r- .,. 
Strathmore 
UNIVERSITY 
Empirical Evaluation of the Relationship Between Bitcoin and 
Domestic Currencies in Africa 
Kalu Elizabeth M'meneni 
070246 
Submitted in partial fulfillment of the requirements for the Degree of 
Bachelors of Business Science in Financial Economics at Strathmore 
University 
Strathmore Institute of Mathematical Sciences, 
--stra-thmnre-University, 
Nairobi, Kenya 
December, 2017 
DECLARATION 
I declare that this work has not been previously submitted and approved for the award of 
a degree by this or any other University. To the best of my knowledge and belief, the 
Research Project contains no material previously published or written by another person 
except where due reference is made in the Research Project itself. 
© No part of this Research Project may be reproduced without the permission of the author 
and Strathmore University 
... . f;J.l :~.etb ............ k:.P. . !~L. .................. [Name ofCandidate] 
.. .... ...... 11/.!4.~f:::. ................................... [Signature] 
........ 1!./.l-?:../.~o.tf.: ................. .............. [Date] 
This Research Project has been submitted for examination with my approval as the 
Supervisor . 
.... .. . . ....... . ....... ....... .. ..... .... ... .... .. .. ......... ... [Name of Supervisor] 
.... ... . .... ..... ... ..... ....... .................... ..... ...... . [Signature] 
...... . . ... . . . . ... .. ... ... ...................... .. ... ..... .. .. . [Date] 
Strathmore Institute of Mathematical Sciences 
Strathmore University 
ABSTRACT 
Bitcoin trading has gained momentum in the African continent as well as in the world. As 
is so, it is crucial to understand how the development of bitcoins would affect a given 
economy, especially because the operation ofbitcoins is beyond the governing eye of any 
Central Banlc This research seeks to study Bitcoin as a currency and establish what impact 
it would have on the domestic currencies of nations which have significant Bitcoin trading 
activity in Africa. This impact will be evaluated using the Random Effects method of 
estimation. The research will further evaluate the long-term relationship, if any, between 
bitcoins and selected domestic currencies in Africa using the Kao and Perdoni residual 
cointegration tests. The countries of focus in this research include Kenya, Morocco, 
Nigeria and South Africa, which were observed between 2014 and 2017. The choice of 
these countries and the duration of observation is due to the ease of availability of data on 
Bitcoin trading. Results confirmed the existence of a statistically significant relationship 
between the amount of bitcoins in circulation and macroeconomic variables such as 
exchange, interest and inflation rates. No long-term relationship was established and the 
Vector Autoregressive test was performed to capture the linear interdependencies among 
the variables. The conclusion drawn from this study is that as the number of bitcoins 
increases in a given economy, the domestic currency suffers devaluation. Due to this, 
governments are recommended to keep a watchful eye over bitcoin transactions in their 
respective economies as well as looking into the development of competitive central 
backed cryptocurrencies over which they would have total control. 
ii 
TABLE OF CONTENTS 
DECLARATION ..... ... ... .... .... ....... .... .. .. ...... .... ... .. .... .... .. .. .. .. .... ... ............... .. ... .. ....... .... ...... .i 
ABSTRACT .. .. ... .......... ................................................... ... ... .................. ........... .. .. ........ .. . ii 
LIST OF FIGURES .. ..... .. ... .... .... .... ... ... .. ....... .. .. .... ..... .. .... .. ........ ...... .... ... ....... ... .. ....... .... .. vi 
LIST OF TABLES .. .. .. ......... ............ .. ........... ... .. .......... .. .. ..................... .. ... ..................... vii 
LIST OF ABBREVIATIONS ....... ................................. .. ...... ..... ... ..... ... .. ... ............ .. ... .. viii 
1 INTRODUCTION ... .... ...... .. ...... .. ... .. ... .... .. .. ...... .... ....... .. ..... ........ .... .................... ... .. .. 1 
1.1. Background of the Study ................... .... .. ... ................................ ....... .... .. ............ 1 
1.1.1 Evolution of Money ....... .... .............. .... ....... ... .. .... .. .... .................................. 1 
1.1.2 The Advent of Digital Currencies ... ............................................ .... ............. 1 
1.1.3 Bitcoin and How It Works .......... ..... .... .. ................. ............... ...... .......... .. .. .. 3 
1.1.4 Blockchain and Bitcoin Transactions ..... ... .. ... ........ .. .... ...... .... ..... .... ....... ... ... 5 
1.1.5 Digital Money Services and Bitcoin Acceptance ......................................... 7 
1.2. Problem Statement ....... .... .... ...... .. ... ... .. ................... ...... .. ....................... ... ..... .. ... 8 
1.3. Research Objectives ........ ......... ..... .... .. ..... ..... ............. ....... ............... ... ... ...... .. .. .. . 8 
1.4. Research Questions .... .... .............. ............ ........................................... ... .. .. .. ....... 8 
1.5. Scope of the Study .. .... ............ ... ..... ... .................. .... .. .. .. ........ ...... .... ......... .. ......... 9 
1.6. Significance of the Study ... .... .. .. .. .... ......... ......... .. ................. ...... ...... ............ ...... 9 
2 LITERATURE REVIEW ............... ..... ....................................... .......... .... .... .. ....... .. . 1 0 
2.1 Introduction ........ ... .......... ... .... ......... .. .............. ...... .. .. ........ ... ..... ... .. ....... ... .... ..... 1 0 
2.2 Theoretical Literature ...... ... .... .. .... ...... .... .. .. .... .......... ........ ... ......... ....... ... ........ ... 1 0 
2.2.1 Traditional Flow Model ................... .. .... .. .. ..... ......... .. ............ .... ....... ......... 1 0 
2.2.2 Purchasing Power Parity (PPP) Theory ... .. .. ............ .. .......... .. .. ...... .. .... .. .. .. 11 
2.2.3 Asset Approach Theory ... .. .. ................ ... .......... .. ...... ....... ................... ... ..... 11 
2.3 Empirical Literature .... ... .. ... .. ........ .. ... ............ ... .. ...... .. ..... .... ....... .... ...... ............ 13 
iii 
2.3.1 Factors affecting Exchange Rates ... .. .... ........... ............... ..... ................. ... .. 13 
2.3.2 The case ofNon-African regions ... ... ... .. .... .. .... ... .... ............... ... ................. 13 
2.3.3 The Case of the African region .. ....... ... .. .. .. .... .. ...... ... ......... .. .... .. .......... ... ... 17 
2.3.4 Bitcoin: A New Perspective? ... ......... ....... .. ........... .... .... ..... .. .... .. ..... .. ....... .. 18 
2.4 Conceptual Framework .......................................... ........ ... ........ .. ... ... ... ............. 19 
2.5 Research Gap .. ......... .................. ..... ....... .. .. ...... .... .... .. ........ ..... ..... ............ .. ... .... . 20 
3 METHODOLOGY ............................... .. .................................... ... ..... .. ...... ............ .. 21 
3.1 Research Design ..... .. .. .. ...... .... .. ..... ..... .. .. .. ..... .. .. .... ...... ...... .. ... ......... .. ... ... .. ..... ... 21 
3.2 Population and Sample .............. ....... ........... ....... ................ .... .. .... .................. ... 21 
3.3 Data Collection ... ... .. ......... .... .... ...... ..... ..... ... ......... ... .. .. ....... ... .. ........... .... ... .. ...... 21 
3.4 Data Analysis ... ...................... .. ................ ............................................ .. ......... .. 21 
3.4.1 Panel Data Regression Analysis .... .... .. .. ... ..... .. .. .. .. .. ... ....... ........... .. ..... ...... 22 
3.4.2 Modelling Long-term Relationships ... .. .... ....... .. .. ...... .. ... ....... .... .. .... .......... 24 
4 DATA ANALYSIS AND FINDINGS ... .... .... .. ..... ....... .. .. ...... ... ... ...... ...... .... .... ...... .. 27 
4.1 Descriptive Statistics ... ..... .. .... ............ ... .. ... .... ... .. ... .. ........ ... .. ........ .... ........ .. ... .. . 27 
4.2 Stationarity Tests: Unit Root Testing ...... ...................... .................. .................. 28 
4.3 Correlation Test .. .. ... ... ...... .. ....... .. ......................... .... ... .... ... ......... .......... ... .. .... ... 29 
4.4 Regression Analysis ..... ............ ... ..... ... .. ... .. ... ... ... .. ..... ........ .... ........... ....... .. ....... 30 
4.4 .1 Regression Coefficients .... ..... .. ... ....... ... ... .. ....... ..... ... ... .............. ...... .. ... ..... 3 2 
4.4.2 Probability (p-value) ... .... ............................ ......... .................... .. ...... ... ..... .. 32 
4.5 Panel Cointegration ... ...... .. .................... .. ......... ..... .. .......... .. ... ..... .... ..... .. .... ..... .. 33 
4.6 Vector Autoregressive Model (VAR) .... .. .... ......................... ... .... ... .................. 34 
4.6.1 Variance Decomposition ............... .... ....... .......... .... ...... ... ....... .... ..... ....... .. .. 36 
5 DISCUSSIONS, CONCLUSIONS AND RECOMMENDATIONS ....... .... ............ 39 
5.1 Summary of Findings ..................... .... ... .. ........ ........ .... ... .......... ..... ...... .. ..... .. ..... 39 
iv 
5.1.1 Discussion ofthe Regression Findings .. ........ .. .. .... .... ....... ... .. .. .... ............. . 39 
5.1.2 Summary of the VAR and Variance Decompositions Findings .. ..... ....... . .40 
5.2 Conclusion of the Study ................................................................ ................... .42 
5.3 Limitations of the Study ... ... .. .... ... ..... .... .... .. .. .... .. ... .. ............ .. ... .... ............ ... .. .. .42 
5.4 Recommendations .. .. .. .. ......... ....... ..... ... ... ............ .. .. .. ...... ... ...................... ......... 43 
5.4.1 Policy Recommendations ........................................................................... 43 
5.4.2 Recommendations for Further Research ............... ....... .......... .. .... .. ..... ...... .44 
6 REFERENCE ............................................................................................................ 45 
7 APPENDICES ... ... .. .. .... .. .. .. .. .. ... ... ............ ... ..... .... ......... ... .... .. ..... .... .. .... .... .. ........... . 50 
Appendix A: BTC/USD Exchange Rate ... ... .... ..... .............. ..... ............ .. ...................... 50 
Appendix B: Bitcoin Price History Chart ........ ..... ... .... ........................ ............ .... .... ..... 50 
Appendix C: Total Number ofbitcoins in Circulation ... .. .. ............ ...... .. .......... .... ........ 51 
Appendix C: Line Graph Plots of the Sample Variables across the Four Crossections . 
................................................................................................................................... ... 51 
Appendix D: Histogram; Normality Test for Regression Residuals ... ... .... .... ... .. ........ . 54 
Appendix E: V AR and Variance Decomposition Tables .. ... .. .. .. .. .. ...... .. ................. ..... 54 
v 
UST OF FIGURES 
Figure 1 :Cryptocurrencies by Market Share .. ............. ....... ..... ... ..... ....... .. ....... ..... ....... .. .. ... 2 
Figure 2 Expected Total Supply ofBitcoins .... ... .......... ................ .. .......... .... ... .. ................ 5 
Figure 3: Bitcoin's Approach to Transaction Flow and Validation ... .... .. ... ......... ..... ... .... .. 6 
Figure 4: Relationship Between Exchange Rate and factors that affect it .. ... .... .. .... ...... .. l9 
vi 
LIST OF TABLES 
Table 1 Description of the Dependent and Independent Variables for Regression Analysis 
.... ... ... ......... .. .. ...... .. ...... .... ...... .. ..... ..... ... .. ... .. .. .. .. ... .... .. ........ .. ................ .. .. .. .. .. .... ... .... ... .. .. 22 
Table 2: Summary of Descriptive Statistics ... ...... .... ...... ...... ... .. .. ...... ............... .. .. ... .. ..... .. 27 
Table 3: Summary of Panel Stationarity Tests .. ..... ..... ........ ........ .... ...... .... ... .. .. .. .. ... .. .. ..... 28 
Table 4: Correlation among Independent Variables .... ...... .... ... .. ... ... ..... ..... .... ....... ... ... ... . 30 
Table 5: Hausman Test Results ... ...... ... ... ...... ..... .. ................... .... ...... ...... ......... .. .. .... .. ...... 31 
Table 6: Summary of Regression Analysis Results ...................... ........ ......... ... ......... ... ... 31 
Table 7: Summary of Panel Cointegration Tests ... .......... ....... ... ... ... ..... ..... ..... ........ .. .. .... . 33 
Table 9: P-values of the V AR model coefficients ....... ...... .... .... .... ............ .. ..... .......... .. ... 34 
Table 10: Variance Decomposition Analysis for the exchange rate(USD _X_LN) ... .... .. 36 
Table 11: Variance Decomposition Analysis for the amount of bitcoins(BTC _ AMT _ LN) 
........... .. .. ......... ... .... ...... ...... ........ .. .. .. ........ .. .. .. ......... ... ..... .. ....... ............... .... .. ............ .. .. ... . 37 
Table 8: Summarized Results of the VAR Model .... ..... ...... ..... .... .... ..... .. .. .. ... ....... ..... .... . 54 
Table 12: Variance Decomposition Analysis for the price ofbitcoins(BTC_PRICE_LN) 
...... ... .. ...... ...... .. .......... ..... .. .... .... ... ... .. ......... .. .. ... .. .. ...... ... .. ... ............ ............. ......... .... .... .... 55 
Table 13:Variance Decomposition Analysis for the inflation rate(INFLATION_RATE) 
.... .. ... ... .. ........... .... .. .... .... ...... .... .. .. .. .. .... .. .. .. .. ... .. .. .. .. .... .... .......... ........ ................ .... ... ... .. .... 55 
Table 14 Variance Decomposition Analysis for the interest rate(INFTEREST_RATE) 56 
vii 
LIST OF ABBREVIATIONS 
BTC: Denomination of the Bitcoin currency 
PPP: Purchasing Power Parity 
USD: United States Dollar 
OLS: Ordinary Least Squares 
ECT: Error Correction Term 
PMG: Pooled Mean Group 
V AR: Vector Autoregressive 
viii 
INTRODUCTION 
1.1. Background of the Study 
1.1.1 Evolution of Money 
At the dawn of humanity, barter trade was used in lieu of money to exchange goods and 
services. The lack of a common currency facilitated the growth of this trading system, 
with communities exchanging what they needed for what they already had in surplus. In 
the African economy, close communities exchanged pastoral and agricultural 
commodities for local consumption while precious stones and metals, rhinoceros horns 
and ivory were shipped overseas (Centrak Bank of Kenya, 2017). There was an 
advancement of the barter system to the use of cowrie shells. The Central Bank of Kenya 
traces this advancement to 1200BC which was mainly to solve the problem of subdivision 
in the barter system. Cowrie shells had appealing qualities such as being durable, easily 
portable and divisible (Glyn, 2002). By 1000 B.C, cowrie shells were already being 
replaced by the use of precious metals. Such metals included bronze, copper, electrum, 
silver and gold (Gascoigne, 1993). These early metal monies developed into primitive 
versions of round coins, which had the faces of various gods and emperors as marks of 
their authenticity (Apsell, 1996). Coins further evolved into bank notes around 1661 AD. 
From coins and notes was birthed credit and debit cards which were introduced in 1946. 
Since then, fiat currency has mostly taken the form of coins and bank notes. Coinage, in 
our digital age, has since developed from tangible coins through to digital coins 
(Callander, 2014). 
J .1.2 The Advent of Digita l Currencies 
Digital currencies, also known as cryptocurrencies, are the most recent invention of 
money. A digital currency is a virtual coinage system. It functions more like the 
conventional currency, enabling users to provide virtual payment for goods and services 
free of a fmancial intermediary as well as a central trusted authority. Digital currencies 
also perform the functions of money of being a medium of exchange, a store of value and 
a unit of account. The cryptocurrency market, even though having a short lifespan, has 
evolved at an unprecedented speed (Farell, 20 15). Since the release of the pioneer 
1 
cryptocun-ency, Bitcoin, in 2009, more than 700 other cryptocun-encies have been 
developed, majority of which have had only a pinch of success (CoinMKTCap, 20 17). In 
studying digital cun-encies, this research will focus solely on Bitcoin, the first successful 
decentralized cryptocun-ency. The focus on Bitcoin is also due to its popularity and it 
having the largest market share of 56.97% in relation to the entire cryptocun-ency market. 
Figure 1 below shows these statistics. 
Figure J:Oyptocurrencies {~Ji A1arket Share 
Et hereum , 15.46°:, 
~ Bitcci n ~ Ethereum 
Cryptocurre ncies by Ma rket Ca p 
coin.dan ce 
Bitcoin , 55. 38° ', 
Oth.: r. 9 . 86~, 
,'\. 
'" ' ..... I'EO, o.s7•;, 
, \\_ '-. r.tonero , 0.9:% 
\ Ltte com , 1. 59 . !;I 
'- D•sh. 1. 7~ , 
\_ Ripple , 3.6a• ·, 
Bitcoin Cash, 10. 5' '' 
Bi tcoin Ca;h 4.: Rippl e 4.: Dash Lit~ coin 4: Monero r· EO • Ot~ e r 
Source: (Coin Dance, 2017) 
2 
l . l.3 Bitco in and How ll Works 
Bitcoin is a decentralized global currency system which was initially designed and 
developed by Satoshi Nakamoto 1 (Barber, Boyen, Shi, and Uzun, 2012). The units of 
currency of Bitcoin are called bitcoins, or BTC, and are used to store and transmit value 
among participants in the Bitcoin network. Essentially the main difference between 
Bitcoin and bitcoins is that Bitcoin, capitalized, refers to the technology and network, as 
well as the currency, whereas bitcoin(s) refer to the unit of account. Unlike conventional 
currencies bitcoins have no physical manifestation as they exist only in the form of 
computer codes. These codes are designed such that they are publicly accessible and can 
be inspected, modified and enhanced. Bitcoin uses peer to peer technology with no 
regulator or financial intermediaryl. Management of transactions and the issuing of 
bitcoins is carried out collectively by the network. Bitcoin, like any other currency, serves 
the purpose of facilitating the exchange of goods and services. However, unlike other 
traditional currencies, it is neither issued nor controlled by a state or even a single authority 
(Decker and Wattenhofer, 2013). 
Since its invention in 2009, Bitcoin has enjoyed a rapid growth, both in value and in 
volumes3. In mid-2010, one bitcoin exchanged for approximately 0.08 USD. This 
exchange rate has since experienced an upward shift with one bitcoin exchanging for as 
much as 8257.47 USD in November 20174. The number ofbitcoins has also been growing 
and there are approximately 16.6 million bitcoins in circulation as of September 2017. 
This success can be attributed to bitcoins being highly liquid, having low transaction costs, 
ease of sending payments quickly across the internet, and being used for micropayments. 
Even though the Bitcoin economy is gradually flourishing, there are concerns over the 
1 The name 'Satoshi Nakamoto' is assumed to be fake by some, and the person bearing that name has 
neither been seen nor heard from since April 2011 
2 In this peer to pee·r technology, transactions do not require authorization by a third party like PayPal, 
Western Union or Visa 
3 See Appendices Band C in the Append ix section . They show graphical representations of the growth of 
Bitcoin in value and volumes. 
4 Appendix A in the Appendices section shows the bitcoin to US dollar exchange rate as at November 
2017 
3 
legal status ofBitcoin and its ability to facilitate money laundering, tax evasion and trade 
in illegal drugs and activities (Grinberg, 2011). Moreover, there are security risks with 
companies that trade bitcoins recording huge losses when they were hacked (Agoya, 20 15) 
and their bitcoins stolen (Pagliery, 2014). 
The process of producing more bitcoins is called mining. Mining is equivalent to the 
printing of fiat money by central and federal banks. This process is canied out by 'miners'. 
The Bitcoin system is based on mathematics and the entire process of mining revolves 
around using mathematical formula to solve some complex mathematical problems. 
Miners are required to use their computing prowess to solve these problems. Once these 
problems are solved, and the solutions approved by all the other miners then new bitcoins 
are produced. Any miner who successfully solves any of the math problems is rewarded 
with a certain number ofbitcoins. This usually serves as an incentive to allow more miners 
to engage and compete in the mining process. An interesting fact about the bitcoin mining 
process is that as the number of miners in the bitcoin network changes, the mathematical 
problem difficulty adjusts to ensure that bitcoins are created at a predetermined rate and 
not faster or slower. Grinberg (20 11) explains that as of 2011, about 50 bitcoins were 
being issued every ten minutes. The rate was halved to 25 bitcoins in 2013 and further 
halved every four years thereafter. At these rates, 10.5 million bitcoins were created in the 
first four years, half that amount in the following four years, half that amount in the years 
thereafter, and so on. This process and time intervals will allow for the number ofbitcoins 
to approach their upper limit of21 million by 2040 as illustrated infigure 2 below. 
4 
Figure 2 Expected Total Supp(y ofBitcoins 
Expected Total Bitcolns 
~ 
~ 
,.....-- ~ 
¥"'_ 
/ 
/ 
I 
/ 
"""" 
2009 201.3 201.7 2022 2025 2029 
Year 
Source: Hastings Science & Technology Law Journal [vol. 4.1], Pg. 164 
Mining is an important part of the Bitcoin system as it ensures integrity, stability, safety 
and security of the bitcoin network. It is also worth noting that bitcoins are divisible to 
eight decimal places, with the smallest unit being: 1 Satoshi=0.00000001 BTC 
(Szczepanski, 2014). 
1.1.4 Blockchain and Bitco in Transactions 
People send bitcoins to each other over the Bitcoin network all the time. Due to lack of a 
central governing authority to keep record of the transactions, the bitcoin network self-
sufficiently collects all the transactions made during a set period into a list called a block. 
The miners have the responsibility to confirm these transactions, and compile different 
blocks into a general ledger known as the 'blockchain'. The blockchain therefore is a long 
list of all the transactions that occurred on the Bitcoin network. This blockchain is also 
available to everyone who participates in the network so that they all know what is going 
on. For every transaction, it can publicly be seen from whom bitcoins were received and 
to whom they were sent to, but the personal information about the parties involved in the 
transactions is never disclosed (Szczepanski, 2014). 
Participants in the Bitcoin network do not just "hold" bitcoins, rather they participate in a 
transaction that can be verified publicly showing from whom they received the bitcoins 
5 
and to whom they are sent to. For instance, according to an example by (Bohme, Nicolas, 
Benjamin, and Moore, 2015) two users, Charlie and Bob carry out Bitcoin transactions. 
These transactions like any other are recorded in the blockchain. Charlie is able to verify 
from the public records on the blockchain that Bob could make payment with regards to 
the transaction because there was a prior transaction in which Bob received three bitcoins 
from Alice, and there was no prior transaction in which Bob spent these three bitcoins. In 
general, bitcoin transactions are ordered recursively by having the input of one transaction 
refer to the output of a previous transaction. Figure 3 further explains this example. 
Figure 3: Bitco in 's Approach to Transaction Flow and Va lidation 
B itcoin. 's Approach to Transaction Flow a n d Valid a tion 
0 
1\ 
.Al ice 
3 b i1c o ins 
A~B: 3 
[ .. . ] 
signeo:d A lic e 
0 
(\ 
Bob 
~ b i tco ins _______ .,. 
s igned B o b 
o l 
{\ 
C h ar 
Peer-to-pee r n e twork 
_ _,.. .. .-
------------------------------ ---------
..... _ .... _ .... ___ .. __ , __ _ 
Transacti o n log (b lock cha i n , c o pie d w all n o des) 
0 
1\[i 11 
~·l i rH: t-s 3ppeu d b loc k s o f tr.u~sactic:u-._, b y so lvin g rn:tthcrn ;l tic a l puz7J~:s o f inc n :a_,ing diffic uh; ·. 
Source: Journal of Economic Perspectives- Vol 29(20 15)-page 216 
6 
"Bitcoin cannot be deposited in a bank, and instead it must be possessed through a system 
of digital wallets" (Y ermack, 2013 ). Every user of bitcoins is assigned a wallet which 
would contain the bitcoins owned, a public and a private key. The public key is seen as 
the public address of the owner. It is through this address that another party is able to send 
bitcoins to the owner of the wallet. The private key acts like a personal identification 
number to a bank account. It is what enables the wallet's owner to send his own bitcoins 
to someone else (Prentis, 20 15). As an analogy, the public key is your postal address, and 
the private key is the key to your front door; others can send mail to your house with just 
your physical address, but no one can remove your belongings without your permission 
(Turpin, 2014). 
1.1.5 Digita l Money Services and Bitcoin Acceptance 
Since its invention, Bitcoin has caused a stir in many economies globally, the main issue 
being whether or not to accept it as legal tender. Another issue is brought about by the 
nature ofbitcoins. The Central Bank or Federal Reserve of any economy cannot control 
bitcoins in the way they do with transactions and taxation of other currencies. As a result, 
a certain part of the monetary system operates outside of the regulators' authority and that 
threatens the stability of the fmancial system of the country (Alina, 2016). With this in 
mind, there are some countries that have indirectly assented to the usage of bitcoins, 
however it is not legally acceptable as a substitute for the country's legal tender. Such 
countries include: The United States, Canada, Australia, Finland, Belgium, the United 
Kingdom, Bulgaria and Germany among others. On the other hand, there are those nations 
that have out-rightly banned this digital currency. Such include: Iceland, Vietnam, 
Ecuador, Bolivia, Kyrgyzstan, China, and Russia (Bajpai, 2015). 
Digital currencies have also gained momentum in Africa with nations such as Ghana, 
Kenya, Morocco, Nigeria, South Africa and Tanzania, among others being indulged 
(CoinPursuit, 2014). Bitcoin in Africa has been made easier by companies such as BitPesa 
which exchanges bitcoins for the Kenyan/Ugandan/Tanzanian Shillings, as well as for 
Nigerian Naira (BitPesa, 2014). Kitiwa is a similar service that operates in Ghana. Other 
digital money services include Bitrefill which allows users to top up any prepaid mobile 
7 
phone using bitcoin (Bitrefill, 20 15). There is also a Bitcoin ATM in South Africa, which 
converts cash into bitcoins. 
l .:?. . Problem Statement 
Due to digital currencies being a fairly new concept, much of the research done in this 
field focuses on explaining what these digital currencies are, how they work, their pros 
and cons and their legality issues. The few empirical studies include (Alina, 2016), who 
carries out a regression analysis of cryptocurrency influence on the US Dollar. A similar 
research was done by (Loseva, 20 16), who also analyzed the influence of cryptocurrency 
on the Russian economy. These studies have been very informative to policy makers in 
their respective countries. Closer home, a research (not yet published) titled "Adoption of 
Bitcoin in Kenya, A Case Study of BitPesa" was done by (Njuguna, 2014). This study 
focused on Bitcoin as a system of transferring funds in relation to other money transfer 
services. 
There has been no research seeking to study the relationship that exists between the 
circulation of bitcoins and the effect on currencies of different countries in Africa. It is 
important to evaluate this relationship because bitcoins are currencies which are viewed 
to be more attractive than traditional fiat currencies. What is uncertain is whether or not 
the Bitcoin system will eventually replace fiat currency. This research seeks to evaluate 
empirically the relationship that exists between bitcoins and the currencies of some 
selected African economies. 
1.3. Resea rch Obj ectives 
This research seeks to achieve the following objective: 
1. To evaluate the relationship between the circulation of bitcoins and selected 
domestic currencies in Africa (Kenya, Morocco, Nigeria and South Africa) 
1.4. Research Questions 
The following research question will be answered in line with meeting the objective of 
this study: 
1. What is the relationship between the circulation of bitcoins and domestic 
currencies in Africa (Kenya, Morocco, Nigeria and South Africa)? 
8 
l .5. Scop~ or the Study 
The scope of this study will be on Kenya, Morocco, Nigeria and South Africa. This is 
because these are the leading countries in Africa in terms of adoption and trading of 
bitcoins (CoinDance, 2017). The choice of these four economies was also due to the ease 
of availability of data on volumes of bitcoin traded since 2013, when they gained traction 
in Africa, to date. 
1.6. Significance of the Study 
The use of digital currencies and most commonly bitcoins is a new concept not only in 
Africa but also globally. This suggests that it is crucial to understand how the development 
of bitcoins would affect any given economy. This research seeks to provide information 
to policy makers and the government on the possible effects to the financial sector brought 
about by usage ofbitcoins. The government will therefore be able to effectively respond 
to their spread and establish ways of using them for the purposes of improving the entire 
economy. 
Secondly, this research seeks to lay a foundation for more extensive research in this field 
which may also be used as a reference by other researchers to build on it. 
9 
2 LITERATURE RE\'IE\V 
2.1 Introduct ion 
In the most simple terms, an exchange rate shows how much units of a foreign currency 
can be purchased with one unit of domestic currency. Exchange rates play a significant 
role in determining the volumes of international trade for any country. The main objective 
therefore for every country is to maintain stable exchange rates and to protect itself, if 
possible, from the risk of exchange rate fluctuations. These exchange rates, if highly 
fluctuating, can influence the decisions of policy makers,and affect the allocation of 
goods, reserve money, exports, imports and balance of payments (Abar, 2015). 
This section seeks to establish the conventional factors that influence exchange rates and 
include a possible consideration of any other factor(s) that could affect exchage rates but 
has(have) not been taken into account by previous literature in this field. 
In reviewing other literature, inclusivity was given to research papers that narrowed their 
scope of study to different countries(regions). This was to show that there is an overall 
acceptance and agreement by researchers on the possible list of factors that affect 
exchange rates. This approach is used to also show how different factors have a more 
significant impact than others in different regions. 
2. 2 Theoretical Li tcm lure 
In this section, theories of exchange rate determination have been reviewed. The focus is 
more on the general theories surrounding exchange rates and much less about the 
proponents of these theories. 
2.2.1 T rad itional Flow Mode l 
Early literature in the field of exchange rates emphasized that the primary determinants of 
exchange rates was international trade flows (Pearce, 1983). This was due, in part, to the 
fact that most governments maintained tight controls on international fmancial capital 
flows. Exchange rates had a role in eliminating international trade imbalances. This role 
was played out where countries with trade surpluses were expected to have appreciating 
currencies whereas those with deficits should have depreciating currencies. Such 
exchange rate changes would then lead to changes in international relative prices that 
would work to eliminate the trade imbalance. In recent years, however, the traditional 
10 
flow model seems not to hold much water. For instance, financial liberalization has led to 
the volumes of international trade in financial assets exceeding trade in goods and 
services. In other instances, there have been countries whose currencies are appreciating 
even though they have trade deficits. 
2.2.2 Purchas ing Power Parity (PPP) Theory 
The purchasing power parity hypothesis states that there exists a proportional relationship 
between the exchange rates and the foreign to domestic price ratio (Stockman, 1980). The 
PPP theory holds if the law of one price is true. There are two forms of the PPP theory. 
The first is the Absolute form of PPP which states that, "the equilibrium exchange rate 
between currenCies of two countries is equal to the ratio of the price levels in the two 
nations". In other words, the prices of the same products in two different countries should 
be equal when measured using a common currency. If, there exists any difference then the 
demand should shift from one country to another in such a way that the prices will have 
to converge. Realistically, this theory may not actually hold due to market imperfections 
brought about by factors such as different levels of technology, costs of production and 
taxation. 
The second form is the Relative form of PPP theory, which states that changes in the 
exchange rate over a certain period of time should be proportional to the relative changes 
in the price levels in the two nations over the same time period. This form of the PPP 
theory accounts for market imperfections as it allows the prices of similar commodities in 
two different countries to not necessarily be the same, even when measured in a common 
currency. When both countries experience some inflation, then the exchange rate will 
automatically adjust itself in such a manner that the difference in the rate of inflation will 
be offset. 
The PPP theories therefore bring out the relationship between inflation and the exchange 
rates of given countries. 
2.2.3 Asset Approach Theory 
The asset approach theory is a form of modem exchange rate theory which put emphasis 
on financial-asset markets. This theory serves as an improvement to the hypothesis held 
by the traditional flow model of exchange rates. Exchange rates under this model, are 
11 
viewed as adjusting to equilibrate international trade in financial assets. Exchange rates 
will therefore change frequently in response to changes in the demand for and supply of 
financial assets of different nations. This is so because prices in financial markets adjust 
way faster than in goods market. This theory assumes perfect capital mobility, which 
simply means that there are no barriers to international capital flows. 
The asset approach theory is further broken down to the monetary approach theory 
(Frenkel, 1976) and the portfolio-balance approach theory (Kouri, 1978). The monetary 
approach theory holds that the relative demand and supply of money between two 
countries is the key determinant of the exchange rate between these countries. An increase 
in domestic money supply would drive prices up and hence reducing the exchange rate. 
An increase in domestic real income causes excess money demand, with a fixed nominal 
money supply, results in a reduction in domestic prices and hence pulls the exchange rate 
up. An increase in the domestic interest rate, which is assumed to reflect higher expected 
inflation, lowers the demand for money, increases prices and lowers the exchange rate 
(Pearce, 1983). Changes in foreign variables would also have the same effects. In 
summary therefore, an increase in foreign money supply leads to a corresponding increase 
the domestic exchange rate. Reducing real foreign income and increasing foreign interest 
rate would also have the same effects on the domestic exchange rate. 
The portfolio-based approach on the other hand holds that the demand and supply of 
relative bonds as well as relative money-market conditions are determinants of the 
exchange rate. Under this model, households in every country are assumed to distribute 
their net fmancial wealth among three assets: domestic government bonds, domestic 
monetary base and net foreign bonds denominated in foreign currency. The domestic and 
foreign bonds are assumed to be imperfect substitutes; thus, the desired proportions of 
these assets depend on their respective yields. An increase in the domestic (foreign) 
interest rate causes investors to increase the desired proportion of their wealth in domestic 
(foreign) bonds and to lower the desired proportions in the monetary base and foreign 
(domestic) bonds. As the supply of domestic bonds rises relative to foreign bonds, there 
will be an increased risk premium on the domestic bonds that will cause the domestic 
currency to depreciate. An increase in net holdings of foreign bonds resulting from a 
12 
current account surplus would increase domestic wealth and disturb portfolio equilibrium. 
In this case, domestic wealth holders would want to hold some of the wealth increment in 
the form of domestic assets. This would lead to a fall in the domestic interest rate and an 
appreciation of the exchange rate (Pearce, 1983). 
2 .3 Empirical Literature 
2.3. 1 Factors affecting Exchange Rates 
In doing an extensive analysis of research done in different regions or countries with 
respect to the factors affecting exchange rate, the regions will be divided into two. The 
first section will consist an analysis of the various factors that affect exchange rates in 
other regions that are not part of the African continent. This includes Asia, Europe and 
America. The section that will follow will consist of the same analysis done in the context 
of the African region. 
2.3.2 The case ofNon-African regions 
Research by (Patel, JPatel, and R.Patel, 2014) in the paper titled " Factors Affecting 
Currency Exchange Rate, Economical Formulas and Prediction Models" clearly listed out 
the various factors affecting currency movements and their characteristics. At the top of 
this list was inflation. This paper explains that a lower rate of inflation tends to result in 
currency appreciation. Taking an example, suppose that the price levels in country X 
reduce by 30%, while the price levels of its foreign trading partners remain relatively 
stable. The goods in country X will seem cheaper thus increasing the volumes of exports. 
This will appreciate the currency of country X. 
Another key factor that is seen to affect exchange rates is interest rates. If the interst rates 
in country X are rising relative to other countries' rates, more investors will want ot invest 
in country X due to the higher returns they will receive on their investments. This causes 
an increase in demand of Country X's currency and therefore an appreciation in its value. 
Capital account balance is another factor. Countries having a surplus can attract more 
capital from other countries and can see an appreciation in the currecny value relative to 
other countries with a deficit. Other factors that affect the exchange rates include public 
debt, role of speculators, gross domestic product(GDP), unemployment rate, relative 
strength of other countries, macroeconomic and geopolitical events. 
13 
In further studying these factors, (Patel et al., 2014) applied some economical formulas in 
testing the parity conditions among currencies as well as to check currency valuation and 
its effects. Some of the models used were the Purchasing Power Parity (PPP) and the 
Interest Rate Parity (IRP) models, which are based on the law of one price. There was also 
the Balance of Payments Theory model which is based on balance of payments. 
Even though the scope of study under this paper was India, (Patel et al. , 2014) concludes 
by advising individual countries to monitor closely the above litsed factors to ensure they 
have a favourable exchange rate, as these factors are not only specific to India but also to 
every other country as well. In this research, (Patel et al., 2014) clearly brings out the 
factors affecting exchange rates but fails to relate these factors to the Indian Economy 
even though it is mentioned that the study was done in India. The empirical analysis in 
this research, fails to actually run the models listed in the paper. Real data is not used to 
prove whether the selected models are actually sufficient in checking for currency 
valuation and its effects, and it is therefore not possible to see the extent to which the 
factors listed affect the Indian rupee. 
Further research by (Abdoh, Yusuf, Zulkifli, Bulot, and Ibrahim, 2016)also studies the 
macroeconomic factors that influence exchange rate fluctuations . This study was carried 
out in the ASEAN countries also agrees that exchange rates are very important. This is 
because they facilitate international trade and commerce, the transfer of funds between 
countries as well as allowing for the comparison of prices of the same goods across 
countries. This research focused on inflation and interest rates as the factors having effect 
on exchange rate fluctuations. Another factor that was added in this research was the 
volume of exports. The research by (Abdoh et al. , 20 16) used actual data to test the effects 
of these three variables on the exchange rates. The data was collected for 10 years which 
started from 2005 until 2014 for Darussalam, Malaysia, Philippines, singapore, Thailand, 
Indonesia, Cambodia, Laos and Vietnam. 
To analyze the factors that affect the exchange rate movement, (Abdoh et al., 2016) used 
a multiple regression analysis. The dependent variable in this case was exchange rate and 
the independent variables: volumes of export, interest rates and inflation rates. The results 
of this research were that volumes of exports show a modest relation to the exchange rate 
14 
at a statistical significance of 5%. There is therefore a positive significant relationship 
between exports and exchange rates. For inflation and interest rates there was found to be 
a statistically insignificant dependence, and as such there is indeed no significant 
relationship between inflation and interest rates, and the exchange rates. These two 
variables were also found to have an inverse relationship with exchange rates. 
To conclude, (Abdoh et al., 2016) established that only one variable, which is volume of 
exports has shown significant relationship with the exchange rate. This means that for the 
ASEAN countries, the volume of exports is an important variable in influencing the 
exchange rate movements. 
Even though the research by (Abdoh et al., 20 16) is consistent with (Patel et al., 20 14) on 
the factors that affect exchange rates, only 3 factors are brought out to affect exchange 
rate. This research is limiting as there are numerous factors that would have effect on the 
exchange rates. This is further backed by the R2 value from the regression of only 4.24%, 
indicating that the model only explains upto 4% of the variability in exchange rates leaving 
out 95.26% which is explained by other macroeconomic variables that were left out in this 
model. 
Abdullah (20 1 0) further contributed to this field of study by researching on the existing 
relationship between inflation and real exchange rates. Unlike other researchers, Abdullah 
(20 1 0) focused solely on inflation rather than multiple factors. This study was centred 
around a comparative analysis in Asia, EU and North America. The research was carried 
out using secondary annual data in the period of 1991-2005. 
Exploratory data analysis was used to study the behavior of data with respect to graphical 
comovement between inflation and the exchange rates. The Granger-causality test was 
then employed to explore the direction of the causal relationship between the two 
variables. In the final step, a panel data model was used to show whether there are 
differences in the relationship between the two variables in Asia, the EU and North 
America. 
15 
From the exploratory data, it was clearly seen that there existed a correlation between the 
inflation and real exchange rate movements, even though the the two variables are seen to 
be more volatile after the Asian financial crisis of 1997. 
From the Granger Causality Test, it was seen that there existed a hi-causal relationship 
between inflation and changes in the nominal exchange rates. This means that the 
depreciation of the nominal exchange rate would have effect on inflation and the inflation 
will cause the nominal exchange rate to depreciate. Similar results were also recorded for 
the relationships between real exchange rates and inflation. When the regions were 
separated however, into Asia and non-Asia, the results obtained were different. In the Asia 
region, the relationship between exchange rates and inflation was unidirecctional. The 
depreciation in the exchange rate had a significant impact on inflation but the converse 
was not true. In the non-asia regions however, inflation was recorded to have a significant 
omapct on the exchange rates, and the converse did not hold. These results indicate that 
Asian countries have a higher vulnerability to exchange rates shocks in comparison to 
European and North American countries. 
Panel data analysis performed in this study, showed that inflation is significantly 
influenced by the lag of real exchange rates and domestic inflation, nominal exchange 
rates and foreign inflation. Coefficients of the model suggest that foreign and domestic 
inflation have a stronger impact in comparison to real and nominal exchange rates. A one 
percentage increase in foreign inflation, for example, would lead to a corresponding 
0.46% increase in domestic inflation. A one percentage depreciation in the exchange rates, 
on the other hand, would lead to a corresponding decrease of about 0.05% in inflation 
rates. 
Furthermore, the results from the study also showed that the behaviors of inflation in the 
Asia Region differed with those of the European Union and North America. 
Abdullah (20 1 0) concludes that there exists a strong relationship between inflation and 
real exchange rates in Asian countries, but no such relationship is seen in the non-Asia 
region (EU and North America) . Another conclusion is that the Asian fmancial crisis was 
seen to have more local impact in Asian countries, than globally in the EU and North 
America. 
16 
This research by Abdullah (20 1 0) brings out very important aspects. It clearly explains 
that the extent to which different factors affect the exchange rate may vary depending on 
the economy or region of study. This research also factors in a larger scope of study 
therefore making it easier to do a comparison on how different countries are afected by 
the same factors. 
2.3.3 The Case of the African region 
In the context of African continent, studies concerning the relationship between inflation, 
among other factors, and exchange rates have also been done to end up with consistent 
results as previous works in this field of study. 
Ndung'u (1999), in his research, 'A Dynamic Model of Inflation for Kenya'showed that 
the level of domestic inflation and exchange rate changes affect each other. This was 
shown using the Granger Causality Test based on Kenyan data during the period of 1970-
1993. The conclusion from this research was that the levels of inflation and changes in 
exchange rate affect each other and changes in exchange rates and reserve changes affect 
each other. Ndung'u (1999)'s research does not exclusively focus on exchange rates but it 
does contribute to this study, and gives conclusions that other researchers mentioned in 
this analysis have proven to be true. 
Further, research by (Mavee, Perrelli, and Schimmelpfennig, 20 16) on the drivers of 
volatility in the South African rand/USD exchange rate(rand volatility) shows other 
possible factors that affect exchange rates. To analyse rand volatility, daily data from 
August 2009 to August 2015 is used. Implied measures of volatility were used to measure 
rand volatility, such that implied rand volatility is a funtion of macroeconomic surprises, 
commodity price volatility, volatility index and domestic political uncertainty. This 
reseach fmds out that all the variables in the model had a significant impact on rand 
volatility and thus adding new factors to the list of factors that affect exchange rates. 
A similar research was done by (Proti, 2013) in Tanzania to explore the exchange rate 
fluctuations in the country. Analysis is done using panel data from 1999 to 2009, and the 
statistical regression analysis model, Ordinary Least Squares model was used. Variables 
regressed against the exchange rate were inflation, real interest rates, national debt, real 
GDP growth, political stability, exports and imports. The variables used were found to 
17 
have a significant impact on the exchangre rate. Proti (2013) also mentions that in the case 
of Tanzania, there is poor control by the proper authorities on the value of the currency 
leaving it depreciating consecutively without appreciating against other foreign 
currencies. 
Lyons (2001) has a different view on the factors that actually affect exchange rates. 
According to this author, macroeconomic variables as the ones previously discussed do 
not adequately account for exchange rate behavior over short time horizons. Lyons (200 1) 
further states that short-run exchange rate movements are attributed to market 
microstructure factors, including inventory management and information aggregation by 
foreign exchange delears. 
2.3.4 Bitco in: A New Perspective? 
As cryptocurrencies have gained momentum and popularity, there have been some 
considerations that these digital currencies might add to the list of factors that affect 
exchange rates. As this is a fairly new concept, most of the research done under this subject 
is mainly aimed at explaining the concept of cryptocurrencies and how they work. 
However, research done by (Alina, 20 16) is one of its kind in taking an empirical approach 
in analyzing cryptocurrencies. In the research, (Alina, 2016)'s objective is to estimate the 
impact of the circulation ofbitcons (the most commonly used cryptocurrency)on the US 
dollar. The goal of this research was achieved by running a regression analysis of the 
exchange rate(EUR _ USD) against the amount of bitcoins, the market price of bitcoins, 
inflation rates, interest rates, unemployment rates and public debt. This regression was run 
using variables for the period from 2009 to 2016. The R2 value obtained from running this 
regression was 0.783, which means that 78.3% of the values of the dependent variable are 
explained by the model used. 
Alina (20 16) established that the more bitcoins there are in circulation, the lesser the 
content of the dollar in one euro. More specifically, as the number ofbitcoins in circulation 
increases by 1 million, the EUR/USD rate rises by 6.1 %, that is, the amount of dollars to 
buy 1 Euro increases. This suggests that the distribution ofbitcoins weakens the US dollar. 
The other variables in the model were also found to have a highly significant impact on 
the exchange rate with a probability of 99%. 
18 
This research by (Alina, 20 16), brings fresh thoughts to this field of study on the factors 
that affect exchange rates. It extends the already conventional knowledge to incorporate 
the new concept of cryptocurrecnies. Similar research has not been done to back up her 
conclusion, but even so she has set the stage for further research to be done with respect 
to whether bitcoins would actually add to the list of factors that affect exchange rates in 
any given country. 
2.4 Conceptua l Fra tne\\·ork 
Figure 4: Relationship Belll'een E.cchange Rate and.fhctors that affect it 
The variable at the center, Exchange rate, serves as the dependent variable for this 
research. The independent variables include volume of bitcoins(Btc_Amt), the price of 
Bitcoin(Btc_price), Inflation and Interest rates. The independent variables include a set of 
variables that may cause volatility in the exchange rates of a given country. Interest and 
Inflation rates are measured as percentages where as Btc _ Amt and Btc _price is measured 
using the common currency, USD. 
19 
I -
_,) Rcsemch Gap 
Most scholars, in studying the factors affecting exchange rates consider only the 
conventional macroeconomic variables as evidenced in section 2.3. Minimal research has 
been done to test whether Bitcoin along with the other macroeconomic variables has an 
effect on exchange rates. Such research has only been done in the United States yet Bitcoin 
usage has gained traction in many other countries including countries in Africa. 
This study aims to fill this gap by studying what impact Bitcoin will have on some African 
currencies. The effect of Bitcoin on exchange rates will be tested to see whether or not it 
can suffice to be added to the conventional list of factors. This study will further test for 
the presence of a long-run relationship between Bitcoin and the exchange rate. 
20 
3 METHODOLOGY 
This chapter highlights the research design, population and sample of the study, variables 
to be used in order to meet the objective of this study, and the analysis method(s) used. 
3.1 Rcsemch Design 
The research design followed by this study was an exploratory design. An exploratory 
approach was taken because the subject on Bitcoin and cryptocurrencies in general is 
fairly new and there are few studies that have been done to address this subject. This 
exploratory approach focused on bringing more insights and familiarity on the concept of 
Bitcoin. Quantitative analysis was also used to formulate facts and uncover the patterns 
around data associated with Bitcoin. 
3.2 Pop ulation and Sample 
The population considered for this study is Africa, more specifically, countries that trade 
in Bitcoin in Africa. These include South Africa, Zimbabwe, Tanzania, Kenya, Nigeria, 
Ghana and Morocco. Of the seven countries, the study focused on South Africa, Kenya, 
Nigeria and Morocco, which formed the sample study. The choice of these four countries 
was because data on the variables of study was easily and readily available. The sample 
size considered included panel data of monthly observations of five different variables in 
the four countries under study for the period of January 2014 to June 2017. 
3.3 Data Co ll ection 
The study used secondary data. Macroeconomic data on interest and inflation rates, 
volumes of bitcoins traded as well as exchange rates of the US dollar to the Kenyan 
shilling, Moroccan dirham, Nigerian naira and the South African rand was used. The data 
for the macroeconomic variables was collected from the individual country's Central 
Bank websites, while data concerning bitcoins was collected from CoinDance.com, which 
is one of the main websites that provide data on Bitcoin. Only monthly variables were 
used in this study. 
3.4 Data Ana lysis 
A quantitative analysis approach was used to make computations on the data. Various 
statistical models were considered to fully analyze data and bring out meaningful 
characteristics about this data. The models used are further discussed below in more detail. 
21 
3.4.1 Panel Data Regression Ana lysis 
Data collected for this research had both time series and cross-sectional elements, and as 
such considered as panel data. The objective of this study was to estimate the impact of 
the circulation ofbitcoins on selected domestic currencies. Panel data regression analysis 
was used to achieve this objective. The hypothesis being tested was as follows : 
HO: Bitcoin circulation in the economy has no significant effect on the domestic 
currency 
Hi: Bitcoin circulation in the economy has a significant effect on the domestic currency 
The dependent and independent variables are identified in the table below. 
Table I DescnjJtion oft/ie Dependent and independent Varioblesfor Regression Analysis 
Dependent Variable Description Data Source 
USD X The USD _X exchange rate Central Bank of Kenya, 
-
(the content of X in one Federal Reserve Bank of 
USD), where X represents South Africa, 
the domestic currency Central Bank ofNigeria, 
(I<ES, ZAR, NGN, MAD) Bank Al-Maghrib, 
To evaluate the phenomenon of 'weakening of domestic currency', the individual 
countries' exchange rates were measured against the US dollar, to monitor their 
fluctuations. 
Independent Variables Description Data Sources 
btc amount The total amount of Coin Dance Website 
-
bitcoins in circulation per 
given country 
btc_price Average market pnce of Coin Dance Website 
bitcoins 
22 
Inflation rate Prevailing inflation rates Central Bank of Kenya, 
Federal Reserve Bank 
South Africa, 
Central Bank of Nigeria, 
Bank Al-Maghrib, 
Interest Rate Central bank rates Central Bank ofKenya, 
Federal Reserve Bank 
South Africa, 
Central Bank ofNigeria, 
Bank Al-Maghrib, 
Given the above variables, the regression equation will therefore be: 
log usn_xit= ai + P1 log btc_amtit + P2 log btc_priceit + P3 inflationit + 
P4 interest rate it+ Eit 
Where, 
log USD_Xit is the dependent variable, where i=count1y and t=time 
ai (i = 4) is the unknown intercept for each country 
Pi ( i = 4) is the coefficient for that independent variable 
Eit is the error tenn 
of 
of 
In assessing the impact, the coefficients of the variables were studied after the regression 
was run. Logarithm transformations were made on some of the variables to help in 
rescaling the data so that the variances in the series are stabilized. Log transformations are 
also directly interpretable as approximate proportional differences. In running the 
regression, there were two models that were used; the fixed effects model which assumes 
correlation between the explanatory variables and the error term, and the random effects 
model which assumes the absence of a correlation between the error term and the 
23 
I · 
explanatory variables. The decision on which of the models to use was made using the 
Hausman Test, which allows the decision to be made using the following hypothesis: 
Ho: random effects are uncorrelated with the explanatory variable 
Ha: random effects are correlated with the explanatory variable. 
The null hypothesis basically holds that the random effects model is most suitable whereas 
the alternative hypothesis holds that the fixed effects model is more appropriate. 
3.4.2 Modelling Long-term Relationships 
In modelling long term relationships, a series of tests was run so as to capture the different 
characteristics of this data. The first step in analyzing this data was to test for stationarity. 
For any tests to be carried on this data, the data first ought to be stationary. A series is said 
to be stationary if it exhibits time-invariant mean and variance properties for each lag. An 
examination of whether a series can be viewed as stationary or not is essential because the 
stationarity or otherwise of a series can strongly influence its behavior and properties. Any 
analysis done on non-stationary series gives spurious or non-sense results. When the first 
difference of a nonstationary process is stationary, the process is said to be integrated of 
order one, /(1). When a linear combination of several/ (1) series is stationary, the series 
are said to be cointegrated (Engle and Granger, 1987). Cointegration tests imply the 
existence of a long-run relationship in /(1) series. Long-run simply means that the series 
moves together over time. 
3.4.2.1 Panel Unil Rool Tests 
The panel unit root test is the first step in the panel based cointegration technique. These 
tests are essential in testing for presence or lack thereof of stationarity. The stationarity 
tests that will be used are the Levin-Lin-Chu unit root test (Levin, Lin, and Chu, 2002). 
The null and alternative hypotheses are as follows: 
Ho: Panels contain unit roots 
Ha: Panels are stationary 
To confirm the results, the IM, Pesaran & Shin W-Stat, ADF _Fisher Chi-square and 
PP _Fisher chi-square panel stationary tests were run. 
24 
3.4.2.2 Cointcgration ,\n;1lysis 
After the precondition of unit root test is met, cointegration tests are then carried out. The 
test of focus will be the (Kao, 1999) test of cointegration. Other tests such as (Perdoni, 
1999) and (Westerlund, 2005) tests of co integration, will be used as confirmatory tests to 
the (Kao, 1999). The null and alternative hypotheses are as follows : 
Ho: Panels are not cointegrated 
Ha: All panels are coilztegrated 
The panel regression model presented by (Kao, 1999) is 
Where Yit and Xit are /(1) and nonintegrated. Xit represents the covariates which are 
required in this test not to be cointegrated among themselves. pi denotes the co integrating 
vector, Yi is a vector of coefficients on Z~t' which is the deterministic term that controls 
for panel-specific effects and linear time trends, and Eit is the error term. 
Having established a cointegrating relationship, the next step is to estimate the long run 
equilibrium relationship given by the Error Correction Term (ECT). The ECT is a measure 
of the extent by which the observed values in time t-1 deviate from the long-run 
equilibrium relationship. Since the variables are cointegrated, any such deviation at time 
t-1 should induce changes in the values of the variables in the next time point in an attempt 
to force the variables back to the long-run equilibrium relationship. The model used in 
this test will be the error correction-based pooled mean group estimator (PMG) developed 
by (Pesaran, Shin, and Smith, 1997) represented as follows: 
p q 
~Yi.t = 0(Yi.t-1 + yXi,t) +I aJ ~Yi,t-j +I Pj ~Xi.t-s + Ei,t 
j=1 s=O 
Where Q) is the enor correction speed of adjustment parameter to be estimated, y IS a 
(kxl) vector of parameters, (Yi.t_1 + yXi,t), is the enor correction term, a are p 
parameters to be estimated and p are q parameters to be estimated, p and q represent the 
number of lags for the variables and Ei,t is the error term. 
25 
3.-L2.3 Vector J\ u toregrcssi\'l~ l'vl od~ l (VAR) 
The V AR model is the second step for building a Granger causality model with a dynamic 
error correction tenn. The V AR model will be used when the panel data is found not to be 
co integrated. V AR is used to capture the linear interdependencies among multiple time 
series. Panel V ARs are particularly suited to analyzing the transmission of idiosyncratic 
shocks across units and time. In V AR models, all variables are treated as endogenous and 
independent. Each of these variables is explained by its own lagged values, the lagged 
values of other model variables and an error term (Ciccarelli, 2013). The appropriate 
number of lags is therefore determined by using the Akaike Information Criterion (AI C) 
and the Schwarz Information Criterion (SIC). The representation ofV AR for a panel data 
IS 
Where Yit is a lxk vector of dependent variables, Xit is a lx/ vector of exogenous 
covariates and Eit is a vector of innovations. The kxk matrices A1, A2 , ... , AP and the lxk 
matrix B are matrices of coefficients to be estimated. 
The vector of endogenous variables to be used in this study is defmed as: 
Xt = l btcamount' btcprice' inflation, intrest rate J 
3.4.2.4 Vnriance Decomposition Analys is 
Variance decompositions give the proportion of the movements in the dependent variables 
that are due to their 'own' shocks, versus shocks to other variables (Brooks, 2008). From 
the variance decompositions, it is usually observed that own series shocks explain most 
of the error variance of the series in a V AR. The variance decomposition was used to aid 
in the interpretation of the results from running the V AR model. 
26 
4 Di\T;\ ,\Ni\LYSIS AND FINDINGS 
This chapter presents the results of the methodology described above as well as an analysis 
and discussion of the generated results and findings. 
4.I Descriptive Statistics 
Descriptive statistics are used to describe the basic features of the data used in this study. 
These statistics, together with graphic plots5, provide simple summaries about the sample 
and form the basis of the quantitative analysis that will be carried out on the data. The 
various descriptive statistics used to generally describe the sample data are presented in 
the table below. 
Table 2: Swnmm y a./Descriptive Statistics 
X USD BTC AMT BT PRICE INFL RATE INT RATE 
MEAN 89.8167 56250499 139203.6 6.5256 7.8750 
MEDIAN 51.3017 483442.3 32042.42 6.4050 7.7500 
MAX 362.4491 1.24E+09 3410433 18.7200 14.0000 
MIN 8.1077 248.0000 2083.888 -0.3000 2.2500 
STD.DEV 99.7894 2.13E+08 406738.9 4.6186 3.9858 
SKEWNESS 1.1894 4.2365 5.4683 0.8730 -0.0222 
KURTOSIS 3.4595 20.5185 36.4847 3.5522 1.7058 
The above table gives a summary of the measures used to describe the sample data. It is 
seen that on average, one-dollar exchanges for 89.8167 of the domestic currencies. The 
mean amount ofbitcoins across the sample is 56250499, which sell at an average price of 
139203.6. It is also seen that the inflation and interest rates average to 6.5256 and 7.8750 
respectively across the selected countries. Skewness is a measure of symmetry and more 
precisely, it measures the lack of symmetry. All the variables except the interest rate 
exhibit positive values for the skewness statistic. This means that the data is skewed to the 
right and therefore lacks symmetry. Interest rates are seen to be negatively skewed as they 
5 The line plots of the different variables are displayed in the appendix section 
27 
report a negative value of skewness. The summary results from the Kurtosis statistic show 
that the exchange, interest and inflation rates have light tails whereas the amount and price 
of bitcoins variables exhibit fat tails. This means that the data fails to follow a normal 
distribution. 
4.2 Stationarity Test : Unit Root Testing 
Stationarity tests were run individually on each variable. The software used for this 
analysis(EViews) was able to run stationary tests using four different tests: the (Levin, 
Lin, & Chu, 2002), (Pesaran, Shin, & Smith, 1997), ADF-Fisher Chi-Square and PP-
Fisher Chi-Square. The results were as follows: 
Table 3: Swnmm y of Panel Stotionari(y Tests 
LEVIN LIN CHU TEST AT LEVEL 1ST DIFFERENCES 
Inflation rate Non-stationary Stationary 
Interest rate Non-stationary Stationary 
Log_btcamt Non-stationary Stationary 
Log_ btcprice Non-stationary Stationary 
USD X Non-stationary Stationary 
IM, PESARAN & SIDN W- AT LEVEL 18T DIFFERENCES 
STAT 
Inflation rate Non-stationary Stationary 
Interest rate Non-stationary Stationary 
Log_btcamt Non-stationary Stationary 
Log_ btcprice Non-stationary Stationary 
USD X Non-stationary Stationary 
ADF FISHER CHI- AT LEVEL 18T DIFFERENCES 
SQUARE 
Inflation rate Non-stationary Stationary 
Interest rate Non-stationary Stationary 
Log_btcamt Non-stationary Stationary 
28 
Log_btcprice Non-stationary Stationary 
USD X Non-stationary Stationary 
PP FISHER CHI- AT LEVEL 1sT DIFFERENCES 
SQUARE 
Inflation rate Non-stationary Stationary 
Interest rate Non-stationary Stationary 
Log_btcamt Non-stationary Stationary 
Log_ btcprice Non-stationary Stationary 
USD X Non-stationary Stationary 
The results of the stationarity tests were consistent across the four tests. All the data was 
found to be non-stationary at level but stationarity was present in the data upon taking first 
differences. This means that any further statistical analyses performed on the data should 
be performed when first differences of the data have been taken. This will enable the 
results make statistical sense and, portray the true behavior and properties of the data6. 
-U Correlat ion Test 
A correlation test was performed to check for multi-collinearity. Multi-collinearity arises 
if there exists a high correlation between any two independent variables. Checking and 
correcting for multi-collinearity is important because it may interfere in determining the 
precise effect of each independent variable. The results of the correlation test are outlined 
in table 4 below. 
6 http://www.eviews.com/help/helpintro.html#page/content%2Fpanei-
Panei_Estimation_Examples.html%23 
29 
Table 4: Correlation among independent Variables 
BTC AMT BTC PRICE INFL RATE INT RATE USD X LN 
- -
BTC AMT LN 1.0000 0.6303 0.4811 0.4991 0.3738 
- -
BTC PRICE LN 0.6303 1.0000 0.1035 0.0705 -0.0871 
- -
INFL RATE 0.4811 0.1035 1.0000 0.8729 0.6278 
INT RATE 0.4991 0.0705 0.8729 1.0000 0.7456 
USD X LN 0.3738 -0.0871 0.6277 0.7456 1.0000 
- -
There was seen to be a high correlation of 87% between inflation and interest rates. This 
is so because any change on the rates of interest will also have an effect on the inflation 
rate, even though indirectly. There was also relatively high correlation of 74% between 
interest rates and the exchange rates. This is in line with theory which holds that an 
increase in interest rates would lead to an increase in exchange rates as well. The interest 
rate variable is therefore an important variable in determining the outcomes of both the 
inflation and exchange rates. 
The test for multi-collinearity usually requires that one of the variables between those with 
high correlations be dropped. In this case however, because interest rates are correlated to 
both an independent variable (inflation rate) and the dependent variable (exchange rate), 
then we fail to drop this variable so as to avoid committing the Omitted Variable 
Bias(OVB). The OVB happens when a variable that is correlated to both the independent 
and dependent variables is excluded from the model of estimation (Buck, 20 14). Exclusion 
of interest rates from the model of estimation will lead to estimation of biased coefficient. 
Therefore, to avoid the OVB, the interest rates will be included in the model estimation. 
4.4 Regress ion Anal ys is 
The purpose of this regression was to check the relationship between the exchange 
rate(USD _X), which forms the dependent variable, and the other independent variables, 
that is, inflation rate, interest rate, bitcoin amount and bitcoin prices. In selecting a suitable 
regression model, both the fixed effects and random effects models were run, and the 
30 
Hausman test performed thereafter to determine the most appropriate model of the two. 
The results of the Hausman test are tabled below. 
Table 5: Hausman Test Results 
Test Summary Chi-sq. Statistic Chi-sq. d.f P-value 
Cross-section random 2.995877 4 0.5585 
The value of interest is the p-value which is compared to the 5% significant level to 
determine whether or not the null hypothesis under this test will be rejected. The p-value 
of 0.5585 is greater than 0.05, and the decision rule is therefore to fail to reject the null 
hypothesis. This means that the random effects model of regression is appropriate. The 
summary of the results from the random effects regression analysis are tabled below. 
Table 6: Summmy a./Regression Analysis Results 
Variable Coefficient Std. Error t-Statistic Pro b. 
c 2.830790 0.590927 4.790422 0.0000 
BTC AMT LN 0.011128 0.003363 3.308860 0.0011 
- -
BTC PRICE LN 0.007233 0.008853 0.817022 0.4150 
- -
INFLATION RATE 0.045344 0.003130 14.48741 0.0000 
INTEREST RATE 0.045084 0.008720 5.170001 0.0000 
When replaced back in the regression equation being tested, the output is as follows: 
log USD_Xit= 2.830790 + 0. 011128log btc_amtit + 0. 007233log btcpriceit 
+ 0. 045344inflationit + 0. 045084interest rate it+ Eit 
The main objective of running this regression was to observe the relationship, magnitude 
and significance of the effect of the independent variables on the dependent variable. It 
can be seen that the all the independent variables have a direct relationship with the 
dependent variable. This means that an increase in either of the independent variable 
results in a corresponding increase in the dependent variable. 
31 
The coefficient of determination (R2) of the model is quite high, and is equal to 0. 7222, 
which means that 72.22% of the variations of the dependent variable are explained by the 
independent variables. This means that the model is a good fit. 
Further, the F-statistic which tests for the joint significance of all the coefficients has a p-
value of less than 0.05 which implies that all the variables used, jointly contribute to 
describing the behavior of the dependent variable. 
4.4 .1 Regression Coefficients 
From the above results, the coefficient column depicts the estimated coefficient values of 
the variables listed in the variables column. These coefficients measure the marginal 
contribution of each of the independent variable to the dependent variable, holding all the 
other variables constant. The first component 'C' in the list of variables, which 
corresponds to the first value under the coefficients column, represents the intercept in 
this regression. 
Taking btc_amt_ln variable, the interpretation of the results would be as follows: holding 
all other variables constant, a one percent increase in the number of bitcoins ( btc _ amt _In) 
leads to a corresponding increase ofO.Ol% in the exchange rate. Similarly, the result for 
the btc _price _In variable coefficient would be interpreted as mentioned above; that a one 
percent change in the price of bitcoins( btc _price _In), holding all the other variables 
constant would lead to a 0.007% increase in the exchange rate. There is a slight difference 
in the interpretation of the results of the inflation and interest rate variables. This is 
because the variables are not in their natural logarithm form and so the coefficient 
interpretation for the inflation rate will be as follows: holding all other variables constant, 
a one percentage point increase in the inflation rate (inflation _rate) leads to a 
corresponding increase of 4.53% in the exchange rate. Additionally, holding all other 
variables constant, a one percentage point increase in the interest rates( interest _rate) leads 
to a corresponding 4.51% increase on the exchange rate. 
4.4.2 Probability (p-value) 
The probability is also known asp-value. Given a p-value, it is then easy to tell whether 
to reject or fail to reject the null hypothesis that the coefficient is zero (insignificant). The 
p-value is compared against a 5%(0.05) significance level. A p-value lower than 0.05 is 
32 
taken as evidence to reject the null hypothesis of a zero coefficient. The p-values from the 
regression results are all less than 0.05, apart from the p-value of the price of 
bitcoins( btc __price _In) variable. The null hypothesis that equates the coefficients to zero 
is rejected for the variables: btc_amt_ln, inflation _rate and interest_rate, and fails to be 
rejected for the btc __price _In) variable. This implies that the coefficients are significant 
and can sufficiently describe the variations in the dependent variable. This implication 
does not hold for the btc __price _In variable which in accordance to the results is seen to be 
insignificant and therefore cannot sufficiently describe the variations on the dependent 
variable. 
4.4 .2 . 1 Normality Tes t 
A series of diagnostic tests are run on the regression residuals. The first test is the Jarque-
Bera Test, which is a test for normality. The residuals are seen to follow a normal 
distribution with the p-value of 0.058 being greater than the significance level of 0.05. 
The null hypothesis that residuals are normally distributed is thus not rejected. The plot of 
the errors is on appendix D. 
4 .5 Panel Co in legmtion 
Cointegration tests were performed after the data was found to have met the cointegration 
condition. This precondition requires that the data be non-stationary at level, but stationary 
upon taking first differences. There were four cointegration tests that were performed on 
the data. These tests were run to check for the existence of a long run relationship mainly 
between the number of bitcoins in an economy and the exchange rates. The variables of 
interest therefore are the btc _ amt _In and the exchange rate. There was consistency of 
results in all the four tests for Cointegration. The table below summarizes the results. 
Tobie 7: Summmy o.lPanel Cointegmtiun Tests 
Test Result 
Perdoni Residual Cointegration tests (No Do not reject Null Hypothesis; No 
Deterministic Trend Assumption) co integration 
33 
Perdoni Residual Co integration tests Do not reject Null Hypothesis; No 
(Deterministic Intercept and Trend co integration 
Assumption) 
Perdoni Residual Cointegration tests (No Do not reject Null Hypothesis; No 
Deterministic Intercept or Trend co integration 
Assumption) 
Kao Residual Cointegration Test Do not reject Null Hypothesis; No 
cointegration 
From the above summary of results, it is clear that there is no co integration between the 
exchange rate and the number of bitcoins in a given economy. This is the consistent 
conclusion across all the models with all the results returningp-values of greater than 5% 
which causes an acceptance of the null hypothesis on no cointegration. 
4.6 Vector Autoregressive Model (VAR) 
A V AR model is performed once panel data is confirmed to have no cointegration. From 
the results outlined in table 7 above, it has been confirmed that there is no existence of 
long run relationships. The unrestricted V AR model is used in this study. The lag order of 
the model was determined using the Schwarz and Akaike Information criteria. The test 
suggested that two lags were suitable. 7 
The results from VAR do not display the p-values and therefore it is difficult to establish 
the significance of the coefficient estimates. However, further regressions are run on the 
individual V AR equations to get p-values of all the coefficients. The corresponding p-
values are outlined in the table below. 
Table 8: P-values ofthe VA R model coefficients 
C{l) 
C(2) 
C(3) 
Coefficient 
1.179666 
-0.183989 
-0.004891 
Std. Error 
0.073863 
0.073909 
0.002880 
t-Statistic 
15.97100 
-2.489403 
-1.698409 
Prob. 
0.0000 
0.0130 
0.0898 
7 The summary table of the VAR results is outlined in the Appendix section. 
34 
C(4) 0.004845 0.002888 1.677553 0.0938 
C(5) -0.004142 0.014158 -0.292547 0.7699 
C(6) 0.002402 0.014253 0.168540 0.8662 
C(7) 0.010437 0.004540 2.298732 0.0218 
C(8) -0.009383 0.004584 -2.046805 0.0410 
C(9) -0.004427 0.006115 -0.723932 0.4693 
C(lO) 0.005428 0.006022 0.901344 0.3677 
C(11) 0.026477 0.024387 1.085728 0.2779 
C(12) -2.342638 1.842224 -1.271636 0.2038 
C(13) 2.184495 1.843369 1.185056 0.2363 
C(14) 0.670034 0.071823 9.328969 0.0000 
C(l5) 0.236056 0.072027 3.277317 0.0011 
C(16) 0.770578 0.353124 2.182176 0.0294 
C(17) -0.675965 0.355482 -1.901543 0.0576 
C(l8) 0.041960 0.113237 0.370550 0.7111 
C(19) -0.047969 0.114332 -0.419562 0.6749 
C(20) 0.188712 0.152511 1.237367 0.2163 
C(21) -0.101742 0.150206 -0.677347 0.4984 
C(22) 0.401982 0.608227 0.660908 0.5088 
C(23) -0.257161 0.390333 -0.658825 0.5102 
C(24) 0.299699 0.390576 0.767326 0.4431 
C(25) 0.001110 0.015218 0.072952 0.9419 
C(26) 0.030421 0.015261 1.993368 0.0465 
C(27) 1.014140 0.074820 13.55433 0.0000 
C(28) -0.022013 0.075320 -0.292260 0.7702 
C(29) -0.020307 0.023993 -0.846357 0.3976 
C(30) 0.025048 0.024225 1.033958 0.3014 
C(31) -0.011743 0.032314 -0.363413 0.7164 
C(32) -0.018498 0.031826 -0.581210 0.5613 
C(33) -0.209814 0.128872 -1 .628077 0.1039 
C(34) 1.589260 1.141722 1.391986 0.1643 
C(35) -1.546302 1.142431 -1.353519 0.1762 
C(36) -0.110294 0.044512 -2.477816 0.0134 
C(37) 0.129918 0.044639 2.910406 0.0037 
C(38) -0.149382 0.218849 -0.682582 0.4951 
C(39) 0.126127 0.220311 0.572494 0.5671 
C(40) 1.363142 0.070179 19.42382 0.0000 
C(41) -0.406482 0.070858 -5.736602 0.0000 
C(42) -0.005563 0.094519 -0.058851 0.9531 
C(43) 0.033055 0.093091 0.355088 0.7226 
C(44) -0.065627 0.376950 -0.174100 0.8618 
C(45) 3.294828 0.890702 3.699136 0.0002 
C(46) -3.064653 0.891256 -3.438579 0.0006 
C(47) 0.003613 0.034726 0.104029 0.9172 
C(48) 0.003804 0.034825 0.109225 0.9130 
C(49) -0.379183 0.170733 -2.220917 0.0266 
C(50) 0.425567 0.171873 2.476050 0.0135 
C(51) 0.080129 0.054749 1.463566 0.1437 
C(52) -0.060183 0.055279 -1.088723 0.2766 
35 
C(53) 
C(54) 
C(55) 
0.895741 
0.001253 
-0.717768 
0.073738 
0.072624 
0.294073 
12.14760 
0.017258 
-2.440780 
0.0000 
0.9862 
0.0149 
Whether the coefficients are significant in explaining the dependent variables or not is 
detennined by observing the respective p-values. The total number of coefficients sums 
up to 55 because the model had 5 regression equations each with 11 independent variables. 
To establish the significance of each of the coefficients, the p-values are compared against 
the 5% (0.05) significance level and checked against the decision rule of whether to reject 
or fail to reject the null hypothesis. From the results tabled above, it is seen that when 
having the exchange rate as the dependent variable, only the lags of the exchange rate and 
the inflation rates have a statistically significant effect on the exchange rate. 
4.6.1 Variance Decomposition 
The variance decomposition analysis allocates the proportion of the movement in the 
dependent variables that are due to their own shocks and shocks due to the other variables. 
The summary and discussion of results are outlined in the section below. 
4. 6. 1. I J ·aria nee Decomposition o(' USD _.i{_LN 
Table 9: Variance Decomposition Analysis for the exchange mte(USD _X_ LN) 
BTC AMT L BTC PRICE INTEREST RINFLATION 
- - - - - -
Period S.E. USD X LN N LN ATE RATE 
0.034198 100.0000 0.000000 0.000000 0.000000 0.000000 
2 0.053204 98.31357 0.529799 0.004756 0.104560 1.047310 
3 0.068187 96.91768 0.679635 0.004337 0.132077 2.266275 
4 0.080857 95.80988 0.751210 0.015234 0.115481 3.308198 
5 0.092039 94.94606 0.772521 0.045471 0.091542 4.144411 
6 0.102174 94.23326 0.771720 0.096183 0.075334 4.823505 
7 0.111533 93.60841 0.761695 0.167496 0.073396 5.389006 
8 0.120299 93.03142 0.749106 0.258683 0.088305 5.872490 
9 0.128602 92.47775 0.737488 0.368670 0.120641 6.295455 
10 0.136534 91 .93267 0.728768 0.496236 0.169923 6.672400 
36 
From Table 10, it can be seen that the exchange rate (USD _X_ LN) explains its own 
variability up to 100% in month one and loses the power of explanation to about 91% in 
month 10. After 3 months, changes in the exchange rate are still greatly attributed to itself, 
with only 0.67% attributed to btc_amt_ln , 0.004% attributed to the price of bitcoins, 
0.13% to interest rates and 2.27% to inflation. 
In a period of 10 months, the major external explanation for the variability in exchange 
rate remains to be the inflation rate which contributes to about 6.67% of variations in the 
exchange rate. The number of bitcoins, bitcoin prices and exchange rates still contribute 
minimally to variation in the exchange rate of about 0.73%, 0.49% and 0.17% 
respectively. 
4.6. 1.2 I ·rtriunc:e Dec:ol/lposition r~/BTC_AMT_LN 
Table 10: Variance Decomposition Ana /ysisfor the amount ofbitcoins(BTC_AMT_LN) 
BTC AMT L BTC PRICE INTEREST RINFLATION 
- - - - - -
Period S.E. USD X LN N LN ATE RATE 
0.852936 0.041014 99.95899 0.000000 0.000000 0.000000 
2 1.044099 0.366947 97.53971 1.508632 0.540756 0.043957 
3 1.209451 0.378879 97.12015 1.635186 0.830177 0.035605 
4 1.340174 0.382592 96.60351 1.891567 1.090471 0.031858 
5 1.449710 0.377038 96.12704 2.144340 1.322172 0.029414 
6 1.543880 0.367759 95.65608 2.419087 1.528912 0.028168 
7 1.626491 0.357032 95.18337 2.720245 1.711444 0.027910 
8 1.700210 0.345909 94.70520 3.050755 1.869594 0.028540 
9 1.766972 0.334953 94.21 894 3.412666 2.003434 0.030012 
10 1.828238 0.324446 93 .72270 3.807159 2.113395 0.032299 
From Table 11 , it can be seen the btc-amt-ln variable explains its own variability up to 
99.95% in the first month and loses the power of explanation to about 93 .72% after 10 
months. The major external explanation for this variability is the price of bitcoins 
(BTC_PRICE_LN), as it explains 0% in the first month and increases to 3.8% after 10 
37 
months. The exchange rates also contribute to the variations in the amount of bitcoins, 
although minimally. In the first three months, the exchange rate contributes up to 0.38% 
of variations in the amount ofbitcoins. This percentage slightly reduces to 0.32% after 10 
months. This implies that in the exchange rate causes minimal variations in the amount of 
bitcoins in a given economy given a period of 10 months. The contribution of the interest 
rates is seen to increase from 0.83% in 3 months to 3.8% after 10 months. The amount of 
bitcoins is seen to be least sensitive to variations in the inflation rates. In a period of three 
months, variations due to the inflation rate amount to just 0.035%. this rate is seen to 
slightly reduce and then increase to 0.032% after ten months.8 
8The variance decomposition results of btc_price_ln, inflation_rate and interest_ rate are in the Appendix 
section labelled Table 11, Table 12 and Table 13 respectively. 
38 
5 DISCUSSIONS, CONCLUSIONS AND RECOtvliVIENDATIONS 
5.1 Summary or Findings 
The objective of this research was to evaluate the relationship between the circulation of 
bitcoins and selected domestic currencies in Africa. The Random effects regression model 
was used to analyze the relationship between the exchange rates and variables such as the 
amount and price of bitcoin, interest and inflation rates. The random effects model 
exploited the direction of the relationship between these variables, the magnitude and 
significance of this relationship. In addition, the test of cointegration was performed to 
check for the existence of any long run relati9nships between the variable. The 
cointegration tests confirmed the absence of a long run relationship and this conclusion 
led to the use of the V AR and Variance decomposition models. The VAR model was used 
to analyze the endogeneity of the variables and explore the effects of lagged independent 
variables on the dependent variable. The variance decomposition analysis was an analysis 
extension of the V AR model. It explored the sensitivity of one variable to its own shocks 
and those of other variables in the V AR system. It also showed how long the shocks 
would last within the V AR system. 
5 .1.1 Discussion of the Regression Findings 
It is clearly seen from the regression results displayed in table 6 that there exists a direct 
relationship between the exchange rates and the all the independent variables used in the 
regression model. The increase in the exchange rate by 0.01% due to a percentage change 
in the amount of bitcoins means that more of the domestic currency will be required to 
exchange for one dollar, thus indicating a loss in value of the domestic currency. On a 
general note, increase in the exchange rate in this context means a depreciation of that 
currency. This is because more of the domestic currency will be required to exchange for 
one dollar. The results also showed a direct relationship between the exchange rates and 
the interest and inflation rates. Economic theory holds that countries with higher inflation 
typically see depreciation in their currency in relation to the currencies of their trading 
partners. 
The inflation regression results are consistent with this economic theory; a percentage 
increase in the inflation rate leads to a corresponding increase of 4.53% in the exchange 
39 
rate, which means that the domestic currency depreciates with respect to the dollar. 
Economic theory holds that increasing interest rates attract foreign capital and thus leads 
to appreciation of the exchange rate. This means that theoretically, there exists an inverse 
relationship between interest and exchange rates. This is not the case with the regression 
results. The results from this model show that interest and exchange rates have a direct 
relationship, with an increase in interest rates resulting in a conesponding increase 
(depreciation) of the exchange rate. These can be said to be contrary to what economic 
theory stipulates. The relationship between the exchange rates and the prices of bitcoins 
is also a direct one. The coefficient of this result is however considered insignificant in 
explaining the dependent variable because the p-value was greater than the level of 
significance, and therefore the null hypothesis that deemed the coefficient insignificant 
failed to be rejected. 
Except for the minimal amount ofbitcoins and small effect ofO.Ol% on the exchange rate, 
the results of this research are found to be consistent with those of (Alina, 20 16), who also 
performed a regression analysis of the exchange rate against the amount ofbitcoins traded 
in the United States. The research titled, 'A Regression Analysis of Cryptocurrency 
Influence on the Us Dollar', yielded similar results with a percentage increase in the 
amount ofbitcoins resulting to a corresponding and larger effect of a 6.1% increase on the 
exchange rate. The conclusion drawn from this research was that the increased distribution 
ofbitcoins weakened the US dollar. 
The conclusion made from this research therefore is that, the increased distribution of 
bitcoins across African countries, that is, Kenya, Morocco, Nigeria and South Africa, 
weakens the domestic cunencies in the respective countries. 
5.1.2 Summary of the VAR and Variance Decompositions Findings 
The results from the V AR and variance decomposition tests further explored the 
relationship between bitcoins and exchange rates. These results contribute further to the 
objective of this study of evaluating the relationship between the circulation of bitcoins 
and the exchange rates. 
40 
5.1.:3.1 Suii !IIIUIT r?f I: 1R results 
The results show that only the lags of the exchange rate and those of the inflation rate have 
a statistically significant relationship with the exchange rate. This means that a critical 
observation of the lags of the exchange and inflation rates would give meaningful 
information about the current value of the exchange rate. 
When the btc _ amt _ln variable became the dependent variable, it was seen that there was 
only a statistically significant relationship between btc _ amt _ln with its lags and those of 
bitcoin prices. It is therefore concluded that observing the lags of the amount of bitcoins 
and those of the prices of bitcoins would inform on the current value of the amount of 
bitcoins. 
Having the btc _price _In variable as the dependent variable showed that it only has a 
statistically significant relationship with its own lags, hence observing the lags of the 
bitcoin prices would greatly inform the current price of bitcoins. When the inflation rate 
became the dependent variable, it was seen to have a statistically significant relationship 
with its own lags and those of the btc_amt_ln variable. 
Finally, setting interest rates as the dependent variable showed that it has statistically 
significant relationships with its own lags, those of the bitcoin prices and the exchange 
rates as well. This means that observing the lagged values of the interest and exchange 
rates, and the bitcoin prices would inform on the current value of interest rate. 
5.1.2.2 Sulllnwrv of' /"oriance Decolllposition !lnall'sis 
The variance decomposition analysis showed that the variations in the exchange rate 
attributed to changes in the amount ofbitcoins are relatively small, with a value ofO. 73%. 
These results however show an increase followed by a decrease in the proportion of 
variation that is accounted for by the amount of bitcoins. Although results show the 
existence of sensitivity between these variables over time, the value of this sensitivity is 
quite low and could be considered almost negligible. 
When the variables are interchanged, the results show that over a period of 10 months, the 
variations in the amount of bitcoins that can be attributed to the exchange rates equal to a 
value of0.32%. This value is seen to be decreasing minimally over the ten-month period. 
41 
It is clearly seen that there exists some relationship between the amount of bitcoins and 
the exchange rates over time. The variations brought on the exchange rate by the amount 
of bitcoins in circulation are inconsistent over time, with periods of an increasing effect 
followed by periods of a decreasing effect. On the other hand, the variations in the amount 
ofbitcoins due to the exchange rates are seen to be consistently decreasing over time. 
Research done by (Asmah, 2013) on the Sources of Real Exchange Rate Fluctuations in 
Ghana, established that the exchange rate variable is highly endogenous and therefore it 
accounts for a great percentage of its own variations. This further supports why very 
minimal changes in the exchange rate are attributed to exogenous variables. 
5.2 Conclusion of the Study 
The results from this study clearly show that there exists a relationship between the 
exchange rates and the amount of bitcoins. From empirical testing, it can be concluded 
that as the amount of bitcoins increase in any economy, the domestic currency in that 
economy is seen to depreciate. This effect however is very minimal and almost negligible. 
This could be attributed to the few bitcoins that are in circulation in the countries of study. 
A similar research when carried out in the United States of America, had the same 
conclusions but with the amount of bitcoins having a greater impact on the exchange rate 
due to a larger number ofbitcoins in circulation in that economy. 
The results further show that the amount of bitcoins in circulation in any given economy 
would also cause minimal variations to the exchange rates, interest rates and inflation rates 
in that economy. 
5.3 Limitations of the Study 
The main limitations to this study are as follows . The variable, btc _amount _In, registered 
minimal data. There was limited data on the amount ofbitcoins in Africa. This is because 
there is still much uncertainty around the concept of Bitcoin and its usage. Bitcoin is not 
as popular in African countries as in other westernized countries. Most African countries 
did not register any volumes of bitcoins in their economies. The countries that did have 
Bitcoin trading, had very minimal volumes, and some did not even have records of this 
kind of data. 
42 
The other limitation was that this research was done over a relatively short time period. 
The sample was of monthly data collected from 2014 to 2017. This time period was so 
because it was not until2014 that the Bitcoin system started to gain popularity and some 
momentum in Africa. This is a relatively short period of examination in which data 
characteristics may not be fully exploited. 
5.-J. Recommendations 
5.4.1 Policy Recommendations 
The exchange rate of any economy is important for the purposes of international trade as 
well as investments. From this study, an increase in the amount of bitcoins in a given 
economy is seen to weaken the domestic currency of that economy. This raises the need 
for governments to keep a watchful eye over the Bitcoin system that is slowly gaining 
momentum in African economies. This research concluded that this impact on the 
domestic currencies by Bitcoin is almost negligible. This has been attributed to the 
minimal volumes ofbitcoins in most African countries. However, as more people become 
more familiar with the Bitcoin system, the volumes are bound to shift upwards, thus 
causing more significant impacts. 
Governments should monitor the traction that the Bitcoin system is garmng m the 
respective economies and come up with risk management measures given that the 
operations of Bitcoin are beyond the monetary and fiscal policies. Governments should 
also seek to establish ways through which they could incorporate the Bitcoin system so as 
to improve operations within their countries. One approach to this is for central banks to 
develop their own central-bank-backed cryptocurrencies. Some central banks that have 
already taken this up include the central banks of Canada, China, Estonia, Eucador, Japan, 
Russia, Senegal, Singapore, Sweden, and Tunisia (Enzozo, 2017). This way, central banks 
will continue to have full control over the monetary system operations. Even in developing 
their own cryptocurrencies, central banks should greatly focus on how the cryptocurrency 
system will be designed, built and maintained, and how it could affect the country's 
fmancial stability. 
Another approach is to explore the technology on which the Bitcoin system is built; the 
blockchain technology which is an online distributed database. The blockchain can be 
43 

adopted in improving public sector operations with the aim of improving the resilience of 
digital infrastmcture. A country such as Estonia is already taking pride in being the first 
'crypto-country' in the world, and has adopted the blockchain in public sector services. 
These include adopting an online voting system, payment of health benefits, and an e-
residency program which allows people around the world to be virtual citizens of Estonia 
(Althauser, 2017). Countries could borrow from what Estonia has done and try to adopt 
and incorporate the blockchain to improve the efficiency of their operations. 
5.4.2 Recommendations for Further Research 
There are two main recommendations for further research. The first is that, the sample 
selected should include a majority of countries in which the Bitcoin system has taken root 
and found acceptance. This will allow for more volumes of bitcoins to be registered and 
hence more significant characteristics of Bitcoin can be explored. 
The fmal recommendation is that; the sample period should be extended to more than four 
years. This will also allow empirical tests to fully explore and capture the behaviors of 
Bitcoin over a longer period. 
44 

2. INTEREST RATE 
INTEREST _RATE 
2 
12 7.2 
6.8 
11 
6.< 
10 tiO 
:i6 
9 
:i2 
'-8 
I n HI N I II Ill N I II Ill rv I II Ill rv I II m ~I I II m N I n m N I II ill rv 
a11! :1015 2)16 2017 2'JI< 2'JI5 2016 2017 
3 
1·5 :3.2 
I< 3.0 
1'J 2.8 
12 2.6 
II 2.4 
10 2.2 
n m N I II Ill ~I l II m N I II m rv I II m N l II Ill N I II m rv I II m N 
'2.11-' 2015 2016 2'JI7 2\)f! 2015 201ol 2017 
3. X USD EXCHANGE RATE 
X_USD_LN 
2 
~70 2.9 
!...S5 2.8 
2.7 
<60 
2.o 
~ 
2.5 
<.:il 2.< 
.!::05 2.J 
I ft Dl N I II 
"' 
rv I n m ~I I n m rv I n m rv I n 
"' 
rv I II m rv I II m rv 
2.11<!. 2015 2'.l1o 2'J17 201< 2315 2010 2'J17 
3 
tiO 2.'.1.1 
sa 2.28 
2.2! 
s.s 
2.2() 
:i< 
2.16 
:i2 2.12 
:iO 2J):') 
I n m Til I n Ill rv I I m rv I ft 111 rv l u m ~I l n 111 N I ft m N I • IB rv 2'J1< 2'J15 2'J16 2'J17 201< 2015 2')16 2017 
52 

l 
4. BTC AMT LN 
BTC_At.1T_LN 
2 
18 2< 
16 2'J 
1< 
16 
12 
10 12 
a 
1 11 Ill lVI 11 Ill IV I II Ill IV I II Ill IV I 11 Ill ~~ I n 1U lV I fl 1>/ I 11 111 IV 
2J1! ~15 :2<)16 2017 201~ 2.)13 2'J16 2'J17 
3 
17 1< 
16 12 
15 10 
1< a 
1J 
" 
12 
I a m 1>/ I 11 01 IV I II n1 IV I n m rv I u 01 rv I ~ m rv1 n m f'll n m rv 
201< 2015 2'J16 2'J17 201! 2'J15 2'J1ol 2'J17 
5. BTC PRICE LN 
BTC_PRICE_LN 
2 
1< 16 
13 15 
1< 
12 
13 
11 
12 
10 11 
9 10 
rv I fl m wv I iJJ H1 IV I n m rv 
2tl1~ 2015 :2016 :2017 2tl1< 2tJ1.5 2l1d 2tl17 
3 4 
12 12 
11 11 
10 10 
9 ~ 9 a a ~~ 
7 
a HI rv 1 D m IV 1 D Ill rv I n m IV 
:201~ :2015 2016 2017 2l15 2>l1S 2017 
53 

From all the graphs: 1, 2, 3, and 4 represent the four cross sections which are Kenya, 
Nigeria, south Africa and Morocco respectively. 
Appendix D: Histogram: Nonnalily Test for Regression Residuals 
Series: Standardized Residuals 
Sample 20141.101 201n112 
Observations 192 
Mean -4.50e-17 
Median -0 .001743 
lilaximum 0.246344 
Minimum -0.270342 
Std. Dev. 0.085052 
Skewness 0.256483 
Kurtosis 3.669326 
Jarque-Bera 5.689052 
Probability 0.058162 
Appendix E: VAR and Va ria nce Decomposition Tables 
Table 11: Summari::ed R esults of'tlie VA R Model 
fNFLA TION _ RATNTEREST _RAT 
X_USD_LN BTC_AMT_LN BTC_PRICE_LN TE E 
X_USD_LN(-1) 1.179666 -2.342638 -0.257161 1.589260 3.294828 
(0.07386) (1 .84222) (0.39033) (1.14172) (0.89070) 
[ 15.9710) [ -1.271 64] [-0.65882) [ 1.39199] [ 3.69914) 
X_USD_LN(-2) -0.183989 2.184495 0.299699 -1.546302 -3.064653 
(0.07391) ( 1.84337) (0.39058) (1.14243) (0.89126) 
[-2.48940] [ 1.18506] [ 0.76733] [ -1.35352) [-3.43858] 
BTC_AMT_LN(-1) -0.004891 0.670034 0.001110 -0.110294 0.003613 
(0.00288) (0.07182) (0.01522) (0.04451) (0.03473) 
(-1.69841 ] [ 9.32897] [ 0.07295) [ -2.47782] [ 0.10403) 
BTC_AMT_LN(-2) 0.004845 0.236056 0.030421 0.129918 0.003804 
(0.00289) (0.07203) (0.01526) (0.04464) (0.03482) 
[ 1.67755) [ 3.27732] [ 1.99337] [ 2.91041] [ 0.10922) 
BTC_PRICE_LN(-1) -0.004142 0.770578 1.014140 -0.149382 -0.379183 
(0.01416) (0.35312) (0.07482) (0.21885) (0.17073) 
[-0.29255) [ 2.18218] [ 13.5543) [-0.68258] [-2.22092) 
BTC_PRICE_LN(-2) 0.002402 -0.675965 -0.022013 0.126127 0.425567 
(0.01425) (0.35548) (0.07532) (0.22031) (0.17187) 
54 

[ 0.16854] [-1 .90154] [-0 .29226] [ 0.57249] [ 2.47605] 
INFLATION_RATE(-1) 0.010437 0.041960 -0.020307 1.363142 0.080129 
(0.00454) (0.11324) (0.02399) (0 07018) (0.05475) 
[ 2.29873] [ 0.37055] [-0.84636] [ 19.4238] [ 1.46357] 
I NFLATION_RATE( -2) -0.009383 -0047969 0.025048 -0.406482 -0.060183 
(0.00458) (0.11433) (0.02422) (0.07086) (0.05528) 
[ -2 .04681 ] [ -0.41956] [ 1 03396] [ -5 .73660] [ -1 .08872] 
INTEREST _RATE( -1) -0 004427 0.188712 -0.0117 43 -0.005563 0.895741 
(0.00611) (0.15251) (0.03231) (0.09452) (0 07374) 
[-0.72393] [ 1 23737] [ -0 .36341] [-0 05885] [ 12.1476] 
INTEREST _RATE( -2) 0.005428 -0.101742 -0.018498 0.033055 0.001253 
(0 00602) (0.15021) (0.03183) (0.09309) (0.07262) 
[ 0.90134] [-0.67735] [ -0.58121] [ 0.35509] [ 0 01726] 
c 0.026477 0.401982 -0.209814 -0 065627 -0.717768 
(0.02439) (0.60823) (0.12887) (0.37695) (0 .29407) 
[ 1.08573] [ 0.66091] [ -1.62808] [-0.17410] [-2.44078] 
N ate.' Standard errors in (}and t-statistics in {} 
T,/),, , 7: \/1ri :1 n r .. c • D r}<;UI1 1{>Cx:> ition An.1/ys i s fnr lttr• 1-.,, . , c:t""' ('")/ 1-:Jilcoin:·:(BTC_PR/CE_LN) 
BTC_AMT_L BT(_PRICE 
-
INTEREST_R INFLATION 
P er-iod S.E. USD_X_ LN N LN ATE RATE 
1 0.180722 0.366065 0.049461 99.58447 0.000000 0.000000 
2 0.256954 0.190206 0.048873 99.55453 0.036407 0.169982 
3 0.318506 0.123964 1.090283 98.20047 0.281885 0.303402 
4 0.375992 0.101515 2.581834 96.34779 0.624616 0.344247 
5 0.431086 0.109615 4.497530 94.02991 1.019384 0.343557 
6 0.485101 0.137782 6.660128 91.45860 1.420744 0.322746 
7 0.538625 0.178586 8.957341 88.76971 1.799638 0.294726 
8 0.591983 0.227029 11.30723 86.06065 2.139326 0.265759 
9 0.645343 0.279728 13.65199 83.39768 2.431982 0.238623 
10 0.698795 0.334371 15.95210 80.82348 2.675707 0.214342 
T ,,,,,., 2: V ,-,, ,,,_e Q,,cr~rnJ'<>Si/ iun A,.,, •lv:>is tor 111<' in l lnli nn r.- ,, ,.(/NFLA T/QN_RI-\ TE) 
Peri o d 
1 
2 
S.E. 
0.528609 
0.891481 
BTC_AMT_L BTC_PRICE_I NTEREST_R INFLATION 
USD_X_LN N LN ATE RATE 
0.010532 
0.241045 
0.528134 
0.411864 
55 
2.039951 
1.451415 
0.017849 
0.013115 
97.40353 
97.88256 

3 1.179199 0.617270 0.310382 1.036766 0.012195 98.02339 
4 1.408671 0.906342 0.240958 0.807502 0.031038 98.01416 
5 1.592823 1.139847 0.189089 0.653675 0.080549 97.93684 
6 1.7 42903 1.339439 0.163841 0.548276 0.167354 97.78109 
7 1.867282 1.519393 0.166706 0.478562 0.293381 97.54196 
8 1.972041 1.688315 0.194620 0.437522 0.457493 97.22205 
' 
9 2.061585 1.851412 0.242521 0.420434 0.6567 48 96 .82888 
., 
10 2.139132 2.011906 0.304903 0.423636 0.887179 96.37238 
T 11">1<' 3 1/.11 i.-,n,-,, Oc•cu,npn.->il i011 An"tvsis '"' Ilk• in /,'rn::;l ', llc(/NFT/::REST_ R/'4 TE} 
BTC_AMT _L BTc_PRICE 
-
INTEREST_R INFLATION 
P e riod S.E. USD_X_LN N LN ATE RATE 
0.412389 0.004150 0.222360 3.188089 96.58540 0.000000 
2 0.576324 3.370834 0.260279 6.587465 89.25529 0.526129 
3 0.686515 5.942775 0.193347 6.942112 85.01556 1.906202 
4 0.771295 7.7 44131 0.153719 6.776286 81.86843 3.457433 
ll~ 5 0.839931 9.144651 0.131424 6.486953 79.32075 4.916223 6 0.897051 10.30720 0.116491 6.162917 77.17630 6.237096 7 0.945544 11.31849 0.105049 5.840576 75 .30689 7.428995 
8 0.987383 12.22633 0.096537 5.533941 73.63105 8.512133 
9 1.023987 13.05898 0.091887 5.249039 72.09564 9.504454 
10 1 056413 13.83389 0.092714 4.988291 70.66562 10.41948 
56 
6 REFEREJ\CE 
Abar, R. R. (2015) . Influence of Macroeconomic Variables on Exchange Rates. Journal 
of Economics, Business and Management, Vol3, No.2, February 2015, 276-281. 
Abdoh, W. M., Yusuf, N. H., Zulkifli, S. A., Bulot, N., & Ibrahim, a. N. (2016). 
Macroeconomic Factors That Influence Exchange Rate Fluctuation in ASEAN 
Countries. International Academic Research Journal of Social Sciences, 89-94. 
Abdullah, N. A. (2010). The Relationship between Inflation and Real Exchange Rate: 
Comparative Study between ASEAN+3, the EU and North America. European 
Journal of Economics, Finance and Administrative Sciences, ISSN 1450-2887, 
Issue 18, 69-76. 
Agoya, V. (2015, January 14). Corporate News. Retrieved from Business daily: 
http://www.businessdailyafrica.com/corporate/Hackers-demanded-Sh6-2m-in-
bitcoins/539550-2589384-3v8fe5/index.htrnl 
Alina, K. (20 16). A Regression Analysis of Cryptocurrency Influence on the US Dollar. 
Journal of Macroeconomics and Monetary Economics, 7. 
A1thauser, J. (2017, October 13). The Cointelegraph; Future of Money. Retrieved from 
Race Towards the First 'Crypto-country' m the World: 
https:/ I cointelegraph.corn/news/race-towards-the-first -crypto-country-in-the-
world 
Apsell, P. (1996, November 26) . The History of Money. Retrieved from NOV A: 
http://www.pbs.org/wgbh/nova/ancientlhistory-money.html 
Asmah, E. E. (2013). Sources of Real Exchange Rate Fluctuations in Ghana. American 
Journal of Economics, 291-302. 
Bajpai, P . (2015, April 15). Forex: Countries where Bitcoin is Legal. Retrieved from 
Investopedia: http :/ /www.investopedia.com/articles/forex/041515/countries-
where-bitcoin-legal-illegal.asp 
Barber, Boyen, a. X. , Shi, a. E. , & Uzun, &. E. (2012). Bitter to Better-How to Make 
Bitcoin a Better Currency. Financial Cryptography and Data Security, 399-414. 
45 
BitPesa. (2014, September 22). Blog: HOW TO USE BITPESA TO ACCEPT PAYMENTS 
FOR YOUR BUSINESS. Retrieved from BitPesa: 
https:/ /www. bitpesa.co/blog/how-to-use-bitpesa-to-accept -payments-for-your-
business/ 
Bitrefill. (2015). About Bitreill. Retrieved from Bitrefill: https://en.bitrefill.corn/about/ 
BlockChain. (2017, 5 5). Information Charts. Retrieved from BlockChain: 
https://blockchain.info/charts/total-bitcoins 
Bohme, R., N. C., B. E., & Moore, a. T. (2015). Bitcoin: Economics, Technology, and 
Governance. Journal ofFinacial Perspectives-Volume 29, Number 2, 2-5. 
Brooks, C. (2008). Introductory Econometrics for Finance. United States of America: 
Cambridge University Press. 
Buck, S. (2014). Omitted Variable Bias versus Multicollinearity. Introductory Applied 
Econometrics, 10. 
Cagan, P. (1958). The Demand For Currency Relative To The Total Money Supply . 
Journal of Political Economy, 303-328. 
Callander, R. (2014, October 20). Money> The history of money: from barter to bitcoin. 
Retrieved from The Telegraph: 
http://www. telegraph.co.uk/fmance/businessclub/money/111740 13/The-history-
of-money-from-barter-to-bitcoin.html 
Centrak Bank of Kenya. (2017, May 3). Currency History. Retrieved from Central Bank 
ofKenya: https://www.centralbank.go.ke/currency-history/ 
Ciccarelli, F. C. (2013) . Panel Vector Autoregressive Models; A Survey. European 
Central Bank Working Paper Series, No 1507, 1-55. 
Coin Dance. (2017, May 4). Bitcoin Statistics . Retrieved from Coin.Dance: 
https:/ /coin.dance/stats#marketcap 
CoinDance. (2017, May). Local Bitcoins Volume charts. Retrieved from Coin Dance: 
https:/ I coin.dance/volume/localbitcoins 
46 
CoinMKTCap. (2017, July 2). Cryptocurrency Market Capitalizations . Retrieved from 
CoinMarketCap.com: http://coinmarketcap.com/all/views/all/ 
CoinPursuit. (2014). Bitcoin Acceptance by Country. Retrieved from CoinPursuit: 
https://www.coinpursuit.com/pages/countries-that-accept-bitcoin/ 
Decker, C., & Wattenhofer, a. R. (2013). Information Propagation in the Bitcoin Network. 
IEEE Journal, 1-10. 
Dinu, A. (2014). The ScarcityofMoney: The case ofCryptocurrencies. 9. 
Engle, R. F., & Granger, C. (1987). Co-Integration and Error Correction: Representation, 
Estimation, and Testing. Econometrica, Vol. 55, No. 2., 251-276. 
Enzozo. (2017, September 27). Verdict. Retrieved from Move Over Bitcoin, These 
Countries are Creating their own Digital Currencies: 
https:/ /www. verdict. co. uklbitcoin-countries-digital-currency/ 
Falko Juessen, L. L. (n.d.). Estimating panel V ARs from macroeconomic data. 
Farell, R . (2015). An Analysis of the Cryptocurrency Industry. Wharton Research 
Scholars Journal, 23. 
Frenkel, J. A. (1976). A Monetary Approach to the Exchange Rate: Doctrinal Aspects and 
Empirical Evidence. Scandinavian Journal of Economics, vol. 78, Issue 2, 200-
224. 
Gascoigne, B. (1993). History of Money. Retrieved from History World: 
http://www .historyworld.net/wrldhis/PlainTextHistories.asp ?historyid=ab 14# 125 
6 
Glyn, D. (2002). A History of Money: From Ancient Times to the Present Day. Cardiff: 
University of Wales Press. 
Grinberg, R. (2011). Bitcoin: An Innovative Alternative Digital Currency. Hastings 
Science and Technology Law Journal, 2-5. 
Kao, C. (1999) . Spurious Regression and Residual-Based Tests for Cointegration in Panel 
Data. Journal of Econometrics, 1-44. 
47 
Kouri, J. B. (1978). Exchange Rates and the International Adjustment Process. Brooldngs 
Papers on Economic Activity, 1-47, Voll. 
Levin, A., Lin, C.-F., & Chu, C.-S. J. (2002). Unit Root Tests in Panel Data: Asymptotic 
and Finite-sample Properties. Journal of Econometrics, Vol 108, Issue 1, 1-24. 
Loseva, A. (2016). Bitcoin: A regression Analysis of Cryptocurrency Influence on 
Russian Economy. Department of economics; Moscow State University , 2-7. 
Lyons, R. K. (2001). The Microstructure Approach to Exchange Rates. Cambridge, 
Massachusetts: The MIT Press. 
Mavee, N., Perrelli, R., & Schimmelpfennig, a. A. (20 16). Surprise, Surprise: What Drives 
the Rand/U.S Dollar Exchange Rate Volatility? International Monetary Fund 
Working Paper Series, WP/16/205, 2-38. 
Nakamoto, S. (2008). Bitcoin: A Peer-to-Peer Electronic Cash System. 
Ndung'u, D. D. (1999). A Dynamic Model of Inflation for Kenya, 1974-1996. 
International Monetary Fund Working Paper, 2-36. 
NIST. (2012, April 7). Measures of Skewness and Kurtosis. Retrieved from NIST 
SEMATECH; ENGINEERING STATISTICS HANDBOOK: 
http://www.itl.nist.gov/div898/handbook/eda/section3/eda35b.htm 
Njuguna, M. E. (2014). Adoption ofBitcoin in Kenya, A Case Study ofBitpesa. 1-34. 
Omri, M. Y. (2013). Are Cryptocurrencies 'Super' Tax Havens? 112 Michigan Law 
Review First Impressions 38(2013), 11. 
Pagliery, J. (2014, February 26). Money.cnn.com. Retrieved from CNNTech: 
http:l/money.cnn.com/2014/02/25/technology/security/bitcoin-mtgox/ 
Patel, P. P., JPatel, D. N., & R.Patel, a. D. (2014). Factors Affecting Currency Exchange 
Rate, Economical Formulas and Prediction Models. International Journal of 
Application or Innovation in Engineering & Management, Volume 3, Issue 3, 53-
56. 
48 
Pearce, D. K. (1983). Alternative Views Of Exchange-Rate Determination . Federal 
Reserve Bank of Kansas City Economic Review, 16-31 . 
Perdoni, P. (1999). Critical Values for Co integration Tests in Heterogenous Panels with 
Multiple Regressors. Oxford Bulletin of Economics and Statistics, Vol.61, Issue 0, 
653-670. 
Pesaran, H., Shin, Y., & Smith, a. R. (1997). Pooled Mean Group Estimation ofDynamic 
Heterogeneous Panels. Journal of the American Statistical Association , 621-634. 
Prentis, M. (20 15). Digital Metal: Regulating Bitcoin as a Commodity. Case Western 
Reserve Law Review, 31 . 
Proti, N. P. (2013). Exchange Rate Fluctuations-Shock m Tanzania; An Empirical 
Analysis. Scholarly Journal of Business Administration, Vol.3(1), 12-19. 
Stockman, A. (1980). A Theory of Exchange Rate Determination. The Journal of Political 
Economy, 673-698, Vol.88, No.4. 
Stockman, A. (1980). A Theory of Exchange Rate Determination. Journal of Political 
Economy 88, no.4, 673-698. 
Szczepanski, M. (2014). Bitcoin: Market, Economics and Regulation. European 
Parliamentary Research Service, 3. 
Turpin, J. B. (2014). Bitcoin: The Economic Case for a Global, Virtual Currency 
Operating in an Unexplored Legal Framework. Indiana Journal of Global Legal 
Studies, 335-339. 
Westerlund, J. (2005) . New Simple Tests for Panel Cointegration. Econometric Reviews, 
Vol. 24, Issue 3, 297-316. 
XE Currence Converter. (2017, 3 22). Retrieved from XE Website: 
http://www.xe.com/currencyconverter/convert/? Amount= 1 &From= XBT &To=U 
SD 
Yermack, D. (2013) . Is Bitcoin A Real Currency? An Economic Appraisal. National 
Bureau of Economic Research Working Paper Series, 4-7. 
49 
7 APPENDICES 
Appendix A: BTC/USD Exchange Rate 
1 Bitcoin equals 
8257.47 US Dollar 
1 Bitcoin 
825747 US Dollar 
Source: (XE Currence Converter, 20 17) 
Appendix B: Bitco in Price History Chart 
... 
... 
... 
... 
12000 
8000 
<1000 
0 
20 13 2014 2015 2016 2017 
Bitcoin Price History Chart 
Zoom 1m 3m 6m YTD 1y From Aug 16. 20 10 j To Nov 23, 2017 
10k 
~ 
0 Sk 
0 
"' :::) 
c 
~ 6k v 
c 
::l 
9 
~ 
£ 4k 
~ 
ci'i 
2k 
Ok 
2011 2012 2013 2014 201 5 2016 20 17 
50 
Appendix C: Totall\ umbcr ul'b itco ins in C ircu lation 
Bitcoins i n c ir-cu lati on 
The t o t'""' ' nun"tbe r o r btt co.n~ tho:~t h.:tve .:t lr e.:ady been n 11ne<J : 1n o th er \.vords. ll"le curr e nt suppty of b 1t Co1n S on th en~ 
1"7.500.000 
1 s .coo. ooo 
12.500.000 
1 O.OOO. CIOO 
- . s oc.oc.-o 
::! . :>oo.oc-o 
.::oo=- :20 1 0 ::: 0 1 1 2012 ::01 ?.. :::01 .. 20 1 s :;;::o 1 s: 
Source: (BlockChain, 20 17) 
Appendix C : Line Graph Plo ts of the Sample Variables aero s the Fo ur C rossectio ns. 
1. INFALTION RATE 
INFLATJON_RATE 
2 
12 7.2 
6.8 
10 ol.< 
ol.O 
a 5.6 
5.2 
6 <8 
« 
<_Q 
11 D1 ~~ 1 n 111 N 1 II Ill IV 1 11 111 N 11 m TV 1 II m IV 1 II m N 1 11 111 N 
2)1! 2J15 2)16 ~li 2)1! 2J15 2016 201i 
3 4 
2J 2.5 
Ill 2.0 
16 
L5 
1< 
LO 
12 
0.5 v 1•J a a.o 
6 . Q.,j 
D m IV I n 111 IV 1 n n1 IV 1 a m ~~ 11 m ~l I n 111 IV 1 n m IV 1 II 10 IV 
2)1! 2015 2)16 2J17 2JT! 2J15 2016 2J17 
51