University of Ghana http://ugspace.ug.edu.gh
UNIVERSITY OF GHANA 
COLLEGE OF HUMANITIES 
 
 
THE IMPACT OF OIL PRICE VOLATILITY ON EXCHANGE RATE 
VOLATILITY: A STUDY OF SOME OIL DEPENDENT ECONOMIES 
 
 
BY 
RICHARD AGYABENG DONKOR 
(ID. NO. 10598713) 
 
 
 
 
THIS THESIS IS SUBMITTED TO THE UNIVERSITY OF GHANA, LEGON IN 
PARTIAL FULFILMENT OF THE REQUIREMENT FOR THE AWARD OF MPHIL 
RISK MANAGEMENT AND INSURANCE DEGREE 
 
DEPARTMENT OF FINANCE 
 
JULY, 2018 
 
University of Ghana http://ugspace.ug.edu.gh
DECLARATION 
 
I hereby declare that; this study is my original work and has not been submitted by anyone 
for an academic award at the University of Ghana or any other university. 
I bear sole responsibility for any shortcomings. 
 
………………………………………    ………………………….. 
RICHARD AGYABENG DONKOR           DATE  
     (10598713) 
 
 
 
 
 
 
 
 
 
 
ii 
 
University of Ghana http://ugspace.ug.edu.gh
CERTIFICATION 
 
This is to certify that this thesis has been supervised with the laid down principles for thesis 
writing at the University. 
 
 
………………………………………..    ………………………… 
          DR. LORD MENSAH        DATE  
                (Supervisor) 
 
………………………………………..    ………………………….. 
DR. EMMANUEL SARPONG – KUMANKOMA    DATE 
             (Co – Supervisor)  
 
 
 
 
 
 
 
iii 
 
University of Ghana http://ugspace.ug.edu.gh
DEDICATION 
This work is dedicated to my supervisor, Dr. Lord Mensah and my patron, Mr. Mathias Tibu 
for their immense support, encouragement and assistance throughout the period of my 
studies.  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
iv 
 
University of Ghana http://ugspace.ug.edu.gh
ACKNOWLEDGEMENT 
I thank the Lord God Almighty for granting me the strength, wisdom, knowledge and 
understanding throughout the period of this study. 
I would like to express my deepest gratitude to my supervisors, Dr. Lord Mensah and Dr. 
Emmanuel Sarpong – Kumankoma for their guidance and constructive criticism which 
shaped and enriched the work.  
My sincere gratitude to my parents, Mr. Kwabena Donkor and Mrs. Helena Kwakyewaa 
Donkor, my beloved friends, Selina Owusu – Nyamekye and Raphael Kuranchie – Pong and 
my colleagues at both work and school for their encouraging words and advice during the 
period when the study was undeniably challenging.  
 
 
 
 
 
 
 
 
 
 
v 
 
University of Ghana http://ugspace.ug.edu.gh
TABLE OF CONTENT 
 
DECLARATION ......................................................................................................................................... ii 
CERTIFICATION ....................................................................................................................................... iii 
DEDICATION ........................................................................................................................................... iv 
ACKNOWLEDGEMENT ............................................................................................................................. v 
TABLE OF CONTENT ............................................................................................................................... vi 
LIST OF TABLES ....................................................................................................................................... ix 
LIST OF FIGURES ...................................................................................................................................... x 
LIST OF ABBREVIATIONS ........................................................................................................................ xi 
ABSTRACT .............................................................................................................................................. xii 
CHAPTER ONE ......................................................................................................................................... 1 
INTRODUCTION ....................................................................................................................................... 1 
1.1 Background of the Study ............................................................................................................... 1 
1.2 Research Problem ......................................................................................................................... 4 
1.3 Research Objectives ...................................................................................................................... 8 
1.4 Research Questions ...................................................................................................................... 8 
1.5 Scope of the Study ........................................................................................................................ 9 
1.6 Significance of the Study ............................................................................................................... 9 
1.7 Limitation of the Study................................................................................................................ 10 
1.8 Chapter Disposition ..................................................................................................................... 11 
1.9 Chapter Summary ....................................................................................................................... 11 
CHAPTER TWO ...................................................................................................................................... 12 
LITERATURE REVIEW ............................................................................................................................. 12 
2.1 Introduction ................................................................................................................................ 12 
2.2 Theoretical Review .......................................................................................................................... 12 
2.2.1 Determinants of Exchange Rate Volatility ............................................................................... 12 
2.2.2 Macroeconomic Variables and Oil Price Volatility ................................................................... 14 
2.2.3 Exchange Rates of Oil Dependent Economies and Oil Price .................................................... 16 
2.3 Empirical Review ............................................................................................................................. 18 
2.3.1 Oil Price Volatility – Exchange rate Volatility Nexus ................................................................ 18 
2.3.2 Oil Price – Exchange Rate Causality Pattern ............................................................................ 19 
2.4 Brief Overview of Selected Oil Dependent Economies ............................................................... 22 
vi 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER THREE .................................................................................................................................... 25 
METHODOLOGY .................................................................................................................................... 25 
3.1 Introduction ................................................................................................................................ 25 
3.2 Research Design .......................................................................................................................... 25 
3.3 Data Source ................................................................................................................................. 26 
3.4 Sample and Sample Period ......................................................................................................... 27 
3.5 Variables .......................................................................................................................................... 28 
3.5.1 Oil Price Volatility ..................................................................................................................... 28 
3.5.2 Exchange Rate Volatility ........................................................................................................... 29 
3.6 Empirical Methodology ................................................................................................................... 31 
3.6.1 The Vector Autoregressive Model ........................................................................................... 31 
3.6.2 Theoretical Framework for the Model..................................................................................... 32 
3.6.3 VAR Model Specification for the Study .................................................................................... 33 
    3.6.4 Likelihood Ratio test…………………………………………………………………………………………………………….32 
    3.6.5 Information Criterion…………………………………………………………………………………………………………..33 
3.7 Granger Causality Test ................................................................................................................ 36 
3.8 Data Analysis ............................................................................................................................... 37 
3.9 Chapter Summary ....................................................................................................................... 37 
CHAPTER FOUR ..................................................................................................................................... 39 
DATA ANALYSIS AND DISCUSSION ........................................................................................................ 39 
4.1 Introduction ................................................................................................................................ 39 
4.2 Descriptive Statistics ................................................................................................................... 39 
4.3 Characteristics of the Oil Price Returns and Exchange Rate Returns ......................................... 43 
4.4 Volatility Trend of Exchange Rate and Oil Price .......................................................................... 43 
4.5 Unit Roots Test Results ............................................................................................................... 46 
4.6 VAR Estimation and Test Results .................................................................................................... 47 
4.6.1 Optimal Lag Selection .............................................................................................................. 47 
4.6.2 The VAR Estimation Results ..................................................................................................... 49 
4.6.3 VAR Estimation Results for the Pre-Financial Crisis ................................................................. 49 
4.6.4 VAR Estimation Results for the Post – Financial Crisis ............................................................. 52 
4.7 Granger Causality Test Result ......................................................................................................... 56 
4.7.1 Granger – Causality for the Pre – Financial Crisis Period ......................................................... 57 
4.7.2 The Granger – Causality Results for Post-Crisis Period ............................................................ 58 
vii 
 
University of Ghana http://ugspace.ug.edu.gh
4.8 Chapter Summary………………………………………………………………………………………………………………………58 
CHAPTER FIVE ....................................................................................................................................... 63 
SUMMARY, CONCLUSION AND RECOMMENDATIONS ......................................................................... 63 
5.1 Introduction ................................................................................................................................ 63 
5.2 Summary ..................................................................................................................................... 63 
5.3 Conclusion ................................................................................................................................... 66 
5.4 Recommendations ...................................................................................................................... 66 
REFERENCES .......................................................................................................................................... 67 
APPENDICES .......................................................................................................................................... 79 
Appendix A: Pictorial Behaviour of the Returns for the Pre-and Post - Crisis Periods ..................... 79 
Appendix B: Pictorial Behaviour of the Volatilities for the Pre-and Post - Crisis Periods ................. 81 
Appendix C: Results of VAR Lag Selection for the Pre - Crisis Period ………………………………………….. 79 
Appendix D: Results of VAR Lag Selection for the Post-Crisis Period ............................................... 86 
 
 
 
 
viii 
 
University of Ghana http://ugspace.ug.edu.gh
LIST OF TABLES 
 
Table 4. 1: Descriptive Statistics of Oil Price and Bilateral Exchange Rate of the Dollar ...... 40 
Table 4. 2: Unit Roots Test Results ......................................................................................... 46 
Table 4. 3: VAR Estimation Results for Pre-Financial Crisis Period ...................................... 50 
Table 4. 4: VAR Estimation Results for Post-Financial Crisis Period .................................... 53 
Table 4. 5: Granger – Causality Results for the Pre – Crisis Period ........................................ 57 
Table 4. 6: Granger – Causality Results for the Post – Crisis Period ...................................... 59 
 Table A. 1VAR Lag Selection of Oil Volatility & Cedi/dollar Exchange Rate Volatility….83 
Table A. 2 VAR Lag Selection of Oil Volatility & Naira/dollar Exchange Rate Volatility ... 83 
Table A. 3VAR Lag Selection of Oil volatility & Rand/dollar Exchange Rate Volatility ...... 84 
Table A. 4: VAR Lag Selection of Oil Volatility & Euro/dollar Exchange Rate Volatility ... 84 
Table A. 5: VAR Lag Selection of Oil Volatility & Ruble/dollar Exchange Rate Volatility.. 85 
Table A. 6: VAR Lag Selection of Oil Volatility & Rupee/dollar Exchange Rate Volatility . 85 
Table A. 7: VAR Lag Selection of Oil Volatility & Cedi/dollar Exchange Rate Volatility.... 86 
Table A. 8: VAR Lag Selection of Oil Volatility & Naira/dollar Exchange Rate Volatility .. 86 
Table A. 9: VAR Lag Selection of Oil Volatility & Rand/dollar Exchange Rate Volatility ... 87 
Table A. 10: VAR Lag Selection of Oil Volatility & Euro/dollar Exchange Rate Volatility . 87 
Table A.11:VAR Lag Selection of Oil Volatility & Ruble/dollar Exchange Rate Volatility.. 88 
Table A.12: VAR Lag Selection of Oil Volatility & Rupee/dollar Exchange Rate Volatility 88 
 
 
 
 
 
 
ix 
 
University of Ghana http://ugspace.ug.edu.gh
LIST OF FIGURES 
Figure A. 1: Panel A: Pictorial Behaviour of the Returns for the Pre – Crisis Period ............. 79 
Figure A. 2: Panel B: Pictorial Behaviour of the Returns for the Post – Crisis Period ........... 80 
Figure A. 3: Panel A: Pictorial Behaviour of the Volatilities for the Pre – Crisis Period ....... 81 
Figure A. 4: Panel A: Pictorial Behaviour of the Volatilities for the Post – Crisis Period ...... 82 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
x 
 
University of Ghana http://ugspace.ug.edu.gh
LIST OF ABBREVIATIONS 
 
ADF  - Augmented Dickey Fuller  
AIC  - Akaike Information Criterion 
ARCH  - Autoregressive Conditional Heteroskedasticity 
EGARCH - Exponential Generalize Autoregressive Conditional Heteroskedasticity 
EURRET - Euro/dollar exchange rate return 
EURVOL - Euro/dollar Exchange Rate Volatility  
EXCRET - Exchange Rate Return 
GARCH - Generalized Autoregressive Conditional Heteroskedasticity 
GHSRET - Ghana Cedi/dollar Exchange Return 
GHSVOL - Ghana Cedi/dollar Exchange Rate Volatility  
LR  - Likelihood Ratio  
NAIRAVOL - Nigeria Naira/dollar Exchange Rate Volatility 
OILRET - Oil Price Return 
OILVOL - Oil Price Volatility 
OPEC  - Organisation of Petroleum Exporting Countries 
PP  - Philips – Perron  
RANDRET - South Africa Rand/dollar exchange rate return 
RANDVOL - South Africa Rand/dollar Exchange Rate Volatility 
RUBRET - Russia Ruble/dollar exchange rate returns  
RUBVOL - Russia Ruble/dollar Exchange Rate Volatility 
RUPRET  - India Rupee/dollar exchange rate returns  
RUPVOL - India Rupee/dollar Exchange Rate Volatility 
USD  - United States Dollar  
VAR  - Vector Autoregressive  
WTI  -  West Texas Intermediate  
xi 
 
University of Ghana http://ugspace.ug.edu.gh
ABSTRACT 
This research investigates the interrelation impact between oil price volatility and bilateral 
exchange rate volatility of selected oil – dependent countries and the causality pattern 
between them in the pre – and post 2008 -2009 global financial recession. Exchange rate 
volatility is for economies that are dependent on oil either for major industrial activities or for 
fiscal revenue. Thus, currencies exchange rates volatility examined in the study are for the 
Ghana cedi, Nigeria naira, South Africa rand, India rupee, Russia ruble; the euro and crude 
oil price is for West Texas Intermediate (WTI). Oil price volatility and exchange rate 
volatility are estimated using nonlinear models and interrelation impact between estimated 
volatilities as well as the causality pattern were done through linear models. 
Empirical findings revealed both unidirectional and bidirectional impact between oil price 
volatility and exchange rate volatility for major oil dependent countries (Russia ruble and 
Nigeria naira, Ghanaian cedi and Nigeria naira). This impact was particularly confirmed in 
the post crisis period, which was not too surprising because higher volatility in oil price and 
exchange rate was evident in that same period (see Appendix B, Panel B). However, Granger 
causality test confirmed this relationship particularly in the post crisis period.    
 
Keywords: Oil price volatility, Exchange rates volatility, Vector Autoregressive (VAR),  
                 GARCH, Exponential GARCH, Granger Causality, Pre-Financial Crisis, Post  
                Financial Crisis. 
 
 
xii 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER ONE 
INTRODUCTION 
 
1.1 Background of the Study 
Countries all around the world, both developed and developing, depend heavily on crude oil for 
industrial production of goods and services and also, for fiscal revenue. Crude oil price and its 
volatility which are usually caused by factors such as demand and supply forces, geographical 
disturbances among others are at core of most countries’ economic activities. Literature have 
empirically shown that oil price volatility is correlated with changes in macroeconomic variables 
such as Gross Domestic Product (GDP), exchange rates, unemployment, interest rates and 
inflations, (Shaari, Hussain & Rahim, 2012; Ling & Jones, 2011; Basher, Hang & Sadorsky, 
2012; Blanchard & Gali, 2007). An increase in crude oil prices contributes significantly to the 
development of economies through the transfer of wealth among oil dependent economies in a 
form of trade balance which leads to trade disequilibrium (Amano &Van Norden, 1998). The 
disequilibrium in trade results in high exchange rates volatility, especially for oil importing 
economies. Over the years, the United States dollar remains the most widely used currency for 
trading crude oil in the international market. In view of this, one will expect that fluctuations in 
the value of the US dollar will have a significant impact on crude oil prices and consequently, 
have an impact on economies that depend heavily on crude oil, especially for revenue generation. 
In other words, oil exporting economies whose currencies’ value are pegged to the value of the 
US dollar in a floating exchange rate system are affected by the volatility of the United States 
dollar. For such economies, their fiscal revenue from crude oil exports may be affected which 
may in turn affect the efficacy of their local currencies relative to the US dollar in a floating 
1 
 
University of Ghana http://ugspace.ug.edu.gh
exchange rate system, even with a stable supply of oil. For economies that are major crude oil 
importers whose local currencies’ values are pegged to the value of the dollar in floating 
exchange rate system, their cost of production fluctuates with the US dollar exchange rates 
volatility. In essence, the economic development vis-?̀?-vis currencies performance of oil 
dependent economies is exposed to the risks that are associated with the volatility of the United 
States dollar in a floating exchange rate system. 
Studies by Ding and Vo (2012), Reboredo (2012), Salisu and Mobolaji (2013) have documented 
that, oil price–exchange rates volatility link appears to be either weak or no relationship of such 
existed in the period preceding the start of the 2007 global financial crisis. However, they 
documented confirmation of such relationship during the 2007 global financial crisis period. 
Despite the attempts of these studies to probe the link between the volatilities of oil price and 
exchange rates, they utilized currency index which tends to neutralize the behavioural 
characteristics of each currency; such characteristics may include the extent of development of a 
particular economy or the degree of dependency on oil by these economies. The use of currency 
index is not able to depict the performance and the strength of each economy’s currency against 
the US dollar in a floating exchange rate system when external shock strikes. This study expands 
existing literature of the impact that exist between the volatilities of oil price and exchange rates 
by adopting the currency exchange rates of six oil dependent economies and analysing how the 
volatilities in these countries exchange rates respectively relates with oil price volatility before 
and after the 2007 global financial crisis period. The currencies exchange rates volatility to be 
examined against oil price volatility include the European union euro, Nigerian naira, Indian 
rupee, Russian ruble, Ghanaian cedi and South African rand. Dependency on oil by these 
2 
 
University of Ghana http://ugspace.ug.edu.gh
economies may be either because fiscal revenue generation is heavily dependent on oil exports or 
industrial activities are primarily dependent on crude oil imports. 
In addition, the causality nexus between oil price and exchange rate has been researched in 
literature. However, the nature (i.e. linear or nonlinear) and the causality direction documented in 
literature between these two markets are unconfirmed. Some studies empirically adopted the 
traditional Granger causality test which is centred on Vector Autoregressive (VAR) model and 
introduced by Granger (1969) and Sim (1972) to examine the linear causal relationship between 
these two markets. These studies, despite their different conclusions on the causality pattern 
between these two markets, documented evidence of linear causality between oil price and 
exchange rates (Amano & Van Norden, 1998; Tentatape, Jui-Chin & Yaya, 2015; Lizardo & 
Mollick, 2010). However, the findings of other empirical studies have criticized the Granger 
causality test as a technique that observed the ignorance of a common information factor (the 
volatility effect) which ends up in spurious conclusions. Moreover, they proposed the application 
of nonlinear techniques when using the Granger causality test that is inferred parametrically 
through autoregressive models, to examine the causal relationship between time series data. In 
this way, while the former captures the volatility effects in the series, the later establishes the 
causal relationship in the context of linear regression models of stochastic processes (Bell, Kay, 
& Malley, 1996; Asimakopoulos, David & Wan, 2000; Peguin–Feissolle, Strikholm & 
Terasvirta, 2008). In view of this, the study employs the Granger Causality test and the 
Generalised Autoregressive Conditional Heteroskedasticity (GARCH) proxies as linear and 
nonlinear techniques respectively in investigating the causality link between oil price volatility 
and exchange rates volatility. Although the Granger causality test has been criticized in 
literature, it is applied in this study because it is proven in literature to be pragmatic, robust and 
3 
 
University of Ghana http://ugspace.ug.edu.gh
standard tool to reveal causal relationships in the context of linear regression models of 
stochastic processes (Granger, 1980; Hu & Liang, 2014; Friston, Stephan, Montague & Dolan, 
2014). In addition, the application of GARCH proxies to generate the volatility series of oil price 
and exchange rates, will cushion the Granger Causality test against its defect to account for the 
volatility effect in the series. In other words, the GARCH proxies used as nonlinear technique in 
the study will neutralize the Granger causality test’s defect and provide a firm ground for its 
application. 
The complexity of oil price volatility in the international market and its continuing sensitive 
impact on macroeconomic variables such exchange rates in high oil dependent economies, 
makes its relevant for academic cross–examination. Hence, the study first employed the 
Generalized Autoregressive Conditional Heteroskedasticity (GARCH) proxies to generated the 
volatility series of oil price and the various currencies exchange rates. Secondly, the concept of 
Vector Autoregressive (VAR) model is employed to obtain the interrelation impact between oil 
price volatility and exchange rates volatility of the selected oil dependent economies. Lastly, the 
Granger Causality test is used to ascertain the linear impact interrelationship results. 
 
1.2 Research Problem 
In the middle of 2007, the United States financial market begun to slip into a disastrous moment 
of financial crisis since the Great recession of the early 1930’s (Tharkor, 2015). The moment of 
crisis in United States led to the collapse of many financial institutions including the Lehman 
Brothers. Indeed, literature have established that, after the ceased of the global economic 
depression in 2010, West Texas Intermediate (WTI) crude oil price started declining steadily, 
dipping as low as 28 US dollars per barrel in February 2016, during that same time, the value of 
4 
 
University of Ghana http://ugspace.ug.edu.gh
the United States dollar surged significantly contrary to most floating currencies including the 
cedi, euro, naira, rupee, rand and the ruble. Earlier in July 2008, crude oil price was traded at 
more than 145 US dollars per barrel (Reboredo, 2012; Mensah, Obi & Bopkin, 2017). Since the 
US dollar remains the most widely used currency for pricing crude oil, the contemporaneous rise 
of the US dollar and fall of oil prices in the post crisis period makes it prying into the 
contribution of oil price to the surged of the US dollar in the post crisis period. Hence, this study 
sought to unravel the contribution of oil price volatility to the significant surge of the US dollar 
against most floating currencies after the 2007 global crisis, with focus on the currencies listed 
above. 
In addition, majority of empirical literature that investigated exchange rate–oil price link mainly 
focused at price level, whether high or low, rather than volatility (Wirjanto & Yousefi, 2004; 
Zhang, Fan, Tsai & Wei, 2008; Groen & Pesenti, 2010). According to Clark (1973) and Ross 
(1989), asset volatility is associated with market interaction in terms of information flow. Thus, 
exchange rate–oil price link should appear not in price levels but in terms of volatility. This 
necessitated studies that probe into how oil price volatility and exchange rate volatility interact to 
provide better insight for traders and regulators in these two markets. In this way, the traders will 
not only consider to their expected returns from their business activities but also their business 
strategy’s vulnerability or exposure to risk in the period of severe volatility in the market. It is 
worth pointing out that some studies have investigated the link between oil price and exchange 
rates in terms of their volatilities in literature (Ding & Vo, 2012; Salisu & Mobolaji, 2013; Phan, 
Sharma, & Narayan, 2016). As mentioned earlier in this research, these studies focused on the 
approach of US dollar index which tends to neutralize the dominance of each oil dependent 
economy’s specific currency against the dollar in the foreign exchange market; such dominance 
5 
 
University of Ghana http://ugspace.ug.edu.gh
could be influenced by their respective domestic economic activities. Also, the currency index 
approach is not able to explain the response of each currency’s exchange rate volatility to oil 
price volatility shock in literature. This study sought to fill this gap which has been abandoned 
by these studies in literature by investigating the respective response of each currency’s 
exchange rate volatility against oil price volatility. This approach unravels answers to how 
volatility in each oil dependent economy’s exchange rate reacts to the volatility in oil price 
which might be of interest in terms of forecasting and investment for market participants in their 
respective oil dependent economies. 
Another important observation from the review of literature on exchange rate and oil price nexus 
is that, empirical cross–country studies in literature about the relationship between these two 
markets are skewed towards the link between developed oil dependent economy’s exchange rates 
and oil price, to the neglect of exchange rates of emerging oil dependent economies in Africa. 
These emerging economies include Ghana, Nigeria, South Africa among others. Studies such as 
Chen and Chen (2007), Lizardo and Mollick (2010), Benassy- Quere, Valerie and Alexis (2007) 
have focused only on developed oil dependent economies’ currencies in terms of investigating 
exchange rates–oil price nexus. Even the few that conducted cross–country analysis of the 
relationship between these two markets that included the currencies’ exchange rates of some 
developing oil dependent economies, neglected exchange rates of emerging oil dependent 
economies in Africa.  
Recently, Ding and Vo (2012), used the currencies of Canada, Mexico and Norway to depict 
exchange rates of oil producing economies; the currencies of Japan and the European union to 
depict developed oil importers and the currencies of India and Brazil depicting developing oil 
consumers. To the best of the researcher’s knowledge, only Mensah et al. (2017) included some 
6 
 
University of Ghana http://ugspace.ug.edu.gh
African countries’ currencies in their research of exchange rates–oil price relationship; however, 
their studies revealed the presence of long run equilibrium relationship existing between these 
two markets in the post crisis period for all oil dependent countries under their study. It is worth 
noting that, Mensah et al. (2017) did not focus on the second moment (variance) aspect between 
these two markets. There has not been much cross–country studies into the volatility interactions 
between these two markets which included emerging economies in Africa that are into the 
exportation and importation of crude oil and other petroleum products. 
 Africa is the home to five of the top 30 oil producing countries in the world. Sub–Saharan 
countries such as Nigeria, Angola among others are major oil producing economies in the world 
with estimated daily production of 5.8 million barrels. In 2013, they exported estimated daily 
output of 5.2 million barrels and imported an estimated 1 million barrels of oil products (Africa 
Energy Outlook report – IEA). Majority of the top oil producing economies in Africa are 
considered as some of the fastest growing economies in the world. However, the rate at which 
the currencies of these developing African countries fluctuate is very high coming from lack of 
proper diversifications, high importation of goods, high demand for foreign currency relative to 
local currency, among others. Crude oil fixated economies such as Ghana, Nigeria and South 
Africa have all witnessed strong currency volatility in recent times with the Ghanaian Cedi and 
the South African rand experiencing over 20% annualized volatility in 2015 (Ecobank Research 
Report). This laid a strong foundation for this study that targets these emerging oil dependent 
economies to determine if oil price volatility contributed to their currency’s volatility. 
 
 
 
7 
 
University of Ghana http://ugspace.ug.edu.gh
1.3 Research Objectives 
The foremost focus of this empirical study was to assess cross–country volatility impact of 
exchange rate–oil price nexus targeting currencies of some developed and emerging oil 
dependent countries in Africa, Europe and Asia. Specifically, this objective is achieved by 
addressing the following sub objectives: 
a. Examine the linear interrelation impact between oil price volatility and exchange rate 
volatility of major crude oil dependent economies. 
 
b. Examine the causality nexus between oil price volatility and exchange rate volatility of 
major oil dependent economies.  
 
1.4 Research Questions  
Adopting appropriate and universally accepted dynamic models that capture how past shocks in 
one variable explains the current variations in another variable in a system, this research sought 
to unravel the answers to the stated questions below and to provide intuitive explanations to them 
in the subsequent chapters: 
a.  Is there a significant linear interrelation impact between oil price volatility and exchange 
rate volatility of major oil dependent economies? 
  
b. Is there causality linking oil price volatility and exchange rate volatility of oil dependent 
countries? 
 
 
8 
 
University of Ghana http://ugspace.ug.edu.gh
1.5 Scope of the Study 
The study centred on determining the extent of interrelation impact linking the volatilities of oil 
price and exchange rate of six selected oil dependent economies. These countries whose 
exchange rates were being considered in this study included Russia, Nigeria, India, South Africa 
Ghana and the European Union. The relevance of using the exchange rates of the listed countries 
above lies in the fact that these countries are either oil-dependent or have their currencies 
exchange rate largely affected significantly by the volatilities that continue to hit the pricing of 
oil. The boundaries of the study could not be extended to include the currencies of other 
countries due to the relatively short time that the study would have to be completed. Hence the 
streamlining of the research to the currencies listed above.  
 
1.6 Significance of the Study 
The research was relevant to future research, practice and policy making. 
Research: This research was intended to serve as a source of literature for academia, although 
there exists a huge expanse of data on oil price volatility, little or no information is readily 
available on how these volatilities that are perceivable in valuing of oil tend to determine 
exchange rates of major currencies as well as the currencies of economies that depend largely on 
oil. The information that postulated from this study had therefore become a reliable tool of 
reference for academia, governments and other relevant players in the macro and micro 
economies of countries across the globe. To add to the above, the result of the study was also 
aimed at acting as a very fertile starting point for other researchers who will also nurture the 
desire to undertake studies in this area.  
9 
 
University of Ghana http://ugspace.ug.edu.gh
Practice: The findings of this study informed traders and investors on the route of causality 
between crude oil prices volatility and exchange rate volatility such that oil price volatility 
causes changes in exchange rate or exchange rate volatility causes fluctuation in oil prices. This 
should catch the awareness of investors, hedgers and financial managers. The reason is because it 
is essential to use the findings of this study to adjust their portfolio holdings since changes in 
currency value might have some effect on crude oil prices. For diversification, the study 
informed investors to rebalance their portfolio to include crude oil and other commodities such 
as tradable goods that are affected by exchange rate fluctuations in opposite route.  
Policy Making: Since the study was also aimed at unearthing the degree at which the volatility 
of exchange rates of currencies’  appear to spill over to affect the pricing of crude oil, the result 
of the study was useful for governments who depend on oil to put in place mechanisms that 
could mitigate against these fluctuations to stabilize their macro and micro economies.  
 
1.7 Research Limitation   
The researcher was unable to cover the exchange rate of all countries in Africa, Europe and Asia 
due to data unavailability and time constraint. In addition, the researcher was unable to look at 
effects of oil price on other macroeconomic variables such as interest rates, inflation among 
others as control variables, although literature supports the assertion that there is a link between 
these macroeconomic variables and oil price (Shaari et al., 2012; Ling & Jones, 2011; Basher, 
Hang & Sadorsky, 2012; Blanchard & Gali, 2007). This was due to the limited time in data 
gathering of all these countries that trade with the currency used in this study.  
 
 
10 
 
University of Ghana http://ugspace.ug.edu.gh
1.8 Chapter Disposition 
The entire research activity was arranged into five chapters. Each of these chapters had sub-
sections or units where applicable. The first chapter introduced the entire research work. The 
chapter also contained the background to the study. The objectives, purpose of the research, the 
relevance of the research engagement, scope of the research and the limitations of the research 
were all developed and espoused in this chapter. Chapter two of the study centered on doing an 
elaborative review of related literature. This was aimed at outlining the various contributions 
made by authors on the subject. The third chapter of the study focused on discussing the 
methodology employed for the study in the collection of data. Following suit to this chapter was 
chapter four which contains analysis of data and discussion. Final chapter of the study was 
devoted to the conclusion to the study as well as the recommendations made after the analysis of 
results. Following suit to this section or chapter were the appendices and bibliography. 
 
1.9 Chapter Summary  
The chapter summarized what this study sought out to do. That was to investigate the impact of 
interrelation of oil price volatility – exchange rate volatility relationship of six oil dependent 
economies in both Africa and Europe. Granger causality test to ascertain the route of causality 
between the second moments of these two markets was discussed. Specifically, introduction of 
the research and background, problem statement, objectives of the research and questions, 
significance and limitation of the research were also included in this chapter. It then concluded 
with the steps that this research follows. 
 
 
11 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER TWO 
LITERATURE REVIEW 
 
2.1 Introduction 
This chapter elaborates a brief discussion concerning the various concepts that determines 
exchange rate volatility, the link between oil prices changes and macroeconomic variables and 
the channel through which oil price volatility affect the exchange rate of oil-dependent 
economies. An empirical review in relation to these concepts and a brief overview of the 
countries whose exchange rates will be examined will be explored subsequently in the study. 
 
2.2 Theoretical Review  
This segment gives a brief in-depth into the theories, concepts, and models connecting oil price 
volatility, exchange rates volatility and the link between economies of oil-dependent countries 
and oil price. 
2.2.1 Determinants of Exchange Rate Volatility  
The volatility of the exchange rate is measured according to variations by considering all 
available information. In a flexible and free currency system, the exchange rate is the value of a 
local currency when traded for an international currency, however, in a constant currency 
system, the government influences the exchange rate system by trading the local currency in the 
international exchange market to keep the peg to another foreign currency (Mensah et al., 2017). 
An increase in trade balance as a result of an innovation that hit foreign price level of 
international tradable goods and nominal interest rate increases the volatility of exchange rate, 
12 
 
University of Ghana http://ugspace.ug.edu.gh
however, volatility in foreign prices decreases exchange rate volatility (Turnovsky & Bhandan, 
1982). 
The theory behind exchange rate volatility was first rigorously researched by Britton (1970) after 
the pioneering work of Laursen and Metzler (1950). Britton (1970) investigated if exchange rate 
volatility is correlated with anticipations by investors due to previous production and 
consumption decisions concerning exports and imports. Britton documented that there is several 
anticipation scheme’s evidence which are created out of necessity which investors are assured to 
follow and hence stabilize the exchange rate. Britton's theory was criticized that the various 
anticipation schemes evidence was anecdotal and was not based on any structural model and 
moreover, unpredictability was not added to their model (Driskill & McCarffety, 1980). Thus, 
Driskill and McCarfferty (1980) adopted Muth's (1961) model to examine an economy with a 
free-floating exchange system. They concluded that when anticipation sets in, it increases 
unpredictability and investors anxiety leads to increase in exchange rate volatility. Thus, increase 
in volatility of exchange rate may occur from the unforeseen risk that hits an economy. 
More recently, Grydaki and Fountas (2008) attempted to investigate the factors that influence 
volatilities of output and exchange rate. They created a model and assumed that price is affected 
by unexpected innovations, rational anticipation holds and capital mobility. Their findings 
conform to the findings of other studies in literature (Driskills & McCarfferty, 1980; Turnovsky 
& Bhandari, 1982; Betts & Devereux, 1996). They showed that uncertainty resulting from 
innovation is positively linked with volatility in exchange rates while at the same time, the 
existence of chosen innovations inflicts either positive or negative impact on volatility in the 
exchange rates. The inclusion of selected shocks impact on the volatility of exchange rate in their 
13 
 
University of Ghana http://ugspace.ug.edu.gh
results permits this study to search whether volatility in oil price as a shock, contribute positively 
or negatively to the variations in the nominal exchange rate of oil-dependent economies. 
 
2.2.2 Macroeconomic Variables and Oil Price Volatility 
Oil price, just like any other goods, has primarily been influenced by demand and supply 
disequilibrium (Cantore, Antimiani & Anciaes, 2012). If the supply of oil increases by major oil 
producing countries, the price will decrease and if supply decreases then price will go up, 
(Bacon, 1991; Kilian, 2009). For example, the oil production cut deal between Organization of 
Petroleum Exporting Countries (OPEC) members and non-OPEC members in 2016 to reduce the 
current oversupply that is keeping the oil price down in order to balance prices in the oil market. 
This led to an estimated daily reduction in the supply of 1.2 million barrels and 300,000 barrels 
by OPEC and non- OPEC members respectively. Moreover, recent fluctuations in oil prices have 
much to do with political disputes in the major oil–exporting countries in the Middle East. 
Current uprising conflict between Bahrain and Libya are cautioning stability within the region, 
which may spill over to nations like Saudi Arabia, the world’s biggest producer of oil (Hamilton, 
2003). These conflicts may affect oil production by closing plants or hindering the transport of 
crude oil in these major oil exporting countries.  
On the demand side, Killian (2009) examined the factors that dictate oil price volatility based on 
the findings of Barsky (2002) and Killian (2004). His framework identified three main ways 
through which oil prices are affected. First, an innovation to recent oil demand is propelled by 
fluctuation in international financial trading activities. Secondly, shocks to oil supply forces and 
finally, an innovation propelled by a shift in precautionary demand for oil (uncertainty). He 
concluded that an integration of oil–specific demand shock and aggregate demand shock have 
14 
 
University of Ghana http://ugspace.ug.edu.gh
been the ultimate drivers of crude oil price disturbance between 1975 and 2007. These findings 
by Killian (2009) were complimented by Dvir and Rogoff (2010). They provided a similar 
theory of precautionary demand channels by showing that inventory hoarding during the period 
of persistent volatility in economic growth increases oil price volatility. 
Literature has shown that the response of macroeconomic variables to the impact of oil price 
volatility can be symmetrical or asymmetrical. Studies that have argued the symmetrical 
relationship indicated that the size of negative impact on macroeconomic variables such as Gross 
National Product (GNP), Gross Domestic Product (GDP), exchange rates, inflation rates and 
interest rates is associated with the degree of oil price volatility (Mork, 1989; Majidi, 2006; 
Hooker, 1986; Hamilton, 1996; Godwin & Gisser, 1985; Mork, Olsen & Terje, 1994). Other 
studies documented the asymmetric effect theory of oil price volatility and economic activities as 
mainly influenced by monetary policy. The restrictive monetary policies implemented by the 
central banks of these oil-dependent economies contributed significantly to the fall of economic 
activities in the period which led to an oil price rise, (Bohi, 1989). This assertion was supported 
by other studies where they blamed some monetary policies by Federal Reserve solely for the 
economic downtown that affected the US (Barsky & Kilian, 2001; Bernanke, Gertler & Watson, 
1997). However, the findings of Brown and Yucel (1999) and Hamilton and Herrera (2001) were 
contrary to their conclusion. Their theory supported the view that counter–inflationary monetary 
policy by the central banks only contributed partially to the hike in oil price which affected the 
economy of the United States. 
However, a study by Lee and Ralti (1995) contended that the asymmetric results of oil price 
volatility and macroeconomic growth is influenced by sectorial reallocation and uncertainty that 
arise from new investment theory. Thus, the sectorial shift is seen from the point of view where a 
15 
 
University of Ghana http://ugspace.ug.edu.gh
decline in oil price leads to resource reallocation with vantage sectors that have a negative 
impact on macroeconomic growth. Thus, economic growth is highly adversely affected when 
negative oil shocks strike whereas the impact of positive oil shocks plays a limited role in 
economic development (Mahrara, 2008). Uncertainty effect arises when a new investment theory 
that hit the investment market deters investors from investing in order to promote economic 
growth. Under such a condition, the positive macroeconomic effect of a decline in oil price will 
be overshadowed by this uncertainty by the investors. 
 
2.2.3 Exchange Rates of Oil Dependent Economies and Oil Price  
Theoretically, the exchange rates of oil-dependent economies are affected by an innovation that 
hit international oil prices via two mediums: the terms of trade medium and wealth effect 
medium, Basher et al. (2016). The trade medium affects oil importing and oil exporting 
economies, though not in the same ways (Backus & Crucini, 2000; Amano & Norden, 1998; 
Cashin, Mohaddes & Raissi, 2014; Chen & Rogoff, 2003; Neavy & Corden, 1982). Oil 
consuming economies are faced with depreciating (appreciate) of trade balance when oil price 
increases (decreases) and subsequently deteriorate (appreciate) their domestic currency 
(Fratzscher, Schneider & Van Robays, 2014). Considering oil producing economies, an 
appreciation (depreciating) of oil price subsequently leads to an appreciating (depreciating) of 
exchange rates for these economies (Buetzer, Habib & Stracca, 2012). The wealth effect channel 
which was developed by Krugman (1983) and Golub (1983) was of view that, a rise in oil price 
increases the income of oil-exporting economies at the expense of oil-importing economies 
which results in downswing (upswing) of the exchange rate of major oil-dependent economies 
16 
 
University of Ghana http://ugspace.ug.edu.gh
via current account disparities and portfolio redistribution (Fratzscher et al., 2014; Buetzer et al., 
2012; Rasmussen & Roitman, 2011).  
Oil and petroleum products in the international market are priced based on US dollar value, so 
most oil-dependent economies peg the value of their local currency on the dollar value which 
exposes their local currencies value in the exchange rate market to the US dollar value volatility 
and volatility of oil prices (Coudert, Miguon & Penot, 2007). As oil price rises it dampens the 
desire to purchase goods and services by oil consumers in oil-importing economies and thus, 
depreciates the dollar (Yousefi & Wirjento, 2004). This is because as oil patronage by consumers 
fall as a result of high oil prices and since the dollar is a major trading currency for oil in the 
international market, it reduces demand for the dollar and hence, depreciates its value in the 
foreign exchange market. If the country’s economy is heavily reliant on imported crude oil, a 
slight increase in actual oil price in the international market may increase its domestic price in 
the home country by a higher proportion than the price in foreign countries and thus, cause a fall 
in the value of the home currency (Chen & Chen, 2007). Thus, a shift in prices of oil and 
petroleum products is negatively linked to the US dollar exchange rate (Cifarelli & Paladino 
2010). This argument is supported by studies such as Lizardo and Mollick (2010) and Akram 
(2009). For United States dollar peggers who are a net exporter of oil, depreciation in dollar 
value decreases their revenue from oil exports, although it somehow strengthens their currency 
against the dollar. This results in some OPEC members adjusting their oil prices to stabilize their 
purchasing power and market share of oil revenue (Yousefi & Wirjanto, 2004). The opposite is 
true for net oil importers whose currencies’ values are pegged against the dollar (Bloomberg & 
Harris, 1996). For such economies, they can increase their demand for oil and build up their 
current account (Mensah et al., 2017; Huntington, 1986; Brown & Philips, 1984). 
17 
 
University of Ghana http://ugspace.ug.edu.gh
2.3 Empirical Review   
This section gives an empirical review of the studies that have been done on the various theories, 
concepts and the exchange rate–oil price relationship and how the study adds to literature. 
2.3.1 Oil Price Volatility–Exchange rate Volatility Nexus  
Several studies have contributed to literature concerning the interaction between the second 
moments of oil price-exchange rate nexus where unidirectional and bidirectional volatility 
interaction was detected between these two markets. Cifarrelli and Paladino (2010) investigated 
whether speculation affects the oil prices volatility. They used a tri-variate Constant Correlation 
Coefficient (CCC) GARCH in mean model with a composite nonlinear conditional expected 
equation in which oil price volatility is linked with exchange rate dynamics and stock market 
behaviour. They concluded that shifts in oil price are contrarily related to exchange rate 
volatility. However, Gosh (2011) examined oil price-exchange rate nexus applying daily data 
and adopting GARCH and EGARCH structure. He concluded that, negative and positive shocks 
of oil price asymmetrically impact the fluctuations of the returns of Nigeria's nominal exchange 
rate. Thus, he suggested that the effect of an oil price shock on the volatility of the exchange rate 
is permanent. 
Moreover, Zhang, Fan, Tsai and Wei (2008) investigated price volatility interactions between the 
oil price market and the US dollar exchange rate market. They found evidence of clustering and 
volatility for the two market prices but interestingly, they did not find any volatility spill over 
between the two markets, proposing that volatility in the exchange rate have no significant effect 
on oil price volatility and the opposite is true. On the contrary, other empirical research has 
documented the presence of volatility spill over between these two markets. Ding and Vo (2012) 
investigated the reciprocal action of oil price and foreign exchange rate nexus to extricate 
18 
 
University of Ghana http://ugspace.ug.edu.gh
information entwined in the two markets for forecasting. Using multivariate GARCH and 
Multivariate Stochastic Volatility (MSV) models, they concluded that in times of crisis, there 
appears to be bidirectional volatility reciprocal between both markets. Thus, the impact of a 
shock in one market that increases volatility in that market tends to impact the other markets 
volatility. Moreover, Salisu and Mobolaji (2013) examined the transmission of return and 
volatility linking oil price and nominal US dollar and the Naira exchange rate. Using daily data, 
they established in their findings the existence of spill over effect emanating from the 
relationship between the oil market and exchange rate market in terms of volatility and returns. 
More recently, Jawadi, Louhichi, Idi-Cheffou and Ameur (2016) investigated oil price volatility 
via an analysis of crude oil markets and the US dollar/euro exchange rate market link. They 
identified that, volatility in US foreign exchange rate is transmitted to oil price, indicating that 
news arriving in the exchange rate market can influence prices of crude oil. These conclusions 
are in conformity with the findings of Razqallah and Smimou (2011) where they found 
significant interactions between these two markets during volatile periods. Intuitively, asset 
volatility is the degree of the flow of information covering related markets considering all 
available information (Touchen & Pitt, 1983; Ross, 1989; Clark, 1973). From the review of the 
literature, no paper has performed a comparative analysis of how the volatilities in these markets 
interact during the period when the market is stable (pre-crisis) and the period after the market 
has experienced downturn (post-crisis period). 
 
2.3.2 Oil Price – Exchange Rate Causality Pattern 
In literature, several studies have probed into the causal link between the return of exchange rate 
and oil price but their conclusions differ and hence there is no solid conclusion as to the direction 
19 
 
University of Ghana http://ugspace.ug.edu.gh
of causality between these two markets. On one hand, there are studies that have concluded that 
changes in the exchange rate may Granger cause oil price changes (Benassey–Quere et al., 2007; 
Sardosky, 2000; Huang & Guo, 2007; Zhang & Wei, 2010). On the other hand, there are studies 
that have documented that causality route from oil price to exchange rate (Amano & Norden, 
1998; Lizardo & Mollick, 2010; Benhmad, 2012; Chaudhuri & Daniel, 1998; Chen & Chen, 
2007).  
Amano and Norden (1998) pioneered the causal tie-in between these two markets. Using the 
Error-Correction Model (ECM), they concluded that oil price causes exchange rates movement 
but not in the opposite direction. Chaudhuri and Daniels (1998) document a similar conclusion; 
they first identified a long-run balance relationship linking these two markets and concluded that 
price changes of crude oil cause changes in the exchange rate of 16 OECD economies. In 
consistence with the above findings in the literature, Lizardo and Mollick (2010) used more 
extensive data spanning from 1970 to 2008 to investigate this relationship between these two 
markets and found that oil price Granger causes volatility in the value of the US dollar relative to 
major trading currencies. Lastly, Chen and Chen (2007) also examined the causal relationship 
between these two markets using a different measure of oil prices including the West Texas 
Intermediate (WTI), Brent, World oil prices and the United Arab Emirates prices of crude oil 
against G7 economies exchange rates. They concluded that these oil prices cause changes in the 
exchange rate of the G7 countries. 
On the contrary, other empirical studies concluded in the opposite direction, indicating that 
changes in exchange rate Granger causes oil prices volatility. Zhang et al. (2008) documented a 
significant co-integration relationship between oil price and exchange rate and found return 
causality from dollar exchange rate to oil price but not in the opposite direction. An Early work 
20 
 
University of Ghana http://ugspace.ug.edu.gh
by Sadorsky (2000) examined empirically the relation connecting future prices of crude oil and 
exchange rate dynamics. He concluded that the exchange rate causes exogenous shock to oil 
prices. Also, Indjehagopian, Lantz, and Simon (2000) used a limited version of Vector 
Autoregression (VAR) model, described in error correction structure to show a short – and long 
run correlation linking exchange rates and oil prices. They identified causality from exchange 
rate variations to oil price variations. In addition, Yousefi and Wirjanto (2004) examined the 
value of the US dollar in the foreign exchange market against crude oil prices of OPEC 
members. They adopted the Generalized Method of Moments and concluded that a change in oil 
prices of OPEC members is due to changes in the value of the US dollar in the foreign exchange 
market. That is, as the value of the dollar changes in the international currency exchange market, 
OPEC members adjust their oil prices accordingly which leads to the conclusion of dollar 
exchange rate causing changes in the prices of OPEC oil. 
Another important observation from literature is that the Granger causality findings of other 
studies revealed some bidirectional relationship between these two markets. Empirical studies 
showed exchange rate Granger causes oil price and the opposite is true in the short and long run 
respectively (Tentatape, Huauq & Sissoko, 2014; Benassy – Quere et al., 2007; Uddin, Tiwari, 
Arouri & Teulou, 2013) 
The above findings from literature emphasized on a linear relationship using Granger Causality 
test. However, some authors in literature have argued that the standard Granger Causality 
assessment of linear relationship is inefficient in revealing exact nonlinear causal relationship 
and therefore, advocated the addition of nonlinear modelling technique of these two markets 
when investigating the existence of causality between these two markets (Baek & Brock, 1992; 
Wang & Wu, 2012; Diks & Panchenko, 2005; Bell, Kay & Malley, 1996;  Li, 2006; Penguin – 
21 
 
University of Ghana http://ugspace.ug.edu.gh
Feisssolle, Strikholm & Terasvirta, 1999; Skalin & Terasvirta 1999; Behnmad, 2012). Rath and 
Bal (2015) investigated nonlinear causality linking exchange rate and crude oil prices, adopting 
Hiemstra and Jones (1994) approach of nonlinear Granger Causality to test the residuals of 
Vector Autoregression (VAR). They identified a significant bidirectional Granger Causality in 
nonlinear form between exchange rates and oil price. 
Based on the argument documented in the literature above concerning nonlinear technique and 
linear Granger causality in testing the relationship between these two markets, the study 
contributes to existing literature by investigating the causality pattern between these two markets 
by adopting a combination of nonlinear modelling technique and the standard Granger causality 
tests. Generalized Autoregression Conditional Heteroscedasticity (GARCH) proxies would be 
adapted to capture the nonlinear dynamics aspects of oil prices and exchange rate respectively. 
 
2.4 Brief Overview of Selected Oil Dependent Economies 
There is an extent of literature on the relevance of oil to the individual economies observed under 
this study. These economies which include Russia, Nigeria, the European Union, India, South 
Africa and Ghana are known to be part of major crude oil producers and consumers in the 
international oil market. Among these economies, Russia and Nigeria are observed as the two 
major oil exporters in Europe and Africa respectively. Although not part of Organization of 
Petroleum Exporting Countries (OPEC), Russia is regarded as the next economy with the biggest 
crude oil reserve in global oil market after Saudi Arabia, a member of Organisation of Petroleum 
Exporting Countries. Russia's estimated output of oil is over 11 million barrels per day which 
makes it a major dominant of the oil exporting industry. As a major reliant on oil export for 
revenue, Russia's economy is sensitive to fluctuations in prices of crude oil in the global market 
22 
 
University of Ghana http://ugspace.ug.edu.gh
(Beck, Demirguc-Kunt, & Levine, 2007; Rautava, 2004; Ito, 2010; Cukrowski, 2004; Tuzova & 
Qayum, 2016). Nigeria, on the other hand, is a member of OPEC and it's one of the major 
producers of oil in Africa and the global economy at large with daily production of 
approximately 2.5 million barrels. The country temporarily lost the top spot as the largest oil 
producer in Africa to Angola because of conflicts in the Niger Delta, a major oil producing 
sector in the Nigerian economy. Nigeria oil holds over 90 percent of its international trade 
earnings and over 70 percent of national budgetary revenue (Salisu & Mobolaji, 2013; Mensah et 
al., 2017). 
The Indian economy and the European union are known to be the world’s major importers of oil 
after US and China. Per the Integrated Energy Planning Report of Indian, over 70 percent of 
India’s demand for crude oil is met via the importation of oil (Bal & Rath, 2015). Unlike the 
European Union which comprises of developed economies, India’s economy is less developed 
with GDP per capita of one sixth of the European Union (Mensah et al., 2017).  
 
Finally, the currency exchange rates of Ghana and South Africa will also be examined against 
the dollar because these two economies are the fastest growing economies in the Sub – Sahara 
African. These two economies are rich in natural resources such as Gold, Cocoa, Bauxite, 
Diamond, Timber, Oil, among others and although they produce oil which contributes 
significantly to their economic growth, they are unable to produce and refine enough oil to meet 
the satisfaction of their local oil consumers due to limited oil fields. Thus, they import more oil 
into their economies. It is worth stating that, recently Ghana’s oil played a major role in the 
recent rise in the country’s Gross Domestic Product (GDP) which stood at 9.3 percent in the first 
23 
 
University of Ghana http://ugspace.ug.edu.gh
quarter of 2018 (Ghana Statistical Service, 2017). This lays a strong foundation for research 
targeting these economies to determine global oil price contributions to the strength of their 
respective currency against the dollar. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
24 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER THREE 
METHODOLOGY 
 
3.1 Introduction 
The previous chapter gave a thorough discussion of existing theories and the literature 
concerning oil price and exchange rates. This assisted the researcher in the choice of 
methodology for the study. The central point of this chapter is the discussion of research 
methodology employed for this study. This chapter presents the research design, variables used 
in the study, data sources and empirical methodologies employed for this study. 
The chapter discusses the research design adopted by the researcher and provides information 
about the data analysis plan. This chapter also explains how the volatilities of the variables of 
interest are generated and explains the causal relationships that exist between the variables’ 
volatilities. 
 
 
3.2 Research Design 
The research employs quantitative techniques by adopting the statistical approach and 
universally accepted scientific method to probe the relation linking the variables in the period 
preceding the start of the 2007 global financial recession and after the recession. In addition, the 
data used were purely quantitative which justifies the appropriateness of quantitative approach to 
this study.  A preliminary estimation of the time series data revealed some non-stationarity and 
serial correlation in the residuals, as a result, the raw oil prices and exchange rates series were 
converted into returns by taking the logarithm, estimating the first difference and multiplying by 
one hundred to obtain the returns of the variables. To determine the nature of oil price volatility, 
Generalized Autoregression Conditional Heteroskedasticity (1, 1) was adopted to generate the 
25 
 
University of Ghana http://ugspace.ug.edu.gh
conditional volatility series of oil price returns and Exponential Generalized Autoregression 
Conditional Heteroskedasticity (1, 1) was also adapted to generate the conditional volatility 
series of the bilateral exchange rate of US dollar and each specific currency. The estimated 
conditional volatility series of the returns of the variables were introduced into the Vector 
Autoregressive (VAR) model to examine the interrelation impact between the volatilities of the 
exchange rates and oil price. Finally, the Granger Causality test was adopted to examine the 
causality pattern that exist between the volatilities of oil price and the respective exchange rates 
of each economy to ascertain the impact relationship between the volatilities in these two 
markets. 
 
3.3 Data Source   
For the research objectives to be addressed, data on bilateral exchange rates and oil price were   
secondarily sourced. Specifically, weekly time series data for one month’s crude oil future 
contracts and dollar exchange rates were obtained for two subsample periods (prior and post 
2007 global financial crisis). The subsample period from January 2000 to December 2007 
represents the pre–crisis period. Also, the subsample period from January 2010 to December 
2016 represents the post– crisis period.  
Crude oil future prices were obtained from West Texas Intermediate (WTI) trading on New York 
Mercantile Exchange (NYME). The settlement data for different maturities are retrieved from 
US Energy Information Administration. Exchange rates are calculated by units of each currency 
per dollar. Nominal exchange rates are for Ghanaian Cedi, Indian rupee, South African rand, 
Russian ruble, Nigerian naira and the Euro. Nominal exchange rates are obtained from Oanda 
and they are calculated as units of each currency per dollar. 
 
26 
 
University of Ghana http://ugspace.ug.edu.gh
3.4 Sample and Sample Period  
The sample selected as discussed in the previous chapter was primarily informed by the degree 
of dependence on crude oil. Unlike other studies that focused only on one specific economy's 
exchange rate in the oil industry (Salisu & Mobolaji, 2013; Gosh, 2011; Yousefi & Wirjanto, 
2003; Tuzova & Qayum, 2016; Jawadi et al., 2016), this research focused on the exchange rates 
of multiple major oil exporting economies that are members of Organization of Petroleum 
Exporting Countries (OPEC), non-Organization of Petroleum Exporting Countries (non-OPEC), 
major importers of oil and neutral economies whose contribution in the oil industry is quite 
insignificant. Specifically, the Russian ruble and the Nigerian naira represent the exchange rates 
of developed and developing non–OPEC and OPEC members respectively. The India rupee and 
the euro respectively represent currencies exchange rates of developing and developed members 
of major oil importers in the world while the Ghana cedi and the South Africa rand was 
considered as neutral currency's exchange rates of economies that are neither major oil importers 
nor oil exporters in the oil industry. 
 
The sample period for this study is from the year 2000 to the year 2016. However, this sample 
period was divided into a sub-sample period (2000 to 2007) representing the period preceding 
the beginning of the 2007 global crisis and another sub-sample period (2010 to 2016) 
representing the period after the crisis. The choice of subsample periods shows the break in 
market valuation due to the housing crisis which resulted in 2007 global financial recession in 
the United States. This division was also necessitated by Mensah et al. (2017) and some other 
researchers’ arguments that the impact of the recent global financial recession on economies 
around the world was much significant and thus, this time periods should be taken into 
consideration when studying financial variables spanning over these periods (2000 to 2016).  
27 
 
University of Ghana http://ugspace.ug.edu.gh
3.5 Variables 
The basic aim of the research is to examine the impact and Granger causality existing between 
the exchange rates of some selected oil dependent economies and oil prices. The selection of the 
variables was influenced by previous studies on the exchange rates – oil price nexus reported in 
the literature. Moreover, these variables were transformed into their returns form for the analysis 
and this is adopted because the variables were not stationary in their raw form.  
 
 
3.5.1 Oil Price Volatility 
In accordance to Lee and Ralti (1995), employing the volatility of oil price provides more robust 
results than employing its returns when examining the effect of oil price variable on 
macroeconomic variables. For this reason, this study employed the volatility of oil price which 
was generated from the returns of oil price. In other words, oil price return was used to estimate 
the oil price volatility series. The oil price return was obtained from the raw oil prices using 
continue compounded returns. The continuously compounded return is applied because of time-
addictive and immaterial of the repetition of compounding (Brooks, 2008). Weekly oil price 
returns were derived by adopting continuously compounded return formulae:  
𝑷𝒕
                                                      𝑶𝑰𝑳𝑹𝑬𝑻𝒊,𝒕 = 𝒍𝒏 ( ) ∗ 𝟏𝟎𝟎                                                     (𝟑. 𝟏) 𝑷𝒕−𝟏
Where 𝑖 = Ghana, Nigeria, Russia, South Africa, India and the Euro, 𝑃𝑡 represents the closing oil 
price at week𝑡, 𝑃𝑡−1 indicates the closing oil price at lag one and 𝑙𝑛 represents natural logarithm. 
 
To obtain oil price volatility from the oil price returns, Bollerslev’s (1986) Generalized ARCH 
(1, 1) was adopted to estimate the volatility of oil price. According to Brooks (2008), GARCH is 
more parsimonious, avoids overfitting and it’s sufficient to capture the clustering in the oil price 
28 
 
University of Ghana http://ugspace.ug.edu.gh
series. As an extension of ARCH, the GARCH is more popular in literature in terms modeling 
second moments of financial time series data. Following Lee and Ralti (1995), the GARCH (1, 
1) model adopted for this study is as follows: 
                                𝑶𝑰𝑳𝑹𝑬𝑻𝒊,𝒕 =  𝜶 +  𝜷𝑶𝑰𝑳𝑹𝑬𝑻𝒊,𝒕−𝟏 + 𝜺𝒊,𝒕                                                           (𝟑. 𝟐) 
𝜺𝒊,𝒕~𝑵(𝟎, 𝜹
𝟐
𝒕)                                                     
                                  𝜹𝟐𝒊,𝒕 = 𝜽𝟎 + 𝜽 𝜺
𝟐
𝟏 𝒊,𝒕−𝟏 +  𝜽𝟐𝜹
𝟐
𝒊,𝒕−𝟏                                                                   (𝟑. 𝟑) 
Where 𝑖 = Ghana, Nigeria, South Africa, Europe, Russia and India, 𝑂𝐼𝐿𝑅𝐸𝑇𝑡 and 𝑂𝐼𝐿𝑅𝐸𝑇𝑡−1 
represent the end of the current week and the previous week’s oil price returns 
respectively, 𝛽, 𝜃0, and 𝜃1 are coefficients to be determined and 𝜀𝑡 is identically independent 
distributed error term. 𝜀𝑡 is also distributed normally with expectation zero and time-dependent 
variance. 𝛿2𝑡 represents the conditional variance of oil price return which depends on the lags of 
the error term and its own previous values. 𝛼 is constant and represents the intercepts to basically 
avoid forcing the regression line to pass through the origin. 
 
 
3.5.2 Exchange Rate Volatility 
The second variable considered for the study is the volatility of exchange rates. Exchange rates 
volatility were considered for this study due to theoretical and empirical literature demonstrating 
its relationship with oil price. To obtain the volatility of exchange rates, the obtained returns of 
exchange rate were used to estimate the exchange rate volatility because of non-stationarity of 
the nominal exchange rate (see Appendix A and its corresponding explanation on page 39). The 
returns of the exchange rate were derived by employing returns continuously compounded 
formulae given as: 
29 
 
University of Ghana http://ugspace.ug.edu.gh
𝑲𝒕
                                            𝑬𝑿𝑪𝑹𝑬𝑻𝒊,𝒕 = 𝒍𝒏 ( ) ∗ 𝟏𝟎𝟎                                                               (𝟑. 𝟒) 𝑲𝒕−𝟏
where 𝐸𝑋𝐶𝑅𝐸𝑇𝑡 is the returns of the nominal exchange rate; 𝑖 = Ghana, Nigeria, South Africa, 
Europe, Russia, and India; where 𝐾𝑡 represents closing nominal exchange rate at week 𝑡, 𝐾𝑡−1 
represents nominal exchange rate at lag one and 𝑙𝑛 represent natural logarithm. 
 
 Exchange rate volatility was estimated by using Exponential GARCH (EGARCH) introduced by 
Nelson (1995). The EGARCH is an improvement over the GARCH model. According to Brooks 
(2008), the EGARCH imposes non-negativity constraints on the parameters as in the case of the 
GARCH. Also, with the leverage effect parameter, asymmetries are allowed for in this modeling 
technique. Leverage effect is the negative relationship between returns and volatility which are 
driven by opposite forces (Kristoufek, 2014). Engle and Bollerslev's GARCH is not able to 
account for these asymmetric impacts. However, Nelson's EGARCH accounts for the 
asymmetric impacts. Hence, the applications of EGARCH (1, 1) in estimating exchange rate 
volatility. The average returns of exchange rate are modeled as follows: 
 
                            𝑬𝑿𝑪𝑹𝑬𝑻𝒊,𝒕 =  𝜷𝟎 +  𝜷𝟏𝑬𝑿𝑪𝑹𝑬𝑻𝒊,𝒕−𝟏 +  𝜺𝒊,𝒕                                                                         (𝟑. 𝟓) 
 
𝜺
𝟐 𝟐 𝒊,𝒕−𝟏
𝜺𝒊,𝒕−𝟏
                    𝒍𝒐𝒈 (𝝈𝒊,𝒕) =  𝝎 + 𝜽 𝐥𝐨𝐠  ( 𝝈𝒊,𝒕−𝟏) +  𝜶 | | +  𝜸                                        (𝟑. 𝟔) 𝝈𝒊,𝒕−𝟏 𝝈𝒊,𝒕−𝟏
where equation 3.5 and 3.6 are the mean equation and conditional variance respectively; 𝑖 = 
Ghana, Nigeria, South Africa, Europe, Russia, and India; 𝐸𝑋𝐶𝑅𝐸𝑇𝑡 and 𝐸𝑋𝐶𝑅𝐸𝑇𝑡−1 represent 
the end of the current week and the previous week’s exchange rate returns respectively; 𝛼, 
𝛽0,𝛽1, 𝜔, 𝜃 and 𝛾 are coefficients to be determined and 𝜀𝑡 is the error term. 𝒍𝒐𝒈 (𝝈
𝟐
𝒊,𝒕) represents 
logarithmic of the conditional variance; which explains the exponential nature of the leverage 
30 
 
University of Ghana http://ugspace.ug.edu.gh
effect, relative to quadratic, so that the predictions of the conditional variance are assured to be 
positive to avoid negativity constraint; The existence of leverage effects can be examined by the 
hypothesis that 𝛾 < 0; The impact is asymmetric if 𝛾 ≠ 0. 
 
 
3.6 Empirical Methodology  
In this study, the exchange rates–oil price volatility nexus is tested by adopting a Vector 
Autoregression (VAR) model and its associated granger-causality.  
 
3.6.1 The Vector Autoregressive Model 
The Vector Autoregression (VAR) is adopted to forecast reciprocally connected systems of time 
series data and for probing the dynamic impact of random interruptions on the system of 
variables. Thus, the VAR model is a system regression model constructed to investigate the lead-
lag relationships among variables and Sims (1980) popularized this model in the econometrics 
literature as the univariate autoregressive model generalization (Brooks, 2008). Typically, a 
VAR is used to estimate several equations simultaneously without indicating which of these 
variables are endogenous and exogenous. The Vector Autoregressive procedure avoids the need 
for structural modeling by considering every endogenous variable in the system as a function of 
the lagged values of all the endogenous variables in the VAR system. In general, the VAR model 
is expressed as:  
𝑲 𝑳
                               𝒀𝒕 = 𝜶 + ∑ 𝑨𝒌𝒀𝒕−𝒌 + ∑ 𝑩𝒍𝑿𝒕−𝒍 + 𝒆𝒕                                                                                     (𝟑. 𝟕) 
𝒌=𝟏 𝒍=𝒐
 
31 
 
University of Ghana http://ugspace.ug.edu.gh
Where 𝑌𝑡 represents 𝑎𝑛 𝑛 × 1 vector of period 𝑡 endogenous variables. 𝑋𝑡 represents control 
variables observations vector in time 𝑡 and 𝑒𝑡 represents an 𝑛 × 1 vector of residuals. The 
regression parameters, 𝐴𝑘 is a 1×m vector which determines the time series nexus linking 
endogenous variables and its lags, 𝐵𝑙 is a 1×m vector which estimates the nexus between 
endogenous variables and the control variables, where 𝑘 represents the number of lagged 
endogenous variables and 𝑙 is the number of lagged exogenous variables. 𝐾 is the last or the nth 
lag observation of the endogenous variables considered and 𝐿 is the last lag or nth lag 
observation of the exogenous variables considered. 
 
3.6.2 Theoretical Framework for the Model  
 The main model adopted for the study is the Vector Autoregressive (VAR) model. VARs are the 
fundamental models which are also known as linear multivariate time series model created to 
capture the reciprocal dynamics of multiple time series variables. At a first glance, VARs seem 
to be simple and uncomplicated generalizations of univariate autoregressive models into a 
multivariate autoregressive model. At another glance, VARs appear to be the most widely used 
and key empirical tool used in contemporary economics, Del Negro and Schorfheide (2011). 
This is because they are simple and flexible alternatives to the traditional multiple equation 
models. Recent studies have applied VAR model to argue exchange rates’ impact on oil price 
(Akram, 2009; Zhang et al, 2008) whereas other empirical studies adopted the same VAR and 
documented oil price’s significant impact on the exchange rate (Huang & Guo, 2007; Tentatape, 
Huauq & Sissoko, 2014). 
 
The beginnings of VARs are closely linked to the work of Christopher Sims (1990). Sims at the 
time in 1980 criticized the large-scale macroeconomics models because of the strong restrictions 
32 
 
University of Ghana http://ugspace.ug.edu.gh
they imposed. These models make a strong presumption concerning the dynamic behaviour of 
the correlation linking macroeconomic variables and they are also largely inconsistent with the 
notion that economic agent prefers present choice to future utility into account. This was 
described as Sims critic, that is, in a world of rational progressive agents, no variable can be 
regarded as exogenous. Sims recommended VARs as alternatives that allows one to model 
macroeconomic data informatively without imposing very strong restrictions. 
 
The study’s adoption of VARs is motivated by the fact that they are deemed as models that often 
provide better forecast relative to the forecast from univariate time series models and sometimes 
complicated theory based on simultaneous equations models. Secondly, it was also motivated by 
their ability to analyse the source of volatilities in business pattern and can produce the standard 
against which the dynamics of modern macroeconomic theories can be assessed. Thus, VARs 
can distinguish between competing theoretical models. 
 
3.6.3 VAR Model Specification for the Study 
It is worth pointing out that Vector Autoregressive model and its associated Granger causality 
test is employed to address research objective one, which is investigating the volatilities’ impact 
of oil price and exchange rates.  
Based on the general VAR model (equation 3.7), a sort of the general model  which consist of 
two endogenous variables, oil price volatility, and exchange rate volatility is constructed as: 
𝑲
𝜹𝟐 𝟐𝒊,𝒕 𝜶𝟏 𝜹 𝒊,𝒕−𝒌 𝜺(𝜹𝟐)𝒊,𝒕
                        [ 𝟐 ]  =  [𝜶 ]  +  ∑ 𝑨𝒌 [ 𝟐 ] +  [𝜺 ]                                                      (𝟑. 𝟖) 𝝈 𝒊,𝒕 𝟐 𝝈 𝒊,𝒕−𝒌 (𝝈
𝟐)𝒊,𝒕
𝒌=𝟏
33 
 
University of Ghana http://ugspace.ug.edu.gh
Where 𝛿2𝑡 and 𝜎
2
𝑡 are the oil price volatility and exchange rates  volatility at week t; 𝑖 = Ghana, 
Nigeria, South Africa, Europe, Russia, and India; the regression coefficients 𝐴𝑘 estimate the time 
series that links exchange rate - oil price volatilities. 
 
Deciding on an appropriate lag order for autocorrelation functions is one of the crucial decisions 
to make to ensure a valid estimation result. However, financial theory and for that matter 
overconfidence bias theories do not specify the suitable lag length that a researcher should adopt 
for a VAR model and its associated dynamic analysis. The model used in this study is a VAR 
system and thus, an optimal lag is necessary for valid estimation results. Therefore, to select the 
VAR model’s optimal lag length, two main approaches are employed which includes 
information criteria approach and Likelihood Ratio (LR) test approach. 
 
3.6.4 Likelihood Ratio Test 
In order to employ the LR test to determine the optimal lag structure, a sequential modification is 
performed on the LR. Beginning from the possible highest lag, a jointly zero hypothesis on the 
estimated coefficients of equation (3.9) is conducted: 
 
                                  𝑳𝑹 = (𝑻 − 𝒎)[𝑳𝒐𝒈|∑?̂?| − 𝒍𝒐𝒈|∑?̂?|]                                                               (𝟑. 𝟗) 
 
where |∑?̂?| refers to the determinant of the variance-covariance matrix resulting from the 
residuals of the restricted VAR model (say lag 4) estimation. Also, |∑?̂?| refers to the variance-
covariance matrix determinant resulting from the unrestricted VAR model (say lag 12) 
estimation. The 𝑇 refers to the sample size whiles the 𝑚 refers to the number of parameters per 
34 
 
University of Ghana http://ugspace.ug.edu.gh
equation. (𝑇 − 𝑚) in Equation (3.9) is Sims (1980) modification for small sample size. The test 
statistics has 𝜒2 distribution with the number of restrictions as the degrees of freedom. To 
achieve the optimal lag, this test “compare the modified LR statistics to the 5% critical values 
starting from the maximum lag, and decreasing the lag one at time until the first rejection is 
obtained. The alternative lag order from the first rejected test is selected as optimal lag” (Eviews 
10, User’s Guide). 
 
3.6.5 Information Criterion 
Another way to choose the optimal lag is to select the model order that minimizes information 
criterion value. The information criterion is performed by fitting VAR (p) models with different 
lag length on the data and the lag length (p) that provides the minimum value of the information 
criterion is selected.  The models used for the information criterion selection of optimal lags 
include Akaike Information Criterion (AIC), Schwarz-Bayesian (SC) and Hannan-Quinn 
Criterion (HQ). 
 
𝟐𝓵 𝟐𝒌
𝐴𝑰𝑪 = − +                                                                       
𝑻 𝑻
𝟐𝓵 𝒌
𝑺𝑪 = − + (𝐥𝐧 𝑻)                                                                  (𝟑. 𝟏𝟎) 
𝑻 𝑻
𝟐𝓵 𝟐𝒌
𝑯𝑸 = − + 𝐥𝐧 (𝐥𝐧 𝑻) 
𝑻 𝑻
 
𝑇
where ℓ = − (1 + ln(2𝜋) + ln(?̂?′?̂?/𝑇)), 𝑘 refers to the number of estimated parameters, 𝑇 is 
2
the sample size. According to Brooks (2008), SC embodies much stiffer penalty term than AIC 
while HC is somewhere in between. 
 
35 
 
University of Ghana http://ugspace.ug.edu.gh
3.7 Granger Causality Test 
The Granger causality approach is adopted to address the second objective of the study. That is, 
examining the Granger causality link between exchange rates volatility and oil price volatility. In 
this study, a Granger causality test based on VAR model estimation is conducted to 
contemporaneously detect the causality of oil price–exchange rates volatilities relationship and 
validate the results of the VAR model. The adoption of the Granger causality is motivated by 
empirical studies in literature (Amano & van Norden, 1998; Lizardo & Mollick, 2010; Chen & 
Chen, 2007; Sardosky, 2000; Huang & Guo, 2007). 
 
Granger (1969) proposed a method to answer the question of whether a series say 𝑥𝑡 causes 
another series say 𝑦𝑡 or if 𝑦𝑡 causes 𝑥𝑡. Thus, Granger’s approach basically seeks to determine 
whether the current values of 𝑦𝑡 is explained by past values of 𝑥𝑡. Therefore, 𝑥𝑡 is claim to 
Granger-causes 𝑦𝑡 if lagged 𝑥′𝑠 explained current 𝑦′𝑠 and thus, 𝑥𝑡 helps in the prediction of 𝑦𝑡. 
Specifically, a bivariate Granger causality test of the form equation (3.9) and (3.10) is adopted 
to empirically test the causal relationship between oil price volatility and exchange rate volatility.  
 
𝒑 𝒑
                           𝜹𝟐𝒕 = 𝜶𝟏𝟏 + ∑ 𝜷
𝟐 𝟐
𝟏𝟏𝒋𝜹 𝒕−𝒋 + ∑ 𝜷𝟏𝟐𝒋𝝈 𝒕−𝒋 + 𝜺𝟏𝒕                                             (𝟑. 𝟏𝟏) 
𝒋=𝟏 𝒋=𝟏
𝒑 𝒑
                          𝝈𝟐𝒕 = 𝜶
𝟐 𝟐
𝟐𝟏 + ∑ 𝜷𝟐𝟏𝒋𝜹 𝒕−𝒋 + ∑ 𝜷𝟐𝟐𝒋𝝈 𝒕−𝒋 + 𝜺𝟐𝒕                                             (𝟑. 𝟏𝟐) 
𝒋=𝟏 𝒋=𝟏
 
where 𝛿2  represents oil price volatility at week 𝑡,  𝜎2𝑡 𝑡 represents exchange rate volatility at 
week 𝑡, The coefficients, 𝛽12𝑗 and 𝛽21𝑗 estimate the time series impacts linking exchange rate 
volatility and oil price volatility. 𝑗 denotes the number of lagged observations and 𝑝 is the last or 
the nth lag observation considered. 𝛼11 and 𝛼21 are constant whiles  𝜀1𝑡 and 𝜀2𝑡 are the error 
36 
 
University of Ghana http://ugspace.ug.edu.gh
terms. Based on equations (3.11) and (3.12) above, a collective hypothesis of the form (3.13) is 
tested and the F-statistics (Wald Statistics) is reported: 
 
                                        𝜷𝟏𝟐𝒋 = 𝜷𝟐𝟏𝒋 = 𝟎                                                                                            (𝟑. 𝟏𝟑) 
 
The null hypothesis for this test states that 𝛿2𝑡 does not Granger-cause 𝜎
2
𝑡 regarding equation 
(3.9) and 𝜎2𝑡 does not Granger-cause 𝛿
2
𝑡 with respect to equation (3.10). The decision criterion 
in this test is that, if the corresponding probability value of the F-statistics test is less than 1% or 
5% or 10% significance level, the null is rejected.  
 
 
3.8 Data Analysis 
In order to address the objectives of this study, the stationarity (unit roots) test was conducted for 
the volatility obtained from the returns of the exchange rates and oil price. The variables which 
were non-stationary became stationary at first difference. Also, to achieve a valid result from the 
VAR model, the optimal lag selection was also conducted using the information criterion. 
Therefore, the VAR model is then estimated and afterwards the Granger causality test was 
conducted. It is worth stating that all these estimations were done using E-views statistical 
package 10. The results of these analyses are presented below in the next chapter. 
 
 
3.9 Chapter Summary 
The chapter summarized the methodology that is deemed appropriate and universally accepted 
for the study and the various assumptions of these models. This includes the research design, 
sample and sample periods and a brief outline of how the volatilities of the variables are 
37 
 
University of Ghana http://ugspace.ug.edu.gh
estimated. In addressing the first objective, the study mainly follows the Vector Autoregressive 
(VAR) approach. A causality test based on Granger (1969) approach is used to address the 
second objective of the study. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
38 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER FOUR 
DATA ANALYSIS AND DISCUSSION 
4.1 Introduction 
The focus of this chapter is the presentation and the discussion of the results obtained from the 
data analysis. The results were obtained by employing the methodologies and the data explained 
in the previous chapter of the study. Specifically, the presentation and the discussion on 
descriptive statistics, the VAR estimation results and its associated Granger Causality results are 
the focus of this chapter.  
4.2 Descriptive Statistics 
The summary statistics on all the variables employed for this study is presented in Table 4.1. 
Thus, the descriptive statistics on the weekly average returns of oil price and weekly returns of 
the exchange rates of the various countries considered for this study are reported in Table 4.1. 
The table presents the summary statistics of the pre-crisis sample period (Panel A) and post-crisis 
sample period (Panel B). Thus, Table 4.1 shows the descriptive statistics for each of the variables 
for the two subs – samples periods (before the crisis–2000 to 2007 and after the crisis–2010 to 
2016). 
From Table 4.1, it is observed that the average weekly return of the oil price for the pre-financial 
crisis period was approximately 32%. The volatility of this return measured by the standard 
deviation (4.0) was high compared to the average weekly return and this shows that the returns 
of oil price during the pre-financial crisis was volatile. The kurtosis value of  5.0995 of the oil 
return is greater than three which indicates oil price returns have significant leptokurtosis. The 
oil price return is negatively skewed (−0.8590) indicating that the distribution has a long-left tail. 
The Jarque-Bera test result rejects the normality hypothesis of the weekly oil return at 
39 
 
University of Ghana http://ugspace.ug.edu.gh
Table 4. 1: Descriptive Statistics of Oil Price and Bilateral Exchange Rate of the Dollar  
 𝐎𝐈𝐋𝐑𝐄𝐓𝐭 𝐆𝐇𝐒𝐑𝐄𝐓𝐭 𝐍𝐀𝐈𝐑𝐀𝐑𝐄𝐓𝐭 𝐑𝐀𝐍𝐃𝐑𝐄𝐓𝐭 𝐄𝐔𝐑𝐑𝐄𝐓𝐭 𝐑𝐔𝐁𝐑𝐄𝐓𝐭 𝐑𝐔𝐏𝐑𝐄𝐓𝐭 
Panel A: Pre-Crisis (𝟑/𝟑/𝟎𝟎 − 𝟐𝟖/𝟏𝟐/𝟎𝟕) - 𝟒𝟏𝟔 Observations 
𝐌𝐞𝐚𝐧  0.3239  0.2403  0.0363  0.0305 −0.0830 −0.0261 −0.0236 
𝐒𝐭𝐝. 𝐃𝐞𝐯.  4.0025  1.1177  1.1041  1.7362  1.0078  0.4599  0.4588 
𝐒𝐤𝐞𝐰𝐧𝐞𝐬𝐬 −0.8590  2.9368 −2.0756  0.7954  0.1622  1.8752 −0.5432 
𝐊𝐮𝐫𝐭𝐨𝐬𝐢𝐬  5.0995  22.6135  29.5581  5.0947  2.8880  20.1741  9.0975 
        
𝐉𝐚𝐫𝐪𝐮𝐞 𝐁𝐞𝐫𝐚  127.5690  7265.957  12524.44  119.9232  2.0403  5356.289  664.9110 
Probability  0.0000  0.0000  0.0000  0.0000  0.3605  0.0000  0.0000 
        
𝐐 (𝟑𝟔) 59.109∗∗∗ 227.05∗∗∗ 68.361∗∗∗ 107.69∗∗∗ 96.880∗∗∗ 40.517 84.159∗∗∗ 
 
Panel B: Post-Crisis (𝟏𝟎/𝟏/𝟏𝟎 − 𝟐𝟗/𝟎𝟓/𝟏𝟔) – 𝟑𝟑𝟑 Observations 
𝐌𝐞𝐚𝐧 −0.1559  0.3002  0.0833  0.2281  0.0759  0.2374 0.1148 
𝐒𝐭𝐝. 𝐃𝐞𝐯.  3.6522  2.2755  1.1936  2.0478  1.0410  2.2100 0.9644 
𝐒𝐤𝐞𝐰𝐧𝐞𝐬𝐬  0.0698 −2.0675  0.4799  0.3790  0.0274 −0.1138 0.4513 
𝐊𝐮𝐫𝐭𝐨𝐬𝐢𝐬  4.4944  24.9134  5.5976  3.9329  3.2300  19.7566 5.5797 
        
𝐉𝐚𝐫𝐪𝐮𝐞 𝐁𝐞𝐫𝐚       31.2566  6899.964  106.4019  20.0475  0.7756  3896.581 103.6409 
Probability  0.0000  0.0000  0.0000  0.0000  0.6785  0.0000 0.0000 
        
𝐐 (𝟑𝟔) 52.879∗∗ 57.094∗∗ 64.910∗∗∗ 40.122      87.456∗∗∗     83.411∗∗∗ 130.62∗∗∗ 
Note: 𝑂𝐼𝐿𝑅𝐸𝑇 = Oil price returns; 𝐺𝐻𝑆𝑅𝐸𝑇 = Ghana cedis/dollar exchange rate returns; 𝑁𝐴𝐼𝑅𝐴𝑅𝐸𝑇 = Nigeria naira/dollar 
exchange rate returns; 𝑅𝐴𝑁𝐷𝑅𝐸𝑇 = South Africa rand/dollar exchange rate returns; 𝐸𝑈𝑅𝑅𝐸𝑇= Euro/dollar exchange rate 
returns;  RUBRET = Russia rubble/dollar exchange rate returns; 𝑅𝑈𝑃𝑅𝐸𝑇 = Indiana rupee/dollar exchange rate returns; Q 
(k) is the Ljung Box Statistic 
 
1% significant level and this implies that the oil returns are not normally distributed.  The Ljung 
Box Statistic (Q (36)) was done to determine whether the series are autocorrelated. The null 
hypothesis of no serial correlation for the oil returns was rejected at 1% and concluded that the 
weekly returns of the oil for the pre-crisis period is autocorrelated and needs to be corrected.   
40 
 
University of Ghana http://ugspace.ug.edu.gh
Similarly, the descriptive statistics on the various currencies’ exchange rates returns for the pre-
crisis period are also reported in Table 4.1. It is observed in the pre-crisis period that the returns 
for the Ghana cedi and dollar exchange rate were approximately 24% whiles the Nigeria naira 
and the dollar exchange rate return was approximately 3.6%. The exchange rate return for South 
Africa rand and the dollar, Euro and the dollar, the Russia ruble and the dollar and the India 
rupee and the dollar are approximately 3.0%, −8.3%, −2.6% and −2.4% respectively. The 
volatility of the returns of these currencies measured using the standard deviation shows that the 
returns of the exchange rates are volatile compared to the average returns of these currencies 
with the South Africa rand/dollar recording the highest volatility value (1.7). The exchange rates 
return for all the countries are positively skewed except for Nigeria naira/dollar exchange rate 
returns and India rupee/dollar exchange rate returns. The positive skewness of the returns shows 
that the distributions of these variables have long right tails whiles the negative skewness shows 
long left tail.  
The returns of these currencies’ kurtosis values were all greater than 3.0 which implies these 
variables distribution have higher peaks or leptokurtic relative to the normal distribution. 
Specifically, the Ghana cedi/dollar returns (22.6), Nigeria naira/dollar returns (29.6) and 
Russian ruble/dollar returns (20.2) recorded kurtosis value above 20 showing how leptokurtic 
these variables are. The Jarque-Bera test result rejects the normality hypothesis of all the weekly 
currencies’ exchange rate returns except for the Euro/dollar exchange rate returns. The null 
hypothesis of no serial correlation was rejected for all the currencies’ exchange rates returns 
except the Russia ruble/dollar exchange rate returns. 
Considering the post-crisis period, it is also observed that the weekly average return of the oil 
price for the post-crisis period was approximately−15.6%. The volatility of this return measured 
41 
 
University of Ghana http://ugspace.ug.edu.gh
by the standard deviation is approximately 3.7. The kurtosis value of  4.4944 of the oil return is 
greater than three which indicates that the oil price returns have significant leptokurtosis. The oil 
price return is positively skewed indicating that the distribution has a long right tail. The Jarque-
Bera test result rejects the normality hypothesis of weekly oil return for the post-crisis period.   
  
Also, the summary statistics on the various currencies’ exchange rates returns for the post-crisis 
period are also reported in Table 4.1. The exchange rates return for Ghana cedi/dollar, Nigeria 
naira/dollar, South Africa rand /dollar, Euro/dollar, the Russia ruble/dollar and the Indian 
rupee/dollar are approximately 30.0%, 8.3%, 22.8%, 7.6%, 23.7% and 11.4% respectively. 
The volatility of the returns of these currencies measured using the standard deviation shows that 
the returns of the exchange rates are less volatile compared to the average returns of these 
currencies and that of the pre-crisis period. The exchange rate returns for all the countries are 
positively skewed except for Ghana cedi/dollar exchange rate returns and Russia ruble/dollar 
exchange rate returns.  
 
The returns of the currencies exchange rates kurtosis values were all greater than 3.0 which 
implies that, these variables distribution have higher peaks or leptokurtic relative to the normal 
distribution. Also, the Jarque-Bera test result rejects the normality hypothesis of all the weekly 
currencies exchange rate returns except for the Euro/dollar exchange rate returns. The null 
hypothesis of no serial correlation for the returns of the exchange rate of these countries using 
Ljung Box Statistic (Q - 36) revealed that all the returns of the currencies’ exchange rates were 
autocorrelated except the South Africa rand/dollar exchange rate returns. 
 
 
42 
 
University of Ghana http://ugspace.ug.edu.gh
4.3 Characteristics of the Oil Price Returns and Exchange Rate Returns  
One of the aims of this study is to investigate the relationship between the volatility of oil price 
and the volatility of the various selected currencies exchange rate. To achieve this aim, the 
method to adopt for the estimation of the volatility of these variables is keen since financial time 
series data have peculiar characteristics from other data. The pictorial behavior of these variables 
as depicted in Appendix A are employed to determine whether to use linear or non-linear models 
(such as Autoregressive Conditional Heteroskedasticity models) to estimate the volatilities of the 
two markets. From Appendix A (Panel A and Panel B), it is observed that large changes in the 
returns follow large changes while small changes follow small changes. Thus, current volatility 
tends to be positively correlated with its immediately preceding period. This implies that there is 
an evidence of volatility pooling or clustering in the returns of these variables. 
 
Also, it is also observed that the volatility occurs in bursts. Thus, the diagram shows that the 
various variables appear to be stable over time and far more volatility is observed in other 
periods. From the diagram, it is also observed that there is evidence of leverage effect since 
negative returns trigger more volatility than the positive returns. All these characteristics are 
common to financial data and one of the best models in the literature to model such behavior in 
the series is the Autoregressive Conditional Heteroskedasticity (ARCH) method (Brooks, 2008), 
as a result, the ARCH models were used to model the volatility of these returns. 
 
4.4 Volatility Trend of Exchange Rate and Oil Price  
The focus of the study is on the aspect of volatilities in analyzing the interrelation impact 
between exchange rates and oil price in the two sub – sample periods considered in the study. 
This is because the estimated volatilities in these two markets were used throughout the              
43 
 
University of Ghana http://ugspace.ug.edu.gh
subsequent analysis to achieve the objectives of the study. This necessitated a graphical 
representation of the estimated volatilities in these two markets to provide a better understanding 
of the trends and sparks of the volatilities in these two markets during the two sub sample 
periods. The graphical behavior of the volatilities in these two markets is depicted in Appendix 
B, consisting of Panel A and Panel B representing the pre- and post-crisis periods respectively.  
 
From Appendix B (Panel A and Panel B), there were no sparks in the trend of the euro exchange 
rate volatility throughout the two subsample periods. However, it can be observed from the pre-
crisis period that the euro volatility was a little stable from the year 2002 to 2004, however, 
volatility decreased earlier from the year 2000 to 2002 and decreased later from the year 2005 to 
2007. The euro volatility was relatively high in the post-crisis period with increasing volatilities 
in the year 2011 and from 2014 to 2015. There was also evidence of a decreasing volatility 
within the periods from the year 2010 to 2011 and from mid-2011 to 2014. 
 
The South Africa rand’s exchange rate volatility was stable throughout the pre-crisis period, 
particularly from the year 2005 to 2007.  However, there were sparks in the volatility between 
the periods of 2001 - 2002 and 2003 – 2004. There was also evidence of volatility pooling in the 
trend of the rand exchange rate volatility in the pre-crisis period. The up and down swings in the 
rand volatility were significant in the post-crisis period. Similar to the pre-crisis period, there was 
a spark between the year 2011 and 2012 and a more serious spark between the years 2015 and 
2016. 
The Ghana cedi exchange rate in Panel A of Appendix B was very volatile from the year 2000 to 
2001. Volatility in the same period was relatively stable after 2001, specifically, from the year 
2002 to 2007. Similarly, there was a relative stability of the Ghana cedi volatility in Panel B of 
44 
 
University of Ghana http://ugspace.ug.edu.gh
Appendix B from the year 2010 to 2013. However, there were high sparks from late 2013 
throughout to 2016. 
 
The Russia ruble exchange rate volatility was very significant throughout the pre-crisis period 
and there was evidence of volatility clustering or pooling in the series throughout the period. 
Specifically, higher volatility turned to follow higher volatility throughout the period. Unlike the 
pre-crisis period, the ruble volatility was relatively stable from 2010 to 2014 but there were 
relatively high sparks from 2014 to 2016. 
 
The India rupee was stable throughout the pre-crisis period but there were severe sparks from the 
year 2004 to 2007. The highest sparks were witnessed in the year 2004. In the post-crisis period, 
the sparks were realized from the year 2011 to 2012 and from 2013 to 2014. Apart from that, 
volatility was relatively stable and also, it was higher in the post-crisis period than the pre-crisis 
period. 
 
The Nigeria naira exchange rate was more volatile from the year 2000 to the late 2003 and 
however, volatility was relatively low from the year 2004 to 2007 in the pre-crisis period. In the 
period after the crisis, the naira exchange rate volatility was relatively stable and low from the 
year 2010 to 2014. It is worth noting that there were some high sparks from 2014to 2015 and 
finally became very low from the year 2015 to 2016.  
Crude oil price volatility was very low throughout the pre-crisis period. However, there was a 
downward spark in the volatility over that period. That is, there was a significant decrease in oil 
price in the year 2001 and the year 2003.  Unlike the pre-crisis period, oil price volatility was 
relatively stable in the post-crisis period from the year 2010 to 2011, but oil price volatility 
started increasing from 2014 through to 2016. 
45 
 
University of Ghana http://ugspace.ug.edu.gh
4.5 Unit Roots Test Results 
According to Brooks (2008), when dealing with time series data, it is very necessary to test the 
stationarity of the series in order to avoid spurious regression results. The results of the  
Table 4. 2: Unit Roots Test Results 
 ADF PP 
Time series Intercept Trend and intercept Intercept  Trend and intercept 
Pre-financial crisis sample period 
OILVOL −13.2934∗∗∗ −13.7661∗∗∗  −13.3310∗∗∗ −13.7783∗∗∗ 
GHSVOL −3.8707∗∗∗  −6.5710∗∗∗ −5.8087∗∗∗ −6.4956∗∗∗ 
NAIRAVOL −12.0817∗∗∗∗ −12.8107∗∗∗ −12.4917∗∗∗  −12.9600∗∗∗ 
RANDVOL −4.4696∗∗∗ −4.4911∗∗∗ −4.2595∗∗∗ −4.2716∗∗∗ 
EURVOL −2.0911 −2.9133 −2.0977           -2.8720 
RUBVOL −4.3386∗∗∗ −4.6742∗∗∙ −4.2867∗∗∗ −4.6249∗∗∗ 
RUPVOL −6.3954∗∗∗ −7.1025∗∗∗ −6.2617∗∗∗ −6.9753∗∗∗ 
Post-financial crisis sample period 
OILVOL −1.1558 −1.7022  −1.0822 −1.6371 
GHSVOL −4.18471∗∗∗ −4.4377∗∗∗ −5.4448∗∗∗ −5.8496∗∗∗ 
NAIRAVOL −12.6086∗∗∗ −12.5931∗∗∗ −12.8619∗∗∗ −12.8459∗∗∗  
RANDVOL −2.9577∗∗ −3.1083 −2.6023 −2.7395 
EURVOL −2.1689 −2.2594 −2.0931 −2.1935 
RUBVOL −5.3012∗∗∗ −5.6626∗∗∗ −3.8673∗∗∗ −4.0748∗∗∗ 
RUPVOL −3.2560∗∗ −3.4360∗∗ −3.3680∗∗ −3.5503∗∗ 
Note: OILVOL= Oil price return volatility; GHSVOL = Ghana Cedi/dollar exchange rate return volatility; NAIRAVOL = 
Nigeria Naira/dollar exchange rate return volatility; RANDVOL = South Africa Rand/dollar exchange rate return 
volatility; EURVOL = Europe Euro/dollar exchange rate return volatility RUBVOL = Russia Rubble/dollar exchange rate 
return volatility; RUPVOL= India Rupee/dollar exchange rate return volatility; ***, ** and * indicates significant at 1%, 
5% and 10% level respectively; 
Augmented Dickey-Fuller (ADF) unit root test and the Phillips-Perron (PP) unit root test 
employed to test the stationarity of the variables used in this study are depicted in Table 4.2. The 
null hypothesis for the ADF and PP test postulates that variable in question has a unit root hence 
it is non-stationary series. Both the ADF and PP test was conducted under the assumption of only 
46 
 
University of Ghana http://ugspace.ug.edu.gh
intercept and both trend and intercept only. In this study, a variable is stationary if the null 
hypothesis of ADF and PP are rejected under all assumptions.  
 
It is observed from Table 4.2 that, oil price volatility (estimated using GARCH model) and the 
currencies returns volatility (estimated using EGARCH) for the pre-financial crisis period were 
all stationary at a level of 1% significant level except for Euro/dollar exchange rate return 
volatility. However, the Euro/dollar exchange rate return volatility became stationary at first 
difference hence all the volatility variables used for the analysis were at level except Euro/dollar 
exchange rate return volatility. On the other hand, over the post-financial crisis period, all the 
variables were found to be stationary at a level of 1% significant level except for the oil price 
volatility and Euro/dollar exchange rate return volatility which became stationary at first 
difference. Therefore, over the post-crisis period, all the volatility variables which were used for 
the analysis were at levels except oil price volatility and the Euro/dollar exchange rate return 
volatility which were at first difference.  
 
4.6 VAR Estimation and Test Results 
4.6.1 Optimal Lag Selection 
The determination of optimal lag length is very important for a valid VAR model results. The 
literature does not specify the optimal lag to use for the VAR model. Therefore, the optimal lag 
order for the VAR analysis for this study is obtained following the Likelihood Ratio and 
information criteria test. E-views computation of the likelihood ratio and different version of 
information criteria until the 8th lags are depicted in Appendix C and D. Appendix C depicts the 
optimal lag selection results of the oil returns volatility and the various currencies exchange rate 
returns volatility for the pre-financial crisis period whiles Appendix D shows the optimal lag 
results of these variables  
47 
 
University of Ghana http://ugspace.ug.edu.gh
over the post-financial crisis period. The decision for the optimal lag selection was based on the 
Akaike Information Criterion (AIC) and Likelihood Ratio (LR). 
Over the pre-crisis period, it is observed from Table 1 of Appendix C that the optimal lag 
selection results for the Ghana cedi/dollar exchange rate return volatility and the oil price return 
volatility revealed that the optimal lag for the VAR using Akaike Information Criterion (AIC) 
was 4 whiles using Likelihood Ratio (LR) was 7. However, the researcher used the AIC as the 
decision criterion because it is the most used criterion in the literature. Therefore, lag length of 4 
was used to construct the VAR involving Ghana cedi/dollar exchange rate return volatility and 
the oil price return volatility over the pre-crisis period.  Similarly, the optimal lag selection 
results presented in Table 2, Table 3, Table 4, Table 5 and Table 6 show the optimal lag length 
selection of Nigeria naira/dollar volatility and oil price return volatility, South Africa rand/dollar 
volatility and oil price return volatility, Euro/dollar return volatility and oil price return volatility, 
Russia ruble/dollar volatility and oil price return volatility and India rupee/dollar volatility and 
oil price return volatility respectively. Both AIC and LR unanimously revealed that lag 1 is the 
optimal lag length for constructing the VAR for these variables. Table 4 of Appendix C revealed 
AIC with a lag of 2 and LR with a lag of 1 but lag 2 was used to construct the VAR.  
 
The optimal lag length selection over the post-crisis period is presented in Appendix D. Table 7, 
Table 8, Table 9, Table 10, Table 11 and Table 12 in Appendix D shows the lag length selection 
results for Ghana cedi/dollar exchange rate return volatility, Nigeria naira/dollar volatility, South 
Africa rand/dollar volatility, Euro/dollar return volatility, Russia ruble/dollar volatility and India 
rupee/dollar volatility with oil price return volatility respectively. Based on the AIC results, the 
optimal lag used to construct the VAR of Ghana cedi/dollar exchange rate return volatility and 
oil price return volatility, Nigeria naira/dollar volatility and oil price return volatility, South 
48 
 
University of Ghana http://ugspace.ug.edu.gh
Africa rand/dollar volatility and oil price return volatility, Euro/dollar return volatility and oil 
price return volatility, Russia ruble/dollar volatility and oil price return volatility and India 
rupee/dollar volatility and oil price return volatility are 8, 1, 8, 1, 6 and 1 respectively. 
 
4.6.2 The VAR Estimation Results 
The VAR system was employed to address the research objective one and the results are reported 
for Table 4.3 (pre-crisis) and Table 4.4 (post-crisis). The VAR system is made up of two 
endogenous variables, namely, oil price return volatility and the currencies exchange rate return 
volatility (Ghana cedi/dollar exchange rate return volatility, Nigeria naira/dollar volatility, South 
Africa rand/dollar volatility, Euro/dollar return volatility, Russia ruble/dollar volatility and India 
rupee/dollar volatility). Both Table 4.3 and Table 4.4 are arranged into columns for endogenous 
variables and rows for the lag terms of the endogenous variables. For each of the variables, the 
estimated coefficient, the test statistics value, and p-value results are reported. To verify if the 
variable estimated coefficient is significantly different from zero, the probability values (p-
values) are employed. Thus, variables with a reported p-value of less than 10% are considered as 
significant in this study. Also, variables with a reported p-value less than 5% or 1% are also 
considered as highly significant in this study.  
 
4.6.3 VAR Estimation Results for the Pre-Financial Crisis 
 From Table 4.3, the Panel A shows the VAR system results for oil price volatility and Ghana 
cedi/dollar exchange rate volatility (using 4 lags). The results show that the oil price volatility is 
very significantly autocorrelated at lag one with standard deviation 0.0410.  It implies that the oil 
price volatility is significantly affected by its own behaviour in the past week and this means that 
market players can use past week oil price return volatility to predict future oil price volatility.  
 
49 
 
University of Ghana http://ugspace.ug.edu.gh
Table 4. 3: VAR Estimation Results for Pre-Financial Crisis Period 
   𝑪𝒐𝒆𝒇𝒇     𝑺𝒕. 𝒆𝒓𝒓𝒐𝒓 𝒑 − 𝒗𝒂𝒍𝒖𝒆 𝑪𝒐𝒆𝒇𝒇 𝑺𝒕. 𝒆𝒓𝒓𝒐𝒓 𝒑 − 𝒗𝒂𝒍𝒖𝒆 
 Panel A: VAR results of Cedi returns volatility and Oil price volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑮𝑯𝑺𝑽𝑶𝑳𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿𝑡−1    0.5290
∗∗∗       0.0410 0.0000 0.0138 0.0867 0.8738 
𝛼    7.2118∗∗∗       0.8718 0.0000 0.0868 1.5123 0.9542 
𝐺𝐻𝑆𝑉𝑂𝐿𝑡−1    0.0467
∗       0.0285 0.0992      0.7925∗∗∗ 0.0490 0.0000 
𝐺𝐻𝑆𝑉𝑂𝐿𝑡−3    0.0244       0.0284 0.3905      0.1753
∗∗∗ 0.0627 0.0053 
𝐺𝐻𝑆𝑉𝑂𝐿𝑡−4    0.0244       0.0284 0.3905  − 0.1969
∗∗∗ 0.0493 0.0001 
𝑹𝟐       0.2968         0.7355  
𝑨𝒅𝒋. 𝑹𝟐          0.2828         0.7303  
 Panel B: VAR results of Naira returns volatility and oil price volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑵𝑨𝑰𝑹𝑨𝑽𝑶𝑳𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿    0.5037∗∗∗𝑡−1        0.0373 0.0000 −0.0579 0.0911 0.5250 
𝛼   7.6652∗∗∗       0.5812 0.0000 1.5291 1.4176 0.2810 
𝑁𝐴𝐼𝑅𝐴𝑉𝑂𝐿  −0.0106       0.0180 0.5564      0.4548∗∗∗𝑡−1  0.0439 0.0000 
𝑹𝟐          0.3081         0.2083  
𝑨𝒅𝒋. 𝑹𝟐          0.3048         0.2045  
                                          Panel C: VAR results of Rand returns volatility and oil price return volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑹𝑨𝑵𝑫𝑽𝑶𝑳𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿𝑡−1   0.5047
∗∗∗ 0.0374 0.0000 −0.0185 0.0172 0.2815 
𝛼   7.6127∗∗∗ 0.5914 0.0000   0.4500∗ 0.2723 0.0988 
𝑅𝐴𝑁𝐷𝑉𝑂𝐿𝑡−1 0.0095 0.0358 0.7905     0.9396
∗∗∗ 0.0165 0.0000 
𝑹𝟐          0.3077         0.8881   
𝑨𝒅𝒋. 𝑹𝟐          0.3043         0.8876  
                                          Panel D: VAR results of Euro returns volatility and oil price return volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑬𝑼𝑹𝑶𝑽𝑶𝑳𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿𝑡−1    0.5359
∗∗∗  0.0499 0.0000 0.0009 0.0011 0.4049 
𝛼    7.1682∗∗∗        0.7348 0.0000 −0.0024 0.0163 0.8841 
𝐸𝑈𝑅𝑂𝑉𝑂𝐿𝑡−2 −0.5841 2.2136 0.7919   −0.1282
∗∗∗ 0.0491 0.0093 
𝑹𝟐          0.3025         0.9917  
𝑨𝒅𝒋. 𝑹𝟐          0.2991         0.9916  
                                         Panel E: VAR results of Ruble returns volatility and oil price return volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑹𝑼𝑩𝑽𝑶𝑳𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿𝑡−1   0.5025
∗∗∗ 0.0376 0.0000   0.0017 0.0018 0.3195 
𝛼   7.6415∗∗∗ 0.5801 0.0000 −0.0117 0.0270 0.6650 
𝑅𝑈𝐵𝑉𝑂𝐿𝑡−1 0.1637 0.4161 0.6941      0.9142
∗∗∗ 0.0194 0.0000 
𝑹𝟐          0.3078         0.8471  
𝑨𝒅𝒋. 𝑹𝟐          0.3044         0.8463  
                                             Panel F: VAR results of Rupee returns volatility and oil price return volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑹𝑼𝑷𝑽𝑶𝑳𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿𝑡−1   0.5049
∗∗∗         0.0373                        0.0000 0.0015 0.0031 0.6253 
𝛼   7.5730∗∗∗ 0.5853 0.0000 0.0147 0.0490 0.7646 
𝑅𝑈𝑃𝑉𝑂𝐿𝑡−1 0.2915 0.3343 0.3836       0.8215
∗∗∗         0.0280 0.0000 
𝑹𝟐          0.3088         0.6772  
𝑨𝒅𝒋. 𝑹𝟐          0.3057         0.6757  
Note: OILVOL = Oil price return volatility; GHSVOL = Ghana Cedi/dollar exchange rate return volatility; NAIRAVOL = Nigeria 
Naira/dollar exchange rate return volatility; RANDVOL = South Africa Rand/dollar exchange rate return volatility; EURVOL = 
Europe Euro/dollar exchange rate return volatility RUBVOL = Russia Ruble/dollar exchange rate return volatility; RUPVOL = India 
Rupee/dollar exchange rate return volatility; ***, ** and * indicates significant at 1%, 5% and 10% level respectively. 
50 
 
University of Ghana http://ugspace.ug.edu.gh
More importantly, only Ghana cedi/dollar exchange rate volatility was found to have significant 
impact on the volatility of oil price returns at lag 1 with coefficient 0.0467 and standard error 
0.0285 in the pre – crisis period. On the other hand, the Ghana cedi/dollar exchange rate 
volatility was very significantly autocorrelated at lag 1, lag 3 and lag 4 with standard errors 
0.0490, 0.0627 and 0.0493 respectively. However, oil price volatility was found not to have 
significant impacts on Ghana cedi/dollar exchange rate volatility. 
The Panel B depicts the VAR system results for oil price volatility and Nigeria naira/dollar 
exchange rate volatility (using 1 lags). The results show that both the oil price volatility and 
Nigeria naira/dollar exchange rate volatility are very significantly autocorrelated at lag one. It 
means that the oil price volatility and Nigeria naira/dollar exchange rate volatility are very 
significantly affected by its own behaviour in past. On the other hand, the Nigeria naira/dollar 
exchange rate volatility was not significantly found to have impacts on oil price volatility and oil 
price volatility was also found not to have significant impacts on the Nigeria naira/dollar 
exchange rate volatility.  
 
Moreover, the Panel C also considers the VAR system for oil price volatility and South Africa 
rand/dollar exchange rate volatility (using lag one). It is observed from Panel C that, both oil 
price return volatility and the South Africa rand/dollar exchange rate volatility are very 
significantly autocorrelated at lag one. Thus, past week oil price return volatility can be used to 
predict future week oil price return volatility. Similarly, past week South Africa rand/dollar 
exchange rate volatility can be used to predict future week volatility of South Africa rand/dollar 
exchange rate volatility.  It is observed that oil price volatility insignificantly has impacts on 
South Africa rand/dollar exchange rate volatility and South Africa rand/dollar exchange rate 
volatility does not have significant impacts on oil price volatility. 
51 
 
University of Ghana http://ugspace.ug.edu.gh
The VAR system results of oil price return volatility and Euro/dollar exchange rate return 
volatility (using 2 lags) is depicted in Panel D above. The results show that the oil price volatility 
is very significantly affected by its past week (lag 1) volatility which means that past week oil 
return volatility can be used by market participants to forecast future oil return volatility. It is 
also observed that the Euro/dollar exchange rate return volatility is very significantly 
autocorrelated at lag 2 with standard error 0.0491. The results also show that oil price volatility 
has no significant impacts on the Euro/dollar exchange rate return volatility and Euro/dollar 
exchange rate return volatility recorded no significant impacts on oil return volatility. 
 
Panel E of Table 4.3 depicts the VAR results of the oil price return volatility and Russia 
ruble/dollar exchange rate return volatility. The results show that the oil price volatility is very 
significantly affected by its past week (lag 1) volatility which means that past week oil return 
volatility can be used by market participants to forecast future oil return volatility. It is also 
observed that Russia ruble/dollar exchange rate return volatility is very significantly 
autocorrelated at lag 2 with standard error 0.0194. The results also show that oil price volatility 
has no significant impacts on the Russia ruble/dollar exchange rate return volatility and Russia 
ruble/dollar exchange rate return volatility recorded no significant impacts on oil return 
volatility.  
Finally, Panel F depicts the VAR results of the oil price return volatility and India rupee/dollar 
exchange rate return volatility. The results show that the oil price volatility is very significantly 
affected by its past week (lag 1) volatility which means that past week oil return volatility can be 
used by market participants to forecast future oil return volatility. It is also observed that India 
rupee/dollar exchange rate return volatility is very significantly autocorrelated at lag 1 with 
standard error 0.0280. 
52 
 
University of Ghana http://ugspace.ug.edu.gh
4.6.4 VAR Estimation Results for the Post – Financial Crisis 
From Table 4.4, the Panel A shows the VAR system results for oil price volatility and Ghana 
cedi/dollar exchange rate volatility (using 8 lags). 
Table 4. 4: VAR Estimation Results for Post-Financial Crisis Period 
 𝑪𝒐𝒆𝒇𝒇 𝑺𝒕. 𝒆𝒓𝒓𝒐𝒓 𝒑 − 𝒗𝒂𝒍𝒖𝒆 𝑪𝒐𝒆𝒇𝒇 𝑺𝒕. 𝒆𝒓𝒓𝒐𝒓 𝒑 − 𝒗𝒂𝒍𝒖𝒆 
                             Panel A: VAR results of Cedi volatility and Oil price volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑮𝑯𝑺𝑽𝑶𝑳𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿𝑡−2    0.0328 0.0570 0.5649 −0.3544
∗∗ 0.1572 0.0245 
𝛼 −0.0564 0.1209 0.6413     0.0868∗∗∗ 0.3334 0.0072 
𝐺𝐻𝑆𝑉𝑂𝐿     0.0079 0.0206 0.7028 −0.6962∗∗∗𝑡−1  0.0569 0.0000 
𝐺𝐻𝑆𝑉𝑂𝐿𝑡−3    0.0237 0.0252 0.3467    0.3748
∗∗∗ 0.0694 0.0000 
𝐺𝐻𝑆𝑉𝑂𝐿𝑡−4 −0.0234 0.0258 0.3650 −0.2995
∗∗∗ 0.0712 0.0000 
𝐺𝐻𝑆𝑉𝑂𝐿𝑡−5    0.0199 0.0258 0.4418    0.2664
∗∗∗ 0.0712 0.0002 
𝐺𝐻𝑆𝑉𝑂𝐿   −0.0533∗∗𝑡−6  0.0253 0.0358 −0.0938 0.0698 0.1798 
𝐺𝐻𝑆𝑉𝑂𝐿 ∗∗𝑡−8     0.0515  0.0205 0.0123 −0.1458
∗∗ 0.0565 0.0102 
𝑹𝟐      0.9627        0.7651  
𝑨𝒅𝒋. 𝑹𝟐      0.9607        0.7529  
                             Panel B: VAR results of Naira returns volatility and oil price volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑵𝑨𝑰𝑹𝑨𝑽𝑶𝑳𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿𝑡−1 −0.0621 0.0551 0.2607 0.1574
∗∗ 0.0398 0.0001 
𝛼    0.1794 0.1372 0.1914 0.8480∗∗∗ 0.0991 0.0000 
𝑁𝐴𝐼𝑅𝐴𝑉𝑂𝐿𝑡−1 −0.0753 0.0702 0.2836 0.3511
∗∗∗ 0.0507 0.0000 
𝑹𝟐      0.9600        0.1209  
𝑨𝒅𝒋. 𝑹𝟐      0.9598        0.1156  
                             Panel C: VAR results of Rand returns volatility and oil price return volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑹𝑨𝑵𝑫𝑽𝑶𝑳𝒍𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿𝑡−1 −0.0505 0.0568 0.3744 −0.0413
∗ 0.0212 0.0516 
𝛼    0.1661 0.2460 0.4999        0.2165∗∗ 0.0917 0.0185 
𝑅𝐴𝑁𝐷𝑉𝑂𝐿     0.0428 0.1543 0.7818     0.8826∗∗∗𝑡−1  0.0575 0.0000 
𝑅𝐴𝑁𝐷𝑉𝑂𝐿𝑡−4 −0.3009 0.2045 0.1418     0.2451
∗∗∗ 0.0762 0.0014 
𝑅𝐴𝑁𝐷𝑉𝑂𝐿𝑡−5      0.5723
∗∗∗ 0.2057 0.0056 −0.1613∗∗ 0.0766 0.0357 
𝑅𝐴𝑁𝐷𝑉𝑂𝐿𝑡−6  −0.4006
∗ 0.2092 0.0560         0.1056 0.0780 0.1759 
𝑅𝐴𝑁𝐷𝑉𝑂𝐿𝑡−7    0.3656
∗ 0.2109 0.0835 0.0299 0.0786 0.7036 
𝑅𝐴𝑁𝐷𝑉𝑂𝐿𝑡−8    −0.5363
∗∗∗ 0.1588 0.0008 −0.1090∗ 0.0592 0.0660 
𝑹𝟐        0.9637        0.8955  
𝑨𝒅𝒋. 𝑹𝟐        0.9619        0.8901  
                             Panel D: VAR results of Euro returns volatility and oil price return volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑬𝑼𝑹𝑶𝑽𝑶𝑳𝒕 
𝐸𝑈𝑅𝑂𝑉𝑂𝐿𝑡−1 −0.0781 0.2135 0.7148      0.9712
∗∗∗ 0.0127 0.0000 
𝛼    0.1553 0.2312 0.5020      0.0260∗∗ 0.0138 0.0595 
𝑹𝟐         0.9599        0.9460  
𝑨𝒅𝒋. 𝑹𝟐         0.9596        0.9456  
53 
 
University of Ghana http://ugspace.ug.edu.gh
                              Panel E: VAR results of Ruble returns volatility and oil price return volatility 
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑹𝑼𝑩𝑽𝑶𝑳𝒕 
𝑂𝐼𝐿𝑉𝑂𝐿𝑡−5 0.0100 0.0553 0.8566   −0.1209
∗ 0.0722 0.0943 
𝛼 −0.0445 0.1134 0.6949       0.3052∗∗ 0.1479 0.0394 
𝑅𝑈𝐵𝑉𝑂𝐿       0.1383∗∗∗ 0.0436 0.0016       0.9532∗∗∗𝑡−1  0.0569 0.0000 
𝑅𝑈𝐵𝑉𝑂𝐿𝑡−2      -0.0734 0.0601 0.2222       0.5275
∗∗∗ 0.0784 0.0000 
𝑅𝑈𝐵𝑉𝑂𝐿𝑡−3   −0.2210
∗∗∗ 0.0646 0.0007    −0.6549∗∗∗ 0.0842 0.0000 
𝑅𝑈𝐵𝑉𝑂𝐿𝑡−4     0.1555
∗∗ 0.0645 0.0163 −0.0075 0.0720 0.9171 
𝑅𝑈𝐵𝑉𝑂𝐿 ∗∗∗𝑡−5      0.2076  0.0601 0.0006 −0.1209 0.0722 0.0943 
𝑅𝑈𝐵𝑉𝑂𝐿𝑡−6   −0.1734
∗∗∗ 0.0437 0.0001 −0.0125 0.0735 0.8654 
𝑹𝟐        0.9642        0.9191  
𝑨𝒅𝒋. 𝑹𝟐        0.9628        0.9161  
                             Panel F: VAR results of Rupee returns volatility and oil price return volatility  
 𝑶𝑰𝑳𝑽𝑶𝑳𝒕 𝑹𝑼𝑷𝑽𝑶𝑳𝒕 
𝑅𝑈𝑃𝑉𝑂𝐿𝑡−1    −0.2519 0.2319 0.2777       0.9368
∗∗∗ 0.0195 0.0000 
𝛼       0.2664 0.1995 0.1822       0.0460∗∗∗ 0.0168 0.0063 
𝑹𝟐         0.9601        0.8758  
𝑨𝒅𝒋. 𝑹𝟐         0.9599        0.8750  
***, ** and * indicates significant at 1%, 5% and 10% level respectively 
The result shows that the Ghana cedi/dollar exchange rate volatility is very significantly 
autocorrelated at lag one, lag three, lag four, lag five with standard errors 0.0596, 0.0694, 0.0712 
and 0.0712 respectively and significantly autocorrelated at lag eight with standard error 0.0565.  
It implies that the Ghana cedi/dollar exchange rate volatility is significantly affected by its own 
behaviour in past week and this means that market players can use Ghana cedi/dollar exchange 
rate volatility to predict future Ghana cedi/dollar exchange rate volatility. It is also observed 
from Table 4.4 that, Ghana cedi/dollar exchange rate volatility was found to have significant 
impacts on oil price return volatility at lag 6 and lag 8 with coefficients -0.0533 and 0.0515 
respectively and standard errors 0.0253 and 0.0205 respectively.  On the other hand, oil price 
volatility was also found to have negative significant impacts on Ghana cedi/dollar exchange rate 
volatility at lag 2 with coefficient -0.3544 and standard error 0.1572.  
 
Panel B of Table 4.4 shows the result of the VAR system of oil price return volatility and Nigeria 
naira/dollar exchange rate volatility. The Nigeria naira/dollar exchange rate volatility was found 
54 
 
University of Ghana http://ugspace.ug.edu.gh
to be highly significantly autocorrelated at lag 1 with standard error 0.0507. Also, the oil price 
return volatility was observed to have significant impacts on the Nigeria naira/dollar exchange 
rate volatility at lag one with coefficient 0.1574 and standard error 0.0398.   
 
The Panel C depicts the results of the VAR system of oil price return volatility and South Africa 
rand/dollar exchange rate volatility. The South Africa rand/dollar exchange rate volatility is 
highly significantly autocorrelated at lag 1 and lag 4 with standard errors 0.0575 and 0.0762 
respectively and significant at lag 5 and lag 8 with standard errors 0.0766 and 0.0592 
respectively. This shows that past weeks South Africa rand/dollar exchange rate volatility can be 
used to predict future volatility. The oil price return volatility at lag 1 was found to have 
significant impact on South Africa rand/dollar exchange rate volatility with coefficient – 0.0413 
and standard error 0.0212. On the other hand, South Africa rand/dollar exchange rate volatility at 
lag 5, lag 6, lag 7 and lag 8 was found to have impacts on the oil price volatility with coefficients 
0.5723, -0.4006, 0.3656 and – 0.5363 respectively and standard errors 0.2057, 0.2092, 0.2109 
and 0.1588 respectively.  
 
The Panel D shows the results of the VAR system of oil price return volatility and Euro/dollar 
exchange rate return volatility. From Panel D, it observed that the Euro/dollar exchange rate 
return volatility is highly significant at lag 1 with standard error 0.0127.     
 
Panel E of Table 4.4 depicts the VAR results of the oil price return volatility and Russia 
ruble/dollar exchange rate return volatility. The results show that the Russia ruble/dollar 
exchange rate return volatility is highly significantly autocorrelated at lag 1, lag 2 and lag 3 with 
standard errors 0.0569, 0.0784 and 0.0842 respectively. It is observed that oil price volatility at 
lag five has significant impacts on Russia ruble/dollar exchange rate return volatility with 
55 
 
University of Ghana http://ugspace.ug.edu.gh
coefficient – 0.1209 and standard error 0.0722. Also, it is observed that Russia ruble/dollar 
exchange rate return volatility at lag 1, lag 3, lag 4, lag 5 and lag 6 has highly significant impacts 
on oil price return volatility with coefficients 0.1383, – 0.2210, 0.1555, 0.2076, -0.1734 and 
standard errors 0.0436, 0.0646,0.0645, 0.0601 and 0.0437 respectively. The impact of oil price 
volatility return on Russia ruble/dollar exchange rate return is consistent with the findings of 
Tuzova and Qayum (2016).  
 
The Panel F shows the VAR results of the oil price return volatility and India rupee/dollar 
exchange rate return volatility. The results revealed that India rupee is autocorrelated at lag 1 
with standard error 0.0195. However, oil price return volatility was found not to have significant 
impacts on the India rupee/dollar exchange rate return volatility. Also, India rupee/dollar 
exchange rate return volatility was found not to have significant impacts on the oil price 
volatility. 
 
4.7 Granger Causality Test Result 
The Granger causality test is conducted to address the second research objective. Granger 
causality test is conducted to detect the causality between oil price return volatility and the 
currencies’ exchange rate return volatility. However, this test is not conducted for such only 
purpose but also is conducted alongside the VAR model estimation in this study to check the 
validity of the results obtained from the VAR model. Employing the various lags of the 
endogenous variables used for the VAR, the results of Granger – causality over the pre-financial 
crisis is depicted in Table 4.5 whiles post-crisis period is depicted in Table 4.6. 
 
 
 
 
56 
 
University of Ghana http://ugspace.ug.edu.gh
4.7.1 Granger – Causality for the Pre – Financial Crisis Period 
Over the pre-crisis period, considering the first item, the null hypothesis that the weekly oil price 
returns volatility does not Granger-cause weekly Ghana cedi/dollar exchange rate return 
volatility (P-value =0.9001) over the pre-financial crisis was not rejected at any conventional 
level (10%, 5% and 1%). This implies that oil price volatility does not Granger – cause the 
exchange rate volatility of Ghana. On the other hand, the null hypothesis that weekly Ghana 
cedi/dollar exchange rate return volatility does not Granger – cause oil price volatility (p-value = 
0.0722) is rejected at 10% significant level. This result implies that, at 10% significant level, 
there is enough evidence to show that weekly Ghana cedi/dollar exchange rate return volatility 
granger – cause oil price volatility and this indicates a unidirectional Granger causality running 
from Ghana cedi/dollar exchange rate return volatility to oil price volatility. 
Table 4. 5: Granger – Causality Results for the Pre – Crisis Period  
Null Hypothesis   df Chi-Sq. P-value Decision 
OILVOL does not Granger cause GHSVOL 4 1.0631 0.9001 Fail to reject 
GHSVOL does not Granger cause OILVOL  4 8.5888 0.0722 Reject 
OILVOL does not Granger cause NAIRAVOL 1 0.4044 0.5248 Fail to reject 
NAIRAVOL does not Granger cause OILVOL  1 0.3462 0.5563 Fail to reject 
OILVOL does not Granger cause RANDVOL 1 1.1614 0.2812 Fail to reject 
RANDVOL does not Granger cause OILVOL  1 0.0706 0.7904 Fail to reject 
OILVOL does not Granger cause EUROVOL 2 0.9486 0.6223 Fail to reject 
EUROVOL does not Granger cause OILVOL  2 1.3790 0.5018 Fail to reject 
OILVOL does not Granger cause RUBVOL 1 2.3931 0.3022 Fail to reject 
RUBVOL does not Granger cause OILVOL  1 1.2789 0.5276 Fail to reject 
OILVOL does not Granger cause RUPVOL 1 0.8768 0.6451 Fail to reject 
RUPVOL does not Granger cause OILVOL  1 2.3924 0.3023 Fail to reject 
 
57 
 
University of Ghana http://ugspace.ug.edu.gh
Also, the Granger – causality between oil price returns volatility and Nigeria naira/dollar 
exchange rate returns volatility is the second item reported in Table 4.5. The results show that at 
any conventional significant level, there is enough evidence to conclude that oil price returns 
volatility does not Granger-cause Nigeria naira/dollar exchange rate returns volatility over the 
pre-crisis period. On the other hand, it is also observed that at any conventional level, the Nigeria 
naira/dollar exchange rate returns volatility does not Granger – cause oil price volatility period. 
 
Similarly, at any conventional level of significance, South Africa rand/dollar volatility, 
Euro/dollar return volatility, Russia ruble/dollar volatility, and India rupee/dollar volatility each 
was documented not to Granger – cause oil price volatility over the pre-crisis period 
respectively. Also, oil price volatility does not Granger – cause South Africa rand/dollar 
volatility, Euro/dollar return volatility, Russia ruble/dollar volatility and India rupee/dollar 
volatility respectively. These results are consistent with the findings of other studies in literature 
such as Ding and Vo (2016), where they concluded when the markets are relatively calm (before 
the crisis), there is no volatility interaction between these two markets. 
   
4.7.2 The Granger – Causality Results for Post-Crisis Period 
The results of the Granger causality among the variables of study for the post-crisis are depicted 
in Table 4.6 below. Considering the first item reported, the null hypothesis that oil price returns 
volatility does not Granger – cause Ghana cedi/dollar exchange rate returns volatility at any 
conventional level (10%, 5%, and 1%) of significance cannot be rejected hence the oil price 
returns does not Granger – cause exchange rate returns volatility of Ghana. However, at 5% 
significant level, there is enough evidence statistical to reject the null hypothesis that Ghana 
58 
 
University of Ghana http://ugspace.ug.edu.gh
cedi/dollar exchange rate returns volatility does not Granger – cause oil price returns volatility 
over the post-crisis period. 
Table 4. 6: Granger – Causality Results for the Post – Crisis Period  
Null Hypothesis    df Chi-Sq. P-value Decision 
OILVOL does not Granger cause GHSVOL 8 9.5005 0.3018 Fail to reject 
GHSVOL does not Granger cause OILVOL  8 15.5196 0.0498 Reject 
OILVOL does not Granger cause NAIRAVOL 1 15.6259 0.0001 Reject 
NAIRAVOL does not Granger cause OILVOL  1 1.1517 0.2832 Fail to reject 
OILVOL does not Granger cause RANDVOL 8 8.9025 0.3506 Fail to reject 
RANDVOL does not Granger cause OILVOL  8 26.8245 0.0008 Reject 
OILVOL does not Granger cause EUROVOL 1 0.2885 0.5912 Fail to reject 
EUROVOL does not Granger cause OILVOL  1 0.1337 0.7147 Fail to reject 
OILVOL does not Granger cause RUBVOL 6 5.4019 0.4934 Fail to reject 
RUBVOL does not Granger cause OILVOL  6 32.3853 0.0000 Reject 
OILVOL does not Granger cause RUPVOL 1 0.1406 0.7077 Fail to reject 
RUPVOL does not Granger cause OILVOL  1 1.1801 0.2773 Fail to reject 
 
This implies there is unidirectional Granger causality running from the exchange rate returns 
volatility over the post-crisis period. 
  
The Granger causality result between Nigeria naira/dollar exchange rate returns volatility and oil 
price returns volatility is the second item depicted in Table 4.6. It is observed that at 1% 
significant level, there is enough statistical evidence to reject the null hypothesis that oil price 
volatility does not Granger – cause Nigeria naira/dollar exchange rate returns volatility and oil 
price returns volatility. This implies that oil price volatility granger – cause Nigeria naira/dollar 
exchange rate returns volatility and oil price returns volatility over the post-crisis period.  
However, the null hypothesis that Nigeria naira/dollar exchange rate returns volatility and oil 
price returns volatility does not Granger – cause oil price returns volatility cannot be rejected at 
59 
 
University of Ghana http://ugspace.ug.edu.gh
any conventional significance level. This implies that there is a unidirectional causality running 
from oil price returns volatility to Nigeria naira/dollar exchange rate returns volatility. This 
unidirectional causality between from oil price volatility to the Nigeria naira/dollar exchange rate 
return volatility contradicts the findings of Salisu and Mobolaji (2013), where they found 
bidirectional returns and volatility interrelations between Nigeria naira exchange rate and oil 
markets. 
  
Considering South Africa rand/dollar exchange rate returns volatility and oil price returns 
volatility, it is observed from the third item of Table 4.6 that, the null hypothesis of oil price 
returns volatility, not Granger – cause South Africa rand/dollar exchange rate returns volatility 
cannot be rejected at any conventional level of significance. However, at 1% significance level, 
there is enough statistical evidence to reject the null hypothesis that South Africa rand/dollar 
exchange rate returns volatility does not Granger-cause oil price returns volatility. Hence, there 
is unidirectional causality between South Africa rand/dollar exchange rate returns volatility and 
oil price returns volatility running from South Africa rand/dollar exchange rate returns volatility 
to oil price volatility. 
 
The results in Table 4.6 shows that there is no statistical evidence of Granger – causality existing 
between oil price returns volatility and euro/dollar exchange rate returns volatility and oil price 
returns volatility and India rupee/dollar exchange rate returns volatility over the post-crisis 
period. In both cases, the null hypotheses cannot be rejected at any conventional significance 
level. 
 
60 
 
University of Ghana http://ugspace.ug.edu.gh
Lastly, considering the causality between oil price return volatility and Russia ruble/dollar 
exchange rate returns volatility, the result in Table 4.6 reveals a unidirectional Granger causality 
running from Russia ruble/dollar exchange rate returns volatility to oil price returns volatility at 
1% significant level after the global financial crisis. 
The study revealed that in the pre-crisis period there were no impact interrelations between oil 
price volatility and the currency exchange rate volatility of the major oil producing economies 
and oil importing economies (Russia, Nigeria, India and the European Union). However, only 
Ghana cedi/dollar exchange rate volatility impacted oil price volatility at lag 1 and at 10% 
significant level. 
In the post-crisis period, the study revealed significant impact interrelations between oil price 
volatility and the currency exchange rate volatility of the major producing economies (Russia 
and Nigeria) as well as economies that are neither major oil exporters nor oil importers (Ghana 
and South Africa). Interestingly, such a relationship did not exist for economies that are 
considered as major oil importers in this study (India and the European Union) 
Granger Causality test showed some variable volatility results. Ghana cedi/dollar exchange rate 
volatility causes oil price volatility but not vice versa in both the pre – and post-crisis period. 
Also, there was unidirectional causality running from the Russia ruble exchange volatility and 
the South Africa rand to oil price volatility. However, oil price volatility granger caused the 
Nigeria naira/dollar exchange rate volatility in the post-crisis period. 
 
4.8 Chapter Summary 
The findings indicated a very significant impact relationship between oil price volatility and 
exchange rates volatility of major oil exporters considered in the study such as Russia and 
61 
 
University of Ghana http://ugspace.ug.edu.gh
Nigeria in the post crisis period. However, such an impact relationship was not significant in the 
period prior to the start the of the global crisis. For the neutral oil-dependent economies such as 
Ghana and South Africa, the fall in oil price over the post-crisis period could be an advantage or 
disadvantage to the value of their currencies   in the exchange rate market. As oil price fell, they 
are propelled to import more oil which increases their current account and consequently, 
increased the value of their local currencies against the US dollar and over that same period, the 
fall in oil price can negatively impact the dollar supply into their economy since they also export 
oil and in turn, increases the dollar demand pressure in these economies which consequently, 
increased the US dollar value against their local currencies. Thus, such economies can be in an 
equilibrium state in a condition of high oil price volatility. 
Test of Granger causality revealed some results of the link between the volatilities of the 
exchange rate and oil price in both subsample periods. The study revealed in the post-crisis 
period  that; currencies exchange rate volatility of the major oil-producing countries granger 
caused oil price volatility even though such a causal relationship never existed in the pre-crisis 
period. Intuitively, these economies, especially Russia, have clout in the oil industry and 
decisions taking by such oil producing economies to protect revenue from oil exports over the 
period when oil price is fallen could affect oil prices in the industry over that period. These 
currencies include the Russia ruble and the Nigeria naira. However, the currencies exchange rate 
volatility of the neutral economies (Ghana and South Africa) granger cause oil price volatility in 
both subsample periods. This confirms the neutral behaviour of the value of those economies 
currencies against the rise and fall of oil prices and moreover, previous information in the 
exchange rate volatility of those currencies can cause changes in oil price irrespective of the 
condition. 
62 
 
University of Ghana http://ugspace.ug.edu.gh
CHAPTER FIVE 
SUMMARY, CONCLUSION AND RECOMMENDATIONS 
 
5.1 Introduction 
The study assessed the link between currencies’ exchange rates returns volatility in oil-dependent 
nations and oil price returns volatility. The purpose of this chapter is to summarize the previous 
chapters as well as to make recommendations to policymakers and future researchers. 
  
5.2 Summary  
The focus of this study was the examination of inter-relation impact between exchange rate 
volatility and oil price volatility and causality pattern between them. This research was 
necessitated because prior studies that examined the volatility interrelation of these two markets 
focused on currency index which neutralises the idiosyncratic features of each currency that may 
arise because of exogenous shock (for instance, volatility of oil price) in the foreign exchange 
market. Also, most empirical evidence documented about the causality nexus between these two 
markets focused on first moments (returns) to the neglect of the common information factor 
(volatility effect). This results in inability to capture all available information in the dynamics of 
these two markets and hence, end up in misleading results. The contribution made by this study 
is twofold. First, the volatilities of these two markets were estimated using nonlinear approach 
(Generalised ARCH proxies) and accordingly, the volatility impact between each currency's 
exchange rate and oil price was further examined. Secondly, the study also examined the 
Granger causality relationship between these two markets with a focus on their second moments 
63 
 
University of Ghana http://ugspace.ug.edu.gh
(volatility) as a way of contributing to literature because most prior empirical studies focused on 
the causality between these two markets in terms of the first moments (return). 
Based on the objectives of this study, a random procedure model designed to capture the linear 
interdependence between oil price returns volatility and exchange rate returns were employed to 
examine the lead-lag nexus between these two variables. Ghana, Nigeria, South Africa, Europe, 
Russia, and India were the six main countries focused on for this study due to the rate of oil 
dependency of these countries and their local currency peg to the value of the US dollar in the 
exchange market and the availability of the data to the researcher. Thus, the Ghana Cedi/USD 
exchange rate returns volatility, Nigeria Naira/USD exchange rate returns volatility, South Africa 
Rand/USD exchange rate returns volatility, Euro/USD exchange rate returns volatility, Russia 
Ruble/USD exchange rate returns volatility and India Rupee/USD exchange rate returns 
volatility and oil price returns volatility were the variables employed for this study. The study 
employed weekly data from 2000 to 2016. However, this sample was sub-divided into pre-
financial crisis subsample and post-financial crisis subsample. The period between 2000 and 
2007 was considered as the pre-financial crisis period whiles 2010 to 2016 was considered as the 
post-financial crisis period. It is worth stating that 2008 to 2009 was considered as Global 
Financial Recession period and the division of the sample into subsample periods depicts the 
malfunction in market valuations because of the 2008 – 2009 Global Financial Recession 
(Mensah et al., 2017). To achieve the aim of the study, the unit root test was conducted and the 
optimal lag selection was also performed before the VAR model was estimated to ensure valid 
results.  
  
The empirical pieces of evidence obtained revealed that over the pre-crisis period, only the 
Ghana cedi/dollar exchange rate returns volatility at lag one had a significant impact on the 
64 
 
University of Ghana http://ugspace.ug.edu.gh
volatility of oil price returns. The remaining exchange rate returns’ volatility over the pre-crisis 
period was found to be autocorrelated.  However, over the post-crisis period, the Ghana 
cedi/dollar exchange rate returns volatility at lag six and eight were observed to have 
significantly impacted the volatility of oil price returns. Also, oil price returns volatility at lag 
two was documented to have significantly impacted the volatility of Ghana cedi/dollar exchange 
rate returns. Similarly, oil price returns volatility at lag one was documented to have significant 
impact on Nigeria naira/dollar exchange rate returns volatility over the post-crisis period. The 
South Africa rand/dollar exchange rate returns volatility at lag five, six, seven and eight were 
observed to have significantly impacted the volatility of oil price return whiles oil price returns 
volatility at lag one was observed to have significant impact on South Africa rand/dollar 
exchange rate returns volatility. The Russia ruble/dollar exchange rate returns volatility at lag 
three, four, five and six were documented to have an impact on oil price volatility whiles oil 
price returns volatility at lag five was observed to have a significant impact on the volatility of 
Russia rubble/dollar exchange rate returns.  
 
The secondary objective of the research was to investigate the Granger causality linking the 
variables. The empirical results revealed that over the pre-crisis period, Ghana cedi/dollar 
exchange rate returns volatility was found to have Granger – caused the oil price returns 
volatility. Similarly, the Ghana cedi/dollar exchange rate returns volatility was found to Granger 
– caused the oil price returns volatility over the post-financial crisis period. Oil price returns 
volatility was also documented to Granger – cause Nigeria naira/dollar exchange rate returns 
volatility over the post-crisis period. Lastly, South Africa rand/dollar exchange rate returns 
volatility and Russia ruble/dollar exchange rate returns volatility were documented to Granger – 
Cause oil price returns volatility in the period after the 2007-2009 financial recession.   
65 
 
University of Ghana http://ugspace.ug.edu.gh
5.3 Conclusion 
The study revealed evidence of a significant interrelation impact between oil price volatility and 
exchange rates volatility, especially in the post crisis period. The interrelation impact was 
particularly evident for currencies’ exchange rates of both oil-importing and major oil-producing 
economies that depend mostly on crude oil for industrial activities and revenue. These economies 
include Russia, Nigeria, Ghana and South Africa.  
In addition, the study revealed evidence of some causality link between oil price volatility and 
exchange rates volatility, particularly in the post crisis period. More specifically, a Granger 
causality link is found to exist between oil price volatility and the dollar exchange rate volatility 
for the currencies of Russia, Nigeria, Ghana and South Africa. The test of Granger Causality 
confirmed the findings of the interrelation impact that exist between the volatilities of these 
currencies and oil price in the post crisis period.  
Interestingly, this post crisis interrelation impact and causality link did not exist between the 
volatilities of oil price and the currencies of these major oil-dependent economies in the pre-
crisis period. 
 
5.4 Recommendations  
The study revealed that oil price volatility significantly impacted variations in the exchange rates 
of major oil exporting economies, especially in the period after the 2008 global financial 
recession. This result should draw the attention of policymakers of oil exporting countries to 
acknowledge that, depreciation of their local currencies against the US dollar could be because 
of their overdependence on revenue from oil exports for their economic growth, particularly in 
the period after the market has experienced a downturn. 
 
66 
 
University of Ghana http://ugspace.ug.edu.gh
Policymakers and government officials should take a clue from this result and implement 
measures to boost their exchange rates to be able to withstand external shocks from oil price 
volatility. The researcher also recommends major oil exporting countries such as Russia and 
Nigeria not to entirely depend on their natural resources (crude oil) for revenue but rather they 
should develop a successful diversification plan that includes at least partially replaced of oil 
exports by the exports of other products. Additionally, they should expand other tradable 
industries such as the gold industry. 
 
Moreover, the study revealed that exchange rate returns volatility of oil exporters (Russia and 
Nigeria) and neutral economies (i.e. neither major oil exporters nor importers such as Ghana and 
South Africa) significantly impacted variations in oil price, particularly in the post-crisis period. 
This information is key for financial managers and investors in oil-dependent economies since 
they can apply such information to rebalance their portfolio and diversify it to include crude oil 
and other tradable goods that are affected by oil price volatility. 
 
Further studies can consider other macroeconomic variables such as GDP, inflations, interest 
rate, unemployment as control variables in the model to examine if oil price volatility is the 
proximate cause of exchange rate volatility of oil-dependent countries. In addition, further 
studies can use nonlinear modelling technique such as Hiemstra & Jones (1994) nonlinear 
causality test to exploit the nonlinear relationship between these variables. Finally, other 
researchers should verify these results for major oil-dependent countries but are not USD – 
peggers in the foreign exchange market. 
 
67 
 
University of Ghana http://ugspace.ug.edu.gh
REFERENCES 
Akaike, H. (1973). Information theory and an extension of the maximum likelihood principle.
 In B.N. Retrov, F. C.Saki (Eds), Proceeding of second international symposium on 
 Information theory Budapest (pp. 267-281). Akademiai Kiado. 
Akram, F.Q. (2009). Commodity prices, interest rates and the dollar. Energy Economic, (31), 
838-851. 
Amano, R., & Norden, S.V. (1998). Oil price and the rise and fall of the US real exchange
 rate. Energy Policy, 17, 299 – 316. 
Asimakopoulos, I., David, A., & Wan, M. M. (2000). Nonlinear Granger causality in the
 currency future returns. Economic Letter, 68, 25- 30. 
Backus, D.K., & Crucini, M.J. (2000). Oil prices and the terms of trade. Journal International
 Economics, 50, 185-213. 
Bacon, R.W. (1991). Rockets and feathers: the asymmetric speed of adjustment of retail
 gasoline prices to cost changes. Energy Economic, 13, 211 – 218.  
Bal, D.P., & Rath, B.N. (2015). Nonlinear causality between crude oil price and exchange
 rate: A comparative study of China and India. Energy Economic Model, 51, 149-156. 
Barsky, B., & Kilian, L. (2004). Economic growth: Determinant, Issue and Lessons.
 Bangladesh: Chukizak printings. 
Barsky, R.B., & Kilian, L. (2001). Do we really know that oil caused the great stagflation? A
 monetary alternative. NBER/Macroeconomics Annual 16(1), 137 - 183. 
68 
 
University of Ghana http://ugspace.ug.edu.gh
Basher, S.A., Hang, A.A., & Sadorsky, P. (2012). Oil prices, exchange rates and emerging
 stock markets. Energy Economics, 34 (1), 227-240 
Beak, E., & Brock, W.A. (1992). A general test for nonlinear Granger causality: bivariate model.
 Working Paper lowa State University of Wisconsin, Madison. 
Bell, D., Kay, J., & Malley, J. (1996). A non – parametric approach to non – linear causality
 testing. Economics Letter, 51, 7 – 18. 
Benassy – Quere, A., Valerie, M., & Alexis, P. (2007). China and relationship between oil
 price and the dollar. Energy Policy 35, 5795 – 5805. 
Benhmad, F. (2012). Modelling nonlinear Granger causality between the oil price and U. S
 dollar: A wavelet - based approach. Economic Model, 29, 1505 – 1514. 
Bernanke, B.S., Gertler, M., & Watson, M. (1997). Systematic Monetary policy and effects
 of oil price shock. Bookings papers on economic activity, 1: 91 – 157. 
Betts, C., & Devereux, M.B. (1996). The exchange rate in a model of pricing – to – market.
 European Economic Review, 40, pp, 1007 – 1021. 
Blanchard, O.J., & Gali, J. (2007). The Macroeconomic Effect of oil shocks: Why are the 2000’s
 so different from the 1970? National Bureau of Economic Research,  
(No. W13368). 
Bloomberg, S.B., & Harris, E.S. (1995). The commodity – consumer price connection: fact or
 fable? Federal Reserve Board of New York. Economic Policy Review, 21 – 38. 
Bohi, D.F. (1989). The economic need for flexibility. Natural gas 5, issue 8. 
69 
 
University of Ghana http://ugspace.ug.edu.gh
Bollerslev, T. (1986). Generalized autoregressive conditional heteroskedasticity. Journal
 Economic 52, 5-59. 
Britton, A.J.C. (1970). The dynamics stability of the foreign exchange market. The Economic
 Journal 317, 91 – 96.  
Brooks, C. (2008). Introduction to econometrics for finance, second edition. The ICMA centre,
 University of Reading. 
Brown, S.P.A., & Philips, K.P. (1984). The effect oil prices and exchange rates on world oil 
 consumption. Federal Reserve Bank Dallas Economic Review, 13 – 21. 
Brown, S.P., & Yucel, M.K. (2002). Energy prices and aggregate economic activities: an          
 interpretative survey. Q. Review Economic Finance 42(2), 193-208. 
Buetzer, S., Habib, M.M., & Stracca, L. (2012). Global exchange rate configurations: do oil
 shocks matter? Working Paper Series No. 1689. European Central Bank. 
Cantore, N., Antimiani, A., & Anciaes, P.R. (2012). Energy Price Shocks: Sweet and sour  
 consequences for developing countries. Overseas Development Institute, London. 
Cashin, P., Mohaddes, K., & Raissi, M. (2014). The differential effects of oil demand and supply
 shocks on the global economy. Energy Economic 44, 113-134. 
Chaudhuri, K., & Daniel, B.C. (1998). Long –run equilibrium real exchange rate and oil prices
 Economic Letter 58, 231 – 238. 
70 
 
University of Ghana http://ugspace.ug.edu.gh
Chen, S.S., & Chen, H.C. (2007). Oil prices and real exchange rates. Energy Econ. 29, 390
 404. 
Chen, Y., Raugarjan, G., Feng, J., & Ding, M. (2004). Analysing multiple nonlinear time series
 with extended Granger causality. Physical Letter A 324, 26 – 35. 
Chen, Y.C., & Rogoff, K. (2003). Commodity currencies. Journal International Economic
 60, 133-160. 
Cifarelli, G., & Pladino, G. (2010). Oil price dynamics and speculations: A multivariate
 financial approach. Energy Economic, 32, 363-372. 
Clark, P. (1973). A Subordinate Stochastic model with finct variance for speculative prices.  
 Econometrica, 41, 135-155. 
Coudert, V., Miguon, V., & Penot, A. (2007). Oil price and the dollar Energy Study. Review
 15(2) Article 3. 
Corden, W. M., & Neary, J.P. (1982). Booming sector and de-industrialization in a small open
 economy. Economic Journal, 92, 825-848. 
Cukrowski, J. (2004). Russia oil: the role of the sector in Russia’s economy. Post-communist 
Economic 16(3), 285 – 296. 
Del Negro, M., & Schorfheide, F. (2012). DSGE model – based forecasting. Handbook of
 Economic Forecasting 2, 57 – 140. 
71 
 
University of Ghana http://ugspace.ug.edu.gh
Diks, C., & Panchenko, V. (2005). A note on the Hiemstra – Jones test for Granger
 noncausality. Study Nonlinear Dynamic Economic 9, 1- 9. 
Ding, L., & Vo, M. (2012). Exchange rates and oil prices: a multi-variate stochastic volatility
 analysis. Q. Rev. Economic Finance 52 (1), 15-37 
Driskill, R., & McCafferty, S. (1980).  Speculation, rational expectations, and stability of the
 foreign exchange market. Journal of International Econometric 10, 91 – 102. 
Dvir, E., & Rogoff, K.S. (2010). Three Epochs of Oil. NBER, Cambridge, MA. 
Engle, R.F., & Granger, C.W.J. (1987). Co-integration and error correction: representation,
 estimation, and testing. Econometrica 55, 251-276 
Fratzscher, M., & Schneider, D., & Van Robays, I. (2014). Oil prices, exchange rates and asset
 prices. Working Paper Series No. 1689. European Central Bank. 
Ghosh, S. (2011). Examining crude oil price – exchange rate nexus for India during the period
 of extreme oil price volatility. Applied Energy 88, 1886 – 1889. 
Gisser, M., & Goodwin,T.H. (1986). Crude oil and the macroeconomy: test of some popular
 notions. Journal of Money Credit Banking 18, 95-103. 
Golub, S. (1983). Oil price and exchange rates. Economic Journal 93, 576 – 593. 
Granger, C.W.J. (1969). Investigating causal relation by econometric models and cross
 spectral methods. Econometrica 37 (3), 424-438. 
Groen, J.J.J., & Pesenti, P.A. (2010). Commodity prices, Commodity Currencies and global
 economic developments. NBER Working Paper 15743. 
72 
 
University of Ghana http://ugspace.ug.edu.gh
Grydaki, M., & Fountas, S. (2008). Exchange rate volatility and output volatility: A
 theoretical approach. Department of Economics, University of Macedonia,
 Thessaloniki, Greece. 
Hamilton, J.D. (1996). Time Series Analysis. Princeton University Press, Princeton, NJ. 
Hamilton, J.D. (2003). What is an oil shock? Journal of Econometrics, 113(2), 363-398. 
Hamilton, J.D., & Herrera, M. (2004). Oil Shocks and aggregate macroeconomic behaviour:
 the role of monetary policy: Journal of Money Credit Banking 36, 265-286. 
Hiemstra, C., & Jones, J.D. (1994). Testing for linear and nonlinear Granger causality in the
 stock price volume relation. Journal of Finance 49, 1639 – 1664. 
Hooker, C. (1986). Effect of oil price and exchange rate variations on government revenue in
 China. Journal of Economics, 2(1): 2 – 3. 
Huang, Y., & Guo, F. (2007). The oil price shocks on China’s real exchange rate. China.
 Economic Review 18, 403 – 416. 
Huntington, H.G. (1986). The US dollar and the world oil market. Energy Policy, 299-306. 
Indjehagopian, J.P., Lantz, F., & Simon, V. (2000). Dynamics of heating oil market prices in
 Europe. Energy Economics 22 (2), 225-252. 
Ito, K. (2010). The impact of oil price volatility on macroeconomic activity in Russia.
 Document Trab. Analysis of Economics 9(5), 1-10. 
Jawadi, F., Louhichi, W., Idi Cheffou, A., & Ameur, H.B. (2016). On Oil-US exchange rate
 volatility relationships: An intraday analysis. Economic Modelling 59, 329-334. 
73 
 
University of Ghana http://ugspace.ug.edu.gh
Kilian, L. (2009). A Comparison of the effects of the exogeneous oil supply shocks on the
 output and inflation in the G7 countries. Journal of the European Economic
 Association, 6(1), 78-121. 
Kristoufek, L. (2014). Leverage effect in energy future. Energy Economics 45, 1-9. 
Krugman, P. (1983a). Oil shocks and exchange rate dynamics. In: Frankel, J.A (Ed.)
 Exchange rate and International macroeconomics. University of Chicago Press,
 Chicago. 
Laursen, S., & Lloyd, A.M. (1950). Flexible exchange rates and the theory of employment.
 Review of economics and statistics 4, 281 – 299. 
Lee, K., & Ralti, R. (1995). Oil shocks and the macroeconomy: the role of price variability.
 Energy  Journal, 16(1), 39 – 56.  
Li, J. (2006). Testing Granger causality in the pressure of thresholds effects. International
 Journal of Forecast 22, 771 – 780. 
Lizardo, R.A., & Mollick, A.V. (2010). Oil price fluctuations and US dollar exchange rates.
 Energy Economics 32, 399 – 408. 
Mahrara, M. (2008). The asymmetric relationship between oil revenue and economic
 activities: the case of oil exporting countries. Energy Policy 36(3), 1164 – 1168. 
Mensah, L., Obi, P., & Bokpin, G. (2017). Cointegration test of oil price and US dollar
 exchange rates for some oil dependent economies. Research in International 
 Business and Finance 42, 304 - 311 
74 
 
University of Ghana http://ugspace.ug.edu.gh
Mork, K.A. (1989). Oil and the macroeconomy when the prices go up and down: An
 extension of Hamilton’s results. Journal of Political Economics 97(30), 740-744. 
Mork, K.A., Olsen, Ø., & Terje M. H. (1994). Macroeconomic responses to oil price
 increases and decreases in seven OECD countries. Energy Journal 15(4), 19 – 35. 
Muth, J. (1961). Rational expectations and the theory of price movements. Econometrica 29,
 315– 335. 
Nelson D.B. (1991). Conditional Heteroskedasticity in asset returns. A new approach.
 Econometrica 59(2), 347 – 770. 
Peguin – Feissolle, A., Strikholm, B., & Terasvirta, T. (2008). Testing the Granger non
 causality hypothesis in stationary nonlinear models of unknown functional form.
 CREATE Research Paper 2008 – 19. University of Aarhus, School of Economics and
 Management. Phan, D.H.B.,  
Peguin – Feissolle, A., & Terasvirta, T. (1999). A general framework for testing the Granger
 non-causality hypothesis. Stockholm School of Economics Working Paper Series in
 Economics and Finance. No.: 343. 
Phan, D.H.B., Sharma, S.S., Narayan, P.K. (2016). Intraday volatility interaction between crude  
oil and equity markets. Journal International Finance Market Institution. Money 40, 1-
13. 
 
Ramussen, T., & Roitman, A. (2011). Oil Shocks in a global perspective: are they really that
 bad? International Monetary Fund Working Paper, WP/11/194. 
75 
 
University of Ghana http://ugspace.ug.edu.gh
Raurava, J. (2004). The role of oil and the real exchange rate in Russia’s economy:A
 cointegration approach. Journal of Comparative Economics 32, 315 – 327. 
Razgallah, B., & Smimou, K. (2011). Oil price and the greenback: it takes two tango.
 Applied Financial Economics 44, 1-17. 
Reboredo, J.C. (2012). Modeling oil price and exchange rate co – movement. Journal of Policy
 Modelling 34, 419 – 440. 
Ross, S. (1989). Information and volatility: The no-arbitrage martingale approach to timing
 and resolution irrelevancy. Journal of Finance 44, 1-17. 
Sadorsky, P. (2000). The empirical relationship between energy future prices and exchange
 rates. Energy Economics 22, 253-266. 
Salisu, A.A., & Mobolaji, H. (2013). Modelling returns and volatility transmission between
 oil price and US-Nigeria exchange rate. Energy Economics 39, 169-176. 
Shaari, M.S., Hussain, N.E., & Rahim, H.A. (2012). The effect of oil price changes and
 exchange rate volatility on unemployment: Evidence from Malaysia. International
 Journal of research business and social science 4, 2147 – 4478. 
Sharma, S., & Narayan, P.K. (2015). Oil price and stock returns of consumers and producers
 of crude oil. International Finance Markets, Institutions and Money 34, 245 – 262. 
Sims, C.A. (1980). Macroeconomics and reality. Econometrica 48, 1-48. 
Skalin, J., & Teresvirta, T. (1999). Another look at Swedish business cycle. Journal of
 Applied Economics 14, 359 – 378. 
76 
 
University of Ghana http://ugspace.ug.edu.gh
Tantatape, B., Huauq, J.C., & Sissoko, Y. (2014). Crude oil prices and exchange rates:
 causality, variance decomposition and impulse response. Energy Economics 44, 407
 412. 
Tharkor, A.V. (2015). The financial crisis of 2007-2009: Why did it happen and what did we  
 learn? The review of Corporate Finance Studies 4, 155-205. 
Toucher, G., Pitts, M. (1983). The price variability-volume relationship on speculative
 markets. Econometrica, 51, 485-505. 
Turnovsky, J.S., Bhandari, J.S. (1982). The degree of capital mobility and stability of an open
 economy under rational expectations. Journal of money, credit and banking, 14, 
 303 – 326. 
Tuzova, Y., & Qayum, F. (2016). Global oil glut and sanctions: The impact on Putin’s Russia.
 Energy 90, 140 – 151. 
Uddin. G.S., Tiwari, A.K., Arouri, M., & Teulou, F. (2013). On the relationship between oil
 price and exchange rates: a wavelet analysis. Journal of Economic Modelling 35, 502
 507. 
Wang, Y., & Wu, C. (2012). Energy prices and exchange rates of U.S Dollar: further
 evidence from linear and nonlinear causality analysis: Journal of Economic Modelling
 29, 2289-2297. 
Yousefi, A., & Wirjanto, T.S. (2004). The empirical role of the exchange rate on the crude-oil
 price formation. Energy Economics 26 (5), 783-799. 
77 
 
University of Ghana http://ugspace.ug.edu.gh
Zhang, Y.J., Fan,.Y., Tsai, H.T,. & Wei, Y.M. (2008). Spillover effect of US Dollar exchange
 rate on oil prices. Journal of Policy Model 30, 973-991. 
Zhang, Y.J., & Wei, Y.M. (2010). The crude oil market and the gold market: evidence for     
 cointegration, causality and price discovery. Resource Policy 35, 168-177. 
 
 
  
 
 
 
 
 
 
 
 
 
 
 
 
 
78 
 
University of Ghana http://ugspace.ug.edu.gh
APPENDICES 
Appendix A: Pictorial Behaviour of the Returns for the Pre-and Post - Crisis Periods 
Panel A: Pictorial Behaviour of the Returns for the Pre – Crisis Period 
OILRET GHSRET
12 12
8
4 8
0
-4 4
-8
-12 0
-16
-20 -4
00 01 02 03 04 05 06 07  00 01 02 03 04 05 06 07  
NAIRARET RANDRET
8 10
8
4
6
4
0
2
-4
0
-2
-8
-4
-12 -6
00 01 02 03 04 05 06 07  00 01 02 03 04 05 06 07  
EURRET RUBRET
3 5
4
2
3
1
2
0
1
-1
0
-2
-1
-3 -2
00 01 02 03 04 05 06 07  00 01 02 03 04 05 06 07  
RUPRET
2
1
0
-1
-2
-3
00 01 02 03 04 05 06 07  
79 
 
University of Ghana http://ugspace.ug.edu.gh
 
Panel B: Pictorial Behaviour of the Returns for the Post – Crisis Period 
OILRET GHSRET
15
10
10
5
0
0
-10
-5
-10
-20
-15
10 11 12 13 14 15 16 10 11 12 13 14 15 16 
NAIRARET RANDRET
6 12
4
8
2
4
0
0
-2
-4
-4
-6 -8
10 11 12 13 14 15 16 10 11 12 13 14 15 16 
EURRET RUBRET
4 20
3
2 10
1
0 0
-1
-2 -10
-3
-4 -20
10 11 12 13 14 15 16 10 11 12 13 14 15 16 
RUPRET
6
4
2
0
-2
-4
10 11 12 13 14 15 16 
80 
 
University of Ghana http://ugspace.ug.edu.gh
Appendix B: Pictorial Behaviour of the Volatilities for the Pre-and Post - Crisis Periods 
Panel A: Pictorial Behaviour of the Volatilities for the Pre – Crisis Period 
 
RANDVOL
EUROVOL
4
1.4
3
1.2
1.0 2
0.8 1
0.6
00 01 02 03 04 05 06 07 0
00 01 02 03 04 05 06 07  
 
RUBVOL
GHSVOL
.8
8
.6
6
.4
4
2 .2
0 .0
00 01 02 03 04 05 06 07 00 01 02 03 04 05 06 07  
 
RUPEEVOL NAIRAVOL
1.2 8
6
0.8
4
0.4
2
0.0 0
00 01 02 03 04 05 06 07 00 01 02 03 04 05 06 07
 
OILVOL
6
4
2
00 01 02 03 04 05 06 07  
81 
 
University of Ghana http://ugspace.ug.edu.gh
 
Panel B: Pictorial Behaviour of the Volatilities for the Post – Crisis Period 
RANDVOL     
EUROVOL
4.0
1.6
1.4 3.5
1.2 3.0
1.0
2.5
0.8
2.0
0.6
0.4 1.5
0.2
10 11 12 13 14 15 16 1.0
10 11 12 13 14 15 16
 
 
RUBVOL
GHSVOL
10 12
8
8
6
4 4
2
0
0
10 11 12 13 14 15 16 10 11 12 13 14 15 16 
 
 
NAIRAVOL
RUPVOL
5
2.0
4
1.6
3
1.2
2
0.8
1
0.4
0
10 11 12 13 14 15 16
10 11 12 13 14 15 16 
OILVOL
8
6
4
2
0
10 11 12 13 14 15 16 
82 
 
University of Ghana http://ugspace.ug.edu.gh
Appendix C: Results of VAR Lag Selection for the Pre-Crisis Period 
Table A. 1VAR Lag Selection of Oil Volatility & Cedi/dollar Exchange Rate Volatility 
       
 Lag LogL LR FPE AIC SC HQ 
       
0 −1774.052 𝑁𝐴   21.15285  8.727529  8.747228  8.735325 
1 −1448.885  645.5406  4.364738  7.149312   7.208410∗   7.172699∗ 
2 −1444.843  7.984171  4.363852  7.149107  7.247603  7.188086 
3 −1443.579  2.485726  4.422924  7.162548  7.300444  7.217119 
4 −1435.141  16.50266   4.327547∗   7.140741∗  7.318034  7.210903 
5 −1433.431  3.327145  4.376577  7.151995  7.368687  7.237749 
6 −1430.428  5.813085  4.398159  7.156897  7.412988  7.258243 
7 −1424.165   12.06513∗  4.349613  7.145774  7.441264  7.262712 
8 −1423.698  0.895887  4.425912  7.163133  7.498022  7.295663 
       
Endogenous variables: OIL OL GHSVOL  
 
 * indicates lag order selected by the criterion 
 LR: sequential modified LR test statistic (each test at 5% level) 
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
 
Table A. 2 VAR Lag Selection of Oil Volatility & Naira/dollar Exchange Rate Volatility 
       
 Lag LogL LR FPE AIC SC HQ 
       
0 −1699.607 𝑁𝐴   14.67218  8.361707  8.381407  8.369503 
1 −1585.930   225.6783∗   8.559143∗   7.822754∗   7.881852∗   7.846141∗ 
2 −1582.569  6.640994  8.586044  7.825890  7.924386  7.864869 
3 −1582.172  0.778714  8.739489  7.843599  7.981494  7.898170 
4 −1581.146  2.007681  8.868192  7.858211  8.035505  7.928373 
5 −1580.266  1.711913  9.005323  7.873544  8.090236  7.959298 
6 −1579.716  1.065944  9.159427  7.890494  8.146585  7.991840 
7 −1578.904  1.564020  9.304267  7.906160  8.201650  8.023098 
8 −1577.398  2.884808  9.419317  7.918419  8.253308  8.050949 
       
Endogenous variables: OILVOL NAIRAVOL 
 * indicates lag order selected by the criterion 
 LR: sequential modified LR test statistic (each test at 5% level) 
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion 
 
 
 
    
83 
 
University of Ghana http://ugspace.ug.edu.gh
Table A. 3VAR Lag Selection of Oil volatility & Rand/dollar Exchange Rate Volatility 
       
  Lag L ogL L R  FPE  AIC  SC H Q 
       
0  −1414 .010 𝑁𝐴    3.605 742  6.958 282  6.977 981  6.966 077 
1 −907.2237  1006.102∗   0.304779∗   4.487586∗   4.546684∗   4.510974∗ 
2 −905.2225  3.953274  0.307788  4.497408  4.595905  4.536387 
3 −899.5123   11.22408  0.305214  4.489004  4.626899  4.543575 
4 −897.9514  3.052719  0.308896  4.500990  4.678284  4.571152 
5 −897.1906  1.480562  0.313856  4.516907  4.733599  4.602661 
6 −896.9239  0.516310  0.319673  4.535252  4.791344  4.636598 
7 −892.5595  8.407128  0.319108  4.533462  4.828952  4.650399 
8 −889.3961  6.062455  0.320432  4.537573  4.872462  4.670102 
       
 Endogenou s variables: OILV OL RANDVOL         
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
 
 
Table A. 4: VAR Lag Selection of Oil Volatility & Euro/dollar Exchange Rate Volatility 
       
  Lag  LogL  LR  FPE  AIC S C H Q 
       
 0   269.9592  𝑁𝐴    0.000916  −1.319996  −1.300260  −1.312185 
1  337.1208   133.3307∗  0.000671 −1.631137  −1.571930∗  −1.607704∗ 
2  341.6122  8.872116   0.000669∗  −1.633558∗ −1.534879 −1.594503 
3  344.8571  6.377968  0.000672 −1.629838 −1.491688 −1.575161 
4  345.4739  1.206205  0.000683 −1.613172 −1.435550 −1.542873 
5  346.6253  2.240352  0.000693 −1.599139 −1.382046 −1.513218 
6  348.3541  3.346911  0.000701 −1.587951 −1.331387 −1.486408 
7  348.5326  0.343757  0.000714 −1.569126 −1.273090 −1.451961 
8  349.7422  2.317894  0.000724 −1.555380 −1.219873 −1.422593 
       
 Endogenou s variables: OILV OL EURVOL         
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion 
 
 
 
 
    
84 
 
University of Ghana http://ugspace.ug.edu.gh
Table A. 5: VAR Lag Selection of Oil Volatility & Rubble/dollar Exchange Rate Volatility 
       
  Lag L ogL L R F PE A IC S C H Q 
       
 0  −412.8487  𝑁𝐴    0.026326   2.038569   2.058268   2.046364 
1  37.63892   894.3342∗   0.002934∗  −0.155474∗  −0.096376∗  −0.132086∗ 
2  38.66509  2.027123  0.002978 −0.140860 −0.042364 −0.101881 
3  40.28620  3.186463  0.003013 −0.129171  0.008725 −0.074600 
4  40.38174  0.186842  0.003071 −0.109984  0.067310 −0.039821 
5  40.65400  0.529803  0.003128 −0.091666  0.125027 −0.005912 
6  42.17023  2.935597  0.003166 −0.079461  0.176631  0.021885 
7  42.61713  0.860867  0.003222 −0.062001  0.233489  0.054937 
8  44.06583  2.776372  0.003263 −0.049464  0.285425  0.083066 
       
 Endogenou s variables: OILV OL RUBVOL        
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
 
 
 
 
Table A. 6: VAR Lag Selection of Oil Volatility & Rupee/dollar Exchange Rate Volatility 
       
  Lag L ogL L R  FPE  AIC S C H Q 
       
 0  −504.0064  𝑁𝐴    0.041203   2.486518   2.506217   2.494314 
1 −211.8906   579.9251∗   0.010001∗   1.070716∗   1.129814∗   1.094103∗ 
2 −210.2793  3.183002  0.010119  1.082454  1.180950  1.121433 
3 −206.6678  7.098803  0.010139  1.084363  1.222258  1.138934 
4 −206.2076  0.900145  0.010317  1.101757  1.279051  1.171920 
5 −205.7326  0.924175  0.010497  1.119079  1.335772  1.204834 
6 −204.9595  1.496860  0.010665  1.134936  1.391027  1.236282 
7 −202.7854  4.187893  0.010762  1.143909  1.439399  1.260846 
8 −202.0915  1.329878  0.010938  1.160155  1.495043  1.292684 
       
 Endogenou s variables: OILV OL RUPVOL        
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
       
 
 
 
85 
 
University of Ghana http://ugspace.ug.edu.gh
 
Appendix D: Results of VAR Lag Selection for the Post-Crisis Period 
 
Table A. 7: VAR Lag Selection of Oil Volatility & Cedi/dollar Exchange Rate Volatility 
       
  Lag  LogL L R F PE  AIC S C H Q 
       
0  −1857 .279 𝑁𝐴    342.6 836  11.51 256  11.53 595  11.52 190 
1 −1654.908  400.9819  100.3343  10.28426   10.35443∗  10.31227 
2 −1645.918  17.70286  97.28163  10.25336  10.37031  10.30005 
3 −1639.923  11.72833  96.08869  10.24101  10.40475  10.30637 
4 −1630.608  18.11115  92.97920  10.20810  10.41862   10.29214∗ 
5 −1628.373  4.318193  94.00304  10.21903  10.47633  10.32174 
6 −1622.664  10.95795  93.01688  10.20845  10.51253  10.32983 
7 −1619.810  5.444333  93.68322  10.21554  10.56641  10.35560 
8 −1611.169   16.37121∗   91.03529∗   10.18681∗  10.58445  10.34554 
       
E ndogenou s variables: OILV OL GHSVOL        
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
 
 
Table A. 8: VAR Lag Selection of Oil Volatility & Naira/dollar Exchange Rate Volatility 
       
  Lag L ogL L R  FPE A IC S C  HQ 
       
 0  −1231.676  𝑁𝐴    7.121437   7.638864   7.662255   7.648201 
1 −1202.278  58.25036   6.085117∗   7.481599∗   7.551772∗   7.509611∗ 
2 −1200.173  4.144345  6.156972  7.493334  7.610290  7.540021 
3 −1194.832  10.45077  6.106108  7.485030  7.648767  7.550392 
4 −1194.212  1.206950  6.235313  7.505954  7.716473  7.589991 
5 −1189.860  8.405895  6.221920  7.503780  7.761081  7.606492 
6 −1183.682  11.86027  6.138753  7.490289  7.794372  7.611675 
7 −1182.589  2.083444  6.250564  7.508292  7.859158  7.648354 
8 −1177.350   9.927252∗  6.203158  7.500618  7.898266  7.659354 
       
 Endogenou s variables: OILV OL NAIRAVOL        
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
 
86 
 
University of Ghana http://ugspace.ug.edu.gh
Table A. 9: VAR Lag Selection of Oil Volatility & Rand/dollar Exchange Rate Volatility 
       
  Lag L ogL L R  FPE A IC  SC  HQ 
       
 0  −1335.082  𝑁𝐴    13.50948   8.279146   8.302537   8.288483 
1 −983.1775  697.2721  1.567035  6.124938   6.195111∗   6.152951∗ 
2 −980.5102  5.252000  1.580026  6.133190  6.250145  6.179877 
3 −978.9844  2.985488  1.604432  6.148510  6.312248  6.213872 
4 −973.5786  10.51043  1.590550  6.139805  6.350325  6.223842 
5 −970.0724  6.773584  1.595459  6.142863  6.400164  6.245575 
6 −962.4542  14.62304  1.560167  6.120460  6.424543  6.241846 
7 −961.1112  2.561291  1.586121  6.136912  6.487777  6.276973 
8 −953.9001   13.66319∗   1.554990∗   6.117029∗  6.514676  6.275765 
       
 Endogenou s variables: OILV OL RANDVOL        
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
 
 
Table A. 10: VAR Lag Selection of Oil Volatility & Euro/dollar Exchange Rate Volatility 
       
  Lag  LogL  LR  FPE A IC  SC  HQ 
       
 0  −868.9472  𝑁𝐴    0.753601   5.392862   5.416253   5.402200 
1 −394.8199   939.4474∗   0.041012∗   2.481857∗   2.552030∗   2.509869∗ 
2 −394.2345  1.152630  0.041888  2.503000  2.619955  2.549687 
3 −390.3982  7.506256  0.041931  2.504014  2.667751  2.569376 
4 −390.1347  0.512328  0.042913  2.527150  2.737669  2.611187 
5 −388.6134  2.938956  0.043578  2.542498  2.799799  2.645210 
6 −386.3422  4.359576  0.044048  2.553203  2.857286  2.674589 
7 −385.3767  1.841434  0.044886  2.571992  2.922857  2.712053 
8 −383.6064  3.354122  0.045513  2.585798  2.983446  2.744535 
       
 Endogenou s variables: OILV OL EUROVOL        
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
 
 
 
87 
 
University of Ghana http://ugspace.ug.edu.gh
Table A. 11: VAR Lag Selection of Oil Volatility & Rubble/dollar Exchange Rate Volatility 
       
  Lag L ogL L R F PE  AIC  SC H Q 
       
 0  −1778.491  𝑁𝐴    210.3884   11.02471   11.04810   11.03405 
1 −1433.874  682.8311  25.53058  8.915630  8.985803  8.943642 
2 −1431.262  5.143626  25.75101  8.924223  9.041178  8.970910 
3 −1371.106  117.7052  18.18806  8.576506   8.740243∗   8.641868∗ 
4 −1366.935  8.108575  18.16915  8.575450  8.785970  8.659487 
5 −1364.762  4.198428  18.37627  8.586762  8.844063  8.689473 
6 −1355.236   18.28552∗   17.75872∗   8.552544∗  8.856627  8.673930 
7 −1354.271  1.840910  18.09642  8.571335  8.922200  8.711396 
8 −1352.840  2.709799  18.38780  8.587247  8.984895  8.745983 
       
       
Endogenous variables: OILVOL RUBVOL    
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
 
 
Table A. 12: VAR Lag Selection of Oil Volatility & Rupee/dollar Exchange Rate Volatility 
       
  Lag L ogL  LR  FPE A IC S C  HQ 
       
 0  −845.1607  𝑁𝐴    0.650394   5.245577   5.268968   5.254915 
1 −508.1858   667.6903∗   0.082749∗   3.183813∗   3.253986∗   3.211825∗ 
2 −505.6051  5.081441  0.083480  3.192601  3.309556  3.239288 
3 −502.1798  6.702227  0.083778  3.196159  3.359897  3.261521 
4 −501.0147  2.265108  0.085263  3.213714  3.424233  3.297750 
5 −500.5885  0.823427  0.087173  3.235842  3.493144  3.338554 
6 −498.8516  3.334071  0.088406  3.249855  3.553938  3.371242 
7 −496.8599  3.798261  0.089517  3.262291  3.613156  3.402352 
8 −493.0057  7.302694  0.089603  3.263193  3.660841  3.421930 
       
 Endogenou s variables: OILV OL RUPVOL        
 * indicates lag order selected by the criterion    
 LR: sequential modified LR test statistic (each test at 5% level)   
 FPE: Final prediction error     
 AIC: Akaike information criterion     
 SC: Schwarz information criterion     
 HQ: Hannan-Quinn information criterion    
 
88