Central Banks' Effect on Bitcoin Returns: An Event Study
Filip Lundqvist, Christian Olivefors
Abstract:
Bitcoin has become established in the financial system and has a volatile price development. Lacking
trust in the financial system and monetary policy has led to increased adoption of Bitcoin. The
regulation of Bitcoin can be hard to succeed with. The thesis will study the impact of central bank
statements from ECB and FED on Bitcoin’s returns.
ECB and FED speeches were analyzed using an event study to see their impact on the Bitcoin price
from 2014 to 2022. Event classifications were positive, negative, and neutral. The abnormal returns
were compared to a reference index based on MA150 and MA15 of the bitcoin price close to the
event.
Few observations had statistical significance. However, the speeches’ relevance and timing could
most likely impact the abnormal returns. The impact of the statements can be questioned because
other monetary policy tools probably impact Bitcoin’s returns more. The central bank credibility can
be questioned due to Bitcoin’s purpose to counter the centralization of the financial systems.
It is not possible to reject the null hypothesis, meaning it is not possible to say if the ECB and FED
can impact Bitcoin’s returns with public statements, but the reason why is hard to define. Bitcoin’s
relationship with the financial system and the difficulty to regulate it can impact who enters the
market or incentivize further regulation.
Future research includes studying, eg., demand and supply influencing factors such as energy prices or
other monetary policy tools, regulations impacting Bitcoin as an asset instead of a currency, and by
dividing the time period into smaller segments to factor in cyclic effects.
Keywords: Event Study, Bitcoin, Efficient Market Hypothesis, Cryptocurrency, Monetary Policy
Tool, Blockchain
Bachelor’s thesis in Economics, 15 credits
Spring Semester 2022
Supervisor: Andreas Dzemski
Department of Economics
School of Business, Economics and Law
University of Gothenburg
Acknowledgement
During the spring of 2022 was this bachelor thesis written as the final initiative to end our educational
effort within economics.
We would like to thank our supervisor Andreas Dzemski for the support and guidance in creating the
thesis.
Filip Lundqvist & Christian Olivefors
Gothenburg
2022-06-09
Nomenclature
ECB - European Central Bank
FED - US Federal Reserve
MA150 - The 150 day moving average for the change of the Bitcoin price
MA15 - The 15 day moving average for the change of the Bitcoin price
Event window - The window of which an event is considered to have an impact, i.e., +- 14 days from
the event
Estimation window - The number of ways to estimate the comparison index, i.e., 150 days to 15 days
before the event.
Table of Content
Acknowledgement 1
Nomenclature 2
Table of Content 3
List of Tables 6
List of Figures 7
1. Introduction 1
1.1 Background 1
1.2 Purpose and Research Questions 2
1.3 Delimitations 2
1.4 Outline of the Study 2
2. Theoretical Framework 3
2.1 Monetary policy 3
2.1.1 European Central Bank Monetary Policy 4
2.1.2 Federal Reserve’s System Monetary Policy 4
2.1.2.1 Open market operations 5
2.1.2.2 Reserve requirements 5
Promoting Financial System Stability 5
2.2 Pricing Mechanisms 6
2.2.1 Weak form markets: 6
2.2.2 Semi-strong form markets 6
2.2.3 Strong form markets 6
2.2.4 Criticism of the efficient market hypothesis 6
2.3 Cryptocurrencies 6
2.3.1 Background of Bitcoin 7
2.3.2 Technology behind Bitcoin 8
2.3.3 Future of Bitcoin 8
3. Methodology 10
3.1 Research Design and Research Methods 10
3.1.1 Theory 11
3.1.2 Hypothesis 11
3.1.3 Data 11
3.1.4 Findings 12
3.1.5 Hypothesis Confirmed or Rejected 12
3.1.6 Revision of Theory 12
3.2 Econometrics Considerations 12
3.2.1 High-quality data 13
3.2.2 Relevant analysis 13
3.2.3 Well executed analysis 13
3.2.4 Well communicated results 13
3.3 Event Studies 13
3.3.1 Event date identification 14
3.3.2 Data/sample selection 14
3.3.3 Event window 15
3.3.4 Measuring abnormal returns 15
3.3.5 Step by step 15
Step 1: Event Identification 15
Step 2: Estimation window, event window, and post-event window selection 16
Step 3: Parameter estimation 16
Step 4: Cleaning data and calculating the event windows and estimation window 16
Step 5: Abnormal returns and regression analysis 17
Step 6: Hypothesis testing and analysis 17
3.4 Method Reflections and Critique 17
3.4.1 Research Design and Research Method 17
3.4.2 Econometric Considerations 17
3.4.3 Event study drawback 17
3.4.4 Quality of Research 18
3.4.5 Ethical considerations 18
3.4.6 Politics 19
4. Results 20
4.1 FED Statements 23
4.1.1 Negative Event 23
4.1.2 Neutral Event 24
4.1.3 Positive Event 25
4.2 ECB Statements 25
4.2.1 Negative Event 26
4.2.2 Neutral Event 27
4.2.3 Positive Event 28
5. Analysis 29
5.1 General Analysis 29
5.2 Significance of results 29
5.3 Timing and content of statements 29
5.4 Impact of statements 30
5.5 Credibility 30
5.6 Comments on the Results and Their Implications 31
5.7 Event Type and Event Window Analysis 31
6. Conclusion 32
6.1 ECB’s and FED’s Impact on Bitcoin Returns 32
7. Future Research 33
References 34

List of Tables
Table 1: Rating criteria for categorization of each event - 12
Table 2: Results of the regression of MA150 of the Bitcoin price and the daily Bitcoin price change -
20
Table 3: Results of the regression of MA15 of the Bitcoin price and the daily Bitcoin price change - 21
List of Figures
Figure 1: How monetary policy work throughout the financial system to meet its goals (Federal
Reserve, 2021) - 3
Figure 2: Division of price drivers for Bitcoin (Hayes, 2017). The topic of study is highlighted in red -
8
Figure 3: Changes in demands affect Bitcoin price in dollars when all coins have been mined. Supply
is a straight line since Bitcoin has a fixed supply of how many coins can be mined. - 9
Figure 4: Description of the Deductive Research Approach (Bryman, 2012) - 10
Figure 5: Illustration of the deductive research approach (Bryman, 2012) - 10
Figure 6: Illustration of event study timeline indicating the timespan settings for estimation window &
event window used in this study. - 14
Figure 7: Illustrating the timing of the events over time and the price of Bitcoin at the time of the
event, and the number of events from FED and ECB by event type. - 20
Figure 8: Scatter plot of the correlation between the MA150 and daily Bitcoin price change. - 21
Figure 9: Scatter plot of the correlation between the MA150 and daily Bitcoin price change. - 22
Figure 10: Visualisation of changes in abnormal returns due to negative FED statements about
Bitcoin for the 7-day and 14-day event windows. The graph below visualizes the trend in cumulative
abnormal returns (marked with a blue line) and the p-value for each day (marked with a grey column)
based on H0: μ = 0 & Ha: μ ≠ 0 for α = 0,05. - 23
Figure 11: Visualisation of changes in abnormal returns due to neutral FED statements about Bitcoin
over the event window period. The graph below visualizes the trend in cumulative abnormal returns
(marked with a blue line) and the p-value for each day (marked with a grey column) based on H0: μ =
0 & Ha: μ ≠ 0 for α = 0,05. - 24
Figure 12: Visualisation of changes in abnormal returns due to positive FED statements about Bitcoin
over the event window period. The graph below visualizes the trend in cumulative abnormal returns
(marked with a blue line) and the p-value for each day (marked with a grey column) based on H0: μ =
0 & Ha: μ ≠ 0 for α = 0,05. - 25
Figure 13: Visualisation of changes in abnormal returns due to negative ECB statements about Bitcoin
over the event window period. The graph below visualizes the trend in cumulative abnormal returns
(marked with a blue line) together with the p-value for each individual day (marked with a grey
column)  based on H0: μ = 0 & Ha: μ ≠ 0 for α = 0,05. - 26
Figure 14: Visualisation of changes in abnormal returns due to neutral ECB statements about Bitcoin
over the event window period. The graph below visualizes the trend in cumulative abnormal returns
(marked with a blue line) and the p-value for each day (marked with a grey column) based on H0: μ =
0 & Ha: μ ≠ 0 for α = 0,05. - 27
Figure 15: Visualisation of changes in abnormal returns due to positive ECB statements about Bitcoin
over the event window period. The graph below visualizes the trend in cumulative abnormal returns
(marked with a blue line) and the p-value for each day (marked with a grey column) based on H0: μ =
0 & Ha: μ ≠ 0 for α = 0,05. - 28
1. Introduction
This chapter serves as an introduction to the bachelor thesis. The following sections are included:
background, purpose and research questions, delimitations, and the outline of the study.
1.1 Background
Cryptocurrencies and blockchain technology have become more established, increasing their
influence on the financial system (Sohaib et al., 2019). The most common form of cryptocurrency and
blockchain technology is Bitcoin, which has had an intense and volatile price development in recent
years with an adoption primarily driven by young adults (Sohaib et al, 2019).
As technologies are incorporated into societies, regulations of the technologies need to increase to
ensure economic development (Castells, 2006). However, the technology behind cryptocurrencies has
several vulnerabilities that might need to be remediated to make the technology attractive for public
institutions (Chowdhury & Mendelson, 2014). According to Castells (2006), it can be one of the
causes of the development and implementation of regulations. Cryptocurrencies, however, might be
hard to regulate due to their decoupling from governments and their regulations (Luther, 2016).
The extensive adoption of Bitcoin and its close relation to the financial system has led to the value of
Bitcoin depending on the decisions made by central banks (Fama, Fumagalli, Lucarelli, 2019). The
trust in public institutions and the potentially inefficient monetary policy can be one of the causes of
Bitcoin's spread so much as it has (Fama, Fumagalli, Lucarelli, 2019). The creation of Bitcoin is
considered the result of the lack of trust in the established banking system (Saiedi, Broström, Ruiz,
2020). However, other causes for the adoption might exist (Parino et al, 2018).
Many central banks investigate cryptocurrencies and develop self-design, blockchain-based, central
bank-based alternatives to Bitcoin (Sparkes, 2021). Nevertheless, their impact on the adoption of
Bitcoin needs to be investigated as the central bank-backed alternatives do not address the problems
found in monetary policy (Fama, Fumagalli, Lucarelli, 2019) or the lack of trust in the financial
system (Saiedi, Broström, Ruiz, 2020).
The lack of clear guidelines, policies, or regulations can discourage actors from adopting
cryptocurrencies. The lack of clarity, or potential changed conditions, entails risks associated with
investing in cryptocurrencies (Mokni, 2021). However, it can be problematic for governments and
public actors to develop clear guidelines, policies, or regulations. As a result, their ability to solve
society's problems has not been sufficient. After that, confidence in public institutions diminished
(Belanche Gracia, Casaló Ariño, 2014). Given the central banks’ mandate to ensure price stability in
the economy (Gottfries, 2013, p. 257), it is relevant to study how central banks can influence and
regulate Bitcoin.
1
1.2 Purpose and Research Questions
The purpose of the thesis is to study the efficient market hypothesis in practice by studying how the
returns of Bitcoin are impacted by the statements of future guidance and policy implications of the
ECB and FED by utilizing an event study. In addition, the purpose is relevant to identifying whether
central banks can influence something meant to be decoupled from the financial system using one of
their monetary policy tools - the public statement. Therefore, the purpose is divided into the following
research questions:
- RQ1: How do the statements of future guidance and policy implications of the European
Central Bank impact the returns of Bitcoin?
- RQ2: How do the statements of future guidance and policy implications of the US Federal
Reserve impact the returns of Bitcoin?
1.3 Delimitations
The data the study will be based on is limited due to several factors: There is a large number of
different cryptocurrencies with varying maturities, adoption, and pricing mechanisms, meaning that
the scope had to be limited, which meant that Bitcoin was selected as the sole focus of the study as it
is the largest cryptocurrency with a relatively high maturity compared to others. Bitcoin was first
publicly traded in July 2010, meaning that no data on prior prices or price changes can be found. Due
to time constraints to be able to analyze the data, the study will only look at data up until April 2022.
The low number of statements before 2014 led to the analyzed time period in scope for the study is
2014-2022 in order to focus the analysis accordingly. Finally, due to the lack of large volumes of
public remarks available directly from many central banks, the study will only include public remarks
from the European Central Bank and the US Federal Reserve. Therefore, the study will be limited to
reviewing the impact of the positive and negative public remarks in speeches made by the European
Central Bank and US Federal Reserve, thereby disregarding any additional impact from other events.
1.4 Outline of the Study
The study will present a theoretical framework and background upon which the analysis will be based
upon. After the theoretical framework, the methodology is presented by describing the process for
achieving the purpose and answering the research questions. Next is the results chapter, which
presents the findings from the event study. The analysis is found after the results chapter and consists
of the analysis of the event study. The conclusions follow the analysis and include the study's overall
conclusions and the answers to the research questions. Lastly is a section of potential future research
found, describing any potential areas that have been identified to be of further interest to study.
2
2. Theoretical Framework
The theoretical framework will be used to enable the analysis in the study. The theoretical framework
consists of a description of monetary economics, monetary policy, monetary policies of ECB and
FED, financial systems, markets and pricing mechanisms, and cryptocurrencies.
2.1 Monetary policy
Monetary policy is the use of tools to influence the economy to reach the central bank goals set up.
Countries with their own currencies can conduct their monetary policy (Araujo & Ferraris, 2021). A
country’s currency will impact the trade, foreign investments, capital markets, and competitiveness of
the country, and therefore is monetary policy an essential tool for steering the economy (Eklund,
2013, p. 219).
The primary way of conducting monetary policies is by adjusting the interest rate to change the
financial conditions. However, the interest rate is just one of many other tools (Federal Reserve,
2021). Monetary policies aim to impact governments, businesses, and people by changing the
conditions of the financial landscape. The most common policy is to adjust the cost of money, and
interest rates, for businesses and people, which changes investing and consumption patterns
(European Central Bank, 2022).
The interest rate set by a central bank can influence the velocity of money and the real GDP and
thereby impact the size of investments in an economy (Gottfries, 2013, p. 261). Methods of how to
conduct monetary policies changed over time. Before the financial crisis of 2008/2009, adjusting
interest rates was the most commonly used tool to influence the investing and consumption of people
and companies (European Central Bank, 2022).
Figure 1: How monetary policy work throughout the financial system to meet its goals (Federal Reserve, 2021)
The projected path of the economy is assessed through different factors and used as a basis for
determining appropriate monetary policy interventions, as seen in figure 1. Some of the factors are the
following (Federal Reserve, 2021):
● Anticipated Factors - Adjustments to spending programs and changes in taxes by
governments
● Demand Shocks- Rapid shifts by unanticipated changes in credit standards improve
confidence in business and consumers' economic outlook, leading to increased demand.
3
● Supply Shocks - Sudden changes in the supply due to, e.g., bad harvest, natural disaster
destroying property, or pandemics which lead to reduced output and higher prices.
2.1.1 European Central Bank Monetary Policy
A country can have its own currency or be part of a monetary union such as the European Monetary
Union, sharing the currency with multiple countries (Gottfries, 2013, p. 405). The European Central
Bank is the entity that has the mandate together with union members' central banks to oversee
monetary politics within the union (European Central Bank, 2022).
The goal of monetary policy is often to apply different tools to adapt the economy to macroeconomic
changes to ensure that the 2% inflation goal is met. One of the most common tools used by the ECB is
setting the short-term rate, and there are three different types of rates that influence the cost of loans
for consumers and companies (European Central Bank, 2022).
● Refinancing operations - ECB lends funds to banks at a predefined interest rate every week.
● Deposit facility - banks make overnight deposits at a predefined interest rate lower than the
one for refinancing operations.
● Marginal lending facility - overnight credit with predefined interest rates for banks with
loans that exceed the refinancing operations loan.
Together, these rates create a rate corridor in the money market, which defines the minimum and
maximum interest rate banks lend to each other overnight (European Central Bank, 2022).
Continuous interest rate cuts have led to using the short-term interest rate as a monetary tool has
started losing its impact on adjusting economic activity. Central banks have therefore needed to start
adopting new tools (European Central Bank, 2022):
● Adjusting the limit on the number of central bank loans at a fixed interest rate a bank can get
as long as it is backed by collateral.
● Imposing negative interest rates keeps banks pressured to lend cheaply to consumers and
companies.
● Offering long term refinancing operations for banks at favorable rates
● Acquiring public and private financial assets
● It affects the sentiment of the macroeconomic landscape by providing future guidance on
economic activity and policy implications.
2.1.2 Federal Reserve’s System Monetary Policy
The central bank of the United States is called the Federal Reserve System (FED). Their purpose is to
promote policies that ensure that the U.S economy and public interest act in an effective operation.
Monetary policy can be described as FED's actions to reach the goals that have been set up by
congress. Their actions consist of different tools to promote that the U.S. reaches its goals for stable
prices and maximum employment in the American economy. (Federal Reserve, 2021)
To ensure this FED is supposed to fulfill five different functions (Federal Reserve, 2021):
● Strive to meet the stable prices in the U.S. and maximum employment goals by conducting
monetary policy for the nation.
● Mitigate and contain systemic risks that threaten the financial system's stability by proactive
work both domestically & abroad.
● Monitors financial institutions' impact on the financial system and takes actions to ensure
safety and soundness in their operations.
4
● Fosters safe and efficient environments for payments and settlement systems in the U.S.
● Via research, influence on consumer laws and regulations, and consumer-focused supervision
ensure consumer protection is functioning well.
In regular times, the essential tools are administered interest rates and open market purchases and
sales of securities. The most common tool among them is the short-term interest rate level, which
changes the sentiment for the time being and in the future. Changing the short-term rate will change
the financial conditions, which has a secondary effect on asset prices, stock prices, exchange rates,
and long-term interest rates. This is then seen in the change of behavior patterns for investment,
consumption, and production but also rates of employment and inflation (Federal Reserve, 2021).
Short-term interest rates refer to short-term debt security like U.S. Treasury bills or commercial paper
for one year or less. The rate can change by either FED trying to manipulate public sentiment by
announcements or changing the federal funds rate target range (Federal Reserve, 2021).
Changes in the short-term rate do not affect the longer-term rates since they can be seen as short-term
rates interpreted over a more extended period. Therefore, long- and medium-term interest rates are
mainly a function of people's beliefs and expectations about the future changes in the federal fund rate
(Federal Reserve, 2021).
2.1.2.1 Open market operations
When the economy is under much pressure, FED may utilize its mandate to purchase longer-term
securities in the open market. Under normal conditions, the open market desk focuses on buying or
selling temporarily or permanently with U.S. government agencies and the treasury (Federal Reserve,
2021)
● Temporary purchase operation (repurchase agreement)
● Overnight reverse repurchase facility
● Overnight reverse repurchase agreement
● Purchases and sales of securities
2.1.2.2 Reserve requirements
Each depository institution has to hold reserves, a certain quantity of cash at the FED while managing
deposit liabilities and transaction accounts. Before the financial crisis of 2008/2009, changing the
reserve requirement was a standard tool for conducting monetary policies, which are used as much
today since banking reserves are much larger today (Federal Reserve, 2021).
Promoting Financial System Stability
One of FEDs' functions is to oversee the financial system to ensure that there is any risk that could
threaten financial stability, e.g., banking panics. The system is seen as stable as long as financial
institutions can provide household and business services and goods to thrive. FED and other
regulators continuously mentor financial institutions to run proper operations to ensure that the system
stays stable. If risks threaten stability, new regulations will be implemented to mitigate the risk. It lies
in everyone's interest to have a financial system that is well functioning and run by financially
sustainable operations, which in the long term, is a vital contributor to a strong economy (Federal
Reserve, 2021).
5
2.2 Pricing Mechanisms
The efficient market hypothesis is based on the assumption that the pricing mechanisms in the market
depend on the available information (Brealey, Myers, & Allen, 2011). The availability of the
information impacts how efficient one market is, with the efficiency ranging from weak form,
semi-strong form, and strong form (Brealey, Myers, & Allen, 2011).
2.2.1 Weak form markets:
Weak form markets mean that the price today results from historical prices but that the historical
prices cannot be used to estimate future prices. Therefore, it is impossible to outperform the market by
studying the historical prices (Brealey, Myears, & Allen, 2011).
2.2.2 Semi-strong form markets
Markets with a semi-strong form mean that the price is the influence of publicly known information as
soon as the information is made available, meaning that all information is already priced into the
market, and it is impossible to predict future prices before the market (Brealey, Myers, & Allen,
2011).
2.2.3 Strong form markets
Strong formed markets are markets where all information, public and private, is priced into the market
and known by the public, resulting in that it is impossible to get an advantage compared to the other
actors on the market by using some information (Brealey, Myers, & Allen, 2011).
2.2.4 Criticism of the efficient market hypothesis
Bubbles are a common argument against the efficient market hypothesis due to bubbles relying on
overpriced assets, which should not be possible in a thoroughly efficient market (Abreu &
Brunnerheimer, 2003). Additionally, the efficient market hypothesis relies on the existence of a
market populated solely by rational actors, which is unlikely due to people’s behavior being
influenced by there, for example, knowledge and confidence (Brealey, Myers, & Allen, 2011) and
level of rationality (Malkiel, 2003). Furthermore, an efficient market would be less volatile than an
inefficient market due to the increased consensus of what the assets and securities are worth (Malkiel,
2003). The efficient market hypothesis also relies on information being available to all actors on the
market simultaneously, which according to Malkiel (2005), is very unlikely due to the imperfectness
of the information flows. Actors can seek and filter out information differently depending on their
preferences, resulting in not all information being available to everyone and people’s biases being
cemented further (Bruns, 2019).
6
2.3 Cryptocurrencies
The following subchapter presents a background of Bitcoin, what defines Bitcoin from a technological
standpoint, and what the future of Bitcoin could be.
2.3.1 Background of Bitcoin
Bitcoin is a cryptocurrency based on blockchain technology (Yaga et al, 2018). Bitcoin was created by
an unknown person or group named Nakamoto in 2008 (Kaushal, Bagga, Sobti, 2017). Before
Nakamoto (2008) created Bitcoin, there were no methods for conducting transactions without a
trusted third party. All previous methods were dependent on a central actor to review all transactions
before execution to minimize the risk of the “double-spending problem,” i.e. a limited sum of
currency is used to pay for multiple transactions (Nakamoto, 2008). Bitcoin is based on
decentralization to reduce the risk of one party having control of the whole system (Kaushal, Bagga,
Sobti, 2017) to minimize systematic risk previously found in the banking system (Nakamoto, 2008).
The decentralization is realized through having a ledger containing all information distributed to all
actors in the system (Yaga et al, 2018).
There are several cryptocurrencies on the market, but Bitcoin is the biggest one in market size, and it
has been more established in the financial system (Sohaib et al, 2019). The volatile price development
of Bitcoin has been driven by the increased adoption, the spread of information, the development of
the blockchain technology (Sohaib et al, 2019), and the lacking and changing policies and regulations
of Bitcoin (Mokni, 2021), together with the uncertainty in the economic and political landscape
(Colon et al, 2021).
The use of Bitcoin as a payment method is not common, and the legal conditions related to Bitcoin are
not clear, but it can be seen as a currency since it fulfills the uses needed for a currency in varying
degrees (Segendorf, 2014).
Bitcoin does not follow the same macroeconomic rules used to define the value of a
government-backed currency (Hayes, 2017). Instead, the price of Bitcoin is impacted by the supply
and demand of the cryptocurrency, general investing potential, and global macro-financial events
(Ciaian, Rajcaniova, & Kancs, 2016). Figure 2 shows how factors affect Bitcoin in a more intuitive
way (Hayes, 2017). In general, does the available information about Bitcoin have a significant impact
on the price (Ciaian, Rajcaniova, & Kancs, 2016).
7
Figure 2 - Division of price drivers for Bitcoin (Hayes, 2017). The topic of study is highlighted in red.
2.3.2 Technology behind Bitcoin
A blockchain is a distributed ledger of the information under the control of all actors on the
blockchain using open source frameworks (Yaga et al, 2018). The lack of a central actor deciding
what is allowed or not is replaced by a consensus model where a majority of the actors have to agree
on all decisions, facts, and changes, including what transactions are trustworthy or not (Yaga et al,
2018). A fork can happen if an agreement cannot be reached, which means that the cryptocurrency
splits into two different cryptocurrencies. It happens due to conflicting views of the information in the
ledger (Shahsavari, Zhang, Talhi, 2019). Agreements are reached through “voting” by measuring the
computing power invested in a particular alternative, meaning that as long as a majority agrees on the
“truth,” it is not possible for a malicious actor to fool the blockchain (Nakamoto, 2008).
The number of units of Bitcoin available to the public is decided by a software algorithm (Ciaian,
Rajcaniova, & Kancs, 2016). Each Bitcoin is “mined” by solving mathematical problems using the
computing power of computers (Segendorf, 2014).
The state of the blockchain and all the transactions are checked every ten minutes. To identify any
new transactions and establish the ownership of all the Bitcoins, resulting in inertia in the system
(Segendorf, 2014).
2.3.3 Future of Bitcoin
The future development and adoption of Bitcoin depend on several factors influencing the uncertainty
(Kaushal, Bagga, Sobti, 2017). First, the governance of Bitcoin is critical in how the adoption, use,
and technology will develop (Segendorf, 2014), even if the governance and regulation might be
8
complex (Luther, 2016). The central banks’ previous monopoly on money has gotten competition
through the existence of Bitcoin, which in turn has led to the will to regulate Bitcoin has increased
further (Sauer, 2015).
The long-term price of Bitcoins will be impacted by the fixed number of Bitcoins specified in the
protocol, which is 21 million (Nakamoto, 2008). This leads to the pricing of Bitcoin can be seen as a
demand shock since there is, in the long term, a fixed supply of Bitcoin (Gottfries, 2013, p.261),
which can be seen in figure 3.
Several risks that might hinder the development of Bitcoin are: One reason is the increasing difficulty
in mining Bitcoin as more Bitcoins have been mined, which will lead to it being more challenging to
have profitable mining operations (Segendorf, 2014). Another reason is the inability to provide
real-time transactions due to the nature of the blockchain (Segendorf, 2014). A lacking overall
infrastructure to enable using Bitcoin as a payment method can impact the willingness to adopt or
develop Bitcoin in the future (Sohaib et al, 2019). Other cryptocurrencies might outcompete and
replace Bitcoin by addressing the above mentioned problems (Segendorf, 2014). The overall inability
to predict changes related to Bitcoin or its surroundings can also impact the future adoption and
development of Bitcoin (Sohaib et al, 2019).
Figure 3: Changes in demands affect Bitcoin price in dollars when all coins have been mined. Supply is a straight line since
Bitcoin has a fixed supply of how many coins can be mined.
9
3. Methodology
The following chapter states how the general research design and research methods were created,
econometric considerations were taken into account, and how the event studies were conducted in
detail.
3.1 Research Design and Research Methods
The research method refers to the various data collection techniques used throughout the thesis
(Bryman & Bell, 2011, p. 168). Depending on which strategy, design, and technique the research uses,
set limitations for if the research question is possible to answer, according to Bryman (2012, p. 24).
The crypto industry is in a fluid state where it is still in its early phase and has not reached any
widespread adoption or existed for an entire business cycle (Yaga et al, 2018). This leads to Bitcoin's
opinions ranging from being a speculative asset to a monetary asset such as gold or having no intrinsic
value (Baur, Hong, Lee, 2018). The lack of consensus on how Bitcoin is affected by the changing
macroeconomic landscape, such as monetary policies, led to the thesis choosing to be based on a
deductive approach, which is illustrated in figure 4. As described by Bryman (2012, p. 43) and shown
in figure 5, the deductive approach means that theory, in this case, is currently a commonly adopted
view of what Bitcoin is being used as a basis for concluding findings and observations. The
conclusions from this thesis are based on what is known today, meaning that a similar study design
might be different when more knowledge about Bitcoin exists.
Figure 4: Description of the Deductive Research Approach (Bryman, 2012)
Figure 5: Illustration of the deductive research approach (Bryman, 2012, p. 43)
10
3.1.1 Theory
The thesis combines both qualitative and quantitative data to follow the deduction process suggested
for social studies by Bryman (2012, p. 20). The thesis started with an initial indication of how
monetary policy affects the price of cryptocurrencies. Then, as described previously, an initial
literature study was conducted to enable the formation of an opinion around what the theory says
about the supply and demand dynamics of cryptocurrencies. To limit the scope was only “Bitcoin”
selected among all cryptocurrencies to enable the study to focus its efforts and due to Bitcoin is the
biggest cryptocurrency based on its market cap (Candila, 2021). The literature was found on databases
such as Google Scholar, the University of Gothenburg’s University Library website, and websites tied
to the ECB, FED, and other central banks in Europe.
Examples of search phrases used for finding literature were: monetary policy, regulation of
cryptocurrencies, monetary economics, efficient market hypothesis, and monetary policy tools.
3.1.2 Hypothesis
The area of interest for the thesis was to study the effect the central banks’ returns has on Bitcoin
price. The initial hypothesis was created based on indications found in the literature study and used to
set the general direction of the thesis (Bryman, 2012, p. 43). Based on the dynamics and nature of
monetary policies, it was concluded that central banks most likely have an effect on managing the
uncertainties of the cryptocurrency markets, as described by Colon et al (2021), meaning that the
effect of central bank statements can be studied.
3.1.3 Data
During this step, several publicly available data sources (ECB, 2022; FED, 2022; Investing.com,
2022) were reviewed, containing numeric and string data, to understand the most suitable types of
studies for testing the research questions and hypothesis. It is important to collect data from multiple
sources to enable triangulation of the information and data (Bryman, 2012, p. 45).
In order to measure the effect of the central banks’ actions were Bitcoin price data collected from
Investing.com (2022), and the central banks’ events came from central bank speeches from the ECB
(2022) and FED (2022). The data was downloaded in the CSV format and then imported into an excel
file to format the data in such a way to enable analysis in Stata. Only speeches and price information
between July 2010 and April 2022 were downloaded for ECB and the Bitcoin price and FED between
July 2010 and June 2020. There were 1369 ECB and 504 FED speeches from 2010 to 2022 analyzed.
Speeches that either mentioned cryptocurrency, crypto, or bitcoin where 44 from ECB and 14 from
FED were sorted out for deeper analysis. Each speech was read through and was assigned a sentiment
ratio according to the convention found in Table 1.
11
Table 1: Rating criteria for categorization of each event
Positive Neutral Negative
Eg. mentioning the potential or Eg. mentioning the terms in Eg. mentioning problems or
upside of cryptocurrencies or neither positive nor negative risks with cryptocurrencies or
Bitcoin ways Bitcoin
3.1.4 Findings
The findings were first compiled and analyzed following the methodology described in the subchapter
“Event Studies.” Then, after the quantitative analysis of the event studies was finalized, a more
qualitative analysis could be conducted where the quantitative results were compared to the literature
framework. Finally, the data was formatted and labeled in such a way to enable the presentation and
analysis and relied on both quantitative and qualitative analysis to improve the quality of the research
(Bryman, 2012, p. 47).
Speeches mentioning cryptocurrency, crypto, or bitcoin were assigned dummy variables, based on the
above terms and the sentiment ratio, and then imported to Stata.
In order to properly test the hypothesis was an event study found to be the most suitable method for
analyzing the central bank’s effect on the Bitcoin price due to the speeches being categorical data.The
method chapter describes how the event study was conducted in the event study section.
3.1.5 Hypothesis Confirmed or Rejected
The combination of quantitative and qualitative analyses requires different tools to confirm or reject
hypotheses (Bryman, 2012, p. 48). For the quantitative analysis, an alpha value of 0,05 was selected
and used as a reference on the t-test, meaning a value of the t-statistic higher than 1,96 means that the
hypothesis can be rejected (Montgomery & Runger, 2013). The detailed analysis leading to a
confirmed or rejected hypothesis is described further in the event study subchapter. For the qualitative
analysis was the confirmation or rejection of the hypothesis based on whether the results had support
from the literature study and the qualitative data or not.
3.1.6 Revision of Theory
Based on the findings and hypothesis can it be relevant to iterate on or modify the literature study to
find new literature that supports the analysis on areas that were not previously identified (Bryman,
2012, p. 48).
When the findings had been analyzed, additional literature was added to support the analysis further,
primarily on the efficient market hypothesis, and FED and ECB specific monetary policy.
3.2 Econometrics Considerations
Econometric methodologies aim to quantify the effects of several observations (Franses, 2002). In
order to get a representative and trustworthy result must, several topics are achieved: high-quality
data, relevant analysis, well-executed analysis, and well-communicated results (Studenmund, 2013)
12
3.2.1 High-quality data
The data being analyzed must be of adequate quality to analyze the relationships (Studenmund, 2013).
For the data to have adequate quality, the data must be trustworthy, have enough observations, have
the correct formatting, and be treated to remove any distorting factors, e.g. outliers or errors
(Studenmund, 2013).
3.2.2 Relevant analysis
The analysis that is done must also be relevant, otherwise is it easy to identify relationships that do not
exist in real life, such as correlation without causation, which can lead to establishing a “truth” that
does not exist and thereby influence future research or decisions in the wrong direction (Studenmund,
2013). In order to mitigate this, it is crucial to only analyze relationships that are of interest and that
might have a theoretical relationship, as compared to p-hacking or similar activities (Studenmund,
2013).
3.2.3 Well executed analysis
The analysis must be done correctly in order to avoid presenting misleading conclusions. It can be
easy to analyze relationships wrongly due to omitted variable bias, meaning an almost but not entirely
correct conclusion is found due to another variable being overlooked (Studenmund, 2013). In order to
find a more correct analysis are iterations of the analysis needed (Studenmund, 2013).
3.2.4 Well communicated results
In order to document and present the results understandably and compellingly must the results follow
some guidelines (Studenmund, 2013). Firstly must the results be presented with an adequate level of
detail, not too little or too much to capture the core idea of the results and not confuse any readers
(Studenmund, 2013). Secondly, a combination of modes should be used, eg. the combination of visual
and written messages to explain the concepts as efficiently as possible (Studenmund, 2013). If the
above guidelines are not followed may the results get misunderstood, meaning the study might not
have the impact it potentially could have otherwise (Studenmund, 2013).
3.3 Event Studies
Event studies enable the analysis of the individual events' impact on a financial asset such as
cryptocurrencies or stocks. The method aims to calculate abnormal returns, i.e. the difference between
actual - expected returns estimated from regression analysis (Subhan et al, 2021). In other words,
abnormal returns can be interpreted as when actual returns deviate from the expected return for the
investment. Usually, this means that the monetary outcome will be higher or lower than expected,
which is not the case for this study. Since expected returns are based on a moving average, MA150
and MA15 in this study, abnormal returns should be interpreted as an abnormal shift in the trend. In
other words does the asset's price change quicker than the previous trend.
The event study method is commonly used in business research and helps researchers analyze the
macro's short- and long-term impact. Or micro-economic events (Subhan et al, 2021). Data analyzing
abnormal returns need to be objective, forward-looking, and not manipulated before being published
(Kaspereit, 2021).
13
The most significant advantage of event studies is isolating individual events and measuring their
impact on a financial asset (Kaspereit, 2021), such as cryptocurrencies or stocks. A broad range of
events such as a CEO getting fired, M&A announcements, or, as in this study, central bank statements
can affect the sentiment (Subhan et al, 2021).
Event studies need to be based on an initial analysis to identify and design the events so that they are
aligned with reality and provide the basis for generating new knowledge (Kaspereit, 2021). In
addition, this gives a better understanding of how the event study should be designed and results can
be interpreted (Kaspereit, 2021). Therefore, a combination of literature research and event studies has
been used to design the study.
Event studies achieve two objectives: Firstly, it gives the researcher an understanding of how
efficiently the market incorporates information by doing hypothesis testing (Subhan et al, 2021).
Secondly, it only shows the impact of publicly available information with its assumptions based on the
efficient market hypothesis (Subhan et al, 2021).
3.3.1 Event date identification
The event date referred to the date when the event of interest occurred. The event date, usually known
as t=0, is the basis of the event study for analyzing the abnormal return -/+ that date. Studying days
before and after the event gives a clear idea of if the information impacts abnormal returns and how
fast the market reacts to new information. If the event date is set wrongly, it will affect the calculation
of the abnormal returns and give a flawed estimation (Subhan, et al., 2021).
3.3.2 Data/sample selection
The data used in the event study model needs to cover the whole event timeline from the estimation
window to the post-event window see figure 6.
In order to isolate the events, it's good to visualize the events on a timeline to identify and exclude any
overlapping events. It is not always true that overlapping events should be discarded since it will
decrease the number of events being studied which might hurt the quality of results from the event
study. There are other ways to avoid overlapping events, including adjusting time windows and what
type of time units (month, week, day, etc) are being studied. (Subhan, et al., 2021)
Figure 6: Illustration of event study timeline indicating the timespan settings for estimation window & event window used in
this study.
14
3.3.3 Event window
The event window is the timeframe for when effects from the event of interest are analyzed within.
There is no strict rule for how long an event window should be or which time unit should be used. The
length is somewhat dependent on the length of the event timeline and sample size. A suggestion is to
keep it short, so it does not overlap with other event windows. (Subhan, et al., 2021)
It has been shown from previous studies that abnormal returns can happen anywhere within the event
window. Therefore, it is up to the researcher to choose an appropriate window based on knowledge of
the topic studied. This is usually done by experimental methods where the researcher tests a few
different event windows e.g -15/15 or -5/5 and determines which suits best by conducting z-test and
t-tests. (Subhan, et al., 2021)
There is a lot of disinformation and noise when it comes to discointing new information into the
returns on the market and therefore was an event window of +- 14 days selected for the thesis. The
relatively short event windows were selected due to the volatility of the Bitcoin market and the fast
transfer of information on the market.
3.3.4 Measuring abnormal returns
With abnormal returns or unexpected returns, the model tries to estimate how much more or less an
investor would make compared if the event did not happen at all. The abnormal returns are evaluated
through the event window and can be on just a daily or accumulative level. (Subhan, et al., 2021)
The formula used for calculating the abnormal returns is:
𝐴𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛𝑠  = 𝐴𝑐𝑡𝑢𝑎𝑙 𝐵𝑖𝑡𝑐𝑜𝑖𝑛 𝑟𝑒𝑡𝑢𝑟𝑛𝑠 − 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝐵𝑖𝑡𝑐𝑜𝑖𝑛 𝑟𝑒𝑡𝑢𝑟𝑛
𝑡 𝑡 𝑡
If abnormal returns > 0, then are the returns higher than expected, if = 0, are the returns as expected,
and if <0 are the returns lower than expected. (Subhan, et al., 2021)
3.3.5 Step by step
The step by step analysis used in the thesis, based on the methodology of Subhan et al (2021), is
described in detail below:
Step 1: Event Identification
The first step for conducting an event study is to decide what type of event should be studied and
collect data about the financial instrument of interest where the event is believed to have an effect.
The result should be one file of table data that shows information about each event, what date the
event occurred, and the event id to tie an event to a certain type of event. For the thesis are the types
of events positive, negative or neutral announcements. (Subhan, et al., 2021)
On the financial instrument’s side, should the data capture the whole period for when one of these
events has occurred. In this study, Bitcoin data has been used where daily returns from daily closing
prices for the timespan 2010-2022, including prices during weekends since Bitcoin is a security that
can be traded every hour of the day. The data set also includes the daily returns of a created reference
index used to calculate the expected returns. In this case, the reference index is based on the 150-day
15
moving average and the 15-day moving average of the Bitcoin price change since this study wants to
determine if central bank announcements affect the Bitcoin price changes. This means that abnormal
returns should be seen as if there is a change in the price trend for Bitcoin during the event time.
(Subhan, et al., 2021)
Step 2: Estimation window, event window, and post-event window selection
The second step includes determining the size of the estimation window which is the basis for
calculating the normal returns. This study uses the market model to understand the normal returns in
an event study. First, the model retrieves the abnormal returns from the event window by reviewing
Bitcoin returns with average market returns, in this case, the moving average (Subhan, et al., 2021).
The market model then models the expected returns by regression analysis for Bitcoin daily returns on
Bitcoin moving average day returns. The estimation window was concluded to be 150 days, due to the
need to combine the creation of a relatively stable long-term trend for a short period. To conduct
sensitivity analyses, were two reference indexes used - one consisting of MA150 and one consisting
MA15.
The step also includes determining the event window, the period analyzed for this event. It was
concluded that the event window should be +/- 14 days due to the volatility of the Bitcoin market. The
post-event window consists of the time after the event in the event window, i.e. up to 14 days after the
event. (Subhan, et al., 2021)
Step 3: Parameter estimation
When the market model for calculating expected returns has been determined with help from the
regression model in the previous step it's time to calculate the expected returns during the event
period. The calculation for expected returns, which in this study is represented by MA15 and MA150,
is based on a linear interpretation of returns from the estimation window described in the previous
step. Below is the formula for calculating expected returns. Since the estimation, the window moves
through time depending on which event is analyzed, α and β for expected returns will differ between
events. (Subhan, et al., 2021)
𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑟𝑒𝑡𝑢𝑟𝑛𝑠  =  α +  β ×  𝐵𝑖𝑡𝑐𝑜𝑖𝑛 𝑑𝑎𝑖𝑙𝑦 𝑟𝑒𝑡𝑢𝑟𝑛
𝑡 𝑡
𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑖𝑜𝑛 𝑤𝑖𝑛𝑑𝑜𝑤:  − 150 ≤ 𝑡 ≤− 15
𝑒𝑣𝑒𝑛𝑡 𝑤𝑖𝑛𝑑𝑜𝑤:  − 14 ≤ 𝑡 ≤ 14
Step 4: Cleaning data and calculating the event windows and estimation window
The data was cleaned and structured to fit the following format to conduct the event study and enable
troubleshooting and analysis of the data. Redundant data points from the data set were removed, and
the data was presented in such a way to enable Stata to process the data efficiently.
The below estimation window was used:
● estimation window: t: -150 to -15
● event window: t: -14 to 14
● pre-event window: t: -14 to -1
● post-event window: t: 1 to 14
● event: t=0
16
Step 5: Abnormal returns and regression analysis
When the data had been cleaned, the regression analysis was conducted over the estimation window to
retrieve and β for the individual events to be able to determine expected Bitcoin returns for the event
in the event window. All event types (positive, negative and neutral) are analyzed separately. (Subhan,
et al., 2021)
Daily abnormal and cumulative abnormal returns are then calculated by:
𝐴𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛𝑠  = 𝐴𝑐𝑡𝑢𝑎𝑙 𝐵𝑖𝑡𝑐𝑜𝑖𝑛 𝑟𝑒𝑡𝑢𝑟𝑛𝑠 − 𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝐵𝑖𝑡𝑐𝑜𝑖𝑛 𝑟𝑒𝑡𝑢𝑟𝑛
𝑡 𝑡 𝑡
where t refers to a specific day within the event window
𝐶𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛 =  Σ14 𝐴𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛𝑠
𝑡=−14 𝑡
The abnormal and cumulative abnormal returns are the keys to understanding an event's impact. By
combining all the abnormal returns for the corresponding event window with the same event type, the
impact can be visualized and confidence intervals for the mean be calculated to get a clearer picture of
the impact. (Subhan, et al., 2021)
Step 6: Hypothesis testing and analysis
The t-test is conducted by having a null hypothesis (H0: μabnormal returns for bitcoin = 0) with Ha: μabnormal returns
for bitcoin ≠ 0) alpha = 0,05 and tested on the mean abnormal return throughout the event window. If the
results from the t-test are higher than 1,96 then the null hypothesis is rejected (Subhan, et al., 2021).
3.4 Method Reflections and Critique
Below is a discussion of the applicability of the methodology presented to identify improvements and
limitations with the current approach.
3.4.1 Research Design and Research Method
The overall methodology follows an established pattern for social and economic research (Bryman,
2012, p. 10), meaning that the approach is relevant for this study. It was critical to have a feedback
loop going back to every stage of the methodology to improve the research approach based on the
input that had been collected (Bryman, 2012, p. 15). One challenge that can impact the results was the
ability to collect data in a large enough volume that was publicly available in a format that was
possible to compile and analyze (Studenmund, 2013). The challenge was mitigated by using multiple
sources (Bryman, 2012, p. 25), both ECB and FED, to identify any potential patterns between the
central bank statements and the price change of Bitcoin.
3.4.2 Econometric Considerations
Intuitive relationships are being studied, reducing the risk of irrelevant or misleading results
(Studenmund, 2013). However, the number of studied observations might lead to insignificant or
misleading results (Studenmund, 2013). The results, however, do point towards areas that should be
studied further to establish additional knowledge on the area.
3.4.3 Event study drawback
All models are based on some assumptions to work as analysis methods and there is no difference in
the event study methodology. For example, an event study model assumes that the markets are always
17
efficient, which means that the price always reflects all publicly available information (Subhan et al,
2021). In reality, will the information flow never be perfect, resulting in that all available information
cannot always be reflected in the asset’s price (Malkiel, 2005).
Since there is no strict way for determining how wide the event window should be this becomes an
area of interpretation and is to some extent subjective (Subhan et al, 2021). The result of the
ambiguity of the event window is that it requires some testing to find an appropriate event window in
the beliefs of the researcher.
Different models can be used for calculating the expected returns where all will give different results
on significant and abnormal returns depending on which method is used (Subhan et al, 2021). The
different models allow the researcher to affect the result using a different model. This study has used a
linear regression model to create an equation for estimating the expected returns.
The low number of events can also impact the event study since there might not be enough data points
to identify a representative result (Schmidheiny & Seigolch, 2019). It is also possible that the
speeches that were used as events might not be the most potent nor representative instruments for the
central banks’ actions.
3.4.4 Quality of Research
The quality of the research was ensured in several ways. Firstly, qualitative data was used from
formal, trusted institutions with large volumes of data made available to establish an adequate
minimum level to base the analysis. Secondly, only relationships with an intuitive relationship based
on the literature were studied to reduce the risk of identifying false positives or irrelevant
relationships. Thirdly, triangulation was achieved by combining data from the ECB, FED, and the
literature study, thereby increasing the likelihood of finding representative conclusions.
3.4.5 Ethical considerations
Since the study has been conducted as a literature study combined with quantitative methods to
answer the research question, it has only collected publicly available information. Most of the data
used in this study come from its primary source e.g ECB, FED or daily Bitcoin prices. Therefore it is
assumed that this data has been published ethically without involving any invasion of privacy or lack
of informed consent. Since this study has not involved any participants it is neglected that anyone
should have experienced adverse effects from the study.
Bitcoin price may not follow many of the traditional theories of the macroeconomy. This is why
Bitcoin in general is seen as a highly controversial asset since it is unregulated and where its use case
often appears in illegal markets. It cannot be discarded that Bitcoin per se may be involved in
deception or harm people (Lee & Choi, 2021). This study does not encourage or use Bitcoin in any
way, meaning that the sole study of the Bitcoin phenomena has no impact on legal or illegal market
activities.
To ensure that the study follows an ethical way of working and does not use methods to alter the
results, the deduction process by Bryman (2012, p. 43) is the foundation of designing the study. Any
analysis or conclusion that has been made based on real-life observations from the collected data.
18
3.4.6 Politics
Since Bitcoin and cryptocurrencies are highly debated topics within the political space (Baur, Hong,
Lee, 2018) it can be hard to determine what is true or not when it comes to the information available
(Craggs & Rashid, 2018). To avoid incorporating too many of these political views in the study, was
quantitative raw data from the primary source used as it is harder to manipulate (Pigott, 2017).
However, it is not possible to altogether avoid easily manipulated data. The likeliness of where
political biases can occur in this study is in the (positive, negative or neutral) labeling of the actual
event being studied. To determine which label each event should have the authors needed to go
through the FED and ECB speeches of interest and manually determine what label that event should
have. This risk incorporates the author's political views and biases. To mitigate this risk, prerequisites
for each label were determined before the labeling process started. The prerequisites for the different
event labels are found in table 1.
19
4. Results
In this chapter, the result of the event study is presented following the findings step in the deductive
process described in the method chapter. The results from FED and ECB are presented in two
different parts for increased visibility, even if the studies have been conducted the same way for both
entities. Every event type is then described from a statistical standpoint where the hypothesis is either
confirmed or rejected.
Figure 7: Illustrating the timing of the events over time and the price of Bitcoin at the time of the event, and the number of
events from FED and ECB by event type.
Figure 7 presented when the ECB and FED held speeches mentioning cryptocurrency, Bitcoin, or
crypto over time and the price of Bitcoin. The number of events is clustered around the hype periods
of 2017, 2018 to 2019, and 2021. FED had five positives, four neutral, and five negative events, while
ECB had six positives, ten neutral, and 28 negative events.
Variable Coefficient Std. Error t-Statistic Prob.
(rmt_ma150)
rit_bit_change 0.0096323 0.001558 6.18 0.000
cons 0.0053775 0.0001239 43.39 0.000
R-squared 0.0092 Number of obs 4138
Adj R-squared 0.0089 F(1, 4136) 38.22
Root MSE 0.00796 Prob > F 0.000
Table 2: Results of the regression of MA150 of the Bitcoin price and the daily Bitcoin price change.
20
Table 2 presents a significant correlation between the MA150 and the daily Bitcoin price change with
a p-value below 0,05. The regression had many observations, over 4000 data points, but the adjusted
R2 value is low at 0,0089.
Figure 8: Scatter plot of the correlation between the MA150 and daily Bitcoin price change.
Figure 8 shows that most of the measurements can be found around 0 on the x-axis and between
-0,005 and 0,01 on the y-axis and a low degree of correlation.
Variable Coefficient Std. Error t-Statistic Prob.
(rmt_ma15)
rit_bit_change 0.0695561 0.0039457 17.63 0.000
cons 0.0047773 0.0003139 15.22 0.000
R-squared 0.0699 Number of obs 4138
Adj R-squared 0.0607 F(1, 4136) 310.75
Root MSE 0.02015 Prob > F 0.000
Table 3: Results of the regression of MA15 of the Bitcoin price and the daily Bitcoin price change.
Table 3 for MA15 shows that daily changes in bitcoin price are also significant for alpha 0,05. The
explanatory power R2 of the model is larger than the one for MA150. On an overall level, the
explanatory power is still low. Both tests are done with an equal amount of data points.
21
Figure 9: Scatter plot of the correlation between the MA150 and daily Bitcoin price change.
Figure 9 correlation between daily changes in bitcoin price and MA15 start to show. It is clear from
the graph that most of the data points are gathered around 0 on the X- & Y-axis, with a more extensive
variation in the daily price changes.
22
4.1 FED Statements
Below is the presentation of the results of the event study of the FED’s statements. The presentation is
divided into the negative, neutral, and positive events for the 14 days event window for MA15 and
MA150.
4.1.1 Negative Event
Figure 10: Visualisation of changes in abnormal returns due to negative FED statements about Bitcoin for the 7-day and
14-day event windows. The graph below visualizes the trend in cumulative abnormal returns (marked with a blue line) and
the p-value for each day (marked with a grey column) based on H0: μ = 0 & Ha: μ ≠ 0 for α = 0,05.
Figure 10 shows the volatility in the abnormal returns for the 14-day event windows for FED’s
negative events. For MA15, the confidence interval is wide initially but gets narrower as time goes by,
and a similar pattern is found for MA150. Based on the hypothesis test on the change of abnormal
returns, were only 3 of the days’ abnormal returns significant for MA15, where the most of them
happened before the event. For MA150 were significant abnormal returns on six days distributed
fairly evenly before and after the event. The cumulative abnormal returns alternated between positive
23
and negative for MA15 ending up on the positive side. For MA150, the cumulative abnormal returns
were negative almost during the whole event window.
4.1.2 Neutral Event
Figure 11: Visualisation of changes in abnormal returns due to neutral FED statements about Bitcoin over the event window
period. The graph below visualizes the trend in cumulative abnormal returns (marked with a blue line) and the p-value for
each day (marked with a grey column) based on H0: μ = 0 & Ha: μ ≠ 0 for α = 0,05
Figure 11 shows the neutral events from FED. The abnormal returns are volatile, ranging from
positive and negative over time, with confidence intervals ranging from relatively large to reasonably
small depending on the number of days before or after the event for MA15 and MA150. For MA15,
there were two days with significant abnormal returns, both happening after the event. For MA150,
there were four days with significant abnormal returns, all happening after the event. The cumulative
abnormal returns for MA15 were initially negative but changed to positive eight days before the
event. For MA150, the cumulative abnormal returns alternated between positive and negative results
but ended with a negative value.
24
4.1.3 Positive Event
Figure 12: Visualisation of changes in abnormal returns due to positive FED statements about Bitcoin for a14 day event
window for both MA15 and MA150. The graph below visualizes the trend in cumulative abnormal returns (marked with a
blue line) together with the p-value for each individual day (marked with a grey column) based on H0: μ = 0 & Ha: μ ≠ 0 for
α = 0,05.
The positive events from FED are shown in figure 12 for the 14-day event window. Both MA15 and
MA150 have volatility in the abnormal returns, leading to narrow and wide confidence intervals
depending on the day. MA15 had three days with significant abnormal returns. MA150 had one day
with significant abnormal returns. The cumulative abnormal returns alternated between positive and
negative values as time went by and ended with a negative value for MA15 and MA150.
4.2 ECB Statements
Below is the presentation of the results of the event study of the ECB’s statements. The presentation is
divided into the negative, neutral, and positive events for the 14 days event window for MA15 and
MA150.
25
4.2.1 Negative Event
Figure 13: Visualisation of changes in abnormal returns due to negative ECB statements about Bitcoin over the event
window period. The graph below visualizes the trend in cumulative abnormal returns (marked with a blue line) together with
the p-value for each individual day (marked with a grey column)  based on H0: μ = 0 & Ha: μ ≠ 0 for α = 0,05.
Figure 13 presents the negative events from the ECB. The volatility in the abnormal returns were
found for both MA15 and MA150 with reasonably wide confidence intervals. MA15 had three days
with significant abnormal returns, with two happening before the event and one after the event, and
the case was the same for MA150. The cumulative abnormal returns were negative from 12 days
before the event for MA15 and alternating between being positive and negative for MA150.
26
4.2.2 Neutral Event
Figure 14: Visualisation of changes in abnormal returns due to neutral ECB statements about Bitcoin over the event window
period. The graph below visualizes the trend in cumulative abnormal returns (marked with a blue line) and the p-value for
each day (marked with a grey column) based on H0: μ = 0 & Ha: μ ≠ 0 for α = 0,05.
ECB’s neutral events are visualized in figure 14. Both MA15 and MA150 had volatility for the
abnormal returns and large confidence intervals. MA15 had two days with significant abnormal
returns. MA150 had no days being significant. The cumulative abnormal returns were negative for
MA15, and changing between positive and negative ended up being negative for MA150.
27
4.2.3 Positive Event
Figure 15: Visualisation of changes in abnormal returns due to positive ECB statements about Bitcoin over the event window
period. The graph below visualizes the trend in cumulative abnormal returns (marked with a blue line) and the p-value for
each day (marked with a grey column) based on H0: μ = 0 & Ha: μ ≠ 0 for α = 0,05.
Figure 15 presents the results of the ECB’s positive events. The abnormal returns of both MA15 and
MA150 varied between being positive and negative with both more narrow and wider confidence
intervals. MA15 had one day with significant abnormal returns with the significant day happening
before the event. For MA150 was one day significant. The cumulative abnormal returns alternated
between positive and negative values for MA15 and were positive almost the whole time for MA150.
28
5. Analysis
The chapter analyzes the results of the event studies using the literature study to aid the creation of the
conclusion for the thesis. The analysis will initially revolve around the general analysis of the event
studies, then on ECB and FED specific analyses, an analysis based on the event type, and lastly, an
analysis of how it would be possible to improve the significance of the results.
5.1 General Analysis
Due to the few days in the different event windows being significant for alpha 0,05, little is shown
about the impact of ECB’s and FED’s statements on the changes in the Bitcoin price. In other words,
the hypothesis testing says that for most days within the event window, we can not say with a high
degree of certainty that the abnormal returns are different from zero. A few days in the different event
windows show statistical significance for alpha 0,05. However, since these are fairly distributed over
the whole event window and do not tend to occur around the event day, no immediate conclusions can
be drawn from these data points. Even if the individual days are not distinguishable from zero, do the
cumulative abnormal returns in many cases have a negative trend over the event window. In other
words, this means that the trend in daily price changes becomes weaker than the moving average trend
over the event window. No immediate conclusions related to the research question should be drawn
from this indication. Nevertheless, it is an interesting phenomenon that could be interesting to
investigate further.
5.2 Significance of results
The lack of significant results can be explained by several factors. One factor is the relatively low
number of observations, meaning that each speech has a large impact on the results. Additionally, the
are the significant observations referring to different t’s, and not compensating for the large number of
t’s, meaning that there is a risk of false positives. Another factor is the unbalance between the event
types for ECB, where a majority of the events were negative, meaning that the study is skewed, thus
impacting the results. The imbalance between the events could be explained by the difficulty of
regulating Bitcoin, which Luther (2016) explained. The central banks’ view of the actors active in the
Bitcoin market due to the actors’ negative view of the financial system, which Fama, Fumagalli
explain, and Lucarelli (2019), and Saiedi, Broström, and Ruiz (2020). The evolving maturity of
Bitcoin, as explained by Sohaib et al (2019), might impact the results, given that the dynamics of the
Bitcoin market are not the same today as it was, eg., five years ago. One factor to consider is the cause
and effect relationship between the speeches and the changes in Bitcoin price. It is not clear whether
ECB and FED brought up the topic to influence the market or due to them “having to” comment on an
event afterward, which according to the European Central Bank (2022), could be a way to influence
the sentiment on the market. As stated by Sohaib et al (2019), Bitcoin has a volatile price
development. The low number of events, imbalance between event types, and evolving maturity
makes it hard to cope with the volatility of the Bitcoin prices.
5.3 Timing and content of statements
A general reflection is that the timing and content of the statements from the ECB and FED could be
due to the potential inefficiency of the monetary policy, as explained by Fama, Fumagalli, and
Lucarelli (2019). Given that there is a negative trend, either strong or weak, for the cumulative
29
abnormal returns for most of the studied event types when ECB or FED made a Bitcoin-related
comment in their speeches, the Bitcoin price decreased compared to the previous trends in the market.
The increased decline of the Bitcoin price could be due to the bad timing of the ECB and FED. i.e.
that the central banks are too slow to react, or that the content is not optimized, i.e. that the message is
irrelevant or counter-effective given the effect that is sought after. The potential impact of the content
and timing of the speeches is aligned with the impact of information on an efficient market, as stated
by Brealey, Myers, and Allen (2011) for semi-strong form markets, although the lack of significance
indicates no clear conclusions can be drawn from the study. Similarly could the lack of impact of the
timing and content of the speeches mean that the market is already aware of the information provided
by the ECB or FED when the speeches are made, meaning that it is already priced into the marked and
it is a strong form market, as supported by Brealey, Myers, and Allen (2011). Lastly, it is possible that
the ECB and FED makes public remarks due to new information about Bitcoin or due to recent events
rather than trying to influence the development of Bitcoin, implying a potential reverse causation in
the relationship between the speeches and Bitcoin and that the central banks try to influence the
sentiment going forward, as proposed by European Central Bank (2022).
5.4 Impact of statements
Based on the number of monetary policy tools available to the central banks, as explained by Gottfries
(2013, p. 261), it is probable that other tools have a larger impact on Bitcoin’s price development than
the speeches done by ECB and FED. The information flow’s impact on the pricing mechanisms of
Bitcoin, which is essential to an efficient market according to Brealey, Myers, and Allen (2011), might
not be too high due to several factors. As explained by Abreu and Brunnerheimer (2003), the
existence of bubbles implies that information does not always impact the price of an asset, which
could be the case for Bitcoin due to the price surge in the last couple of years. Similarly, does the
efficient market hypothesis depend on rational actors (Brealey, Myers, & Allen, 2011) which is
unlikely to be the case due to the volatility of Bitcoin. It could be interpreted as there is no consensus
of the price of Bitcoin, which according to Malkiel (2003) could imply it being an inefficient market.
The ECB and FED making the information of their intentions available through their speeches does
not necessarily impact the returns on the market as the direction of causality is not clear, either due to
the imperfectness of the information flow on the markets, as explained by Malkiel (2005), or the
inherent biases of the actors on the market, as Bruns (2019) explained. As the information is likely to
have a large impact on the price of Bitcoin, as stated by Ciaian, Rajcaniova, and Kancs (2016), is
another source of information more likely than the ECB and FED.
5.5 Credibility
The credibility of the ECB and FED on the Bitcoin market can be questioned due to Bitcoin’s
existence is a reaction to minimize the systematic risk of the banking system, as described by
Nakamoto (2008). Furthermore, relying on decentralization to minimize the control of one sole actor,
as described by Kaushal, Bagga, and Sobti (2017). Additionally, could the overall lack of trust in
public institutions and the monetary policy, as described by both Fama, Fumagalli, and Lucarelli
(2019) and Saiedi, Broström, and Ruiz (2020), explain why the statements made by the ECB and FED
does not have a significant effect on the Bitcoin price changes.
30
5.6 Comments on the Results and Their Implications
As stated above, are the results for the ECB and FED events not significant, which effectively means
that it is not possible to reject the hypothesis that the events have an effect of zero on the returns of
Bitcoin. However, the results indicate that the cumulative abnormal returns remain negative or grow
increasingly negative for the different event types, with some exceptions. The volatility of Bitcoin,
and the abnormal returns, points towards that the price and financial stability, as described by Federal
Reserve (2021), is not achieved. The lack of stability indicates that the ECB and FED might develop
new ways of influencing and regulating Bitcoin, given that Segendorf (2014) states that governance
and regulation is critical for how Bitcoin evolves, although it is hard due to the complexity of the
regulations, as stated by Luther (2016). The ECB’s and FED’s way of regulating Bitcoin is likely to be
influenced by their will to have control over the currency system, as supported by Suaer (2015).
5.7 Event Type and Event Window Analysis
When viewing the different event types, regardless of if ECB or FED made the speech, it can be seen
that, even if the results are not significant, there is no clear trend for a specific event type. The events
could result in both positive and negative cumulative abnormal returns regardless of the event types,
with the central bank and the event window being the deciding factor instead.
31
6. Conclusion
The chapter consists of the final reflections of the thesis and concludes if the research questions have
been answered and to what extent, as well as the societal implications.
6.1 ECB’s and FED’s Impact on Bitcoin Returns
The study aims to study how the statements of future guidance and policy implications of the
European Central Bank or US Federal Reserve impact the returns of Bitcoin. The analysis cannot
conclude that abnormal returns from the events are different from zero and, therefore, can the null
hypotheses not be rejected. Furthermore, there are no clear patterns of significant changes in the
abnormal returns before or after the event day. The result is similar all over the line, independent of
the event's type or entity. The answer to the research question is that statements of future guidance and
policy implications cannot be shown to have an impact on the returns of Bitcoin. The literature
however, indicates that it is likely that some impact should exist.
The answer to why central bank announcements from ECB and FED cannot be concluded to have an
impact is harder to point out. From the perspective of the efficient market hypothesis, the event study
does not provide any indications for supporting the semi-strong or strong form of information
distribution. Since the study shows that some observations are significant several days even before the
event has occurred, which is highly unlikely to be discounted by the market that far ahead of the event
day. This theory drives the conclusion that central bank statements are not in the interest of affecting
the bitcoin price. In order to provide any clear evidence of the impact of monetary policy on the
sentiment of the Bitcoin market is further research needed.
32
7. Future Research
The following chapter describes several topics that can be investigated in future research to clarify
any ambiguous or new areas.
This study does not provide evidence that central bank speeches, i.e. speaking of regulations, have any
significant effect on the returns of Bitcoin. Therefore it is suggested that further research on other
areas that affect the price from a supply and demand perspective presents in figure 2 by Hayes (2017).
One area to further expand on is investigating other tools from the monetary policy toolboxes, such as
the effect of short-term interest rates, quantitative easing, or quantitative tightening on the bitcoin
price. It would be possible to conclude whether the other tools and the general economic situation
impact the Bitcoin price changes or if the Bitcoin market is separated from the overall economic
development. The issue with investigating this approach further is that bitcoin has not existed through
a whole business cycle. Interest rates and inflation have been at historically low levels, so Bitcoin
needs to have experienced another economic climate to do a proper study to retrieve reliable results.
Another area of considerable interest in bitcoin is electrical energy consumption since proof of work is
used to validate all transactions on the blockchain (Yaga et al, 2018), as well as the voting in the
blockchain is done by measuring the computing power invested in each alternative (Nakamoto, 2008).
Analyzing how energy prices change the marginal cost for bitcoin mining activity would be an
interesting study from the supply side since the energy cost can be a big part of the marginal cost for
mining Bitcoin. Instead of focusing on how the central banks affect bitcoin prices directly, it is more
likely more interesting to understand how they indirectly affect prices through the energy market.
Supporters of Bitcoin often claim that energy used for mining activities comes from waste energy or
renewable energy on a large scale. It is possible to investigate if this statement is accurate by seeing
central banks' effect on oil prices and how those price changes transfer into the bitcoin prices.
The thesis has focused on analyzing how Bitcoin is influenced by monetary policy on the basis of
Bitcoin being a currency. To continue the research, it is possible to conduct another research effort to
investigate how Bitcoin is influenced by financial policy based on the assumption that Bitcoin is an
asset only, and not a currency. It could be possible to establish a clearer relationship between the
financial policy and the development of Bitcoin price than what was achieved for the monetary policy
speeches.
Lastly, it could be relevant to conduct another analysis by dividing the time period 2014-2022 into
smaller parts where each part represents a certain maturity of Bitcoin and its market in order to
identify and analyze the effects of the speeches over time and minimize the distorting effects of
speeches from early in the time period due to the large changes of maturity over time as it has not
existed for an entire business cycle, as state by Yaga et al (2018).
33
References
Abreu, D., Brunnerheimer, M. (2003). Bubbles and Crashes. Econometrica. 71(1), 173-204.
Araujo, L., & Ferraris, L. (2021). Social Benefits of Multiple Currencies. Journal of Money, Credit
and Banking. 53(8). 2201-2214.
Baur, D., Hong, K., Lee, A. (2018). Bitcoin: a medium of exchange or speculative asset? Journal of
International FInancial Markets, Institutions, and Money. 54(May 2018). 177-189.
Brealey, R., Myers, S., Allen, F. (2011). Principles of Corporate Finance (10th ed.). McGraw-Hill.
New York.
Bruns, A. (2019). Filter Bubble. Internet Policy Review. 8(4).
Bryman, A., (2012). Social Research Methods. 4th ed. Oxford: Oxford University Press.
Bryman, A., Bell, E. (2011). Business Research Methods. Oxford: Oxford University Press.
Candila, V. (2021). Multivariate Analysis of Cryptocurrencies. Econometrics. 9(3). 28-45.
Ciaian, P., Rajcaniova, M., & Kancs, D. A. (2016). The economics of Bitcoin price formation. Applied
economics, 48(19), 1799-1815.
Colon, F., Kim, C., Kim, H., Kim, W. (2021). The effect of political and economic uncertainty on the
crypto market. Finance Research Letters. 39 (March 2021). 101621.
Craggs, B., Rashid, A. (2018, February). Misplacing trust in bitcoin information sources. In MIS2:
Misinformation and Misbehavior Mining on the Web in Proceedings of the Eleventh ACM
International Conference on Web Search and Data Mining. ACM.
Eklund, K. (2013). Vår Ekonomi. 13 ed. Lund. Studentlitteratur.
ECB (2022). Speeches Dataset. Frankfurt am Main: European Central Bank. [Online] Available at:
https://www.ecb.europa.eu/press/key/html/downloads.en.html [Accessed April 12 2022].
European Central Bank, (2022). Monetary policy decisions. Frankfurt am Main: European Central
Bank. [Online] Available at: https://www.ecb.europa.eu/mopo/decisions/html/index.en.html
[Accessed 1 May 2022].
Fama, M., Fumagalli, A., Lucarelli, S. (2019). Cryptocurrencies, monetary policy, and new forms of
monetary sovereignty. International journal of political economy. 48:2. 174-194.
FED. (2022). Speeches of Federal Reserve Officials. Washington DC: Federal Reserve. [Online]
Available at: https://www.federalreserve.gov/newsevents/speeches.htm [Accessed April 12 2022].
Federal Reserve. (2021). The Fed Explained: What the Central Bank Does, Washington DC: Federal
Reserve.
34
Franses, P.H. (2002). A concise introduction to econometrics: an intuitive guide. New York:
Cambridge University Press.
Gottfries, N. (2013). Macroeconomics. New York. Palgrave Macmillan.
Hayes, A. S. (2017). Cryptocurrency value formation: An empirical study leading to a cost of
production model for valuing bitcoin. Telematics and Informatics, 34(7), pp. 1308-1321.
Investing.com, (2022). Bitcoin Historical Data. Nicosia: Investing.com. [Online] Available at:
https://www.investing.com/crypto/bitcoin/historical-data [Accessed April 12 2022].
Kaspereit, T. (2021). Event studies with daily stock returns in Stata: Which command to use? The
Stata Journal. 21(2). 462-497.
Kaushal, P., Bagga, A, Sobti, R. (2017). Evolution of Bitcoin and Security Risk in Bitcoin Wallets. In
2017 International Conference on Computer, Communications and Electronics (Comptelix) Manipal
University Jaipur, Malaviya National Institute of Technology Jaipur & IRISWORLD, July 01-02,
2017.
Lee, H. & Choi, K.-S., 2021. Interrelationship between Bitcoin, Ransomware, and Terrorist Activities:
Criminal Opportunity Assessment via Cyber-Routine Activities Theoretical Framework. Victims &
Offenders, 16(3), pp. 363-384.
Malkiel, B. (2003). The Efficient Market Hypothesis and Its Critics. Journal of Economic Perspective.
17(1). 59-82.
Malkiel, B. (2005). Reflection on the Efficient Market Hypothesis: 30 years later. The Financial
Review. 40(2005). 1-9.
Mokni, K. (2021). When, where, and how economic policy uncertainty predicts Bitcoin returns and
volatility? A quantiles-based analysis. The Quarterly Reviews of Economics and Finance. 80(2021).
65-73.
Montgomery, D., Runger, G. (2013). Applied Statistics and Probability for Engineers. Hoboken.
Wiley.
Nakamoto, S., (2008). Bitcoin: A Peer-to-Peer Electronic Cash System. Bitcoin.org. [Online]
Available at: https://bitcoin.org/bitcoin.pdf [Accessed 20 April 2022].
Parino, F., Beiró, M., Gauvin, L. (2018). Analysis of the Bitcoin Blockchain: socio-economic factors
behind the adoption. EPJ Data Science. 7(1). 38-71.
Pigott, T. D., (2017). The Role of Theory in Quantitative Data Analysis, Chicago: Loyola eCommons,
Education: School of Education Faculty Publications and Other Works.
Saiedi, E. Broström, A., Ruiz, F. (2020). Global drivers of cryptocurrency infrastructure adoption.
Small Business Economics. 57. 353-406.
35
Sauer, B. (2015). Central Bank Behaviour Concerning the Level of Bitcoin Regulation as a Policy
Variable. Athens Journal of Business and Economics. 1(4). 273-286.
Segendorf, B. (2014). What is bitcoin? Sveriges Riksbank Economic Review, 2014:2. 71-87.
Shahsavari, Y., Zhang, K., Talhi, C. (2019, July). A Theoretical Model for Fork Analysis in the
Bitcoin Network. In 2019 IEEE International Conference on Blockchain. 237-244. IEEE.
Sohaib, O., Hussain, W., Asif, M., Ahmad, M., Mazzara, M. (2019). A PLS-SEM neural network
approach for understanding cryptocurrency adoption. IEEE Access. 8(2019). 13138-13150.
Studenmund, A.H. (2013). Using Econometrics: A Practical Guide. Pearson. Harlow.
Subhan, U., Ghasem, Z., Rizwan, A. & Danson, K., (2021). How to apply the event study
methodology in STATA: An overview and a step-by-step guide for authors. Industrial Marketing
Management, 99(1), pp. A1-A12.
Yaga, D., Mell, P., Roby, N., Scarfone, K. (2018). Blockchain Technology Overview.
Gaithersburg.National Institute of Standards and Technology – US. Department of Commerce.
36