9/11 (September 11, 2001) 2. Iraq war (March 20, 2003)

3. Subprime mortgage crisis1_{(August 9, 2007)}

4. Bernard Madoff arrested due to 65 billion dollars Ponzi’s scheme (December 8, 2008) The event study on Bitcoin’s volatility has been performed based on the following significant events:

1. Chinese government ban for Financial Institutions from using Bitcoins (December 5, 2013)

1_{The date chosen to indicate the start of the subprime mortgage crisis is different from the more popular date}

when the US government allowed the investment bank Lehman Brothers to go bankrupt. The first stage of the crisis is here considered to have begun with the seizure in the banking system precipitated by BNP Paribas announcing that it was ceasing activity in three hedge funds that specialized in US mortgage debt. This was the moment it became clear that there were tens of trillions of dollars-worth of dodgy derivatives swilling round which were worth a lot less than the bankers had previously imagined. From that moment it took more than one year to arrive at Lehman bankruptcy, which should be considered the most disruptive effect, more than the starting point of the crisis.

(21)

21

2. Mt.Gox stoppage of Bitcoin trading after being hacked (February 7, 2014). Complete shut down of the platform followed with loss of thousands of Bitcoins

3. Bitcoin featured on front page of The Economist (October 31, 2015). This event is considered the first big public exposure of the cryptocurrency

4. Mike Hearn quits Bitcoin (January 14, 2016). Hearn represents one of Bitcoin’s first developers and stronger supporter, but in January 2016 he quit every kind of involvement in the cryptocurrency, justifying it with the failure of Bitcoin. A crisis of faith in the cryptocurrency started.

For both the groups of significant events, the event window has been defined as an interval (T1, T2) of 20 days, where the event date τ = T1 + 5 days. The estimation window has been

defined for both the groups of significant events as an interval (T0, T1) of 40 days. The choice

of the windows’ length is based on the following objectives: (a) having a time horizon to define normal returns longer than the one representing the event window; (b) accounting for market capability to anticipate events; (c) considering asymmetry of information which can influence detecting the effective event date; (d) accounting for market reaction time after the events; (e) obtaining event windows which are as much as possible not influenced by other significant market events. Following Morse (1984), shorter sampling interval have been chosen due to event time accurately known. In this situation a shorter sampling interval is considered having higher ability to identify the events effect (Morse, 1984). Nonetheless the effects of using short intervals and windows should not be underestimated in terms of autocorrelation of data and forecasting power of the models (i.e. standard errors in the definition of regressors coefficients). Figure 1 presents the structure for the definition of time windows associated to the events.

(22)

22

2.2 USDX normal return

The analysis of U.S. Dollar volatility will be performed using the USD Index (USDX) as a proxy for the currency’s price. The USDX is obtained as a geometric weighted average of the change in six foreign currency exchange rates against U.S. Dollar, using March 1973 as a base year. The foreign currency exchange rates used for the USDX and their weights are the followings (Liu Z. , 2006): Euro (57.6%), Japanese Yen (13.6%), UK Pound (11.9%), Canadian Dollar (9.1%), Swedish Krona (4.2%) and Swiss Franc (3.6%). The USDX presents a unique feature among all currency indexes: its composition has received only one change since 1973, a relevant adjustment needed after the introduction of Euro in 1999. The stability of USDX makes it a reference point for spot and future currency markets.

An estimation of the USDX normal return will be presented in this paper based on the equilibrium value of the index, as obtained applying the Behavioural Equilibrium Exchange Rate (BEER) model introduced by Liu &Han (2012), in a version which has been adapted to the purposes of this paper2_{. After Clark & MacDonald (1998) developed a theoretical framework}

for the BEER model, it has been widely used by international institutions, especially the International Monetary Fund to asses under or overvaluation of currencies. The BEER model is based on the econometric relationship between the exchange rate and its fundamental determinants, representing both internal and external balance proxies (Liu & Han, 2012). The model presented in this paper uses the following variables as determinants of the USDX equilibrium value:

• Money Supply (M2) as a proxy for monetary policy

• Federal Funds Rate (FFR) to account for monetary policy performed without affecting the stock of money base

2_{The model presented in this paper differs from the one introduced by Liu & Han in terms of variables considered}

to determine the USDX equilibrium level. In this paper the Foreign Exchange Reserves and GDP growth rate are excluded from the model. This choice is based on the purpose of this paper: being the analysis focused on short-term effects of the events, variables with data at least weekly updated have been considered preferable, while GDP and Foreign Exchange Reserves data do not present higher frequency than quarterly or yearly availability.

(23)

23

• Gold price change rate (GOLD) as a proxy for inflation3

The BEER model adopted in this paper to estimate the equilibrium price of USDX in logarithmic scale is presented in the following equation:

(𝑎) 𝑢𝑠𝑑𝑥 = 𝛼0+ 𝛼1𝑚2 + 𝛼2𝑓𝑓𝑟 + 𝛼3∆𝐺𝑂𝐿𝐷 + 𝜀

where the lowercase indicates the logarithmic transformation of the variable. Starting from the model presented in equation (𝑎), which provides USDX expected logarithmic value, normal returns (𝑅𝑈𝑆𝐷𝑋) can be obtained as difference between logarithmic values over time:

(b) 𝑅𝑈𝑆𝐷𝑋 = 𝑢𝑠𝑑𝑥𝑡 − 𝑢𝑠𝑑𝑥𝑡−1

2.3 Bitcoin normal return

Bitcoin is a relatively recent phenomenon with a significant degree of volatility, the peculiar nature of which makes any widely accepted market pricing model (such as CAPM for example) not applicable. Despite a widely accepted pricing model supported by empirical results does not exist, several studies (e.g. Bouoiyour, Selmi, Tiwari, & Olayeni, 2016; Buchholz, Delaney, Warren, & Parker, 2012; Ciaian, Rajcaniova, & Kancs, 2016; van Wijk, 2013) have identified three main drivers of Bitcoin price formation: supply and demand forces; attractiveness; global macroeconomic and financial developments. Based on the results presented in previous studies (Bouoiyour, Selmi, Tiwari, & Olayeni, 2016; Ciaian, Rajcaniova, & Kancs, 2016), global macro-financial developments will be excluded from the model used in this paper to estimate Bitcoin price and consequently its normal return.

The pricing model presented by van Wijk (2013) is based only on macroeconomic and financial variables such as stock exchange indexes and oil prices. These variables are considered relevant for the Bitcoin price formation based on several theories. Oil price is one of the main source of demand and cost pressure, which makes it a good indicator of inflation trends (Palombizio & Morris, 2012). Inflation can be considered an indicator of general

3_{The choice of gold price movements as a proxy for inflation is based on Le & Chang (2011), who studied the}

trivariate relationship among oil price, gold price and the USDX. Among other results, this study shows the correlation between USDX and gold price movements being stronger than the effect of oil price movements on the index value. This result supports the choice of using gold price movements as proxy for the inflation in the model used in this paper to obtain normal returns of the USDX

(24)

24

macroeconomic and financial development, which may stimulate the use of Bitcoins for trading, increasing the demand for this currency and, consequently, the price (van Wijk, 2013). The second variable, the stock exchange indexes trend, is considered representing the opportunity cost of investing in Bitcoins as an alternative (van Wijk, 2013). Despite the pricing model presented by van Wijk (2013) shows statistically significance of macrofinancial variables, other studies (Bouoiyour, Selmi, Tiwari, & Olayeni, 2016; Ciaian, Rajcaniova, & Kancs, 2016) demonstrated how this significance decreases when other variables are introduced. When the other two variables, namely demand-supply market forces and attractiveness, are introduced in regression models, the statiscal significance of macro-financial variables decreases (Bouoiyour, Selmi, Tiwari, & Olayeni, 2016; Ciaian, Rajcaniova, & Kancs, 2016). This finding is in line with the argument proposed by Yermack (2014), who defined Bitcoin as a relatively ineffective risk mitigation tool against macroeconomic trends, due to the low level of correlation.

The regression model presented in this paper to estimate Bitcoins normal return will be based on two group of variables: (1) the market forces of Bitcoin supply and demand and (2) Bitcoin attractiveness among market agents.

Market forces of Bitcoin supply and demand

According to Buchholz, Delaney, Warren, & Parker (2012), the interaction between supply and demand forces represents an important driver of Bitcoin’s price. The interaction between the demand and supply forces can be obtained by an adapted version of the model presented by Barro (1979) for the gold standard4_{. The basic assumption, following Barro’s study, is that all}

goods and services are traded in U.S. Dollars, meaning that firms need to convert Bitcoins in the fiat currency to operate in the market. Consequence of this assumption is that the stock of money base of Bitcoins will be presented in terms of its value in U.S. Dollars.

4_{The monetary quantitative model presented by Barro (1979) has been produced for an economy based on the}

gold standard. The gold standard has two main differences with the Bitcoin system, which have to be considered while evaluating the results of the model: (1) the commodity-currency demand in the gold standard system is driven by both the intrinsic value and the its future exchange value, while Bitcoins lack the intrinsic value component; (2) a commodity-currency supply has an important endogenous component, depending also on technology improvements in the mining industry, while the supply of Bitcoins is exogenous, having the algorithm already set a maximum amount

(25)

25

Total Bitcoin money supply (𝑀𝑆) is defined as the the total stock of BitCoins in circulation (B) multiplied by the value of each Bitcoin expressed in USD (𝑃𝐵).

(c) 𝑀𝑆 _{= 𝐵 ∗ 𝑃}𝐵

Bitcoin’s demand (𝑀𝐷) is presented as a function of the general price level of goods and services (P), the size of Bitcoin economy (Y) and the velocity of Bitcoin circulation (V).

(d) 𝑀𝐷 ₌𝑃 ∗ 𝑌

𝑉

Setting a demand-supply equilibrium, an equilibrium price which equals demand and supply for Bitcoins can be expressed as:

(e) 𝑃𝐵 ₌𝑃 ∗ 𝑌

𝑉 ∗ 𝐵

Equation (e) shows that a market equilibrium of demand and supply is represented by an inverse proportionality between Bitcoin’s price and both the circulating stock and its velocity; at the opposite, the price increases with both the size of Bitcoin’s economy and the general price level. In order to analyse the equilibrium price, the following proxies will be used for the supply-demand fundamentals: the dependent variable 𝑃𝐵is represented by Bitcoin’s price in U.S. Dollars (mktpr); the total stock circulating, 𝐵, is represented by the number of Bitcoins which has been already mined (totbtc); the size of Bitcoin’s economy, 𝑌, is represented by the number of transactions per day (ntran); following Ciaian, Rajcaniova, & Kancs (2016), the general price level of global economy, 𝑃, is represented by the USD/EUR exchange rate (fxrate); following Bouoiyour, Selmi, Tiwari, & Olayeni (2016), the velocity of Bitcoin circulation, 𝑉, meaning the frequency at which one unit of the currency is used to purchase products, is represented using Bitcoin days destroyed for any given transaction (btcdd).Bitcoin days destroyed is a measure obtained by multiplying the number of Bitcoins in a transaction and the number of days passed since the same Bitcoins were last spent

The logarithmic transformation of the equilibrium equation (e), provides the basis for the Bitcoin’s normal return model, as presented in equation (f), where the logarithmic transformation of the variables is express in lowercase:

(f) 𝑝_𝑡𝐵_{= 𝛽}

(26)

26

Bitcoin attractiveness among market agents

The differences Bitcoins have with regards to other currencies require an update of the equilibrium model presented above, in order to consider those differences. Bitcoins lack any intrinsic value and they represent a form of decentralized currency, features which makes their value estimation different from the one of commodity currencies and fiat ones. As already highlighted in the previous chapter, network externalities represent an important leverage for the future development of this currency. An increase in the network of Bitcoin users could enhance the currency’s effectiveness as a currency in terms of its three main functions (i.e. medium of exchange, unit of account, store of value). The lack of any central authority, the trust in which enhances the stability of a currency, along with the lack of any intrinsic value and the relevance of network externalities, switch the attention on the attractiveness of Bitcoins as a variable impacting the value of the currency. In order to account for the attractiveness of Bitcoin among the investors, the model presented in equation (f) is modified, as in the following:

(g) 𝑝_𝑡𝐵 _{= 𝛽}

0 + 𝛽1𝑝𝑡+ 𝛽2𝑦𝑡+ 𝛽3𝑣𝑡+ 𝛽4𝑏𝑡+ 𝛽5𝑎𝑡+ 𝜀𝑡

Where the independent variable 𝑎𝑡 represents the logarithm of the dimension expressing the

attractiveness of Bitcoins. Following Kristoufek (2013), the attractiveness of Bitcoins could be represented by users researches about the topic. Data provided by Google Trends (btcsearch) are used as a proxy for the attractiveness of Bitcoins in the model presented in this paper. This variable consists of worldwide daily search queries for the word “Bitcoin” on Google. The values this variable assumes over a time interval consist of a scaled range between 0 and 100, where 100 represent the highest amount of search queries in the interval of time.

Starting from the results of the model presented in equation (𝑔), normal returns (𝑅𝐵𝑇𝐶) can be obtained as the difference between Bitcoin’s logarithmic prices over time:

(h) 𝑅𝐵𝑇𝐶 = 𝑝_𝑡𝐵_{− 𝑝} 𝑡−1𝐵

(27)

27

2.4 Data and variables hypothesis testing

All data series are composed of daily observations. Table 2 summarizes data and sources used for both the models presented in equations (a) and (g) to estimate the equilibrium value of USDX and equilibrium price of Bitcoin.

Table 2 Data definition and sources

Model Variable Definition Sources

USDX equilibrium USDX U.S. Dollar Index value https://www.investing.com

M2 Money supply https://www.quandl.com

FED Federal funds rate https://www.quandl.com GOLDGROWTH Gold price growth (%) https://www.quandl.com Bitcoin equilibrium MKTPR Bitcoin price in USD https://www.blockchain.com

FXRATE Economy price level https://www.quandl.com NTRAN Size of Bitcoin economy https://www.blockchain.com BTCDD Bitcoin velocity https://www.quandl.com TOTBTC Number of Bitcoins https://www.blockchain.com BTCSEARCH Bitcoin attractiveness https://trends.google.com

Before testing hypothesis with regards to the two models presented in equations (a) and (g), an analysis of the variables with regards to their stationarity will be presented. A time series is defined as (weakly or covariance) stationary if its mean, variance and autocovariances of the series do not change over time (Brooks, 2014). A difference stationary series is defined integrated and it is denominated as I(d), where d is the order of integration. The order of integration represents the number of unit roots, meaning the number of differencing operations needed to make the series stationary. An example is a random walk, typically defined as 𝑦𝑡 = 𝑦𝑡−1+ 𝜀𝑡, where ε is a random constant. A random walk is integrated of order

one, I(1), being its first differential 𝑦𝑡− 𝑦𝑡−1= 𝜀𝑡, which is a series with an increasing over

time variance. A stationary series can be also defined as integrated of order 0. The Augmented Dickey-Fuller (ADF) test has been performed to determine the stationarity of the variables. The ADF tests the null hypothesis (H0) of non-stationarity against the hypothesis (H1) of

(28)

28

in the regression (Brooks, 2014). Including both the constant and the time trend is the choice applied for all the variables, excluding goldgrowth. Being all those variables logarithmic transformations, they display an upward trend, meaning that a time trend has to be taken into account. For the last variable (goldgrowth), only a constant has been added to the regression to perform the ADF test, being not clear whether the variable follows a linear trend. For the sake of the ADF, the number of lags used for each dependent variable is determined by the Akaike Information Criterion (AIC). Results of the ADF test on the variables used in both the models are presented in table 3.

Table 3ADF Test results for the variables included in the regressions and their first difference

Model Series Lags ADF t-statistic 5% confidence level

USDX equilibrium _logUSDX ₀ _-1.723602 _-3.412184

logM2 21 -1.387813 -3.412211 logFED 15 -3.115094 -3.412154 goldgrowth 5 -18.84952 -2.862837 D(logUSDX) 0 -43.88382 -3.412185 D(logM2) 25 -8.378537 -3.412217 D(logFED) 14 -15.41248 -3.412154 D(goldgrowth) 18 -18.00452 -2.862848 Bitcoin equilibrium logMKTPR 11 -4.053937 -3.415312

logFXRATE 0 -1.784982 -3.417586 logTOTBTC 12 -4.385089 -3.415325 logNTRAN 20 -2.632172 -3.415373 logBTCDD 6 -6.259045 -3.415339 logBTCSEARCH 0 -2.712239 -3.415239 D(logMKTPR) 10 -6.499043 -3.415312 D(logFXRATE) 0 -3.973984 -3.131224 D(logTOTBTC) 11 -4.934049 -3.415325 D(logNTRAN) 19 -9.814768 -3.415373 D(logBTCDD) 18 -10.18778 -3.415522 D(logBTCSEARCH) 0 -28.74892 -3.415245

(29)

29

Focusing on the first difference of the variables, the ADF t-statistics are always higher in absolute values compared to the critical value with a 5% confidence. These results suggests that the null hypothesis of no-stationarity should be rejected for the first difference of the variables. Saying that all the variables are stationary in their first difference means they are integrated of order one I(1). This result fulfils the criteria for estimating any long-term relationship among variables.

The regressions presented in equations (a) and (g) have been applied at the four estimation windows in order to obtain the normal values and returns for USDX and Bitcoins. This results in two groups, each one composed of four models, the goodness of which has been tested in terms of heteroskedasticity of the residuals. Heteroskedasticity literally refers to a variance which changes over time and across observations. In presence of heteroskedasticity, the ordinary least squares (OLS) estimators are still consistent, but they are not defined as best linear unbiased estimators (BLUE).

The test performed in this work is the White’s Test, which tests the null hypothesis (H0) of no

heteroskedasticity against heteroskedasticity of any level. The White’s test statistic is defined as the product of the number of observations (Obs) and the centred R2_{from the test}

regression. White’s test statistic is asymptotically distributed as a χ2_{with degrees of freedom}

equal to the number of independent variables in the regression (n). Results of the White’s heteroskedasticity test are presented in table 4.

Table 4 White's Heteroskedasticity Test on the models for the returns over the estimation windows

Model White’s t-statistic Prob. χ2_(n)

USDX model_01 8.476354 0.0371 USDX model_02 3.609596 0.3068 USDX model_03 1.457301 0.6922 USDX model_04 3.098729 0.6848 Bitcoin model_01 10.47335 0.06290 Bitcoin model_02 11.59598 0.04080 Bitcoin model_03 4.836951 0.43610 Bitcoin model_04 3.098729 0.68480

(30)

30

Being the White’s test statistic always higher than the critical value represented by the χ2

distribution function, the null hypothesis of homoskedasticity (i.e. no heteroskedasticity) should be rejected, meaning that all the eight models used to obtain normal returns over the estimation windows present heteroskedasticity. This result shows that even if the OLS estimators analysed are still unbiased and consistent, they are no longer BLUE, meaning that estimators with a lower variance exist.

After the test of variables stationarity and models heteroskedasticity, the equilibrium models will be presented in detail and used to calculate the abnormal returns.

2.5 Abnormal returns calculation and discussion of results

Performing OLS regressions based on data from each of the eight estimation windows, eight equations have been obtained as models for the equilibrium values (i.e. four equations for USDX value and four for Bitcoin price). Each of these equations has been used to calculate normal returns expected in the event windows. The results of the OLS regressions are reported in Exhibit 5.

Applying equations (b) and (h), normal logarithmic expected returns have been obtained for both the observed values of USDX and Bitcoins and for the estimated figures. The difference between observed return and estimated return over the event window represents what it is defined as Abnormal Return (AR). Exhibit 6 and exhibit 7 show the calculation of ARs for each of the four event windows related to, in order, both USDX and Bitcoin. Exhibit 6 and exhibit 7 report also a graphical representation of the comparison between observed and estimated logarithmic returns (i.e. LogReturns), accompanied by the AR trend over the estimation window.

Analysing the results of the event study performed in this paper, it is possible to appreciate how the level of the abnormal returns is higher for Bitcoins than for the USDX. Bitcoin abnormal returns show both higher average value and standard deviation compared to ARs measured for USDX. Table 5 represents for both USDX and Bitcoin models a summary of the main statistics: average, standard deviation and the maximum absolute value measured.

(31)

31

Table 5 Absolute Returns statistics for the models applied to each of the event windows

Model Average Standard

deviation Maximum absolute value USDX model_01 -0.10% 0.87% 1.61% USDX model_02 -0.04% 0.52% 1.14% USDX model_03 -0.24% 0.30% 0.90% USDX model_04 0.32% 1.09% 2.28% Bitcoin model_01 8.15% 9.05% 26.07% Bitcoin model_02 1.58% 3.73% 7.75% Bitcoin model_03 -0.90% 7.36% 18.37% Bitcoin model_04 1.62% 4.55% 10.47%

Based on the results of this event study, it is possible to conclude that Bitcoins returns are more affected by significative events. The effects of the shocks appear to move the returns from the equilibrium trend they used to show before the events, towards an equilibrium characterized by different relationships among the variables. This result appears as a further confirmation of the higher stability of fiat currencies, in this case represented by the U.S. Dollar through its index value, compared to cryptocurrencies, represented in this analysis by Bitcoin price. The stability of fiat currency in absolute terms, as it has been presented in exhibit 2, is enhanced by the results of this event study, which show how the value of the U.S. Dollar is less affected by the effects of significative events than Bitcoin price is. The event study performed shows how the models obtained for the USDX over the estimation periods do not provide too much divergent results when applied to the event windows, meaning that the shocks generated by the significant events do not strongly affect the relationship among the variables.

One of the reasons for the higher resilience of USDX value compared to the Bitcoins could be related to one of the features analysed in the first chapter as an element differentiating fiat currencies from cryptocurrencies: the presence of a central authority. Being the Bitcoin system out of the control of a central authority, deprives the currency of an entity able to amortise the effects of the shocks. The role of the central banks as lender of last resorts, at the opposite and as strongly demonstrated in the States after the subprime mortgage crisis of

(32)

32

2007, provides the economic system with a mechanism to exit from the crises, with a consequent enhancing effect on the currency used in that system.

The role of the central authority is also relevant in terms of maintaining and enhancing trust in the currency. Fiat currency value is completely backed by the trust the market agents put in the central authority. The concept of trust is related to the network and the security of it when the analysis moves to cryptocurrencies. Among the shocking events for the Bitcoins, the platform Mt.Gox being hacked and the moment when Hearn left Bitcoin industry demonstrate how the lack of a central authority brings a more fragile trust in the currency, potentially affected by several elements.

(33)

33

Conclusions

The definition of currency, for the purposes of this paper, has been linked to the three main functions literature use to address to this item. Bitcoin and USD comparison has started with regards to the features that affect these three functions, in order to individuate strengths and weaknesses of Bitcoin as a currency. The medium of exchange function is positively affected by the lower transaction costs related to Bitcoins and the anonymity and privacy this currency offers. With regards to the medium of exchange function, Bitcoin presents weaknesses due to: the fact it has not been recognized as legal tender; the fixed costs necessary to adopt a Bitcoin-based payment system; the absence of a dispute resolution mechanism. The unit of account function is positively influenced by the theoretically infinite divisibility of Bitcoin, but at the same time there is a low level of relative price comparability and a high level of volatility which negatively affect Bitcoin’s effectiveness as a unit of account. In terms of effectiveness as a store of value, the non-inflationary supply represents a strength of Bitcoin, but issues related to a deflationary pressure and cyber security represent relevant aspects which cannot be underestimated.

The result of the literature review comparing Bitcoin features with the ones which made fiat currency the standard for the actual economy, has highlighted strong limits of the cryptocurrency. These weaknesses represent the reasons why Bitcoin cannot be considered an alternative to fiat currency, under the actual scenario.

The analysis performed in the second part focused on the volatility of Bitcoin compared to the one of the most important fiat currency, the U.S. Dollar. The volatility comparison has been performed in terms of reaction of the two currencies to events considered significant, meaning events which can potentially affect their values. After having defined two groups of four significative events, one for Bitcoin and one for the U.S. Dollar Index, normal return models for the currencies have been obtained by an OLS regression over each estimation window. The models have been used to calculate normal returns over the event windows, which have then been compared to the observed returns, for both Bitcoin and USDX. The differences between the observed returns and the normal returns over the event windows

(34)

34

(i.e. abnormal returns), have represented the effect of the events on both the currencies values.

The results of the event analysis showed that the abnormal returns are more relevant when they refer to Bitcoin, meaning that significant events move the cryptocurrency from the equilibrium trend it used to follow before the event occurred. This result has enforced the conclusion that volatility is a feature which strongly differentiate Bitcoins from fiat currency. One of the reasons of this different volatility behaviour could be identified in the anarchy that characterizes Bitcoins, meaning the absence of any central authority. The role of the central authority is to enhance trust in the currency, with effects on its volatility, especially in periods of market shocks. The central authority role as lender of last resort is a typical example of how this category of actors has the power to influence the economic system and make the currency more stable mainly using monetary policy instruments. This feature does not characterize Bitcoins, which are consequently more strongly affected by market shocks.

In the light of the comparison between Bitcoin and U.S. Dollar, the increasing attention central banks are paying on cryptocurrencies could be interpreted as a signal of innovation. This increasing interest should not be only seen as the necessity for establishing an effective regulation for this new phenomenon, but also as the possibility to create a hybrid currency, in order to exploit the advantages of Bitcoins, without suffering from the downsides this currency actually presents.

(35)

35

Exhibit 1

(36)

36