International Journal of Energy Economics and

Policy
ISSN: 2146-4553

available at http: www.econjournals.com

International Journal of Energy Economics and Policy, 2018, 8(1), 144-158.

The Impact of Oil Price Volatility to Oil and Gas Company Stock
Returns and Emerging Economies

Veysel Ulusoy1, Caner Özdurak2*

Department of International Finance, Yeditepe University, Istanbul, Turkey, 2Department of Financial Economics, Yeditepe
1

University, Istanbul, Turkey. *Email: cozdurak@yahoo.com

ABSTRACT
In this paper, we examine the impact of oil price shocks on both selected companies and emerging markets. The novelties of this study can be described
as: (i) It also includes the recent oil price crisis compared to previous articles in this field, (ii) our study considers in details the oil and gas company
business acumen to explain the results of the econometric models which is not the case in previous studies, (iii) we also include the impact of oil price
volatility on emerging markets since oil prices have an importance as explanatory variable of exchange rate movements which makes out study a very
comprehensive one. As mostly preferred in many previous studies in this literature, we employed the exponential GARCH (EGARCH) estimation
methodology, we concluded that the volatility effect of a given shock to the oil prices and oil and gas company stock price returns are highly persistent
and the successive forecasts of the conditional variance converge to the steady state slowly. In addition, we also present the news impact curves which
indicate that the behavior of commodity prices and company stock prices react differently to bad and good news.
Keywords: Oil Prices, Time Series, Asymmetric Volatility, Stock Returns, Oil and Gas Companies, News Impact Curves
JEL Classifications: Q4, Q43

1. INTRODUCTION expected to be a hot topic again while renewables continue to

rise worldwide. Diminishing oil prices in the last years forced the
Crude oil prices fluctuated heavily in the past three decades and energy giants such as BP, Exxon, Shell, Total, Chevron etc. BP had
become more volatile than they were over the period from World to contend with collapse of crude oil prices while at the same time
War II to the early 1970s. Since oil prices are denominated in US paying out tens billions of dollars in compensation and clean-up
dollars, and so their fluctuations in domestic currencies depend costs caused of UK group’s 2010 oil spill in the Gulf of Mexico.
closely on the dollar exchange rates, oil and non-oil economies
are not affected by the oil price increases in the same manner. A recovery in the industry is crucial for those companies in an
environment where renewable energy grows substantially with the
High volatility and specialness in energy markets result in more support of Paris Agreement. However Royal Dutch and Chevron
challenging trade execution, larger capital requirements and better positioned themselves compared to BP and Exxon by
decreasing effectiveness of benchmark hedging compared to other investing aggressively amid the downturn. Shell acquired BG
asset classes. Major sources of energy are often discovered at Group of the UK for 35bn £ during the depths of the oil crash.
considerable distances from the locations of ultimate consumption This strong portfolio acquired from BG including energy assets
which created regional imbalances with consequences raging from in Australia and deep water oil fields of Brazil recovered the
international capital flows to geopolitical risks to the reliability Anglo-Dutch Company.
of energy supply.
On the other hand Exxon is interested in new asset acquisitions
With an enthusiast for fossil fuels in the white house and former and it struck deals worth up to 6.6bn £ to buy shale oil companies
head of Exxon Mobil as US secretary of state the oil industry is with drilling right on a large area of the Permian basin in New

144 International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Mexico. BP is also rebuilding however the oil price needed to persistent increases in global oil prices. Aloui et al. (2013) claim
cover investment and its dividend-to 60$ per barrel this year from that the negative relationship between the oil prices and the price
55$ at the end of last year. of dollar can be explained by the fact that oil is a hedge against
rising inflation and serves as a safe haven against growing risk.
The stabilization of oil prices is more important than price itself
since volatility makes it difficult to predict both for the major In the study of Lizardo and Mollick (2010), cointegration tests
players and the countries as well. Oil price dynamics influenced and forecasts show that increases in real oil prices lead to a
economic activity, equity markets and the strategies of the significant depreciation of the USD dollar against currencies of
energy companies. In this context modeling and forecasting the net oil exporting countries (Canada, Mexico and Russia). On
comovements between oil priced and the dollar exchange rates the other hand the value of dollar relative to currencies of net
are crucial. oil importing countries such as Japan increases when the real
oil prices go up.
The paper is organized as follows: Section 2 includes the literature
review on previous research on the interaction between oil prices Moreover, it is documented that oil shocks may have an
and exchange rates along with macroeconomic implications of oil asymmetric impact on macroeconomic variables. Federer (1996)
price shocks. Section 3 describes the empirical methodology and and Lee et al. (1995) have found that changes in oil price volatility
Section 4 presents the data used and we discussed our empirical significantly affect macroeconomic variables.
results in Section 5. Finally, section 6 provides conclusion remarks
and further study areas within this topic. After more than two decades of research on volatility forecasting,
there is still considerable disagreement on how volatility should
Results show that the volatility of a given shock to the oil prices be modeled. One respectful example of volatility forecasting is
and oil and gas company stock prices are highly persistent and the observation that equity returns and volatility are negative
the successive forecasts of the conditional variance converge to correlated. The phenomenon can be explained by a leverage
the steady state slowly. The news impact curve (NIC) indicates effect, or a volatility feed-back effect. Takaishi (2017) propose
that the behavior of commodity prices and company stock prices a new ARCH-type model that uses a rational function to capture
react differently to bad and good news. the asymmetric response of volatility to returns, which is leverage
effect. Coherently, we also included analysis to find out the effect
2. LITERATURE REVIEW of shocks on stock returns of the major industry players in to this
study.
Before global financial crisis, there was a positive relationship
between oil price prices and dollar value. Chen and Chen (2007) In addition to macroeconomic impact, commodity prices such
studied the long run relationship between real oil prices and real as oil have significant effects of company stock returns. Jorion
exchange rates and concluded that world oil prices constitute the (1990) estimates exchange rate exposure of US multinationals
dominant source of exchange rate movements. Narayan et al. over the period from January 1971 to December 1987. Blose and
(2008) examined the relationship between oil prices and the Fiji- Shieh (1995) examine the impact of gold prices’ changes on the
US exchange rate and concluded that a rise in oil prices leads to returns of gold mining stocks. Due to their findings the gold price
an appreciation of the Fijian-dollar. Krugman (1983) and Golub sensitivity of a mining stock was found to be greater than one.
(1983) document the potential importance of oil prices as an The hypothesis of unity gold price sensitivity was not rejected
explanatory variable of exchange rate movements. Kang et al. using monthly data over the period 1981–1990 for a sample of
(2015) examine the effects of global oil price shocks on the stock commonly traded companies. Those studies guide us to analyze
market return and volatility contemporaneous relation using a the impact of oil price volatility on emerging market currencies to
structural VAR model which they conclude that the spillover index understand the macroeconomics aspect of energy price movements
between the structural oil price shocks and covariance of stock since for most of those countries it is the most important input of
return and volatility is large and highly statistically significant. the whole economics activity.

Coherently, Ratti and Vespignani (2016) state that global money, Due to the results of the previous literature there is a clear
global industrial production and global oil prices are cointegrated. asymmetric behavior between oil prices and other assets classes
A rise in oil prices result in significant increases in global interest like company equities and currencies. Also since the effect of oil
rates. Causality goes from global liquidity to oil prices and from price shocks can be persistent for a long time period there are
oil prices to the global interest rate, global industrial production cyclical impacts on both microeconomics and macroeconomics
and global CPI. Positive shocks to global M21, to global CPI and indicators. In this respect one of the crucial points of this study
to global industrial production lead to statistically significant and is that it includes the recent oil price crisis period in the dataset.
Narayan and Narayan (2007) paper appears to be the only notable
1 M2 is a measure of the money supply that includes all elements of M1 as paper that has attempted to model oil price volatility using different
well as “near money.” M1 includes cash and checking deposits, while near sub periods in order to judge the robustness of their results. This
money refers to savings deposits, money market securities, mutual funds
and other time deposits. These assets are less liquid than M1 and not as
is the main reason why we will also use three sub periods in our
suitable as exchange mediums, but they can be quickly converted into cash analysis which will cover both 2008 global crisis and 2014 oil
or checking deposits. price crisis.

International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018 145
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

3. METHODOLOGY Figure 1: News impact curves

Firstly, we used exponential GARCH (EGARCH) instruments to

model the volatility behavior of oil prices. Major advantage of
the model is that, instead of considering heteroskedasticity as a
problem to be corrected, ARCH and GARCH models treat it as a
variance to be modeled. Usually financial data suggests that some
time periods are riskier than others; that is, the expected value of
the magnitude of error terms at sometimes is greater than at others.
The goal of such models is to provide a volatility measure, like a
standard deviation, then can be used in financial decisions related
with risk analysis, portfolio selection and derivative pricing (Engle
1982; 1993; 2001).

ARCH model assumes that the variance of t ut in period t, σ2t

depends on the square of the error term in t−1 period, ut−1.
however, if the news is “good” (i.e., if ԑt is positive), volatility
In this context, ARCH(q) and GARCH(q) models are as follows;
increases along ab (or abᶦ for EGARCH model). If the news is
“bad,” volatility increases along ac (or acᶦ for EGARCH model).
α0>0, αi>0
Since ac and acᶦ are steeper than ab and abᶦ, a positive ԑt shock will
have a smaller effect on volatility than a negative shock of these
h t =α 0 +α1ε t −12 + α 2 ε t − 2 2 + ... + α q ε t − q 2 +ν t  (1)
same magnitude (Figure 1).
GARCH models which express the generalized form of ARCH
Asymmetric or leverage volatility models, in which good news
models were developed by Engle (1982) and Bollerslev (1986)
and bad news have different predictability for the future volatility
to provide reliable estimations and predictions. GARCH models
are the most interesting approaches in the literature. Overall, Chen
consist of conditional variance, in equation (2) in addition to
and Ghysels (2010) found that moderately good (intra-daily)
conditional mean in equation (1).
news reduces volatility (the next day), while both very good news
q p (unusual high intra-daily positive returns) and bad news (negative
h t =α 0 + ∑
i=1
αi r 2 t −i
+ ∑β h
i=!
j t− j  (2) returns) increase volatility, with the latter having a more severe
impact. The asymmetries disappear over longer horizons.

In this context, restrictions of variance model are as follows; The NIC characterizes the impact of past return shocks on the
return volatility which is implicit in a volatility model. In the
for αi≥0 and βi≥0, αi+βi<1 next sections we discuss several models of oil price and oil and
company stock prices volatility and present the NIC.
If αi+βi≥1 it is termed as non-stationary in variance. For non-
stationarity in variance, the conditional variance forecasts will Coherent with that GARCH models allow us to test the effect of
not converge on their unconditional value as the horizon increases news on oil prices quantitatively and help us to understand if the
(Brooks, 2008). markets absorb these closely tracked data by people. For most
financial assets, the distribution function for the rate of return is
In this context ARCH and GARCH models have become very fat-tailed. A fat-tailed distribution has more weight in the tails than
popular as they enable the econometrician to estimate the variance a normal distribution. Suppose that the rate of return on a particular
of a series at a particular point in time. Clearly asset pricing models stock has a higher probability of a very large loss (or gain) than
indicate that the risk premium will depend on the expected return indicated by the normal distribution. As such, you might not want
and the variance of that return (Enders, 2004). to perform a maximum likelihood estimation using a normal
distribution. Figure 2 below compares the standardized normal
An interesting feature of asset prices is that “bad” news seem to distribution to a t-distribution. You can see that the t-distributions
have a more pronounced effect on volatility than does “good” places a greater likelihood on large realizations than does the
news. For many stocks, there is a strong negative correlation normal distribution. As such, many computer packages allow you
between the current return and the future volatility. The tendency to estimate a GARCH model using a t-distribution.
for volatility to decline when returns rise and to rise when returns
fall is often called the leverage effect. Another model that allows for asymmetric effect of news is the
EGARCH model. One problem with a standard GARCH model
The idea of the leverage effect is captured in the Figure 1, where is that it is necessary to ensure that all of the estimate coefficients
“new information” is measured by the size of ԑt−1. If ԑt−1=0, are positive. Nelson (1991) proposed a specification that does not
expected volatility (ht) is 0a. Any news increases volatility; require nonnegativity constrains.

146 International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Figure 2: Normal distribution versus students-t distribution Consider EGARCH (1,1)

ln(ht)=α0+βln(ht−1)+α1zt−1+γ(|zt−1|)-E(|zt−1)|) (4)
εt
Where z t = σ t . The NIC is

 α1 +γ 
 
Aexp h t  for ε t −1 >0
ht =    (5)
Aexp α1 − γ  for ε t −1 <0
 h t 


A ≡ h βt exp α 0 − γ 2 / π   (6)

α1<0 α1+γ>0 (7)

Consider:
Although in our analysis GARCH and EGARCH model results
ε  ε did not differ from each other significantly we proceeded with
ln ( h t ) =α 0 +α1  +t −0.15  + λ1 t0−.51+β1ln (h t −1 )  (3)
 h t −1  h t −1 EGARCH models for NIC. Further details can be found in
Section 5.
Equation (3) is called the exponential-GARCH or EGARCH
model. There are three interesting features to notice about ECONOMETRIC DATA DESCRIPTION
EGARCH model:
1. The equation for the conditional variance is in log-linear The NYMEX WTI futures contract is one of the world energy
form. Regardless of the magnitude of ln (ht), the implied value benchmarks. The notional quantity for one contract is 1000 barrels,
of ht can never be negative. Hence, it is permissible for the which, as mentioned earlier, is one lot. As with all futures, trading
coefficients to be negative. for a given contract month ceases at a defined futures expiration
2. Instead of using the value of ε 2t−1 , the EGARCH model uses date prior to the contract month.
the level of standardized value of ε 2t−1 [i.e.,,  ε 2t−1 divided by
In the case of the WTI contract, this is roughly two-thirds of the
(ht−1)0.5]. Nelson argues that this standardization allows for a way through previous contract month. However, in the recent
more natural interpretation of the size and persistence of years the idiosyncrasies related to the delivery location of the
shocks. After all, the standardized value of ε 2t−1 is a unit-free WTI contract resulted in substantial and prolonged decoupling
measure. from global crude oil prices. As a result, despite complications of
3. The EGARCH model allows the leverage effects. If ε 2t−1 its own, the Brent futures contract which trades on ICE2 is now
(ht−1)0.5 is positive, the effect of the shock on the log of viewed as the dominant crude oil benchmark. The settlement and
conditional variance is α1+λ1. If ε 2t−1 (ht−1)0.5 is negative, the delivery mechanism of Brent contracts are more complex than
WTI futures. The Brent contract is described by the exchange
effect of the shock on the log of the conditional variance is physically settling with an option to settle financially on the
−α1+λ1. ICE Brent Index. However Salisu and Fasanya (2012) state that
their choice of the crude oil price is underscored by the fact that
Most financial decisions involve a trade-off between future WTI has remained dominant in the world oil market. The crude
risks and asset returns. The volatilities and correlations oil is either traded themselves or their prices are reflected in
of securities are often major components of risk. Second other types of crude oil. In this respect we also decided to use
moments evolve over-time as the economy changes and new WTI in our models however we also incorporated Brent in the
information is released. Volatilities and correlations measured same models instead of WTI and experienced no significant
from historical data may miss changes in risk unless methods result changes.
are carefully designed to update estimates rapidly (Cappiello
and Engle, 2006). Our dataset contains daily crude oil, hard currencies such as
Canadian dollar (CAD), Euro (EUR), Swiss Franc (CHF), UK
In the case of EGARCH models the NIC has its minimum at
ԑt−1=0 and is exponentially increasing in both directions but with 2 Intercontinental Exchange (traded as ICE) is an American business
different parameters. and finance company founded on May 11, 2000 by Jeffrey Sprecher,
headquartered in Atlanta, Georgia. It owns exchanges and clearing houses
for financial and commodity markets, and operates 23 regulated exchanges
The NIC are drawn by using the estimated conditional variances
and marketplaces. ICE futures exchanges in the United States, Canada and
equation for the model under consideration, with its given Europe, Liffe futures exchanges in the US and Europe, New York Stock
coefficient estimates and with the lagged conditional variance set Exchange, Equity options exchanges, OTC energy, credit and equity
to the unconditional variance. markets.

International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018 147
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Table 1: Model dataset descriptions

Variable Description Frequency
RWTI Returns** of NYMEX light sweet crude oil (WTI) closing prices Daily
RWTI (−1) 1 day lagged returns of NYMEX light sweet crude oil (WTI) closing prices Daily
RCAD Returns of USD Dollar/Canadian Dollar (CAD) closing prices Daily
REUR Returns of USD Dollar/Euro (EUR) closing prices Daily
RCHF Returns of USD Dollar/Swiss Franc (CHF) closing prices Daily
RGBP Returns of USD Dollar/UK Pound Sterling (GBP) closing prices Daily
RJPY Returns of USD Dollar/Japanese Yen (JPY) closing prices Daily
RDXY Returns of USD Dollar index* Daily
RMXN Returns of USD Dollar/Mexican Peso (MXN) closing prices Daily
RRUB Returns of USD Dollar/Russian Ruble (RUB) closing prices Daily
RBrent Returns of ICE brent crude electronic energy futures closing prices Daily
RBrent (−1) 1 day lagged returns of ICE brent crude electronic energy futures closing prices Daily
RTRY Returns of USD Dollar/Turkish Lira (TRY) closing prices Daily
RXOM Returns of Exxon Mobil Corp (XOM) stock closing prices Daily
RCVX.N Returns of Chevron Corp (CVX.N) stock closing prices Daily
RCOP.N Returns of Conoco Phillips (COP.N) stock closing prices Daily
RHES.N Returns of Hess Corp (HES.N) stock closing prices Daily
RMRO.N Returns of Marathon Oil Corp (MRO.N) stock closing prices Daily
RBP.L Returns of BP PLC (RBP.L) stock closing prices Daily
RDSa.AS Returns of Royal Dutch Shell PLC (RDSa.AS) stock closing prices Daily
RTOTF.PA Returns of Total SA (TOTF.PA) stock closing prices Daily
*The US Dollar Index is an index of the value of the United States dollar relative to a basket (57.6% Euro, 13.6% Japanese Yen, 11.9% pound sterling, 9.1% Canadian dollar, 4.2%
 x 
Swedish krona, 3.6% Swiss franc) of foreign currencies, often referred to as a basket of US trade partners’ currencies **returns are calculated as ln  t 
 x t−1 

Pound Sterling (GBP), Japanese Yen (JPY) as well as emerging Descriptive statistics and distributional characteristics of return
currencies such as such as Turkish Lira (TRY), Mexican Peso series are reported in Appendix Tables  2 and 3. The normal
(MXN), Russian Ruble (RUB) and dollar index (DXY) over distribution has a skewness of zero. But in reality, data points may
the period from January 4, 2000 to February 9, 20173 (Table1). not be perfectly symmetric. So, an understanding of the skewness
Furthermore, we have major industry players’ daily stock prices of the dataset indicates whether deviations from the mean are
which are Exxon Mobil, Chevron Corp, Conoco Phillips, Hess going to be positive or negative. The hard currency returns like
Corp, Marathon Oil Corp, BP, Shell and Total. Detailed business GBP and CAD are negative skewed which means that the left tail
descriptions of the mentioned companies are exhibited at Appendix is longer; the mass of the distribution is concentrated on the right
Table 1 in Appendices part. We computed the returns on crude of the figure. Emerging market currency returns like TRY and ARS
oil price indices, exchange rates and stock prices by taking the are positive skewed which means that the right tail is longer; the
difference in logarithm of the two successive daily prices. mass of the distribution is concentrated on the left of the figure.

At a glance all the currencies and oil prices fluctuate significantly The kurtosis of any univariate normal distribution is 3. It is
on 2008 global financial crisis as we can see in Figure  3. In common to compare the kurtosis of a distribution to this value.
addition we narrowed the period from September 15, 2008 to Distributions with kurtosis less than 3 are said to be platykurtic
February 9, 2017 which we will emphasize as “Global Financial which has thinner tails. It means the distribution produces fewer
Crisis Period” in our GARCH models. In Figure 4 after global and less extreme outliers than does the normal distribution.
financial crisis we can clearly observe that after from 2014 to Distributions with kurtosis greater than 3 are said to be leptokurtic.
present there is an increase in oil price return volatility (RBRent All the series in our dataset is highly leptokurtic which has fatter
and RWTI) as well as emerging market currencies go on to tails which is expected for financial assets.
fluctuate after 2008 crisis.
Thus we will also analyze the oil prices in a third sub-sample
Oil prices have fallen sharply since mid-2014 and reached a namely “oil price crisis” which includes the data between
10-year low in early 2016. From their peak in June 2014 to the November 1st, 2014 and February 7th, 2017. We will also analyze
trough in January 2016, Brent crude oil prices dropped by USD industry company stock prices in the same sub-period in order to
find out the effect of oil price volatility on company stock returns
82 per barrel (70%).
and their business strategies.
There are five key moments in oil price decline which are:
i. November 2014: OPEC decides not to cut output APPLICATIONS AND FINDINGS
ii. April 2015: Shell and Total delay west African projects
iii. January 2016: Brent hits 12 years low We applied all our models by using Brent instead of WTI and
iv. November 2016: OPEC agrees to reduce output any significant difference was not detected. The analysis for
v. December 2016: BP approves expansion of Mad Dog field. countries and company stock returns are exhibited in two separate
sub-sections in order to make the reader focus easier on the
3 Dataset is provided by Thomson Reuters Eikon. fundamental differences of results and NIC behaviors of the assets.

148 International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Figure 3: Return graph for overall period

Figure 4: Return graph for global financial crisis period

Country Implication Analysis In all EGARCH (1,1) models exhibited in Tables  3, Appendix
We present our results in Tables  2 and 3 by fitting Tables 5 and 6, λ1 is negative for all periods. This validates the
GARCH and EGARCH models with both normal and student-t conclusion that negative shocks have the tendency of reducing
distributions. The series were modeled by GARCH (1,1) and volatility more than positive shocks thereby suggesting asymmetric
EGARCH (1,1) satisfactory. Note that for all models the parameter effects in the volatility of crude oil prices.
β is close to 0,9 (even 1,0 in EGARCH model with student-t
distribution) highly significant which thus indicates that conditional Moreover, we included first lags of WTI, CAD, EUR, CHF,
volatility is past dependent and very persistent over time. GBP, JPY, DXY, MXN and RUB returns in the mean equation
for the all models. Russia and Canada are among top world
While Tables 2 and 3 exhibit models for overall period Appendix
oil producers while Switzerland is a net oil importer without
Tables 4 and 5 exhibit models for global financial crisis period in
domestic oil production. Japan’s current account balance and
which we see that the parameter β is still close to 0.9 and highly
reliance on nuclear energy help weakening the dependence
significant but slightly less than overall period models. The effect
of ԑt and past values of ԑt on yt is the effect of shocks which include of the Yen value on changes in the price oil even though this
news effect or extra ordinary days. country is one of the biggest oil importers following China, US,
India and South Korea. Mail EU countries such as Germany,
In Appendix Tables 6 and 7 models exhibit results for oil crisis period Italy, Netherland and France are also in the list of biggest oil
which shows that the effect of news and extra ordinary days increase importers.
compared to overall period and global financial crisis period. It
appears that there is a high level of persistence in the oil price In this context, EUR, CAD, RUB returns have a negative effect
volatility that may be associates with crisis such as 2008 and 2014. on WTI returns while JPY returns are expected to have a positive

International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018 149
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Table 2: RWTI GARCH model for overall period

Distribution Normal distribution Student t‑distribution
Mean equation Variance equation Mean equation Variance equation
Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats
RWTI (−1) −0.04 −2.69 −0.04 −3.17
RCAD −0.88 −14.29 −0.89 −14.67
REUR −0.52 −5.22 −0.40 −3.44
RCHF 0.00 0.00 0.08 1.27
RGBP 0.04 0.65 0.08 1.20
RJPY 0.29 6.66 0.25 5.32
RDXY −0.87 −6.95 −0.74 −4.84
RMXN −0.09 −2.02 −0.12 −2.73
RRUB −0.66 −21.75 −0.66 −18.28
α0 0.00 4.61
α1 0.07 14.00 0.06 8.31
β1 0.92 155.23 0.94 132.1
Observations 4266 4266
R2 0.177 0.178
DW 2.029 2.087

Table 3: RWTI EGARCH model for overall period

Distribution Normal distribution Student t‑distribution
Mean equation Variance equation Mean equation Variance equation
Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats
RWTI (−1) −0.03 −2.61 −0.04 −3.16
RCAD −0.89 −14.37 −0.90 −14.79
REUR −0.48 −4.78 −0.37 −3.26
RCHF −0.02 −0.40 0.06 1.02
RGBP 0.08 1.33 0.09 1.48
RJPY 0.26 6.01 0.24 5.19
RDXY −0.76 −5.74 −0.69 −4.50
RMXN −0.10 −2.35 −0.13 −2.84
RRUB −0.64 −20.03 −0.63 −17.22
α0 −0.21 −9.20 −0.16 −6.38
α1 0.15 15.17 0.12 8.70
λ1 −0.04 −6.68 −0.04 −4.58
β1 0.99 423.1 1.0 416.6
Observations 4266 4266
R2 0.180 0.178
DW 2.030 2.080

effect. Since Canada4 and Russia5 are oil producers and exporters findings the causality between CAD and oil prices is one way from
while Japan is an importer country, the signs of the coefficients are CAD to oil prices however it is just in the opposite way for RUB.
quite coherent with macroeconomics theory. Oil prices are traded Flexible exchange rates can provide a measure of protection to
as US dollar denominated. Therefore when Russia and Canada countries like Russia which mitigated some of the impact of low
export oil, RUB and CAD will come down since there will be US oil prices with fallen ruble: In dollar terms. Lower oil revenues
dollar inflow in to these markets while JPY will rise as there will are offset by cheaper domestic expenditures. Consequently since
be US dollar outflow from Japanese market. Mexico6 is an oil producer two way causality between crude oil
prices (Brent and WTI) and MXN is also relevant.
Wang et al. (2013) found that the magnitude, duration, and even
direction of response by stock market in a country to oil price In Appendix Table 9 results show that there is one-way causality
shocks highly depend on whether the country is a net importer or from crude oil prices (Brent and WTI) to TRY given by the fact
exporter in the world oil market, and whether changes in oil price that Turkey is an oil importing country. Oil prices increase pressure
are driven by supply or aggregate demand. In Appendix Table 8, over TRY since oil is traded as US dollar denominated. As Berk
we performed Granger Causality tests in order to understand the and Aydoğan (2012) suggest that the global financial liquidity
signs of coefficients better in the mean equation. Based on our conditions are the most plausible explanation for the changes in
Turkish stock market returns. There exists some evidence that
4 215.5 million tonnes in 2015 (4.9% of total production), BP Statistical purified oil price shocks still have an impact on stock market
Review of World Energy June 2016.
5 540.7 million tonnes in 2015 (12.4% of total production), BP Statistical 6 127.6 million tonnes in 2015 (2.9% of total production), BP Statistical
Review of World Energy June 2016. Review of World Energy June 2016.

150 International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

returns where this effect is smaller and less significant than the Figure 5: News impact curves
liquidity constraints.

Hamilton (1985) stated that a given oil price increases seems to

have a smaller macroeconomic effect after 1973 than an increase
of same magnitude would have had before 1973. The article
concluded with the statement: “The political history of the Middle
East makes it almost inevitable that sometime within the next
decade economists will be granted some more data with which
to assess the economic effects of oil supply disruptions.” This
is exactly what happened in 1990 when Iraq invaded Kuwait,
and surely this oil shock was a key factor in the recession that
followed (Hamilton, 1996). Considering the latest oil price crisis
and macroeconomic developments in the world we can conclude
that Hamilton’s statement is still valid.
impact of the linear specification of oil price on stock returns within
In Figure  57, the NIC of the EGARCH (1,1) is compared for 2 days after the shock and then a positive impact which is absorbed
overall and global financial crisis sub-periods. As we can see both within nearly one week. In all cases, the impulse responses revert
NIC are asymmetric, with negative shocks having more impact to zero usually within 6–8 days. The impulse response analysis was
on future volatility than positive shocks of the same magnitude. also tested with overall period data and any significant behavior
However, after 2008 global financial crisis we can see that NI on change was not detected.
volatility decreases.
NI curves of the major industry players’ differentiate from oil price
Company Analysis NI curves in a way that most of them are either close to symmetric
Appendix Table 10 exhibits models for oil crisis period in which or good news increase volatility more than bad news (Figure 7). NI
we see that the parameter β is below 0.9 and highly significant. curves of Exxon, Conoco Phillips and Shell are almost symmetric
The effect of ԑt and past values of ԑt on yt is the effect of shocks which shows that both good and bad news increase volatility in
which include news effect or extra ordinary days. Table shows the same way. For Chevron, BP and Total good news increase
that the effect of news and extra ordinary days in company stocks volatility more while for Hess Corp and Marathon Oil bad news
volatility forecast models has more effect compared to oil price have more impact to increase volatility of stock returns.
volatility forecast models.
NI curves of asymmetric ARCH-type models exhibit for BP stock
In most of the EGARCH (1,1) models exhibited in Appendix returns that the higher volatility response to negative stocks in the
Tables  10 λ1 is positive which validates that negative shocks oil prices period while it is the opposite for the oil prices period.
generate less volatility than positive shocks in company stock The tragic accident and oil spill in the Gulf of Mexico may be one
returns. Diaz et  al. (2016) investigate the role of real oil price of the key factors in this behavior change however further analysis
shocks on real stock returns of four oil and gas corporations8 can be provided by using proxy variables in the EGARCH variance
where they conclude that both linear and non-linear specifications equations to drill down more.
oil price changes have a positive significant impact on real stock
returns of these companies in the short-run. Exxon and Chevron react to the shocks equally based on their
NI curves which suggest that bad or good, any kind of news
Initially we display the impulse response functions using the oil increases the volatility of the stock returns significantly more in
price changes of WTI (Figure 6). In summary, we detect a negative the oil prices period compared to overall period. As of January
31st, 2017 Exxon and Chevron have stronger financial positions
7 We follow the EViews syntax and firs generate the conditional variance compared to Shell, Total, BP and others. Both companies focused
series (GARCH01). Next, we store the median by entering “scalar med
=@median(garch01)” in the command window, where GARCH01 is the on investing in higher return investments with shorter-cycles and
name of the conditional variance series produced in Table 2 and 3 above optimizing their costs via flexible capex programs which enabled
(If Eviews give a range error in this step use “pagestruct” command them to improve their cash flows in the oil crisis period.
before creating series z and make your data type “unstructed/unfitted”).
Third, we generate the z series, which is the x -axis amount of the news For Hess and Marathon, negative shocks have greater impact
impact curve, using the commands “smpl 1 100” and “series z = −10 + @
trend(1)*20/100,” which constructs an equispaced series between -10 and on volatility both in overall and oil prices period. Considering
10. Fourth, we generate the ht series using the variance equation in Table Shell NIC, we can conclude that while in overall period good
2-3 and the command “series log(SIG2) = eq01.c(3) + eq01.c(6)*log(med) news increase the volatility more in the oil crises period both
+ eq01.c(4)*abs(z) + eq01.c(5)*z , where SIG2 is the name for the ht series. bad and good news have a balanced impact on the volatility of
Finally, the EViews automatically creates the series SIG2 from the log stock returns.
specification. Highlighting the two series Z and SIG2 shows a customized
graph depicting the estimated news impact curves from EGARCH model
fitted to the oil prices and company stock prices. Finally for total we see that news effect changes for overall and
8 Exxon, BP, Chevron, Shell. oil crises period significantly where good news have a significant

International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018 151
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Figure 6: Impulse response analysis for oil and gas company stock returns

impact on volatility in the oil prices period. Total has a diversified commodity. First we analyzed the country specific models and
portfolio of gas developments both in downstream and upstream both EGARCH models and causality tests showed that based
and implements its strategy through portfolio management. on if the country is an oil exporter or importer, the magnitude
However, countries do not have such flexibilities to optimize and sign of the currency of the related country as an explanatory
their spending or changing the dynamics of macroeconomy and variable in oil price change compatible with macroeconomics
production schemes against oil shocks in order to adjust themselves theory.
like oil and gas companies can do.
We also showed that bad news increase volatility more than
We should also keep in mind that Shell and Total stocks are quoted bad news for oil prices which is coherent with the theory. It
in Amsterdam stock exchange and Euronext respectfully. Since is quite coherent with the theory since a slowdown in global
the exchange markets of the other companies are US, there will economy is likely to result in a further decline in crude oil prices.
be different systemic and unsystemic risks for the stocks that can The view expressed in Hamilton (1998a, b) is that oil shocks
affect the returns and volatiles rather than oil price fluctuations. affect the macroeconomy preliminary by depressing demand
for key consumption and investment goods. If that is indeed
CONCLUSION the mechanism by which oil shocks affect the economy, then
a decrease in oil price would not create a positive effect on the
In this paper we examined the oil price and oil and gas company economy. However on the supply side, significant investment
stocks volatility forecast and the impact of oil price shocks to and technological innovations (especially in shale oil extraction)
both selected companies and emerging markets. The innovation caused oil production to fluctuate in a slowing world growth
part in this paper are: (i) We analyze the oil price across three putting downward pressure on oil prices.
sub-periods which also includes the recent oil price crisis, (ii) we
use alternative models to model volatility forecast, (iii) we make NI curves of the company stock returns clearly exhibits that NI on
the analysis both in macroeconomics and microeconomics level volatility changed significantly during oil price crisis compared
considering both production and consumption areas of oil as a to overall period.

152 International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Figure 7: News impact curves for major industry player’s EGARCH models

All those companies we have chosen for the analysis operate both in provided them room for maneuvers in oil price crisis period. This
upstream and downstream businesses along with alternative energy is one of the major reasons why bad and good NI differentiates for
segments. Leveraging their business portfolio and dividend payments commodity prices and industry company stock prices.

International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018 153
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Strong (1991) analyses the ability of oil equities portfolios to hedge volatility. The Journal of Finance, 48(5), 1749-1778.
oil price risk. Using monthly data over the period 1975–1987, Federer, J.P. (1996), Oil price volatility and the macroeconomy. Journal
the oil price sensitivity of firms’ returns appears to be low or not of Macroeconomics, 18, 1-26.
significant, and on average the percentage of oil price changes Golub, S.S. (1983), Oil prices and exchange rates. The Economic Journal,
offset by the returns of the hedge portfolio is only about one-third. 93, 576-593.
Hamilton, J.D. (1985), Historical causes of postwar oil shocks and
In this context, further improvements and additional studies can
recessions. Energy Journal, 6, 97-116.
be achieved by examining the relation between the commodity
Hamilton, J.D. (1988a), A neoclassical model of unemployment and the
hedge market and the underlying commodity itself along with business cycle. Journal of Political Enonomy, 96, 593-617.
its permanent effects on the real industry and macroeconomics Hamilton, J.D. (1988b), Are the Macroeconomic Effect of Oil Price
activities. Changes Symmetric? A Comment, Carnegie-Rochester Conference
Series on Public Policy, 28, 369-378.
REFERENCES Hamilton, J.D. (1996), This is what happened to the oil price-
macroeconomy relationship. Journal of Monetary Economics, 38,
Aloui., R., Aïssa, B.S.M., Nguyen, K.D. (2013), Conditional dependence 215-220.
structure between oil prices and exchange rates: A copula-GARCH Jorion, P. (1990), The exchange rate exposure of US multinationals.
approach. Journal of International Money and Finance, 32, 719-738. Journal of Business, 63, 331-341.
Berk, I., Aydogan, B. (2012), Crude Oil Price Shocks and Stock Returns: Kang, W., Ratti, A.R., Yoon, H.K. (2015), The impact of oil price shocks
Evidence from Turkish Stock Market under Global Liquidity on the stock market return and volatility relationship. Journal of
Conditions, EWI Working Paper, No  12/15, Institute of Energy International Financial Markets, Institutions and Money, 34, 41-54.
Economics at the University of Cologne (EWI). Krugman, P. (1983), Oil shocks and exchange rate dynamics. In: NBER
Blose, E.L., Shieh, P.C.J. (1995), The impact of gold price on the value of Book Exchange Rates and International Macroeconomics. Chicago:
gold mining stock. Review of Financial Economics, 4(2), 125-139. University of Chicago Press. pp259-284.
Bollerslev, T. (1986), Generalized autoregressive conditional Lee, K., Shawn, N., Ratti, R. (1995), Oil shocks and the macroeconomy,
Heteroscedasticity. Journal of Econometrics, 31, 307-327. the role of price variability. Energy Journal, 16, 39-56.
Brooks, C. (2008), Introductory Econometrics for Finance. 2nd ed. Lizardo, R.A., Mollick, A. (2010), Oil price fluctuations and U.S. dollar
Cambridge: Cambridge University Press, exchange rates. Energy Economics, 32, 399-408.
Cappiello, L., Engle, S.K. (2006), Asymmetric dynamics in the Narayan, P., Narayan, S. (2007), Modelling oil price volatility. Energy
correlations of global equity and bond returns. Journal of Financial Policy, 35, 6549-6553.
Econometrics, 4(4), 537-572. Narayan, P.K., Narayan, S., Prasad, A. (2008), Understanding the oil
Chen, S.S., Chen, H.C. (2007), Oil prices and real exchange rates. Energy price-exchange rate nexus for the Fiji Islands. Energy Economics,
Economics, 29(3), 390-404. 30, 2686-2696.
Chen, X., Ghysels, E. (2010), News-good or Bad-and its Impact on Nelson, B.D. (1991), Conditional heteroskedasticity in asset returns:
Volatility Predictions Over Multiple Horizons. The Society for A new approach. Econometrica, 59(2), 347-370.
Financial Studies. Oxford University Press. Ratti, R., Vespignani, J. (2016), Oil prices and global factor macroeconomic
Diaz, E.M., Molero, J.C., de Gracia, F.P. (2016), Oil price volatility and variables.0 Energy Journal, 59, 198-212.
stock returns in the G7economies. Energy Economics, 54, 417-430. Salisu, A.A., Fasanya, O.I. (2012), Comparative performance of volatility
Enders, W. (2004), Applied Econometric Time Series. 2nd ed. New York: models for oil price. International Journal of Energy Economics and
Wiley. Policy, 2(3), 167-183.
Engle, R. (1982), Autoregressive conditional heteroscedasticity with Strong, J. (1991), Using oil share portfolios to hedge oil price risk. The
estimates of the variance of United Kingdom inflation. Econometrica, Quarterly Review of Economics Business, 31, 48-63.
50, 987-1007. Takaishi, T. (2017), Rational GARCH model: An empirical test for stock
Engle, R. (2001), GARCH 101: The Use of ARCH/GARCH models in returns. Physica A, 473, 451-460.
applied econometrics. Journal of Economic Perspectives, 15(4), Wang, Y., Wu, C., Yang, L. (2013), Oil price shocks and stock market
157-168. activities: Evidence from oil-importing and oil-exporting countries.
Engle, R., Ng, K.V. (1993), Measuring and testing the impact of news on Journal of Comparative Economics, 41, 1220-1239.

154 International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

APPENDICES

Table 1: Business descriptions of selected companies for models (1/2)

Company name Business description Country of Exchange
Exxon Mobil Exxon Mobil Corporation is an energy company. The Company is engaged in the United States of America
Corporation exploration and production of crude oil and natural gas, manufacturing of petroleum
products, and transportation and sale of crude oil, natural gas and petroleum products. The
Company also manufactures and markets petrochemicals, including olefins, aromatics,
polyethylene and polypropylene plastics, and various specialty products. The Company
operates through the Upstream, Downstream, Chemical, and Corporate and Financing
segments. The Upstream segment operates to explore for and produce crude oil and natural
gas. The Downstream segment operates to manufacture and sell petroleum products.
The Chemical segment operates to manufacture and sell petrochemicals. The Company’s
projects include the Kearl project, Heidelberg project, the Point Thomson project, the
Hadrian South project, the Lucius project, the Barzan project, the Arkutun‑Dagi project,
and the Upper Zakum 750 project, among others
Chevron Chevron Corporation (Chevron) manages its investments in subsidiaries and affiliates United States of America
Corporation and provides administrative, financial, management and technology support to the United
States and international subsidiaries that engage in integrated energy and chemicals
operations. Chevron operates through two business segments: Upstream and Downstream.
The Upstream segment consists primarily of exploring for, developing and producing crude
oil and natural gas; transporting crude oil by international oil export pipelines; transporting,
storage and marketing of natural gas, and a gas‑to‑liquids plant. Downstream operations
consist primarily of refining crude oil into petroleum products; marketing of crude oil
and refined products; transporting crude oil and refined products by pipeline, marine
vessel, motor equipment and rail car, and manufacturing and marketing of commodity
petrochemicals, plastics for industrial uses and fuel and lubricant additives
ConocoPhillips ConocoPhillips (ConocoPhillips) is an independent exploration and production company. United States of America
The Company explores for, produces, transports and markets crude oil, bitumen, natural
gas and liquefied natural gas (LNG). The Company operates through six operating
segments, which are primarily defined by geographic region: Alaska; Lower 48; Canada;
Europe and North Africa; Asia Pacific and Middle East, and Other International. The
Company’s continuing operations are producing in the United States, Norway, the United
Kingdom, Canada, Australia, Timor‑Leste, Indonesia, China, Malaysia and Qatar. The
Company’s portfolio includes North American unconventional assets and oil sands assets
in Canada; assets in North America, Europe, Asia and Australia; several international
developments, and an inventory of global conventional and unconventional exploration
prospects.
Hess Corp Hess Corporation is an exploration and production (E&P) company. The Company is United States of America
engaged in exploration, development, production, transportation, purchase and sale of
crude oil, natural gas liquids, and natural gas. Its segments include E&P, which is engaged
in the sale of crude oil, natural gas liquids and natural gas, and Bakken Midstream, which
provides services, including crude oil and natural gas gathering, processing of natural
gas and the fractionation of natural gas liquids, transportation of crude oil by rail car,
terminaling and loading crude oil and natural gas liquids, and the storage and terminaling
of propane, located in the Bakken shale play of North Dakota. Its Bakken Midstream assets
include Tioga gas plant, Tioga gas plant, Crude oil train units, Ramberg truck facility,
Gathering pipelines and Gathering pipelines. It has production operations located in the
United States, Denmark, Equatorial Guinea, the Joint Development Area of Malaysia/
Thailand, Malaysia and Norway

International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018 155
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Table 2: Descriptive statistics (1/2)

RARS RBRENT RCAD RCHF RDXY REUR RGBP RJPY RMXN RRUB RTRY RWTI
Mean 0.0006 0.0002 0.0000 −0.0001 0.0000 0.0000 −0.0001 0.0000 0.0002 0.0002 0.0005 0.0002
Median 0.0000 0.0006 −0.0001 −0.0001 0.0000 0.0001 0.0001 0.0000 −0.0001 0.0000 0.0000 0.0006
Maximum 0.4616 0.1271 0.0330 0.0893 0.0390 0.0372 0.0304 0.0622 0.0877 0.1240 0.3567 0.1641
Minimum −0.1036 −0.1444 −0.0377 −0.1714 −0.0437 −0.0318 −0.0841 −0.0378 −0.0654 −0.1288 −0.2513 ‑0.1654
SD 0.01 0.02 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02
Skewness 21.88 ‑0.16 0.14 −2.52 −0.05 −0.01 −1.10 −0.07 1.09 0.56 5.56 ‑0.10
Kurtosis 775.07 5.90 5.83 74.39 6.20 4.62 15.44 7.47 18.64 45.51 228.35 6.93
Observations 4.267 4.267 4.267 4.267 4.267 4.267 4.267 4.267 4.267 4.267 4.267 4,267

Table 3: Descriptive statistics (2/2)

RBP_L RCVX_N RHES_N RTOTF_PA RXOM RCOP_N RRDSA_AS RMRO_N
Mean −0.0001 0.0003 0.0003 0.0001 0.0002 0.0002 24.9696 0.0002
Median 0.0000 0.0007 0.0006 0.0006 0.0003 0.0006 25.2050 0.0006
Maximum 0.1058 0.1894 0.1544 0.1279 0.1586 0.1536 37.5000 0.2099
Minimum −0.1404 −0.1334 −0.2127 −0.0964 −0.1503 −0.1487 15.3800 0.2177
SD 0.02 0.02 0.03 0.02 0.02 0.02 4.08 0.02
Skewness −0.14 0.09 −0.66 0.00 0.04 −0.30 0.34 0.17
Kurtosis 7.51 13.43 11.05 7.46 13.06 8.53 2.81 10.68
Observations 4.189 4.189 4.189 4.189 4.189 4.189 4.189 4.189

Table 4: RWTI GARCH model for global financial crisis period

Distribution Normal distribution Student t‑distribution
Mean equation Variance equation Mean equation Variance equation
Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats
RWTI (−1) −0.05 −2.95 −0.05 −2.92
RCAD −1.12 −14.49 −1.12 −14.72
REUR −0.64 −4.99 −0.51 −3.89
RCHF 0.10 1.55 0.15 2.03
RGBP 0.06 0.77 0.10 1.45
RJPY 0.42 8.00 0.37 6.71
RDXY −1.14 −6.52 −0.96 −5.36
RMXN 0.00 −0.01 −0.04 −0.84
RRUB −0.61 −20.87   −0.59 −16.53
α0 0.00 4.71 0.00 3.26
α1 0.09 9.64 0.08 6.50
β1 0.89 88.79 0.90 66.3
Observations 2118 2118
R2 0.321 0.322
DW 2.078 2.079

Table 5: RWTI EGARCH model for global financial crisis period

Distribution Normal distribution Student t‑distribution
Mean equation Variance equation Mean equation Variance equation
Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats
RWTI (−1) −0.05 −2.79 −0.05 −2.88
RCAD −1.09 −13.95 −1.12 −14.84
REUR −0.56 −4.35 −0.47 −3.53
RCHF 0.06 0.98 0.12 1.80
RGBP 0.06 0.85 0.09 1.34
RJPY 0.38 7.48 0.35 6.63
RDXY −1.00 −5.55 −0.89 −4.96
RMXN −0.03 −0.62 −0.05 −1.04
RRUB −0.57 −18.34 −0.56 −15.32
α0 −0.25 −7.90 −0.21 −5.70
α1 0.15 10.20 0.14 7.15
λ1 −0.06 −6.39 −0.06 −4.55
β1 0.98 314.74 0.99 282.4
Observations 2118 2118
R2 0.320 0.322
DW 2.081 2.078

156 International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018
 Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

Table 6: RWTI GARCH model for oil crisis period

Distribution Normal distribution Student t‑distribution
Mean equation Variance equation Mean equation Variance equation
Coefficient z‑Stats Coefficient Z‑Stats Coefficient z‑Stats Coefficient z‑Stats
RWTI (−1) −0.05 −1.54 −0.05 −1.61
RCAD −2.09 −11.19 −2.12 −11.01
REUR −0.19 −0.54 −0.09 −0.26
RCHF 0.16 0.77 0.19 0.79
RGBP 0.08 0.50 0.07 0.49
RJPY 0.45 3.00 0.42 2.76
RDXY −0.03 −0.06 0.10 0.22
RMXN 0.03 0.31 0.01 0.14
RRUB −0.54 −13.44 −0.54 −11.47
α0 0.00 2.75 0.00 2.01
α1 0.11 4.02 0.10 2.99
β1 0.83 20.85 0.84 16.2
Observations 594 594
R2 0.440 0.441
DW 2.110 2.115

Table 7: RWTI EGARCH model for oil crisis period

Distribution Normal distribution Student t‑distribution
Mean equation Variance equation Mean equation Variance equation
Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats
RWTI (−1) −0.05 −1.50 −0.07 −2.24
RCAD −2.02 −10.64 −2.10 −10.82
REUR −0.20 −0.65 −0.08 −0.25
RCHF 0.08 0.72 0.14 0.90
RGBP 0.08 0.54 0.04 0.28
RJPY 0.40 2.76 0.36 2.39
RDXY 0.06 0.13 0.15 0.37
RMXN −0.01 −0.06 −0.04 −0.42
RRUB −0.53 −13.47 −0.51 −11.41
α0 −0.47 −2.92 −0.22 −1.82
α1 0.18 4.42 0.09 2.14
λ1 −0.06 −2.40 −0.08 −3.39
β1 0.96 51.43 0.98 72.3
Observations 594 594
R2 0.439 0.442
DW 2.130 2.088

Table 8: Granger causality tests for major oil exporter countries

Null hypothesis Overall period Global financial crisis period
F‑Statistic P F‑Statistic P
WTI does not granger cause CAD 1.26 0.17 1.31 0.14
CAD does not granger cause WTI 1.62 0.03 1.34 0.12
WTI does not granger cause RUB 2.21 0.00 1.79 0.01
RUB does not granger cause WTI 0.92 0.58 1.08 0.36
MXN does not granger cause WTI 2.70 0.00 1.82 0.01
WTI does not granger cause MXN 2.67 0.00 1.90 0.01
BRENT does not granger cause CAD 1.05 0.40 1.20 0.23
CAD does not granger cause BRENT 2.25 0.00 2.03 0.00
BRENT does not granger cause RUB 2.05 0.00 1.49 0.06
RUB does not granger cause BRENT 0.91 0.58 0.85 0.67
MXN does not granger cause BRENT 3.29 0.00 1.51 0.05
BRENT does not granger cause MXN 2.27 0.00 2.29 0.00
Observations 4243 2095
Lags 24 24

International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018 157
 158
Table 9: Granger causality tests for Turkey
Null hypothesis Overall period Global financial crisis period
F‑Statistic P F‑Statistic P
WTI does not granger cause TRY 1.72 0.02 1.58 0.04
TRY does not granger cause WTI 1.13 0.30 0.98 0.49
TRY does not granger cause BRENT 1.09 0.35 1.07 0.37
BRENT does not granger cause TRY 1.21 0.22 1.67 0.02
TRY does not granger cause EUR 0.89 0.61 1.29 0.16
EUR does not granger cause TRY 1.18 0.24 1.13 0.30
TRY does not granger cause MXN 1.40 0.09 1.39 0.10
MXN does not granger cause TRY 5.79 0.00 2.92 0.00
TRY does not granger cause RUB 2.19 0.00 2.31 0.00
RUB does not granger cause TRY 1.81 0.01 2.08 0.00
Observations 4243 2095
Lags 24 24

Table 10: Major industry players’ EGARCH model for oil price crisis period with student‑t distribution
RXOM Mean equation Variance equation RCVX.N Mean equation Variance equation RCOP.N Mean equation Variance equation RHES.N Mean equation Variance equation
Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats
RCVXN 0.63 25.81 RXOM 0.66 23.76 RCVXN 0.28 5.60 RCOPN 0.39 10.33
RHESN 0.04 2.56 RWTI 0.02 1.68 RXOM 0.17 3.47 RXOM 0.26 5.40
RWTI 0.01 0.80 RBPL 0.10 5.50 RWTI 0.03 1.76 RWTI 0.12 5.27
REUR −0.06 −1.55 RCOPN 0.20 11.24 RBPL 0.04 1.70 RMRON 0.27 11.92
RMXN −0.08 −2.63 RHESN 0.27 9.59
RMRON 0.21 11.47
α0 −1.41 −2.68 α0 −2.27 −3.26 α0 −1.58 −2.37 α0 −0.37 −2.09
α1 0.33 3.62 α1 0.43 4.62 α1 0.29 3.50 α1 0.13 2.95
λ1 0.00 −0.08 λ1 0.00 0.05 λ1 0.06 1.17 λ1 −0.02 −0.59
β1 0.88 17.76 β1 0.80 11.9 β1 0.85 12.30 β1 0.97 52.3
Observations 564 564 564 564
R2 0.682 0.757 0.785 0.747
DW 2.0353 2.045 1.788 1.832
RMRO.N Mean equation Variance equation RBP.L Mean equation Variance equation RDSa.AS Mean equation Variance equation RTOTF.PA Mean equation Variance equation
Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats Coefficient z‑Stats
RHESN 0.48 12.04 RCOPN 0.09 3.98 RWTI 0.04 2.22 RMRON 0.03 1.94
RCOPN 0.59 12.67 RWTI 0.07 4.10 RTOTFPA 0.45 13.67 RBPL 0.73 29.94
RBPL 0.10 2.65 RTOTFPA 0.68 28.26 RGBP 0.24 3.34 RXOM 0.15 3.99
RWTI 0.08 3.22 REUR −0.28 −3.91
RXOM 0.09 2.71
RBPL 0.42 14.72
α0 −0.20 −2.27 α0 −1.03 −1.27 α0 −15.13 −9.84 α0 −12.50 −1.42
α1 0.15 3.16 α1 0.07 1.16 α1 0.22 3.23 α1 0.07 0.64
Ulusoy and Özdurak: The Impact of Oil Price Volatility to Oil and Gas Company Stock Returns and Emerging Economies

λ1 −0.04 −1.42 λ1 0.05 1.28 λ1 0.08 1.75 λ1 0.05 0.70

International Journal of Energy Economics and Policy | Vol 8 • Issue 1 • 2018

β1 0.99 112.43 β1 0.89 10.2 β1 −0.58 −3.5 β1 −0.37 −0.4
Observations 564 564 564 564
R2 0.752 0.649 0.78 0.648
DW 1.916 1.940 1.945 2.037