NATIONAL OPEN UNIVERSITY OF NIGERIA

INTRODUCTION TO ECONOMETRICS II
ECO 356

FACULTY OF SOCIAL SCIENCES

COURSE GUIDE

Course Developers:

Dr. Adesina-Uthman G. A.
Faculty of Social Sciences
Department of Economics
National Open University of Nigeria.
and
Okojie, Daniel Esene
E-mail: danelektrik@gmail.com
School of Postgraduate Studies (SPGS)
University of Lagos (UNILAG)
Akoka, Yaba, Lagos State
Nigeria.

Content Editor:
Prof. Ismael Ogboru
Department of Economics,
University of Jos,
INTRODUCTION TO ECONOMETRICS II ECO 306

Jos, Plateau State.

COURSE CONTENT:
Main Introduction
Course Outline
Aims
Course Objectives
Working through the Course
Course Materials
Study Units
Textbooks and Reference Resources
Assignment Folder
Presentation Plan
Assessment
Tutor-Marked Assignments (TMAs)
Concluding Examination and Grading
Marking Scheme
Overview
Makingthe Most ofthis Course
Tutors and Tutorials
Summary

NOUN 2
INTRODUCTION TO ECONOMETRICS II ECO 306

Main Introduction
ECO 306 is a logical extension of the first-semester course on regression analysis. As such, it
introduces the concept of thesimultaneous equation and their estimation. Essentially, this
course examines the possible solutions to problems arising from the breakdown of the
ordinary least squares assumptions and sampling theories. To this end, it covers topics
likemulticollinearity, heteroscedasticity, autocorrelation and Econometrics Modeling:
Specification and Diagnostic Testing. It also examines the use of regression analyses,
correlation, variance and dummy variables.For this reason, experiential case studies that
apply the techniques to real-life data are stressed and discussed throughout the course,
and students are required to get acquaintedwith their several models and theories that
deal with the measurement of economic relationships.
The course would be a very useful material to you in your academic pursuit and could
helpto broaden your understanding further in this case. Once this understanding and
application areestablished, you are then able to have a broadened knowledge of
econometrics while distinguishing it from mathematical economics.
This course is therefore developed in a manner to guide you further on what
econometrics entails, what course materials in line with a course learning structure
you will be using. The learning structure suggests some general guidelines for a time
frame required of you on each unit to achieve the course aims and objectives
effectively. Further work in this course would expose you to introductory levels of
topics like; vector autoregressions, unit roots, cointegration, time-series analysis and
errors in variables.
Course Outline
ECO 306 is made up of five modules with seventeen units spread across twelve
lectures weeks. The modules cover areas such as theconcept of thesimultaneous
equation and their estimation, ordinary least squares assumptions, multicollinearity,
heteroscedasticity, autocorrelation and econometrics modeling: Specification and
Diagnostic Testing, use of dummy variables and time-lags as independent variables.
Aims
The aimof this course is to give you thorough understanding and an appreciative
importance of econometrics being concerned with more than measurement in
economics. But more importantly, how econometrics as a method of causal inference
isapplied to economics. That is, this method of causal inference is a statistical
inference combined with the logic of causal order; which is to infer or learn something
about the real world by analysing a sample of data.
Specifically, the aims of the course are to:
 Equip you with the application of statistical methods to the measurement and
critical assessment of assumed economic relationships using data.
 Provide an improvedintroductory understanding of how the economy works, at
either the microeconomic or macroeconomic level.

NOUN 3
INTRODUCTION TO ECONOMETRICS II ECO 306

Course Objectives
To achieve the aims mentioned above alsoto the overall stated course objectives.
Each unit, in the beginning, has its specific objectives. You should read them before
you start working through the unit. You may want to refer to them during your study
of the unit to check on your progress and should always take a look back at the
objectives after completion. In this way, you can be certain you have done what was
necessary to you by the unit. The course objectives are set below for you to achieve
the aims of the course. On successful conclusion of the course, you should be able
to:
 Know the basic principles of econometric analysis
 Express relationships between economic variables using mathematical concepts and
theories
 Understand both the fundamental techniques and wide array of applications
involving linear regression estimation
 Analyse the strengths and weaknesses of the basic regression model.
 Outline the assumptions of the normal linear regression model and discuss the
significance of these assumptions
 Explain the method of ordinary least squares
 Test hypotheses of model parameters and joint hypotheses concerning more than
one variable
 Discuss the consequences of multicollinearity, the procedures for identifying
multicollinearity, and the techniques for dealing with it
 Explain what ismeant by heteroscedasticity, and the consequences for ordinary
least square (OLS) estimators and prediction based on those estimators
 Assess the methods used to identify heteroscedasticity, including data plots and
more formal tests, and the various techniques to deal with heteroscedasticity,
including model transformations and estimation by weighted least squares
 Explain autocorrelation, and discuss the consequences of autocorrelated
disturbances for the properties of OLS estimator and prediction based on those
estimators
 Outline and discuss the methods used to identify autocorrelated disturbances, and
what can be done about it, including estimation by generalised least squares
 Discuss the consequences of disturbance terms not being normally distributed, tests
for non-normal disturbances, and methods to deal with non-normal disturbances,
including the use of dummy variables
 Discuss the consequences of specifying equations incorrectly
 Discuss the tests used to identify correct model specification and statistical criteria
for choosing between models
Working through the Course
This course highlights on critical thinking and the application of both logical and
quantitative skills.It also stresses on the application of econometric methods to
economic theory and practical problems. Therefore, to complete this course, you are
required to read the study units, referenced books and other materials on the course.
Each unit contains self-assessment exercises called Student Assessment Exercises
(SAE). At some points in the course, you will be required to submit assignments for
assessment purposes. At the end of the course, there is a final examination. This

NOUN 4
INTRODUCTION TO ECONOMETRICS II ECO 306

course should take about twelve weeks to complete, and some components of the
course areoutlined under the course material subsection.
Course Material
The major component of the course, what you have to do and how you should allocate
your time to each unit to complete the course successfully and on time are as follow:
1. Course guide
2. Study unit
3. Textbook
4. Assignment file
5. Presentation schedule

Study Unit
In this course, there are five modules that subdivided into 17 units which should be
studied thoroughly.

Module 1: Sampling Theory, Variance, and Correlation

Unit 1: Random Variables and Sampling Theory
Unit 2: Covariance and Variance
Unit 3: Correlation

Module 2: Regression Models, Hypotheses Testing, and Dummy Variables

Unit 4: Simple Regression Analyses
Unit 5: Properties of the Regression Coefficients and Hypothesis Testing
Unit 6: Multiple Regression Analysisand Multicollinearity
Unit 7: Transformations of Variables
Unit 8: Dummy Variables
Unit 9: Specification of regression variables: A preliminary skirmish

Module 3: Heteroscedasticity/Heteroskedasticity
Unit 10: Heteroscedasticity and Its Implications
Unit 11: Solution to Heteroscedasticity Problem
Unit 12: Other Tests

Module 4: Autocorrelation, Error, and Econometric Modelling

Unit 13: Stochastic Regression and measurement errors
Unit 14: Autocorrelation
Unit 15: Econometric Modellingand Models Using Time Series Data

Module 5: Simultaneous Equation, Binary Choice, and Maximum Likelihood Estimation

Unit 16: Simultaneous Equations
Unit 17: Binary Choice and Maximum Likelihood Estimation.

The general aim of module 1 (units 1-3) is to provide you with a thorough
understanding of the basic statistical tools needed for regression analyses in the
subsequent modules. The Random variables and sampling theory, covariance,

NOUN 5
INTRODUCTION TO ECONOMETRICS II ECO 306

variance, and correlationaredemystified for proper understanding. By the end of this

module, you would have been able to understand the basics of regression analysis.
Module 2 (units 4-9) explains single-equation regression models. It shows how a
hypothetical linear relationship between two variables can be quantified using
appropriate data. The principle of least squares regression analysis explained, and
expressions for the coefficients are derived.Multicollinearityand multiple regression
analysis looked at in units 6. Transformations of Variablesdiscussed in unit 7 while
dummy variables as well preliminary skirmish of the specification of regression
variables are the topics in units 8 and 9. An exploration of what happens when there is
a violation of one of the classical assumptions; equal variances (homoscedastic) is
carried out in module 3. It demonstrates how properties of estimators of the regression
coefficients depend on the properties of the disturbance term in the regression model.
Also, in this module, we shall look at some of the problems that arise when violations
of the Gauss–Markov conditions; the assumptions relating to the disturbance term, are
not satisfied. Basic understanding of heteroscedasticity (unequal-variances) will gain
thoroughexplanation.
The module 4 (unit 13-15) covers an understanding of the basics of econometric
modelling. It goes further to give some details on stochastic regression and
measurement errors, autocorrelation, econometric modelling and models using time
series data. More detail description of an introduction to Consequences of
Measurement Errors.Intercorrelation among the Explanatory Variables and
Measurement Errors in the Dependent Variable are brought to the students‟
knowledge here. Also, possible causes of Autocorrelation and Detection of First-Order
Autocorrelation using the Durbin–Watson Test are presented in units 14and 15 of the
same module 4. While module 5 with units 16 and 17provide you with a thorough
understanding of the basic rudiments of Simultaneous Equation, Binary Choice, and
Maximum Likelihood Estimation.
Respectively, study unit will take at least two hours which include anintroduction,
objective, main content, examples, In-Text Questions (ITQ) and their solutions, self-
assessment exercise, conclusion, summary, and reference. Additional areas border on
the Tutor-Marked Assessment (TMA) questions. Some of the ITQ and self-assessment
exercise will require you free-associating and solve with some of your colleagues.
You are advised to do so to grasp and get familiar with how significant econometrics
is in being concerned with measurement and also as a method of causal inference
application to economics.
There are also econometrics materials, textbooks under the reference and other (on-
line and off-line) resources for further studies. These are intended to give you extra
facts whenever you allow yourself of such prospect. You are required to study the
materials; practise the ITQ, self-assessment exercise and TMA questions for better
and thorough understanding of the course. In doing these, the identified learning
objectives of the course would have been attained.
For further reading in this course, the following reference texts and materials are
suggested:

NOUN 6
INTRODUCTION TO ECONOMETRICS II ECO 306

Textbooks and References

Robert D. Coleman, 2006, The Aims and Methodology of Econometrics Harvard
Business School, USA
Gujarati, Damodar N., 1988, Basic Econometrics, Second Edition. New York:
McGraw-Hill
Dougherty C., 2014, Elements of Econometrics; anUndergraduate study in
Economics, Management, Finance and the Social Sciences, London School of
Economics and Political Science, Oxford Revised Edition.
Hill, R. Carter, William E. Griffiths and George G. Judge, 2001, Undergraduate
Econometrics, second edition. New York: John Wiley & Sons
Maddala, G.S., 1992, Introduction to Econometrics, second edition. New York:
Macmillan Publishing Company.
Assignment Folder
The assignments given in this course are for you to attempt all of them by following
the timetable recommended regarding when to do them and submission of same for
grading by your lecturer. The marks you obtain for these assignments will count
toward the final mark you obtain for this course. Further information on assignments
will be found in the Assignment File itself and later in this Course Guide in the
section on Assessment.

There are fiveassignments in this course:

Assignment 1 - All TMAs‟ question in Units 1 – 3 (Module 1)
Assignment 2 - All TMAs' question in Units 4 – 8 (Module 2)
Assignment 3 - All TMAs' question in Units 9 – 11 (Module 3)
Assignment 4 - All TMAs' question in Units 12 – 14 (Module 4)
Assignment 5 - All TMAs' question in Units 15 – 17 (Module 5)

Presentation Plan
The presentation plan included in your course materials gives you the important dates
in the year for the completion of tutor-marking assignments and tutorial attendance.
Remember, you are required to submit all your assignments by thedue date. You
should guide against dropping behind in your assignments submission.

Assessment
Two types of assessments are available in this course; Tutor-Marked Assignment and
a written examination at the end of the course.

For the assignments, you are expected to apply lessons learnt during the course. The
assignments must be submitted to your lecturer for proper valuation in agreement with
the deadlines stated in the Presentation Schedule and the Assignments File. The
assignment works you are to submit to your lecturer for evaluation would count for
30% of your total course grade.

NOUN 7
INTRODUCTION TO ECONOMETRICS II ECO 306

At the end of the course, you will need to sit for a final written examination of three
hours duration. This examination will also count for 70% of your total course grade.

Tutor-Marked Assignments (TMAs)

There are six tutor-marked assignments in this course. You will submit all the
assignments. You are encouraged to attempt all the questions carefully. The TMAs
constitute 30% of the total marks.

Assignment questions for the units contained in this course are in the Assignment File.
You will be able to complete your assignments from the information and materials
contained in your textbooks and study units. However, it is desirable that you
demonstrate that you have readand solved a lot of problems relating to each topic in a
module. You could use other reference materials to have a broader viewpoint of each
subject in this course.

When you have completed each assignment, send it together with a TMA form to your
lecturer. Make sure that each assignment reaches your lecturer on or before the due
dates given in the Presentation File. If for any reason, you cannot complete your
assignment on time, contact your lecturer before the assignment is due, so as to
discuss the possibility of an extension. Extensions will not be granted after the due
date unless there are exceptional circumstances.

Concluding Examination and Grading

Final examination on the course will be for three hours duration and has a value of
70% of the total course grade. The examination will consist of questions which reflect
the types of self-assessment practice exercises and tutor-marked problems you have
previously encountered. There is an evaluation of all areas of the course.
You are advised to use the time between finishing the last unit and sitting for the
examination to revise the entire course materials. You might find it useful to review
your In-Text Questions (ITQ) and self-assessment exercises, tutor-marked
assignments and comments on them before the examination. The final examination
covers the entire course outline.

Marking Scheme
Table 1 presents the total marks (100%) allocation.

Table 1: Mark Allotment

Assessment Marks

Assignment (Best three assignment out of the five marked) 30%

Final Examination 70%

Total 100%

NOUN 8
INTRODUCTION TO ECONOMETRICS II ECO 306

Overview
Table 2 shows the units, number of weeks and assignments to be taken by you to
complete the course successfully;Introduction to Econometrics (ECO 306).

NOUN 9
INTRODUCTION TO ECONOMETRICS II ECO 306

Table 2: Assignment Schedule

* Comprise of a single module (Module 3) not broken into theunit.

Week’s
Unit Unit Title Activity Assessment (end of unit)
Course Guide
Sampling Theory, Variance, and Correlation

1 Random variables and sampling theory Week 1
2 Covariance and Variance Week 2
Correlation
3 Week 3 Assignment 1
Regression Models, Hypotheses Testing, and Dummy Variables

4 Simple Regression Analyses Week 4
Properties of the regression coefficients and hypothesis testing

5 Week 5
6 Multiple regression analysis and Multicollinearity Week 6
7 Transformations of Variables Week 7
8 Dummy Variables Week 8
9 Specification of regression variables: A preliminary skirmish Week 9 Assignment 2

 Heteroscedasticity/Heteroskedasticity
*10 Heteroscedasticity and its Effects Week 10
*11 Solution to Heteroscedasticity Problem Week 11
*12 Other Tests Week12 Assignment 3

 Autocorrelation, Error and Econometric Modelling

13 Stochastic Regression and measurement errors Week 13
14 Autocorrelation Week 14
15 Econometric Modellingand Models Using Time Series Data Week 15 Assignment 4

 Simultaneous Equation, Binary Choice, and Maximum Likelihood Estimation

16 Simultaneous Equation Week 16
17 Binary Choice and Maximum Likelihood Estimation Week 17 Assignment 5
Total 17 Weeks
Examination
Making the Most of this Course
An advantage of the distance learning is that the study units replace the university
lecturer. You can read and work through specially designed study materials at your
tempo and at a time and place that goes well with you.
Consider doing it yourself insolving and providing solutions to econometric problems
in the lecture instead of listening and copying solution being provided by a lecturer. In
the same way, that a lecturer might set you some practice exercisesand ITQ to do, the
study units tell you when to solve problems and read your books or other material, and
when to embark on adiscussion with your colleagues. Just as a lecturer might give you
an in-class exercise, your study units provide exercises for you to do at appropriate
points.

NOUN 10
INTRODUCTION TO ECONOMETRICS II ECO 306

Each of the study units follows a common format. The first item is an introduction to
the subject matter of the unit and how a particular unit is integrated with the other
units and the course as a whole. Nextis a set of learning objectives. These objectives
let you know what you should be able to do by the time you have completed the unit.
You should use these objectives to guide your study. When you have finished the unit,
you must go back and check whether you have achieved the objectives. If you make a
habit of doing this, you will significantly improve your chances of passing the course
and getting the best grade.

The main body of the unit guides you through the required understanding from other
sources. This will usually be either from your textbooks or a reading section. Some
units require you to undertake apractical overview of real life econometric events.
You will find when you need to embark on discussion and guided through the tasks
you must do.
The purpose of the practical overview of real life econometric events is in twofold.
First, it will enhance your understanding of the material in the unit. Second, it will
give you practical experience and skills to evaluate economic arguments, and
understand the roles of econometric in guiding current economic problems,
measurements, analysis, solutions and debates outside your studies. In any event, most
of the critical thinking skills you will develop during studying are applicable in
normal working practice, so it is important that you encounter them during your
studies.

Self-assessments are available throughout the units, and answers are at the ends of the
units. Working through these tests will help you to achieve the objectives of the unit
and prepare you for the assignments and the examination. You should do each self-
assessment exercises as you come to it in the study unit. Also, ensure to master some
major econometricstheorems and models while studying the material.

The following is a practical strategy for working through the course. If students
encounter any difficulties they may consult the lecturer concerned. Remember that
part of the lecturers‟ responsibility is to help the students when necessary. When in
need of help, do not hesitate to contact the lecturer though the available channels, for
the required assist.

1. Read this Course Guide thoroughly.

2. Organize a study schedule. Refer to the `Course overview' for more details.
Note the time you are expected to spend on each unit and how the assignments
relate to the units. Important information, e.g. details of your tutorials, and the
date of the first day of the semester is available from study centre. You need to
gather together all this information in one place, such as your dairy or a wall
calendar. Whatever method you choose to use, you should decide on and write
in your dates for working breach unit.
3. Once you have created your study schedule, do everything you can to stick to
it. The major reason students fail is that they get behind with their course work.
If you get into difficulties with your schedule, please let your tutor know before
it is too late for help.
4. Turn to Unit 1 and read the introduction and the objectives for the unit.

NOUN 11
INTRODUCTION TO ECONOMETRICS II ECO 306

5. Assemble the study materials. Information about what you need for a unit is
available in the `Overview' at the beginning of each unit. You will also need
both the study unit you are working on and one of your textbooks on your desk
at the same time.
6. Work through the unit. The content of the unit itself has been arranged to
provide a sequence for you to follow. As you work through the unit, you will
be instructed to read sections from your textbooks or other articles. Use the unit
to guide your reading.
7. Up-to-date course information will be deliveredcontinuously to you at the study
centre.
8. Work before the relevant due date (about four weeks before due dates) get the
Assignment File for the next required assignment. Keep in mind that you will
learn a lot by doing the assignments carefully. They have been designed to help
you meet the objectives of the course and, therefore, will help you pass the
exam. Submit all assignments no later than the due date.
9. Review the objectives for each study unit to confirm that you have achieved
them. If you feel unsure about any of the objectives, review the study material
or consult your lecturer.
10. When you are confident that you have achieved a unit's objectives, you can
then start on the next unit. Proceed unit by unit through the course and try to
pace your study so that you keep yourself on schedule.
11. When you have submitted an assignment to your tutor for marking, do not wait
for its return `before starting on the next units. Keep to your schedule. When
returning the assignment, pay particular attention to your lecturer's comments,
both on the tutor-marked assignment form and also written on the assignment.
Consult your lecturer as soon as possible if you have any questions or
problems.
12. After completing the last unit, review the course and prepare yourself for the
final examination. Check that you have achieved the unit objectives (listed at
the beginning of each unit) and the course objectives (listed in this Course
Guide).

Tutors and Tutorials

There are some hours of tutorials (2-hours sessions) provided in support of this course.
You should get notifications of dates, times, and location for these tutorials. Together
with the name and phone number of your lecturer, as soon as thetutorial group
allocated are made.

Your lecturer will mark and comment on your assignments, keep a close watch on
your progress and on any difficulties you might encounter, and provide assistance to
you during the course. You must mail your tutor-marked assignments to your lecturer
well before the due date (at least two working days are required). They will be marked
by your lecturer and returned to you as soon as possible.

Do not hesitate to contact your lecturer by telephone, e-mail, or discussion board if

you need help. The following might be circumstances in which you would find help
necessary. Contact your lecturer if.
• You do not understand any part of the study units or the assigned readings
• You have difficulty with the self-assessment exercises

NOUN 12
INTRODUCTION TO ECONOMETRICS II ECO 306

• You have a question or problem with an assignment, with your lecturer's comments
on an assignment or with the grading of an assignment.

You should try your best to attend the tutorials. Such avenues are the only chance to
have face to face contact with your lecturer and to ask questions which are given
instant answers instantly. You can raise any problem encountered in the course of
your study. To gain the maximum benefit from course tutorials, prepare a question list
before attending them. You will learn a lot from participating in discussions actively.

Summary
The course, Introduction to Econometrics II (ECO 306) presents you with general
background and applications of the concept of Random Variables, Sampling Theory and
how to be able to identify functions and problems associated with estimation. This course
also examines ordinary least squares assumptions and sampling theories. Topics like,
multicollinearity, heteroscedasticity, autocorrelation and Econometrics modeling had
illustrative examples used for further explanations. For this reason, use of regression
analyses, correlation, variance and dummy variables with experiential case studies that
apply the techniques to real-life data are stressed and discussed throughout the course.

This course is therefore developed in a manner to guide you further on what

econometrics entails, what course materials in line with a course learning structure
you will be using. The learning structure suggested some general guidelines for a time
frame required of you on each unit to achieve the course aims and objectives.
Conclusively, you would have developed critical thinking skills with the material
necessary for anefficient introductory understanding of econometrics. Nevertheless, to
achieve a lot more from the course, please try to solve econometrics problems
independently, do presentation and interpretation of findings in any assignment given
both in your academic programme and other spheres of life. Further work in this
course would expose you to introductory levels of topics like; vector autoregressions,
unit roots, cointegration, and time-series analysis.
We wish you the very best in your schoolwork.

NOUN 13
INTRODUCTION TO ECONOMETRICS II ECO 306

NATIONAL OPEN UNIVERSITY OF NIGERIA

Course Code: ECO 356

Course Title: Introduction to Econometrics II

Course Developer/Writer:
Dr. Adesina-Uthman G. A.
Faculty of Social Sciences
Department of Economics
National Open University of Nigeria.

and

OKOJIE, Daniel Esene

School of Post Graduate Studies (SPGS)
University of Lagos, Akoka-Yaba
Lagos.

ContentEditor: Prof. Ismael Ogboru

Department of Economics,
University of Jos, Jos,

NOUN 14
INTRODUCTION TO ECONOMETRICS II ECO 306

Plateau State.

December, 2017

INTRODUCTION TO ECONOMETRICS II

NOUN 15
INTRODUCTION TO ECONOMETRICS II ECO 306

INTRODUCTION TO ECONOMETRICS II
CONTENTS
PAGES
Module 1: Sampling Theory, Variance and Correlation
Unit 1: Random variables and sampling
theory.......................................................................4
Unit 2: Covariance and
Variance...........................................................................................13
Unit 3:
CorrelationCoefficient...........................................................................................
...20

Module 2: Simple Equation Regression Models

Unit 1: Simple Regression
Analyses......................................................................................25
Unit 2: Properties of the regression coefficients and hypothesis
testing.................................36
Unit 3: Multiple regression analysis and
Multicollinearity....................................................54
Unit 4: Transformations of
Variables.....................................................................................66
Unit 5: Dummy
Variables......................................................................................................69
Unit 6: Specification of regression variables: A preliminary
skirmish...................................74

Module 3: Heteroscedasticity/Heteroskedasticity
- Heteroscedasticity and Its Implications
…......................................................76
- Solution to Heteroscedasticity
Problem...........................................................84
- Other Tests/ Consequences of
Heteroscedasticity…......................................84

NOUN 16
INTRODUCTION TO ECONOMETRICS II ECO 306

Module 4: Autocorrelation, Error and Econometric Modelling

Unit 1: Stochastic Regression and measurement
errors..........................................................86
Unit 2:
Autocorrelation......................................................................................................
....93
Unit 3: Econometric Modelling and Models Using Time Series
Data...................................99
Module 5: Simultaneous Equation, Binary Choice, and Maximum Likelihood
Estimation
Unit 1: Simultaneous
Equations...........................................................................................103
Unit 2: Binary Choice and Maximum Likelihood
Estimation..............................................107

MODULE 1 SAMPLING THEORY, VARIANCE, AND CORRELATION

The general aim of this module is to provide the student with a thorough
understanding of the basic statistical tools that will be needed for regression analyses
in the subsequent module. As well as random variables and sampling theory,
Covariance, variance, and correlation are to bedemystified for proper understanding.
By the end of this module,students would have been able to understand the basic parts
of regression analysis.
The units to be studied in this module are;

Unit 1: Random variables and sampling theory

Unit 2: Covariance and Variance
Unit 3: Correlation

UNIT 1: RANDOM VARIABLES AND SAMPLING THEORY

CONTENTS

NOUN 17
INTRODUCTION TO ECONOMETRICS II ECO 306

1.1.1.0 Introduction
1.1.2.0 Objectives
1.1.3.0 Main Content
1.1.3.1 Random Variables and Sampling Theory
1.1.3.2 Expected values of discrete random variable
1.1.3.3 Expected value rules
1.1.3.4 Sampling theory
1.1.3.4.1 Some terminology
1.1.3.4.2 Reasons for sampling
1.1.3.4.3 Types of sampling technique
1.1.3.4.4 Simple Random Sampling technique
1.1.3.5 Estimation of Population Mean
1.1.4.0 Summary
1.1.5.0 Conclusion
1.1.6.0 Tutor-Marked Assignment
1.1.7.0 References/Further Reading

1.1.1.0 INTRODUCTION

Random variable or stochastic variable is a variable whose possible values are

numerical products of a chance occurrence. As a function, a random variable is
required to be quantifiable, which rules out certain uncontrolled circumstances where
the quantity which the random variable returns is considerably sensitive to small
changes in the outcome. It is common that these outcomes depend on some physical
variables that are not well understood. For example, when you toss a coin, which
outcome will be observed is not certain. The domain of a random variable is the set of
possible outcomes. In the case of the coin, there are only two possible outcomes,
namely heads or tails. The domain of the random variable leads us into the concept of
sampling theory which is concerned with the theory involved in the selection of a
subset of individuals from within a statistical population estimates the characteristics
of the whole population.

NOUN 18
INTRODUCTION TO ECONOMETRICS II ECO 306

1.1.2.0 OBJECTIVE

The main objective of this unit is to provide a broad understanding of the topic,
Random Variables, and Sampling Theory, which is preparatory to the more widely
used simple and multiple regression analyses.

1.1.3.0 MAIN CONTENTS

1.1.3.1 Random Variables and Sampling Theory

A variable X is said to be a random variable if for every real number a there exist a
probability P( X  a) that X takes on a value less than or equal to a. That is, a Random
variable is a variable whose value cannot be predicted exactly. It can assume any
value. Random variables could be discrete or continuous. A discreterandom variable is
one that has a specific set of possible values or a finite set of values. An example is a
total score when two dice are thrown. A continuous variable, e.g. the temperature in a
particular room, is a variable that can assume any value in thecertain range. It can take
any form of the continuing range of values.

The set of all possible values of a random variable is known as apopulation where
thesample or a random variable can be drawn for inferential analysis.

1.1.3.2 Expected values of discrete random variable

The expected value of a discrete random variable is the weighted average of all its
possible values, taking the probability of each outcome as its weight. It can be
calculated by multiplying each possible value of the random variable by its probability
and adding. In mathematical terms, if X denotes the random variable, its expected
value is denoted by E(X).
Let us suppose that X can take nparticular values of , ,..., and that the probability
of is .
Then,
n
E ( X )  xi pi  ...  xn pn   xi pi ...[1.01]
i 1

Table 1.0 shows an example of expected value of variable X with two dice.
Table 1.0:Expected value of variable X with two dice

NOUN 19
INTRODUCTION TO ECONOMETRICS II ECO 306

X P X .P

10

11

12

( ) ∑

In the case of the two dice, the values … were the numbers 2 … 12: = 2, =
3... =12, and = , = ... = . As shown in table 1.0, the expected value is
7. Also, the expected value of a random variable is described as population mean. In
the case of the random variable X, the population mean is given as .

1.1.3.3 Expected value rules

There are three main rules of expected values that are equally valid for both discrete
and continuous random variables. These are;

Rule 1: The expected value of the sum of several variables is equal to the sum of their
respective expected values. For example, if you have three random variables X, Y, and
Z,

( ) ( ) ( ) ( ) …[1.02]

NOUN 20
INTRODUCTION TO ECONOMETRICS II ECO 306

Rule 2: If you multiply a random variable by a constant, you multiply its expected
value by
the same constant. If X is a random variable and b is a constant,

( ) ( ) …[1.03]

Rule 3: The expected value of a constant is that constant. For example, if b is a

constant.

( ) …[1.04]

Putting the three rules together; suppose we wish to calculate E(Y), where we have

…[1.05]

and and are constants.

Then,

( ) ( ) …[1.06]

( ) ( ) ( ) …[1.07

= ( ) ( )…[1.08]

1.1.3.4 Sampling theory

The goals of a sample survey and an experiment are very different. The role of
randomisation also differs. In both cases, without randomisation, there can be no
inference. Without randomisation, the analyst can only describe the observations and
cannot generalise the results. In the sample survey, randomisation is used to reduce
bias and to allow the results of the sample to be generalised to the population from
which the sample was drawn. In an experiment, randomisation is used to balance the
effects of confounding variables. The objective of asample survey is often to estimate
a population mean and variance.

1.1.3.4.1 Some terminology

NOUN 21
INTRODUCTION TO ECONOMETRICS II ECO 306

i. Element: An element is an object on which a measurement is made,which

could be a voter in an area, a product as it comes off the assembly line or a
plant in a field that has either flowered or not.
ii. Population: A population is a collection of elements about which we wish to
make an inference. The population must be clearly defined before the sample is
taken.
iii. Sampling units: These are some overlapping collections of elements from the
population that covers the entire population. The sampling units partition the
population of interest. The sampling units could be households or individual
voters.
iv. Frame: Is a list of sampling units.
v. Sample: This is a collection of sampling units drawn from a frame or frames.
Data are obtained from the sample and are used to describe characteristics of
the population. Two advantages of sampling are that the cost is lower and data
collection is faster than measuring the entire population.
vi. Census: The enumeration of the total element of the population.

Example 1: Suppose we are interested in what voters in a particular area think about
the drilling of oil in the national wildlife preserves. The elements are the registered
voters in the area. The population is the collection of registered voters. The sampling
units will likely be households in which there may be several registered voters. The
frame is a list of households in the area.

1.1.3.4.2 Reasons for sampling

Information could be obtained by taking a complete enumeration of the whole

population or aggregate. This is usually difficult as information on every element is
rarely available. Therefore, it is better to employ sampling method to obtained
information than complete enumeration for the following reasons:

i. Reduce cost: if data are secured from only a small fraction of the aggregate,
expenditures are smaller than if a complete census is attempted. With large
populations result accurate enough to be useful can be obtained from samples
that represent only a small fraction of the population.

NOUN 22
INTRODUCTION TO ECONOMETRICS II ECO 306

ii. Greater speed: for the same reason the data can be collected and summarized
more quickly with a sample than with a complete count. This is a vital
consideration when the information is urgently needed.

iii. Greater Scope: a complete census is impracticable; the choice lies between
obtaining the information by sampling or not at all. Thus surveys that rely on
sampling have more scope and flexibility regarding the kind of information that
can be obtained.
iv. Greater Accuracy: here, personnel of higher quality can be employed and
given intense training. This would allow for much more careful supervision of
the field work. Processing and analysing of the results become feasible because
the volume of work is now reduced. The sample would most likely produce
amore accurate result than the complete enumeration.

1.1.3.4.3 Types of sampling technique

i. Probability sampling technique: Simple random sampling, systematic

random sampling, stratified random sampling, cluster sampling, etc.
ii. Non-probability sampling technique: Snowball sampling, quota sampling
technique, accidental or convenient sampling technique, etc.

Sample designs that utilise planned randomness are calledprobability sampleswhile

non-probability doesn‟t apply randomness as it is based on the subjective dictate of
the researcher since all elements are not given equal chance of being selected. The
most fundamental probability sample is the simple random sample. In a simple
random sample, a sample of n sampling units is selected in such a way that each
sample of size n has the same chance of being selected. In practice, other more
sophisticated probability sampling methods are commonly used, but we would focus
here on simple random sampling technique.

1.1.3.4.4 Simple Random Sampling technique

Suppose the observations , are to be sampled from a population with mean,

standard deviation, and size N in such a way that every possible sample of size n has
an equal chance of being selected. Then the sample , was selected in a
simple random sample. If the sample mean is denoted by ̅ then we have;

NOUN 23
INTRODUCTION TO ECONOMETRICS II ECO 306

( ̅) …[1.09]

V (̅) . / …[1.10]

The term . / in the above expression is known as the finite population correction

factor. For the sample variance , it can be shown that

( ) . / …[1.11]

When using as an estimate of , we must adjust with

. / ( ) …[1.12]

Consequently, an unbiased estimator of the variance of the sample mean is given by

̂ (̅) . / …[1.13]

As a rule of thumb, the correction factor . / can be ignored if it is greater than

0.9, or if the sample is less than 10% of the population.

Example 2; Consider the finite population with N = 4 elements* +. For this

population = 3 and = 5. Simple random samples without replacement of size n =
2 are selected from the population. All possible samples along with their summary
statistics are listed in table 1.1.1.

Table 1.1.1 Simple Random Sampling

Samples Probability mean Variance

(0,2) 1 2

(0,4) 2 8

(0,6) 3 18

(2,4) 3 2

NOUN 24
INTRODUCTION TO ECONOMETRICS II ECO 306

(2,6) 4 8

(4,6) 5 2

We see in this example that;

V ( ̅) . / . / . /

Similarly,

( ) . / …[1.14]

Could also be obtained from table 1.1.1

1.1.3.5 Estimation of Population Mean

If we are interested in estimating a population mean from a simple random sample, we

have;

∑
̂ ̅ …[1.15]

If we are interested in estimating population variance from a simple random sample,

we have;

̂ (̅) . / …[1.16]

Where,

∑ ( ̅)
…[1.17]

When the margin of error is two standard errors, we have;

√ ̂ (̂) √ . / …[1.18]

1.1.4.0SUMMARY

NOUN 25
INTRODUCTION TO ECONOMETRICS II ECO 306

In this unit, the students are presented with the essentials and applications of the
concept of random variables and sampling theory and their estimations. Also, by now
the students should be able to identify functions and problems associated with the
estimation.

1.1.5.0CONCLUSION

In this unit, the concepts of random variables and sampling theory have been
introduced and discussed and the associated estimation explained. The students are
made to understand that random variables are of two quantitative (discrete and
continuous random) variable types, whose values may be determined through the outcome of
a random trials. Discreteand continuous random variables were both explained with
examples. Further definition of random variable as it relates to a population is
explained. Furthermore, inference is made on how the student may apply the random
variable in other statistical analysis like mean, variance, standard deviation, etc.

Sampling theory is introduced in this unit in form of comparison to random variable.

Some terminologies and reasons associated with sampling are also discussed.
Probability and non-

probability sampling techniques are presented as two sample designs methods that
utilise planned randomness. Although, in practice, other more sophisticated
probability sampling and estimation methods are commonly used, the unit focused on
simple random sampling and population mean techniques as starting points.

1.1.6.0 TUTOR-MARKED ASSIGNMENT

1.) A random variable X is defined to be the difference between the higher value and
the lower
value when two dice are thrown. If they have the same value, X is defined to be 0.
Find the probability distribution for X.

2.) A random variable X is defined to be the larger of the two values when two dice
are thrown, or
the value if the values are the same. Find the probability distribution for X.

NOUN 26
INTRODUCTION TO ECONOMETRICS II ECO 306

1.1.7.0 REFERENCES /FURTHER READING

Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.

Stock, J. H., & Watson, M. W. (2015).Introduction to econometrics. Pearson.

Dougherty, C. (2014). Elements of econometrics.London: University of London.

Dougherty, C. (2003). Numeracy, literacy and earnings: evidence from the National
Longitudinal Survey of Youth. Economics of education review, 22(5), 511-521.

UNIT 2: CO-VARIANCE AND VARIANCE

CONTENTS
1.2.1.0 Introduction
1.2.2.0 Objectives
1.2.3.0 Main Content
1.2.3.1 CoVariance and Variance
1.2.3.2 Some Basic Covariance rule
1.2.3.3 Population CoVariance
1.2.3.4 Sample Variance
1.2.3.5 Variance Rule
1.2.4.0 Conclusion
1.2.5.0 Summary
1.2.6.0 Tutor-Marked Assignment
1.2.7.0 References/Further Reading

1.2.1.0 INTRODUCTION

The previous unit in this module introduced and discussed random variable and
associated sampling theories. In other to further equip the students with the adequately
understanding of more basic tools needed for regression analysis in the next module

NOUN 27
 INTRODUCTION TO ECONOMETRICS II ECO 306

and in general statistical analyses, this unit discusses covariance and variance. The
unit explains that, variance and covariance are two measures used in statistics. While
variance is an intuitive concept that measures the scatter of the data, covariance on the
other hand, is not that intuitive at first but gives a mathematical indication of the
degree of change of two random variables together.

1.2.2.0 OBJECTIVE

The main objective of this unit is to provide a broad understanding of the topics
Covariance and Variance which is preparatory to the more widely used simple and
multiple regression analyses.

1.2.3.0 MAIN CONTENTS

1.2.3.1 Covariance and Variance

Sample covariance is a measure of association between two variables. The sample

covariance, Cov(X, Y), is a statistic that enables you to summarize this association
with a single number. In general, given n observations on two variables X and Y, the
sample covariance between X and Y is given by;

∑ ( ̅ )( ̅) …[2.19]

Where the bar over the variable signifies the sample mean. Therefore, a positive
association would be summarized by a positive sample covariance while a negative
sample covariance would summarise a negative association.

1.2.3.2 Some Basic Covariance rules

i. Co-variance Rule 1: If Y = V + W, Cov(X, Y) = Cov(X, V) + Cov(X, W)

ii. Co-variance Rule 2: If Y = bZ, where b is a constant and Z is a variable, Cov(X, Y)
= bCov(X, Z)
iii. Co-Variance Rule 3: If Y = b, where b is a constant, Cov(X, Y) = 0

For example, Tables 1.2.0(a) and (b) show years of schoolingS, and hourly earningsY,
for a subset of 20 households in theUnitedStates. We are required to calculate the
covariance.

NOUN 28
INTRODUCTION TO ECONOMETRICS II ECO 306

Table 1.2.0(a) Covariance table

Observation S Y

1 15 17.24

2 16 15.00

3 8 14.91

4 6 4.50

5 15 18.00

6 12 6.29

7 12 19.23

8 18 18.69

9 12 7.21

10 20 42.06

11 17 15.38

12 12 12.70

13 12 26.00

14 9 7.50

NOUN 29
INTRODUCTION TO ECONOMETRICS II ECO 306

15 15 5.00

16 12 21.63

17 16 12.10

18 12 5.55

19 12 7.50

20 14 8.00

Table 1.2.0(b)Covariance table

( ̅)( ̅)
Observation S Y ̅ ̅

1 15 17.24 1.75 3.016 5.277

2 16 15.00 2.75 0.775 2.133

3 8 14.91 -5.25 0.685 -3.599

4 6 4.50 -7.25 -9.725 70.503

5 15 18.00 1.75 3.776 6.607

6 12 6.29 -1.25 -7.935 9.918

7 12 19.23 -1.25 5.006 -6.257

8 18 18.69 4.75 4.466 21.211

9 12 7.21 -1.25 -7.015 8.768

10 20 42.06 6.75 27.836 187.890

11 17 15.38 3.75 1.156 4.333

12 12 12.70 -1.25 -1.525 1.906

13 12 26.00 -1.25 11.776 -14.719

NOUN 30
INTRODUCTION TO ECONOMETRICS II ECO 306

14 9 7.50 -1.45 -6.725 28.579

15 15 5.00 1.75 -9.225 -16.143

16 12 21.63 -1.25 7.406 -9.257

17 16 12.10 2.75 -2.125 -5.842

18 12 5.55 -1.25 -8.675 10.843

19 12 7.50 -1.25 -6.725 8.406

20 14 8.00 0.75 -6.225 -4.668

Total 265 284.49 - - 305.888

Average 13.250 14.225 - - 15.294

Note from the above example that the association is positive. This is given by the
positive covariance.

1.2.3.3 Population Covariance

If X and Y are random variables, the expected value of the product of their deviations
from their means is defined to be the population covariance :

,( )( )- …[2.20]

Where and are the population means of X and Y, respectively.

As you would expect, if the population covariance is unknown, the sample covariance
will
provide an estimate of it, given a sample of observations. However, the estimate will
be biased downwards, for

, ( )- …[2.21]

The reason is that the sample deviations are measured from the sample means of X
and Y and tend to underestimate the deviations from the true means. Therefore, we can
construct an unbiased estimator by multiplying the sample estimate by n/(n–1).

1.2.3.4 Sample Variance

NOUN 31
INTRODUCTION TO ECONOMETRICS II ECO 306

For a sample of n observations, , ..., , the sample variance will be defined as the
average squared deviation in the sample:

∑ ( ̅)
( ) …[2.22]

The sample variance, thus defined, is a biased estimator of the population variance.
The reason for the underestimation is because it is calculated as the average squared
deviation from the sample mean rather than the true mean. This is because the sample
mean is automatically in the centre of the sample, the deviations from it will tend to be
smaller than those from the population mean. Therefore, sample variance as an
unbiased estimate of population variance is given as:

∑ ( ̅)
…[2.23]

1.2.3.5 Variance Rule

Variance rule 1: If Y = V + W, Var(Y) = Var(V) + Var(W) + 2Cov(V, W)

Variance rule 2: If Y = bZ, where b is a constant, Var(Y) = Var (Z)

Variance rule 3: If Y = b, where b is a constant, Var(Y) = 0.

Variance rule 4: If Y = V + b, where b is a constant, Var(Y) = Var(V) since the

variance of a constant is 0.

1.2.4.0 SUMMARY

While explaining the variance and covariance, the temptations to make comparison of
the two concepts may not be completely overcome. The unit briefly describes variance
as the measure of spread in a population while covariance is considered as a measure
of variation of two random variables. Furthermore, the unit showed that variance and
covariance are dependent on the magnitude of the data values and cannot be
compared; therefore, regulated. This means, covariance is dividing by the product of
the standard deviations of the two random variables and variance is normalised into
the standard deviation by taking the square root of it.

1.2.5.0 CONCLUSION

NOUN 32
INTRODUCTION TO ECONOMETRICS II ECO 306

Variance and covariance concepts as statistical tools are discussed in this unit. How
they are estimated were also explained using some basic covariance and variance
rules. The existence of population covariance and sample variance estimations were
briefly introduced. The introductions and discussions of these two concepts point out
that variance can be considered as a special case of covariance.

1.2.6.0 TUTOR-MARKED ASSIGNMENT

1.) In a large bureaucracy the annual salary of each, Y, is determined by the formula

Where,S is the number of years of schooling of the individual and T is the length of
time, in years, of employment. X is the individual‟s age. Calculate Cov(X, Y), Cov(X,
S), and Cov(X, T) for the sample of five individuals shown below and verify that

( ) ( ) ( )

2.) In a certain country the tax paid by a firm, T, is determined by the rule

Where,P is profits, and I is aninvestment, the third term being the effect of an
investment incentive. S is sales. All variables are measured in $ million at annual
rates. Calculate Cov(S, T), Cov(S, P), and Cov(S, I) for the sample of four firms
shown below and verify that

( ) ( ) ( )

1.2.7.0 REFERENCES /FURTHER READING

Dominick, S., & Derrick, R. (2002).Theory and problems of statistics and

econometrics.Schaum‟s Outline Series.
Dougherty, C. (2003). Numeracy, literacy and earnings: evidence from the National
Longitudinal Survey of Youth.Economics of education review, 22(5), 511-521.
Dougherty, C. (2014). Elements of econometrics.London: University of London.

NOUN 33
INTRODUCTION TO ECONOMETRICS II ECO 306

UNIT 3: CORRELATION CO-EFFICIENT

CONTENTS
1.3.1.0 Introduction
1.3.2.0 Objectives
1.3.3.0 Main Content
1.3.3.1 Properties of the regression coefficients and hypothesis testing
1.3.4.0 Summary
1.3.5.0 Conclusion
1.3.6.0 Tutor-Marked Assignment
1.3.7.0 References/Further Reading

1.3.1.0 INTRODUCTION

This unit introduces a statistic called correlation coefficient. The correlation

coefficient describes the direction, whether positive or negative and further measures
the degree of relationship that exist between two different variables.

1.3.2.0 OBJECTIVE

The main objective of this unit is to provide ways for which the studentmay have a
simpler understanding of the topic „correlation‟.

1.3.3.0 MAIN CONTENTS

Correlation measures the degree of association between two or more variables.

1.3.3.1 Properties of the regression coefficients and hypothesis testing

Like variance and covariance, the correlation coefficient comes in two forms,
population and sample. ρ traditionally denotes the population correlation coefficient,

NOUN 34
INTRODUCTION TO ECONOMETRICS II ECO 306

the Greek letter that is the equivalent of “r”, and pronounced “row”, as in row a boat.
For variables X and Y it is defined by

 XY
 XY  …[3.24]
 X2  Y2

If X and Y are independent, will be equal to 0 because the population covariance

will be 0. If there is a positive association between them, then we have ,
otherwise will still be positive.

If there is an exact positive linear relationship will assume its maximum value of
1. Similarly, if there is a negative relationship will be negative, with minimum
value of –1.

The sample correlation coefficient, , is defined by replacing the population

covariance and variances by their unbiased estimators. We have seen that these may
be obtained by multiplying the sample variances and co-variances by ( nn 1) . Hence,

n
cov( XY )
rXY  n 1 …[3.25]
n n
var( X ) var(Y )
n 1 n 1

The factors n
( n1) could be cancelled out so we can conveniently define the sample

correlation by

( )
( )
…[3.26]
√ ( )

 XY
 …[3.27]
 XY

Like ρ, r has maximum value 1, which is attained when there is a perfect positive
association
between the sample values of X and Y (when you plot the scatter diagram, the points
lie exactly on an upward-sloping straight line). Similarly, it has minimum value –1,
attained when there is a perfect negative association (the points lying exactly on a

NOUN 35
INTRODUCTION TO ECONOMETRICS II ECO 306

downward-sloping straight line). A value of 0 indicates that there is no association

between the observations on X and Y in the sample. Of course the fact that r = 0 does
not necessarily imply that ρ = 0 or vice versa.

That is;

1  corr ( X , Y )  1
corr ( X , Y )  1 means perfect positive linear association
corr ( X , Y )  1 means perfect negative linear association
corr ( X , Y )  0 means no linear association

Figures 1.3(a) to (d) below give more graphical explanations;

Figures 1.3(a) to (d);Scattered diagrams of correlation coefficient as a measure of

linear association

Example: For illustration, using the education and earning example, the sample
correlation coefficient can be estimated. This is shown below:

NOUN 36
INTRODUCTION TO ECONOMETRICS II ECO 306

Table 1.2.1 Sample correlation coefficient

Observ. S Y ̅ ̅ ( ̅) ( ̅) ( ̅)( ̅)

1 15 17.24 1.75 3.016 3.063 9.093 5.277

2 16 15.00 2.75 0.775 7.563 0.601 2.133

3 8 14.91 -5.25 0.685 27.563 0.470 -3.599

4 6 4.50 -7.25 -9.725 52.563 94.566 70.503

5 15 18.00 1.75 3.776 3.063 14.254 6.607

6 12 6.29 -1.25 -7.935 1.563 62.956 9.918

7 12 19.23 -1.25 5.006 1.563 25.055 -6.257

8 18 18.69 4.75 4.466 22.563 19.941 21.211

9 12 7.21 -1.25 -7.015 1.563 49.203 8.768

10 20 42.06 6.75 27.836 45.563 774.815 187.890

11 17 15.38 3.75 1.156 14.063 1.335 4.333

12 12 12.70 -1.25 -1.525 1.563 2.324 1.906

13 12 26.00 -1.25 11.776 1.563 138.662 -14.719

14 9 7.50 -1.45 -6.725 18.063 45.219 28.579

15 15 5.00 1.75 -9.225 3.063 85.091 -16.143

16 12 21.63 -1.25 7.406 1.563 54.841 -9.257

NOUN 37
INTRODUCTION TO ECONOMETRICS II ECO 306

17 16 12.10 2.75 -2.125 7.563 4.514 -5.842

18 12 5.55 -1.25 -8.675 1.563 75.247 10.843

19 12 7.50 -1.25 -6.725 1.563 45.219 8.406

20 14 8.00 0.75 -6.225 0.563 38.744 -4.668

Total 265 284.49 - - 217.750 1,542.150 305.888

Average 13.250 14.225 - - 10.888 77.108 15.294

From column 6 and 7, you can see that Var (S) is 10.888 and Var (Y) is 77.108,
therefore,

1.3.4.0 SUMMARY

This unit briefly introduced correlation coefficient and showed some properties of
regression coefficients and hypothesis testing. Four figures and a table were used to
further illustrate the correlation coefficient as a measure of linear association and how
sample correlation may be estimated.

1.3.5.0 CONCLUSION

The brief discussion on correlation in this unit is to let the students aware that
correlation may be approached as a statistical tool that precedes the basic introduction
to econometrics. Although correlation measures may be clouded by relationships that
exist with other variables, figures and table were however used to show correlation as
the linear relationships between two variables.

1.3.6.0 TUTOR-MARKED ASSIGNMENT

NOUN 38
INTRODUCTION TO ECONOMETRICS II ECO 306

1.) Demonstrate that, in general; the sample correlation coefficient is not affected

by a change inthe unit of measurement of one of the variables.

2.) Suppose that the observations on two variables X and Y lie on a straight line

Demonstrate that ( ) ( ) and that Var(Y) = ( ) and

hence that the sample correlation coefficient is equal to 1 if the slope of the line is
positive, –1 if it is negative.

1.3.7.0 REFERENCES /FURTHER READING

Maddala, G. S., &Lahiri, K. (1992).Introduction to econometrics (Vol. 2). New York:

Macmillan.Dougherty, C. (2007).Introduction to econometrics. Oxford University

Press, USA.

Stock, J. H., & Watson, M. W. (2015).Introduction to econometrics. Pearson.

Dougherty, C. (2014). Elements of econometrics.London: University of London.

NOUN 39
INTRODUCTION TO ECONOMETRICS II ECO 306

MODULE 2: SIMPLE EQUATION REGRESSION MODELS

The general aim of this module is to provide studentswith a thorough understanding of

the basic rudiments of simple equation regression models. It shows how a theoretical
linear relationship between two variables can be quantified using appropriate data.
The principle of least squares regression analysis is explained, and expressions for the
coefficients are derived. By the end of this module,students should be able to
understand the basic parts of regression analysis. The units to be studied are;

Unit 1: Simple Regression Analyses

Unit 2: Properties of the regression coefficients and hypothesis testing

Unit 3: Multiple regression analysis and Multicollinearity

Unit 4: Transformations of Variables

Unit 5: Dummy Variables

Unit 6: Specification of regression variables: A preliminary skirmish.

UNIT 1: SIMPLE REGRESSION ANALYSES

CONTENTS
2.1.1.0 Introduction
2.1.2.0 Objectives
2.1.3.0 Main Content
2.1.3.1 Simple Regression Analyses
2.1.3.2 Causes of the Existence of the Disturbance Term
2.1.3.3 Least Squares Regression
2.1.3.3.1 Least Squares Regression with One Explanatory Variable
2.1.3.3.2 Alternative Expressions for b2

NOUN 40
INTRODUCTION TO ECONOMETRICS II ECO 306

2.1.4.0 Summary
2.1.5.0 Conclusion
2.1.6.0 Tutor-Marked Assignment
2.1.7.0 References/Further Reading

2.1.1.0 INTRODUCTION

In this unit, an analytic method to measure the association of one or more independent
variables with a dependent variable is discussed. Simple regression analyses is a
statistical method which can be used to summarize and study relationships existing
between two continuous quantitative variables. This unit further introduces the simple
regression as a concept that is better understood after a thorough understanding of the
basic correlation procedures.

2.1.2.0 OBJECTIVE

The main objective of this unit is to acquaint students with the rudiments of
identifying and differentiating simple equation model from multiple regression.

2.1.3.0 MAIN CONTENTS

2.1.3.1 Simple Regression Analyses

The correlation coefficient may indicate that two variables (bivariate regression
model) are associated with one another, but it does not give any idea of the kind of
relationship involved. While regression predicts the value of the dependent variable
based on the known value of the independent variable. In this module further step is
taken for cases which we are willing to hypothesize on, than one variable dependence
on another. It must be statedimmediately that one would not expect to find an exact
relationship between any two economic variables unless it is true as a matter of
definition. In textbook expositions of economic theory, the usual way of dealing with
this awkward fact is to write down the relationship as if it were exact and to warn the
reader that it is only an approximation. However, in statistical analysis, one
acknowledges the fact that the relationship is not exact by explicitly including in it a

NOUN 41
INTRODUCTION TO ECONOMETRICS II ECO 306

random factor known as the disturbance term. We shall start with the simplest
possible model:

...[2.01]

, the value of the dependent variable in observation i, has two components: (1) the
non-random (deterministic term) component , being described as the
explanatory (or independent/descriptive) variable and the fixed quantities and as
the parameters of the equation, and (2) the disturbance (stochastic term), .

Figure 2.0 illustrates how these two components combine to determine Y. X1, X2, X3,
and X4, which are four hypothetical values of the explanatory variable. If the
relationship between Y and X were

exact, the corresponding values of Y would be represented by the points Q1 – Q4 on

the line. The disturbance term causes the actual values of Y to be different. In the
diagram, the disturbance term has been assumed to be positive in the first and fourth
observations and negative in the other two, with the result that, if one plots the actual
values of Y against the values of X, one obtains the points P1 – P4.

Figure 2.0Illustration of independent component combination to give a dependent

variable

In practice, the P points are all not what can be seen in Figure 2.0. The actual values
of and and hence the location of the Q points, are unknown, as these are the
values of the disturbance term in the observations. The task of regression analysis is to

NOUN 42
INTRODUCTION TO ECONOMETRICS II ECO 306

obtain estimates of and , and hence an estimate of the location of the line, given
the P points. As it is, it‟s somehow curious. The question “Why then does the
disturbance term exist”? would therefore arise. There are several reasons.

2.1.3.2 Reasons for inclusion of Disturbance Term

i. The omission of explanatory variables: The relationship between Y and X is
almost certain to be a simplification. In reality, there will be other factors
affecting Y that have been left out of the non-random dependent component,
and their influence will cause the points to lie on the line. It often happens
that there are variables that you would like to include in the regression
equation but cannot because you are unable to measure them. All of these
other factors contribute to the disturbance term.
ii. Aggregation of variables: In many cases, the relationship is an attempt to
summarise in aggregate somemicroeconomic relationships. For example,
the aggregate consumption function is an attempt to summarize a set of
individual expenditure decisions. Since the individual relationships are
likely to have different parameters, any attempt to relate

aggregate expenditure to aggregate income can only be an approximation.

The discrepancy is attributed to the disturbance term.
iii. Model misspecification: The model may be misspecifiedregarding its
structure. Just to give one of the many possible examples, if the relationship
refers to time series data, the value of Y may depend not on the actual value
of X but on the value that had been anticipated in the previous period. If the
anticipated and actual values are closely related, there will appear to be a
relationship between Y and X, but it will only be an approximation, and
again the disturbance term will pick up the discrepancy.
iv. Functional misspecification: The functional relationship between Y and X
may be misspecified mathematically. For example, the true relationship
may be non-linear instead of linear. Obviously, one should try to avoid this
problem by using an appropriate mathematical specification, but even the
most sophisticated specification is likely to be only an approximation, and
the discrepancy contributes to the disturbance term.

NOUN 43
INTRODUCTION TO ECONOMETRICS II ECO 306

v. Measurement error: If the measurement of one or more of the variables in

the relationship is subject to error, the observed values will not appear to
conform to an exact relationship, and the discrepancy contributes to the
disturbance term.

The disturbance term is the collective outcome of all these factors. Obviously, if you
were
concerned only with measuring the effect of X on Y, it would be much more
convenient if the
disturbance term did not exist. Were it not for its presence, the P points in Figure 2.1
would coincide with the Q points. Therefore, it would be known that every change in
Y from observation to observation was due to a change in X, and you would be able to
calculate and , exactly. However, part of each change in Y is due to a change in μ,
and this makes life more difficult. For this reason, μ is sometimes described as noise.

2.1.3.3Least Squares Regression

Suppose that you are given the four observations on X and Y represented in Figure 2.1
and you are asked to obtain estimates of the values of and , in [2.01]. As a rough
approximation, you could do this by plotting the four P points and drawing a line to fit
them as best you can, as shown in Figure 2.2 The intersection of the line with the Y-
axis provides an estimate of the intercept , which will be denoted b1 and the slope
provides an estimate of the slope coefficient , which will be denoted b2. The fitted
line will be written as;

NOUN 44
INTRODUCTION TO ECONOMETRICS II ECO 306

Figure 2.2 Plotting of Observations

Figure 2.3 fitting Plotted Observations

̂ …[2.02]

The hat mark over Y in [2.02] indicates that it is the fitted value of Y corresponding to
X and not the actual value. In Figure 2.3 the fitted points are represented by the points
R1 – R4. One thing that should be accepted from the beginning is that however much
care you take in drawing the line; you can never discover the true values of and .
b1 and b2 are only estimates, and they may be good or bad. Once in a while your
estimates may be absolutely accurate, but this can only be by coincidence and even
then you will have no way of knowing that you have hit the target exactly.

This remains the case even when you use more sophisticated techniques. Drawing a
regression line by eye is all very well, but it leaves a lot to subjective judgment.
Furthermore, as will become obvious, it is not even possible when you have a variable
Y depending on two or more explanatory variables instead of only one. The question
arises, is there a way of calculating good estimates of

and algebraically? The answer is yes! The first step is to define what is known as
a residual for each observation. This is the difference between the actual value of Y in
any observation and the fitted value given by the regression line, that is, the vertical
distance between Piand Riin observation i. Which will be denoted by ei.

NOUN 45
INTRODUCTION TO ECONOMETRICS II ECO 306

̂ ...[2.03]

The residuals for the four observations are shown in Figure 2.3

Substituting [2.02] into [2.03], we obtain

...[2.04]

and hence the residual in each observation depends on our choice of b1 and b2.
Obviously, we wish to fit the regression line, that is, choose b1 and b2, in such a way
as to make the residuals as small as possible. Equally obvious, a line that fits some
observations well will fit others badly and vice versa. We need to devise a criterion of
fit that takes account of the size of all the residuals simultaneously. There are some
possible criteria, some of which work better than others. It is useless minimizing the
sum of the residuals, for example. The sum will automatically be equal to 0 if you
make b1 equal to ̅ and b2 equal to 0, obtaining the horizontal line Y = ̅ . The positive
residuals will then exactly balance the negative ones but other than that, the line will
not fit the observations.

One way of overcoming the problem is to minimize RSS (sum of the squares of the
residuals).

...[2.05]

According to this criterion, the smaller one can make RSS the better is the fit. If one
could reduce RSS to 0, one would have a perfect fit, for this would imply that all the
residuals are equal to 0. The line would go through all the points, but of course, in
general, the disturbance term makes this impossible. There are other quite reasonable
solutions, but the least squares criterion yields estimates of b1 and b2 that are unbiased
and the most efficient of their type, provided that certain conditions are satisfied. For
this reason, the least squares technique is far and away the most popular in
uncomplicated applications of regression analysis. The form used here is usually
referred to as ordinary least squares and abbreviated OLS.

NOUN 46
INTRODUCTION TO ECONOMETRICS II ECO 306

Table 2.1
X Y ̂ e

1 3

2 5

We shall assume that the true model is;

...[2.06]

And we shall estimate the coefficients b1 and b2 of the equation using;

̂ ...[2.07]

When X is equal to 1, according to the regression line ̂ is equal to (b1 + b2). When X
is equal to 2, ̂ is equal to (b1 + 2b2). Therefore, we can set up Table 2.1.0. So the
residual for the first observation, e1, which is given by (Y1 – ̂ 1), is equal to (3 – b1 –
b2), and e2, given by (Y2 –
̂ 2), is equal to (5 – b1 – 2b2). Hence

( ) ( )

...[2.08]

Now we want to choose b1 and b2 so as to minimize RSS. To do this, we use the

calculus and find the values of b1 and b2 that satisfy

…[2.09]

Taking partial differentials of [2.08];

NOUN 47
INTRODUCTION TO ECONOMETRICS II ECO 306

...[2.10]

And

...[2.11]

And so we have

And

Solving these two equations, we obtain b1 = 1 and b2 = 2, and hence the regression
equation

Just to check that we have come to the right conclusion, we shall calculate the
residuals:

e1 = 3 – b1 – b2 = 3 – 1 – 2 = 0

e2 = 5 – b1 – 2b2 = 5 – 1 – 4 = 0

Thus both residuals are equal to 0, implying that the line passes exactly through both
points.

2.1.3.3.1 Least Squares Regression with One Explanatory Variable

We shall now consider the general case where there are n observations on two
variables X and Y and supposing Y to depend on X; we will fit the equation

̂ ...[2.12]

The fitted value of the dependent variable in observation i.

̂ will be (b1 + b2Xi) and the residual will be (Yi– b1 – b2Xi). We wish to choose b1
and b2 so as to minimize the residual sum of the squares RSS given by

NOUN 48
INTRODUCTION TO ECONOMETRICS II ECO 306

∑ ...[2.13]

We will find that RSS is minimised when

( )
…[2.14]
( )

And

̅ ̅ …[2.15]

The derivation of the expressions for b1 and b2 will follow the same procedure as the
derivation in the preceding example, and you can compare the general version with
the examples at each step.

We will begin by expressing the square of the residual in observation iregardingb1, b2

and the data on X and Y:

( ̂) ( )
…[2.16]

Summing over all the nobservations, we can write RSS as

( ) ( )

∑ ∑ ∑ ∑ ∑
…[2.17]

Note that RSS is effectively a quadratic expression in b1 and b2, with numerical
coefficients
determined by the data on X and Y in the sample. We can influence the size of RSS
only through our choice of b1 and b2. The data on X and Y, which determine the
locations of the observations in the scatter diagram and are fixed once we have taken
the sample. This equation [2.17] is the generalized version of the equations.

The first order conditions for a minimum,

…[2.18]

Yield the following equations:

NOUN 49
INTRODUCTION TO ECONOMETRICS II ECO 306

∑ ∑

∑ ∑ ∑ …[2.19]

Noting that

̅ ∑ ̅ ∑ …[2.20]

may be rewritten as

̅ ̅ …[2.21]

and hence

̅ ̅ …[2.22]

Substituting for and again noting that ∑ ̅ we obtain

∑ ∑ (̅ ̅) ̅ …[2.23]

Separating the terms involving b2 and not involving b2 on opposite sides of the
equation, we have

,(∑ ) ̅ - ∑ ̅̅̅̅ …[2.24]

Dividing both sides by 2n,

0 (∑ ) ̅ 1 (∑ ) ̅̅̅̅ …[2.25]

Using the alternative expressions for sample variance and covariance, this may be
rewritten as;

( ) ( )

( )
…[2.26]
( )

b2 is from [2.23], b1 is equally from [2.22]. Those who know about the second-order
conditions will have no difficulty confirming that we have minimized RSS.

2.1.3.3.2 Alternative Expressions for b2

NOUN 50
INTRODUCTION TO ECONOMETRICS II ECO 306

From the definitions of Cov(X, Y) and Var(X) one can obtain alternative expressions
for b2

( ) ∑ ( ̅ )( ̅) ∑ ( ̅ )( ̅)
( ) ∑ ( ̅) ∑ ( ̅)

 xy 
 i 1
n
…[2.27]
x
i 1
2

where,

xX X

y  Y Y

2.1.4.0 SUMMARY

In this unit, you are expected to have learnt the essentials and applications of the
concept of simple regression analyses and its estimation.

2.1.5.0 CONCLUSION

In conclusion, the concept of simple regression analyses and its estimation are
explained.

2.1.6.0 TUTOR-MARKED ASSIGNMENT

1.) A researcher obtaineddata on the aggregate expenditure on services Y, and

aggregate disposable
personal income X, both measured in N billion at constant prices, for each of
the U.S. states and fits the equation

̂ ̂ ̂

The researcher initially fits the equation using OLS regression analysis. However,
suspecting
that tax evasion causes both Y and X to be substantially underestimated, the researcher

NOUN 51
INTRODUCTION TO ECONOMETRICS II ECO 306

adopts
two alternative methods of compensating for the under-reporting:

a.) The researcher adds N90 billion to the data for Y in each state and N200 billion
to the data for X.
b.) The researcher increases the figures for both Y and X in each state by 10
percent.

2.) Derive from first principles the least squares estimator of in the
model

2.1.7.0 REFERENCES /FURTHER READING

Dominick, S., & Derrick, R. (2002).Theory and problems of statistics and

econometrics.Schaum‟s Outline Series.
Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.
N Gujaratti, D. (2004). Basic econometrics. McGraw-Hill, New York.
Stock, J. H., & Watson, M. W. (2015).Introduction to econometrics. Pearson.
Smith, G. (2013). Econometric Principles and Data Analysis.Centre for Financial and
Management Studies SOAS, University of London, London.

NOUN 52
INTRODUCTION TO ECONOMETRICS II ECO 306

UNIT 2: PROPERTIES OF THE REGRESSION COEFFICIENTS AND

HYPOTHESIS TESTING

CONTENTS
2.2.1.0 Introduction
2.2.2.0 Objectives
2.2.3.0 Main Content
2.2.3.1 The Random Components of the Regression Coefficients
2.2.3.2 Assumptions Concerning the Disturbance Term
2.2.3.2.1 Gauss–Markov Condition 1: E(μi) = 0 for All Observations
2.2.3.2.2 Gauss–Markov Condition 2: Population Variance of μi Constant for All
Observations
2.2.3.2.3 Gauss–Markov Condition 3: μi Distributed Independently of μj ( )
2.2.3.2.4 Gauss–Markov Condition 4: u Distributed Independently of the
Explanatory Variables
2.2.3.3 The Normality Assumption
2.2.3.4 Unbiasedness of the Regression Coefficients
2.2.3.5 Precision of the Regression Coefficients
2.2.3.6 Testing Hypotheses Relating to the Regression Coefficients
2.2.3.6.1 Formulation of a Null Hypothesis
2.2.3.6.2 Developing the Implications of a Hypothesis
2.2.3.7 Compatibility, Freakiness, and the Significance Level
2.2.3.8 What Happens if the Standard Deviation of is Not Known
2.2.4.0 Conclusion
2.2.5.0 Summary
2.2.6.0 Tutor-Marked Assignment
2.2.7.0 References/Further Reading

2.2.1.0 INTRODUCTION

NOUN 53
INTRODUCTION TO ECONOMETRICS II ECO 306

This unit firstly attempts giving an appropriate explanation to the concept of GAUSS-
MARKOV THEOREM before proceeding into the discussion of the properties of
regression coefficients and hypothesis testing. However, to properly understand this
unit, a brief discussion would be made of some knowledge areas like;

i. Estimators
ii. Assumptions underlying the Classical Linear Regression
Model (CLRM)
iii. Properties of Ordinary Least Square (OLS) estimator
iv. Statistical inference (hypothesis testing) like; null and
alternative hypotheses.

The basic knowledge of the aforementioned areas are what the students must be
equipped with before proceeding in this unit.

- Estimators
A rule for calculating an estimate of a given quantity based on observed
data is referred to as an estimator. Hence in the calculation three quantities
are distinguished; the quantity of interest, referred to as an estimand, the
result (estimate) and the rule (estimator).
The properties of estimators are the concerns of Estimation theory. The
theory defines and determines properties that can be used under given
circumstances to relate different estimators or different rules for creating
estimates for the same quantity which are built on the same data.
A simple example of an estimator is given by a sample mean equation of a
population mean shown below.
1 N
k  ki
N i 1

where, k is an estimator for the population mean  .

- Assumptions underlying Classical Linear Regression Model (CLRM)

This model is the basis of most econometric theory and provides a
description of the method of ordinary least squares (OLS). It also explains

NOUN 54
INTRODUCTION TO ECONOMETRICS II ECO 306

how the OLS, using sample data, can estimate unknown parameters of a
regression equation. Furthermore, it makes available opportunity to ask
whether statements about the true unknown parameters of the model, based
on our estimated values can be made. In doing this, there is need to make a
number of assumptions. These assumptions, if satisfied, ensure that the
estimators being used are accurate and efficient. Precise predictions about
the unknown model parameters can also be make through satisfactory
assumptions. Here is a summary of the 10 assumptions in CLRM:

Assumption 1: Linear regression model. The regression model

is linear in the parameters,as shown in[01]


    Yi  1  2 Xi  i   [01]

where,
Y is the regressandY and the regressorX may be
nonlinear.

Assumption 2: X values are fixed in repeated sampling.

Values taken by the regressorXare considered fixed in repeated
samples. More precisely, X is assumed to be nonstochastic.

Assumption 3: Zero means value of disturbance i . That is,

given the value of X, the mean value of the random disturbance
term i is zero. This shows that the uncertain mean value of i is
zero, as shown in[02]

E ( i | Xi )  0 …[02]

Assumption 4: Equal variance of i . If given the value of X, the

variance of i is the same for all observations. Which means that
the uncertain variances of i are alike, as shown in[03].

NOUN 55
INTRODUCTION TO ECONOMETRICS II ECO 306

var( i | Xi )  E[ i  E ( i | Xi )]2

 E ( i2 | Xi ) (due to
assumption 3)
 2 …[03]
where,
var is variance.

Assumption 5: No autocorrelation between the disturbances.

If given any two X values, X i and X j (i  j ), the autocorrelation
between any two i and  j (i  j ) is zero, as shown in[04].

cov( i ,  j | Xi , X j )  E{[ i  E ( i )] | Xi }{[  j  E (  j )] | X j }

 E ( i | Xi )(  j | X j ) ( shows i and  j are uncorrelated )

…[04]
0
where,
(i and j ) are different observations and cov is covariance.

Assumption 6: Zero covariance between (i and Xi ) . As earlier

expressed;
cov(i , Xi )  E[i  E(i )] Xi  E( Xi )]

sin ce E ( i )  0

 E[i ( Xi  E ( Xi ))]

 E(i Xi )  E ( Xi ) E (i )

sin ce E ( Xi ) is nonstochastic and E( i )  0

NOUN 56
INTRODUCTION TO ECONOMETRICS II ECO 306

E(i Xi )  0 …[05]

Assumption 7: The number of observations n must be greater

than the number of parameters to be estimated. On the other
hand, the number of observations n must be greater than the
number of descriptive (explanatory) variables.

Assumption 8: Variability in X values. The X values in a given

sample must not all be the same. That is, var(X) must be a finite
positive number.

Assumption 9: The regression model is correctly specified. On

the other hand, there is no error in the model used in observed
analysis.

Assumption 10: There is no perfect multicollinearity. That is,

there are no perfect linear relationships among the descriptive
variables.
Multicollinearity as a problem associated with CLRM is
discussed in unit 3.

- Properties of Ordinary Least Square (OLS) estimator

The following properties are associated with OLS estimators:

1. Linearity
2. Unbiasedness
3. Efficient: it has the minimum variance
4. Consistency
5. Asymptotic Unbiasedness

NOUN 57
INTRODUCTION TO ECONOMETRICS II ECO 306

The OLS estimator is sometimes referred to as the CLRM and in data analysis the best
estimator is refer to as BLUE (best linear unbiased estimator). Therefore, the OLS
estimator requires that the descriptive variables are received outside a data group and
there is no perfect multicollinearity. Also, OLS is best in the class of linear unbiased
estimators when the errors are vector of random variables and successively
uncorrelated. Within these conditions, the OLS offers minimum-variance mean-
unbiased estimation when the errors have fixed variances. Again, the OLS is a
maximum likelihood estimator under the additional assumption that the errors are
normally distributed. So, whenever students are planning to use a linear regression
model by means of OLS, each time check for the OLS assumptions. In as much as the
OLS assumptions are satisfied, the analysis becomes simpler. Through the Gauss-
Markov theorem (as will be seen later in this unit) students can directly use OLS for
the best results. When the OLS estimator is asymptotically normal and a consistent
estimator of the asymptotic covariance matrix is available to carry out hypothesis tests
on the coefficients of a linear regression model.

2.2.2.0 OBJECTIVE

The main objective of this unit is to provide basic understanding of the topic,
properties of regression coefficients and hypothesis testing. As well as how these
properties form the basis for prediction and forecasting analyses. Focus will also be on
the use of regression analysis to recognise which among the independent variables are
related to the dependent variable and to explore the forms of these relationships.

2.2.3.0 MAIN CONTENTS

With the aid of regression analysis, we can obtain estimates of the parameters of a
relationship. However, they are only estimates. The next question to ask is, how
reliable are they? We shall

answer this first in general terms, investigating the conditions for unbiasedness and
the factors governing their variance. Secondly, building on those conditions for
unbiasedness and their variances, we shall develop a means of testing whether a
regression estimate is compatible with a specific prior hypothesis concerning the true
value of a parameter. Hence, we shall derive a confidence interval for the true value,

NOUN 58
INTRODUCTION TO ECONOMETRICS II ECO 306

that is, the set of all hypothetical values not contradicted by the experimental result.
We shall also see how to test whether the goodness of fit of a regression equation is
better than what might be expectedby pure chance.

2.2.3.1 The Random Components of the Regression Coefficients

The least squares regression coefficient is a special form of arandom variable whose
properties depend on those of the disturbance term in the equation. This will be
demonstrated first theoretically and then using a controlled experiment. In particular,
we will investigate the implications for the regression coefficients of certain
assumptions concerning the disturbance term. Throughout the discussion, we shall
continue to work with the simple regression model where Y depends on X according to
the relationship
And we fit the regression equation ̂ given a sample of n observations.

We shall also continue to assume that X is a non-stochastic exogenous (not external

randomly determined) variable; that is, that its value in each observation may be
considered to be predetermined by factors unconnected with the present relationship.
First, note that has two components. It has non-random component ( ),
which owes nothing to the laws of chance ( may be unknown, but
nevertheless they are fixed constants) and it has the random component . This
implies that, when we calculate b2 according to the usual formula;

( )
…[2.28]
( )

b2 would also have a random component ( ). ( )depends on the values

of Y, and the values of Y depend on the values of μ. If the values of the disturbance
term had been different in the n observations, we would have obtained different values
of Y, hence of Cov (X, Y), and hence of b2.Thus we have shown that the regression
coefficient b2 obtained from any sample consists of (1) a fixed component, equal to
the true value , and (2) a random component dependent on Cov(X, μ), which is
responsible for its variations around this central tendency. Similarly, one may easily
show that b1 has a fixed component equal to the true value , plus a random
component that depends on the random factor μ.

2.2.3.2 Assumptions Concerning the Disturbance Term

NOUN 59
INTRODUCTION TO ECONOMETRICS II ECO 306

It is thus obvious that the properties of the regression coefficients depend critically on
the properties of the disturbance term. Indeed the latter has to satisfy four conditions,
known as the Gauss–Markov conditions, if ordinary least squares regression analysis
is to give the best possible results. If they are not satisfied, the user should be aware of
the fact. If remedial action is possible, he or she should be capable of taking it. If it is
not possible, he or she should be able to judge how seriously the results may have
been affected.

2.2.3.2.1 Gauss–Markov Condition 1: E(μi) = 0 for All Observations

The first condition is that the expected value of the disturbance term in any
observation should be 0. Sometimes it will be positive, sometimes negative, but it
should not have a systematic tendency in either direction. If an intercept is included in
the regression equation, it is usually reasonable to assume that this condition is
satisfied automatically since the role of the intercept is to pick up any systematic but
constant tendency in Y not accounted for by the explanatory variables included in the
regression equation.

2.2.3.2.2 Gauss–Markov Condition 2: Population Variance of μi Constant for All

Observations

The second condition is that the population variance of the disturbance term should be
constant for all observations. Sometimes the disturbance term will be greater,
sometimes smaller, but there should not be any a priori reason for it to be more erratic
in some observations than in others. The constant is usually denoted by , often
abbreviated to , and the condition is written as,
Since E(μi)is 0, the population variance of μiis equal to ( ), so the condition can
also be written

( ) , of course is unknown. One of the tasks of regression analysis

is to estimate the standard deviation of the disturbance term. If this condition is not
satisfied, the OLS regression coefficients will be inefficient, but you should be able to
obtain more reliable results by using a modification of the regression technique.

2.2.3.2.3 Gauss–Markov Condition 3: μi Distributed Independently of μj ( )

NOUN 60
INTRODUCTION TO ECONOMETRICS II ECO 306

This condition states that there should be no systematic association between the values
of the
disturbance term in any two observations. For example, just because the disturbance
term is large and positive in one observation, there should be no tendency for it to be
large and positive in the next (or large and negative, for that matter, or small and
positive, or small and negative). The values of the disturbance term should be
independent of one another. The condition implies that μiμj, the population
covariance between μiandμj, is 0, because;

μiμj =E[(μi– μu)(μj– μu)] = E(μiμj) = E(μi)E(μj) = 0

…[2.29]

where, u is a value in μ as shown in ( u1 ) of Figure 2.0

Note that the population means of μiandμjare 0, by the first Gauss–Markov condition,
and that E(μiμj) can be decomposed as E(μi)E(μj) if μiandμjare generated
independently. If this condition is not satisfied, OLS will again give inefficient
estimates.

2.2.3.2.4 Gauss–Markov Condition 4: u Distributed Independently of the

Explanatory Variables

The final condition comes in two versions, weak and strong. The strong version is that
the
explanatory variables should be non-stochastic, that is, not have random components.
This is very unrealistic for economic variables, and we will eventually switch to the
weak version of the condition, where the explanatory variables are allowed to have
random components provided that they are distributed independently of the
disturbance term. However, the strong version is usually used because it simplifies the
analysis of the properties of the estimators.

X i ui ,* X i ( X i )+* ui +- ( Xi X i ) ( ui ) …[2.30]

2.2.3.3 The Normality Assumption

NOUN 61
INTRODUCTION TO ECONOMETRICS II ECO 306

In addition to the Gauss–Markov conditions, one usually assumes that the disturbance
term is
normally distributed. The reason is that if u is normally distributed, so will be the
regression coefficients, and this is useful when performing tests of hypotheses and
constructing confidence intervals for and using the regression results. The
justification for the assumption depends on the Central Limit Theorem; that, if a
random variable is the composite result of the effects of a large number of other
random variables, it will have an approximately normal distribution even if its
components do not, provided that none of them is dominant. The disturbance term u is
composed

of a number of factors not appearing explicitly in the regression equation so, even if
we know nothing about the distribution of these factors (or even their identity), we are
entitled to assume that they are normally distributed.

2.2.3.4 Unbiasedness of the Regression Coefficients

We can show that b2 must be an unbiased estimator of if the fourth Gauss–Markov

condition is satisfied:

( ) ( )
( ) 0 ( )
1 0 ( )
1 …[2.31]

since is a constant. If we adopt the strong version of the fourth Gauss–Markov

condition and assume that X is non-random, we may also take Var(X) as a given
constant, and so

( ) , ( )- …[2.32]
( )

To demonstrate that , ( )- :

, ( )- 0 ∑ ( ̅ )( ̅ )1 ∑ ,( ̅ )(

̅ )-

∑ ( ̅ ) ,( ̅ )- …[2.33]

NOUN 62
INTRODUCTION TO ECONOMETRICS II ECO 306

In the second line, the second expected value rule has been used to bring ( ) out of
the expression as a common factor, and the first rule has been used to break up the
expectation of the sum into the sum of the expectations. In the third line, the term
involving has been brought out because X is non-stochastic. By virtue of the first
Gauss–Markov condition, ( )is , and hence ( ) is also 0. Therefore
, ( )- is 0 and

( ) …[2.34]

In other words, b2 is an unbiased estimator of . We can obtain the same result with
the weak version of the fourth Gauss–Markov condition (allowing X to have a random
component but assuming that it is distributed independently of u), unless the random
factor in the nobservations happens to cancel out exactly, which can happen only by
coincidence. b2 will be different from for any given sample, but in view of unbiased
regression coefficient, there will be no systematic tendency for it to be either higher or
lower. The same is true for the regression coefficient b1.

Using [2.22]

̅ ̅ …[2.35]

Hence

( ) ( ̅) ̅ ( ) …[2.36]

Since is determined by

We have

( ) ( ) …[2.37]

because ( )is 0 if the first Gauss–Markov condition is satisfied. Hence

( ̅) ̅ …[2.38]

Substituting this into [2.36], and using the result that ( )

NOUN 63
INTRODUCTION TO ECONOMETRICS II ECO 306

( ) ( ̅) ̅ …[2.39-

Thus b1 is an unbiased estimator of provided that the Gauss–Markov conditions 1

and 4 are satisfied. Of course in any given sample the random factor will cause b1 to
differ from .

2.2.3.5 Precision of the Regression Coefficients

Now we shall consider and , the population variances of b1 and b2 about their
population means.

The following expressions give these

̅
, ] and …[2.40]
( ) ( )

Equation [2.40] has three obvious implications. First, the variances of both b1 and b2
are directly inversely proportional to the number of observations in the sample. This
makes good sense. The more information you have, the more accurate your estimates
are likely to be.

Second, the variances are proportional to the variance of the disturbance term. The
bigger the
the variance of the random factor in the relationship, the worse the estimates of the
parameters are likely to be.

Third, the variance of the regression coefficients is inversely related to the variance of
X. What is the reason for this? Remember that (1) the regression coefficients are
calculated on the assumption that the observed variations in Y are due to variations in
X, but (2) they are in reality partly due to variations in X and partly to variations in u.
The smaller the variance of X, the greater is likely to be the relative influence of the
random factor in determining the variations in Yand the more likely is regression
analysis give inaccurate estimates.

2.2.3.6 Testing Hypotheses Relating to the Regression Coefficients

NOUN 64
INTRODUCTION TO ECONOMETRICS II ECO 306

Which comes first, theoretical hypothesizing or empirical research? In practice,

theorizing and experimentation feed on each other, and questions of this type cannot
be answered. For this reason, we will approach the topic of hypothesis testing from
both directions. On the one hand, we may suppose that the theory has come first and
that the purpose of the experiment is to evaluate its acceptability. This will lead to the
execution of significance tests. Alternatively, we may perform the experiment first
and then consider what theoretical hypotheses would be consistent with the results.
This will lead to the construction of confidence intervals.

Students would already have encountered the logic underlying significance tests and
confidence
intervals in an introductory statistics course. Students will thus be familiar with most
of the concepts in the following applications to regression analysis. There is, however,
one topic that may be new: the use of one-tailed tests. Such tests are used very
frequently in regression analysis. Indeed, they are, or they ought to be, more common
than the traditional textbook two-tailed tests. It is, therefore, important that you
understand the rationale for their use, and this involves a sequence of small analytical
steps. None of this should present any difficulty, but be warned that, if students
attempt to use a shortcut or, worse, try to reduce the whole business to the mechanical
use of a few formulae, you will be asking for trouble.

2.2.3.6.1 Formulation of a Null Hypothesis

We will start by assuming that the theory precedes the experiment and that you have
some
thehypothetical relationship in your mind. For example, you may believe that the
percentage rate

of price inflation in an economy, p, depends on the percentage rate of wage inflation,

w, according to the linear equation

…[2.41]

NOUN 65
INTRODUCTION TO ECONOMETRICS II ECO 306

where and are parameters and u is a disturbance term. You might further
hypothesize that, apart from the effects of the disturbance term, price inflation is equal
to wage inflation. Under these circumstances you would say that the hypothesis that
you are going to test, known as your nullhypothesis and denoted H0, is that is equal
to 1. We also define an alternative hypothesis, denoted H1, which represents your
conclusion if the experimental test indicates that H0 is false. In the present case H1, is
simply that is not equal to 1. The two hypotheses are stated using the notation

H0
H1

In this particular case, if we believe that price inflation is equal to wage inflation, we
are trying to establish the credibility of H0 by subjecting it to the strictest possible test
and hoping that it emerges intact. In practice, however, it is more usual to set up a null
hypothesis and attack it with the objective of establishing the alternative hypothesis as
the correct conclusion. For example, consider the simple earnings function

…[2.42]

WhereEARNINGS is hourly earnings in dollars and S is years of schooling. On very

reasonable theoretical grounds, you expect earnings to be dependent on schooling, but
your theory is not strong enough to enable you to specify a particular value for .
You can nevertheless establish the dependence of earnings on schooling by the inverse
procedure in which you take as your null hypothesis the assertion that earnings does
not depend on schooling, that is, that is 0. Your alternative hypothesis is that is
not equal to 0, that is, that schooling doesaffect earnings. If you can reject the null
hypothesis, you have established the relationship, at least in general terms. Using the
conventional notation, your null and alternative hypotheses are H0
H1 , respectively.

The following discussion uses the simple regression model

NOUN 66
INTRODUCTION TO ECONOMETRICS II ECO 306

It will be confined to the slope coefficient, , but exactly the same procedures are
applied to the constant term, . We will take the general case, where you have
defined a null hypothesis that is equal to some specific value, say , and the
alternative hypothesis is that is not equal to this value (H0
); you may be attempting to attack or defend the null hypothesis as it suits your
purpose. We will assume that the four Gauss–Markov conditions are satisfied.

2.2.3.6.2 Developing the Implications of a Hypothesis

If H0 is correct, values of b2 obtained using regression analysis in repeated samples

will be distributed with mean and we will now introduce the assumption
( )

that u has a normal distribution. If this is the case, b2 will also be normally distributed,
In view of the structure of the normal distribution, most values of b2 will lie within
two standard deviations of (if is true).

2.2.3.7 Compatibility, Freakiness, and the Significance Level

Now, suppose that we take an actual sample of observations on average rates of price
inflation and wage inflation over the past five years for a sample of countries and
estimate using regression analysis. If the estimate is close to 1.0, we should almost
certainly be satisfied with the null hypothesis, since it and the sample result are
compatible with one another. But suppose, on the other hand, that the estimate is a
long way from 1.0. Suppose that it is equal to 0.7. This is three standard deviations
below 1.0. If the null hypothesis is correct, the probability of being three standard
deviations away from the mean, positive or negative, is only 0.0027, which is very
low. You could come to either of two conclusions about this worrisome result:

You could continue to maintain that your null hypothesis H0: = 1 is correct, and
that the
experiment has given a freak result. You concede that the probability of such a low
value of
b2 is very small, nevertheless it does occur 0.27 percent of the time and you reckon
that
this is one of those times.

NOUN 67
INTRODUCTION TO ECONOMETRICS II ECO 306

Or you could conclude that the regression result contradicts the hypothesis. You are
not convinced by the explanation in (1) because the probability is so small and you
think that a much more likely explanation is that is not really equal to 1. In other
words, you adopt the alternative hypothesis H1 instead.

We can summarize this decision rule mathematically by saying that we will reject the
null
hypothesis if

…[2.43]

wherez is the number of standard deviations between the regression estimate and the
hypothetical value of :

…[2.44]
( )

The null hypothesis will not be rejected if

This condition can be expressedregardingb2 and by substituting for z from

…[2.45]
( )

Multiplying through by the standard deviation of b2, one obtains

( ) ( ) …[2.46]

from which one obtains

( ) ( ) …[2.47]

[2.47] gives the set of values of which will not lead to the rejection of a specific

null hypothesis . It is known as the acceptance regionfor , at the 5 percent

significance level.

2.2.3.8 What Happens if the Standard Deviation of is Not Known

NOUN 68
INTRODUCTION TO ECONOMETRICS II ECO 306

So far we have assumed that the standard deviation of is known, which is most
unlikely in practice. It has to be estimated by the standard error of . This causes two
modifications to the test procedure. First, z is now defined using s.e.( ) instead of
s.d.( ) and it is referred to as the t statistic:

…[2.48]
( )

Second, the critical levels of t depend on upon what is known as a tdistribution instead
of a normal distribution. We will not go into the reasons for this, or even describe the t
distribution mathematically. But enough to say that it is a partner of the normal
distribution. Its exact shape depends on the number of degrees of freedom in the
regression and approximates the normal distribution increasingly closely as the
number of degrees of freedom increases. You will certainly have encountered the t
distribution in your introductory statistics course.

The estimation of each parameter in a regression equation consumes one degree of

freedom in the sample. Hence the number of degrees of freedom is equal to the
number of observations in the sample minus the number of parameters estimated. The
parameters are constant (assuming that this is specified in the regression model) and
the coefficients of the explanatory variables. In the present case of simple regression
analysis, only two parameters, , are estimated and hence the number of
degrees of freedom is n – 2. It should be emphasized that a more general expression
will be required when we come to multiple regression analysis.

The critical value of t, which we will denotetcrit, replaces the number 1.96 in [2.43], so
the
condition that a regression estimate should not lead to the rejection of a null
hypothesis H0: is

…[2.49]
( )

Hence we have the decision rule:

NOUN 69
INTRODUCTION TO ECONOMETRICS II ECO 306

rejectH0 if | |> ,
( )

do not reject if | |
( )

Where | | is the absolute value (numerical value, neglecting the sign) of t.

( )

2.2.4.0 SUMMARY

In this unit, basic understanding of the properties of regression coefficients and

hypotheses testing has been explained. As introductory step to understanding this unit,
some knowledge areas like estimators, assumptions underlying CLRM and properties
of OLS estimators were briefly discussed. These knowledge areas acquainted the
students of what is expected to be known for

better understanding of the different aspect of regression Coefficients, hypotheses

testing and the assumptions associated with studying these topics in this unit.

2.2.5.0 CONCLUSION

Assumptions and basic knowledge areas needed for the students to be acquainted with
the properties of regression coefficients and hypotheses testing have been introduced
and discussed in this unit. Most especially, the respective Gauss-Markov conditions in
the assumptions concerning disturbance term of regression analyses were discussed in
a manner that the students would be able to properly understand. Also, formulation of
null hypothesis and developing implications of a hypothesis were discussed as testing
hypotheses relating to the regression coefficients.

2.2.6.0 TUTOR-MARKED ASSIGNMENT

1.) Where performance on a game of skill is measured numerically, the improvement

that comes with practice is called a learning curve. This is especially obvious with
some arcade-type games. The first time players try a new one; they are likely to score
very little. With more attempts, their scores should gradually improve as they become

NOUN 70
INTRODUCTION TO ECONOMETRICS II ECO 306

accustomed to the game, although, obviously, there will be variations caused by the
luck factor. Suppose that the learning curve determines their scores

Yi  500  100 Xi  i

where,Y is the score, X is the number of times that they have played before, and  is a
disturbance term.
The following table gives the results of the first 20 games of a new player. X
automatically goes from 0 to 19;  was set equal to 400 times the numbers generated
by a normally distributed random variable with 0 mean and unit variance, and X and 
determined Yaccording to the learning curve.

Observation X  Y

1 0 -236 264

2 1 -96 504

3 2 -332 368

4 3 12 812

5 4 -152 748

6 5 -876 124

7 6 412 1,512

8 7 96 1,296

9 8 1,012 2,312

10 9 -52 1,348

11 10 636 2,136

12 11 -368 1,232

13 12 -284 1,416

14 13 -100 1,700

15 14 676 2,576

16 15 60 2,060

NOUN 71
INTRODUCTION TO ECONOMETRICS II ECO 306

17 16 8 2,108

18 17 -44 2,156

19 18 -364 1,936

20 19 568 2,968

Regressing Y on X, one obtains the equation (standard errors in parentheses):

S.E.E (190) (17.1)

Why is the constant in this equation not equal to 500 and the coefficient of X not equal
to 100?

What is the meaning of the standard errors?

2.) The experiment is repeated with nine other new players (the disturbance term
being generated by 400 times a different set of 20 random numbers in each
case), and the regression results for all ten players are shown in the following
table. Why do the constant, the coefficient of X, and the standard errors vary
from sample to sample?

Player Constant Standard error of constant Coefficient of X Standard error of coefficient of X

1 369 190 116.8 17.1

2 699 184 90.1 16.5

3 531 169 78.5 15.2

4 555 158 99.5 14.2

5 407 120 122.6 10.8

6 427 194 104.3 17.5

7 412 175 123.8 15.8

8 613 192 95.8 17.3

NOUN 72
INTRODUCTION TO ECONOMETRICS II ECO 306

9 234 146 130.1 13.1

10 485 146 109.6 13.1

The variance of X is equal to 33.25,and the population variance of  is equal to

160,000. Using appropriate equation, show that the standard deviation of the
probability density function of the coefficient of X is equal to 15.5. Are the standard
errors in the table good estimates of this standard deviation?

2.2.7.0 REFERENCES /FURTHER READING

Maddala, G. S., &Lahiri, K. (1992).Introduction to econometrics (Vol. 2). New York.

Carter, H. R., Griffiths, W. E., & Judge, G. (2001).Undergraduate econometrics.2nd
Ed. New York: John Wiley and Sons.
Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.
Smith, G. (2013). Econometric Principles and Data Analysis.Centre for Financial and
Management Studies SOAS, University of London, London.
Dougherty, C. (2014). Elements of econometrics.London: University of London.

UNIT 3 MULTIPLE REGRESSION ANALYSIS AND

MULTICOLLINEARITY

CONTENTS
2.3.1.0 Introduction
2.3.2.0 Objectives
2.3.3.0 Main Content
2.3.3.1 Multiple Regression Coefficients Interpretation
2.3.3.2 Properties of the Multiple Regression Coefficients
2.3.3.3 t Tests and Confidence Intervals
2.3.3.4 Consistency
2.3.4.0 Multicollinearity
2.3.4.1 Multicollinearity in Models with More Than Two Explanatory Variables
2.3.4.2 Ways to alleviate multicollinearity problems
2.3.5.0 Summary

NOUN 73
INTRODUCTION TO ECONOMETRICS II ECO 306

2.3.6.0 Conclusion
2.3.7.0 Tutor-Marked Assignment
2.3.8.0 References/Further Reading

2.3.1.0 INTRODUCTION

The multiple regression analysis is an extension of simple regression analysis. It

covers cases in which the dependent variable is hypothesized to depend on more than
one descriptive variable. Most of the multiple regression analysis is a direct extension
of the simple regression model but hasonly two new dimensions. First, when
evaluating the influence of a given descriptive variable on the dependent variable, we
would now have to face the problem of discriminating between its effects and the
effects of the other descriptive variables. Second, we shall have to tackle the problem
of model specification. Often some variables might be thought to influence the
behaviour of the dependent variable; though, they might be unconnected. We shall
have to decide which should be included in the regression equation and which should
be omitted. However, the arrangement of flow for the multiple regression analysis is
to firstly, carry out derivation of formula, then estimation procedures using values,
followed by presentation of results and lastly interpretations. As an extension of unit
2, we shall discuss multicollinearity being a problem associated with CLRM.

2.3.2.0 OBJECTIVE
The main objective of this unit is to provide broad understanding of the topic; multiple
regression analysis and appropriate alleviation measures associated with
multicollinearity problems. This understanding includes the knowledge of the
properties, principles behind the derivation of and how to interpret multiple regression
coefficients.

2.3.3.0 MAIN CONTENTS

2.3.3.1The Multiple Regression Coefficients Derivation

In the simple regression case, the values of the regression coefficients were chosen to
make the fit as good as possible in the hope of obtaining most satisfactory estimates of

NOUN 74
INTRODUCTION TO ECONOMETRICS II ECO 306

the true unknown parameters. Our earlier stated definition of goodness of fit; is the
minimization of RSS, which is the sum of squares of the residuals:
∑ …[2.50]

Where eiis again, the residual in observation i, the difference between the actual value
Yiin that observation and the value ̂ predicted by the regression equation:

̂ …[2.51]

̂ …[2.52]

It could be observed that the X variables now have two subscripts. The first identifies
the X variable and the second identifies the observation.

Applying [2.52] into [2.50];

…[2.53]

From first-order conditions for a minimum;

[2.52] will give the following equations:

∑ ( ) …[2.54]

∑ ( ) …[2.55]

∑ ( ) …[2.56]

Resulting in three equations from the three unknowns, b1, b2, and b3.

The first can easily be rearranged to express b1regardingb2, b3, and the data on Y, X2,
and X3:

̅ ̅̅̅ ̅̅̅ …[2.57]

NOUN 75
INTRODUCTION TO ECONOMETRICS II ECO 306

From [3.57] and working through [3.55] to [3.56], the following expression for b2is
obtained:

( ) ( ) ( ) ( )
( ) ( ) , ( )-

,∑ ( )( )- ,∑( )( )-
,∑( )∑( ) ∑,( )-
…[2.58]

Similarly, theexpressionofb3 can be obtained by switching X2 and X3 in [2.58].

Clearly, the principles behind the derivation of the regression coefficients have been
shown to be the same for multiple regression as that of the simple regression. But, it
should also be observed that the expressions are however different and so should not
try to use expressions derived for simple regression in a multiple regression situations.

A generalized framework for the multiple regression model is

Yi  1  2 X2i  ...k Xki  i …[2.59]
We may write [2.59] for three variables as,
Yi  1  2 X2i  3 X3i  i …[2.60]
whereY is the dependent variable, X2 and Xk (kth term) the regressors,  the stochastic
disturbance term and i the ith (tth, if in time series) observation. Also 1 , 2 and k are
the partial regression coefficients but 1 is the intercept term which gives the mean
effect on Y of all the variables excluded from the model. That is, in the case of [2.50],
when X2 and Xk are set equal to zero.

Zero mean value of i in [2.60] is;

E (i | X2i , X3i )  0 …[2.61]

2.3.3.1 Multiple Regression Coefficients Interpretation

Discriminate between the effects of the explanatory variables and making allowance
for the fact that they may be correlated is enabled in multiple regression analysis. The
regression coefficient of each X variable provides an estimate of its influence on Y.

NOUN 76
INTRODUCTION TO ECONOMETRICS II ECO 306

There are two ways in which this can be demonstrated. First is the case where there
are only two explanatory variables; to demonstrate that the estimators are unbiased if
the model is correctly specified and the Gauss–Markov conditions are fulfilled.

The second method is to run a simple regression of Y on one of the X variables, having
first purged both Y and the X variable of the components that could be accounted for
by the other explanatory variables. The estimate of the slope coefficient and its
standard error thus obtained are the same as in the multiple regression. It follows that
a scatter diagram plotting the purged Y against the purged X variable will provide a
valid graphical representation of their relationship that can be obtained in no other
way.

…[2.62]

If thegraphical illustration is particularly interested in, in the relationship between

earnings and schooling; a direct plot of EARNINGS on S would give a distorted view
of the relationship. This is because ASVABC is positively correlated with S and having
some consequences as S increases. These are [1] EARNINGS will likely increase,
because is positive; [2] ASVABC will tend to increase, because S and ASVABC are
positively correlated; and [3] EARNINGS will receive a lift due to the increase in
ASVABC and the fact that is positive. That is, the variations in EARNINGS will
overstate the apparent influence of S because in part they will be due to associated
variations in ASVABC. And the outcome of this is that in a simple regression the
estimator of will be biased. The graphical illustration is shown in Figure 3.1.

NOUN 77
INTRODUCTION TO ECONOMETRICS II ECO 306

Figure 3.1: Regression of EARNINGS residuals on S residuals

2.3.3.2 Properties of the Multiple Regression Coefficients

Concerning simple regression analysis, the regression coefficients should be thought

of as different categories of random variables whose random components are related
to the existence of the disturbance term in the model. Each regression coefficient is
calculated as a function of the values of Y and the explanatory variables in the sample.
Y, in turn, is determined by the explanatory variables and the disturbance term. It
follows that the regression coefficients are indeed determined by the values of the
explanatory variables and the disturbance term, in which their properties depend on
critically upon the properties of the disturbance term.
In continuation of the assumption that the Gauss–Markov conditions are satisfied,
which are:
(i) that the expected value of uin any observation is 0
(ii) that the population variance of its distribution is the same for all observations
(iii) that the population covariance of its values in any two observations is 0, and
(iv) that it is distributed independently of any explanatory variable.

The first three conditions are the same as for simple regression analysis but (iv) is a
generalization of (i) to (iii).
Furthermore, there are two practical requirements to be met.

NOUN 78
INTRODUCTION TO ECONOMETRICS II ECO 306

(i) There must be enough data to fit the regression line. That is, there must be
at least as many (independent) observations as there are parameters to be
estimated.
(ii) There must not be an exact linear relationship among the explanatory
variables.

2.3.3.3 t Tests and Confidence Intervals

The t tests on the regression coefficients are performed in the same way as for simple
regression analysis. Particular attention should, however,be taken when looking up the
critical level of t at any given significance level. It depends on the number of degrees
of freedom (n – k); the number of observations n minus the number of parameters
estimated k.
The confidence intervals are also obtained in the same manner as in simple regression
analysis and equally based on the number of degrees of freedom (n – k).

2.3.3.4 Consistency

Once the fourth Gauss–Markov condition is satisfied, OLS yields consistent estimates
in the multiple regression models, as is the case in thesimple regression model. One
condition for consistency is that when n becomes large, the population variance of the
estimator of each regression coefficient tends to 0, and the distribution falls to a spike.
The other condition for consistency is since the estimator is unbiased, the spike would
be located at the true value.

2.3.4.0 MULTICOLLINEARITY

In most situations, the available data for use in multiple regression analysis would not
provide significant solutions to problems at hand. The reason being that the standard
errors are very high, or the t test ratios are very low.Which means the confidence
intervals for such parameters are very wide. A situation of this nature occurs when the
explanatory variables show little variation and high intercorrelations. Multicollinearity
is the aspect of the situation where the explanatory variables are highly
intercorrelated.

NOUN 79
INTRODUCTION TO ECONOMETRICS II ECO 306

Let‟s look at multicollinearity in a model with two explanatory variables. It would be

observed that the higher the correlation between the explanatory variables, the larger
the population variances of the distributions of their coefficients and the greater the
possibility of attaining irregular estimates of the coefficients.
You should, however, bear in mind that a high correlation does not necessarily lead to
poor estimates. If all the other elements determining the variances of the regression
coefficients are properly in the number of observations and the sample variances of
the explanatory variables are large and the variance of the disturbance term small,
good estimates could still be obtained. Multicollinearity, therefore, must be caused by
a mixtureof a high correlation and one or more of the other elements being
inappropriate. This is a matter of degree and not kind of element of which any
regression will suffer from it to some extent unless all the explanatory variables are

uncorrelated. But the consequence is only taken into consideration when it is

obviously going to have aserious effect on the regression results.

It is a common problem in time series regressions, particularly where the data consists
of a series of observations on the variables over a number of time periods. Which may
give rise to multicollinearity if two or more of the explanatory variables are highly
correlated in a strong time trend.

Using Table 3.1 as an example let‟s consider first the case of exact multicollinearity
where the explanatory variables are perfectly correlated.

Table 3.1

X2 X3 Y Change in X2 Change in X3 Approximate change in Y

10 19 51+u1 1 1 5

11 21 56+ u2 1 1 5

12 23 61+ u3 1 1 5

13 25 66+ u4 1 1 5

14 27 71+ u5 1 1 5

NOUN 80
INTRODUCTION TO ECONOMETRICS II ECO 306

15 29 76+ u6 1 1 5

Let [2.40] be the true relationship, that is;

…[2.63]

Suppose that there is a linear relationship between and :

…[2.64]

and suppose that X2 increases by one unit in each observation. X3 will increase by two
units, and Y by approximately five units as indicated in Table 3.1. Applying the linear
relationship between X2 andX3 in manipulating [2.40] will result in different
conclusions for Y.

In such a situation it is impossible for regression analysis, or any other technique for
that matter, to distinguish between these possibilities. You would not even be able to
calculate the regression

coefficients because both the numerator and the denominator of the regression
coefficients would collapse to 0. This willbe demonstrated with the general two-
variable case. Suppose

…[2.65]

And

…[2.66]

Substituting for in [3.58] gives

( ) ( ) (, - ) ( , -)
( ) ( ) , ( , -)-

( ) ( ) ( ) ( )
= ( ) ( ) , (
…[2.67]
-)-

From Variance Rule 4, the additive in the variances can be dropped. A similar rule
could be developed for covariances, since an additive does not affect them either.

NOUN 81
INTRODUCTION TO ECONOMETRICS II ECO 306

Therefore,

( ) ( ) ( ) ( )
( ) ( ) , (
…[2.68]
-)-

( ) ( ) ( ) ( )
( ) ( ) , (
…[2.69]
-)-

Which is unusual for there to be an exact relationship among the explanatory variables
in a regression. So, when this occurs, it is typical because there is a logical error in the
specification.

2.3.4.1 Multicollinearity in Models with More Than Two Explanatory Variables

The previous discussion of multicollinearitywas restricted to the case where there are
two
explanatory variables. In models with a greater number of explanatory variables,
multicollinearity may be caused by an approximately linear relationship among them.
It may be difficult to discriminate between the effects of one variable and those of a
linear combination of the remainder. In the model with two explanatory variables, an
approximately linear relationship automatically means a high correlation, but when
there are three or more, this is not necessarily the case. A linear relationship does not
inevitably imply high pairwise correlations between any of the variables. The effects
of multicollinearity are the same as in the case with two explanatory variables and as
in that

case, the problem may not be serious if the population variance of the disturbance
term is small, the number of observations large and the variances of the explanatory
variables are equally large.

2.3.4.2 Ways to alleviate multicollinearity problems

Two categories exist to alleviate multicollinearity problems:

i. The direct attempts to improve the four conditions responsible for the
reliability of the regression estimates, and

NOUN 82
INTRODUCTION TO ECONOMETRICS II ECO 306

ii. The indirect methods.

First, you may try to reduce . The disturbance term is the joint effect of all the
variables
influencing Y that you have not included explicitly in the regression equation. If you
can think of an important variable that you have omitted, and is therefore contributing
to u, you will reduce the population variance of the disturbance term if you add it to
the regression equation.

Second, consider n, the number of observations. If you are working with cross-section
data (individuals, households, enterprises, etc.) and you are undertaking a survey, you
could increase the size of the sample by negotiating a bigger budget. Alternatively,
you could make a fixed budget go further by using a technique known as clustering.

A further way of dealing with the problem of multicollinearity is to use minor

information, if available, concerning the coefficient of one of the variables.

…[2.70]

For example, suppose that Y in equation is the aggregate demand for a category of
consumer expenditure, X is aggregate disposable personal income, and P is a price
index for the category. To fit a model of this type, you would use time series data. If X
and P possess strong time trends and are therefore highly correlated, which is often
the case with time series variables, multicollinearity is likely to be a problem.
Suppose, however, that you also have cross-section data on Y and X derived from a
separate household survey. These variables will be denotedY' and X' to indicate that
the data are household data, not aggregate data. Assuming that all the households in
the survey were paying roughly the same price for the commodity, one would fit the
simple regression

̂ …[2.71]

Now substitute for in the time series model

…[2.72]

Subtract from both sides,

NOUN 83
INTRODUCTION TO ECONOMETRICS II ECO 306

…[2.73]

And regress on price. This is a simple regression, so multicollinearity

has been eliminated.

There are, however, two possible problems with this technique.

First, the estimate of depends on the accuracy of the estimate of b2', and this of
course is subject to sampling error.
Second, you are assuming that the income coefficient has the same meaning in time
series and cross-section contexts, and this may not be the case.
For many commodities, the short-run and long-run effects of changes in income may
differ because expenditure patterns are subject to inertia. A change in income can
affect expenditure both directly, by altering the budget constraint, and indirectly,
through causing a change in lifestyle, and the indirect effect is much slower than the
direct one. As a first approximation, it is commonly argued that time series
regressions, particularly those using short sample periods, estimate short-run effects
while cross-section regressions estimate long-run ones.

For the indirect methods to alleviate multicollinearity problems. If the correlated

variables are similar conceptually, it may be reasonable to combine them into some
overall index.

2.3.7.0 SUMMARY

In this unit, we discussed the multiple regression model, a model in which there is
more than one descriptive variable but a direct extension of the simple regression
model having two new dimensions. We introduced the arrangement of flow for the
multiple regression analysis and started with the derivation of formula by showing the
linkage of the simple regression model and the multiple regression. However, we
briefly discussed on interpretation and properties of multiple regression coefficients.
For more understanding, the student may use the reference materials to look up
estimation procedures using values and presentation of results, aspects of arrangement
of flow for the multiple regression analysis not discussed. Finally, the concept of
multicollinearity as an existence of linear relationship in the midst of regressors was
equally discussed in this unit. Students are shown two ways to alleviate
multicollinearity being a problem associated with CLRM.

NOUN 84
INTRODUCTION TO ECONOMETRICS II ECO 306

First, when evaluating the influence of a given descriptive variable on the dependent
variable, we would now have to face the problem of discriminating between its effects
and the effects of the other descriptive variables. Second, we shall have to tackle the
problem of model specification. Often some variables might be thought to influence
the behaviour of the dependent variable; though, they might be unconnected. We shall
have to decide which should be included in the regression equation and which should
be omitted. However, the arrangement of flow for the multiple regression analysis is
to firstly, carry out derivation of formula, then estimation procedures using values,
followed by presentation of results and lastly interpretations.

2.3.6.0 CONCLUSION

The features of multiple regression analyses and multicollinearity introduced in this

unit are extension of unit 2. Here, we pointed out some of the complications arising
from the introduction of several descriptive variables. In the discussions, we explained
that when we go beyond the two-variable model and consider multiple regression
models we add the assumption that there is no perfect multicollinearity (assumption
10 of CLRM). That is, there are no perfect linear relationships among the descriptive
variables when two or more of these variables move together and difficult to
determine their separate influences.

2.3.7.0 TUTOR-MARKED ASSIGNMENT

1.) The following earnings functions were fitted separately for males and females
(standard errors in parentheses):

Males
̂
S.E.E (2.8420) (0.2434) (0.0600)
Females
̂
S.E.E (2.6315) (0.1910) (0.0577)

3.) Explain why the standard errors of the coefficients of S and ASVABC are
greater for the male subsample than for the female subsample, and why the
difference in the standard errors are relatively large for S.

NOUN 85
INTRODUCTION TO ECONOMETRICS II ECO 306

2.3.8.0 REFERENCES /FURTHER READING

Maddala, G. S., &Lahiri, K. (1992).Introduction to econometrics (Vol. 2). New York.

Dougherty, C. (2003). Numeracy, literacy and earnings: evidence from the National
Longitudinal Survey of Youth. Economics of education review, 22(5), 511-521.
Smith, G. (2013). Econometric Principles and Data Analysis.Centre for Financial and
Management Studies SOAS, University of London, London.
James H. Stock and Mark W. Watson (2010).Introduction to Econometrics.3rd Ed.
Addison-Wesley Series in Economics.

NOUN 86
INTRODUCTION TO ECONOMETRICS II ECO 306

UNIT 4: TRANSFORMATIONS OFVARIABLES

CONTENTS
2.4.1.0 Introduction
2.4.2.0 Objectives
2.4.3.0 Main Content
2.4.4.0 Summary
2.4.5.0 Conclusion
2.4.6.0 References/Further Reading

2.4.1.0 INTRODUCTION

In model transformation, the functional form of an equation or model determines the

estimation techniques and interpretation of results obtained from it. Transforming a
variable involves using mathematical procedure to modify its measured values. Single
equation (or any other form of equation) may be in different forms. There are two
kinds of transformations and generally, models can be of the form;
i. Linear transformation; this preserves the linear relationships between variables
(parameters and variables are linear). That is the correlation between xand y
(say) would be unchanged after a linear transformation.
Examples of a linear transformation to variable x would be multiplying x by a
constant, dividing x by a constant, or adding a constant to x.
ii. Nonlinear transformation; A nonlinear transformation changes (increases or
decreases) linear relationships between variables and, thus, changes the
correlation between variables.
Examples of a nonlinear transformation of variable x would be taking the square root
of x or the reciprocal of x. By extension, nonlinear transformation is a non-linear
model that can be made linear. For example;
Yt  AXt eut ... is an example of production function that can be made
linear by taking logarithms, that is; ln Yt  ln A   ln Xt  ut .

NOUN 87
INTRODUCTION TO ECONOMETRICS II ECO 306

In regression, however, a transformation to achieve linearity is a special kind of

nonlinear transformation. It is a nonlinear transformation that increases the linear
relationship between two variables.

2.4.2.0 OBJECTIVE
The main objective of this unit is to show that regression analysis can be extended to
fit nonlinear models through transformation of nonlinear model that can be made
linear.

2.4.3.0 MAIN CONTENT

A limitation out of other limitations of linear regression analysis is that it is contained

in its very name, in that it can be used to fit only linear equations where every
explanatory term, except the constant, is written in the form of a coefficient multiplied
by variable:

…[2.74]

Y equations such as the two below are non-linear

Y  1  2 1
X …[2.75]

And

…[2.76]

Nevertheless, both [2.75] and [2.76] have been suggested as suitable forms for Engel
curves, (the relationship between the demand for a particular commodity, Y and
income, X). As an illustration, given data on Y and X, how could one estimate the
parameters in these equations? Actually, in both cases, with a little
preparation one can actually use linear regression analysis.

Here, first, note that [2.74] is linear in two ways. The right side is linear in variables
because the variables are included exactly as defined, rather than as functions. It,
therefore, consists of a weighted sum of the variables, the parameters being the
weights. The right side is also linear in the parameters since it consists of a weighted
sum of these as well, the X variables being the weights in this respect.

NOUN 88
INTRODUCTION TO ECONOMETRICS II ECO 306

For the purpose of linear regression analysis, only the second type of linearity is
important.
Nonlinearity in the variables can always be sidestepped by using appropriate
definitions.

For example, suppose that the relationship was of the form

√ …[2.77]

By defining Z2= , Z3=√ , Z4 = etc, the relationship can be rewritten

…[2.78]

and it is now linear in variables as well as in parameters. This type of transformation

is only beautifying, and you will usually see the regression equation presented with
the variables written in their nonlinear form. This avoids the need for explanation and
extra notation.

But [2.76] is nonlinear in both parameters and variables and cannot be handled by a
mere redefinition. That is, even if attempted, the equation cannot be made linear by
defining Z = and replacing with Z; since you do not know , you have no way
of calculating sample data for Z.

1
However, you could define Z  , the equation now becomes
X

…[2.79]

and this is linear, which is the regress of Y onZ. The constant term in the regression
will be an estimate of and the coefficient of Z will be an estimate of .

2.4.4.0 SUMMARY

This unit discussed transformation of variables. Two kinds of transformations were

discussed, the linear transformation in which the linear relationships between the
parameters and variables are preserved after transformation. As well as the nonlinear

NOUN 89
INTRODUCTION TO ECONOMETRICS II ECO 306

transformation in which there is increase or decrease in the linear relationship of the

variables involved.

2.4.5.0 CONCLUSION

In this unit the concept of transformation of variables is discussed to show that

regression analysis can be extended to fit nonlinear models through transformation of
non-linear models. That nonlinearity in the variables can always be sidestepped by
using appropriate definitions. Example of these definitions is taking logarithm of the
nonlinear model and application of the least squares principle when the model cannot
be linearised.

2.4.6.0 REFERENCES /FURTHER READING

Maddala, G. S., &Lahiri, K. (1992).Introduction to econometrics (Vol. 2). New York.

Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.
Dougherty, C. (2014). Elements of econometrics.London: University of London.

UNIT 5: DUMMY VARIABLES

NOUN 90
INTRODUCTION TO ECONOMETRICS II ECO 306

CONTENTS
2.5.1.0 Introduction
2.5.2.0 Objectives
2.5.3.0 Main Content
2.5.3.1 The Dummy Variable Trap
2.5.3.2 Change of Reference Category
2.5.3.3 Slope Dummy Variables
2.5.4.0 Summary
2.5.5.0 Conclusion
2.5.6.0 Tutor-Marked Assignment
2.5.7.0 References/Further Reading

2.5.1.0 INTRODUCTION

It sometimes happens that some descriptive variables do exist in our regression

equation,and/or the factors that you would like to introduce into a regression model
are qualitative (racial, sex or age differences) in nature and therefore not measurable
in numerical terms. In such circumstances, dummy variables are utilised.

2.5.2.0 OBJECTIVE

The main objective of this unit is to provide basic understanding of the topic „Dummy
Variable‟ through the use of imitation variables existing or being introduced into a
regression equation to solve some variables that are qualitative or immeasurable in
numerical terms.

2.5.3.0 MAIN CONTENTS

The inherent assumption for the application of dummy variables is that the regression
lines for the different groups differ only in the intercept term but have the same slope
coefficients. For example; (1). You are investigating the relationship between
schooling x and earnings y, and you have both males and females in your sample. You
would like to see if the sex of the respondent makes a difference.

NOUN 91
INTRODUCTION TO ECONOMETRICS II ECO 306

(2). You are investigating the relationship between income and expenditure in
Cameroun, and your sample includes both English-speaking and French-speaking
households. You would like to find out whether the ethnic difference is relevant.

(3). You have data on the growth rate of GDP per capita and foreign aid per capital for
a sample of developing countries, of which some are democracies and some are not.
You would like to investigate whether the impact of foreign aid on growth is affected
by the type of government.

A solution to these examples would be to run separate regressions for the two
categories and see if the coefficients are different. Alternatively, you could run a
single regression using all the observations together, measuring the effect of the
qualitative factor with what is known as a dummy variable. This effect has the two
important advantages of providing a simple way of testing whether the effect of the
qualitative factor is significant

The qualitative variable has four categories, and we need to develop a more elaborate
set
of dummy variables. The standard procedure is to choose one category as the
reference category to which the basic equation applies, and then to define dummy
variables for each of the other categories. In general, it is good practice to select the
dominant or most normal category, if there is one, as the reference category.

Accordingly, we will define dummy variables for the other three types. TECH will be
the dummy variable for the technical schools: TECH is equal to 1 if the observation
relates to a technical school, 0 otherwise. Similarly, we will define dummy variables
WORKER and VOC for the skilled workers‟ schools and the vocational schools. The
regression model is now

…[2.80]

Where are coefficients that represent the extra overhead costs of the
technical, skilled workers‟, and vocational schools, relative to the cost of a general
school. Note that you do not include a dummy variable for the reference category, and
that is the reason that the reference category is usually described as the omitted

NOUN 92
INTRODUCTION TO ECONOMETRICS II ECO 306

category. Note that we do not make any prior assumption about the size, or even the
sign, of the coefficients.

2.5.3.1 The Dummy Variable Trap

What would happen if you included a dummy variable for the reference category?
There would be two consequences.

i. Were it is possible to compute regression coefficients, you would not be

able to give them an interpretation. The coefficient b1 is a basic estimate of
the intercept, and the coefficients of the dummies are the estimates of the
increase in the intercept from this basic level, but now there is no definition
of what is basic, so the interpretation collapses.
ii. The other consequence is that the numerical procedure for calculating the
regression coefficients will break down, and the computer will simply send
you an error message (or possibly, in sophisticated applications, drop one of
the dummies for you). Suppose that there are m dummy categories, and you
define dummy variables D1... Dm.

Then, in observation i, ∑ because one of the dummy variables will be equal

to 1 and all the others will be equal to 0. But the intercept is really the product of
the parameter and a special variable whose value is 1 in all observations. Hence,
for all observations, the sum of the dummy variables is equal to this special variable,
and one has an exact linear relationship among the variables in the regression model.
As a consequence the model is subject to a special case of exact multicollinearity,
making it impossible to compute regression coefficients.

2.5.3.2 Change of Reference Category

The skilled workers' schools are considerably less academic than the others, even the
technical schools. Suppose that we wish to investigate whether their costs are
significantly different from the others. The easiest way to do this is to make them the
omitted category (reference category). Then the coefficients of the dummy variables
become estimates of the differences between the overhead costs of the other types of
school and those of the skilled workers' schools. Since skilled workers' schools are

NOUN 93
INTRODUCTION TO ECONOMETRICS II ECO 306

now the reference category, we need a dummy variable, which will be called GEN, for
the general academic schools. The model becomes

…[2.81]

where are the extra costs of technical, vocational, and general schools
relative to skilled workers‟ schools.

2.5.3.3 Slope Dummy Variables

We have so far assumed that the qualitative variables we have introduced into the
regression model are responsible only for shifts in the intercept of the regression line.
We have implicitly assumed that the slope of the regression line is the same for each
category of the qualitative variables.

This is not necessarily a plausible assumption, and we will now see how to relax it,
and test it, using the device known as a slope dummy variable (also sometimes known
as an interactive dummy variable).

The assumption that the marginal cost per student is the same for occupational and
regular schools is unrealistic. Because occupational schools incur expenditure on
training materials related to the number of students, and the staff-student ratio has to
be higher in occupational schools because workshop groups cannot be, or at least
should not be, as large as academic classes. We can relax the assumption by
introducing the slope dummy variable, NOCC, defined as the product of N and OCC:

…[2.82]

If this is rewritten

( ) …[2.83]

it can be seen that the effect of the slope dummy variable is to allow the coefficient of
N for
occupational schools to be greater than that for regular schools. If OCC is 0, so is
NOCC and the equation becomes

…[2.84]

If OCC is 1, NOCC is equal to N and the equation becomes

NOUN 94
INTRODUCTION TO ECONOMETRICS II ECO 306

( ) …[2.85]

is thus the incremental marginal cost associated with occupational schools, in the
same way that is the incremental overhead cost associated with them.

2.5.5.0 SUMMARY

In this unit, the essentials and applications of the concept of dummy variable
estimation was introduced. By way of illustrative example to the students, an
investigation is made which showed that in the application of dummy variables the
regression lines for the different groups differ only in the intercept term but have the
same slope coefficients. We also discussed two traps that may occur in the application
of dummy variable. First, change in reference category, in which a reference variable
is created out of the variables in consideration for analysis. Second, the slope
dummyvariables; which considered the slope of the regression line not being the same
for each category of the qualitative variables.

2.5.4.0 CONCLUSION

This unit discussed the use of dummy variable estimation to solve some variables that
are existing or being introduced into a regression equation which are qualitative or
immeasurable in numerical terms. In this discussion, students are made aware that the
inherent assumption for the application of dummy variables is that the regression lines
for the different groups differ only in the intercept term but have the same slope
coefficients.

2.5.6.0 REFERENCES /FURTHER READING

N Gujaratti, D. (2004). Basic econometrics. McGraw-Hill, New York.

Maddala, G. S., &Lahiri, K. (1992).Introduction to econometrics (Vol. 2). New York:
Macmillan.Dougherty, C. (2007). Introduction to econometrics. Oxford University
Press, USA.
Dougherty, C. (2014). Elements of econometrics.London: University of London.

NOUN 95
INTRODUCTION TO ECONOMETRICS II ECO 306

UNIT 6: SPECIFICATION OFREGRESSION VARIABLES:A PRELIMINARY

SKIRMISH

CONTENTS
2.6.1.0 Introduction
2.6.2.0 Objectives
2.6.3.0 Main Content
2.6.3.1 Model Specification of Regression Variables
2.6.4.0 Summary
2.6.5.0 Conclusion
2.6.6.0 References/Further Reading

2.6.1.0 INTRODUCTION

The construction of an economic model involves the specification of the relationships

that constitute it, the specification of the variables that participate in each relationship
and the mathematical function representing each relationship.

2.6.2.0 OBJECTIVE

The main objective of this unit is to provide a general understanding of the topic
„Specification of Regression Variable‟. This include the creating an opportunity for
the students to know that model specification denotes the determination of which
independent variables may be included in or omitted from a regression equation.

2.6.3.0 MAIN CONTENTS

2.6.3.1 Model Specification

The knowledge of exactly which descriptive variables ought to be included in the

equation helps when we undertake regression analysis, our task would equally be
limited to calculating estimates of their coefficients, confidence intervals for these
estimates and so on. In practice, however, we can never be sure that we have specified
the equation properly. Economic theory ought to provide a guide, but thetheory is
never flawless. Unaware, we might be including some variables that ought not to be in
the model and we might be leaving out others that ought to be incorporated.

NOUN 96
INTRODUCTION TO ECONOMETRICS II ECO 306

Existing properties of the regression estimates of the coefficients depend significantly

on the validity of the specification of the model. The consequences of misspecification
of the variables in a relationship are stated below.

i. When a variable that ought to be included is left out, the regression

estimates are in general (but not always) biased. The standard errors of the
coefficients and the corresponding t tests are in general invalid. Another
serious consequence of omitting a variable that ought to be included in the
regression is that the standard errors of the coefficients and the test statistics
are in general invalidated. This means of course that you are not in principle
able to test any hypotheses with your regression results.
ii. On the other hand, if you include a variable that ought not to be in the
equation, the regression coefficients are in general (but not always)
inefficient but not biased. The standard errors are in general valid but,
because the regression estimation is inefficient, they will be needlessly
large.

2.6.4.0 SUMMARY

In this unit, the specification of regression variables at a preliminary skirmish is

discussed. In general, an introductory opportunity is created for the students to know
that the specification of a regression model should be based primarily on theoretical
considerations rather than what may be obtainable in practice.

2.6.5.0 CONCLUSION

The specification of regression variables at a preliminary skirmish was explained in

this unit. And briefly made the students aware that this is one of the foundational
econometrics topic that prepares the readers for intermediate econometrics. That is,
specification of model is the first and most critical of the stages in regression analysis
which students may use to identify functions and solve problems associated with all
other topics discussed in this module.

2.6.6.0 REFERENCES /FURTHER READING

NOUN 97
INTRODUCTION TO ECONOMETRICS II ECO 306

Maddala, G. S., &Lahiri, K. (1992).Introduction to econometrics (Vol. 2). New York:

Macmillan.Dougherty, C. (2007). Introduction to econometrics. Oxford University
Press, USA.
Stock, J. H., & Watson, M. W. (2015).Introduction to econometrics. Pearson.

NOUN 98
INTRODUCTION TO ECONOMETRICS II ECO 306

MODULE 3: HETEROSCEDASTICITY

CONTENTS
3.1.1.0 Introduction
3.1.2.0 Objectives
3.1.3.0 Main Content
3.1.3.1 Heteroscedasticity and Its Effects
3.1.3.2 Likely Sources of Heteroscedasticity
3.1.3.3 Detection of Heteroscedasticity
3.1.3.4 The Spearman Rank Correlation Test
3.1.3.5 The Goldfeld–Quandt Test
3.1.3.6 The Glejser Test
3.1.3.6 Solution to Heteroscedasticity
3.1.3.7 Consequencesof Heteroscedasticity
3.1.4.0 Summary
3.1.5.0 Conclusion
3.1.6.0 Tutor-Marked Assignment
3.1.7.0 References/Further Reading

3.1.1.0 INTRODUCTION

The general aim of this module is to provide you with a thorough understanding of the
violation of one of the classical assumptions, equal variances (homoscedastic). The
properties of the estimators of the regression coefficients depend on the properties of
the disturbance term in the regression model. In this module, we shall be looking at
some of the problems that arise when violations of the Gauss–Markov conditions, the
assumptions relating to the disturbance term, are not satisfied. Basic understanding of
heteroscedasticity (unequal-variances) will be likewise explained.

3.1.2.0 OBJECTIVE

The main objective of this unit is to provide a platform for the students to understand
that in statistic, heteroscedasticity is a collection of random variables and the absence
of it is homoscedasticity.

NOUN 99
INTRODUCTION TO ECONOMETRICS II ECO 306

3.1.3.0 MAIN CONTENTS

3.1.3.1 Heteroscedasticity and Its Effects

Gauss–Markov second conditions listed in the previous module states; that the
variance of the disturbance term in each observation should be constant. This sounds
peculiar and needs a bit of explanation. The disturbance term in each observation has
only one value, so what can be meant by its "variance"?
The focus point of discussion here is, its potential behaviour before the sample is
generated. So when the model is written as;

…[3.01]

Figure 1.1 Homoscedasticity

[3.01]has in it the first two Gauss–Markov conditions stating that the disturbance
terms , ..., in the nobservations are drawn from probability distributions that have
0 mean and the same variance. Their actual values in the sample will sometimes be
positive, sometimes negative, sometimes relatively far from 0, sometimes relatively
close, but there will be no a priori reason to anticipate a particularly erratic value in
any given observation. To put it another way, the probability of ureaching a given
positive or negative value will be the same in all observations. This condition is
known as homoscedasticity, which means "same dispersion".

NOUN 100
INTRODUCTION TO ECONOMETRICS II ECO 306

Figure 1.1 is a depiction of homoscedasticity. For a simple illustration, the sample in

Figure 1.1 contains only five observations. Let us start with the first observation,
where X has the value X1. If

there were no disturbance term in the model, the observation would be represented by
the circle vertically above X1 on the line The effect of the disturbance
term is to shift the observation upwards or downwards vertically. The
potentialdistribution of the disturbance term, before the observation has been
generated, is shown by the normal distribution centered on the circle. The actual value
of the disturbance term for this observation turned out to be negative, the observation
being represented by the darkened indicator. The potential distribution of the
disturbance term, and the actual outcome, are shown in a similar way for the other
four observations. Although homoscedasticity is often taken for granted in regression
analysis, in some contexts it may be more reasonable to suppose that the potential
distribution of the disturbance term is different for different observations in the
sample. This is illustrated in Figure 1.2 where the variance of the potential distribution
of the disturbance term is increasing as X increases. This does not mean that
thedisturbance term will necessarilyhave a particularly large (positive or negative)
value in anobservation where X is large, but it does mean that the a priori probabilityof
having an erratic value will be relatively high. This is an example of
heteroscedasticity, which means "differing dispersion".

Mathematically, homoscedasticity and heteroscedasticity may be defined:

Homoscedasticity: same for all observations
Heteroscedasticity: not the same for all observations

NOUN 101
INTRODUCTION TO ECONOMETRICS II ECO 306

Figure 1.2 Heteroscedasticity

Figure 1.3 Model with a heteroscedastic disturbance term

Figure 1.3shows how a typical scatter diagram would look if Y were an increasing
function of X and the heteroscedasticity were of the type shown in Figure 1.2. It could
be seen that, although the observations are not necessarily further away from the non-
stochastic component of the relationship, represented by the line ,
there is a tendency for their dispersion to increase as X increases. Thus this particular
Gauss–Markov condition does not seem to have been used anywhere in the analysis so

NOUN 102
INTRODUCTION TO ECONOMETRICS II ECO 306

far, so it might look almost irrelevant. In particular, the proofs of the unbiasedness of
the OLS regression coefficients did not use this condition. There are however two
explanations for the presence of heteroscedasticity.

The first explanation has to do with making the variances of the regression
coefficients as small as possible, so that in a probabilistic sense, maximum precision is
achieved. If there is no heteroscedasticity and if the other Gauss–Markov conditions
are satisfied, the OLS regression coefficients have the lowest variances of all the
unbiased estimators that are linear functions of the observations of Y. If
heteroscedasticity is present, the OLS estimators are inefficient because there are still
other estimators that have smaller variances and are still unbiased.

The other reason is that the estimators of the standard errors of the regression
coefficients will be wrong. This is because their computation is based on the
assumption that the distribution of the disturbance term is homoscedastic. Otherwise,
they are biased. As a consequence, the ttests and also the usual Ftests will be invalid.
It is therefore quite likely that the standard errors will be underestimated, so the
tstatistics will be overestimated which will have a misleading impression of the
precision of the regression coefficients. The coefficient may appear significantly
different from 0, at a given significance level, when in fact, it is not. The inefficiency
property can be explained quite easily assuming that heteroscedasticity of the type
displayed in Figures 1.2 and 1.3 is present.

Which is an observation where the potential distribution of the disturbance term has a
small standard deviation, similar to that of Figure 1.1.

3.1.3.2 Likely Sources of Heteroscedasticity

For heteroscedasticity, it is likely to be a problem when the values of the variables in

the sample vary substantially in different observations. Given that ,
the variations in the omitted variables and the measurement errors that are jointly
responsible for the disturbance term (u) would be somewhat small when Yand Xare
small and large when they are large. This is simply because economic variables in
such a true relationship tend to move in size together.

3.1.3.3 Detection of Heteroscedasticity

NOUN 103
INTRODUCTION TO ECONOMETRICS II ECO 306

There seems to be no limit to the different possible types of heteroscedasticity, and

consequently, a large number of different tests appropriate for different conditions
have been suggested. The attention here would, however,be focused on three tests that
hypothesize a relationship between the variance of the disturbance term and the size of
the explanatory variable(s). These would be the Spearman rank correlation, Goldfeld–
Quandt, and Glejser tests.

3.1.3.4 The Spearman Rank Correlation Test

This test assumes that the variance of the disturbance term is either increasing or
decreasing as Xincreases and that there will be a correlation between the absolute size
of the residuals and the size of Xin an OLS regression. The data on Xand the absolute
values of the residuals are both ranked, and the rank correlation coefficient is defined
as

∑
…[3.02]
( )

whereDiis the difference between the rank of Xand the rank of ein observation i.
Under the assumption that the population correlation coefficient is 0, the rank
correlation
1
coefficient has a normal distribution with 0 mean and variance in large
(n  1)

samples. Theappropriate test statistic is therefore √ and the null hypothesis of

homoscedasticity will be rejected at the 5 percent level if its absolute value is greater
than 1.96 and at the 1 percent level if its absolute value is greater than 2.58, using two-
tailed tests. If there is more than one explanatory variable in the model, the test may
be performed with any one of them.

NOUN 104
INTRODUCTION TO ECONOMETRICS II ECO 306

Example

Table 1.1

Using the data in Table 1.1 above, an OLS regression of manufacturing output on
GDP yields the following result (standard errors in parentheses):

S.E.E. (5700) (0.013)

This implies that manufacturing accounts for $194,000 out of every $1 million
increase in GDP in the cross-section. The residuals from the regression and GDP are
both ranked in Table 1.2 and Di and are computed.

NOUN 105
INTRODUCTION TO ECONOMETRICS II ECO 306

Table 1.2

The sum of the latter came to 1608. The rank correlation coefficient is thus

and the test statistic is √ . This is above 2.58 and hence the null
hypothesis of
homoscedasticity is rejected at the 1 percent level.

3.1.3.5 The Goldfeld–Quandt Test

Goldfeld and Quandt (1965) are so far attributed with the most common formal test
for heteroscedasticity. The test assumes that  u the standard deviation of the
i

probability distribution of the disturbance term in observation i, is about the size of Xi.
It also assumes that the disturbance term is distributed and satisfies the other Gauss–
Markov conditions. The sizeof X ordersthe nobservationsinthesampleand separate
regressions are carried out for the first n'and the last n' observations, the middle (n–
2n') observations being dropped completely. If heteroscedasticity is present, and if the
assumption regarding its nature is correct, the variance of uin the last n' observations
will be more than that in the first n' and this will be reflected in the RSS in the two

NOUN 106
INTRODUCTION TO ECONOMETRICS II ECO 306

sub-regressions. Representing these by RSS1 and RSS2 for the sub-regressions with
the first n' and the last n' observations, respectively.The ratio RSS2/RSS1 will be
distributed as an Fstatistic with (n' – k) and (n' – k) degrees of freedom, where kis the
number of parameters in the equation, under the

null hypothesis of homoscedasticity. The power of the test depends on the choice of
n'about n. As a result of some experiments undertaken by Goldfeld and Quandt, they
recommend that in general, n' should be about 11 when nis 30 and about 22 when nis
3
60. Which clearly shows that n' should be about of n.
8
If there is more than one explanatory variable in the model, the observations should be
ordered by that which is hypothesized to be associated with the null hypothesis for the
test is that RSS2 is not significantly greater than RSS1, and the alternative hypothesis
is that it is significantly greater. If RSS2 turns out to be smaller than RSS1, the null
hypothesis should not be rejected; it only means that there would not be any point in
computing the test statisticRSS2/RSS1. However, the Goldfeld–Quandt test can also be
used for the case where the standard deviation of the disturbance term is hypothesized
to be inversely proportional to Xi. The procedure is the same as before, but the test
statistic is now RSS1/RSS2, and it will again be distributed as anF-statistic with (n' –
k) and (n' – k) degrees of freedom under the null hypothesis of homoscedasticity.

3.1.3.6 The Glejser Test

This test permits you to search the nature of the heteroscedasticity a little more
closely. Here, the assumption that  u is a relative quantity to Xi is relaxed,and you can
i

then investigate whether some other efficient form may be more suitable, for example

…[3.03]
 ui

To use the procedure, you regress Yon Xusing OLS and then fit the absolute values of
the
residuals, | | to the function for a given value of . You may fit several such
functions, varying the choice of . In each case the null hypothesis of
homoscedasticity will be rejected if the estimate of is significantly different from 0.

NOUN 107
INTRODUCTION TO ECONOMETRICS II ECO 306

If more than one function gives rise to a significant estimate of , that with the best
fit may be a guide to the nature of the heteroscedasticity.

3.1.3.6 Solution to Heteroscedasticity Problem

Suppose that the true relationship is

…[3.04]

Let the standard deviation of the disturbance term in observation ibe  u . If you i

happened to now  u for each observation, you could eliminate the heteroscedasticity
i

by dividing each observation by its value of . The model becomes

ui
…[3.05]
 ui  ui  ui  ui

ui
The disturbance term becomes homoscedastic because the population variance of
 ui

ui
is
i

ui
{( ) } ( )  u2 …[3.06]
 ui  2
ui  2
ui
i

That is, every observation will have a disturbance term drawn from a distribution with
population variance 1, and the model will be homoscedastic. The revised model may
be rewritten as;

Yi '  1hi   2 X i'  ui' …[3.07]

Yi Xi 1
where Yi '  , X i'  , his a new variable whose value in observation iis and
u i
u i
u i

ui
ui' 
u i

NOUN 108
INTRODUCTION TO ECONOMETRICS II ECO 306

Note that there should not be a constant term in the equation. By regressing Y' on hand
X',
you will obtain efficient estimates of and with unbiased standard errors.

3.1.3.7 Consequences of Heteroscedasticity

The seriousness of the consequences of heteroscedasticity will depend on the nature of

the occurred heteroscedasticity, and there are no general rules. In the case of the
heteroscedasticity, where the standard deviation of the disturbance term is
proportional to Xand the values of Xare integers from 5 to 44. Here, the population
variance of the OLS estimator of the slope coefficient is approximately double that of
the estimator, where the heteroscedasticity has been eliminated by dividing through

byX. Further, the standard errors of the OLS estimators are underestimated, giving a
misleading impression of the precision of the OLS coefficients.

3.1.4.0 SUMMARY

This unit begins with a general discussion ofheteroscedasticity and its meaning. Also
discussed are the reasons why the distribution of a disturbance term may be subject to
heteroscedasticity, and the consequences of the heteroscedasticityproblem for OLS
estimators. We continued with presentation of several tests for heteroscedasticity and
methods of alleviating the problem.

3.1.5.0 CONCLUSION

The discussion in this unit concludes with an awareness to the students that the
concept of heteroscedasticity and associated problems are indications of how an
apparent case of heteroscedasticity may be caused by model misspecification.

3.1.6.0 TUTOR-MARKED ASSIGNMENT

A researcher investigating whether government expenditure tends to crowd out

investment fits the regression (standard errors in parentheses):

̂
S.E.E. (7.79) (0.14) (0.02)

NOUN 109
INTRODUCTION TO ECONOMETRICS II ECO 306

She sorts the observations by increasing size of Y and runs the regression again for the
11
countries with smallest Y and the 11 countries with largest Y. RSS for these regressions
is 321
and 28101, respectively. Perform a Goldfeld–Quandt test for heteroscedasticity.

3.1.7.0 REFERENCES /FURTHER READING

Maddala, G. S., &Lahiri, K. (1992).Introduction to econometrics (Vol. 2). New York

Carter, H. R., Griffiths, W. E., & Judge, G. (2001).Undergraduate econometrics.2nd

Ed. New York: John Wiley and Sons.

Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.

MODULE 4: ECONOMETRIC MODELLING AND AUTOCORRELATION

The general aim of this module is to provide you with a thorough understanding of the
basic rudiments of econometric modelling. Stochastic Regression and Measurement
Errors, autocorrelation, econometric modelling and models using time series data are
explained. By the end of this module, you would have been able to understand the
components of the module stated below. The units to be studied are;

Unit 1: Stochastic Regression and Measurement Errors

Unit 2: Autocorrelation

Unit 3: Econometric Modelling and Models Using Time Series Data

UNIT 1: STOCHASTIC REGRESSORS ANDMEASUREMENT ERRORS

CONTENTS
4.1.1.0 Introduction
4.1.2.0 Objectives
4.1.3.0 Main Content
4.1.3.1 Stochastic Regressors
4.1.3.2 Unbiasedness

NOUN 110
INTRODUCTION TO ECONOMETRICS II ECO 306

4.1.3.3 Consistency
4.1.3.4 The Consequences of Measurement Errors
4.1.3.5 Measurement Errors in the Explanatory Variable(s)
4.1.3.6 Measurement Errors in the Dependent Variable
4.1.4.0 Summary
4.1. 5.0 Conclusion
4.1.6.0 Tutor-Marked Assignment
4.1.7.0 References/Further Reading

4.1.1.0 INTRODUCTION

The least squares regression model assumed that the explanatory variables
arenonstochastic, that is, that they do not have random components. Although relaxing
this assumption does not in itself undermine the OLS regression technique, it is
typically an unrealistic assumption, so it is important

you know the consequences of relaxing it. We shall see that in some contexts we can
continue to use OLS, but in others, for example when one or more explanatory
variables are subject to measurement error, it is a biased and inconsistent estimator.

4.1.2.0 OBJECTIVE

The main objective of this unit is to provide a general understanding of the topic
„stochastic regressors and measurement errors and point out that random element in a
regression model is not the only disturbance term but that the variables themselves do
have random components.

4.1.3.0 MAIN CONTENTS

4.1.3.1 Stochastic Regressors

Based on the adopted assumption that the regressors, which is the explanatory
variables in the regression model are nonstochastic, their values in the sample are
therefore fixed and unaffected by the way the sample is generated. Perhaps the best
example of a nonstochastic variable is time, which, as we will see when we come to
time series analysis, is sometimes included in the regression model as a proxy for

NOUN 111
INTRODUCTION TO ECONOMETRICS II ECO 306

variables that are difficult to measure, such as technical progress or changes in tastes.
Nonstochastic explanatory variables are unusual in regression analysis.

A rationale for making the nonstochastic assumption has been one of simplifying the
analysis of the properties of the regression estimators. For example, we saw that in the
regression model

…[4.01]

the OLS estimator of the slope coefficient may be decomposed as follows:

( ) ( )
…[4.02]
( ) ( )

Here, if X is nonstochastic, so is ( )
and the expected value of the error term can be written , ( )- ( ).
Also if X is nonstochastic, , ( )- is 0.
Which easily helps us to prove that b2 is an unbiased estimator of .

The desirable properties of the OLS estimators remain unchanged even if the
descriptive variables have stochastic components, provided that these components are
distributed independently of the disturbance term, and provided that their distributions
do not depend on the

parameters u Let us demonstrate the unbiasedness and consistency

properties and as typical, taking an efficient approach.

4.1.3.2 Unbiasedness

Once X is stochastic, ( ) cannot be treated as a scalar, so we cannot

rewrite , ( ) ( )- as , ( )- ( ) Hence the previous proof of
unbiasedness is blocked. However, we can find another route by decomposing the
error term:

( ) ∑ ( ̅ )( ̅) ( ̅)
∑ . /( ̅) ∑ ( )( ̅)
( ) ( ) ( )

…[4.03]

NOUN 112
INTRODUCTION TO ECONOMETRICS II ECO 306

(Xi  X )
where f (X i )  . Now, if X and uare independently distributed,
Var ( X )

E[ f ( X i )(ui  u)]

may be decomposed as the product of E[ f (x i )] and E[(ui  u)] . Hence

E[ f (Xi )(ui  u)] = E[ f (Xi )E(ui  u)]  E[ f (Xi )] 0 …[4.04]

since by assumption E (ui ) is 0 in each observation. This implies,of course, that E (u ) is

also 0.

Hence, when we take the expectation of ∑ ( )( ̅), each term within the

summation has expected value 0. Thus the error term as a whole has expected value 0
and b2 is an unbiased estimator of .

4.1.3.3 Consistency

Generally stated, ( ) is equal to ( ) ( ) where A and B are any

two stochastic quantities, on condition that both ( ) and ( ) exist and that
( ) is nonzero ( is the limiting value as the sample size becomes large).
As also stated, sample expressions tend to their population counterparts as the sample
size becomes large, so ( ) is the population covariance of X and u and
( ) is , the population variance of X. If X and u are independent, the
population covariance of X and u is 0 and we can write that:

( )
…[4.05]
( )

4.1.3.4 The Consequences of Measurement Errors

As it is in other human activities, it habitually happens in economics that, when

investigating a relationship, the variables involved could be measured defectively. For
example, surveys often contain errors caused by the person being interviewed not
remembering properly or not understanding the question correctly. However,
misreporting is not the only source of inaccuracy. It sometimes happens that you have
defined a variable in your model in a certain way, but the available data correspond to
a slightly different definition.

NOUN 113
INTRODUCTION TO ECONOMETRICS II ECO 306

4.1.3.5 Measurement Errors in the Descriptive Variable(s)

To keep the analysis simple, we will confine it to the simple regression model. Let us
suppose that a variable Y depends on a variable Z according to the relationship

…[4.06]

wherev is a disturbance term with mean 0 and variance , distributed independently

of Z. We shall suppose that Z cannot be measured absolutely accurately, and we shall
use X to denote its measured value. In observation is equal to the true value, ,
plus the measurement error, :

…[4.07]

We shall suppose that w has mean 0 and variance , that Z has population
variance , and that w is distributed independently of Z and v.

[4.07] into [4.06], will yield

( ) …[4.08]

Two random components are present in [4.08], the original disturbance term v and the
measurement error (multiplied by ). Together they form a composite disturbance
term, which we shall call u:

…[4.09]

Therefore, [4.08] becomes

…[4.10]

You have your data on Y (which, for the time being, we shall assume has been
measured accurately) and X, and you unsuspectingly regress Y on X.

As usual, the regression coefficient b is given by

( ) ( )
…[4.11]
( ) ( )

NOUN 114
INTRODUCTION TO ECONOMETRICS II ECO 306

Looking at the error term, we can see that it is going to behave badly. By [4.07] and
[4.09], both Xiand depend on . The population covariance between X and u is
nonzero and, so b2 is an inconsistent estimator of . Even if you had a very large
sample, your estimate would be inaccurate. In the limit it would underestimate by
an amount

…[4.12]

4.1.3.6 Measurement Errors in the Dependent Variable

These measurement errors in the dependent variable do not matter as much. In

practice, they can be thought of as contributing to the disturbance term. They are
undesirable, because anything that increases the noise in the model will tend to make
the regression estimates less accurate, but they will not cause the regression estimates
to be biased.

By assumption, let the true value of the dependent variable be Q, and the true
relationship be

…[4.13]

wherev is a disturbance term. If is the measured value of the dependent variable in

observation i, and is the measurement error,

…[4.14]

which may be rewritten

…[4.15]

whereu is the composite disturbance term (v + r)

The only difference from the usual model is that the disturbance term in [4.15] has
two
components: the original disturbance term and the error in measuring Y. The important
thing is that the explanatory variable X has not been affected. Hence OLS still yields
unbiased estimates provided that X is nonstochastic or that it is distributed

NOUN 115
INTRODUCTION TO ECONOMETRICS II ECO 306

independently of v and r. The population variance of the slope coefficient will be

given by

…[4.16]

and so will be greater than it would have been in the absence of measurement error,
reducing the precision of the estimator. The standard errors remain valid but will be
larger than they would have been in the absence of the measurement error, reflecting
the loss of precision.

4.1.4.0 SUMMARY

In this unit, in other for the students to have understanding of the topic stochastic
regressors and measurement errors, we explained conditions under which OLS
estimator remain unbiased when the variable in a regression model possessing random
components. A demonstration of the unbiasedness and consistency properties was also
approached. Equally, the consequences of measurement errors, errors in descriptive
and dependents variables were discussed.

4.1.5.0 CONCLUSION

The unit concludes that under general conditions the regression model remain
unchanged even if the descriptive variables have stochastic components. Provided that
these components are distributed independently of the disturbance term and
considering measurement errors in the descriptive and dependent variables.

4.1.6.0 TUTOR-MARKED ASSIGNMENT

In a certain industry, firms relate their stocks of finished goods, Y, to their expected
annual
sales, , according to a linear relationship

Actual sales,X, differ from expected sales by a random quantity u that is distributed
with mean 0 and constant variance:

NOUN 116
INTRODUCTION TO ECONOMETRICS II ECO 306

uis distributed independently of An investigator has data on Y and X (but not on

) for a cross-section of firms in the industry. Describe the problems that would be
encountered if OLS were used to estimate and , regressing Y on X.

4.1.7.0 REFERENCES /FURTHER READING

Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.

Dominick, S., & Derrick, R. (2002).Theory and problems of statistics and

econometrics.Schaum‟s Outline Series.

N Gujaratti, D. (2004). Basic econometrics. McGraw-Hill, New York.

Stock, J. H., & Watson, M. W. (2015).Introduction to econometrics. Pearson.

NOUN 117
INTRODUCTION TO ECONOMETRICS II ECO 306

UNIT 2: AUTOCORRELATION

CONTENTS
4.2.1.0 Introduction
4.2.2.0 Objectives
4.2.3.0 Main Content
4.2.3.1 Possible Causes of Autocorrelation
4.2.3.2 Detection of First-Order Autocorrelation: the Durbin–Watson Test
4.2.4.0 Summary
4.2.5.0 Conclusion
4.2.6.0 Tutor-Marked Assignment
4.2.7.0 References/Further Reading

4.2.1.0 INTRODUCTION

Autocorrelation is the correlation between the error terms arising in time series data.
Such correlation in the error terms often arises from the correlation of the omitted
variables that the error term captures. Furthermore, the assumption in the third Gauss–
Markov condition is that the value taken by the disturbance term in any observation
and determined independently of its values in all the other observations, is satisfied,
and hence that the population covariance of and is 0 for i ≠ j. When the condition
is not satisfied, the disturbance term is said to be subject to autocorrelation, often
called serial correlation or cross-autocorrelation.

4.2.2.0 OBJECTIVE

The main objective of this unit is to provide a basic understanding that autocorrelation
may arise as a consequence of the exclusion of a significant variable or the
mathematical misspecification of regression model.

NOUN 118
INTRODUCTION TO ECONOMETRICS II ECO 306

4.2.3.0 MAIN CONTENTS

The significances of autocorrelation for OLS are to some extent comparable to those
of heteroscedasticity. The regression coefficients remain unbiased, but OLS is
inefficient because one can find an alternative unbiased estimator with smaller
variance. The other main concern, which should not be mixed up with the first, is that
the standard errors are estimated wrongly, probably being biased downwards. Finally,
although in general autocorrelation does not cause OLS estimates to be biased, there is
an important special case where it does.

4.2.3.1 Possible Causes of Autocorrelation

There is two forms autocorrelation occurrence, which could either be positive and
negative. Persistent effects of excluded variables are probably the most frequent cause
of positive autocorrelation, the usual type of economic analysis. In Figure 4.1, Y
depends on X and some minor variables not included explicitly in the specification.
The disturbance term in the model is generated by the combined effects of these
excluded variables. In the first observation, the excluded variables have a net positive
effect and the disturbance term is positive. If the excluded variables change slowly,
their positive effect will persist, and the disturbance term will remain positive. In time
the balance will change, and the net effect of the excluded variables becomes
negative. Here, the persistence effect works the other way, and the disturbance term
remains negative for a few observations. The duration and amplitude of each positive
and negative sequence are essentially random, but overall there will be a tendency for
positive values of the disturbance term to be followed by positive ones and for
negative values to be followed by negative ones. However, a factor to note is that
autocorrelation is on the whole more likely to be a problem for shorter intervals
between observations.

NOUN 119
INTRODUCTION TO ECONOMETRICS II ECO 306

Figure 4.1 Positive Autocorrelation

Negative autocorrelation means that the correlation between successive values of the
disturbance term is negative. A positive value in one observation is more likely to be
followed by a negative value than a positive value in the next, and vice versa; this is
shown by an illustrative scatter diagram in Figure 4.2. A line joining successive
observations to one another would cross the line relating Y to X with greater frequency
than one would expect if the values of the disturbance term were independent of each
other. Economic examples of negative autocorrelation are relatively uncommon, but
sometimes it is induced by manipulations used to transform the original specification
of a model into a form suitable for regression analysis.

NOUN 120
INTRODUCTION TO ECONOMETRICS II ECO 306

Figure 4.2 Negative Autocorrelation

When an error term at time period t is correlated with error terms in time series, the
correlation between and is called an autocorrelation of order k. The
correlation between and is the first-order autocorrelation and is usually
denoted by The correlation between and is called the second order
autocorrelation and is denoted by , and so on. There are (n - 1) such autocorrelations
if we have n observations. However, we cannot hope to estimate all of these from our
data. Hence we often assume that these (n - 1) autocorrelations can be represented in
terms of one or two parameters.

4.2.3.2 Detection of First-Order Autocorrelation: the Durbin–Watson Test

NOUN 121
INTRODUCTION TO ECONOMETRICS II ECO 306

We will mostly be concerned with first-order autoregressive autocorrelation, often

denoted AR (1). AR (1) appears to be the most common type of autocorrelation
approximation. It is described as positive or negative according to the sign of ρ. Note
that if ρ is 0, there is no autocorrelation occurrence.

There are two major things that will be discussed in this unit, which are:

1. Test for the presence of serial correlation.

2.Estimate the regression equation when the errors are serially correlated.

Durbin-Watson Test (DW)

The simplest and most commonly used model is one where the errors and
have a correlation . For this model one can think of testing hypotheses about on the
basis of , the

correlation between the least squares residuals and . A commonly used

statistic for this purpose which is related to is the DW statistic, which will be denote
by . It is defined as

∑ ( )
∑
…[4.17]

Where is the estimated residual for period . DW can be re-written as

∑ ∑ ∑
∑ ∑ ∑
…[4.18]

Since ∑ and ∑ are approximately equal if the sample is large, we have

( ) If

The sampling distribution of depends on the values of the explanatory variables and
hence DW derived upper ( ) limits and lower ( ) limits for the significance levels
for . There are tables to test the hypothesis of zero autocorrelation against the
hypothesis of first-order positive autocorrelation. (For negative autocorrelation we
interchange ( ) ( )), hence;

NOUN 122
INTRODUCTION TO ECONOMETRICS II ECO 306

If ( ) , we reject the null hypothesis of no autocorrelation.

If ( ) we do not reject the null hypothesis.
If ( ) ( )the test is inconclusive.

The upper bound of the DW statistic is a good approximation to its distribution when
the regressors are slowly changing. DW argue that economic time series are slowly
changing, and hence one can use ( )as the correct significance point.

The significance points in the DW tables are tabulated for testing = 0 against > 0.
If d > 2 and we wish to test the hypothesis = 0 against <0, we consider 4…d and
refer to the DW tables as if we are testing for positive autocorrelation. Although we

have said that → ( ) this approximation is valid only in large samples. The
mean of when has been shown to be given approximately by

( )
( ) …[4.19]

wherek is the number of regression parameters estimated (including the constant

term), and n is the sample size. Thus, even for zero serial correlation, the statistic is
biased upward from 2. If k = 5 and n= 15, the bias is as large as 0.8.

4.2.5.0 SUMMARY

The unit explained the concept of autocorrelation at first order, its possible causes and
detection (with particular interest on the first-order autoregressive autocorrelation,
denoted by AR (1)) using Durbin-Watson test.

4.2.4.0 CONCLUSION

In this unit, autocorrelation is statistically explained as a random process that

measures the linear correlation between values of the process at different times, as a
function of time or of the time lag. The significances of autocorrelation for OLS are
shown to be comparable to those of heteroscedasticity and have two forms of
occurrences, which could either be positive or negative. Students are advised to use
the further reading materials to look at more autocorrelation techniques and study
more on the correlation between the error terms arising in time series data as indicated
in the introduction section of this unit.

NOUN 123
INTRODUCTION TO ECONOMETRICS II ECO 306

4.2.6.0 REFERENCES /FURTHER READING

Maddala, G. S., &Lahiri, K. (1992).Introduction to econometrics (Vol. 2). New York.

Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.
Stock, J. H., & Watson, M. W. (2015).Introduction to econometrics. Pearson.
Smith, G. (2013). Econometric Principles and Data Analysis.Centre for Financial and
Management Studies SOAS, University of London, London.
Dougherty, C. (2014). Elements of econometrics.London: University of London.

NOUN 124
INTRODUCTION TO ECONOMETRICS II ECO 306

UNIT 3: ECONOMETRIC MODELLING AND MODEL USING TIME-

SERIES DATA

CONTENTS
4.3.1.0 Introduction
4.3.2.0 Objectives
4.3.3.0 Main Content
4.3.3.1 The Adaptive Expectations Model
4.3.4.0 Summary
4.3.5.0 Conclusion
4.3.6.0 Tutor-Marked Assignment
4.3.7.0 References/Further Reading

4.3.1.0 INTRODUCTION

The modelling of expectations using time series data is often an important and
difficult task of the applied economist. This is especially true in macroeconomics, in
that investment, saving, and the demand for assets are all sensitive to expectations
about the future. Unfortunately, there is no satisfactory way of measuring expectations
directly for macroeconomic purposes. Consequently, macroeconomic models tend not
to give particularly accurate forecasts, and this makes economic management difficult.

4.3.2.0 OBJECTIVE

The main objective of this unit is to introduce the application of regression analysis to
time series data, starting with static models and then continuing to dynamic models
with lagged variables used as descriptive variables.

4.3.3.0 MAIN CONTENTS

4.3.3.1 The Adaptive Expectations Model

As a makeshift solution, some models use an indirect technique known as the adaptive
expectations process. This involves a simple learning process in which, in each period,
the actual value of the variable is compared with the value that had been expected. If

NOUN 125
INTRODUCTION TO ECONOMETRICS II ECO 306

the actual value is greater, The expected value is adjusted upwards for the next
period.If it is lower, the expected value is

adjusted downwards. The size of the adjustment is hypothesized to be proportional to

the discrepancy between the actual and expected value.

If X is the variable in question, and is the value expected in time period t given the
information available at time period t–1,

( )( ) …[4.20]

This canbe rewritten

( ) ( ) …[4.21]

Which states that the expected value of X in the next period is a weighted average of
the actual value of X in the current period and the value that had been expected. The
larger the value of  , the quicker the expected value adjusts to previous actual
outcomes.

For example, suppose that you hypothesize that a dependent variable, , is related to
the expected value of the descriptive variable, X, in year t+1, :

…[4.22]

expresses in terms of , which is unobservable and must somehow be replaced by

observable variables, that is, by actual current and lagged values of X, and perhaps
lagged values of Y. We start by substituting for ,

( ( ) ) ( ) )
…[4.23]

Of course, we still have unobservable variable as an descriptive variable, but if it is

true for time period t, it is also true for time period t–1:

( ) …[4.24]

Substituting for , in [4.23] we now have

NOUN 126
INTRODUCTION TO ECONOMETRICS II ECO 306

( ) ( )
( ) ( )
…[4.25]

Now it is reasonable to suppose that lies between 0 and 1, in which case (1 – ) will
also lie between 0 and 1. Thus ( ) becomes progressively smaller as s increases.
Eventually, there will be a point where the term ( ) is so small that it
can be neglected and we have a model in which all the variables are observable.

A lag structure with geometricallydeclining weights, such as this one, is described as

having a Koyck distribution. It is highly sparing regarding its constraint, requiring
only one parameter more than the static version. Since it is nonlinear in the
parameters, OLS should not be used to fit it, for two reasons. First, multicollinearity
would almost certainly make the estimates of the coefficients so erratic that they
would be worthless – it is precisely this problem that caused us to search for another
way of specifying a lag structure. Second, the point estimates of the coefficients
would yield conflicting estimates of the parameters.

4.3.4.0 SUMMARY

In this unit, an indirect technique known as the adaptive expectations process is

explained as a makeshift solution used in some models. This involves simple learning
process for which, in each period, the size of adjustment is proportional to the
discrepancy between the actual and expected value.

4.3.5.0 CONCLUSION

The unit introduces the application of regression analysis to time series data, starting
with static models and then proceeding to dynamic models with lagged variables used
as descriptive variables. In this unit, the adaptive expectations process was mainly
used for explanation but the multicollinearity problem in time series models,
especially dynamic ones with lagged descriptive variables was also explained. The
students may use the reference materials for more understanding and further readings.

4.3.6.0 TUTOR-MARKED ASSIGNMENT

NOUN 127
INTRODUCTION TO ECONOMETRICS II ECO 306

1.) The results of linear and logarithmic regressions of consumer expenditure on food,
FOOD, on DPI and a relative price index series for food, PRELFOOD, using the
Demand Functions data set, are summarized below. Provide an economic
interpretation of the coefficients and perform appropriate statistical tests.

̂
S.E.E.(31.9) (0.002) (0.332)

̂
S.E.E(0.28) (0.01) (0.07)

2.) Sometimes a time trend is included in a regression as an explanatory variable,

acting as a proxy for some gradual change not associated with income or price.
Changing tastes might be an example. However, in the present case, the addition of a
time trend might give rise to a problem of multicollinearity because it will be highly
correlated with the income series and perhaps also the price series. Calculate the
correlations between the TIME variable in the data set, LGDPI, and the logarithm of
expenditure on your category. Regress the logarithm of expenditure on your category
on LGDPI, the logarithm of the relative price series and TIME (not the logarithm of
TIME). Provide an interpretation of the regression coefficients, perform appropriate
statistical tests, and compare the regression results with those of the same regression
without TIME.

4.3.7.0 REFERENCES /FURTHER READING

Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.

Stock, J. H., & Watson, M. W. (2015).Introduction to econometrics. Pearson.

Smith, G. (2013). Econometric Principles and Data Analysis.Centre for Financial and
Management Studies SOAS, University of London, London.

NOUN 128
INTRODUCTION TO ECONOMETRICS II ECO 306

MODULE 5: SIMULTANEOUS EQUATION, BINARY CHOICE, AND

MAXIMUM LIKELIHOOD ESTIMATION

The general aim of this module is to provide you with a thorough understanding of the
basic rudiments of Simultaneous Equation, Binary Choice, and Maximum Likelihood
Estimation. By the end of this module, you should be able to understand the
components of the module stated below. The units to be studied are;

Unit 1: Simultaneous Equation

Unit 2: Binary Choice and Limited Dependent Models with Maximum

Likelihood Estimation

UNIT 1: SIMULTANEOUS EQUATIONSESTIMATION

CONTENTS
5.1.1.0 Introduction
5.1.2.0 Objectives
5.1.3.0 Main Content
5.1.3.1 Simultaneous Equations Models: Structural and Reduced Form Equations

5.1.3.2 Simultaneous Equations Bias

5.1.4.0 Summary
5.1.5.0 Conclusion
5.1.6.0 Tutor-Marked Assignment
5.1.7.0 References/Further Reading

5.1.1.0 INTRODUCTION

In the engagement of OLS to estimate the factors of an equation that is set in a

simultaneous equations model, it is likely that the estimates will be biased and erratic
which would invariably make the statistical tests invalid and inconsistent.

NOUN 129
INTRODUCTION TO ECONOMETRICS II ECO 306

5.1.2.0 OBJECTIVE

The main objective of this unit is to demonstrate to the students that in practice most
economic relationships interact with others in a system of simultaneous equations and
when this is the case the application of OLS to a single relationship in isolation yields
biased estimates.

5.1.3.0 MAIN CONTENTS

5.1.3.1 Simultaneous Equations Models: Structural and Reduced Form

Equations
As explained earlier in other modules, measurement error is not the only probable
cause why the fourth Gauss–Markov condition may not be satisfied. Simultaneous
equations bias is another. To illustrate this; suppose there is an investigation on the
determinants of price inflation and wage inflation. For ease, it would be better to start
with a very simple model that supposes that p, the annual rate of growth of prices, is
related to w, the annual rate of growth of wages, it being assumed that increases in
wage costs force prices upwards:

That is;

…[5.01]

Here, w is related to pand U, the rate of unemployment, workers protecting their real
wages by demanding increases in wages as prices rise, but their ability to do so being
the weaker, the higher the rate of unemployment ( ). Which is stated as:

…[5.02]

where, are disturbance terms

Clearly, this simultaneous equations model involves a certain amount

ofcomplexity:wdetermines p in the first equation [5.01], and in turn,p helps to
determine w in the second [5.02]. For better clarity in resolving this complexity, we
need to make a distinction between endogenousand exogenous variables. Endogenous
variables are variables whose values are determined by the interaction of the
relationships in the model. Exogenous ones are those whose values are determined

NOUN 130
INTRODUCTION TO ECONOMETRICS II ECO 306

externally. Thus in the present case,p and ware both endogenous, and U is exogenous.
The exogenous variables and the disturbance terms ultimately determine the values of
the endogenous variables, once the complexity is cleared. The mathematical
relationships expressing the endogenous variables regarding the exogenous variables
and disturbance terms are known as the reduced form equations. The original
equations that we wrote down when specifying the model are

described as the structural equations. We will derive the reduced form equations for p
and w. To obtain that for p, we take the structural equation for p and substitute for w
from the second equation:

( ) …[5.03]

Hence,

( ) …[5.04]

and so we have the reduced form equation for p;

…[5.05]
( )

Similarly we obtain the reduced form equation for w:

( )
…[5.06]

Hence

( ) …[5.07]

and so

…[5.08]

5.1.3.2 Simultaneous Equations Bias

In almost all simultaneous equations models, the reduced form equations express the
endogenous variables regarding all of the exogenous variables and all of the

NOUN 131
INTRODUCTION TO ECONOMETRICS II ECO 306

disturbance terms. You can see that this is the case with the price inflation/wage
inflation model. In this model, there is only one exogenous variable, U.

wdepends on it directly; p does not depend on it directly but does so indirectly because
w determines it. Similarly, both p and wdepend on , p directly and w indirectly. And
both depend on , w directly and p indirectly.

The dependence of w on means that OLS would yield inconsistent estimates if used
to fit
equation [5.01], the structural equation for p. w is a stochastic regressor and its
random component is not distributed independently of the disturbance term .
Similarly the dependence of p on

means that OLS would yield inconsistent estimates if used to fit [5.02]. Since [5.01] is
a simple regression equation, it is easy to analyze the large-sample bias in the OLS
estimator of .

5.1.5.0 SUMMARY

In this unit, we started by explaining structural and reduced form of equations which
was illustrated by showing that measurement error is not the only probable cause why
the fourth Gauss–Markov condition may not be satisfied for which the biasness of
simultaneous equations is an example. We then went further to explain the
simultaneous equations bias.

5.1.4.0 CONCLUSION

In this unit, Simultaneous equations estimation is discussed. The structural and

reduced form of equations as it relates to Simultaneous equations model bias is also
explained. Clearly, for deeper understanding of the equation models, the students
should make a distinction between endogenousand exogenous variables in
Simultaneous equations estimation.

5.1.6.0 TUTOR-MARKED ASSIGNMENT

NOUN 132
INTRODUCTION TO ECONOMETRICS II ECO 306

1.) Simple macroeconomic model consists of a consumption function and an income

identity:

whereC is aggregate consumption, I isaggregate investment, Y is aggregate income,

and u is a disturbance term. On the assumption that I is exogenous, derive the reduced
form equations for C and Y.

2.) From the model above, demonstrate that OLS would yield inconsistent results if
used to fit the consumption function, and investigate the direction of the bias in the
slope coefficient.

5.1.7.0 REFERENCES /FURTHER READING

Maddala, G. S., &Lahiri, K. (1992).Introduction to econometrics (Vol. 2). New York.

Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.

NOUN 133
INTRODUCTION TO ECONOMETRICS II ECO 306

Unit 2: Binary Choice and LimitedDependent Models andMaximum Likelihood

Estimation

CONTENTS
5.2.1.0 Introduction
5.2.2.0 Objectives
5.2.3.0 Main Content
5.2.3.1 The Linear Probability Model
5.2.3.2 Goodness of Fit and Statistical Tests
5.2.4.0 Summary
5.2.5.0 Conclusion
5.2.6.0 Tutor-Marked Assignment
5.2.7.0 References/Further Reading

5.2.1.0 INTRODUCTION

Most times economists are known to be interested in the factors behind the decision-
making of individuals or enterprises. Examples are:

- Why do some people go to college while others do not?

- Why do some women enter the labour force while others do not?
- Why do some people buy houses while others rent?
- Why do some people migrate while others stay put?

Models have been developed to proffer solutions to these interest, and they are known
as abinary choice or qualitative response models. The outcome will be denoted by Y,
and assigned a value of 1 if the event occurs and 0 otherwise. Models with more than
two possible outcomes have also been developed, but let us restrict our scope to
abinary choice. The linear probability model apart, binary choice models are fitted
using maximum likelihood estimation.

5.2.2.0 OBJECTIVE

The main objective of this unit is to provide the students with a clear understanding
that apart from the linear probability model, binary choice models are fitted using
maximum likelihood estimation.

NOUN 134
INTRODUCTION TO ECONOMETRICS II ECO 306

5.2.3.0 MAIN CONTENTS

5.2.3.1 The Linear Probability Model

The simplest binary choice model is the linear probability model where, as the name
implies, the probability of the event occurring, p, is assumed to be a linear function of
a set of descriptive variable(s). That is:

( ) …[5.09]

For one descriptive variable, the relationship is as shown in Figure 5.1. Of course,p is
unobservable, and as expected there is only one data Y, on the outcome. In the linear
probability model, this is used as a dummy variable for the dependent variable.

Figure 5.1. Linear Probability Model

Regrettably, the linear probability model though simple still has some serious defects.
First, there are problems with the disturbance term. As usual, the value of the
dependent variable in observation i,has a nonstochastic component and a random
component. The nonstochastic component depends on and the parameters and is the
expected value of given ( | ). The random component is the disturbance
term.

( | ) …[5.10]

NOUN 135
INTRODUCTION TO ECONOMETRICS II ECO 306

It is simple to compute the nonstochastic component in observation i because Y can

take only two values. It is 1 with probability and 0 with probability (1 – ):

( ) ( ) …[5.11]

The expected value in observation i is therefore This means that we can

rewrite the model as;

…[5.12]

Probability function is thus also the nonstochastic component of the relationship

between Y and X. It follows that, for the outcome variable to be equal to 1, as
represented by the point A in Figure 5.2, the disturbance term must be equal to
( ) For the outcome to be 0, as represented by the point B, the
disturbance term must be ( ) Thus the distribution of the disturbance term
consists of just two specific values.

Figure 5.2. Linear Probability Model

Which means that the standard errors and the usual test statistics are invalidated. For
good measure, the two possible values of the disturbance term change with X, so the
distribution is heteroscedastic as well. It can be shown that the population variance of
is ( )( ) and this varies with .

NOUN 136
INTRODUCTION TO ECONOMETRICS II ECO 306

The other problem is that the predicted probability may be greater than 1 or less than 0
for extreme values of X. The first problem is dealt with by fitting the model with a
technique known as maximum likelihood estimation.

The second problem involves elaborating the model as follows. Define a variable Z
that is a linear function of the descriptive variables. In the present case, since we have
only one descriptive variable, this function is;

…[5.13]

5.2.3.2 Goodness of Fit and Statistical Tests

Even though numerous measures have been proposed for comparing alternative model
specifications, there is still no measure of goodness of fit equivalent to R2 in
maximum likelihood estimation. Denoting the actual outcome in observation
if the event occurs and 0 if it does not, and denoting the predicted
probability of the event occurring ̂ the measures include the following:

i. the number of outcomes correctly predicted, taking the prediction in

observation ̂ is greater than 0.5 and 0 if it is less;

ii. the sum of the squared residuals ∑ ( ̂) t

iii. the correlation between the outcomes and predicted probabilities, ̂

iv. the pseudo- in the logit output,

Every of these measures has its shortcomings, and it is recommended to consider more
than one and compare their results. Nevertheless, the standard significance tests are
similar to those for the standard regression model. The significance of an individual
coefficient can be evaluated via its t statistic. However, since the standard error is
valid only asymptotically (in large samples), the same goes for the t statistic, and since
the t distribution converges to the normal distribution in large samples, the critical
values of the latter should be used. The counterpart of the F test of the explanatory
power of the model (H0: all the slope coefficients are 0, H1: at least one is nonzero) is
a chi-squared test with the chi-squared statistic in the logit output distributed under H0
with degrees of freedom equal to the number of explanatory variables.

NOUN 137
INTRODUCTION TO ECONOMETRICS II ECO 306

5.2.4.0 SUMMARY

In this unit, we started with the linear probability model being the simplest binary
choice model where the probability of the event occurring is assumed to be a linear
function of a set of descriptive variables. We then proceeded to goodness of fit and
statistical tests using maximum likelihood estimation as a method of estimating the
parameters of a model given observations, by finding the parameter values that
maximise the likelihood of making the observations given the parameters.

5.2.5.0 CONCLUSION

Although numerous measures have been proposed for comparing alternative model
specifications, there is still no measure of goodness of fit equivalent to maximum
likelihood estimation. The students should be of the opinion that every of the
estimation measure has its shortcomings and it is recommended to consider more than
one and compare their results.

5.2.6.0TUTOR-MARKED ASSIGNMENT

A researcher, using a sample of 2,868 individuals from the NLSY (National

Longitudinal Survey of Young Men), is investigating how the probability of a
respondent obtaining a bachelor‟s degree from a four-year college is related to the
respondent‟s score on ASVABC. 26.7 percent of the respondents earned bachelor‟s
degrees. ASVABC ranged from 22 to 65, with mean value 50.2, and most scores were
in the range 40 to 60. Defining a variable BACH to be equal to 1 if the respondent has
a bachelor‟s degree (or higher degree) and 0 otherwise, the researcher fitted the OLS
regression (standard errors in parentheses):

̂
S.E.E. (0.042) (0.001)
The researcher also fitted the following logit regression:

S.E.E. (0.487) (0.009)

whereZ is the variable in the logit function. Using this regression, the researcher
plotted the probability and marginal effect functions shown in the diagram below.

NOUN 138
INTRODUCTION TO ECONOMETRICS II ECO 306

a.) Give an interpretation of the OLS regression and explain why OLS is not a

satisfactory estimation method for this kind of model.

b.) With reference to the diagram below, discuss the variation of the marginal

effect of the ASVABC score implicit in the logit regression and compare it with
that in the OLS regression.

c.) Sketch the probability and marginal effect diagrams for the OLS regression and

compare them with those for the logit regression. (In your discussion, make use
of the information in the first paragraph of this question.)

5.2.7.0 REFERENCES /FURTHER READING

Dominick, S., & Derrick, R. (2002).Theory and problems of statistics and

econometrics.Schaum‟s Outline Series.

Dougherty, C. (2007). Introduction to econometrics. Oxford University Press, USA.

Stock, J. H., & Watson, M. W. (2015).Introduction to econometrics. Pearson.

NOUN 139