Analysis of Inflation Rates of Some Selected Petroleum Products

ANALYSIS OF INFLATION RATES OF SOME SELECTED PETROLEUM
PRODUCTS (PETROL, DIESEL AND KEROSENE) USING ECONOMETRIC
APPROACH
BY
ADAM ABDULRASAQ OLUWATOYIN
MATRIC NO: 1911320026
DEPARTMENT OF MATHEMATICS AND STATISTICS, FACULTY OF PURE AND
APPLIED SCIENCES, OSUN STATE COLLEGE OF TECHNOLOGY, ESA OKE,
OSUN STATE
IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF
HIGHER NATIONAL DIPLOMA (HND) IN STATISTICS
AUGUST, 2021
1
CERTIFICATION
This is to certify that this project was carried out by Adam Abdulrasaq Oluwatoyin of the
Department of Mathematics and Statistics, Osun State College of Technology, Esa Oke under
my supervision.
………………………… . ……………………….
Mr Akinrinmade Date
Supervisor
……………………………. ……………………….
Mrs Akinawoniran Date
Head of Department
………………………… ………………………..
External moderator Date
2
DEDICATION
This project is dedicated to my parents Mr. and Mrs. Adam who provided me a strong
foundation.
3
ACKNOWLEDGEMENT
Unless the lord builds the house, the builders labour in vain. I am much grateful to almighty God,
the giver of life, the only wise and ever-knowing God, for his grace upon my life, who has been
my source and tower of strength all through the course of my study.
My gratefulness extends to my project supervisor Mr Akinrinmade for his efforts and guidance
towards making this project a success. And also to all the lecturers in the department who have
impacted knowledge in me ever since my early days on campus. The lord will take you all to
greater heights.
I am greatly indebted to my ever loving parents Mr. and Mrs. Adam for their hard labor, love and
care for my prayer is that God should please grant them long life so that they can eat the fruits of
their hard labor.
My family means the world to me, so thank you Mr. Adam Abdulquadri, for your love and
support. I love you more than you can ever imagine.
And to my wonderful friends like family members Adam Abdulganiyu, Adam Abdulsemiu you
guys have really being friends in need.
My undiluted thanks also goes to my wonderful family on campus-MSSN mosque, you are all a
wonderful set to be with. My appreciation also goes to my spiritual parents, Mr and Mrs Adam.
I especially want to thank my best friend Lasisi Samuel lati for his exceptional support. I also
express my gratitude to friends like Olamilekan Sodiq and all my course mates in HND2
statistics, see you all at the top.
4
ABSTRACT
The use of econometric model has over the time being a unique tool that allows test,
detection correlation or remodeling to give the close to perfect response to the presented
problems. This study has carefully shown the nature of relationship that exists, when Inflation
(Y) was regress on Petrol (X1), Kerosene (X2) and Diesel (X3). The model fitted is
Ŷ=30.37209+0.197102X1-0.642958X2-0.008665X3.
Petrol contributes positively to the forecasting power of inflation rate while kerosene (X 2) and
diesel (X3) did not show any evidence of contribution. The value of adjusted R-square
(R2=0.245) depict that the combinations of Petrol, Diesel and Kerosene explained about 25% of
total variation in inflation rate of the petroleum products ,leaving the remaining 75% to be
explained by other unknown factors. The data was tested for stationarity and this was stationary
at third difference using graphical approach and unit root test. The test of significance of each
parameter, shows that multicollinearity exist.
5
TABLE OF CONTENTS
Page
Title Page i
Certification ii
Dedication iii
Acknowledgement iv-v
Abstract vi
Table of Content vii-xi
CHAPTER ONE: INTRODUCTION
1.1 Background 1-2
1.2 Objective of the Study 3-4
1.3 Hypotheses to Be Tested 4
1.4 Source of Data 5
1.5 Scope of the Study 5
6
CHAPTER TWO: LITERATURE REVIEW
2.1 Conceptual Framework 15
2.2 Theoretical Framework 16
2.4 Appraisal of the Reviewed Literature 17
CHAPTER THREE: METHODOLOGY
3.1 Definition of Some Terms 10
3.2 Types of Econometric Model 11-12
3.3 Steps in Econometric Analysis 12-13
3.4 Types of Economic Data 13
3.5 Regression Analysis 14
3.6 Simple Linear Regression Model 14-15
3.6.1 Assumptions of Linear Regression 15-16
3.6.2 Multiple Linear Regression 17
3.6.3 Assumptions of Multiple Linear Regression Model 17-18
3.7 Coefficient of Determination (R2) 18
3.8 Econometric Problems Relating To A Single Equation 19
3.8.1 Multicollinearity 19
7
3.8.2 Causes of Multicollinearity 19
3.8.3 Consequences of Multicollinearity 20
3.8.4 Test for Multicollinearity 20-21
3.8.5 Solution to Multicollinearity 21
3.9 Autocorrelation 22
3.9.1 Causes of Autocorrelation 22
3.9.2 Consequences of Autocorrelation 23
3.9.3 Test for Autocorrelation 23-25
3.9.4 Correcting Autocorrelation 26
3.10 Heteroscedasticity 26-27
3.10.1 Consequences of Heteroscedasticity 27
3.10.2 Testing for Heteroscedasticity 27
3.10.2.1Goldfield Quandit Test 27-28
3.10.2.2 Spearmanʹs Rank Correlation Analysis 28
3.11 Overcoming Heteroscedasticity 28
3.12 Testing for Predictive Power 28-29
3.12.1 Hypothesis 29
8
3.12.2 Conclusion 29
CHAPTER FOUR: DATA ANALYSIS AND INTERPRETATION
4.0 Data Analysis and Interpretation 30
4.1 Presentation of Data 31
4.2 Time Plot of Variables 32-33
4.3 Descriptive Statistics of Variables 34-35
4.4 Regression Analysis 35
4.4.1 Test for Joint Significance 35-37
4.5 Hypotheses Testing and Interpretation 37
4.5.1 Hypothesis One 37-38
4.5.2 Hypothesis Two 38-39
4.5.3 Hypothesis Three 40-41
4.6 Tests for Multicollinearity 41-42
4.7 Test for Autocorrelation 42-43
4.8 Correlation and its Significance 44
4.9 Test for Heteroscedasticity 45-46
4.10 Forecasting Evaluation 47-48
9
CHAPTER FIVE: SUMMARY, CONCLUSION AND RECOMMENDATION
5.1 Findings 49
5.2 Conclusion 50
10
CHAPTER ONE: INTRODUCTION
1.1 Background of the Study
The contributions of the petroleum industry to growth and development of the
Nigerian economy can be enumerated in terms of the industry’s impacts on the economic
variables responsible for economic growth in Nigeria. The contributions of petroleum
industry can also be analyzed in terms of its share of revenue generation in the Nigerian
economy. The petroleum industry has contributed immensely in both foreign exchange
reserves and government revenues (Onyemaechi 2012). It has been observed that the
government share of crude oil revenue as a result of various joint venture agreements
with the international oil producing companies are roughly 70 percent of revenues
accruing from crude oil transactions. He said further that the petroleum industry can also
contribute significantly to growth and development of the Nigerian economy through
foreign direct investment (FDI). Foreign direct investment (FDI) has been referred to as
real investment interactions of the rest of the world with a given domestic economy.
Whether these interactions encourage or discourage economic growth depends on the
area of strength of the economy concerned and purpose of the investment.
The origin of the persistent rise in prices of goods and services can be traced to
government and its use of the oil revenue. It is evident that export of petroleum earns
valuable foreign exchange in Nigeria, and that petroleum and money are so important to
modern living that shortages disrupt essential transactions. Indeed, the occasional petrol
shortages experienced by Nigerian towns and villages due to inefficient distribution,
11
which is as a result of incompetence and corruption on the part of bureaucrats and the
business class. The links between petroleum and money are easily obvious. More
important, a proper grasp of the relationship between the domestic and the world
economy is essential to the identification of these relationships. The monetization of
petro dollars shows how closely related petroleum is to money stock, which in essence
has direct bearing on the inflationary pressures in the economy (Sikkam 1999).
According to (Osagie 1981), the annual growth rate of money supply especially in
1974 and 1975 far outstrips those of all the developed industrial countries as well as the
developing countries of the world. Even compared with other OPEC countries where the
growth rate of the money supply has been generally high, the recent growth rates of
money supply in Nigeria is alarming. The alarming rate of money supply is influenced by
the earnings from petroleum.
The oil industry is very important to the Nigerian economy. It provides among
other things the greatest part of the foreign exchange earnings and total revenue needed
for socio-economic and political development of Nigeria. The bulk of Nigerian crude oil
is sold unrefined and when refined, the products range from petrol to heavy liquids for
road tarring. Government has been the custodian of petroleum and its products in Nigeria.
The persistent instability of crude oil prices in the global market has adversely affected
all the sectors of the Nigerian economy negatively. This is because Nigeria is a
monoculture economy. From 1990 to 2011, the prices of petroleum products were
reviewed more than ten times. The adjustment in 2000 under the democratically elected
government marked a turning point in the economy as petrol moved up to N30 per litre,
12
diesel to N29 and kerosene to N27. And as of today, petrol goes for N87, kerosene N120
and diesel N150. According to the government, the upward review of domestic prices of
petroleum products was necessitated by the high spot market price of crude oil and the
need for higher margins for the Nigerian National Petroleum Corporation (NNPC) to
meet operational and capital costs. The upward adjustments of petroleum products have
resulted in inflation, high cost of living, and inequitable distribution of income in Nigeria.
However, this has led to one major problem which is instability of the prices of goods
and services in virtually all the sectors of the economy.
1.2 OBJECTIVE OF THE STUDY
1. To fit a model that show the relationship between inflation rate (endogenous
variable), and petrol, diesel and kerosene (exogenous variable).
2. To test the data for multicolinearity, autocorrelation and heteroscedasticity.
3. To test the hypothesis that PMS, DPK and AGO have effect on the inflation rate.
4. To predict the value of inflation rate for any given price of PMS, DPK and AGO.
1.3 HYPOTHESIS TO BE TESTED

Hypothesis I
H0: The exogenous variables have no effect on the endogenous variable.
H1: The exogenous variables have effect on the endogenous variable.
Hypothesis II
H0: Multicolinearity does not exist between the variables.
H1: Multicolinearity exists between the variables.
Hypothesis III
13
H0: Autocorrelation does not exist between the variables.
H1: Autocorrelation exists between the variables.
Hypothesis IV
H0: Heteroscedasticity does not exist.
H1: Heteroscedasticity exists.
1.4 SOURCE OF DATA
The data used for this project work was obtained from “CIA World Fact Book, National
Bureau of Statistics, and statistical bulletin (Central Bank of Nigeria) (CBN)”
1.5 SCOPE OF THE STUDY
This project work is carried out to study the relationship between the inflation rate
and the price of petrol, diesel and kerosene from 1990 to 2014.
14
CHAPTER TWO: LITERATURE REVIEW
2.1 Conceptual Framework
Onwioduokit and Adenuga, (2000) in their work “empirical analysis of the
demand for petroleum products in Nigeria”, were able to find that there has been
perennial petroleum products scarcity in the country over the past few years; inadequate
energy planning had compounded the problem of scarcity. The empirical findings reveal
that urbanization was one of the principal factors that have a positive impact on the
consumption of liquefied petroleum gas and premium motor spirit. The impact of
urbanization on the consumption of household kerosene is negative, showing that
kerosene is not urbanization elastic.
Siddy (1999) asserted that the causes of price instability is attributed to scarcity
caused by refinery maintenance and rehabilitation problem, low capacity utilization,
supply, and demand inequality. The political change that Nigeria went through, which
turned over the administration and endured a lingering economic down turn is enough
reason to cause price instability of oil products in Nigeria. The author opined that trailing
oil products prices down to crude oil prices has revealed that the instability in the prices
of oil products was due to cost of refining, storing, transporting distributing and
inefficiencies in the process.
Nwosu (2009) in her work “the impact of fuel price on inflation”, which used the
variance Autoregressive analysis model to assess the relative contribution of fuel price on
inflation. The study used available quarterly data series spanning 1995 to 2008. The
15
finding of the study revealed that the policy of subsidizing the price of fuel should be
continued so as to help cushion the economy from the adverse effects of oil-price shock.
2.2 Theoretical Framework
Mbendi (2000) argued that in theory, Nigeria’s refineries capacity is sufficient to
meet its domestic consumption requirement. In practice, however, according to the
author, the country has experienced frequent shortage of refined products since it
refineries have poor configuration and operation inefficiency. The author stated that it has
been estimated that smuggling amounts to over 320,000 barrels per day largely to Benin
Republic, Niger, Chad, and Cameroon. The author noted that Nigeria has become a large
importer of light petroleum products, importing thousands of tons of refined products.
Arinze (2011) in his work “the impact of oil price on the Nigerian economy,”
asserted that there is a direct relationship between fuel price increase and inflation rate in
Nigeria and also recommended that more resources should be tapped to diversify the
economy. In this paper, effort has been made to look at the adverse effect of petroleum
product prices increase on the Nigerian economy. Is the discovery of oil in the country a
blessing or a curse to Nigerians? Works reviewed have shown concern in this area of
study, while the main objective of the paper is to examine the effect of price increase of
petroleum products on the Nigerian economy.
Hooker (2002), De Gregorio, et al. (2007), Blanchard and Gali (2007), and Shioji
and Uchino (2010) made similar conclusions that oil price pass-through has declined in a
number of countries such as the US, Japan and other industrialized countries. They
attributed the developments to the intensity with which oil is used in production in those
16
countries, improved monetary policy, greater wage flexibility and the presence of off-
setting shocks.
2.3 Appraisal of the Reviewed Literature
On the contrary, Berument and Tasci (2002) investigated the effects of oil prices
in Turkey and found that when wages and other three factors of income (profit, interest
and rent) are adjusted to the general price level that include oil price increases, the
inflationary effect of oil prices becomes significant.
Brown, Oppedahl and Yucel (1995) in their study on how oil prices transmit
through various channels of the US economy to influence inflation suggest that monetary
policy generally accommodated the inflationary pressure of oil price shocks.
Hamilton (2003) investigated the role of monetary policy in eliminating
recessionary consequences of an oil shock and concludes that the potential of monetary
policy to avert the contractionary consequences of an oil price shock is little or not as
great as suggested by the analysis of Bernanke, Gertler, and Watson (1997).
A study by Bouakez, et al (2008), using a Dynamic Stochastic General Equilibrium
(DSGE), analyzes how high oil prices would lead to an increase in inflation by a much
greater magnitude under managed than under a fixed exchange rate regime.
Furthermore, De Fiore, et al. (2006) looked at simple policy rules and found that
oil price shocks brought about a trade-off between inflation and output stabilization and,
thus, monetary policy partially accommodated oil-price increase.
Ewa and Agu (2003) shared their view that the dominance of petroleum in
Nigerian economy has led to instability in the economy, which as a result makes price
17
instability of oil products to be more prevalent in Nigeria than other countries. The author
observed that smuggling is attractive and profitable due to price differential. This act of
smuggling oil products from Nigeria to her neighboring countries is one of the factors
which made price instability of oil products to be prevalent in Nigeria.
Runl (2010) asserted that people say Nigeria is dominated by oil and they are right
because Nigeria seems to be exporting nothing but oil. The government revenues are so
dependent on oil, which has been managed quite protectively. But it’s still extremely
undesirable that internally generated revenue are such a small part of Nigeria’s revenue
because essentially, it means that all the revenues of the government is just coming down
from heaven. It’s like a gift and it is easy to waste a gift. The author noted that Nigeria is
poor because of oil.
18
CHAPTER THREE: METHODOLOGY
3.1 DEFINITION OF SOME TERMS
1. TEST STATISTIC: This is a numerical index that is expected to take a value if
Ho is correct and is expected to take some other values if H1 is correct.
2. MODEL: This is the representation of problem in mathematical form that can be
estimated.
3. ECONOMETRIC MODEL: This is the specification of the relationship between
the variables into a mathematical form.
The model for this project is
Ŷ= ^
ẞ0 + ^
ẞ 1X1i + ^
ẞ 2X2i + ^
ẞ 3X3i + Ui
Where:
Y = Inflation Rates
^
ẞ 0 = Intercept
X1i = Petrol
X2i = Diesel
X3i = Kerosene
Ui = Error Term
^
ẞ 1, ^
ẞ 2 and ^
ẞ 3 = Coefficients of the explanatory variables
19
3.2 TYPES OF ECONOMETRIC MODEL
Two types of econometrics discussed in this project are:
 SINGLE EQUATION MODEL: this is a form of equation in which a dependent
variable is expressed as a function of independent variables. The model may be
linear or non-linear. Linear is in the form of Y=a + bx + u
b
While the form of non linear is Y =a+
x
 MULTIPLE EQUATION MODEL: This is the type in which a dependent variable
Y1 is determined by two or more explanatory variables
X1,X2,X3………………..XK.
This will be expressed in the matrix form:
Y=Xẞ+U
Where
Y1 X11 X12 . . . . . X1k

Y2 X21 X22 . . . . . X2k
Y= . X= . . .
. . . .
Yn X1T X2T . . . . . XkT
20
B0 U1
B1 U2
B= . and U= .
. .
BK UT
3.3 STEPS IN ECONOMETRIC ANALYSIS
The process involved in econometric analysis can be divided into five major steps:
1. SPECIFICATION OF THE MODEL: It involves the expression of econometric
theory in mathematical form. At this stage, relevant variables in econometric relationship
are identified and the patterns of relationship are defined.
2. COLLECTION OF DATA: It involves the collection of data often related to
random sample drawn from the population of variables.
3. ESTIMATION: This involves the use of appropriate econometric and statistical
techniques to obtain numerical values for model using collected data. This step covers the
initial processing of data to fit estimation requirements, the device of estimation
techniques and maybe the determination of the relationship among variables in a model.
4. VERIFICATION: once the model has been estimated, one should proceed with
the evaluation of the estimates, i.e. Decide on the basis of certain criteria, whether the
estimates are satisfactory and reliable. The criteria for evaluation are economics,
21
statistical and econometrics, the emphasis put on this aspect derives mainly from the need
to establish the validity of methods of statistical inference applied to econometric models.
5. FORECASTING: this involves the use of estimated model to forecast unknown
future values of variables.
3.4. REGRESSION ANALYSIS
According to Montgomery and Peak, the regression analysis is the study of the
relationship among variables and its purpose is to predict, or estimate the values from
known or assumed values of other variables related to it. Regression analysis is a
statistical technique for modeling and investing the relationship between two variables or
one variable. In other words, it is a statistical techniques used for forecasting which seeks
to establish “the line of best fit” to the observed data.
There are two types of variables involved in regression analysis. The first type is
called the independent variable, explanatory variable or regressor while the second type
is called the dependent variable, the response variable or endogenous variables.
3.5 SIMPLE LINEAR REGRESSION MODEL
This is a special but simple case regression, which contains one independent
variable and also one dependent variable. Also according to G.M.Clarks and D.Cooks,
the two variables regression model is the mathematical linear relationship assumed
between a dependent variable-Y and independent variable-X .Thus, the model for the
regression is given as: Y = B0 + BIX + e1.
22
Where: Y=dependent variable
X=independent variable
ẞ0 and ẞ1 are unknown parameters
e= Error term.
3.5.1 ASSUMPTIONS OF THE LINEAR REGRESSION
i. The model is linear i.e. Y=B0 + B1X1 + e1
ii. The error structure is additive.
iii. The random error.
a. Zero mean i.e.∑(U)=0
U1 0
U2 0
E= . = .
. .
UT 0
b. Equal variances are mutually uncorrected.
E(UU1) =σ2 In, .i.e. homoscedasticity
U1
U2
E (UU1) = . U1 U2 . . . . . Un
.
Un
23
Where UU1= n*n matrix, since (n*1) (n*1) =n* n i.e.
Cov (U) =E (UU1)

E(U12) E(U1U2) . . . . . E(U1Un)
E(U2U1) E(U22) . . . . . E(U2Un)
E(UU1) = . . .
. . .
E(UnU1) E(U2Un) . . . . . E(Un2)
Since E (UiUj) =0 i≠j and E (UiUj) =σ2 i=j
Then:
σ2 0 .....0 1 0....0
0 σ2 . . . . .0 0 1....0
E(UU1) = . . . = σ2 . . .
. . . . . .
0 0 . . . . . σ2 0 0....1
=σ2 In
iv. X is non-stochastic. It is a set of fixed numbers. That is, they are independent of
the error term U.
v. X has rank K<T i.e. X is of full rank and the number of observation exceeds the
number of parameters to be estimated.
3.5.2 MULTIPLE LINEAR REGRESSION
Multiple regression analysis is the extension of two variable linear regression to
make account of more than one independent variable. It is the appropriate technique used
24
to investigate the effect of X’s variables on Y. Other variables influencing Y in a multiple
regression analysis must be included because of the following reasons:
1. To reduce stochastic error: the objective here is to reduce the residual variance σ 2
and hence increase the strength of our statistical test.
2. To remove or eliminate bias that might result if we ignore a variable that
substantially affects Y.
The general multiple regression model is:
Y=ẞ0+ ẞ1X1 + ẞ2X2 + ẞ3X3 +……………….ẞKXK + Ɛt
3.5.3 ASSUMPTIONS OF MULTIPLE LINEAR REGRESSION MODEL
1. Yt = ẞ1 + ẞ2Xt2 +…………….+ ẞkXtk + Ɛt, t=1,…….T
2. E(yt) = ẞ1 + ẞ2Xt2 +………….+ ẞkXtk E(Ɛt)=0
3. Var(yt) = var(Ɛt) = σ2
4. Cov(yt,ys)=Cov (et,es)=0
5. The values of Xtk are not random and are not exact linear functions of the other
explanatory variable.
6. yt ~ N{(ẞ1 + ẞ2Xt2 +…………….ẞkXtk),σ2}↔et~N(0,σ2)
25
3.6 COEFFICIENT OF DETERMINATION (R2)
The major reasons for analyzing the model are:
1. To explain how the dependent variable (yt) changes as the dependent variable (xt)
changes.
2. To predict yo given. Closely allied with the 2 point mentioned above is the desire
to use xt to explain as much of the variation in the dependent variable yt as possible.
Therefore, it is used to determine whether the regression line provides a good fit and
measures the proportion of the variation of Y (explanatory variable) explained by
variation in the independent variables. It lies between 0 and 1 i.e. 0<R 2<1.the closer R2 is
to 1, the better the job we have done in explaining the variation in y t with ýt=b1+b2xt, and
the greater is the predictive ability of our model.
It is given by:
2 SSR 1−SSE
R= ∨
SST SST
3.7 ECONOMETRIC PROBLEMS RELATING TO A SINGLE EQUATION
3.7.1 MULTICOLLINEARITY
The problem of multicolinearity arises when two or more independent variables in
the regression model are highly correlated. Thereby making it difficult or impossible to
isolate the individual effect on dependent variable, when this problem arises, it becomes
difficult to estimate the parameter ẞ using least square estimate procedure. Also,
26
determinant of (x1x) will be close to zero and hence it is inverse, (x 1x)-1 becomes very
large. Hence, an indicator of multicolinearity is very high standard errors of ẞ
coefficient. The assumptions of the ordinary least squares regression techniques of
independent variables are not highly correlated in the sense that they do not move
together in the same pattern i.e. There is no linear relationship between the individual
variables.
3.7.2 CAUSES OF MULTICOLLINEARITY
I. The incorporation of lagged values of explanatory variables as separate
regression.
II. The inherent tendency of economic variables to move together especially
overtime but over different agents at the same point in time.
3.7.3 CONSEQUENCES OF MULTICOLLINEARITY
(I) The fit of the simple regression equation is unaffected.
(II) The variances and the standard errors of the parameter estimates will increase.
(III) The ordinary least square estimator is still unbiased.
3.7.4 TEST FOR MULTICOLLINEARITY
I. One of the most reliable tests for detecting multicolinearity in a data is the Farar-
Glauber test. The test is conducted in three stages as follows:
a) Chi-square test for detecting the existence and severity of multicolinearity:
27
The hypotheses are:
H0: Multicolinearity exists.
H1: Multicolinearity does not exist.
The procedure involved is:
i) Compute the matrix (rij) of the sample correlation coefficients between X 1, X2 and
X3 or independent variables.
ii) Find the determinant of the matrix rij i.e /rij/.
iii) Use Chi-square test to detect the existence and severity of multicolinearity using
this formula:
{ 1
}
χ 2= ( T −1 )− ( 2 k +5 ) log e D
6
Where T = number of observations.
K= number of X variables.
iv) Check χ2αwith ½ k (k-1) degrees of freedom. (α is the level of significance).
(b) F-test to locate the variables that are intercorrelated or collinear.
 Compute the multiple correlation among the explanatory variables.
 Test the statistical significant coefficient with F-test.
 The combined use of r2, R2 and standard error of coefficient, if the explanatory
variable neither improve nor impair R2, it can be assumed that the variable is
28
useless and it will be eliminated from the equation. Then, if the new variable
improves R2 and impairs the quality of parameter estimate then the variable has
introduced multicolinearity in the equation.
3.7.5 SOLUTION TO MULTICOLLINEARITY
i. Change of functional form

ii. Addition of equations to model
iii. Exclusion of collinear variables
iv. Extraneous information techniques
v. Increase in sample size
3.8 AUTOCORRELATION
The least square regression model assumes that the error or disturbances (U 1)
corresponding to different observations are independent pair wise uncorrelated (that is
there is no correlation between the error terms). If this assumption fails to hold, the result
is autocorrelation between members of series of observations ordered in time or space, it
is also referred to as serial correlation.
Autocorrelation is more common in time series studies when the errors associated
with observations in a given time period is carried over into the future time periods. The
serial correlation can either be positive or negative depending on the data.
3.8.1 CAUSES OF AUTOCORRELATION
1. Missing or omission of explanatory variables which are relevant to the model.
29
2. Fitting the wrong model into a data e.g. using a linear model when non-linear
model is appropriate.
3. Interpolation in the statistical observations.
4. Presence of random error in the data used.
5. Mis-specification of error term U in the dependent variable Y.
3.8.2 CONSEQUENCES OF AUTOCORRELATION
 The estimated variances and covariance of the OLS estimate are biased and
inconsistent.
 The “F” and “t” test become invalid because they are likely to overstate the
statistical significance of parameters.
 The least square parameter estimates, may not be precise especially when sample
is small.
 The best linear unbiased estimator claimed for the least squares parameter
estimate may no longer hold. This is as a result of autocorrelation error which may
yield large standard error.
3.8.3 TEST FOR AUTOCORRELATION
30
There are various ways by which autocorrelation can be detected. We have:
(i) The scatter diagram method: this is a method of testing for the existence or
otherwise of autocorrelation. In the method, the residual U i = (Yi-Ŷi) are plotted against
time.
(ii) Trial Estimation of Autoregressive Equation: this method is followed when there
is no prior basis for assuming a particular form of autocorrelation.
(iii) The Durbin-Watson coefficient, d, tests for autocorrelation: the value of the ranges
from 0 to 4, values close to 0 indicates extreme positive autocorrelation, close to 4
indicates extreme negative autocorrelation, and close to 2 indicates no serial
autocorrelation. As a rule of thumb, d should be between 1.5 and 2.5 to indicate
independence of observations. Positive autocorrelation means standard errors of the ẞ
coefficient are too small. Negative autocorrelation means standard errors are too large.
Hence, it is the most reliable and conventional.
The procedure is as follows:
(i) Define the hypothesis
H0: d =2. There is no autocorrelation.
H1:d≠2. There is autocorrelation.
(ii) Fit in the estimated parameters into the equation that is :
Ŷ=ẞ0+ẞ1X1+ẞ2X2+ẞ3X3
(iii) List all the values of observations under Y and calculate Ŷ (estimated Y’s) for the
given values of X’s.
31
(iv) Calculate et’s which is Y-Ŷ and list them out.
(v) Calculate et-et-1 and list them out.
(vi) Calculate (et-et-1)2 and find its sum; ∑(et-et-1)2.
(vii) Calculate et2 and hence ∑et2.
The Durbin-Watson test is given as:
2
n
( U t−1−U t −2 )
D=∑ ❑
i=2
∑
❑
2
Ut
Consequently, the relationship between ρ and D shows that;
d = Σ¿ ¿ ¿
2 2
(U ¿¿ t −2 U t U t −1+U t −1)
d= Σ 2
¿
Σ Ut
But recall that, U t =U t −1
2
(U t −2U t U t−1 +U t )
d= Σ 2
ΣU t
2
(U ¿¿ t −2 U t U t −1)
d= Σ 2
¿
Σ Ut
2
ΣU t ΣU t U t−1
d=2 2
− 2
ΣU t Σ Ut
32
Σ U t U t−1
Also recall that, 2 = ρ, hence
Σ Ut
d = 2(1-ρ)
Interpretation goes thus:
Range of d: 0≤d≤4
Uncorrelated residual=2
Positive autocorrelation d<2
Strong positive autocorrelation d=0
Negative autocorrelation d>2
Strong negative autocorrelation d= 4.
3.8.4 CORRECTING AUTOCORRELATION
Autocorrelation can be corrected by transforming the original data into a new
model. The step by step procedure involved is stated below:
1. Replace the original regression model:
Y=ẞ0+ẞ1X1i+ẞ2X2i+……………+ẞkXki + ei
With model Yi-rYi-1=ẞ0(1-i) + ẞ1(X1i-rX1i-1) + ẞ2(X2i-rX2i-1)+…..+ẞk(Xki-rXki-1) +Ui
Where r =
∑ U t U t−1
∑ U 2t−1
1. Calculate r and substitute in the equation.
2. Use the values of Y’s and X's in the original data to calculate the new values.
33
Y*=Yi-rYi-1
X*1=X1i-rX1i-1 and so on for X2 and X3.
3. Use the least square estimation to obtain the parameters b 0,b1,b2,b3 to get the new
model: Y=ẞ0+ẞ1X1i+ ẞ2X2i +…………..+ẞkXki
4. Test for autocorrelation in the transformed data.
3.9 HETEROSCEDASTICITY
When the assumption that the disturbance variance is constant at each observation
point is strongly violated, then we have heteroscedasticity, i.e.
E(UU1)=σ21 (assumption of spherical disturbance) does not hold. It is more common with
cross-sectional data than with time series data. Rather than remaining constant, the
variance of error varies with the X’s.
3.9.1 CONSEQUENCES OF HETEROSCEDASTICITY
1. Parameter estimates b=(X1X)-1X1Y are unbiased.
2. The least squares formula for computing the variance of parameter estimates
maybe unreliable due to heteroscedastic term.
3. Prediction based on the least squares estimates maybe inefficient and unreliable.
4. Variance of parameter estimate will tend to be very large.
34
3.9.2 TESTING FOR HETEROSCEDASTICITY
The methods used for testing for the presence of heteroscedasticity are:
i. Scatter diagram.
ii. Spearman’s Rank Correlation analysis.
iii. Goldfield Quandit test.
iv. Glejser test.
3.9.2.1 GOLDFIELD QUANDIT TEST: This test is applicable to large samples. The
number of observation must be more than twice the parameters to be estimated. The test
assumes normality and serially independent Ui. The hypothesis to be tested is:
H0: U’s are Homoscedastic

H1: U’s are Heteroscedastic
3.9.2.2 SPEARMAN’S RANK CORRELATION ANALYSIS
In this project work, the method of Spearman’s rank correlation analysis will be
used in carrying out the necessary calculations.
3.10 OVERCOMING HETEROSCEDASTICITY
If the test is completed and heteroscedasticity is confirmed, then it can be removed by:
1. Transformation of the original data to reduce heteroscedasticity in the residuals of
the variances.
2. Deflating all the variables by some measure of size.
35
CHAPTER FOUR: PRESENTATION OF DATA AND ANALYSIS
4.1 PRESENTATION OF DATA
Table 1: Data on Variables of Interest
Year INFLATION PETROL DIESEL KEROSENE

1990 7.5 0.6 0.5 0.4
1991 12.7 0.7 0.5 0.5
1992 44.81 0.7 0.55 0.5
1993 57.17 3.25 3 2.75
1994 53.03 11 9 6
1995 72.81 11 9 6
1996 29.29 11 9 6
1997 10.67 11 9 6
1998 7.86 11 9 6
1999 6.62 20 19 17
2000 6.94 22 21 17
2001 18.87 22 21 17
2002 12.89 26 26 24
2003 14.03 40 38 38
2004 15.01 49 48 48
2005 17.85 65 60 50
2006 8.24 65 60 50
2007 5.38 65 60 50
2008 11.6 65 60 50
2009 11.5 65 145 50
2010 11.9 65 145 50
2011 12.6 65 145 50
2012 10.9 67 191 51
2013 9.2 87 230 53
2014 7.7 87 278 55
36
4.2. DESCRIPTIVE STATISTICS OF VARIABLES
Table 2: Descriptive statistics of Inflation, Petrol, Diesel and Kerosene (1990 – 2014)
5.
DIESEL INFLATION KEROSENE PETROL
Mean 63.86200 19.08280 28.16600 37.41000
Median 26.00000 11.90000 24.00000 26.00000
Maximum 278.0000 72.81000 55.00000 87.00000
Minimum 0.500000 5.380000 0.400000 0.600000
Std. Dev. 78.59785 18.04948 21.87191 29.49308
Skewness 1.394798 1.816933 -0.043572 0.204768
Kurtosis 3.840134 5.068165 1.206786 1.515529
Jarque-Bera 8.841324 18.21072 3.357510 2.470182
Probability 0.012026 0.000111 0.186606 0.290808
Sum 1596.550 477.0700 704.1500 935.2500
Sum Sq. Dev. 148262.9 7818.806 11481.13 20876.20
Observations 25 25 25 25
The summary of the descriptive statistics of the variables of this study are
presented in Table 2. The skewness is an indicator of the asymmetry or deviation of the
variables from a normal distribution with an expected value of zero. The kurtosis defines
the degree of flattening or peakedness of a distribution with an expected value of three.
Jarque-Bera statistic determines the normality or otherwise of a distribution.
The variables Inflation, Diesel and Petrol have a skewness greater than zero
(Positively skewed) while Kerosene has a skewness lesser than zero (Negatively skewed),
kurtosis of two particular variables (Kerosene and Petrol) is merely average (platykurtic
37
since the values are less than 3) while two other variables (Inflation and Diesel) are
leptokurtic since they both greater than 3 and their Jarque-Bera statistic denotes that its
errors are not normally distributed except that of Diesel and Inflation (Significant with p-
values of 0.012 and 0.00 respectively).
4.3 TESTING FOR STATIONARITY OF THE DATA USING GRAPHICAL

APPROACH AND UNIT ROOT TEST
The figures (1-12) shows the graphs illustrating whether the data for the data is
stationary or not. At level, first difference and second difference the series are not
stationary, but at third difference, it was fairly stationary. So als in the unit root tables
(tables 1-12) the augmented Dickey Fuller test conducted shows that the series is only
stationary at third difference. The implication of this is that the analysis could be
performed due to stationarity achieved.
Graph of the original data, first difference, second difference and third difference
are given in the figures below:
Figure 1:
PETROL
90
80
70
60
50
40
30
20
10
0
90 92 94 96 98 00 02 04 06 08 10 12 14
38
Figure 2:
DIESEL
300
250
200
150
100
50
0
90 92 94 96 98 00 02 04 06 08 10 12 14
Figure 3:
KEROSENE
60
50
40
30
20
10
0
90 92 94 96 98 00 02 04 06 08 10 12 14
Figure 4:
ISTDIFPET
24
20
16
12
0
90 92 94 96 98 00 02 04 06 08 10 12 14
39
Figure 5:
ISTDIFDIESEL
90
80
70
60
50
40
30
20
10
0
90 92 94 96 98 00 02 04 06 08 10 12 14
Figure 6:
ISTDIFKERO
16
14
12
10
0
90 92 94 96 98 00 02 04 06 08 10 12 14
Figure 7:
SECONDDIFPET
20
10
-10
-20
-30
90 92 94 96 98 00 02 04 06 08 10 12 14
Figure 8:
40
SECONDDIFDIESEL
100
75
50
25
-25
-50
-75
-100
90 92 94 96 98 00 02 04 06 08 10 12 14
Figure 9:
SECONDDIFKERO
12
-4
-8
-12
90 92 94 96 98 00 02 04 06 08 10 12 14
Figure 10:
THIRDDIFPET
20
10
-10
-20
-30
-40
90 92 94 96 98 00 02 04 06 08 10 12 14
Figure 11:
41
THIRDDIFDIESEL
100
50
-50
-100
-150
-200
90 92 94 96 98 00 02 04 06 08 10 12 14
Figure 12:
THIRDDIFKERO
12
-4
-8
-12
-16
-20
-24
90 92 94 96 98 00 02 04 06 08 10 12 14
The unit root test of the original data, first difference, second difference and third
difference are given in the tables below:
Table 1:
Null Hypothesis: PETROL has a unit root
Exogenous: Constant
Lag Length: 0 (Automatic - based on AIC, maxlag=5)
t-Statistic Prob.*
42
Augmented Dickey-Fuller test statistic 0.219653 0.9681
Test critical values: 1% level -3.737853
5% level -2.991878
10% level -2.635542
*MacKinnon (1996) one-sided p-values.
Augmented Dickey-Fuller Test Equation

Dependent Variable: D(PETROL)
Method: Least Squares
Date: 08/10/15 Time: 08:56
Sample (adjusted): 1991 2014
Included observations: 24 after adjustments
Variable Coefficient Std. Error t-Statistic Prob.
PETROL(-1) 0.009737 0.044330 0.219653 0.8282

C 3.255851 1.988578 1.637276 0.1158
R-squared 0.002188 Mean dependent var 3.600000

Adjusted R-squared -0.043167 S.D. dependent var 5.873800
S.E. of regression 5.999237 Akaike info criterion 6.500797
Sum squared resid 791.7985 Schwarz criterion 6.598968
Log likelihood -76.00956 Hannan-Quinn criter. 6.526842
F-statistic 0.048247 Durbin-Watson stat 1.787535
Prob(F-statistic) 0.828167
Table 2:
Null Hypothesis: DIESEL has a unit root
Exogenous: Constant
t-Statistic Prob.*

5% level -3.029970
10% level -2.655194

Warning: Probabilities and critical values calculated for 20 observations
and may not be accurate for a sample size of 19

Dependent Variable: D(DIESEL)
Date: 08/10/15 Time: 08:57
43
DIESEL(-1) 1.759421 0.415395 4.235538 0.0012
D(DIESEL(-1)) -2.753029 0.625457 -4.401626 0.0009
D(DIESEL(-2)) -2.982338 0.694841 -4.292115 0.0010
D(DIESEL(-3)) -2.533460 0.711852 -3.558968 0.0039
D(DIESEL(-4)) -2.040084 0.636638 -3.204464 0.0076
D(DIESEL(-5)) -1.359102 0.528621 -2.571033 0.0245
C -0.042325 4.745895 -0.008918 0.9930

Adjusted R-squared 0.649508 S.D. dependent var 23.42189
Table 3:
Null Hypothesis: KEROSENE has a unit root
Exogenous: Constant
t-Statistic Prob.*
Augmented Dickey-Fuller test statistic -0.882420 0.7752

5% level -2.998064
10% level -2.638752

Dependent Variable: D(KEROSENE)
Date: 08/10/15 Time: 08:58
KEROSENE(-1) -0.034084 0.038626 -0.882420 0.3880

D(KEROSENE(-1)) 0.411334 0.200615 2.050369 0.0537
C 2.390262 1.397999 1.709774 0.1028
44
Table 4:
Null Hypothesis: ISTDIFPET has a unit root
Exogenous: Constant
t-Statistic Prob.*

5% level -2.998064
10% level -2.638752

Dependent Variable: D(ISTDIFPET)
Date: 08/10/15 Time: 09:01
ISTDIFPET(-1) -0.897983 0.217184 -4.134672 0.0005

C 3.368944 1.505083 2.238377 0.0362
R-squared 0.448754 Mean dependent var -0.004348

Table 5:
Null Hypothesis: ISTDIFDIESEL has a unit root
Exogenous: Constant
t-Statistic Prob.*

45
5% level -3.040391
10% level -2.660551


Dependent Variable: D(ISTDIFDIESEL)
Date: 08/10/15 Time: 09:02
ISTDIFDIESEL(-1) 3.718642 1.216323 3.057282 0.0109

D(ISTDIFDIESEL(-1)) -5.194271 1.297677 -4.002746 0.0021
D(ISTDIFDIESEL(-2)) -5.614667 1.374705 -4.084269 0.0018
D(ISTDIFDIESEL(-3)) -5.101799 1.311522 -3.889983 0.0025
D(ISTDIFDIESEL(-4)) -4.340800 1.209756 -3.588162 0.0043
D(ISTDIFDIESEL(-5)) -3.298580 1.075009 -3.068421 0.0107
C -4.388543 6.679287 -0.657038 0.5247

Table 6:
Null Hypothesis: ISTDIFKERO has a unit root
Exogenous: Constant
t-Statistic Prob.*

5% level -2.998064
10% level -2.638752

Dependent Variable: D(ISTDIFKERO)
Date: 08/10/15 Time: 09:03
46
ISTDIFKERO(-1) -0.603951 0.198809 -3.037837 0.0063

C 1.463817 0.918232 1.594170 0.1258

Table 7:
Null Hypothesis: SECONDDIFPET has a unit root
Exogenous: Constant
t-Statistic Prob.*

5% level -3.004861
10% level -2.642242

Dependent Variable: D(SECONDDIFPET)
Date: 08/10/15 Time: 09:11
SECONDDIFPET(-1) -1.574712 0.233951 -6.730953 0.0000

C 0.519853 1.578569 0.329319 0.7453

Table 8:
Null Hypothesis: SECONDDIFDIESEL has a unit root
Exogenous: Constant
47
t-Statistic Prob.*

5% level -3.029970
10% level -2.655194


Dependent Variable: D(SECONDDIFDIESEL)
Date: 08/10/15 Time: 09:12
SECONDDIFDIESEL(-1) -5.442472 1.140443 -4.772245 0.0003

D(SECONDDIFDIESEL(-1)) 3.208540 0.944295 3.397814 0.0043
C 7.925485 4.843844 1.636197 0.1241

Table 9:
Null Hypothesis: SECONDDIFKERO has a unit root
Exogenous: Constant
t-Statistic Prob.*

5% level -3.012363
10% level -2.646119

Dependent Variable: D(SECONDDIFKERO)
Date: 08/10/15 Time: 09:13
48
SECONDDIFKERO(-1) -1.545208 0.323807 -4.772000 0.0002

D(SECONDDIFKERO(-1)) 0.382505 0.216615 1.765827 0.0944
C 0.019984 0.986628 0.020255 0.9841

Table 10:
Null Hypothesis: THIRDDIFPET has a unit root
Exogenous: Constant
t-Statistic Prob.*

5% level -3.052169
10% level -2.666593


Dependent Variable: D(THIRDDIFPET)
Date: 08/12/15 Time: 09:28
THIRDDIFPET(-1) -6.415305 1.696978 -3.780430 0.0030

D(THIRDDIFPET(-1)) 3.933011 1.508707 2.606876 0.0244
D(THIRDDIFPET(-2)) 2.655810 1.161546 2.286444 0.0430
D(THIRDDIFPET(-3)) 1.535700 0.768015 1.999571 0.0709
D(THIRDDIFPET(-4)) 0.549017 0.347258 1.581006 0.1422
C 0.638134 2.415330 0.264202 0.7965

49
Table 11:
Null Hypothesis: THIRDDIFDIESEL has a unit root
Exogenous: Constant
t-Statistic Prob.*

5% level -3.052169
10% level -2.666593


Dependent Variable: D(THIRDDIFDIESEL)
Date: 08/12/15 Time: 09:29
THIRDDIFDIESEL(-1) -14.86222 2.827394 -5.256508 0.0003

D(THIRDDIFDIESEL(-1)) 11.87998 2.664333 4.458893 0.0010
C 4.597885 5.759905 0.798257 0.4416

Table 12:
Null Hypothesis: THIRDDIFKERO has a unit root
Exogenous: Constant
t-Statistic Prob.*
50
5% level -3.040391
10% level -2.660551


Dependent Variable: D(THIRDDIFKERO)
Date: 08/12/15 Time: 09:31
THIRDDIFKERO(-1) -4.101116 0.904033 -4.536469 0.0006

D(THIRDDIFKERO(-1)) 2.152687 0.719874 2.990369 0.0104
D(THIRDDIFKERO(-2)) 1.199490 0.485114 2.472592 0.0280
D(THIRDDIFKERO(-3)) 0.421042 0.247340 1.702282 0.1125
C 0.128968 1.353157 0.095309 0.9255

4.4 REGRESSION ANALYSIS
Table 3 shown below contains multivariate regression results for the regression
model. The results indicated that the constant is statistically significant while the
coefficients of petrol, kerosene and diesel are statistically insignificant.
4.4.1 TEST FOR JOINT SIGNIFICANCE -- USING F-TEST
Main Hypothesis
H0: β0 = β1= β2= β3
51
H0: β0 ≠ β1≠ β2≠ β3
α = 5% = 0.05
Table 3: Output of Regression Analysis using OLS (Ordinary Least Square Method)
Dependent Variable: INF

Sample: 1990 2014
Included observations: 25
INF = C(1) + C(2) *PET + C(3)*KER + C(4)*DIE
Coefficient Std. Error t-Statistic Prob.
C(1) 30.37209 5.778957 5.255636 0.0000

C(2) 0.197102 0.876783 0.224802 0.8243
C(3) -0.642958 0.961329 -0.668822 0.5109
C(4) -0.008665 0.106631 -0.081258 0.9360

The F-statistic 2.279598, which is a measure of the joint significance of the
explanatory variables, is found to be statistically insignificant at 5 percent level as
indicated by the corresponding probability value 0.108965 which less than α (0.05), it
leads to the acceptance of the null hypothesis because of low evidence and conclude that
there is no relationship between the dependent variable and independent variables. It
52
implies that the Petrol, Kerosene and Diesel are not that related and have no great
influence on Inflation rate in Nigeria Economy.
The R2 = 0.245657 (25%) implies that 25 percent total variation in the Inflation
rate is explained by the explanatory variables (Petrol, Kerosene, and Diesel) in the
regression equation. Coincidentally, the goodness of fit of the regression remained not
high enough after adjusting for the degree of freedom as indicated by the adjusted R 2
(R2= 0.137 or 14%). After observing the Durbin-Watson statistic 0.729 in table 3 to be
higher than R2 (0.246) indicating that the model is non-spurious, so the need for a unit
root test became unnecessary.
4.5 HYPOTHESES TESTING AND INTERPRETATIONS
4.5.1 HYPOTHESIS ONE:
H0: Petrol has no significant impact on the Inflation rate of Nigeria Economy
H1: Petrol has significant impact on the Inflation rate of Nigeria Economy
α = 5% = 0.05
Table 4(a) Model Summary
Model R R Square Adjusted R Std. Error of

Square the Estimate
1 .472a .223 .189 16.25163
a. Predictors: (Constant), Petrol
Table 4(b) ANOVAa
53
Model Sum of Df Mean Square F Sig. t Sig.
Squares
Regression 1744.149 1 1744.149 6.604 .017b Const 5.623 0.000
1 Residual 6074.657 23 264.116 Petrol -2.570 0.017
Total 7818.806 24
a. Dependent Variable: Inflation
b. Predictors: (Constant), Petrol
It is observed from the above Table 4 that the probability value is less than the
level of significance i.e. sig. <α (0.017 < 0.05) which indicates that there is enough
evidence to reject the null hypothesis leading to the conclusion of statistically
significance of Individual Test (named t-test) and thereby concluded that Petrol has
significant impact on the Inflation rate of Nigeria Economy.
The R 2 = 0.223 (22%) implies that 22 percent total variation in the Inflation is
explained by the Petrol in the regression equation Yi = β0 + β1Xi.
4.5.2 HYPOTHESIS TWO:
H0: Kerosene has no significant impact on the Inflation rate of Nigeria Economy
H1: Kerosene has significant impact on the Inflation rate of Nigeria Economy
α = 5% = 0.05
Table 5(a): Model Summary

Model R R Square Adjusted R Std. Error of the
Square Estimate
1 .493a .243 .210 16.04084
a. Predictors: (Constant), Kerosine
Table 5(b) ANOVAa
54
Model Sum of df Mean Square F Sig. t Sig.
Squares
1 Residual 5918.099 23 257.309 Petrol -2. 718 0.012
Total 7818.806 24
The R2 = 0.243 (24%) implies that 24 percent total variation in the Inflation is
explained by the Kerosene in the regression equation Yi = β0 + β1Xi.
It has revealed from the above Table 5 that the probability value is less than the
level of significance i.e. sig. <α (0.012< 0.05) which indicates that there is enough
evidence to reject the null hypothesis leading to the conclusion of statistically
significance of Individual Test (named t-test) and thereby concluded that Kerosene has
4.5.3 HYPOTHESIS THREE:
H0: Diesel has no significant impact on the Inflation rate of Nigeria Economy
H1: Diesel has significant impact on the Inflation rate of Nigeria Economy
α = 5% = 0.05
Table 6 (a): Model Summary

Model R R Square Adjusted R Std. Error of the
Square Estimate
1 .361a .130 .093 17.19288
a. Predictors: (Constant), Diesel
Table 6(b) ANOVAa
55
Model Sum of df Mean Square F Sig. t Sig.
Squares
1 Residual 6798.688 23 295.595 Petrol -1.858 0.076
Total 7818.806 24
It has shown from the Table 6 above that the probability value is greater than the
level of significance i.e. sig. ˃α (0.076 ˃ 0.05) which indicates that there is enough
evidence to accept the null hypothesis leading to the conclusion of statistically
significance of Individual Test (named t-test) and thereby concluded that Kerosene has no
The R2 = 0.13 (13%) implies that 13 percent total variation in the Inflation is explained by
the Kerosene in the regression equation Yi = β0 + β1Xi.
4.6 TESTS FOR MULTICOLINEARITY
This test is to determine the presence of multicolinearity in the data.
Decision Rule:
If Variance Inflation Factor (VIF) is greater than 5, there is multicolinearity
If Variance Inflation Factor (VIF) is greater than 10, multicolinearity is severe
Table 7(a): Inflation and its components in terms of Multicolinearity (Coefficientsa)
Model Unstandardized Coefficients Standardized T Sig. Collinearity Statistics
Coefficients
56
B Std. Error Beta Tolerance VIF
(Constant) 30.372 5.779 5.256 .000
Petrol .197 .877 .322 .225 .824 .018 57.141

1
Diesel -.009 .107 -.038 -.081 .936 .167 6.002
Kerosine -.643 .961 -.779 -.669 .511 .026 37.778
The Variance Inflation Factor (VIF) of petrol, diesel and kerosene show the
presence of multicolinearity since their values are all greater than 5 (five).The remedy in
solving the problem of multicolinearity is by removing the variable with highest VIF
which is X1 (Petrol) to notice how the result will look like.
Table 7(b): Inflation and its components in terms of Multicollinearity (Coefficientsa)

Model Unstandardized Coefficients Standardized T Sig. Collinearity Statistics
Coefficients
B Std. Error Beta Tolerance VIF
(Constant) 30.679 5.493 5.585 .000
1 Diesel .010 .066 .043 .147 .884 .410 2.439
Kerosine -.434 .239 -.526 -1.816 .083 .410 2.439
Since the Variance Inflation Factor (VIF) in Diesel and Kerosene do not greater
than 5, that is the (VIF) s are both less than 5 (2.439< 5) which lead to the conclusion of
no multicolinearity in the remaining explanatory variables X 2 and X3 (Diesel and
Kerosene) .
4.7 TESTS FOR AUTOCORRELATION
57
The Durbin Watson test statistics will be used to test for the presence of
autocorrelation. If d is < 2 (Positive Autocorrelation) and If d is > 2 (Negative
Autocorrelation).
Since the d < 2, the hypothesis is given thus:-
H0: There is no positive first order autocorrelation
H1: There is a positive first order autocorrelation
α = 5% = 0.05
Model: Ŷ = 30.37209 + 0.197102X1 – 0.642958X2 – 0.008665X3
Table 8: Autocorrelation Model Summaryb
Model Change Statistics Durbin-Watson
R Square F Change df1 df2 Sig. F Change
Change
1 .246a 2.280 3 21 .109 .729
a. Predictors: (Constant), Kerosine, Diesel, Petrol
b. Dependent Variable: Inflation
Hence, since d=0.729, Positive Autocorrelation is to be tested from d=2(1-ρ)
Conclusion: Since the p-value > α (0.109> 0.05), there is no enough evidence to reject
the null hypothesis and therefore conclude that there is no positive first order
58
autocorrelation between the dependent variable (Inflation) and independent variables
(Petrol, Diesel, Kerosene).
Inflation Petrol Diesel Kerosene

*
Pearson Correlation 1 -.472 -.361 -.493*
Inflation Sig. (2-tailed) .017 .076 .012
N 25 25 25 25
Pearson Correlation -.472* 1 .854** .978**
Petrol Sig. (2-tailed) .017 .000 .000
N 25 25 25 25
Pearson Correlation -.361 .854** 1 .768**
Diesel Sig. (2-tailed) .076 .000 .000
N 25 25 25 25
Pearson Correlation -.493* .978** .768** 1
Kerosine Sig. (2-tailed) .012 .000 .000
N 25 25 25 25
*. Correlation is significant at the 0.05 level (2-tailed).
**. Correlation is significant at the 0.01 level (2-tailed).
The Table 9 above indicated that there is a relationship between Inflation and
petrol, Inflation and kerosene since sig. < α (0.17, 0.012< 0.05) with weak negative
correlation values of -0.472 and -0.493 respectively as the extent of relationship, and that
there is no relationship between Inflation and diesel since sig. > α (0.076> 0.05) with
weak negative correlation value of -0.361 as the degree of relationship.
4.9 TESTS FOR HETEROSCEDASTICITY
Table 10: Heteroscedasticity Test: Breusch-Pagan-Godfrey
F-statistic 3.541201 Prob. F(3,21) 0.0322

Obs*R-squared 8.398477 Prob. Chi-Square(3) 0.0385
Scaled explained SS 9.048143 Prob. Chi-Square(3) 0.0287
59
Test Equation:
Dependent Variable: RESID^2
Date: 06/19/15 Time: 06:51
Sample: 1990 2014
Included observations: 25
C 530.1028 126.4018 4.193792 0.0004

PET 12.03112 19.17767 0.627351 0.5372
KER -24.84867 21.02692 -1.181755 0.2505
DIE -0.694888 2.332307 -0.297940 0.7687

Sum squared resid 2821740. Schwarz criterion 14.98689
A test for Heteroscedasticity was carried out and the below Hypothesis is set;
Alternatively;
H0: The error variances are homoscedastic
H1: The error variances are heteroscedastic
α = 5% = 0.05
60
Since the P-value (0.0322) < , the null hypothesis (H0) is therefore rejected
by concluding that the error variances are not equal (heteroscedastic).
The Heteroscedasticity can also be described using Graphical Method as shown in
the Figure 4 above.
Figure 5: Actual Fitted Residual Graph
80
60 4.10
FORECASTING
60 40
EVALUATION
40 20
Table 11:
20 Forecasted Values
0
of variables for
0
the next 5 years
-20
-40
90 92 94 96 98 00 02 04 06 08 10 12 14
Residual Actual Fitted

Kerosen
Year Inflation Petrol e Diesel
6.46129 91.4962 57.0609
2015 1 4 2 335.877
5.42640 97.2479 59.2768 405.790
2016 7 5 2 4
4.55727 103.361 61.5787 490.256
2017 7 2 6 4
2018 3.82735 109.858 63.9701 592.304
61
3 8 2
3.21433 116.764 715.593
2019 9 9 66.4543 4
Figure 6: Forecasted diagram of the four variables
160
60
150
40
140
20 130
0 120
110
-20
100
-40
90
-60 80
2015 2016 2017 2018 2019
70
INFF ± 2 S.E. 2015 2016 2017 2018 2019
100 PETF ± 2 S.E.
800
90
80 700
70
600
60
500
50
40 400
30
2015 2016 2017 300 2018 2019
2015 2016 2017 2018 2019
KERF ± 2 S.E.
DIEF ± 2 S.E.
62
CHAPTER FIVE: FINDINGS, CONCLUSIONS AND RECOMMENDATIONS
5.1 FINDINGS.
After all the necessary tests have been carried out on the data, these are the
findings in accordance with the main objectives stated earlier.
The appropriate model to fit is:
Ŷ= 30.37209+0.197102X1-0.642958X2-0.008665X3
The above model can be interpreted by considering the coefficient of X variables.
For the coefficient of X1 variable, it is noticed that the value is positive i.e. (0.197102)
which shows a fairly positive relationship between Petrol and inflation rate. Also, it is
realized from the model that the coefficient of X 2 variable is -0.642958 which shows a
negative relationship between Diesel and inflation rate. Lastly, the coefficient of X 3 is -
0.008665 which implies that it does not contribute or has no effect on inflation rate.
For the individual test carried out, it was observed that;
I. Petrol has significant impact on the inflation rate of Nigeria economy.
II. Kerosene has significant impact on the inflation rate of Nigeria economy.
III. Diesel has no significance impact on the inflation rate of Nigeria economy.
63
Among the objectives of the study is to show the relationship between inflation rate
and Petro l, Diesel and Kerosene .This has been proved with the model stated above.
Moreso, Multiolinearity was detected, the variance inflation factor of Petrol ,Diesel and
Kerosene show the presence of Multicollinearity since their values are all greater than
5.the remedy in solving this problem is by removing the variable with the highest
variance inflation factor which is the X1 (Petrol) to notice what the result will look like.
The Variance Inflation (VIF)Factor in Diesel and Kerosene is not greater than 5
which lead to the conclusion that Multicollinearity does not exist in the remaining
explanatory variables X2 and X3 (Diesel and Kerosene).
Durbin Watson test was used to test for the presence of autocorrelation. From the
result, the p-value was greater than α (0.109˃0.05) which implies there is no positive first
order autocorrelation between the dependent variable (inflation rate) and independent
variables (Petrol, Diesel and Kerosene).
Also, the correlation significance table indicated that there is a relationship
between inflation rate and Petrol, Inflation rate and kerosene since sig <α (0.17,
0.012<0.05) with a weak negative correlation values of -0.472 and -0.493 respectively as
the extent of the relationship, and that there is no relationship between inflation rate and
diesel since sig >α (0.076>0.05) with weak negative correlation value of -0.361 as the
degree of relationship.
64
Lastly, the original data is not stationary and so the first, second and third
difference were done and it was stationary at the third difference.
5.2 CONCLUSION
Based on the findings stated earlier, it has been seen that there is a direct
relationship between inflation rate and Petrol, Kerosene while diesel is fairly related. The
coefficient of determination R2 implies that 25% total variation in the inflation rate is
explained by the explanatory variables (Petrol, Diesel, and Kerosene).
5.3 RECCOMMENDATION
It is obvious based on the findings that petrol and kerosene have significant impact
on the inflation rate of Nigerian economy and also contribute to the forecasting power of
inflation rate while diesel did not.
The following measures are recommended:
1. There should not be upward adjustment of petroleum products (which in time past
have resulted in inflation, high cost of living, and inequitable distribution of income).
2. It is evident that export of petroleum earns valuable foreign exchange in Nigeria, and
the links between petroleum and money are easily obvious, Government should
invest more on oil refineries to enhance operation efficiency.
3. In order to prevent instability of oil products, Government should curb the act of
smuggling oil products from Nigeria to her neighboring countries.
65
4. Government should employ experts in the area of statistical records and good analyst
for planning and determining the progress and growth of its economy.
REFERENCES
Arinze P. E. (2011) “The impact of oil price on the Nigerian economy”

www.transcampus.org/journals. www.ajol.info/journals/jorind JORIND (9)1.
Central Bank of Nigeria (2005) Economic and Financial Review, Vol.14, CBN.
Gbadamosi, N. A., Kupolokun, P. O. and Oluleye, A. B. (2007) “Deregulation of the

Nigerian Downstream Oil sector”. Abuja.
Mbendi, M. L. (2000) “Law and Petroleum Industry in Nigeria Lagos”, Nigeria: African
Books Publishers.
Olorunfemi, S. (2003) “Natural Gas option in Nigeria Industrial Development Process”,

CBN Economic and Financial Review, Vol.14.
Onwioduokitanda, E. A. and Adenuga, O. (2000) “Empirical Analysis of the Demand for

Petroleum Products in Nigeria”. C.B.N 2000 Journal Publication pp 33-57.
Nwosu, N. L. and Chioma P. (2009) “Import of Fuel Prices on Inflation: Evidences from
Nigeria”. Research Department, Central Bank of Nigeria.
Runl, O. (2010) “Nigeria is poor because of oil alternative view publication of change
Africa New work Africa Centre for Leadership Strategy and Development”,
August, 2010.
Sikkam B. S. (1999) “Inflationary Effects of the Petroleum Industry on the Nigerian

Economy” pp 128-129.
66

Analysis of Inflation Rates of Some Selected Petroleum Products

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Analysis of Inflation Rates of Some Selected Petroleum Products

Uploaded by

Copyright:

Available Formats

ANALYSIS OF INFLATION RATES OF SOME SELECTED PETROLEUM

PRODUCTS (PETROL, DIESEL AND KEROSENE) USING ECONOMETRIC

ADAM ABDULRASAQ OLUWATOYIN

MATRIC NO: 1911320026

DEPARTMENT OF MATHEMATICS AND STATISTICS, FACULTY OF PURE AND

APPLIED SCIENCES, OSUN STATE COLLEGE OF TECHNOLOGY, ESA OKE,

IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF

HIGHER NATIONAL DIPLOMA (HND) IN STATISTICS

Mrs Akinawoniran Date

External moderator Date

my source and tower of strength all through the course of my study.

their hard labor.

support. I love you more than you can ever imagine.

guys have really being friends in need.

statistics, see you all at the top.

parameter, shows that multicollinearity exist.

Table of Content vii-xi

CHAPTER ONE: INTRODUCTION

1.1 Background 1-2

1.2 Objective of the Study 3-4

1.3 Hypotheses to Be Tested 4

1.4 Source of Data 5

1.5 Scope of the Study 5

2.1 Conceptual Framework 15

2.2 Theoretical Framework 16

2.4 Appraisal of the Reviewed Literature 17

CHAPTER THREE: METHODOLOGY

3.1 Definition of Some Terms 10

3.2 Types of Econometric Model 11-12

3.3 Steps in Econometric Analysis 12-13

3.4 Types of Economic Data 13

3.5 Regression Analysis 14

3.6 Simple Linear Regression Model 14-15

3.6.1 Assumptions of Linear Regression 15-16

3.6.2 Multiple Linear Regression 17

3.6.3 Assumptions of Multiple Linear Regression Model 17-18

3.7 Coefficient of Determination (R2) 18

3.8 Econometric Problems Relating To A Single Equation 19

3.8.3 Consequences of Multicollinearity 20

3.8.4 Test for Multicollinearity 20-21

3.8.5 Solution to Multicollinearity 21

3.9.1 Causes of Autocorrelation 22

3.9.2 Consequences of Autocorrelation 23

3.9.3 Test for Autocorrelation 23-25

3.9.4 Correcting Autocorrelation 26

3.10 Heteroscedasticity 26-27

3.10.1 Consequences of Heteroscedasticity 27

3.10.2 Testing for Heteroscedasticity 27

3.10.2.1Goldfield Quandit Test 27-28

3.10.2.2 Spearmanʹs Rank Correlation Analysis 28

3.11 Overcoming Heteroscedasticity 28

3.12 Testing for Predictive Power 28-29

CHAPTER FOUR: DATA ANALYSIS AND INTERPRETATION

4.0 Data Analysis and Interpretation 30

4.1 Presentation of Data 31

4.2 Time Plot of Variables 32-33

4.3 Descriptive Statistics of Variables 34-35

4.4 Regression Analysis 35

4.4.1 Test for Joint Significance 35-37

4.5 Hypotheses Testing and Interpretation 37

4.5.1 Hypothesis One 37-38

4.5.2 Hypothesis Two 38-39