Estimation of crude oil import demand of OECD countries

UNIVERSITY OF ECONOMICS
HO CHI MINH CITY
VIETNAM

ERASMUS UNVERSITY ROTTERDAM
INSTITUTE OF SOCIAL STUDIE
THE NETHERLANDS

VIETNAM – THE NETHERLANDS
PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

ESTIMATION OF CRUDE OIL IMPORT
DEMAND OF OECD COUNTRIES

BY

HOANG TUNG DIEP

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, SEPTEMBER 2016


UNIVERSITY OF ECONOMICS
HO CHIMINH CITY
VIETNAM

INSTITUTE OF SOCIAL STUDIES
THE HAGUE
THE NETHERLANDS

VIETNAM - NETHERLANDS
PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

ESTIMATION OF CRUDE OIL IMPORT DEMAND
OF OECD COUNTRIES
A thesis submitted in partial fulfilment of the requirements for the degree of
MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By

HOANG TUNG DIEP

Academic Supervisor:
DR. PHAM THI BICH NGOC

HO CHI MINH CITY, SEPTEMBER 2016


DECLARATION
I declare that: “Estimation of crude oil import demand of OECD countries” is my own
work; it has not been submitted for any degree at other universities.
I confirm that I have made all possible effort and applied all knowledge for finishing this
thesis to the best of my ability.

Ho Chi Minh City, September 2016
Hoang Tung Diep

iii

ACKNOWLEDGEMENTS
I would like to express deepest gratitude to my academic supervisor, Dr Pham ThiBich
Ngoc who gives me helpful comments, excellent guidance. Her patience and caring brings the
motivation for me.
I would like to send my special thanks to Dr. Truong Dang Thuy who always gives me
good advices whenever I got stuck. Additionally, he is the person push me to finish this thesis
and always cares of my thesis process.
I am also grateful to Prof. Dr. Nguyen Trong Hoai and Dr. Pham Khanh Nam and all of
Vietnam – Netherland staffs who always support us for the two-year of studying.
Last but not least, my sincerest thanks are for my family, my friends. Without their
frequent encouragement as well as spiritual support, I would not have been able to complete this
thesis.

iv


ABSTRACT
Crude oil is considerably the most important energy for world economy and most of all countries
were affected by the crude oil price no matter they are played as the producers or consumers or
both. Applying the data of 27 OECD countries from 1988 to 2013, this thesis conducts the
import demand model to estimate the income elasticity and price elasticity for OECD together
with the impact of financial crisis on 2008, the domestic crude oil production, the exchange rate,
the population growth. The estimation of price elasticity for the whole region is -0.155 suggests
that OECD is not sensitive with the increase in crude oil import price. Additionally, the income
elasticity of whole region is 0.562 implies that income raise would lead to the increase in
economic activities, so that the demand for crude oil increases. Finally, the impact of world
financial crisis is confirmed in the estimation.
Keywords: crude oil, demand, import, OECD, price elasticity, income elasticity, crisis

v

ABBREVIATIONS
OECD

Organization of Economic Cooperation and Development

IEA

International Energy Agency

EEC

European Economic Community

IMF

International Money Fund

UK

United Kingdom

USA

United State of America

GDP

Gross Domestic Product

GNP

Gross National Product

OLS

Ordinary Least Square

FE

Fixed effects

RE

Random effects

ARDL

Autoregressive Distributed Lags

FM

Fully Modified

UECM

Unrestricted Error Correction Model

RE GLS

Random effect Generalize Least Square

vi


TABLE OF CONTENTS
CHAPTER 1 INTRODUCTION .................................................................................................. 1
1.1 Problem statement ..................................................................................................................... 1
1.2 Research Objective ................................................................................................................... 2
1.3 Research questions .................................................................................................................... 2
1.4 Research scope, data and methodology .................................................................................... 3
1.5 Thesis structure ......................................................................................................................... 3
CHAPTER 2 LITERATURE REVIEW ......................................................................................... 4
2.1. Some concepts ......................................................................................................................... 4
2.1.1 Crude oil................................................................................................................................. 4
2.1.2 Import demand ....................................................................................................................... 4
2.1.3 Price elasticity and income elasticity ..................................................................................... 5
2.2 Theoretical literature ................................................................................................................. 5
2.2.1 The traditional trade theory.................................................................................................... 5
2.2.2 The new trade theory ............................................................................................................. 6
2.2.3 The standard trade model ....................................................................................................... 6
2.3 Empirical literature ................................................................................................................... 8
2.3.1 Impact of import price and income on import demand.......................................................... 8
2.3.2 Impact of financial crisis on oil import demand .................................................................. 10
2.3.3 Applied control variables ..................................................................................................... 12
2.3.4 Import demand estimation approach .................................................................................... 13
2.4 Analytical Framework ............................................................................................................ 17
CHAPTER 3 RESEARCH METHODOLOGY ......................................................................... 18
3.1. Model specification ................................................................................................................ 18

viii


3.1.1 The crude oil import demand model .................................................................................... 18
3.1.2 Price elasticity and income elasticity ................................................................................... 20
3.1.3 Variables and expected signs ............................................................................................... 21
3.2 Research methodology ............................................................................................................ 23
3.2.1 Estimation models................................................................................................................ 23
3.2.2 Hypothesis tests ................................................................................................................... 26
3.3 Data description ...................................................................................................................... 27
3.3.1 Data source........................................................................................................................... 27
3.3.2 The OECD economic development and oil demand............................................................ 27
3.3.3 Data description ................................................................................................................... 28
CHAPTER 4

EMPIRICAL RESULTS .................................................................................... 33

4.1. The OLS, FE and RE results: ................................................................................................. 33
4.2 Hypothesis tests ...................................................................................................................... 35
4.2.1 Hausman test for Fixed versus Random effects model ........................................................ 35
4.2.2 Random effects Generalized Least Square estimation......................................................... 35
4.2.3 Cointegration test for panel data .......................................................................................... 36
4.3 Estimation of the price elasticity and income elasticity.......................................................... 37
CHAPTER 5

CONCLUSION ................................................................................................... 40

5.1 Main findings .......................................................................................................................... 40
5.2 Policy implications.................................................................................................................. 41
5.3 Limitations and further research ............................................................................................. 42

REFERENCES ............................................................................................................................. 43

ix


LIST OF TABLES
Table 2.3 Summary of empirical researches ................................................................................. 16
Table 3.1 Variables expected signs ............................................................................................... 23
Table 3.3.3 Variables summary .................................................................................................... 29
Table 3.3.4 Mean of Import price (in logarithm) and Income (in logarithm) ............................... 30
Table 4.1 Summarized Estimation of specification ...................................................................... 33
Table 4.2 Hausman test ................................................................................................................. 35
Table 4.2.2 Random effects Generalized Least Square model ..................................................... 36
Table 4.2.3 Cointegration test for panel data ................................................................................ 37
Table 4.3.1 Results of individual income elasticity and price elasticity ....................................... 37
Table 4.3.2 Summarized of price elasticity and income elasticity of crude oil import demand ... 38

x


LIST OF FIGURES
Figure 2.1.2 Import demand curve .................................................................................................. 5
Figure 2.2.3a General equilibrium theory ....................................................................................... 7
Figure 2.2.3b Consumer behavior theory ....................................................................................... 8
Figured 3.3.2 OECD versus Worldcrude oil import from 1988 to 2012 ...................................... 28
Figure 3.3.3 Total Crude oil import to OECD from 1988 to 2013 ............................................... 30
Figure 3.3.4a Interrelationship between crude oil import and income ......................................... 31
Figure 3.3.4b Interrelationship between crude oil import and crude oil price .............................. 32

xi

CHAPTER 1 INTRODUCTION
1.1 Problem statement
It is no doubt about the vital role of crude oil in the world economy and most of all countries
were affected by the crude oil price no matter they are played as the producers or consumers or
both (Natural resource Canada, 2010)(1). Oil has provided about 38 % energy needs in 2014 and it
will contribute to the world economy in further (Berdzenadze, 2015). Although there is new
trend of using the renewable energy which is friendly with environment, it seems that we still not
find energy that can replace for the oil due to its unique attribution such as easy to access or low
cost of refinery.
According to the IEA statistic, Global demand for crude oil in 2014 was about 92.6 million
barrels a day and daily demand for crude oil turned to 94 million in 2015. The increase in oil
demand could be explained by the development of economic activities and human income which
push expenditure on vehicles and other energy application (Zhao and Wu, 2007).
OECD stands for Organization of Economic Cooperation and Development includes 35
members and their mission is: “promote policies that will improve the economic and social wellbeing of people around the world.” The majority of OECD is the developed countries which tend
to use large amount of oil for economic activities. In the 20th century, OECD oil demand was
accounted 70 – 80% of total world demand. The IEA reported that OECD oil demand held 53
percent of total world demand in 2010 and the majority of oil was imported.
Empirical research suggested that energy demand was affected intensely by the financial
crisis. The Global Financial crisis in 2008 was recorded the greatest financial crisis since the
Great Depression (1930s) (Zang et al, 2009) when the crude oil price dropped off more than 72%
percent within five months, from the all-time high at $145.31 per barrel on July 3rd, 2008 to the
lowest value at $41 per barrel on December 5, 2008. The economists found that the financial
crisis started from the raise of subprime mortgages in America in 2007 which leaded to the
reduction in economic activities of OECD. Yet, the economic slowdown made the OECD cut off
their expense hence the crude oil demand declined.
Natural resource Canada is an organization seeks to enhance the responsible development and use of Canada’s
natural resources and the competitiveness of Canada’s natural resources products.

(1)

1


There are many researches related to oil such as oil consumption, oil export/import in
various forms of estimation. The most popular approach is to analyze income elasticity and price
elasticity of oil import. Empirical researches proved that income is inelastic in short run and
more elastic in long run (Ghosh, 2009; Kim and Beak, 2013; Narayan and Smyth, 2007…)
whereas the price elasticity is inelastic both in short and long run (Ziramba, 2010; Yaprakli and
Kaplan, 2015; Altinary, 2007…). It is interesting that energy demand studies often estimate the
particular demand of one country whereas oil demand often analyze the demand for group of
countries or region (Altinary, 2007). Although there are numerous researches that estimate the
energy import demand for particular country, the OECD crude oil import demand is completely
rare in empirical research. Additionally, due to vital role of OECD in the world economy and the
high demand of oil consumption together with high rate of crude oil import, the model of crude
oil demand of OECD needed to be employed to help exporters make prediction of impact of
pricing on future oil consumption and help the importer regulator decide to tax or subsidy on
imported oil. That was the motivations for author to conduct a study of estimation for crude oil
import demand of OECD with the effects of financial crisis in 2008.
The results of this thesis are consisted with previous studies which stated that the price and
income has impact to the quantity of crude oil imported to OECD. Additionally, the author found
that financial crisis made the demand for crude oil import shift down.
1.2 Research Objective
The thesis aims to identify the responses of crude oil quantity imported to OECD with the
change in crude oil imported price and national income in the period 1988 – 2013.
To achieve the above objective, this thesis is conducting the import demand model to
estimate the income elasticity and price elasticity for OECD together with the impact of financial
crisis on 2008, the domestic crude oil production, the exchange rate and the population growth.
1.3 Research questions

This research is designed to answer:
(i) What is a crude oil import demand model of OECD in the period 1988-2013?
(ii) What are price elasticity and income elasticity of crude oil import demand of OECD?
(iii) Does the financial crisis have effect on the crude oil import demand in OECD?
2


1.4 Research scope, data and methodology
The reason for choosing OECD countries for conducting the thesis is the huge demand of oil
in OECD. The empirical statistic showed that OECD oil consumption was 53 percent of total
world demand in 2010, this means the OCDE plays a very important role in the oil market.
Understanding the responses of demand for crude oil to the change in import price as well as the
change in OECD income would be very helpful for the exporters. In addition, there was few
researches paid attention to estimate the OECD crude oil import demand for a long time so that
this motivate the author to investigate the thesis in this region.
The empirical analysis section of the thesis will be implemented by employing unbalanced
panel data from 27 countries in OECD region over the period from 1988 to 2013. The data was
collected from OECD library.
To achieve the research objective, this thesis applied the OLS, the Fixed effects, the Random
effects and the Random effects Generalized Least Squared methods to estimate the price
elasticity and income elasticity of crude oil demand import to OECD and the impact of financial
crisis to the import demand.
1.5 Thesis structure
The outline of paper is as follow. Chapter I introduces the research objectives, the research
questions and the research scope, data and methodology. Next, Chapter 2 provides literature of
import demand and some empirical evidences. Chapter 3 describes the model specification;
research methodology including the OLS model, the FE model, the RE model and the Random
effects Generalized Least Squared model. The empirical results will be presented in Chapter 4.
Finally, Chapter 5 summarizes the conclusions then gives some policy implications and
suggestion of further research.

3


CHAPTER 2 LITERATURE REVIEW
This chapter is a review of oil import demand with the evidences in theory and empirical
evidences. Firstly, there will be some basic concepts that bring the overall images of crude oil,
import demand, price and income elasticity. Secondly, the author will restructure the theory of
trade. In the next step, many previous researches were summarized to give a board learning of oil
and energy import demand. This chapter will end with the Analytical framework.
2.1. Some concepts
2.1.1 Crude oil
Crude oil is the component of crude oil production including crude oil, natural gas liquids
(NGLs) and additives. According to OECD, Crude oil is a mineral oil consisting of a mixture of
hydrocarbons of natural origin, yellow to black in color, and of variable density and viscosity.
Additionally, the IEA organization defines crude oil is a mineral oil of fossil origin extracted
from underground reservoirs and which comprises liquid or near‐liquid hydrocarbons and
associated impurities, such as sulphur and metals.
2.1.2 Import demand
Import demand refers to the total demand for foreign goods and services of population
within particular country. It could be considered the component of Gross National Product
(GNP). According to Krugman (1978) is “An import demand curve is the difference between the
quantity that Home consumers demand minus the quantity that Home producers supply, at each
price”
The Home import demand equation is described:
MD = D – S
The import demand curve shows the maximum quantity of imports that Home country
would like to consume at each price of import good. The import demand curve is downward
sloping indicates that the increase in oil price import lead to the decrease in quantity imported
demand.

4


Figure 2.1.2 Import demand curve

Source: Krugman (1978)
At P1, the Home country solely produces goods at S1 whereas the Home country demand for
goods is D1. The shortage of demand of Home country will be imported from foreign country
and this was estimated by: D1- S1
Similar pattern is found at P2 where the import quantity is calculated by: D2 – S2
2.1.3 Price elasticity and income elasticity
The price elasticity of demand is simply the proportion change in quantity of demand given
by 1 percent change in price.

Ep=

∆𝑄/𝑄 ∆𝑄 𝑃

= x

∆𝑃/𝑃 ∆𝑃 𝑄

The income elasticity of demand is the ratio of the percentage change in quantity demand
when the income increases 1 percent.

EI=

∆𝑄/𝑄
∆𝐼/𝐼

∆𝑄 𝐼

= x

∆𝐼 𝑄

2.2 Theoretical literature
2.2.1 The traditional trade theory
According to Ricadian theory, the country that can produce goods with lower opportunity
cost has comparative advantage. Suppose that the country can produce two kind of goods A and
B. The Ricandian theory of trade stated that: “The economy will specialize in producing A if the
5


relative price of A exceeds its opportunity cost and the economy will specialize in producing B if
the relative price of A is less than its opportunity cost”
Additionally, Hecksher-Ohlin Theory found that difference in comparative advantage is due
to the difference in factor endowment. For instant, the Home country with higher labor to land
ratio is the labor abundant, the Foreign country with higher land to labor is land abundant. The
cloth production requires more labor than the food production whereas the food production
required more land than cloth production. We call that cloth production is labor-intensive and
food production is land-intensive.
When Home country and Foreign country trade together, the relative price will converge.
The relative price of cloth will increase in Home country and decrease in Foreign country. The
raise in cloth price motivates the cloth production and the cloth consumption will decrease in
Home. As the result, the Home becomes the cloth exporter and food importer.
The Hecksher-Ohlin Theory stated that: “Countries tend to export goods whose production
is intensive in factors with which they are abundantly endowed.”
2.2.2 The new trade theory

According to traditional theory, the comparative advantage comes from difference of
technology or factor endowment between countries. Empirical analysis proves that trade between
countries with similar technology or factor endowment still occurs. Therefore, the new trade
theory was developed in 1980s (Krugman 1980). What makes new trade different from the
traditional is the source of comparative advantage. These sources consist of economic scale,
imperfect markets, and product differentiation.
The new trade theory claimed that country tend to import products from another country in
order to better specialize in other products to achieve economies of scale.
2.2.3 The standard trade model
According to the neoclassical trade theory which relies on the assumption of General
Equilibrium Theory and Consumer Behavior Theory, international trade is a function of relative
price. Therefore, the standard trade model was described by Kurgman and Obstfeld (2011) as
follow:

6


General equilibrium theory
A productive ability of a country is defined by the production possibility frontier line and
points lie on the production possibility frontier line depend on the relative price of products that
the economy produced. The market value of output is indicated by the isovalue lines which the
value of output is constant. The economy will achieve highest output level at the point where the
production possibility frontier line tangent to the isovalue line. At this point, the economy yields
the highest welfare. If relative price of one product increase (called the product F), the economy
will produce more F, indicating the decrease in producing others (called the product C).
Therefore, the equilibrium output point will shift to new equilibrium position.
Figure 2.2.3a General equilibrium theory

Consumer behavior theory
The economy choice of point on the isovalue line depends on the consumers’ tastes. The

individual tastes are designed by the indifference curves – the combination of products that
maintain the consumer well fare. The indifference curves are downward slopping indicating that
if individual consumes less C, to achieve the well fare, he must use more F.

7


As consequence, the economy produces more C and consumes more F than consumer
demand. Therefore the excess of C will used for exporting and the shortage of F must be
imported from other countries.
Figure 2.2.3b Consumer behavior theory

2.3 Empirical literature
2.3.1 Impact of import price and income on import demand
The demand of crude oil imported had been developed soon by Kouris and Robinson (1977)
using the data of EEC countries in 1956 to 1985. This research had two stages: the estimation of
EEC petroleum products consumption and the relationship between crude oil imported to EEC
and the quantity of petroleum products consumed in EEC.
At the first stage, petroleum product consumption was dominated by the income, the relative
price of crude oil, the previous petroleum product consumption and the temperature. Applying
the pooled cross-section time series approach to all observations of each country, the two authors
had found that all variables had expected sign and statistically significant even at 1%. This
suggested a positive correlation between the EEC petroleum consumption and the GDP. Also,
the research verified the increase of EEC petroleum consumption is associated with the decrease
8


of oil price. Surprisingly, a computed equation in the second stage showed approximately one-toone association of oil imported and oil consumption which indicated that all of EEC oil
consumption was imported from foreign countries.
In 1993, Hungtington introduced the response surfaces for nine different world oil models to

estimate oil demand for OECD in the period of 21 years (1989-2010) with key variables were
world oil price and GDP. The main purpose of the response surfaces model was to generate the
policy insights rather than the exactly evaluation of each model. The median results revealed that
OECD oil demand was price inelastic (-0.075) in short run and -0.4 in long run whereas the
income elasticity was closely to unity. More recently, Gately and Huntington (2002) conducted
the study to find the asymmetric effects of changes in price and income on energy and oil
demand. The estimation of 96 largest economies including the OECD in the period of 1971 to
1997 showed that long run income elasticity was 0.5 and oil demand responded more with
increase than decrease in oil price/income.
In attempt to estimate the short-run and long-rung elasticity of import demand for crude oil
for Turkey, Altinary (2007) conducted a research covered the period 1980-2005. The models
included price of energy, income and two dummy variables of economy crisis and boom period.
The estimation of short run and long run elasticity of demand applied the ARDL method which
developed by Pesaran et al (2001). The bound test exposed the lag length was 1 which suggested
the long run relationship between variables. The results indicated the long run income elasticity
was 0.61 and nominal price elasticity was -0.18. In term of short run, the result revealed the short
run income elasticity was 0.64, slightly higher than the long-run, whereas the short-run nominal
price elasticity was -0.1. The dummy variable presented for the earthquake was not significant
but the dummy variables stand for War was significant.
For a long time Middle East was recognized as one of main region of exporting crude oil in
the world but there were a truth that this region has reached to bigger oil consumers. Therefore,
the study of Narayan & Smyth (2007) focused on the demand of oil in Middle East using panel
cointegration analysis. This study was consisted with many previous researches that the oil
demand was the function of per capita real income, real price of oil. The first step of panel unit
root test presented that the panel data is stationary and all variables had unit root. As a result, the
author proceed to test the cointegration using the seven Pedroni’s heterogeneous panel

9

cointegration method which presented that that oil demand, real income and real oil price were
cointegrated at least 5 percent. This gave the implication that there was a long run relationship
between variables. For more details, the two authors tried to conduct a long run impact of income
and price on oil demand in Middle East by Dynamic ordinary least square (DOLS) estimator. It
was clear that for full panel, oil demand was income inelastic in short run but it was slightly
elastic in long run while the individual panel presented the strong positive relationship between
oil demand and real income. In term of prices, the result exposed that demand for oil in Middle
East was highly price elastic in both full panel and individual countries.
In effort to estimate the oil import demand in Africa, Ziramba (2010) processed a
multivariate regression analysis for the period of 1980 -2006. The model is simple with two
independent variables: the relative price and the income. Also, the author followed the traditional
method which included Dicky – Fuller test of unit root test of time- series data and the
cointegration test proposed by Johansen technique. The result showed one vector of
cointegration. Unfortunately, the income elasticity and price elasticity were not statistically
significant in short run. In term of long-run, the income price elasticity is 0.429 indicated that
imported crude oil is a normal good whereas the price elasticity is -0.147.
With the same purpose, the study of Ghosh (2009) estimated the oil import demand for
crude oil in India for 1970 to 2006. The Autoregressive Distributed Lags (ARDL) test approach
of cointegration is developed to test the long run relationship between the crude oil import
quantity and income, and imported price. The short run price and income elasticities were not
significant with any level of confident. The long run income elasticity estimated was 1.97 and
highly statistically significant whereas the long run price elasticity was not significant.
Additionally, the bound test indicated one cointegrated vector which motivated the author
processed to the Granger-causality tests. The Granger-causality test result suggested the long run
relationship between the India economic growth and the crude oil import.
2.3.2 Impact of financial crisis on oil import demand
Most of all the researches above were implemented in the oil import-countries where the
crude oil resource was considerably rare. Kolluri and Torrisi (1987) conducted a very interesting
research which estimated an aggregate import demand for five developing oil exporting countries
using the time series data from 1960 to 1982 (Saudi Arabia, Nigeria, Indonesia, Mexico and

10


Venezuela). In details, the oil import demand model is a function of GDP, relative prices index,
the lagged of individual total reverse and dummy variable presented for the impact of oil price
shock in 1973. The OLS estimation seemed to be face with serial correlation in residual and the
maximum likelihood method was applied. The empirical results appeared to be very statistically
significant with right expected signs. Estimated income elasticities were fluctuate from 0.57 for
Saudi Arabia to 1.25 for Indonesia, with average income elasticity for five countries was 0.82
before 1973 and 1.63 since 1973. The separate domestic and import price yield significant and
correct signs with average elasticity is -1.04. The impact of dummy variable has negative value
indicate that the oil price shock in 1973 made the import demand shift down.
Despite the growing of renewable energy, crude oil holds it vital role in production and daily
life in Korea. Therefore, Kim and Baek (2013) examined crude oil import demand in Korea for
quarterly data from 1986 to 2010 using the traditional Autoregressive distributed lag (ARDL)
bound test approach. The model showed that determinants of crude oil import to Korea were
relied on the import price and income. One thing that made this research different from others
was the dummy variable captured the market shocks. The ARDL test required to formulate the
demand equation with lag variables and the selected variables was said to be cointegrated if all
the lagged variables were statistically significant. The authors revealed the appropriate lag length
for Korea estimation was p = 2 and the cointegration test showed the long rung relationship
between crude oil import and income as well as oil price. In details, the long run income
elasticity was 1.31 and the price elasticity held the lower value at -0.43 in long run. Additionally,
the market shocks dummy variable yielded significant indicated that the Asian financial crisis
made the demand for crude oil import in Korea raised.
In 2015, Yaprakli and Kaplan built a model based on traditional import demand model
which including income and relative price variables. For deeper investigation, these authors tried
to find the impact of financial crisis on the crude oil demand in Turkey for the period 1970 –
2013. To handle time series data problems, the paper used co-integration method which
contained three steps: the unit root test, test for cointegration and the dynamic ordinary least

squared (DOLS) estimations. The cointegration test supported long run relationship between
variables and the empirical results showed the inelasticity of income and price in the long run.
Moreover, the authors concluded that external crisis had bigger impact on crude oil import
demand the internal crisis did.
11


2.3.3 Applied control variables
The relationship between oil domestic energy production, population and oil import demand
were introduced in the study of Adewuji (2016). This study estimated import demand for total
and specific refined petroleum products in Nigeria from 1984 to 2013. The autoregressive
distributed lag (ARDL) bound test estimation was applied to find the long run and short run
determinants of import demand. The results showed that domestic energy production and
population growth rate were drivers of demand of import of refines kerosene and motor spirit in
long – run. In the short-run, significant effect of domestic energy production on the import
demand for refined products was not found but the population growth rate did. Additionally, the
real effective exchange rate was main determinant of total and specific petroleum products in
Nigeria due to it significant effect in almost estimation. For example, the long run elasticity of
the real effective exchange rate is -1.08 for total import of refined petroleum products and long
run elasticity of the real effective exchange rate for motor spirit product and refined distillate fuel
were -0.77 and -1.1 respectively. The effect of real effective exchange rate in short-run of motor
spirit product was 0.4 and it rose to 1.82 for refined distillate fuel.
The role of exchange rate was confirmed in the research of Schryder and Peersman (2012)
by the estimations of oil demand for 23 OECD countries, 42 non-OECD countries and total 65
oil importing countries respectively for the period of 1971 to 2008. This study was applied the
advanced recent panel technique to examine the effect of income, oil price in US dollar and the
US dollar exchange rate on oil demand. The results seemed to be very consisted with the theory
which revealed the short run income elasticity of OECD was 0.568 and it was slightly higher in
non-OECD (0.639) and the total sample of countries (0.614) whereas the long run income
elasticity was 0.674, 0.885 and 0.775 respectively. In term of price elasticity, the short run price

elasticity was -0.051, -0.026 and -0.035; the long run price elasticity was higher which yielded to
-0.150 for OECD and-0.104 for non OECD. What made this study different from the others were
the findings of impact of US exchange rate on the oil demand. In details, the 1 percent
appreciation of US dollar exchange rate leaded to 0.24 percent decrease in oil demand of OECD
countries and it was timed twice to -0.150 for 65 countries.
Another research related to energy import demand was proposed by Goldar and
Mukhopadhyay (1990). The analysis of seven petroleum products import demand for India was

12


estimated for the period of 1970 – 1986 using a very simple OLS method. In particular, the
domestic demand function for petroleum products was dominated by the price index of each
products, the price of substitution possibilities exist, the other exogenous variables included the
number of tractors, the number of vehicles, industrial production index, gross power generation,
the per capita income and the population. The results suggested that domestic demand for most
of all seven petroleum products were not influent much on price changes. The other explanation
variables were considerably statistically significant with expected signs. In the next step, the
author tried to generate the import function of petroleum and found that the gap between
domestic crude oil production and the domestic demand for petroleum was the main
determinants of quantity of petroleum imports to India. What is important conclusion of this
research was the depreciation of domestic currency (rupee) which leaded to higher price of
imports could not reduce the domestic demand.
2.3.4 Import demand estimation approach
The Monte Carlo approach
For more than quarter of century, the traditional import demand function was estimated
without econometric problems. This problem was the nonstationary of variables which could
lead to the unreliable results. Therefore, Sehadji (1997) introduced the Monte Carlo approach in
which the Ordinary Least Square (OLS) estimator was dominated by Fully –Modified (FM)
estimator. The empirical research had showed that the OLS estimator faced the problem of bias

higher than the FM estimator. The FM estimator would achieve the minimum bias when the
relative price of imports and the activity variables were not endogenous. As consequence, the
FM estimator was an optimal single-equation method based on the use of OLS with serial
correlation corrections and potential endogeneity prevention of the dependent variables. In
addition, the Monte Carlo approach was very appropriate for the estimation with small sample.
Also, the FM estimator could be recognized similar to the full systems maximum likelihood
estimators.
The Dynamic - Optimizing approach
Although the Monte Carlo approach had solved the nonstationary problem of time series
data, there were several limitations with this method. In details, the time-series approach
assumed that the output was stationary and followed AR (1) process while it was nonstationaryin
13


real. In addition, the investment and government activity was not mentioned in the model which
could lead to the unstable results. Thanks to Dynamic - Optimizing approach conducted by Xu
(2000), the import demand function had taken into account both the growing economy,
investment and government activity. The Dynamic – Optimizing approach offered a
conventional import demand equation that included the “national cash flow” variable (GDP-I-GEX), the relative prices and time trend. Additionally, the investment and government was
included as the activity variables while the activity variable in the research of Shehadji (1997)
was the GDP minus exports. The Dynamic – Optimizing approach was considered more
advantage than the previous methods because it created the more general and flexible import
demand model.
The Cointegration approach
The Cointegration approach is the test of existence of a cointegrated relationship between
the aggregate import demand function and its determinants. This process is developed by Engle
and Granger (1987) and Johasen (1988, 1991). If there is an existence of a cointegration, the
long-run relationship between the aggregate import demand function and its determinants is
determined.
The Engle – Granger’s residual –based ADF method has two steps:

1. Applying OLS on the nonstationary variables to test the parameters of cointegrating
regression.
2. Appling Aguement Dickey – Fuller (ADF) to test residuals stationary.
The Johansen –Juselius (JJ) method is considerably better than the Engle – Granger’s
residual –based ADF method due to some reasons:
- The JJ method could make the combined framework for interpreting the distinction of
cointergating vectors.
- The JJ method allows us to test the magnitude and the volume of the elasticity estimation.
The Autoregressive Distributed Lags (ARDL) bound test approach
The bound test approach is generally a test based on the estimation of Unrestricted Error
Correction Model (UECM). It has two advantages:

14


- While the Johansen (1988) and Johansen and Juselius (1991) estimation are solely
appropriate for the nonstationary time-series with the same integrated level A(1), the bound test
provides the estimation for A(0); A (1) level or fractionally integrated. It means no need of test
for non-stationay properties and order of intergration of variables.
- The bound test is very powerful in cointegration analysis for model with small data.
According to Pesaran et al (2001), the bound test is based on the value of F-statistic. The
null hypothesis here is: whether the explanatory variables are I(1) or A (0), there are no
cointegration between examined variables. If the F-statistic value is higher than critical value,
I(1) then rejects the null hypothesis. In case F-statistic value is lower than critical value, I(0), the
null hypothesis cannot be reject. Note that in case the F-statistic falls within the critical value
bounds, the bounds test is solely capable use for I(1) and I (0) regressors.
- Furthermore, the bound test help to estimate the long-run elasticity by estimated coefficient
of one lagged dependent variable divided to estimated coefficient of one lagged independent
variable. In addition, the short run elasticity is estimated coefficient of first difference of
variables in UECM model.

The Panel Cointegration approach
Pedroni’s test (2004) is broadly known as the cointegration approach for heterogeneous
panels. It provides seven statistics for the test of null hypothesis no cointegration between the
dependent variable and independent variables. Among seven statistics, four are preferred for the
“within-dimension” (panel-v, panel-ρ, semi-parametric panel-t and parametric panel-t) and three
preferred for “between-dimension” (group-ρ, semi-parametric group-t and parametric group-t).
The former are calculated by “summing up the numerator and the denominator over N crosssections separately, while the latter statistics are calculated by dividing the numerator and the
denominator before summing up over N cross-sections” (Pedroni, 2004). If the statistics exceed
the critical value which calculated by Pedroni (2004), the null hypothesis of no cointegration is
rejected. Hence, there is a long run relationship between examined variables.

15