Study e Material
Uploaded by
Tanuj SharmaStudy e Material
Uploaded by
Tanuj SharmaBASIC ECONOMETRICS
Study E Material
Dr. M. Chitra
Title: BASIC ECONOMETRICS
Study E Material
Author’s Name: Dr. M. Chitra
Published by: Shanlax Publications,
Vasantha Nagar, Madurai – 625003,
Tamil Nadu, India
Publisher’s Address: 61, 66 T.P.K. Main Road,
Vasantha Nagar, Madurai – 625003,
Tamil Nadu, India
Printer’s Details: Shanlax Press, 66 T.P.K. Main Road,
Vasantha Nagar, Madurai – 625003,
Tamil Nadu, India
Edition Details (I,II,III): I
ISBN: 978-93-95422-76-5
Month & Year: December, 2022
Copyright @ Dr. M. Chitra
Pages: 115
Price: `330
Study E - Material
BASIC ECONOMETRICS
MADURAI KAMARAJ UNIVERSITY
School of Economics
Department of Econometrics
Prepared by
Dr.M.CHITRA1
(For M.A Economics Programme)
1
Dr.M.Chitra is a faculty in Department of Econometrics, studied B.Sc Mathematics, M.Sc Mathematical
Economics, M.A Economics, M.Phil with Applied Econometrics, Ph.D in Economics. National and International
experienced in teaching and lecturing.
Syllabus- M.A Economics
Basic Econometrics
Credits: 4 Teaching Hours: 60
Course Objectives
To introduce the basic terminologies, concepts, scope and methodology of
econometrics
To equip the students with basic theory of econometrics and relevant applications
of the methods
Unit I: Introduction to Basic Econometrics
Econometrics – Meaning, Features, objectives and Scope, – Tools and Types of
Econometrics - Significance of stochastic disturbance term – specification of the
econometric model - Methodology of Econometrics – Limitations of econometrics.
Unit II:Simple linear Regression Model
Simple linear Regression model: Specification of SLRM- OLS Estimation& Assumptions
- Properties of OLS Estimators - Gauss Markov Theorem - Evaluation of SLRM : The
Coefficient of determination - Application of t& f test in testing of hypothesis –
interpreting and reporting the SLRM, Simple numerical problems in SLRM.
Unit III: Multiple Linear Regression Model
Elementary ideas on Multiple Linear Regression Model - Estimation - Testing of
coefficients basic problems interpreting and reporting the MLRM.
Unit IV: Problems of Single Equation Models
Violation of OLS assumptions: Multicollinearity, Heteroscedasticity, Auto correlation:
Meaning, sources, consequences, Detecting and remedial measures – Specification &
measurement errors.
Unit V:Dynamic &Qualitative Regression Models
Instantaneous & Dynamic models- DL,AR, MA concepts – Pure and Mixed Dynamic models
– Estimation of distribute lag models: Adhoc, Koyck and Almon’s Approach, Regression on
qualitative independent variables : ANOVA and ANCOVA models - features &advantages –
regression on qualitative dependent variables: LPM&Logit models.
References
1. Baltagi, B.H. (1998), Econometrics, Springer, New York.
2. Greene, W. (1997), Econometric Analysis, New York:Prentice Hall.
3. Griffith, W.F., R.H. Hill and G.G. Judge (1993), Learning and Practicing Econometrics, New
York:JohnWiley.
4. Gujrati, D. (1995), Basic Econometrics, (3rd Edition), New Delhi:McGraw Hill.
5. Intrilligator, M.D. (1978), Econometric Methods, Techniques and Applications, Prentice Hall,
Englewood Cliffs, New Jersey.
6. Kmenta, J. (1997), Elements of Econometrics, Michigan Press, New York.
7. Koutsoyiannis, A. (1977), Theory of Econometrics (2nd ed.), The Macmillan Press Ltd., London.
8. Maddala, G.S. (1993), Econometrics — An Introduction, New York:McGraw-Hill.
9. Pindyck, R.S. and D.L. Rubinfeld, Econometric Models and Econometric Forecasts, 2nd Ed.,
McGraw-Hill Book Company, New York, 1981.
10. Wooldridge, J.M., 2013, Introductory Econometrics: A Modern Approach, NewDelhi: Cengage.
CONTENTS
Page
S.NO UNITS Headings
No
1 Unit-1 Introduction to Basic Econometrics 1
2 Unit-2 Simple linear Regression Model 15
3 Unit-3 Multiple Linear Regression Model 36
4 Unit-4 Problems of Single Equation Models 52
5 Unit-5 Dynamic &Qualitative Regression Models 65
Additional Unit – Gretel Software
6 Unit-6 81
(Not in Syllabus )
GLOSSARY 90
Sample Question Papers 102
BASIC ECONOMETRICS STUDY E MATERIAL
UNIT I
Structure
1.1 Objectives
1.2 Introduction
1.3 Meaning and Definition
1.4 Nature of Econometrics
1.4.1Objectives of Econometrics
1.4.2 Features of Econometric equations
1.4.3Econometrics is a separate discipline. Why?
1.4.4 Tools of Econometrics
1.4.5 Raw materials of Econometrics
1.4.6 Methodology of Econometrics
1.4.7 Economic Model Vs Econometric Model
1.4.8Types of Econometrics
1.5 Scope of Econometrics
1.6 Goals of Econometrics
1.7 Let us sum up
1.8 Unit End Exercise
1.9 Reference Books
1.1 Objectives
After reading the unit you will be able to:
gain insights into nature of econometrics
understand and be able to articulate, both orally and in writing, the scope of
econometricsinto reality and day today life
1.2 Introduction
Econometric methods are widely used in economic research. Research required
different variety of techniques, which is varied from one subject to another. In recent
decades an increased emphasis has been laid down on the development and use of
statistical techniques for the analysis of the economic problems. Prof. Ragnar Frisch, a
Norwegian economist and statistician first of all named this science as “Econometrics” in
1926. Econometrics emerged as an independent discipline studying economics phenomena.
1
BASIC ECONOMETRICS STUDY E MATERIAL
But it recognized and got attention after the world war. In 1931, the realization of the
necessity of econometric work had become so evident, which made to form “Econometric
Society”. This International association includes practically all the worker in the field. The
society published a periodical called “Econometrica” which disseminates the result of
econometric research work. The electronic gadgets like computers have stimulated the
utilization of econometrics in recent days.
1.3 Meaning and Definition
a) Meaning
Econometrics means economic measurement. Econometrics deals with the
measurement of economic relationships. It‟s an amalgamation of economic theory with
mathematics and statistics.
It is a science which combines economic theory with economic statistics and tries by
mathematical and statistical methods to investigate the empirical support of general
economic law established by economic theory.
The term econometrics is formed from two words of Greek origin, „oukovouia’ meaning
economy and „uetpov’ meaning measure.
b) Definitions
The book „Econometric Theory‟ was authored byArthur S Goldberger, and defined
econometrics in that book as“Econometrics may be defined as the social science in which
the tools of economic theory, mathematics and statistical inference are applied to the
analysis of economic phenomena”.
Gerhard Tinbergen points out that “Econometrics, as a result of certain outlook on the
role of economics, consists of application of mathematical statistics to economic data to
lend empirical support to the models constructed by mathematical economics and to
obtain numerical results”.
H Theil “Econometrics is concerned with the empirical determination of economic
laws”
In the words of Ragnar Frisch “The mutual penetration of quantitative econometric
theory and statistical observation is the essence of econometrics”.
Thus, econometrics may be considered as the integration of economics, mathematics
and statistics for the purpose of providing numerical values for the parameters of
economic relationships and verifying economic theories. It is a special type of economic
analysis and research in which the general economic theory, formulated in mathematical
terms, is combined with empirical measurement of economic phenomena.
1.4 Nature of Econometrics
The nature of econometrics given in a systematic and logical way starts from its
objectives, features of its equations, the uniqueness of econometrics, tools , raw materials,
anatomy of econometrics, the differences of econometric models with economic models,
and types of econometrics in following pages.
2
BASIC ECONOMETRICS STUDY E MATERIAL
1.4.1 Objectives of Econometrics
The general objective of Econometrics is to give empirical content to economic theory.
Empirical study means study based upon data.
It helps to explainthe behavior of a forth coming period that is forecasting economic
phenomena.
It helps to prove the old and established relationships among the variables or
between the variables
It helps to establish new theories and new relationships.
It helps to test the hypotheses and estimation of the parameter.
1.4.2 Features of An Equation
Econometric theory is mainly concerned with quantitative relationships among
economic variables. Quantitative statements are usually expressed in the form of equation
with specified numerical coefficients. Prof. Carl.F.Christ expressed that the equation must
have the following features:
1. An economic equation should be relevant to the phenomenon being studied.
2. Equation should be simple to understand.
3. Equation should be consistent and consider only the relevant part of the theory.
4. Equation relating to a problem should be consistent with available relevant data.
5. The co-efficient of an equation will affect the economic inferences, so it is desirable
to have an accurate knowledge about the co-efficient
6. Equation must have forecasting ability, because econometric study concerned with
future. The above all features can be simplified as follows:
“An equation may have relevance, simplicity, theoretical probability, explanatory
ability, accuracy of co-efficient and forecasting ability”.
1.4.3 Econmetics Is a Separate Discipline. Why?
In the practice of econometrics, economic theory, institutional information and other
assumptions are relied upon to formulate a statistical model, or a set of statistical
hypotheses to explain the phenomena in question.
a. Economic theory makes statements or hypotheses that are mostly qualitative in
nature, where Econometrics gives empirical content to most economic theory.
b. Econometrics differs from mathematical economics. The main concern of the
mathematical economics is to express economic theory in mathematical form
(equations) without regard to measurability or empirical verification of the theory.
As noted above, econometrics is mainly interested in the empirical verification of
economic theory. The econometrician often uses the mathematical equations
proposed by mathematical economist but put these equations in such a form that
they lend themselves to empirical testing.
c. Further, although econometrics presupposes the expression of economic
relationships in mathematical form, like mathematical economics it does not
3
BASIC ECONOMETRICS STUDY E MATERIAL
assume that economic relationships that are exact. On the contrary, econometrics
assumes that economic relationships are not exact but stochastic. Econometric
methods are designed to take into account random disturbances which create
deviations from exact behavioural patterns suggested by economic theory and
mathematical economics. Econometric methods are designed in such a way that
they take into account the random disturbances.
d. Econometrics differs both from mathematical statistics and economic statistics. An
economic statistician gathers empirical data, records them or charts them, and then
attempts to describe the pattern in their development over time and detects some
relationship between various economic magnitudes. Economic statistics is mainly
descriptive aspect of economics. It does not provide explanations of the
development of the various variables and measurement of the parameters of
economic relationships.
e. On the contrary, mathematical statistics deals with methods of measurement which
are developed on the basis of controlled experiments in laboratories. Statistical
methods of measurement are not appropriate for economic relationships, which
cannot be measured on the basis of evidence provided by controlled experiments,
because such experiments cannot be designed for economic phenomena.
Econometrics uses statistical methods for adapting them to the problems of economic
life. These adapted statistical methods are called econometric methods. In particular,
econometric methods are adjusted so that they become appropriate for the measurement of
economic relationships which are stochastic, that is, they include random elements. Hence,
Econometrics is a separate discipline.
1.4.4 Tools of Econometrics
Tools of Econometrics are Mathematics and Statistics. Econometrics transforms
economic theory into mathematical terms and utilizes statistical methods to derive
economic relationships under certain assumptions. Algebra, properties of number system,
Calculus, Statistical Data, statistical methods of sampling and testing the hypothesis are the
tools of Econometrics.
1.4.5 Raw Materials of Econometrics
Data is the prime raw materials of Econometrics collected from two sources as (a)
Primary and (b) Secondary. A primary source gives the first hand information and called
as Primary data. The information which is already collected for some other uses is termed
as Secondary data. The Secondary Data can be again classified as time series data, Cross
section Data and Panel Data.
A time series is a set of observations on the values that a variable takes at different
times. That is, time series data give information about the numerical values of variables
from period to period. Such data may be collected at regular time intervals such as daily,
weekly, monthly, quarterly, annually, quinquennially or decennially. The data thus
collected may be quantitative or qualitative. Thus, data on one or more variables collected
4
BASIC ECONOMETRICS STUDY E MATERIAL
over a period of time is called time series data. That is, values of one or more variables for
several time periods pertaining to a single economic entity are given such data set is called
time series data.
Cross-sectional data are data on one or more variables collected at the same point of
time. These data give information on the variables concerning individual agents at a given
point of time.
Pooled data is a combination of time series and cross sectional data. That is, in the
pooled data are elements of both time series and cross-sectional data.
1.4.6 Methodolgy of Econometrics
Broadly speaking, traditional or classical econometric methodology consists of the
following steps.
1) Statement of the theory or hypothesis
2) Specification of the mathematical model of the theory
3) Specification of the econometric model of the theory
4) Obtaining the data
5) Estimation of the parameters of the econometric model
6) Hypothesis testing
7) Forecasting or prediction
8) Using the model for control or policy purposes.
Flow Chart of Anatomy / Methodology of Econometrics
Statement of the theory or hypothesis
Specification of the mathematical model of the theory
Specification of the econometric model of the theory
Obtaining the data
Estimation of the parameters of the econometric model
Forecasting or prediction
Hypothesis testing
Using the model for control or policy purposes
To illustrate the preceding steps, let us consider the well-known psychological law of
consumption.
1) Statement of theory or hypothesis
Keynes stated “the fundamental psychological law......is that men (women) are
disposed, as a rule and on average, to increase their consumption as their income increases,
but not as much as the increase in their income”. In short, Keynes postulated that the
marginal propensity to consume (MPC), that is, the rate of change in consumption as a
result of change in income, is greater than zero, but less than one. That is 0<MPC<1.
5
BASIC ECONOMETRICS STUDY E MATERIAL
2) Specification on the mathematical model of consumption
Mathematical model is specifying mathematical equations that describe the
relationships between economic variables as proposed by the economic theory. Although
Keynes postulated a positive relationship between consumption and income, he did not
specify the precise form of functional relationship between the two. For simplicity, a
mathematical economist might suggest the following form of the Keynesian consumption
function:
Yi =β1+β2Xi 0< β2<1 (1.1)
Where Yi = consumption expenditure, Xi= income and β1and β2, known as parameters of
the model are intercept and slope coefficients respectively. The slope coefficient β2
measures the MPC.
In the above equation (1.1), the variable appearing on the left side of the equality sign is
called the dependent variable and the variables on the right side are called the independent
or explanatory variables. Thus, in the Keynesian consumption function, consumption
expenditure is the dependent variable and income is the explanatory variable.
(3) Specification of the econometric model of consumption
The purely mathematical model of the consumption function as in equation (1.1) is an
exact or deterministic relationship between consumption and income. But relationships
between economic variables are generally inexact. This is because of the fact that in
addition to income other variables affect consumption expenditure. For example, size of
family, ages of the members in the family, family religion etc are likely to exert some
influence on consumption.
To allow for the in exact relationship between economic variables, the econometrician
would modify the deterministic consumption function as follows
Yi =β1+β2Xi+Ui (1.2)
Where Ui is known as the disturbance or error term, which is a random or stochastic
variable. The disturbance term Uirepresents all those factors that affect consumption but
are not taken into account explicitly. This equation is an example of an econometric model.
More technically, it is an example of linear regression model. The econometric
consumption function hypotheses that the dependent variable Y (consumption) is linearly
related to the explanatory variable X (income) but that the relationship between the two is
not exact; it is subject to individual variation. The econometric model of consumption
function is shown as 1.2
6
BASIC ECONOMETRICS STUDY E MATERIAL
(4) Obtaining Data
To estimate the econometric model given in equation (1.2), that is, to obtain the
numerical values of β1and β2, one need data. Consumption expenditure, Income are
collected from 5 respondents which are as follows
Income and expenditure of household in Madurai
S.No Income Consumption Expenditure
1 3000 2500
2 3500 2750
3 2750 1800
4 6500 5400
5 1780 1470
4 4000 3700
5 2570 3500
Source: Primary Data
(5) Estimation of the econometric model
After the model has been specified and data has been collected, the econometrician
must proceed with its estimation. The task is to estimate the parameters of the
consumption function, that is, β1 and β2. The numerical estimates of the parameters gives
empirical content to the consumption function. Choice of the appropriate econometric
technique for the estimation of the function and critical examination of the assumptions of
the chosen technique is a crucial step.
(6) Hypothesis Testing
A hypothesis is a theoretical proposition that is capable of empirical verification or
disproof. It may be viewed as an explanation of some event or events, and which may be
true or false explanation. Confirmation or refutation of economic theories on the basis of
sample evidence is based on a branch of statistical theory known as statistical inference or
hypothesis testing.The rate of change in consumption as a result of change in income is
greater than zero, but less than one. That is 0<MPC<1 will be the hypothesis.
(7) Forecasting or Prediction
To predict the future values of the dependent or forecast variable Y, on the basis of
known value or expected values of the explanatory, or predictor, variable X.
(8) Use the model for control or policy purposes
Suppose the estimated Keynesian consumption function, and then the government can
use it for control or policy purposes such as to determine the level of income that will
guarantee the target amount of consumption expenditure. In other words, an estimated
model may be used for control or policy purposes. By appropriate fiscal and monetary
policy mix, the government can manipulate the control variable X to produce the desired
level the target variable Y.
The above process is illustrated in the following figure for better understanding:
7
BASIC ECONOMETRICS STUDY E MATERIAL
ESTIMATE
Theories Estimation of parameters
Model Confidence regions
Assumptions Tests of hypotheses
Data Graphical displays
Statistical methods
Diagnosis, validation and criticism
1.4.7 Economic vs Econometric Model
Model is an abstract representation of reality which clears what is relevant to a
particular question at a particular point of time and neglects all other aspects. The
Economic and Econometric models study economic phenomena but two are differs in the
following aspects.
(1) An Economic model is a logical representation of whatever a theoretical
knowledge. A set of definition and assumption that can be used to explain a
particular economic events, while
An Econometric model is an integration of endogenous variables and exogenous
variables able to analyses the particular events and its spill over effects on third
parties of that events.
(2) An Economic model is adapted to yield a definite and precise formulation of the
economic processes at work. While
An econometric model represents a set of hypothesis that permits statistical
inference from the particular data under review.
(3) Economic model needs exact and precise knowledge, while for Econometric model
needs the understanding of what is relevant to the particular observations at hand.
(4) Economic model are based upon abstract economic theory; Economic theorists set
great store by generality, so economic models are insufficient to permit an
empirical application, While Econometric models are appropriate to the particular
situation base on common sense.
(5) An economic model contains only established facts and theories while Econometric
models can be introduced new relation into an economic phenomenon.
(6) Economic models are concerned with the explanation of economic laws While
Econometric models are designed to forecast about economic phenomena and to
serve as aids to policy formation, thee models are called as policy models.
(7) Economic models are prepared after formulation of economic laws While
Construction of Econometric model is the starting point of any econometric
investigation.
1.4.8 Types of Econometrics
Econometrics is divided into two broad categories as (a) Theoretical Econometrics and
(b) Applied Econometrics.
8
BASIC ECONOMETRICS STUDY E MATERIAL
Theoretical Econometrics is concerned with the development of appropriate methods
for measuring economic relationship specified by econometric models. While Applied
Econometrics use the tools of theoretical econometrics to study the field like economics and
business specifically production function, investment function, demand and supply
function etc.
1.5 Scope of Econometrics
Scope and areas of application of econometrics is expanding constantly. It includes
simple as well as sophisticated mathematical and statistical techniques. Econometrics is the
application of specific methods in the general field of economics science. In this sense, it
plays a service role to economic analysis. By establishing new relationships and theories it
serves the policy makers.
Government Aspect:
Suppose government want to devalue its currency to correct the BOP position. For
estimating the consequences of devaluation, the government is concerned with price
elasticity‟s of imports and exports. The price elasticity is to be estimated with the help of
demand function of import and export commodities. Here, the econometric tools will be
applied.
Producer Aspect:
Suppose a producer wants to maximize his profit, the producer will choose the level of
production which gives him maximum surplus . That is minimum cost of production and
maximum output, which will be solved with help of econometric methods.
In capitalistic economy too, the econometric help the producers in making rational
calculations, Demand function, Price elasticity‟s and constraints help a producer to choose
his field of investment.
Econometrics help in establish new relationships and prove old theorems.
Econometrics is the outstanding method for the verification of economic theorem.
Consumer Aspect:
Effect of the taxation on consumers or effects of government expenditures on
consumers standard of living are come under the purview of econometric analysis.
Optimum allocation of resources has been solved with the development of the theory of
programming.
Professor Oscar Lange explained the scope around three groups of questions.
(1) Earlier studies were centered round the main problem of capitalistic economy that
is forecasting of business cycle. This type of study was a thing of past.
(2) Secondly econometric researches were connected with market research. Analysis of
demand function, Production function, Cost function, Supply function, Distribution
of wealth. etc all problems connected with market analysis.
(3) The third group of question related to theory of programming. It includes the
questions relating to the whole of the economy. This field is related with planned
and socialistic economies. These studies have been stimulated with the growth of
communistic countries.
9
BASIC ECONOMETRICS STUDY E MATERIAL
Now – a – days encompass mainly testing hypotheses, estimation of the parameters,
usages of estimates of the parameter, ascertaining the proper functional form of economic
relations, measuring the effects of imperfect data and study of the feedback relationships.
Hence, whatsoever may the part of economy, or types of markets, the econometric tools are
very useful for interpreting them. Whether a producer or consumer, supplier or buyer,
government or public, econometrics will help in rational calculation in economic
phenomena. Econometrics provides equally valuable assistance to normative as well as
positive economics.
1.6 Goals of Econometrics
The three main goalsof econometrics are as follows:
1. Analysis: Econometrics primarily aims at the verification of economic theories. In
this case we say that the purpose of the research is analysis. That is, the economic models
are formulated in an empirically testable form, to decide how well they explain the
observed behavior of the economic units. Several econometric models can be derived from
an economic model. Such models differ due to different choice of functional form,
specification of stochastic structure of the variables etc. So, a strong analysis will be carried
out by econometrics as a prime goal to verify any economic theory and economic
phenomena.
2. Policy Making: The models are estimated on the basis of observed set of data and
are tested for their suitability. This is the part of statistical inference of the modeling.
Various estimation procedures are used to know the numerical values of the unknown
parameters of the model. Based on various formulations of statistical models, a suitable
and appropriate model is selected. The inference or the knowledge obtain from the
numerical value of the coefficients are important for decision making of firms as well as
formulation of the economic policy of the government. It helps to compare the effects of
alternate policy decision.
3. Forecasting: The obtained models are used for forecasting and policy formulation
which is an essential part in any policy decision. Such forecasts help the policy makers to
judge the goodness of fitted model and take necessary measures in order to re-adjust the
relevant economic variables.
1.7 Significance of Stochastic Disturbance Term
The disturbance term is a surrogate for all those variables that are omitted from the
model but that collectively affect Y. The reasons for to introduce the stochastic disturbance
term𝑈𝑖 are as follows:
1. Vagueness of theory: The theory, if any, determining the behavior of Y may be,
and often is, incomplete. We might know for certain that weekly income X
influences weekly consumption expenditure Y, but we might be ignorant or unsure
about the other variables affecting Y. Therefore, ui may be used as a substitute for
all the excluded or omitted variables from the model.
10
BASIC ECONOMETRICS STUDY E MATERIAL
2. Unavailability of data: Even if we know what some of the excluded variables are
and therefore consider a multiple regression rather than a simple regression, we
may not have quantitative information about these variables. It is a common
experience in empirical analysis that the data we would ideally like to have often
are not available. For example, in principle we could introduce family wealth as an
explanatory variable in addition to the income variable to explain family
consumption expenditure. But unfortunately, information on family wealth
generally is not available. Therefore, we may be forced to omit the wealth variable
from our model despite its great theoretical relevance in explaining consumption
expenditure.
3. Core variables versus peripheral variables: Assume in our consumption-income
example that besides income X1, the number of children per family X2, sex X3,
religion X4, education X5, and geographical region X6 also affect consumption
expenditure. But it is quite possible that the joint influence of all or some of these
variables may be so small and at best nonsystematic or random that as a practical
matter and for cost considerations it does not pay to introduce them into the model
explicitly. One hopes that their combined effect can be treated as a random variable
ui
4. Intrinsic randomness in human behavior: Even if we succeed in introducing all
the relevant variables into the model, there is bound to be some "intrinsic"
randomness in individual Y's that cannot be explained no matter how hard we try.
The disturbances, the ui‟s, may very well reflect this intrinsic randomness.
5. Poor proxy variables: Although the classical regression model assumes that the
variables Y and X are measured accurately, in practice the data may be plagued by
errors of measurement. Consider, for example, Keynes well-known theory of the
Psychological law of consumption function regards consumption expenditure (Yp)
as a function of income (Xp). But since data on these variables are not directly
observable, in practice we use proxy variables, such as current
consumptionexpenditre (Y) and current income (X), which can be observable. Since
the observed Y and X may not equal Yp and Xp, there is the problem of errors of
measurement. The disturbance term u may in this case then also represent the
errors of measurement. As we will see in a later chapter, if there are such errors of
measurement, they can have serious implications for estimating the regression
coefficients, the p's.
6. Principle of parsimony: Following we would like to keep our regression model as
simple as possible. If we can explain the behavior of Y "substantially" with two or
three explanatory variables and if our theory is not strong enough to suggest what
other variables might be included, why introduce more variables? Let ui represent
all other variables. Of course, we should not exclude relevant and important
variables just to keep the regression model simple.
11
BASIC ECONOMETRICS STUDY E MATERIAL
7. Wrong functional form: Even if we have theoretically correct variables explaining a
phenomenon and even if we can obtain data on these variables, very often we do
not know the form of the functional relationship between the regressand and the
regressors. Is consumption expenditure a linear (invariable) function of income or a
nonlinear (invariable) function? In two-variable models the functional form of the
relationship can often be judged from the scatter diagram. But in a multiple
regression model, it is not easy to determine the appropriate functional form, for
graphically we cannot visualize scatter diagrams in multiple dimensions.
1.7. Limitations of Econometrics
a. Applicable only to quantifiable phenomena
b. Lack of moral judgments, possibility of spurious regressions.
c. Irrational human behavior leads challenges in specifying variables and model
construction for estimation.
d. Econometric model construction and data analysis are time consuming, tedious
and complex because of mathematical statistical knowledge and economic
theoretical knowledge are needed.
e. Data are scarce relative to the number of parameter needed to be estimated
f. Econometrics is sometime criticized for relying too heavily on the interpretation of
raw data without linking it to establish economic theory or looking for casual
machinist
g. It tests the hypotheses, but neglects the concerns of error
1.7 Let us Sum Up
This chapter gave the meaning, definition, aims, nature, scope, methodology of
econometrics, significance of Ui term, goals of econometrics and Limitations of
Econometrics.
1.8 End Chapter Exercise
A. Multiple Choice Questions
1. The book „Econometric Theory‟ was authored by-----------------.
A. T C Koopman B.Arther S Goldberger
C. H. Theil D. J R N Stone
2. The name „Econometrics‟ was coined by ---------------------------.
A. Irving Fischer B. J.M Keynes
C. Ragnar Frisch D. Alfred Marshall
3. Econometric is an -----------------------.
A. amalgamation of mathematics and statistics
B. organization of Mathematics and Economics
C. integration of Mathematics, Statistics and Economics
D. integration of Economics and Statistics
12
BASIC ECONOMETRICS STUDY E MATERIAL
4. The first step in methodology of Econometrics is --------------------.
A. Constructing Schedule C. collecting data
B. Statement of a theory D. analyzing data.
5. The Quantitative analysis -Econometrics is based on---------------------.
A. concurrent development of theory.
B. concurrent development of observation.
C. Both a and b.
D. concurrent development of theory and observation, and these are related by
appropriate methods of inference.
6. Why does government use econometrics?
A. To aid decisions on prices. C. To aid decision on policy making.
B. To aid decision on production. D. To aid profit making.
7. Why do firm owners use econometric?
A. To aid decisions on prices.
B. To aid decision on inventory.
C. To aid decision on production.
D. All of the above.
8. The term „u‟ in an econometric model is usually referred to as the---------------.
A. dependent variable B. parameter
C. error term D. hypothesis
9. The Stochastic Disturbance term Ui is introduced for representing
A. Omitted Variable B. Vagueness of the theory
C. Unavailability of data D. All.
B. Short and Essay Type Questions
1. What is Econometrics?What are the goals of Econometrics?
2. Differentiate the Economic model vs.Econometric model.
3. State the objectives and features of Econometrics
4. Econometrics is a separate discipline. Why?
5. What are the limitations of Econometrics?
6. Explicate the significance of stochastic error term.
7. Discuss the role of Econometrics in a Society
8. Enumerate the Methodology of Econometrics.
1.9 Reference Books
1. DamodarN.Gujarati and Sangeetha, “ Basic Econometrics” Special Indian Edition , Tata
McGraw Hill Education Privated Limited (Sixth Print 2010) , ISBN: 978-0-07-066005-2
2. P.G.Apte,” Text book of Econometrics” Tata McGraw – Hill Publishing Company
Limited
3. Dhanasekaran,” Econometrics” 2nd Edition, Vrinda Publications (P) Ltd, Delhi-53 ,2011
ISBN: 978-81-8281-388-5
13
BASIC ECONOMETRICS STUDY E MATERIAL
4. HumbertoBarreto and Frank M. Howland,” Introductory Econometrics” Cambridge
University Press, First South Asian Edition 2009, ISBN: 978-0-521-12358-9.
5. Dilip M. Nachane,” Econometrics: Theoretical Foundations and Empirical
Perspectives” Oxford University Press, Second Impression 2010 , ISBN: 978-0-19-
564790-7
6. S.Shyamala,RavdeepKaur, Arul Pragasam,” A text book on Econometrics: Theory and
Applications” Vishal Publishing Co., Jalandhar 2017, ISBN: 81-88-646-98-9
7. S.P.Singh, Anil.K. Parashar, H.P. Singh,” Econometrics and Mathematical Economics”
Second Revised Editions,S.Chand and Company Ltd, New Delhi -55
8. A. Koutsoyiannis,” Theory of Econometrics” Second Edition Palgrave – New
York,2004,ISBN: 0-333-77822-7
9. Maddala .G.S.(1997), “Econometrics” , McGraw Hill, New York.
10. Johnston. (1997),” Econometric Methods” McGraw Hill, 4th Edition, New Delhi.
14
BASIC ECONOMETRICS STUDY E MATERIAL
UNIT-II
2.1 Objectives
2.2. Introduction
2.3 Meaning of Simple Regression
2.4 The concept of Population Regression Function (PRF)
2.5 Estimation of parameters using Ordinary Least Square (OLS) method
2.6 The Classical Linear Regression model (CLRM): The assumptions
2.7 Properties of OLS estimators
2.7.1 Property:1 Estimators are linear in parameters
2.7.2 Property: 2 Estimators are unbiased
2.7.3 Property:3 Estimators have a minimum variance
2.7.4 Gauss-Markov theorem
2.8. Goodness of Fit
2.9 Tests of Hypotheses
2.10 Simple soled problems
2.11 Let us sum up
2.12 Unit -End Exercises
2.13 Reference Books
2.1 Objectives
To learn the procedure of estimation of Ordinary Least Square Method,
To explore the valid interpretation of the regression estimates with assumptions on
independent variables and error term
To know the properties of estimators with proof
2.2 Introduction
We know, economics is the study of allocating the scarce resources to meet the
unlimited needs and wants of human being. During the allocation process, by a mixed
economy of country like India, the Demand for the goods and services or Supply of goods
and services are determine by few factors directly and indirectly. If the policy makers and
planners know the exact significant determinants which are influencing the demand for
and supply of particular good or service, then it will be easy to allocate the resources in apt
way to satisfy the consumer and supplier and study the impact into the economy of their
policy implementation. Further the exact manipulated elasticity of supply and demand
values will be helpful to know the scenario of an economy. Regression is tool to support in
micro and macro analysis for analyzing the influencing factors, elasticity and impact of any
15
BASIC ECONOMETRICS STUDY E MATERIAL
programme, event, policy, etc,.Hence in this chapter , an attempt is made to explain the
estimation procedure, assumptions made about the independent variable and error term
and properties of estimators.
2.3 Meaning of Simple Regression
Simple regression is the study of dependence of one variable (Y) with respect to
another variable (X) in order to estimate and predict the unknown parameter with known
values of dependent and independent variable.
2.4 The concept of Population Regression Function (PRF)
The regression Y on X is the conditional mean values of Y against the various X values.
The joining of all conditional values will be resulted the regression line Y on X otherwise
Population Regression Line. Hence, a Population Regression Line is simply the locus of the
conditional means of the dependent variable for the fixed values of the explanatory
variable. This can be explained by the following figure2.1 from the data table 2.1.
Figure 2.1 Population Regression Function
Hypothetical Example
A total population of 30 families in a village and their family income (X) and their
weekly consumption expenditure (Y) given in Indian rupees. The 30 families are divided
into 8 income groups from 100 to 240. Therefore 8 fixed values of X and the corresponding
Y values against each of the X values in order to speak about sub population.
Table 2.1 Distribution of weekly income and weekly consumption expenditure
Y ↓ X → 100 120 140 160 180 200 220 240
65 90 85 100 120 170 200 185
75 70 75 110 130 140 190 175
Weekly family consumption expenditure Y in Rs. 85 80 95 115 125 150 0 145
0 0 0 120 135 145 0 195
0 0 0 125 140 160 0 0
Total 225 240 255 570 650 765 390 700
Conditional mean of X E(Y/Xi) 75 80 85 114 130 153 195 175
16
BASIC ECONOMETRICS STUDY E MATERIAL
The figure 2.1 shows that for each X, there is a population of Y values, which are spread
around the mean of those values. From this diagrammatic explanation, it is clear that each
condition mean E(Y/Xi) is a function of Xi, where Xiis a given value of X
Symbolically, E(Y/Xi) = f (xi)
E(Y/Xi) is a linear function of Xi; it is called as Conditional Expectation
FunctionorPopulation Regression Function (PRF). Since E(Y/Xi) is a linear function of Xi,
say of the type, E(Y/Xi) = α + β Xi where α and β are unknown parameters, where α is
known as intercept and β is known as slope coefficients. The term regression, regression
equation and regression model will be used synonymously.
2.5 Estimation of parameters using Ordinary Least Square (OLS) method
The method of Ordinary Least Squares is attributed to Carl Fredrisch Gauss, a German
Mathematician. It is one of methods of estimation of regression analysis which is
frequently used by econometricians.
The Population Regression Function is E(Y/Xi) = α + β Xi + ui, the population
regression function is not directly observable. So , we have to estimate it from the sample
^ ^ ^
regression function, which is as follows:The sample regression function is Y = α + βXi
^ ^ ^ ^
Yi = Y + ei where Y = α + βXi
^
=> Yi - Y = ei(Since ei is the difference between actual and estimated value of Y)
According to OLS assumption Σei2 should be minimum.
Σe 2 ^ )2
= Σ (Y - Y
i i
^ ^
= Σ (Yi-α + βXi )2 must be minimum ------ (1)
ӘΣei2 ӘΣei2
= =0
Ә( ^
α) Ә( ^
β)
Equation (1) is partially derivated with respect to ^
α then
ӘΣei 2
^ ^
^ = Σ 2(Yi - α + βXi ) (-1) = 0
Ә( α )
^ ^
= -2 Σ (Yi - α - βXi ) = 0
^ ^
= Σ Yi – n α + β ΣXi = 0
^ ^
=> Σyi = n α + β ΣXi ------------- A
^
Equation (1) is partially derivated with respect to β then
ӘΣei2 ^ ^
^ = Σ 2(Yi - α + βXi ) (-xi) = 0
Ә( β )
^ ^
= -2 Σ (Yi - α - βXi )xi = 0
17
BASIC ECONOMETRICS STUDY E MATERIAL
^ ^
= Σ Yi Xi– α Σxi- β ΣXi2= 0
^ ^
=> Σyixi = α Σxi+ βΣXi2 ------------- B
A and B are called as normal equation
^ ^
=> Σyi = n α + β ΣXi
^ ^
=> Σyixi = α Σxi+ βΣXi2 Solving these two equation by using elimination method, one
^ ^
can obtain the value of α and β , The other simple alternative way is as follows:
^ ^
From (A) ΣYi= n α + β ΣXi
Σyi ^ ^ ΣXi
Divide by n throughout or both sides} n = α + β n
^ ^
Y =α+β x ------------- (2 ) {Since ΣYi/n = y and ΣXi/n= x }
From (B) normal equation, by changing its origin (0,0) to ( x , y ) we get
^
Σ (Xi- x ) (yi - y ) = ^
α Σ (xi - x ) + β (xi - x )2
Assume Σ (Xi - x ) = xi and (Yi - y ) = yi ,then we know Σ (xi - x ) =0 always, therefore
the first term in right hand side is equal to zero ( ^
α Σ (xi - x ) )and in the second term
^
keeping β and bringing the (xi - x )2to left hand side denominator then arriving the value
^
of β
Σ (xi - x ) (yi - y ) ^
= β -------- Result R1
Σ (xi - x )2
Σ xi yi ^ ^
i.e., Σx 2 = β and substituting this β value in 2 we get,
i
^ Σ xi yi
Y = α + Σx 2 x
i
Σ xi yi ^
i.e., Y - x Σxi2 = α --------Result R2
^ Σ xi yi ^ Σ xi yi
Hence, α = Y - x Σx 2 and β = Σx 2
i i
2.6 The Classical linear Regression model: The assumptions underlying the Method of
Ordinary Least Squares.
Our intention is to study the method of estimation, obtain the values of unknown
parameter and draw inferences about the true parameter. In constructing the econometric
18
BASIC ECONOMETRICS STUDY E MATERIAL
model, it is essential to depict specifically about how the independent variables and error
term are created or generated for a critical valid interpretation of the regression estimators.
There are ten assumptions in the context two variable regression model or Simple
regression model. The assumptions are as follows:
1. The regression model is linear in the parameter. That is Yi = α + β Xi + ui
2. X is assumed to be nonstochastic or X values are fixed in repeated sampling.
3. Given the value of Xi the mean or expected value of the random disturbance term ui
is zero. Technically, the conditional mean value of ui is zero.
4. Given the value of Xi,the variance of ui is same for all observations. That is the
conditional variances of ui are identical. Technically represents the assumptions of
Homoscedasticity.
5. Give any two X values, Xi and Xj (i ≠j), the correlation between any two u i and uj(i
≠j) is zero, where i and j are two different observations. Technically, this
assumption represents that no serial correlation or autocorrelation.
6. The disturbance term ui and explanatory variable X are uncorrelated. Technically
there exist zero covariance between ui and Xi.
7. The number of observation „n‟ must be greater than the number of parameters to be
estimated. Otherwise the number of observation „n‟ must be greater than the
number of explanatory variables.
8. The X values in a given sample must not all be the same. Technically var (X) must
be a finite positive number.
9. The regression model is correctly specified. Otherwise there is no specification bias
or error.
10. There is no perfect linear relationship among the explanatory variables. Technically
there is no Multicollinearity.
2.7 Properties of OLS Estimators
The Ordinary least square estimators possesses the following properties
1. The estimators are linear in parameters
2. The estimators are unbiased
3. The estimators have minimum variance or least variance is known as an efficient
estimator
4. The estimators which is linear, unbiased and with minimum variance or least
variance is called as „Best linear unbiased estimator‟ (BLUE) called Gauss Markov
Theorem.
5. An unbiased estimator is said to be consistent estimator when its sample size „n‟
tends to infinity and its variance tends to zero.
2.7.1 The estimators are linear in parameters
Proof: The OLS estimators are linear function of actual observation y.
^ Σ xi yi
OLS estimators of β = Σx 2
i
19
BASIC ECONOMETRICS STUDY E MATERIAL
Σxi (yi - y )
= Σxi2
Σxi yi - Σxi y
= Σxi2
Σ xi yi Σ xi y
= Σx 2 - Σx 2
i i
^ Σ xi yi Σ xi
β = Σx 2 - y Σx 2
i i
Σ xi yi
= Σx 2 - 0 (∵Σxi = 0)
i
^ Σ xi yi
β = Σx 2
i
^ ^ Σxi
β = Σ wiyi , β is linear. Where Σx 2 = Σwi
i
Now OLS estimator of ^
α to be prove as linear, let us take
^
α = Y -^ β x
Σyi ^
= n -β x
Σyi
= n - Σwiyi x Σwi = 0 ∵Σxi = 0
1 Σxi Σxi 1
= Σyi n - wi x Σwi2 =
Σxi2 = Σxi2
^ 1 Σxi Σxi Σxi2
α = Σyizi , linear ,where zi = n - wi x Σwixi = Σx 2 = Σx 2 =1
i i
^
Therefore ^
α and β are linear
2.7.2 Property – OLS estimators are unbiased
Proof: Let us take ^
β = Σw Yi i
= Σwi (α + βxi + ui)
= Σwi α + Σwi βxi + Σwiui
= α Σwi + Σwi βxi + Σwiui
= α. 0 + β Σwi xi + Σwiui (∵Σwi = 0, Σwixi = 1)
= 0 + β + Σwiui
= β + Σwiui
Taking expectation on both left hand side and right hand side
E ^ ()
β = E (β + Σwiui)
= E (β) + wiE(ui)
=β+0 (∵E(ui) = 0)
20
BASIC ECONOMETRICS STUDY E MATERIAL
()
E ^
β =β
E (^
β) = β is an unbiased estimator of β
Let us take,^
α value as
α = y -^
^ β x
Taking expectation on both sides we get
α) = E( y - ^
E(^ β x )
α) = E( y ) - E( ^
E(^ β x ) (∵ E ( y )=αo + β1 x ]
= αo + β 1 x - ^
β x
E (^
α ) = αo, Thus αo is an unbiased estimator.
2.7.3. The estimators have a minimum variance or efficient estimators:
Proof: To find the variance of ^
α and ^
β
Var (^β) = E [(^
β – E (^
β)2]
= E [(^
β –^β)2] (∵ E (^
β) = β)
= E [Σwiui]2
= E [Σ (wiui)2 + 2 wiwj uiuj j
i ≠j
= E (Σ wiui)2
1
= Σ wi2 E (ui)2 (∵ E(ui)2= σu2) ∵ Σwi2 = Σx 2
i
1
Var (^
β) = σu2 x Σx 2
i
Let us take the variance of intercept α
Var (^
α) = E [(^
α- E (^α)2] (∵ E (^
α) = α)
= E (^
α) - α)2
= E (ziui)2
= E [Σ zi2 ui2 + 2 zizj uiuj ]
i ≠j
= E [Σzi2ui2] + o (∵ Assumption of OLS)
= Σzi2 E (ui2) (∵ E (ui)2 = σu2)
= Σzi2 σu2
2
1
= σu2 Σ n - x wi
21
BASIC ECONOMETRICS STUDY E MATERIAL
1 2
= σu2 Σ n2 + x- 2 wi2 - n x- wi
n 2 Σw 2 = 1
= σu2 n2 + x- 2 Σwi2 - n x- Σwi (∵Σwi=0) i Σxi2
1 -
x
= σu2n + Σx 2
i
Σxi2 + n x- 2
= σu2 nΣx 2
i
Σ(xi- -x) 2 + n x- 2
= σu2
nΣxi2
Σxi2 + Σ x- 2 -2 Σxi x- +n x- 2
= σu2 (∵ Σ x- 2 = nx- )
n Σxi2
Σxi2-2n x- 2 + 2n x- 2
= σu
2 ( Second term and third term in bracket are cancelled)
nΣxi2
Σxi
2
= σu2 nΣx 2
i
σu2 Σxi2
= Σx 2 x n (x stands for multiplication of first term and second term)
i
Σxi2 σ u2
Var (^
α) = Var (^
β) . ∵Σxi2 = Var (^
β)
n
Σxi2
Var (^α) = Var (^β) . n
To find the co-variance of (^
α) and (^
β)
^
Cov (^
α^β) = E [(^
α –E(α)(^
β –E(β)] [∵α = Y - ^
β x and α = Y - β x ]
=E [(^
α –α)(^
β –β)]
Cov (^
α^β) = E [x- (β - ^
β) (^
β - β)] => Since/ ∵^
α–α= Y -^
β x - Y -β x
= E [x- (β - ^
β)2] => =-x- (β - ^
β) (both Y bar cancelled
= -x- E (^
β - β)2
= -x- Var ^ β
σ u2 ∵Var ^ σ u2
= - x- Σx 2 β = Σx 2
i i
To prove the minimum variance of least square estimators:
Proof: Say ^
β = Σw Y i i
Var (^
β) = Var (Σwi Yi)
= ΣwiVar Yi (∵Var Yi = Varui= σu2)
= σui2 Σwi2
22
BASIC ECONOMETRICS STUDY E MATERIAL
xi xi 2
= σ2 Σ wi - Σx 2 + Σx => The bracket terms / it‟s of the form (a+b)2 So,
i
i
xi 2 Σxi2 xi xi
= σ2 Σ wi - Σx 2 + σ2 (Σx 2)2 + 2 σ2 Σ wi - Σx 2 + Σx
i i i i
xi 2 1
= σ2 Σ wi - Σx 2 + σ2 Σx 2 (∵Σwi = 0)
i i
σ2 ∵ w = xi
= Σx 2 i
Σxi2
i
= Var (^
β)
The variance of the linear estimator ^
β is equal to the variance of least square estimator
^
β. Otherwise var (^ β) >var (β). Hence, ^ β is the minimum variance of linear unbiased
estimator of β.
2.7.4 Gauss-Markov theorem
Statement:OLS estimators areBest Linear Unbiased Estimator (BLUE)
Proof: To show that OLS estimators are BLUE
We have shown that already in properties that
^ ^
i. β = Σwiyiand E (β) = β ………Linear Property
ii. ^
α = Σz y and E (^
i i α) = α ……….Unbiased Property
σu2 Σxi2
iii. Var ^
β = Σx 2 and Var^ α = σu2 n Σx 2 ……Minimum Variance Property
i i
^
Hence the OLS estimators ^ α and β are linear functions of the independent variables
^ ^
and also unbiased estimators of ^ α and β respectively. Now to prove that only ^ α and β are
also best estimators, it is essential to show that among all the unbiased estimators, the
variance of the OLS estimators is the least, otherwise OLS estimators is BLUE.
^
^
Let β = Σciyibe any other linear estimator of β, where ci = wi + di and di being any
arbitrary constant other than zero. Then to prove an unbiased estimator as follows:
^
^ ^
^
E ( β) = β (∵ β is an unbiased estimator of β)
^
^
E ( β) = E (Σciyi)
= E (Σci (α + βxi + ui)]
= E (Σciα + β Σcixi +Σ ui ci)]
= ΣciE (α) + E (β) Σcixi +Σci E (ui)
^
^
E ( β) = β……Unbiased Proved. (only if Σei=0, and Σxici=1, E(ui)=0)
Rough work :i.e., If Σ(wi+di) = 0 and Σ (wi+di) xi = 1
=> Σdi= 0, Σwixi + Σdixi = 1
23
BASIC ECONOMETRICS STUDY E MATERIAL
=> 1+Σdixi=1 (∵Σwixi=1)
=> Σdixi=1-1=0
^
^
Now ,To prove the minimum variance property Let us take var ( β ) = E [Σciui]2
^
^
Var ( β ) = E [Σciui]2
= E [Σci2ui2]
= σu2 Σci2 (∵E(ui2) = σu2)
= σu Σ(wi+di)
2 2
= σu2 (Σ wi2+ Σ di2+ 2 Σ widi)
Σdixi
= σu2Σ wi2+ σu2 Σ di2+ σu2 2 Σ widi Now Σ diwi = Σx 2
i
= Var ^ β + σu2 Σ di2+ σu2 2 Σ widi where Σ di xi = Σ di (xi - x- )
= Var ^ β + σ 2 Σ d2+ 0
u i = Σ d x - Σ d x- = 0
i i i
= Var ^
β + σu2 Σ di2
^
^
Hence Var ( β ) = Var ^ β + a positive quantity
^
^ ^
^
∴Var ( β ) - Var ^β > 0 (or) Var ( β ) >Var ^
β
Thus, the variance of the OLS estimators is the least among all linear unbiased
estimators.
Similarly, for ^α
^
Let ^
α = Σci*yi be any other linear unbiased estimator of α where ci* = wi*+di*, di* being
any arbitrary constant other than zero.
Then,
^
^ = Σci*yi…….. Linear and to prove unbiased, Let us take
α
^
^ Σci* (α + βxi + ui)]
α=
= Σci*α + β Σci*xi +Σ ui ci* (only if Σci*=1 and Σci*xi = 0 and E (ui) = 0)
Taking expectation on both sides
^
^ = E (α) + Σci*E (ui) we get
E (α)
^
^ = α + Σci*E(ui)
α (since E (ui) = 0)
=α+0
^
^ =α
E (α)
Rough work : if Σci* = 1 = Σ(wi*+di*) = 1+Σ di* => Σ di*= 0 where Σ wi*= 1
Σci*xi = Σ(wi*+di*) xi = Σwi*xi + Σdi*xi = 0 (∵Σwi*xi = 0, Σdi*xi = 0)
Σwi*xi = Σdi*xi + 0
1 = Σdi*xi => Σdi*xi = -1
24
BASIC ECONOMETRICS STUDY E MATERIAL
^
Var (^
α) = E [Σci*ui]2
= E [Σci*2ui2]
= σu2 Σ ci*2
= σu2 Σ (wi* + di*)2
= σu2 Σ wi*2 + σu2Σdi*2 + σu2 2 Σwi*di*
= var (^
α) + a positive quantity
^
Var ( ^α) >var (^
α)
^
αis a linear unbiased estimator. The OLS estimators have the least variance, which is
the best linear unbiased estimator (BLUE).
2.8. Goodness of Fit
Goodness of fit refers the summary measure of how well the sample regression line fits
the data. Goodness of fit called as coefficient of determination ( r2) means that the
proportion or percentage of the total variation in Y explained by the regression model. The
properties of r2 are (a) Non negative quantity and (b) lies between 0 to 1.If the r2 value is
equal to one then it is perfect fit. If the value of r2 is zero means then there is relationship
between the regressor and regressand or no relationship between the dependent variable
and the independent variable. The coefficient of determination is the ratio of Explained
Sum Square with respect to Total Sum Square. It tells about what proportion of the
variation in the dependent variable or regressand is explained by the explanatory variable
or regressor
2.9 Tests of Hypotheses
The theory of hypothesis testing is concerned with developing rules or procedure for
deciding whether to reject or not reject the null hypothesis devised. There are two mutually
complementary approached for devising such rules, namely confidence interval and test of
significance. Both these approaches predict that the variable statistic or estimator.
Confidence interval approach is the concept of interval estimation. An interval
estimator is an interval or range constructed in such a manner that it has a specified
probability of including within it limits the true value of unknown parameter. The interval
thus constructed is known as confidence interval which is often stated in percent form such
as 90 percent or 95 percent.
In the significance test procedure, one develops a test statistic and examines its
sampling distribution under null hypothesis. The test statistics are usually a well-defined
probability distribution such as normal, t, F or chi-square. Once the test statistic is
calculated or computed from the data, the its ρ value will be easily taken from the
statistical tables. If the ρ value is small one can reject the null hypothesis. In choosing the ρ
value the investigator has to bear in mind the probabilities of committing Type I and Type
II error.
25
BASIC ECONOMETRICS STUDY E MATERIAL
2.10 Solved Numerical Problems
Illustration: 1.Using the following data-
Investment (Y) 65 57 57 54 66
Change in output (X) 26 13 16 -7 27
Estimate the regression line Y= + X, test the hypothesis that =0 against the
alternative < 0 at 5% level of significance, also construct 95% confidence interval for .
Solution: The estimated line is Y= ̂ + ˆ X. Now first of all we will calculate the
parameters of the equation.
Calculation of Parameters and Error Term
^ ^
X Y X2 XY Y e=Y- Y e2
26 65 676 1690 54.55+.35x26=63.65 1.35 1.82
13 57 169 741 54.55+.35x13=59.10 -2.10 4.41
16 57 256 912 54.55+.35x16=60.15 -3.15 9.92
-7 54 49 -378 54.55+.35x-7=52.10 1.90 3.36
27 66 726 1782 54.55+.35x27=64.00 2.00 4.00
75 299 1879 4747 299 23.51
X. Y 75x 299
XY n
4747
5
̂ yx =
X 2 75
2
X n
2
1879
5
4747 4485 262
= .35(approx.)
1879 1125 754
and also X =15 and Y =59.8
ˆ ˆ X on passing through mean Y
Since the equation is Y= ˆ ˆ X
On passing through mean
Y ˆ ˆ X
59.8= ˆ .35x15
̂ =59.8-5.25=54.55
Regression equation will be –
Ŷ =54.55+0.35X
On putting given values of X, the corresponding values of Ŷ can be calculated as
shown in the table.Now, we will test the hypothesis,
Suppose the null hypothesis is =0. The formula of t is
ˆ
t= . x Thus, t 0.35x
2 754
x3 0.35x9.81 3.433
e 2
i /n2 i
23.51
26
BASIC ECONOMETRICS STUDY E MATERIAL
Tabulated value of„t‟ as 3 degree of freedom is 2.353.Since tabulated value is less than
calculated value of t, the hypothesis is to be rejected and alternative hypothesis will be
accepted. Thus is different from zero.
Sˆ e 2
23.51
0.102
n 2. x 2
3x 754
Confidence interval at 95% level is 0.35 (3.182x0.102) ,0.35 0.325
Illustration:2. The following table gives the production of steel in different years at a steel
bx
factory. Find out the equation y= a.e expressing the relationship between production and
year
Years 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
Production (000
10.2 12.0 13.9 15.9 17.9 20.1 22.7 26.0 29.0 32.5 36.1
tons)
bx
Solution: The given equation is y= a.e where a andb are the constant and e the exponential
constant. Taking log to the base e we have-
log e y log e a bx On putting Y= log e y and a 0 log e a; our equation will be
y= a 0 bx Now least squares method will be applied to estimate a 0 and b.
,
Year x Production y Y = logey1= log10y×2.3025 x2 xY
2011 -5 10.2 1.0086x2.3025=2.3223 25 -11.6115
2012 -4 12.0 1.0792x2.3025=2.4848 16 -9.9392
2013 -3 13.9 1.1430x2.3025=2.6318 9 -78954
2014 -2 15.9 1.2014x2.3025=2.7662 4 -5.5324
2015 -1 17.9 1.2529x2.3025=2.8848 1 -2.88848
2016 0 20.1 1.3032x2.3025=3.0006 0 0
2017 1 22.7 1.3560x2.3025=3.1222 1 3.1223
2018 2 26.0 1.4150x2.3025=3.2580 4 6.5160
2019 3 29.0 1.4624x2.3025=3.2580 9 10.1016
2020 4 32.5 1.5119x2.3025=3.4811 16 13.9244
2021 5 36.1 1.5575x2.3025=3.5861 25 17.9305
Total 32.9015 110 13.7315
The two normal equations are
Y na 0b x ……………(i)
xY a x b x
0
2
……………..(ii)
1
Since we are given with the log table to base of 10, to change the base to ‘e’ we will have to multiply the
usual value of log to the base 10 with the value of ‘log10e’ which is equal to 2.3025.
27
BASIC ECONOMETRICS STUDY E MATERIAL
On putting values in the above equations
32.9051= 11a0+0 , a0 = 2.9914
13.7315=0+110b , b =0.1248
Since a0 =logea=2.9914
a= 19.92 and the equation is Y= 19.92e.01248x.
The reciprocal method: The given relationship may be of the following type-
1
Y . In such a situation the reciprocals of X will be taken and rest of the
X
procedure will remain the same.
Illustration 3- Following are the observations on two variables X and Y.
X: 2 3 4 5
Y: 3.1 2.9 2.7 2.6
Estimate the equation Y= / X
Solution: We have to establish the relationship between Y and reciprocals of X(X*).
Y X X* X*2 X*Y
3.1 2 0.50 0.2500 1.550
2.9 3 0.33 0.1089 0.957
2.7 4 0.25 0.0625 0.675
2.6 5. 0.20 0.0400 0.520
11.3 1.28 0.4614 3.702
ˆ ˆ X *
The estimated relationship is Y
where X* is the reciprocal of X. By formula
X*Y
X* Y
3.702
1.28X11.3
ˆ
X
n 4
1.28
X
*2 2
*2 0.4614
n 4
3.702 3.616 0.086
1.66
0.4614 0.4096 0.0518
Y ˆ 1.66X * or ˆ Y 1.66X *
^
=2.825-(1.66x.32)=2.825-.531
1
The equation is Y=2.294+1.66X* or Y=2.29+1.66
X
Illustration:4.The following statistical coefficients were deduced in the course of an
examination of the relation between yield of wheat and the amount of rainfall.
Yield in Ib Annual rainfall
per acre in inches
28
BASIC ECONOMETRICS STUDY E MATERIAL
Mean 985.0 12.8
Standard deviation 70.1 1.6
Correlation coefficient between yield and rainfall +0.52
Estimate the linear regression of yield on rainfall. Calculate the most likely yield of wheat
per acre when the amount of rainfall is 9.2 inches.
Solution: Let yield is denoted by Y and annual rainfall is denoted by X.
Let the linear regression of yield on rainfall is Y X u
Here, Given X =12.8, Y =985, =70.1, x =1.6, r=0.52
y 70.1
We know that ˆ yx r. 0.52x =22.78
x 1.6
ˆ Y ˆ X 985 -22.78x12.8=693.416
Hence regression line of yield on rainfall is
Ŷ =693.416+22.78X
When X=9.2 (given)then Y= 693.416+(22.78x9.2)=902.992
So when the rainfall is 9.2” then yield will be 902.992 Ibs.
Illustration: 5. A sample of 20 observations on X and Y gave the following data:-
Y =21.9 Y Y 86.9
2
X =186.2 X X X XY Y 106.4
2
215.4
Answer the following:-
a) Estimate the regression of Y on X
b) Estimate the regression of X on Y
c) Compute the mean value of Y corresponding to X=10.
d) Compute the mean value of X corresponding to Y=1.5.
Solution
a) Let regression line of Y on X be
Y X
we know that
ˆ yx
X X Y Y 106.4 0.49
XX 2
215.4
X= X = 186.2 =9.31 and Y =
Y = 21.9 =1.09
n 20 n 20
ˆ
So, Y X 1.09 -(0.49x9.31) = -3.47
ˆ
Thus, estimated regression line of Y on X is
Y= -3.47+0.49X
b) Now, let the regression line of X on Y be
X= Y
29
BASIC ECONOMETRICS STUDY E MATERIAL
ˆ yx
X X Y Y 106.4 1.22
YY 2
86.9
ˆ X ˆ Y 9.31 -(1.22X1.09) =7.98
Thus estimated regression line of X on Y is
X= 7.98+1.22Y
c) when X=10
then Y=-3.47+(0.49x10) =1.43
d) when Y=1.5
then X=7.98+(1.22x1.5)= 9.81
Illustration 6. The following data were obtained in a sample study:-
X =56, Y =40, X 2 =524 Y 2 =256
X =364, N=20
Answer all of the following:
a) Estimate the regression line Y X
b) Estimate the regression line X Y
c) Compute the value of Y corresponding to a value 7 for X
d) Compute the value of X corresponding to a value 3 for Y
Solution:
a) Estimated regression line is Ŷ X
XY
X. Y
where ˆ n
X
X
2
2
n
56x 40
364
= 20 252 0.686
56x56 367.2
524
20
X
X 56 2.8 , Y
Y 40 2
n 20 n 20
ˆ
ˆ Y X 2 -0.686x2.8=2-1.921=0.079
Thus estimated regression line becomes Y=0.079+686X
b) Estimated regression line is X Y
XY n
X. Y
^
where,
Y
Y
2
2
30
BASIC ECONOMETRICS STUDY E MATERIAL
56x 40
364
= 20 252
40x 40 256 80
256
20
252
= 1.43
176
ˆ X ˆ Y 2.8 -1.43x2=2.8-2.86=-0.06
Now the estimated regression line becomes
X=0.06+1.43Y
c) When X=7
Then Y=0.079+.686x7=4.881
d) When Y=3
then X=-0.06+1.43x3=4.23
Illustration: 7 The following table gives ages in year of 10 husbands and their wives:-
Age of husband (X) 18 19 20 21 22 23 24 25 26 27
Age of wife (Y) 17 17 18 18 18 19 19 20 21 22
a) Estimate the linear regression of the ages of wives (Y) on the ages of husbands (X).
b) Plot the regression line on the scatter diagram.
c) Are the age of wives dependent on the ages of their husbands? Use 5% level of
significance.
d) Estimate the age of the wife whose husband is 28 years old.
Solution a) Let the estimated Y on X be
Ŷ X
XY
X Y
where, ˆ n
X
X
2
2
n
and ˆ y ˆ X
To solve the ̂ and ̂ we shall construct the following table-
X Y x-X-a (a=23) x2 y=Y-a (a=19) xy y2
18 17 -5 25 -2 10 4
19 17 -4 16 -2 8 4
20 18 -3 9 -1 3 1
21 18 -2 4 -1 2 1
22 18 -1 1 -1 1 1
23 19 0 0 0 0 0
24 19 +1 1 0 0 0
31
BASIC ECONOMETRICS STUDY E MATERIAL
25 20 +2 4 +1 2 1
26 21 +3 9 +2 6 4
27 22 +4 16 +3 12 9
X Y X =-5 X 2
y =-1 xy y 2
=25
=225 =189 =85 =44
X
X 225 22.5 , Y
Y 189 18.9
n 10 n 10
xy x y 44
(5) x (1)
43.5
Again, ˆ n = 10 = =0.527
x 2 5
2
82.5
x 2
n
85
10
ˆ y ˆ X =18.9-(0.527x22.5) =7.044
The regression line becomes Y= 7.044+0.527X
d) when X=28 ,Putting the value of X in the regression line
Y=7.044+0.527x28=218 ,Thus when the age of husband is 28 years, the age of wife will be
22 years (approx)
(c) To test the hypothesis we shall apply the „t‟ test
t= x x .2
e2
n2
To determine „e‟ we shall construct table as following-
x y Ŷ e=Y- Ŷ e2
18 17 16.529 0.471 0.2197
19 17 17.053 -0.053 0.0028
20 18 17.583 0.417 0.1738
21 18 18.110 -0.110 0.0121
22 18 18.637 -0.637 0.4056
23 19 19.164 -0.164 0.0268
24 19 19.691 -0.691 0.4764
25 20 20.218 -0.218 0.0475
26 21 20.745 0.255 0.0650
27 22 21.272 0.272 0.5299
e 2
=1.9596
Let =0
32
BASIC ECONOMETRICS STUDY E MATERIAL
.527 .527 x 82.5x8
t = x 82.5
1.9596 1.9596
10 2
0.527 x 25.69 13.539
= 9.678 t calculated = 9.678
1.399 1.399 ,
t- tab at 95% confidence interval at (n-2)=10-2=8 degress of freedom =1.860
Therefore t.cal>t. tab .Since the tabulated values of t is less than its calculated value,
thus the hypothesis is to be rejected i.e. 0 and alternative hypothesis will be accepted.
Illustration: 8 The following data were collected from 5 different plants in a certain
industry.
Total cost (Y) 80 44 51 70 61
Production (X) 12 4 6 11 8
Answer the following questions
a) Estimate a linear total cost function Y= X for the industry.
b) What is the economic significance of the estimate of and .
c) Estimate the total cost for a level of production of 10.
Solution: Our regression line is Y= X
XY
X Y
where, ˆ n ˆ Y ˆ X
and,
X
X
2
2
To estimate the ̂ and ̂ values we shall construct the following table
Y X X=x-a (a=8) X2 Y=y-a (a=61) XY
80 12 +4 16 +19 76
44 4 -4 16 -17 68
51 6 -2 4 -10 20
70 11 +3 9 +9 27
61 8 0 0 0 0
Y =306 X =41 X =+1 X 2
=45 Y =+1 xy =191
191 1 / 5 954
yx = =4.25
45 1 / 5 224
X
X 41 8.2 Y Y 306 61.2
n 5 n 5
ˆ Y ˆ X =61.2-4.25x8.2=26.35
33
BASIC ECONOMETRICS STUDY E MATERIAL
Regression line of Y on X is Y=26.35+4.25Y
b) Economic significance of „ ‟and „ ‟
Our estimated linear total cost equation is
Y=26.35+4.25Xwhere, =26.35 and =4.25
Hence, = total fixed cost and =marginal cost
Thus total cost =fixed cost +marginal cost x production
=total fixed cost +total variable cost.
Either production takes place or not the total fixed cost remains constant while total
variable cost vary with production. The marginal cost plays an important role in the
economic field. Each firm will get maximum profit where marginal cost is equal to the
marginal revenue. In the field of distribution each entrepreneur should employ the factors
of production to the point where marginal cost is equal to the marginal revenue product
(M.R.P).In the above example, (26.35) units) as a part of total cost is always fixed. By
knowing the entrepreneur can estimate his total of production. The accuracy of and
is necessary for the good results
c)when X=10 , then Y=26.35+4.25x10=68.85
2.11 Let us sum up
This unit explains the simple linear regression model, its method of estimation,
properties of the estimators, and goodness of fit and soled numerical problems.
2.12 Unit -End Exercises
A. Multiple Choice Questions
1. One of the following assumption is not in OLS
A. E(ui /xi)= 0
B. Cov(uiuj/xixj) = 0
C. No Autocorrelation
D. No Perfect Multicollinearity
2. Find out which is not the property of a parameter?
A. Linear in Parameter
B. Parameters are Unbiased
C. Parameters has the Minimum Variance
D. Biased
3. The Co-efficient of Determination Measures
A. The correlation between the X and Y
B. Error
C. Goodness of fit of the model
D. TSS
4. The formula for testing the co-efficient with t- test is
34
BASIC ECONOMETRICS STUDY E MATERIAL
A. Co-efficient÷ Std.Error C. Std.Error ÷ Co-efficient
B. both (A) and (C) D. larger variance ÷ Smaller Variance
5. The value of β in simple regression is
A. B. (X‟X)-1 X‟Y C. 𝑦 2 D. (X‟X)-1 XY
6. The number of Explanatory variables in a simple regression is------------.
A. Zero B.Two C. One D. More than Two
7. The property of an estimator E(𝛽1) = β1 is termed as….
A. Linearity B.Efficiency C.Unbiasedness D. Accuracy
2
8. Choose the correct one from the following about 𝑅
A. 𝑅 2 = TSS/RSS C. 𝑅 2 = RSS/TSS
2
B. 𝑅 = ESS/TSS D. 𝑅 2 = TSS/ESS
B. Short Answer and Essay type questions
1. State the assumptions of classical linear regression.
2. Derive the value of unknown parameter of α, β for the given regression equation
^ ^ ^
Y = α + βXi
3. What are the properties of estimator?
4. Prove that the estimator is linear and unbiased
5. State and prove the Gauss Markov theorem.
2.13 Reference Books
1. DamodarN.Gujarati and Sangeetha, “ Basic Econometrics” Special Indian Edition , Tata
McGraw Hill Education Privated Limited (Sixth Print 2010) , ISBN: 978-0-07-066005-2
2. P.G.Apte,” Text book of Econometrics” Tata McGraw – Hill Publishing Company
Limited
3. Dhanasekaran,” Econometrics” 2nd Edition, Vrinda Publications (P) Ltd, Delhi-53 ,2011
ISBN: 978-81-8281-388-5
4. HumbertoBarreto and Frank M. Howland,” Introductory Econometrics” Cambridge
University Press, First South Asian Edition 2009, ISBN: 978-0-521-12358-9.
5. Dilip M. Nachane,” Econometrics: Theoretical Foundations and Empirical
Perspectives” Oxford University Press, Second Impression 2010 , ISBN: 978-0-19-
564790-7
6. S.Shyamala,RavdeepKaur, Arul Pragasam,” A text book on Econometrics: Theory and
Applications” Vishal Publishing Co., Jalandhar 2017, ISBN: 81-88-646-98-9
7. S.P.Singh, Anil.K. Parashar, H.P. Singh,” Econometrics and Mathematical Economics”
Second Revised Editions,S.Chand and Company Ltd, New Delhi -55
8. A. Koutsoyiannis,” Theory of Econometrics” Second Edition Palgrave – New
York,2004,ISBN: 0-333-77822-7
9. Maddala .G.S.(1997), “Econometrics” , McGraw Hill, New York.
10. Johnston. (1997),” Econometric Methods” McGraw Hill, 4th Edition, New Delhi.
35
BASIC ECONOMETRICS STUDY E MATERIAL
UNIT3
3.1 Objectives
3.2 Introduction
3.3 Meaning of Multiple Regression
3.4 Assumptions underlying the Method of OLS
3.5 Estimation of Multiple Linear Regression Model
3.6 Properties of OLS estimators
3.6.1 Estimators are linear in parameters
3.6.2 Estimators are unbiased
3.6.3. Estimators have a minimum variance
3.7. Goodness of Fit-R2 and the adjusted R2
3.8 Solved numerical Problems
3.9 Let us sum up
3.10 Unit -End Exercises
3.11 Reference Books
3.1 Objectives
The specific objective of this chapter intends towards the student are as follows
To make them with familiarity of multiple regression and its properties
To understand the multiple regressions for using in reality as a researcher, as a field
surveyor, and desk researcher.
3. 2 Introduction
This unit made an attempt to explain the Multiple Regression Model and its properties.
Any economic activity is not with single factor, but determined by more than one factors.
The daily routine life demand in early morning tea or coffee determined the factors like
input cost of its preparation, taste and preference, and its competitive beverages prices.
Every economic activity determined by more than one variable from micro to macro.
Hence it is essential to study this chapter for dealing such a kind of situation.
3.3 Meaning of Multiple regression
Multiple regression models are an extended form of more than one explanatory
variable. The multiple regression models are designed to describe economic relationships
as an extension of the simple regression model.
36
BASIC ECONOMETRICS STUDY E MATERIAL
Let us start with the theoretical proposition that changes in one variable can be
explained by change in several other variables. Such a relationship is described in simple
way by a multiple linear regression equation of the form
Yi 1 2X2i 3X3i ............ k Xki u i ……. (1)
where Y denotes the dependent variable, the X‟s are explanatory variables and U is a
stochastic disturbance term.
3.4 The assumptions underlying the Method of Ordinary Least Squares
Basic assumption which are as follow
a. ui is normally distributed
b. E (ui)=0
c. E(ui2)= u
2
d. E(uiuj )= 0 for i j
e. Each of the explanatory variables is non-stochastic with fixed value in repeated
X
n
2
samples and such that for any sample size ki X k / n is a finite number
i 1
different from zero.
f. The number of observations exceeds the number of coefficients to be estimated.
g. No exact linear relation exists between any of the explanatory (X‟s) variables.
3.5 Estimation of Model by Method of Ordinary Least Square
Ordinary Least squares principle applies over the regressions model which is
expressed in matrix. As the regression model was
Y= X u where
y1 x 21x 31...x k1 2 u1
y 2 x x ...x u
22 32 k 2 3 2
y3 . .
y , X , . u
. . . .
. . . .
yn x 2 n x 3n ...x kn k u n
Sum of the squared residuals is
e1
e
n 2
ui
2
e i e' e . e1e 2 ...e n e12 e 22 ........ e 2n
2
i 1
.
e n
= Y Xˆ Y Xˆ
37
BASIC ECONOMETRICS STUDY E MATERIAL
= YY ˆ XY YXˆ ˆ XXˆ
= YY 2ˆ XY ˆ XXˆ
(Since XY is a scalar, thus it will equal to YXˆ )
e 2
i YY 2ˆ XY ˆ XXˆ
For least squares first differentiation should be equal to zero
or
ei2
2XY 2XXˆ
ˆ
2XXˆ 2XY 0
XXˆ XY
Premultiplying by XX to both the sides we have,
1
XX1XXˆ XX1 XY
ˆ XX1 XY where X is column vector, X=row vector.
This is the fundamental result for the least squares estimators. To determine the 0 we shall
use Y ˆ 1 X1 ˆ 2 X2 ... ˆ k Xk ˆ 0
3.6 Properties of Ordinary Least Squares Estimators
Least squares estimators are best linear and unbiased (BLUE)
Let us take linear model be- Y=X +u
3.6.1 i) The Least Square Estimators are Unbiased Estimators-
Since ˆ XX XY
1
Y=X β+u
ˆ XX XX u
1
XX XX XX Xu
1 1
XX Xu
1
taking expectations
E ˆ XX XE(u)
1
Since E(u)=0
Ê
Thus are unbiased estimators of
3.6.2 ii) The Least Square Estimators are linear estimators- Since least squares estimators
have linear relation with Y, so they are linear.
38
BASIC ECONOMETRICS STUDY E MATERIAL
3.6.3 iii) The Least Square Estimators are best estimators-Now we will prove that our
estimators are best among all the estimators, since
ˆ XX Xu ˆ XX1 Xu
1
Var . ̂ = E
ˆ
ˆ
E
= X
1
X X u
XX 1
Xu
= E X X X u.u XX X
1 1
Since XX XX
1 1
XX X.E(uu).XXX
1 1
XX X.2u .In XXX
1 1
2u .In .XX .XXXX 2u XX
1 1 1
Var ̂ var (u). Hence ̂ is an best estimator.
Now we shall prove the more general result which is a special case. The more general
result also has its applications in predication problem. Let us consider a relation
b=(A+B)Y where Y=X +u, A = XX X and B is a constant.
1
or b=(A+B) (X +u)
=(A+B)X +(A+B)u
Taking expectation of both sides we have
E(b) = (A+B)X since E(u)=0
=AX +BX
= +BX since AX=In
E(b) = only if BX=0.
Thus b is an unbiased estimator.
Again b= +(A+B)u
b- =(A+B)u
and b A Bu u(A B)
Eb b E A BuuA B
= Euu(A B)(A B)
2u (A B)(A B)
2u AA BA AB BB
39
BASIC ECONOMETRICS STUDY E MATERIAL
2u XX XXXX BXXX XX XB BB
1 1 1 1
Since XX XX In ; BX 0
1
Var.(b) = 2n XX1 BB
= n XX n BB
2 1 2
ˆ
=Var. u BB
2
var (b) –var ˆ 0
Let us consider the vector Ci with unity in the ith position and zero elsewhere, then
var (b) Var
Thus other estimators will either have greater or at least equal values to least squares
estimator. So we can say that only least squares estimators have smallest variance among
all linear unbiased estimators. Hence our least squares estimators are the best linear and
unbiased (BLUE).
3.7 R2 and the adjusted R2 (Goodness of Fit)
R2 is a non-decreasing function of the number of explanatory variables or regressors
present in a model; as the number of regressors increases, R2 almost invariably increases
and never decreases. Stated differently, an additional X variable will not decrease R2. It
explains the goodness of fit of a model.
ESS
R2
TSS Where ESS stands for Explained Sum Square and TSS stands for Total Sum
Square.
RSS
1
TSS
1
û i2
y 2
i
Now yi is independent of the number of X variables in the model because it is
2
simply (Yi Y) . The RSS, û i , however, depends on the number of regressors
2 2
present in the model. Intuitively, it is clear that as the number of X variables increases,
û 2
i is likely to decrease (at least it will not increase); hence R2 as defined it will increase.
In view of this, in comparing two regression models with the same dependent variable but
differing number of X variables, one should be very wary of choosing the model with the
highest R2.
To compare two R2 terms, one must take into account the number of X variables
present in the model. This can be done readily if we consider an alternative coefficient of
determination, which is as follows:
40
BASIC ECONOMETRICS STUDY E MATERIAL
R 2 1
û i2 / (n k )
where k = the number of parameters in the model including the
yi2 / (n 1)
intercept term. (In the three-variable regression, k = 3. Why?) The R2 thus defined is known
as the adjusted R2, denoted by R 2 . The term adjusted means adjusted for the degrees of
freedom associated with the sums of squares entering into: û 2
i has n-k degrees of
freedom in a model involving k parameters, which include the intercept term, and y 2
i
has n-1 degree of freedom. (Why?) For the three-variable case, we know that û 2
i has n-3
ˆ 2
degrees of freedom. R 1 where ̂ 2 is the residual variance, an unbiased estimator of
2
2
SY
true 2 , and SY is the sample variance of Y. It is easy to see that R 2 and R2 are related
2
n 1
because, substituting the value of R2, obtain R 2 1 (1 R 2 ) It implies that as the
nk .
number of X variables increases, the adjusted R2 increases less that the unadjusted R2; and
R 2 can be negative, although R2 is necessarily nonnegative. In case R 2 turns out to be
negative in an application, its value is taken as zero. Which R2should one use in practice?
As Theil notes: ….it is good practice to use R 2 rather than R2 because R2 tends to give an
overly optimistic picture of the fit of the regression, particularly when the number of
explanatory variables is not very small compared with the number of observations,
explanatory variables is not very small compared with the number of observations.
3.8 Solved Numerical Problems
Illustration: 1. A random sample of five families yields the following data
Family A B C D E
Saving S(in hundred Rs.) 6 12 10 7 3
Income Y (in thousand Rs) 8 11 9 6 6
No. of children, N 5 2 1 3 4
Estimate the regression line of S on Y and N
Solution: The linear regression line will be
S 0 1Y 2 N u
So estimated line will be
Ŝ ˆ 0 ˆ 1Y ˆ 2 N
To apply least square method, the three normal equation are
S n.ˆ ˆ Y ˆ N
0 1 2
SY ˆ . Y ˆ Y ˆ NY
0 1
2
2
SN ˆ . N ˆ NY ˆ N
0 1 2
2
Values of these equations will be calculated as follows:-
41
BASIC ECONOMETRICS STUDY E MATERIAL
Family S Y N Y2 N2 SY SN NY
A 6 8 5 64 25 48 30 40
B 12 11 2 121 4 132 24 22
C 10 9 1 81 1 90 10 9
D 7 6 3 36 9 42 21 18
E 3 6 4 36 16 18 12 24
Total 38 40 15 338 55 330 97 113
On putting values in normal equations,
38 5ˆ 0 40ˆ 1 15ˆ 2 ……….(1)
330 40ˆ 338ˆ 113ˆ
0 1 2 ……… (2)
97 15ˆ 0 113ˆ 1 55ˆ 2 …….....(3)
On multiplying equation (1) by 8 and subtracting it from equation (2)
18ˆ 1 7ˆ 2 26 ………(4)
On multiplying equation (1) by 3 and substracting it from equation (3)
7ˆ 1 10ˆ 2 17 ………(5)
On multiplying equation (4) by 10 and (5) by 7
180ˆ 1 70ˆ 260
49ˆ 1 70ˆ 2 119
on subtracting
131ˆ 141
1
141
or ˆ 1 1.076
131
On putting the value of ̂1 in equation (5)
7x1.076- 10ˆ 2 =17
- 10ˆ 2 =9.466 or ˆ 2 .9466 .947
Now the equation will pass through mean values. So , S ˆ 0 ˆ 1 Y ˆ 2 N
7.6 ̂0 +(8x1.076)+3(-947) = ̂0 =1.833
Our estimated relation will be
S=1.833+1.076Y-.947N
Illustration: 2 The following matrix gives the variances and covariances of three variables:-
X1=log food consumption per capita
X2=log food price
X3=log disposable income per capita
X1 X2 X3
X1 7.59 3.12 26.99
X 2 29.16 30.80
X 3 133.0
42
BASIC ECONOMETRICS STUDY E MATERIAL
On the assumption that the demand relationship may be adequately represented by a
function of the firm Y1= AY2 Y3 (where Xi=log Yi) estimate the income elasticity of
demand.
Solution: From the above matrix we have,
X12 =7.59 X1X2=3.12 X1X3=26.99
X 22 =29.16 X2X3=30.80
X 32 =133.0
Given regression line is
Y1= AY2 Y3 ….(1)
Taking log of both sides we have,
log Y1=log A+ logY2+ log Y3 .
or X1= 0 +X2+X3 . …(2)
where Xi=log Yi (given) and log A= 0
From the regression line we have
1 = , X=[X2 X3]
X X2 X 2 X3 29.16 30.80
XX 2 X 2 X3 = 2
X3 X 2 X 3 X32 30.80 133.0
Since we know,
XX 1 Adj.of X X
XX
29.16 30.80 133X29.16 30.80X30.80
XX
30.80 133.0 3878.28 94.64 2929.64
29.16 30.80
Transpose of XX
30.80 133.0
Cofactor of 29.16 =133
Cofactor of30.80 =30.80 {Using (-1)i+jwhere i=no of rows, j=no of columns}
Cofactor of 30.80 =30.80
Cofactor of 133 =29.66
133.0 30.80
Adjoint of XX
30.80 29.16
133.0 30.80
XX 1 1
2929.64 30.80 29.16
43
BASIC ECONOMETRICS STUDY E MATERIAL
133 30.80
2929.64 2929.64 .045 .010
30.80 29.16 .010 .009
2929.64 2929.64
X 2 X X 3.12
Now XX1 X1 1 2
X3 X1X3 26.99
.045 .010 3.12
ˆ XX XX1
1
.010 .009 26.99
3.12X.045 26.99X.010 0.1404 0.2699
3.12X.010 26.99X.009 0.0312 0.2429
0.1295
Thus =0.1295 and =0.2177
0.2117
Our given equation is, X1= 0 + X2+ X3 .where we have estimated =0.1295 and
=0.2117
Here =0.2117 is known as income elasticity of demand.
(since there are some structural relationships, which describe the behaviour of the
individuals in the economy. These are, for instance demand function, production function
and supply function. These structural relationships also involve structural parameters,
which are to be estimated by statistical methods. Examples of such parameters are
elasticity of demand with respect to price, elasticity of demand with respect to income,
marginal propensity to consume and marginal production. )
Illustration: 3 Three related variates X1, X2, X3 take the following sets of values:-
X1 1 2 3 4 5
X2 2 1 5 4 3
X3 3 1 4 5 2
a) Show that the regression plan of X1 on X2 and X3 is
18X1-17X2+10X3=33
b) Also test the null hypothesis H0( 2=0) against alternative hypothesis H1( 2 0) at 5%
level of significance.
Solution: Let the regression line be
X1= 1+ 2X2+ 3X3.
2 X
We have ˆ XX XX1 where, ˆ X X 2X3 and X' 2
1
3 , X3
44
BASIC ECONOMETRICS STUDY E MATERIAL
We shall estimate ̂ as following
X1 X2 X3 X12 X 22 X 32 X1X 2 X1X 3 X 2X3
1 2 3 1 4 9 2 3 6
2 1 1 4 1 1 2 2 1
3 5 4 9 25 16 15 12 20
4 4 5 16 16 25 16 20 20
5 3 2 25 9 4 15 10 6
15 15 15 55 5 55 50 47 53
X1
X 1
15
3 X 2 3, X3 3
n 5
Now we shall construct the following quantities in terms of deviation around the
means
X
x X n
2
15X15
2
2
2
2
2
55 10
5
15X15
x 2
3 55
5
10, x12 10
x1x 2 X1X2 1n 2 50 5 5
X X 15x15
x1x 3 47 45 2, x 2 x 3 53 45 8
X x2 x 2 x 3 10 8
XX 2 X 2 X3 2
X3 x 2 x 3 x 3 8 10
10 8
XX 100 64 36
8 10
Cofactor of 10=10, cofactor of 8=-8, cofactor of 8=-8, Cofactor of 10=10.
Adjo int of XX 10 8
XX 1 ,Adjoint of XX
Deter min antof XX 8 10
10 8 0.278 0.222
XX 1 1
36 8 10 0.222 0.278
x x x 5
XX1 2 x1 2 1
x3 x 3 x1 2
0.278 0.222 5
XX 1 XX1
0.222 0.278 2
45
BASIC ECONOMETRICS STUDY E MATERIAL
0.278x5 0.222x 2
0.222x5 0.278x 2
0.946 2
ˆ XX XX1
1
̂ 2 =0.946, ̂ 2 =-0.554
0.554 3
X1 ˆ 2 X2 ˆ 3 X3 ˆ 1
3-.946X3+.554X3= ̂1
3-.946X3+.554X3= ̂1
̂1 =1.824
Regression line of X1 on X2 and X3 is
X1=1.824+0.946X2-0.554X3.
or we can write it in another way; on multiplying both the sides of eq. by18
18X1=32.83+17.028X2-9.97X3.
or 18X1=33+17X2-10X3
or 18X1=17X2+10X3=33. (proved)
b) Test of significance
1 X̂1
e X1 X̂1 e2
1 2.054 -1.054 1.110
2 2.216 -0.216 0.0467
3 4.338 -1.338 1.790
4 1.838 +2.162 4.674
5 3.554 +1.446 2.091
e 2
=7.875
ˆ 2 2
Applying the „t‟ test- t a ii 2 =hypothetical parameter , e= error term,
e2
nk
k= no.ofparameters , aii= ith diagonal element in XX
1
Putting the values in the above formula
0.946 0.946
t X 0.278 1.108 degrees of freedom 5-3=2
7.875 0.854
52
t- tab at 5% level of significance for 2 degrees of freedom =2.920 and t.cal.=1.108
t-cal.< t-tab. Thus accept the hypothesis 2 =0. Thus we shall accept the null
hypothesis H0( 2 =0) i.e There is no relationship between X1 and X2.
46
BASIC ECONOMETRICS STUDY E MATERIAL
Illustration:4 The following table shows the weights (X1) to the nearest pound, heights (X2)
to the nearest inch and ages (X3) to the nearest year of 12 boys:-
Weight (X1) Height (X2) Age (X3)
64 57 8
71 59 10
53 49 6
67 62 11
55 51 8
58 50 7
77 55 10
57 48 9
56 52 10
51 42 6
76 61 12
68 57 9
Estimate the least squares regression line to predict the weight of a boy of given height
and age.
Solution: Let the regression line of X1on X2 and X3 be X1 1 2 X 2 3X3
To determine the values of parameters I e. 1 2 and 3 we shall construct the following
table-
X1 X2 X3 X1X 2 1X2 X2X3 X12 X 22 X 32
64 57 8 3648 512 456 4096 3249 64
71 59 10 4189 710 590 5041 3481 100
53 49 6 2597 318 294 2809 2401 36
67 62 11 4145 737 682 4489 3844 121
55 51 8 2850 440 408 3025 2601 64
58 50 7 2900 406 350 3364 2500 49
77 55 10 4235 770 550 5929 3025 100
57 48 9 2736 513 432 3249 2304 81
56 52 10 2912 560 520 3136 2704 100
51 42 6 2142 306 252 2601 1764 36
76 61 12 4636 912 732 5776 3721 144
68 57 9 3876 612 5133 4624 3249 81
753 643 106 40830 6796 5779 48139 34843 976
753 643 106
X1 62.75, X 2 53.58, X3 8.83
12 15 12
Setting of value to the actual mean-
x x X / n 34843 643
2
/ 12 388.92
2 2 2
2 2 2
47
BASIC ECONOMETRICS STUDY E MATERIAL
x x X / n 976 106
2
/ 12 39.67
2 2 2
3 3 3
x x x x X . X / n 40830 753X643 / 12 481.75
1 2 1 2 1 2
x x x x X . X / n 6796 753X106 / 12 144.5
1 3 1 3 1 3
x x x x X . X / n 5779 643X106 / 12 99.17
2 3 2 3 2 3
on applying the formula
ˆ 2
x x . x x x . x x = (481.75x39.67) (144.5 99.17)
1 2
2
3 1 3 2 3
x . x x x 388.92x39.67 99.17
2 2 2 2
2 3 2 3
19111.02 14330.06 4780.96
0.85
15428.46 9834.69 5593.77
ˆ 3
x x . x x x . x x = 44.5x388.92 481.75x99.17
1 3
2
2 1 2 2 3
x . x x x 388.92x39.67 99.17
2 2 2 2
2 3 2 3
56198.94 47775.15 8423.79
= 1.51
15428.46 9834.69 5593.77
Now ˆ X1 ˆ X 2 ˆ X3
1 2 3
=62.75-(0.85X53.58)-(1.51X8.83)
=62.75-45.54+13.33=3.88 Thus regression line is X1=3.88+0.85X2+1.51X3.
Illustration: 5 From the following data compute the regression line of X1 on X2 and 3.
Year 2011 2012 2013 2014 2015 2016 2017 2018 2019
X1 100 106 107 120 110 116 123 133 137
X2 100 104 106 111 111 115 120 124 126
X3 100 99 110 126 113 103 102 103 98
Where X1= Index of imports of goods and services to U.S.A at constant (2000) prices.
X2=Index of gross U.S.A product at 2000 prices.
X3= Ratio of indices of prices of imports and general U.S.A output respectively.
Solution: Let the estimated regression line of X1 onX2 and X3 be
X1 1 2 X 2 3X3 From the above table firstly we shall compute the mean
n=9 X 1 =1052, X 2 =1017, X 3 =954
X1
X 1052 116.9 ; X 2
1017
113 ; X 3
954
106
n 9 9 9
48
BASIC ECONOMETRICS STUDY E MATERIAL
X X =119,750,
1 2 X1X3=111.433 X X2 3 =107, 690
X =124,288
2
1 X 2
2 =115,571, X 2
3 =101,772
Now we shall compute in terms of deviation from actual mean
x x x1x 2
X . X
1 2
119750
1052x1017
874
1 2
n 9
x x x1x 3
X1. X3 111433
1052x954
79
1 3
n 9
x x X X n 107690 9 112
X. X 1017 x954
2 3
2 3 2 3
X
x x n 124228 9 126089
2
2 2 1052x1052
1
1 1
X
x x n 115571 9
2
1017 x1017
2
2
2
2
2
650
X
x x n 101772 9 648
2
2 2 954x954
3
3 3
x 2
We know that ˆ XX XX1
1
where, X= [x2 x3] and X
x3
x 2 x 22 x 2 x 3 650 112
Now X X x 2 x 3
x3 x 2 x 3 x 32 112 648
650 112
XX 650X648 112X112 408656
112 648
Cofactor of 650=648 since sign for cofactor is (-1)i+j
Cofactor of -112=112 where i=no. of rows,
Cofactor of -112=112 j=no. of columns
Cofactor of 648=650
648 112
Adjoint of XX
Adjo int
Since, Inverse =
112 650 Deter min ant
648 112
648 112 408656 408656
XX 1 1
408656 112 650 112 650
408656 408656
(Since [A]=[ A]where, A is a matrix and is scalar)
XX 1
0.00158 0.00027 x x x
XX 1 2 X1 1 2
874
0.00027 0.00159 , x3 x1x 3 79
49
BASIC ECONOMETRICS STUDY E MATERIAL
0.00158 0.00027 874
ˆ XX 1 XX1
0.00027 0.000159 79
0.00158X874 0.00027X 79 1.364 ˆ 2
= = = ̂ 2 =1.364 ̂3 =0.113
0.00027X874 0.00159X 79 0.114 ˆ 3
Now ˆ X1 ˆ X 2 ˆ X3 =116.9-1.364x113-0.114x106=-49.33
1 2 3
The estimated regression line becomes X1=-49.33+1.364X2+0.114X2.
3.9. Let us sum up
If we have several repressors in a regression model, how do we estimate and what are
the assumptions of that model, properties of estimators such as linear, unbiased, minimum
variance, explained followed by goodness of fit of model with solved numerical problems.
3.10 Unit -End Exercises
A. Multiple Choice Questions
1. The term Multiple regression Stands for
A. Regressing more than one explanatory variables
B. Regressing no variables
C. Many regression
D. Regressing one explanatory variable
2. An estimator is consistent if..?
A. It converges to the true value as the sample size remains same.
B. It converges to the true value as the sample size gets smaller.
C. It converges to the true value as the sample size gets larger.
D. Above all.
3. The value of 𝛽0 is
A. ˆ0 y 1 x ˆ y ˆ x
B. 0 C. . Zero D,. One
4. Regression model in which more than one independent variable is used to predict the
dependent variable is called
A. a simple linear regression model C. a multiple regression model
B. an independent model D. none of the above
5. The Value of 𝛽 estimator is
A. ˆ XX XY
1
ˆ X X X X
1
B.
ˆ X Y X X
1
C.
ˆ X Y Y 'Y
1
D.
50
BASIC ECONOMETRICS STUDY E MATERIAL
B. Short Answer and Essay type Questions
1. What is Multiple Regression?
2. Describe the method of estimation of unknown parameter with matrix.
3. Derive the β value in terms of matrix as (X‟X)-1X‟Y
4. Enumerate the properties of linear, unbiased and minimum variance of estimator.
5. What do meant by goodness of fit? Explain it.
3.11. Reference Books
1. DamodarN.Gujarati and Sangeetha, “ Basic Econometrics” Special Indian Edition , Tata
McGraw Hill Education Privated Limited (Sixth Print 2010) , ISBN: 978-0-07-066005-2
2. P.G.Apte,” Text book of Econometrics” Tata McGraw – Hill Publishing Company
Limited.
3. Dhanasekaran,” Econometrics” 2nd Edition, Vrinda Publications (P) Ltd, Delhi-53 ,2011
ISBN: 978-81-8281-388-5
4. HumbertoBarreto and Frank M. Howland,” Introductory Econometrics” Cambridge
University Press, First South Asian Edition 2009, ISBN: 978-0-521-12358-9.
5. Dilip M. Nachane,” Econometrics: Theoretical Foundations and Empirical
Perspectives” Oxford University Press, Second Impression 2010 , ISBN: 978-0-19-
564790-7
6. S.Shyamala,RavdeepKaur, Arul Pragasam,” A text book on Econometrics: Theory and
Applications” Vishal Publishing Co., Jalandhar 2017, ISBN: 81-88-646-98-9
7. S.P.Singh, Anil.K. Parashar, H.P. Singh,” Econometrics and Mathematical Economics”
Second Revised Editions,S.Chand and Company Ltd, New Delhi -55
8. A. Koutsoyiannis,” Theory of Econometrics” Second Edition Palgrave – New
York,2004,ISBN: 0-333-77822-7
9. Maddala .G.S.(1997), “Econometrics” , McGraw Hill, New York.
10. Johnston. (1997),” Econometric Methods” McGraw Hill, 4th Edition, New Delhi.
51
BASIC ECONOMETRICS STUDY E MATERIAL
UNIT 4
Structure
4.1 Objectives
4.2 Introduction
4.3 Violation of OLS assumptions
4.4 Multicollinearity
4.4.1 Meaning and types
4.4.2 Causes, Consequences
4.4.3 Deduction and Remedial Measures
4.5 Heteroscedasticity
4.5.1 Meaning
4.5.2 Causes, Consequences
4.5.3 Deduction and Remedial Measures
4.6 Autocorrelation
4.6.1 Meaning
4.6.2 Causes, Consequences
4.6.3 Deduction and Remedial Measures
4.7 Specifications
4.7.1 Meaning, Reasons and types
4.7.2 Causes, Consequences, Tests
4.8 Let us sum up
4.9 Unit End Exercise
4.10 Reference Books
4.1 Objectives
After going through the unit you will be able to:
Understand the concepts and issues of violation of assumptions
Sense the causes and consequences of violation of assumptions.
Know the method of detection and remedies for violation of assumptions.
4.2 Introduction
The econometric models are constructed by introducing the random variable '𝑈𝑖 ' for to
take into account of influence of various errors, such as (a) errors of omitted variable (b)
errors of the mathematical form of the model (c) errors of measurement of the dependent
variable and (d) the effects of the erratic element which is inherent in human behaviour.
52
BASIC ECONOMETRICS STUDY E MATERIAL
We studied the role of random variable '𝑈𝑖 ‟ and the reason for it is introduced into model
in unit one and in unit two, under assumptions. Further to get valid and representative
results one must be familiar with the expected consequences from non-fulfillment of an
assumption. Hence, this chapter, we will discuss the causes, consequences, detection and
remedies which are to be made if any one of the basic assumption is violated.
4.3 Violations of Assumptions
There are ten assumptions framed for execution of classical linear regression model
under Ordinary Least Square method of estimation. In that the assumptions related with
multicollineariy, Heteroscedasticity, Autocorrelation and Specification error are important
one for getting the consistent and efficient parameters of intercept(α) Slope(β) and
stochastic error term(𝑈𝑖 .
4.4.Multicollinearity
The classical linear regression model (CLRM) assumes that there is no multicollinearity
among the regressors included in the regression model.
4.4.1. Meaning:
Multicollinearity refers to the existence of more than one exact linear relationship, and
collinearity refers to the existence of a single linear relationship. Originally,
Multicollinearity meant that the existence of a “perfect” or exact, linear relationship among
some or all explanatory variables of a regression model.
In Classical linear regression model assume that there is no multicollinearity, among
the explanatory variables (Xs). The reasoning is this: if multicollinearity is perfect. Then
the regression coefficients of explanatory variables are in determinate and their standard
errors are determinate and their standard errors are infinite. If multicollinearity is less than
perfect, the regression coefficients possess large standard errors, which means the
coefficients cannot be estimated with great precision of accuracy
53
BASIC ECONOMETRICS STUDY E MATERIAL
Types of Multicollinearity
The types of multicollinearity are of four types as follows:
1. High Multicollinearity: It signifies a high or strong correlation between two or more
independent variables, but not a perfect one.
2. Perfect Multicollinearity: This degree of collinearity indicates an exact linear
relationship between two or more independent variables.
3. Data-based Multicollinearity: The possibility of collinearity, in this case, arises out
of the selected dataset.
4. Structural Multicollinearity: This issue arises when researchers have a poorly
designed framework for the regression analysis.
4.4.2 Causes of Multicollinearity
Multicollinearity may be due to the following factors:
1) The data collection method employed
For example, Sampling over a limited range of values taken by the regressors in the
population.
2) Constraints on the model or in the population being sampled.
For example in the regression of electric energy consumption on income and house sine
here is a physical constraint in the population in that families with higher incomes
generally have larger homes than families in lower is comes.
3) Model Specification
For example adding polynomial terms to a regression model, especially when the range
of the „X‟-variable is small.
4) An over determined model
This happens when the model has more explanatory variables than the number of
observations. This would happen in medical research, health economics where there
may be a small number of patients about whom information is collected on large no of
variables.
Consequence ofMulticollinearity.
The main consequences of multicollinearity are the following
1) The precision of estimation falls, so that it becomes very difficult, to separate the
relative influence of various X variables This loss of precision has three aspects;
(a) Specific estimates may have very large errors,
(b) These errors may be highly correlated and
(c) the sampling variances of the coefficients may be very large
2) Investigators are sometimes led to door variables in correctly from an analysis because
their coefficients are not significantly different from zero but the true situations may be
that a variable has no effect but Simply because the set of sample data has not enabled
as to pick it up.
3) Estimates of coefficients become very sensitive to particular set of sample data and the
addition of a few more observations can sometimes produce dramatic shifts in some of
the coefficients.
54
BASIC ECONOMETRICS STUDY E MATERIAL
4.4.3. Detection of Multicollinearity
Here the question arises as, How does you know that collinearity is present in any
given models involving more than two explanatory variables?
The answer is as follows as:
1) High R2 value but few significant„t‟ ratios which is a classic symptom of presence
of Multicollinearity
2) There exist a high zero order correlation coefficient between two regressors or High
pair wise correlation among regressors, then multi collinearly o existing and a
serious problem too.
3) The examination of partial correlation coefficients may suggest that the explanatory
variables are intercorrelated or not. But this method was criticized by John „O‟
Hagan and Brendan McCabe.
4) The presence of Multicollinearity can be detected by using Auxilliary regression to
the main regression of Y on the X‟s.
If the computed f exceeds the critical F; at the chosen level of significance, it is taken
to mean that the particular X is collinear with other X's. If it does not exceed the
critical Fi, then there is no collinear with other X‟s.
5) The Eigenvalues and Condition index aids to diagnose multicollinearity. Condition
Index is the ratio of the square root of maximum eignvalue with respect to
minimum eigen value. If the condition index is between 10 and 30, there is
moderate to strong multicollinearity, and if it exceeds so there is severe
multicollinearity.
6) The larger the value of variance inflation factors VIF the more collinear the variable
X‟s. The tolerance value is closer to zero, the greater the degrees of collinearity of
that variable with the other regressors. If the tolerance (TOL) is to I, then there no
collinearity between Xj with other regressor
A variance inflation factor (VIF) is a measure of the amount of multicollinearity in
regression analysis. Multicollinearity exists when there is a correlation between multiple
independent variables in a multiple regression model. This can adversely affect the
regression results.
Small VIF values, VIF < 3, indicate low correlation among variables under ideal
conditions. The default VIF cutoff value is 5; only variables with a VIF less than 5 will be
included in the model. However, note that many sources say that a VIF of less than 10 is
acceptable.
Remedial measures for Multicollinearity
The multicollinearity can be solved by two options
i) Do nothing or
ii) Follow some rules of thumb. The rules of thumb are as follows:
a) Using extraneous or pool information,
b) Combining cross sectional and time series data
55
BASIC ECONOMETRICS STUDY E MATERIAL
c) Omitting a highly collinear variable,
d) Transforming data and
e) Obtaining additional or new data
4.5 Heterocedasticity
The classical linear regression model assumes that, the variance of each disturbance
term U; is same for all observations. That o the conditional variances of U; are identical
U
symbolically Var i = where var stands for variance. This is the assumption of
2
Xi
Hemoscedasticily, or equal variance. If this assumption is violated, heteroscedasticity
arises
4.5.1 Meaning
Heteroscedasticity refers to for given vales of X's the variance of each disturbance term
Ui is not constant number equal to
2
Ui
var 2
Xi
4.5.2 Causes Heteroscedasticity
The causes of variances of Ui may be variable are as follows:
1) The error of learning models, as people - learn their errors of behaviour become
smaller over time and is expected to decrease
2) The existence of discretionary of human behaviour is an another cause of variance
of Ui may be variable.
3) The method of data collection and data processing is as a cruising factors a variance
varied nature
4) The presence of outliers or an outlying e observation in relation to the observation
in the sample alter the results of regression analysis and causes the
heteroscedasticity.
5) The misspecification of model is also a reason for heteroscedasticity
6) Hetetoscedasticity arises due to the skewness is the distribution of one or more
regressors included in the model.
7) Incorrect data transformation and incorrect functional form are sources of
hetroscedasticity.
Consequence ofHetroscedasticity
If the assumptions of homoscedastic, disturbance is not fulfilled, we have the following
consequences:
1) We cannot apply the formula of the variances of the coefficients to conduct tests of
significance and construct confidence intervals. The tests are replicable.
56
BASIC ECONOMETRICS STUDY E MATERIAL
2) If is heteroscedastic, the OLS estimate do not have the minimum variance
property in the Class of unbiased estimators, that is they are inefficient in small
samples and large samples.
3) The presence of heteroscedasticitythat is the variance of the “U”; “S” are not
constant does not require for unbiasaedness of the least square estimates. It means
that the coefficients estimates would be statistically significant even in presence of
heteroscedasticity.
4) The prediction of Y for given value of X based on the estimates from the original
data would be have a high variance, that is the prediction would be inefficient.
Because the variance of the prediction includes the variances of 'U' and of the
parameter estimates, which are not minimal due to the incidence of
heteroscedasticity.
4.5.3. Detecting Hetéroscedasticity
Various tests have been suggested for establishing homoscedasticity. Here in following
pages, the test which are conceptually and computationally simple one to apply are
presented.
1. Park Test: Park
ti
Suggested a functional form as i xX i e (or)
2 2
taking log in both sides
log 2 log 2 log X;Vi
where V is the Stochastic disturbance term. Since is generally not known park suggests
2
using U2 as a proxy and running the following regression logU2=logX2+ log Xi+ ;
If turns out to be statistically significant, it would be suggest that heteroscedasticity is
present in the data.
2. The Spearman rank correlation test
This is the simplest test, which may be applied for small or large samples. The Steps
are as follows.
1. Regress Y on X .
Y X1 U and obtain the residuals, ê ‟s which are estimates of the U's .
2. Arrange the x values, and the e‟s in ascending or descending order and compute the
rank correlation Coefficient
6 Di2
re.x 1 where
n (n 2 1)
DI = difference between the ranks corresponding pairs of X and e.
57
BASIC ECONOMETRICS STUDY E MATERIAL
n = observation in the sample.
3. The result of high rank correlation
Coefficient is high suggests that the presence of heterscedasticity, otherwise if the rank
correlation coefficient is low means the presence of homoscedasticity
3. Gold feld - Quant Tests
The regression model
Yi Xi Ui Then
Step-1 Order or rank the observations according to to the values Xi, beginning with the
lowest Value „X‟
Step. 2 Omit „c‟ central observations, where 'c' is specified a priori, and divide the
remaining (n-c) observations into two groups each of (n-c)/2 Observations.
Step.3 Fit separate, OLS regressions to me first (n-c)/2 Observations and the last (n-
c)/2 observations, and obtain the respective residual sum of squares RSS1 , and.
RSS 2
Residual sum square ( RSS1 ) corresponding to smaller X i values and RSS 2 is
from the larger X i values.
(n 2)
The degree of freedom is k where „k‟ is the number of parameters to
2
be estimated including the intercept.
Step 4 Compute the ratio
RSS 2 / df
Where df Stands for degrees of freedom.
RSS1 / df
The F , so the computed is greater than Critical F at the chosen level of
significance Then we can say that heteroscedasticity is existing in model estimation
process.
Remedial measures
The presence of heteroscedasticity does not destroy the unbiasedness and consistence
properties of ous estimators, but They are no longer efficient, and not even asymptotically.
Therefore remedial measures needed to solve the problem of heteroscedasticity
There are two approaches to remediation: When i is known and when i is not
2 2
known.
(a) When i is known, then to correct the problem of heterosedasticity is by means of
2
Weighted least squares, for the estimators thus obtained are BLUE.
(b) When i is not known, then use the data transformation method(1) based on the
2
assumption of the error covariance is proportional to X i (iii) the error variance is
proportional to the Square of the mean values of Y. (iv) log transformation Which
Compresses the scales in which the Variables are measured ..
58
BASIC ECONOMETRICS STUDY E MATERIAL
To conclude the above discussion of the remedial measures which of the
Transformation discussed will work depends on the nature of the problem and severity
ofheteroscedasticity.
4.6 Auto correlation meaning
Auto correlation is also called „Serial correlation‟. Auto, which means self, signifies that
a series is correlated with itself. Auto correlation refers to the relationship between
successive error terms.
But the classical linear regression model assumption is that successive disturbance
terms are drawn at random that is
E Ui U j =0 for i j = 1,2,3 ……n.
It implies that when observations are made over time the effect of the disturbance to
occuring in one period does not carry over to another period. It means no Auto correlation.
4.6.1. Causes Auto Correlation
The causes of autocorrelation are as follows:
1) Inertia or sluggishness of economic time series leads to successive observations are
likely to be interdependent.
2) Misspecification of the relationship or
3) Excluding important variables, functional form.
4) Contains errors of measurement.
5) The nature of Cobweb phenomena means that decision making at a variable
depends upon its past and data messaging and data transformation.
Consequences of Auto Correlation
The U is auto correlated, and then the following Consequences will be as follows as
(a) The presence of auto correlation the OLS estimators remain unbiased, Consistent
and asymptotically normally distributed, but they are no longer efficient.
(b) If U's are autocorrected, then the predictions based on OLs estimators will be
inefficient
(c) The confidence intervals are unnecessarily widen and the tests of significance „E‟
and „F‟ and X are no longer valid and if applied are likely to give seriously
misleading conclusions about the statistical significance of the estimated regression
coefficients.
4.6.2. Detection and Remedies.
There are formal and informal methods of detecting the presence of
autocorrelation.Among the informal methods one can simply plot the actual or
standardized residuals, or plot current residuals against past residuals. In formal methods,
one can use the runs test, Durbin-Watson‟s‟ test, asymptotic normality test, Berenblutt -
Webb test, and Boot Breusch Godfrey (BG) test. Durbin Watson„d‟ test is commonly used
with its limitations. So, it is better to use Breusch Godfrey test.
59
BASIC ECONOMETRICS STUDY E MATERIAL
The Remedial measures are based on the knowledge. One has about the nature of
interdependence among the disturbances that is knowledge about the structure of
autocorrelation. Then the remedial tan measures can be grouped as when to know and
is not known. The problem of autocorrelation can be removed by Markov first order
autoregressive scheme, known as AR(I) Scheme. When is not known. This scheme
assumes that the disturbance in the current time period is linearly related to the
disturbance term in the precious time period, the coefficient of autocorrelation e providing
the extent of the autocorrelation of providing the extent of the interdependence.
If the value of l is known, then the problem of autocorrelation can be removed by using
Durbin - Watson a Theil- Nagar Modified and Cochrane - Orcutt (C-O) iterative procedure.
4.7. Specification Error
Specification of a regression refers to formulate the regression equation. Specification of
a model is the logical idea of economic theory. But due to mishandling or inadequate
knowledge of economic fact and economic theories, the errors occur due to this
misspecification is known as "Specification Error?
4.7.1. Meaning
In simple words, Specification error means the error that occur because of mistake in
variables inclusion or exclusion or assumption of the model.
Reasons for Specification errors
Specification error is common and intractable problem in economics. In reality, any
economic variable is determined by a number of factors and all of which will not be
included in regression analysis. The inclusion of large number of explanatory variables will
of course reduce the number of degrees of freedom in the analysis, and making the
estimates of the parameters in precise.
Some of the explanatory variables may not be quantifiable and there for difficult to
incorporate into numerical analysis.
Some of the variables may be or rifted by mistake because their relevance is
unrecognized, or may be the lack of knowledge of researcher. Hence the error of
specification arises due to:
a) Omission of a relevant explanatory variable (S)
b) Inclusion of an irrelevant explanatory variable (S)
c) Discarding of a qualitative change in one of the explanatory variables and
d) Incorrect mathematical form of the regression equation.
Types of Specification error
A regression model will have a specification error when at least one of the following
problem occur in that model
1. Inclusion of irrelevant explanatory variable
2. Omission of relevant explanatory variable and
3. Incorrect functional form
60
BASIC ECONOMETRICS STUDY E MATERIAL
4.7.2 Consequences
The inclusion of irrelevant variable (S) does not affect the relationship between other
variables and the dependent variable,Because, the estimator for such a variable turns out to
be zero. The estimates of inclusion of irrelevant variable in a model are unbiased and
consistent. However the estimates are not efficient because of the variance are larger than
they would have been in the model excluding the irrelevant variable. Further the
model estimators violate the properties of „BLUE‟, the concept of regression because the
estimators are inefficient. If the specification error is due to qualitative change in one or
more explanatory variables, then also, the estimations will be biased.
Another sort of specification error arises when the functional relationship is incorrect.
The magnitude of bias will depend upon the size of coefficients. Thus the estimated
parameters will be biased if we calculate the parameters without taking into account of the
errors committed.
Comparison of specification error of exclusion/ omission of relevant explanatory
variable and inclusion of irrelevant variable.
Category of Omission\
S.No Inclusion model
information Exclusion model
Estimation
l Biased unbiased.
of Coefficient
2 Efficiency Generally declines Declines
Estimation of
3 over estimate unbiased.
disturbance term
Convention test of hypothesis and invalid and faculty valid through
4
confidence region inferences erroneous
Tests of specification error
To detect equation Specification errors, there are several test are used. The prime tests
are (a) examination of residuals (b) the Durbin Watson d statistic (c)Ramsey's RESET test
and (d) the Lagranage multiplier test
4.8 Let us Sum up.
If the assumptions of the Classical linear regression model that the errors term or
disturbance term „ U i ‟
(a) (i) entering into the population regression function (PRF) are random or
uncorrelated, (autocorrelation) (ii) have all the same variance (homoscedasticity) and
2
(b) There is no linear relationship among the explanatory variables X‟s (multcolltinearity)
are violated causes of violation what will happen in estimation? How to identify the factors
influencing for violations of assumption? and what are the remedial measures to solve the
issue of violation of assumption were studied in a descriptive way not in empirical way.
61
BASIC ECONOMETRICS STUDY E MATERIAL
4.9. Unit End Exercises
A. Multiple Choice Questions (MCQ)
1. Find out which is not the violation of assumption .
A. Autocorrelation C. Multicollinearity
B. Heteroscedasticity D. Dummy variable
2. Homoscedasticity means that
A. Var(Ui/Xi) =σ2 B. Var(Ui) = 0 C. Var(Ui) =1 D. Var(Ui) =∞
3. Which of the following pair is not correctly matched?
A. Dicky- Fuller test – Hetroscedasticity
B. Durbin‟s h test – Autocorrelation in autoregressive models
C. F test – Overall significance of the regression model
D. Distributed lag models – Koyck approach
4. Park Test is used for what purpose?
A. Detecting Hetroscedasticity
B. Solving Hetroscedasticity
C. Detecting Multi-collinearity
D. Solving Multi-collinearity
5. E (Ui,Uj) ≠ 0, when i≠j is termed as,
A. Auto-Correlation B. Hetroscedasticity
C. Multi-collinearity D. Homoscedasticity
6. The IV estimator can be used to potentially eliminate bias resulting from
A. Multicollinearity B. serial correlation
C. errors in variables D. heteroskedasticity
7. Variance Inflation Factor is used for…
A. Detecting Hetroscedasticity B. Solving Hetroscedasticity
C. Detecting Multi-collinearity D. Solving Multi-collinearity
8. If the value of Durbin-Watson‟s d statistic = 0, there is………
A. No Auto-correlation B. Positive Auto-correlation
C. Negative Auto-correlation D. None of these
9. The assumption of homoscedasticity was expressed as……….
A. E(𝑈𝑖 ) 2= σ2𝑖 B. E(Ui)2 = σ 2 C. E(Ui)2 = 0 D. E(Ui) = 0
10. Which of the following is a multi-collinearity diagnostic?
A. Condition Index. B. Park test.
C. Glejsertes. D. Durbin's h test.
11. Which of the following is used to detect specification errors?
A. The Park test. B. Ramsey's RESET test.
C. Chow test. D The Runs test.
12. What is the meaning of the term 'heteroscedasticity'?
A. The variance of the errors is not constant.
B. The variance of the dependent variable is not constant.
62
BASIC ECONOMETRICS STUDY E MATERIAL
C. The errors are not linearly independent of one another.
D. The errors have non-zero mean.
13. When one or more of the regressors are linear combinations of the other regressors, it is
called --------------.
A. Autocorrelation. B Heteroscedastity.
C. Multicollinearity. D. Serial correlation.
14. As a rule of thumb, a variable is said to be highly collinear if the Variance Inflation
Factor (VIF) is ------------.
A. Exactly 10. B Exceeds 10. C. Less than 10. D. None of the above.
15. Specification bias or specification error means
A. Leaving out important explanatory variables
B. Including unnecessary variables
C. Choosing the wrong functional form between Y and X variables
D. All of the above
Part B. Short Answers and Essay type Questions
1. Elucidate the causes, consequences and remedies of Autocorrelation.
2. Enumerate the causes, consequences and remedies of Multicollinearity.
3. Explicate the causes and consequences of Heteroscdasticiy.
4. What are diagnostic test used for Autocorrelation, Mulitcolloinerity, and
Hetroscedasticity?
5. What is specification error? Compare the effects of exclusion of relevant variable with
inclusion of irrelevant variable.
4.10 Reference Books
1. DamodarN.Gujarati and Sangeetha, “ Basic Econometrics” Special Indian Edition , Tata
McGraw Hill Education Privated Limited (Sixth Print 2010) , ISBN: 978-0-07-066005-2
2. P.G.Apte,” Text book of Econometrics” Tata McGraw – Hill Publishing Company
Limited.
3. Dhanasekaran,” Econometrics” 2nd Edition, Vrinda Publications (P) Ltd, Delhi-53 ,2011
ISBN: 978-81-8281-388-5
4. HumbertoBarreto and Frank M. Howland,” Introductory Econometrics” Cambridge
University Press, First South Asian Edition 2009, ISBN: 978-0-521-12358-9.
5. Dilip M. Nachane,” Econometrics: Theoretical Foundations and Empirical
Perspectives” Oxford University Press, Second Impression 2010 , ISBN: 978-0-19-
564790-7
6. S.Shyamala,RavdeepKaur, Arul Pragasam,” A text book on Econometrics: Theory and
Applications” Vishal Publishing Co., Jalandhar 2017, ISBN: 81-88-646-98-9
7. S.P.Singh, Anil.K. Parashar, H.P. Singh,” Econometrics and Mathematical Economics”
Second Revised Editions,S.Chand and Company Ltd, New Delhi -55
63
BASIC ECONOMETRICS STUDY E MATERIAL
8. A. Koutsoyiannis,” Theory of Econometrics” Second Edition Palgrave – New
York,2004,ISBN: 0-333-77822-7
9. Maddala .G.S.(1997), “Econometrics” , McGraw Hill, New York.
10. Johnston. (1997),” Econometric Methods” McGraw Hill, 4th Edition, New Delhi.
64
BASIC ECONOMETRICS STUDY E MATERIAL
UNIT V
5.1 Objectives
5.2 Introduction
5.3 Lag and Reasons for introducing Lag
5.4 DL, AR, MA
5.5. Adhoc Estimation drawbacks
5.6 Koyck approach and feature
5.7 Dummy variable
5.7.1 Meaning of Dummy variable
5.7.2 Nature of Dummy variable
5.7.3 Types of Dummy variable
5.7.4. Caution in use of Dummy variable
5.8 ANOVA and ANCOVA
5.8.1Meaning ANOVA
5.8.2. Types of ANOVA
5.8.3 Advantages and Disadvantages of ANOVA
5.8.4 Meaning of ANCOVA
5.8.5. Assumptions of ANCOVA
5.8.6 Advantages and Disadvantages of ANCOVA
5.8.7 Comparison of ANOVA and ANCOVA
5.9 Regression on qualitative dependent variables
5.10 Let us sum up
5.11 Unit End Exercise
5.12 Reference Books
5.1. Objectives
To learn and apply the knowledge in real data set by construction of econometric
modeling with dummies
To understand the lag and its reason for introduction in analysis
To study the ANOVA and ANCOVA
To learn regression on qualitative independent variables and qualitative
dependent variables
65
BASIC ECONOMETRICS STUDY E MATERIAL
5.2 Introduction
Econometric researches incorporate many economic variables. Some of them are
quantifiable or measurable while some variables are qualitative and hence are not
measurable directly. In general, the explanatory variables in any regression analysis are
assumed to be quantitative in nature. For example, the variables like temperature, distance,
age etc. are quantitative in the sense that they are recorded on a well-defined scale. But in
reality all variables in an economic activity may not be measureable, in such qualitative
cases of variables, this unit knowledge will be used for finding the qualitative variables
influence on quantitative and qualitative variable. In this unit, the readers may get the
knowledge of lag variable usage for to study the implications of past period and shocks.
Further reader learns about the qualitative variables usage as independent and dependent
in a regression analysis as dummies.
5.3 Lag and Reasons for Lag
In economics the dependence of a variable Y (the dependent variable) on another
variable(s) X (the explanatory variable) is rarely instantaneous. Very often, Y responds to X
with a lapse of time. Such a lapse of time is called a lag. There are three main reasons for
lag:
1. Psychological reasons. As a result of the force of habit (inertia), people do not
change their consumption habits immediately following a price decrease or an
income increase perhaps because the process of change may involve some
immediate disutility. Thus, those who become instant millionaires by winning
lotteries may not change the life styles to which they were accustomed for a long
time because they may not know how to react to such a windfall gain immediately.
Of course, given reasonable time, they may learn to live with their newly acquired
fortune. Also, people may not know whether a change is “permanent‟‟ or
“transitory.‟‟ Thus, my reaction to an increase in my income will depend on
whether or not the increase is permanent. If it is only a nonrecurring increase and in
succeeding periods my income returns to its previous level, I may save the entire
increase, whereas someone else in my position might decide to “live it up.‟‟
2. Technological reasons. Suppose the price of capital relative to labor declines,
making substitution of capital for labor economically feasible. Of course, addition
of capital takes time (the gestation period). Moreover, if the drop in price is
expected to be temporary, firms may not rush to substitute capital for labor,
especially if they expect that after the temporary drop the price of capital may
increase beyond its previous level. Sometimes, imperfect knowledge also accounts
for lags. At present the market for personal computers is glutted with all kinds of
computers with varying features and prices. Moreover, since their introduction in
the late 1970s, the prices of most personal computers have dropped dramatically.
As a result, prospective consumers for the personal computer may hesitate to buy
until they have had time to look into the features and prices of all the competing
66
BASIC ECONOMETRICS STUDY E MATERIAL
brands. Moreover, they may hesitate to buy in the expectation of further decline in
price or innovations.
3. Institutional reasons. These reasons also contribute to lags. For example,
contractual obligations may prevent firms from switching from one source of labor
or raw material to another. As another example, those who have placed funds in
long-term savings accounts for fixed durations such as one year, three years, or
seven years are essentially “locked in‟‟ even though money market conditions may
be such that higher yields are available else where. Similarly, employers often give
their employees a choice among several health insurance plans, but once a choice is
made, an employee may not switch to another plan for at least one year. Although
this may be done for administrative convenience, the employee is locked in for one
year
For psychological, technological, and institutional reasons, a regress and may respond to a
regressor(s) with a time lag. Regression models that take into account time lags are known
as dynamic or lagged regression models. There are two types of lagged models:
distributed-lag and autoregressive. In the former, the current and lagged values of
regressors are explanatory variables. In the latter, the lagged value(s) of the regress and
appears as an explanatory variable(s).
5.4 Distributed Lag Model (DL)
In regression analysis involving time series data, if the regression model includes not
only the current but also the lagged (past) values of the explanatory variables (the X‟s), it is
called a distributed-lag model. Thus, Yt= α + β0Xt + β1Xt−1 + β2Xt−2 + ut represents a
distributed-lag model
Autoregressive Model (AR)
If the model includes one or more lagged values of the dependent variable among its
explanatory variables, it is called an autoregressive model.
Yt= α + βXt+ γ Yt−1 + ut
is an example of an autoregressive model. It also known as dynamic models since they
portray the time path of the dependent variable in relation to its past value(s).
Estimation of Distributed Lag Model (DL)
A purely distributed-lag model can be estimated by OLS, but in that case there is the
problem of multi collinearity since successive lagged values of a regress or tend to be
correlated. As a result, some shortcut methods have been devised. These include the
Koyck, the Adaptive expectations, and partial adjustment mechanisms, the first being a
purely algebraic approach and the other two being based on economic principles. Aunique
feature of the Koyck, adaptive expectations, and partial adjustment models is that they all
are autoregressive in nature in that the lagged value(s) of the regress and appears as one of
the explanatory variables.
67
BASIC ECONOMETRICS STUDY E MATERIAL
5.5. Adhocestimation suffers from many drawbacks, such as the following:
1. There is no a priori guide as to what is the maximum length of the lag.
2. As one estimates successive lags, there are fewer degrees of freedom left, making
statistical inference somewhat shaky. Economists are not usually that lucky to have a
long series of data so that they can go on estimating numerous lags.
3. More importantly, in economic time series data, successive values (lags) tend to be
highly correlated; hence multi collinearity rears its ugly head. Multi collinearity leads
to imprecise estimation; that is, the standard errors tend to be largein relation to the
estimated coefficients. As a result, based on the routinely computed tratios, one may
tend to declare (erroneously), that a lagged coefficient(s) is statistically insignificant.
4. The sequential search for the lag length opens the researcher to the charge of data
mining.
5. In view of the preceding problems, the adhoc estimation procedure has very little to
recommend it.
5.6 Koyck transformation
Koyck has proposed an ingenious method of estimating distributed-lag models.
Assuming that the β’s are all of thesame sign, Koyck assumes that they decline geometrically
as follows
βk= β0λk where k = 0, 1, . . .Eq. (1)
where λ, such that 0 < λ <1, is known as the rate of decline, or decay, of the distributed lag
and where 1 − λ is known as the speed of adjustment. Eq. (1) postulates is that each
successive β coefficient is numerically less than each preceding β (this statement follows
since λ <1), implying that as one goes back into the distant past, the effect of that lag on Yt
becomes progressively smaller, a quite plausible assumption. After all, current and recent
past incomes are expected to affect current consumption expenditure more heavily than
income in the distant past.
Equation one may be written as
Yt= α + β0Xt + β0λXt−1 + β0λ2Xt−2 + · · ·+ut
Yt−1 = α + β0Xt−1 + β0λXt−2 + β0λ2Xt−3 + · · ·+ut−1
λYt−1 = λα + λβ0Xt−1 + β0λ2Xt−2 + β0λ3Xt−3 + · · ·+λut−1
Yt− λYt−1 = α(1 − λ) + β0Xt + (ut− λut−1)
Yt= α(1 − λ) + β0Xt + λYt−1 + vt
Where vt= (ut− λut−1), a moving average of ut and ut−1.The procedure just described is
known as the Koyck transformation.
The following features of the Koyck transformation:
1. Any One Researcher started with a distributed-lag model but ended up with an
autoregressive model because Yt−1 appears as one of the explanatory variables.
This transformation shows how one can “convert‟‟ a distributed-lag model into an
autoregressive model.
68
BASIC ECONOMETRICS STUDY E MATERIAL
2. The appearance of Yt−1 is likely to create some statistical problems. Yt−1, like Yt, is
stochastic, which means that we have a stochastic explanatory variable in the
model. Recall that the classical least-squares theory is predicated on the assumption
that the explanatory variables either are non stochastic or, if stochastic, are
distributed independently of the stochastic disturbance term. Hence, we must find
out if Yt−1 satisfies this assumption.
3. In the original model the disturbance term was ut, whereas in the transformed
model it is vt= (ut− λut−1). The statistical properties of vt depend on what is
assumed about the statistical properties of ut, for, as shown later, if the original ut‟s
are serially uncorrelated, the vt‟s are serially correlated. Therefore, we may have to
face up to the serial correlation problem in addition to the stochastic explanatory
variable Yt−1.
4. The presence of lagged Y violates one of the assumptions underlying the Durbin–
Watson d test. Therefore, we will have to develop an alternative to test for serial
correlation in the presence of lagged Y. One alternative is the Durbin h test,
Auto regressiveness poses estimation challenges; if the lagged regress and is correlated
with the error term, OLS estimators of such models are not only biased but also are
inconsistent. Bias and inconsistency are the case with the Koyck and the adaptive
expectations models; the partial adjustment model is different in that it can be consistently
estimated by OLS despite the presence of the lagged regress and.
To estimate the Koyck and adaptive expectations models consistently, the most
popular method is the method of instrumental variable. The instrumental variable is
aproxy variable for the lagged regress and but with the property that it is uncorrelated
with the error term.
An alternative to the lagged regression models just discussed is the Almon polynomial
distributed-lag model, which avoids the estimation problems associated with
theautoregressive models. The major problem with the Almon approach, however, is
thatone must prespecifyboth the lag length and the degree of the polynomial. There areboth
formal and informal methods of resolving the choice of the lag length and thedegree of the
polynomial.
5.7 Dummy Variable (qualitative variable as explanatory)
In regression analysis the dependent variable, or regressand, is frequently influenced
notonly by ratio scale variables (e.g., income, output, prices, costs, height, temperature)
butalso by variables that are essentially qualitative, or nominal scale, in nature, such as
sex,race, color, religion, nationality, geographical region, political upheavals, and party
affiliation.For example, holding all other factors constant, female workers are found to earn
lessthan their male counterparts or nonwhite workers are found to earn less than whites.
Thispattern may result from sex or racial discrimination, but whatever the reason,
qualitativevariables such as sex and race seem to influence the regressand and clearly
69
BASIC ECONOMETRICS STUDY E MATERIAL
should beincluded among the explanatory variables, or the regressors.Since such variables
usually indicate the presence or absence of a “quality” or an attribute.
5.7.1 Meaning of Dummy Variable
A dummy variable is a variable that takes values of 0 and 1, where the values indicate
the presence or absence of feature variables. Where a categorical variable has more than
two categories, it can be represented by a set of dummy variables, with one variable for
each category. Numeric variables can also be a dummy coded to explore nonlinear effects.
Dummy variables are also known as indicator variables, design variables, contrasts, one-
hot coding, and binary basis variables.
Technically, dummy variables are dichotomous, quantitative variables; they can take
on any two quantitative values. As a practical matter, regression results are easier to
interpret when dummy variables take on two specific values, 1 or 0. Typically, 1 represents
the presence of a qualitative attribute, and 0 represents the absence.
5.7.2 Nature of Dummy Variables
1. A dummy variable can only take on 2 values (0 or 1), we call the condition in which
the dummy variable is 0 the base condition. Dummy variables are discrete variables
taking a value of „0‟ or „1‟. They are often called „on‟ „off‟ variables, being „on‟ when
they are 1
2. Dummy variables can be used either as explanatory variables or as the dependent
variable. When they act as the dependent variable there are specific problems with
how the regression is interpreted, however when they act as explanatory variables
they can be interpreted in the same way as other variables.
3. The coefficient of the dummy variable represents the difference between being in
the base condition and not being in the base condition.
4. The dummy variable affects the intercept of the regression model, not the slope
5.7.3.Types of Dummy Variables
1. Qualitative dummy variables: i.e. age, sex, race, health.
2. Seasonal dummy variables: depends on the nature of the data, so quarterly data
requires three dummy variables etc.
3. Dummy variables that represent a change in policy:
(a) Intercept dummy variables, that pick up a change in the intercept of the
regression
(b) Slope dummy variables, that pick up a change in the slope of the regression
Broadly speaking this specially designed variable represents the following
effects based on their types of dummy variable utilized in a regression analysis.
(i) Temporal Effects: An investigator sometimes finds that behavioral
relationship shifts from one period to another. As sales receipts of a
shopkeeper might have a tendency to increase during first week of every
month; Government expenditure might be expected to shoe an upward shift
70
BASIC ECONOMETRICS STUDY E MATERIAL
during war period; consumption expenditure might also change during war
period. Similarly, these might be temporary change in relation during the
different seasons, periods or even during different political regimes.
(ii) Spatial Effects: Sometimes economic functions change with a change in
country, economic structure or other regional differences. For example,
consumption functions for U.S.A include some variables but when this
consumption function is applied to the Indian population, necessary
corrections should be made before hand. The reason is that behaviour
pattern of American consumption will certainly be different from that of
their Indian counter-parts. They would also be facing an environment
different from that of Indian consumers. Thus consumption function for
India may include the effects of different economic setting.
(iii) Qualitative Variable‟s Effects: Economic behaviour is also influenced by
qualitative phenomena such as sex, occupation social status, material status
etc. For example, consumption pattern of a newly married couple is bound
to be different from that of an elderly couple. Similarly, the expenditure of
white collared lobourers might be different from that of manual labourers.
Thus these effects must be incorporated in the estimation process.
Effect of all above causes can be incorporated into our regression model by the
specification of appropriate dummy variables. In practice we find several types of models
containing dummy variables.
5.7.4 Caution in the Use of Dummy Variables
Although they are easy to incorporate in the regression models, one must use the
dummy variables carefully. In particular, consider the following aspects
1. If a qualitative variable has m categories, introduce only (m − 1) dummy variables.
If anyone do not follow will fall in the problem of dummy variable trap, that is, the
situation of perfect collinearity or perfect multicollinearity, if there is more than one
exact relationship among the variables For each qualitative regressor, the number
of dummy variables introduced must be one less than the categories of that
variable.
2. The category for which no dummy variable is assigned is known as the base,
benchmark, control, comparison, reference, or omitted category. And all
comparisons are made in relation to the benchmark category.
3. The intercept value (β1) represents the mean value of the benchmark category.
4. The coefficients attached to the dummy variables in a Equation are known as the
differential intercept coefficients because they tell by how much the value of the
category that receives the value of 1 differs from the intercept coefficient of the
benchmark category.
71
BASIC ECONOMETRICS STUDY E MATERIAL
5. If a qualitative variable has more than one category, the choice of the benchmark
category is strictly up to the researcher. Sometimes the choice of the benchmark is
dictated by the particular problem at hand.
6. if a model has several qualitative variables with several classes, introduction of
dummy variables can consume a large number of degrees of freedom. Therefore,
one should always weigh the number of dummy variables to be introduced against
the total number of observations available for analysis.
5.8 ANOVA and ANCOVA
ANOVA, explicated by Ronald A. Fisher ( 1924, 1932, 1935b) for to analyses the data
obtained from agricultural experiments, for to compare the means of any number of
experimental groups or conditions without increasing the Type I error rate. Fisher ( 1932)
also described ANCOVA with an approximate adjusted treatment sum of squares, before
describing the exact adjusted treatment sum of squares a few years later (Fisher, 1935b, and
see Cox and McCullagh, 1982, for a brief history). In early recognition of his work, the F-
distribution was named after him by G.W. Snedecor (1934). ANOVA procedures culminate
in an assessment of the ratio of two variances based on a pertinent F-distribution and this
quickly became known as an F-test.
5.8.1 Meaning ANOVA
ANOVA expands to the Analysis of Variance, is described as a statistical technique
used to determine the difference in the means of two or more populations, by examining
the amount of variation within the samples corresponding to the amount of variation
between the samples. It bifurcates the total amount of variation in the dataset into two
parts, i.e. the amount ascribed to chance and the amount ascribed to specific causes.
Dummy variables can be incorporated in regression models just as easily as
quantitativevariables. As a matter of fact, a regression model may contain regressors that
are all exclusivelydummy, or qualitative, in nature. Such models are called Analysis of
Variance(ANOVA) models.ANOVA models are used to assess the statistical significance
of the relationship between a quantitativeregressand and qualitative or dummy regressors.
They are often used to compare the differencesin the mean values of two or more groups
or categories, and are therefore more general than the ttest, which can be used to compare
the means of two groups or categories only.
5.8.2. Types of ANOVA
It is a method of analysing the factors which are hypothesised or affect the dependent
variable. It can also be used to study the variations amongst different categories, within the
factors, that consist of numerous possible values. It is of two types:
a. One way ANOVA: When one factor is used to investigate the difference amongst
different categories, having many possible values.
b. Two way ANOVA: When two factors are investigated simultaneously to measure
the interaction of the two factors influencing the values of a variable
72
BASIC ECONOMETRICS STUDY E MATERIAL
consider the following model:
Yi = β1 + β2D2i + β3iD3i + ui
whereYi = (average) salary of public school teacher in state i
D2i = 1 if the state is in the Northeast or North Central
= 0 otherwise (i.e., in other regions of the country)
D3i = 1 if the state is in the South
= 0 otherwise (i.e., in other regions of the country
5.8.3 Advantages and Disadvantages of ANOVA
Advantages of ANOVA
1. Whereas the Z test can only be used to compare the means of two populations, the
ANOVA test can be used to compare the means of three or more populations.
2. If there are two different treatments/factors affecting the dependent variable, then
we can use the two way ANOVA test to analyse the effect due to each treatment.
The test will tell us whether the difference due to each of the treatments is significnt
or not.
3. We can check equality of three or more populations means by repeatedly applying
Z test pairwise. But this increases the Type 1 error. On the other hand, the same
comparison done by the ANOVA technique has low Type 1 error. This means that
ANOVAtest is a statistically powerful test.
4. The ANOVA method is used in clinical testing to check for the effectiveness of
experimental medicines.
5. The calculations involved in calculating the F statistics are easy and involve
elementary operations such as squaring, summing up and dividing. The decision
criteria for rejecting or accepting the null hypothesis are easy to understand.
Disadvantages of ANOVA
1. It often happens that the parent populations do not follow the normal distribution.
For example, the lifetimes of products generally follow the Weibull distribution. In
such cases the ANOVA method cannot be used. For instance, we may not be able to
use the ANOVA technique to compare the mean life of bulbs produced by three
companies.
2. If there are two or more dependent variables then the ANOVA technique cannot be
applied. The MANOVA test must be used in such cases.
3. It rarely happens that all the population variances are equal. If the assumption of
homoscedasticity is violated then the use of ANOVA cannot be justified.
4. If the null hypothesis is rejected we can only conclude that some population means
are unequal. The ANOVA test does not tell us anything about which of them are
unequal. Some post hoc tests must be carried out in order to know about that.
5. Checking all the background assumptions such as independence, normality,
homoscedasticity, etc. is in and of itself a difficult task.
73
BASIC ECONOMETRICS STUDY E MATERIAL
6. Although the calculations involved are elementary they are still tedious to peform
by hand. But ANOVA tests are usually carries out using statistical software so this is
not a huge barrier.
5.8.4 Meaning of ANCOVA
ANCOVA stands for Analysis of Covariance, is an extended form of ANOVA, that
eliminates the effect of one or more interval-scaled extraneous variable, from the
dependent variable before carrying out research. Regression models containing an
admixture of quantitative and qualitative variables are called analysis of covariance
(ANCOVA) models. ANCOVA models are an extension of the ANOVA models in that
they provide a method of statistically controlling the effects of quantitative regressors,
called covariates or control variables, in a model that includes both quantitative and
qualitative, or dummy, regressors.
When in a set of independent variable consist of both factor (categorical independent
variable) and covariate (metric independent variable), the technique used is known as
ANCOVA. The difference in dependent variables because of the covariate is taken off by
an adjustment of the dependent variable‟s mean value within each treatment condition.
Yi = β1 + β2D2i + β3D3i + β4Xi + ui
whereYi = average annual salary of public school teachers in state ($)
Xi = spending on public school per pupil ($)
D2i = 1, if the state is in the Northeast or North Central
= 0, otherwise
D3i = 1, if the state is in the South
= 0,otherwise
5.8.5. Assumptions of ANCOVA
This technique is appropriate when the metric independent variable is linearly
associated with the dependent variable and not to the other factors. It is based on certain
assumptions which are:
a. There is some relationship between dependent and uncontrolled variable.
b. The relationship is linear and is identical from one group to another.
c. Various treatment groups are picked up at random from the population.
d. Groups are homogeneous in variability.
5.8.6 Advantages and Disadvantages of ANCOVA
Advantages of ANCOVA
Advantages of ANCOVA include better power, improved ability to detect and estimate
interactions, and the availability of extensions to deal with measurement error in the
covariates.
a. A covariate can be identified and used after the fact to save an experiment when
significance is just being missed. Examples: in educational research, can use IQ or
achievement test scores taken before experiment
74
BASIC ECONOMETRICS STUDY E MATERIAL
b. ANCOVA is more precise than blocking if the correlation between the covariate
and the criterion is greater than .6Remember, ANCOVA not only reduces bias, but
it also improves sensitivity
Disadvantages of ANCOVA
The main disadvantage of ANCOVA is the underlying assumption of no difference
across groups or treatment arms in terms of the covariate used in the analysis and the
homogeneity of regression slopes
a. More assumptions to be violated with ANCOVA and effects of violations of those
assumptions not always clear.
b. Skill of Computational labor needed, if doing by hand, takes laborious.
c. Blocking is more precise than ANCOVA when correlation between covariate and
criterion is less than .4
5.8.7 Comparison of ANOVA and ANCOVA
a. The technique of identifying the variance among the means of multiple groups for
homogeneity is known as Analysis of Variance or ANOVA. A statistical process
which is used to take off the impact of one or more metric-scaled undesirable
variable from dependent variable before undertaking research is known as
ANCOVA.
b. While ANOVA uses both linear and non-linear model. On the contrary, ANCOVA
uses only linear model.
c. ANOVA entails only categorical independent variable, i.e. factor. As against this,
ANCOVA encompasses a categorical and a metric independent variable.
d. A covariate is not taken into account, in ANOVA, but considered in ANCOVA.
e. ANOVA characterises between group variations, exclusively to treatment. In
contrast, ANCOVA divides between group variations to treatment and covariate.
f. ANOVA exhibits within group variations, particularly to individual differences.
Unlike ANCOVA, that bifurcates within group variance in individual differences
and covariate.
The above explanation of difference between ANOVA and ANCOVA can be put in
tabular form for easy understanding and put it in memory for the readers based on six
bases as follows:
Comparison of ANOVA and ANCOVA
BASIS FOR
ANOVA ANCOVA
COMPARISON
ANOVA is a process of ANCOVA is techniques that remove
examining the difference the impact of one or more metric-
Meaning among the means of multiple scaled undesirable variable from
groups of data for dependent variable before
homogeneity. undertaking research.
75
BASIC ECONOMETRICS STUDY E MATERIAL
Both linear and non-linear
Uses Only linear model is used.
model are used.
Includes Categorical variable. Categorical and interval variable.
Covariate Ignored Considered
Divides Between Group (BG)
Attributes Between Group
BG variation variation, into treatment and
(BG) variation, to treatment.
covariate.
Attributes Within Group (WG) Divides Within Group (WG)
WG variation variation, to individual variation, into individual differences
differences. and covariate.
5.9 The Qualitative Response Models
Qualitative response regression models refer to models in which the response, or
regressand, variable is not quantitative or an interval scale. The simplest possible
qualitative response regression model is the binary model in which the regressand is of the
yes/no or presence/absence type.
Linear Probability Model:
The simplest possible binary regression model is the linear probability model (LPM)in
which the binary response variable is regressed on the relevant explanatory variablesby
using the standard OLS methodology. Simplicity may not be a virtue here, forthe LPM
suffers from several estimation problems. Even if some of the estimationproblems can be
overcome, the fundamental weakness of the LPM is that it assumesthat the probability of
something happening increases linearly with the level of the regressor.LPM is plagued by
several problems, such as (1) non-normality of ui, (2) heteroscedasticity of ui, (3) possibility
of Yˆilying outside the 0–1 range, and (4) thegenerally lower R2 values
Limitation of LPM
1. The error term is not normally distributed; it also follows the Bernoulli distribution.
2. The variance of the error term is heteroskedastistic. The variance for the Bernoulli
distribution is p(1-p), where p is the probability of a success.
3. The value of the R-squared statistic is limited, given the distribution of the LPMs.
4. Possibly the most problematic aspect of the LPM is the non-fulfilment of the
requirement that the estimated value of the dependent variable y lies between 0
and 1.
5. One way around the problem is to assume that all values below 0 and above 1 are
actually 0 or 1 respectively
6. An alternative and much better remedy to the problem is to use an alternative
technique such as the Logit or Probit models.
7. The final problem with the LPM is that it is a linear model and assumes that the
probability of the dependent variable equalling 1 is linearly related to the
explanatory variable.
76
BASIC ECONOMETRICS STUDY E MATERIAL
For example if we have a model where the dependent variable takes the value of 1 if a
student has extension contact and 0 otherwise, regressed on the student education level.
The probability of contacting an extension employer will rise as education level rises.
LPM model example
The following model of Machine Learning (ML) was estimated, with Python
Knowledgemarks(d) and Econometric Course Education marks (e ) as the explanatory
variables. Regression using OLS gives the following result.
tˆi 3.12 0.6ei 0.12d i
(2.10) (0.06) (0.04)
R 2
0.25, DW 1.78
1 ML
0 Not Ml
t
The coefficients are interpreted as in the usual OLS models, i.e. a 1% rise in econometric
course educationmarks , gives a 0.60% increase in the probability of Machine learning
technology adoption.The R-squared statistic is low, but this is probably due to the LPM
approach, so we would usually ignore it. The t-statistics are interpreted in the usual way.
Logit Model
In the logit model the dependent variable is the log of the odds ratio, which is a
linearfunction of the regressors. The probability function that underlies the logit model is
thelogistic distribution. If the data are available in grouped form, we can use OLS
toestimate the parameters of the logit model, provided we take into account explicitly
theheteroscedastic nature of the error term. If the data are available at the individual,
ormicro, level, nonlinear-in-the-parameter estimating procedures are called for.
Features of the Logit model
1. As P goes from 0 to 1 (i.e., as Z varies from −∞ to +∞), the logitL goes from −∞ to
+∞. That is, although the probabilities (of necessity) lie between 0 and 1, the logits
are not so bounded.
2. Although L is linear in X, the probabilities themselves are not. This property is
incontrast with the LPM model where the probabilities increase linearly with X.
3. Although we have included only a single X variable, or regressor, in the
precedingmodel, one can add as many regressors as may be dictated by the
underlying theory.
4. If L, the logit, is positive, it means that when the value of the regressor(s)
increases,the odds that the regressand equals 1 (meaning some event of interest
happens) increases.If L is negative, the odds that the regressand equals 1 decrease
as the value of X increases.To put it differently, the logit becomes negative and
77
BASIC ECONOMETRICS STUDY E MATERIAL
increasingly large in magnitude as theodds ratio decreases from 1 to 0 and becomes
increasingly large and positive as the oddsratio increases from 1 to infinity.
5. More formally, the interpretation of the logit model given in Eq.is as follows:
β2, the slope, measures the change in L for a unit change in X, that is, it tells how the
log-oddsin favor of a categorical variable (owning a house) change as
(income)explanatory variable changes by a unit, say, $1,000. Theintercept β1 is the
value of the log-odds in favor of owning a house if income is zero. Likemost
interpretations of intercepts, this interpretation may not have any physical
meaning.
6. Given a certain level of income, say, X*, if we actually want to estimate not the
oddsin favor of owning a house but the probability of owning a house itself, this
can be done directlyfrom Eq. once the estimates of β1 and β2 are available.
7. Whereas the LPM assumes that Pi is linearly related to Xi, the logit model
assumesthat the log of the odds ratio is linearly related to Xi .
5.10 Let us Sum up
From this unit the reader may learned briefly about the concepts used in dynamic
model, reasons for using lag in a regression model, usage of dummies in regression and
theoretical background about the ANOVA and ANCOVA , linear probability model and
logit model. The basic knowledge obtained from this unit may help and induce the reader
to go further reading in detailed way of these topics for their academic needs.
5.11. Unit End Exercises
A. Multiple Choice Questions
1. Including relevant lagged values of the dependent variable on the right hand side of a
regression equation could lead to which one of the following?
A. Biased but consistent coefficient estimate
B. Biased and inconsistent coefficient estimate
C. Unbiased but inconsistent coefficient estimate
D. Unbiased and consistent but inefficient coefficient estimate
2. If in our regression model, one of the explanatory variables included is the lagged
value of the dependent variable, then the model is referred to as
A. Best fit model C. Dynamic model
B. Autoregressive model D. First-difference form
3. Regressionn models containing a mixture of quantitative and qualitative variables are
called :
A. ANOVA models. C. ANCOVA models.
B. Parallel regressions. D. Coincident regressions.
4. In Linear Probability Model, the:
A. Regressand is dichotomous C. Regressand is ordinal variable
B. Regressor is dichotomous D. Regressor is ordina
5. Which of the following models is used to regress on dummy dependent variable ?
78
BASIC ECONOMETRICS STUDY E MATERIAL
A.The LPM model. B. The tobit model.
C. The logit model. D. All of the above.
6. A binary variable is often called as -------------------- in econometrics.
A. dummy variable. C. dependent variable.
B. residual. D. power of a test.
7. The difference between the Autoregressive and Distributed lag model is in its lag of
A. Dependent variable placing among explanatory variable
B. Dependent variable
C. Independent variable
D. Dummy variable.
8. The linear probability model is
A. the application of the multiple regression model with a continuous left-hand
sidevariable and a binary variable as at least one of the regressors.
B. an example ofprobit estimation.
C. another word for logit estimation.
D. the application of the linear multiple regression model to a binary
dependentvariabl
9. Logistic regression is used when you want to:
A. Predict a dichotomous variable from continuous or dichotomous variables.
B. Predict a continuous variable from dichotomous variables.
C. Predict any categorical variable from several other categorical variables.
D. Predict a continuous variable from dichotomous or continuous variables.
PART B
Short answer and Essay type questions
1. What is Lag in Regression?
2. Write a note DL, AR, and MA
3. Find the difference between ANOVA and ANCOVA
4. Clarify the reasons for Lags in Econometric Models.
5. What are reasons for the introduction dummy variable and what are the cautions in use
of dummies in a regression?
6. Enumerate the advantage and disadvantages of ANOVA and ANCOVA
7. What are the features of Logit Model?
5.12 Reference Books
1. DamodarN.Gujarati and Sangeetha, “ Basic Econometrics” Special Indian Edition , Tata
McGraw Hill Education Privated Limited (Sixth Print 2010) , ISBN: 978-0-07-066005-2
2. P.G.Apte,” Text book of Econometrics” Tata McGraw – Hill Publishing Company
Limited.
3. Dhanasekaran,” Econometrics” 2nd Edition, Vrinda Publications (P) Ltd, Delhi-53 ,2011
ISBN: 978-81-8281-388-5
79
BASIC ECONOMETRICS STUDY E MATERIAL
4. HumbertoBarreto and Frank M. Howland,” Introductory Econometrics” Cambridge
University Press, First South Asian Edition 2009, ISBN: 978-0-521-12358-9.
5. Dilip M. Nachane,” Econometrics: Theoretical Foundations and Empirical
Perspectives” Oxford University Press, Second Impression 2010 , ISBN: 978-0-19-
564790-7
6. S.Shyamala,RavdeepKaur, Arul Pragasam,” A text book on Econometrics: Theory and
Applications” Vishal Publishing Co., Jalandhar 2017, ISBN: 81-88-646-98-9
7. S.P.Singh, Anil.K. Parashar, H.P. Singh,” Econometrics and Mathematical Economics”
Second Revised Editions,S.Chand and Company Ltd, New Delhi -55
8. A. Koutsoyiannis,” Theory of Econometrics” Second Edition Palgrave – New
York,2004,ISBN: 0-333-77822-7
9. Maddala .G.S.(1997), “Econometrics” , McGraw Hill, New York.
10. Johnston. (1997),” Econometric Methods” McGraw Hill, 4th Edition, New Delhi.
80
BASIC ECONOMETRICS STUDY E MATERIAL
Additional Unit 6 (Not included in Syllabus MKU)
Structure
6.1 Objectives
6.2 Introduction to Gretl
6.3 Features of Gretl Software
6.4 Installation of Gretl Software (Downloading Gretl from the Internet for Free)
6.5 Creating Data Sets and Reading them into Gretl
6.6 Simple Descriptive Statistics in Gretl
6.7 Let us sum up
6.8 Unit -End Exercises
6.9 Answer to Check Your Progress
6.10 Suggested Readings
Econometrics requires skills in (licensed) software packages for testing the existing
theory, creating a new theory, for evaluating any policy implications in an economy, for
examining the impact on public by government funded programme accessibility, and
utilization in a nation and for to study the cause and effect of a fact.
6.1 Objectives
This Unit is designed to provide students with the basic tools to work with data using
the open source package gretl. In this Chapter students will learn how
To write a script file.
To Import native-type data sets and several other types of data sets.
To explore your data.
To Run basic statistical tests and to Run OLS regressions.
To create graphs and plots.
6.2 Introduction to Gretl
GRETL is a useful tool and free software for teaching econometrics. GRETL has been
written by Allin Cottrell based on ESL (Econometrics Software Library) code written by
RamuRamanathan of the University of California, San Diego. It can be obtained from the
World Wide Web at http://gretl.sourceforge.net/, where the source package and binary
distributions running on GNU/Linux and Microsoft Windows in the form of a self-
extracting executable can be downloaded.
81
BASIC ECONOMETRICS STUDY E MATERIAL
GRETL is the first complete econometric software package released under the GNU
software license. The software consists of a shared library, a command-line client program,
and a graphical client program. It comes with many sample data files from Greene (2000)
and Ramanathan, (2002), which are immediately accessible from the menu. It supports
several least-squares based statistical estimators (including two-stage least squares and
panel data methods), time series models (including the Cochrane–Orcutt procedure and
VARs), and some maximum likelihood methods (logit and probit). It also has built-in
commands for several econometric tests (including the Chow, Hausman, and Dickey–
Fuller tests). A copy of Gretl can be downloaded from the Internet at
http://www.sourceforge.net. It is approximately 7.5MB in size. An important item that
can be found on the Gretl window is the option for defining a new variable. Often new
variables must be created.
6.3 Features of Gretl Software
Gretl is an econometrics package, including a shared library, a command-line client
program and a graphical user interface.
a. User-friendly: Gretl offers an intuitive user interface; it is very easy to get up and
running with econometric analysis.
b. Flexible: You can choose your preferred point on the spectrum from interactive
point-and-click to complex scripting, and can easily combine these approaches.
c. Cross-platform: Gretl‟s “home” platform is Linux but it is also available for MS
Windows and Mac OS X, and should work on any Unix-like system that has the
appropriate basic libraries
d. Open source: The full source code for Gretl is available to anyone who wants to
critique it, patch it, or extend it.
e. Sophisticated: Gretl offers a full range of least-squares based estimators, either for
single equations or for systems, including vector autoregressions and vector error
correction models. Several specific maximum likelihood estimators (e.g. probit,
ARIMA, GARCH) are also provided natively; more advanced estimation methods
can be implemented by the user via generic maximum likelihood or nonlinear
GMM.
f. Extensible: Users can enhance gretl by writing their own functions and procedures
in gretl‟s scripting language, which includes a wide range of matrix functions.
g. Accurate: Gretl has been thoroughly tested on several benchmarks, among which
the NIST reference datasets.
h. Internet ready: Gretl can fetch materials such databases, collections of textbook
data files and add-on packages over the internet.
i. International:Gretl will produce its output in English, French, Italian, Spanish,
Polish, Portuguese, German, Basque, Turkish, Russian, Albanian or Greek
depending on your computer‟s native language setting.
82
BASIC ECONOMETRICS STUDY E MATERIAL
Check Your Progress:
a. What is Gretl?
b. Is Gretl Paid software?
c. Who wrote the code to Gretl?
d. Is Gretl able to open in MS Window?
gre
Software?
6.4 Installation of Gretl Software
1. DownloadingGretl from the Internet for Free
GRETL econometric software can be downloaded at the following site:
http://gretl.sourceforge.net/
1. On the left hand side of this web site, double left-click on “Gretl for Windows”
,You will be directed to another web page that looks like the following:
2. Double left-click on “gretl_install.exe”which is about 7.5 MB You will be again
directed to another web page like the following:
83
BASIC ECONOMETRICS STUDY E MATERIAL
3. Choose a “preferred mirror”– for example, “unc”as shown above.
4. Right-click on “download 7483 kb” say from “Chapel Hill, NC”
5. Save the file to your desktop
6. After downloading this file, open it and follow the installation wizard.
7. Gretl will create a shortcut on your desktop that looks like the following:
6.5 Creating Data Sets and Reading them into Gretl
After you have successfully downloaded Gretl you will find the package was saved in
the default target c:\userdata\gretl and a small icon was probably placed on your desktop.
The icon looks like
You can double left-click on the gretl icon and it will start the software package. A
main gretl window will open and looks like the window on the following page. The
options which run along the top of the window are
File Utilities Session Data Sample Variable Model and Help
The blank part of the window will be filled when you open your data file (to be
explained shortly). It will show a list of the variables in your data set. You can then choose
the type of analysis that you wish to perform. You can run Gretl by simply clicking on
options or you can write a gretl program and run it either interactively or in batch modes.
Unless you wish to explore the software on your own, the first step in a Gretl session is
the creation of a data set which Gretl can understand. The simplest way this can be done is
to use Notepad (or Wordpad) and the Gretl Command editor. You don‟t need Excel to
create this kind of data set. Let‟s take an example.
84
BASIC ECONOMETRICS STUDY E MATERIAL
Data Set Type I: Let‟s suppose that you have 4 observations on two variables X1 and
X2, which you can write as
Date X1 X2
1970 Q1 1 3
1970 Q2 3 7
1970 Q3 3 6
1970 Q4 5 8
1971 Q1 6 12
You may have simply keyed these in yourself, or you may have found the data on the
Internet and used a cut and paste function to create a file. Let‟s suppose that you use
Notepad (or Wordpad) to create a data file called mydata.txt and which you have saved on
your desktop.
The Notepad (or Wordpad) file looks like the following:
85
BASIC ECONOMETRICS STUDY E MATERIAL
Note that this quarterly data does not have any names, like X1 or X2, in the Notepad
(or Wordpad) file. It also does not have any dates, like 1970 Q3 or 1971 Q1.
Gretl will not read this data file as it is. You must do two things.
(1) You save the file again (using Notepad (or Wordpad)) as mydat.gdt
You should save the file to c:\userdata\gretl\user
(2) Next, you start Gretl (by clicking on the icon) and select File then
choose
New Command File and then choose Regular Script
(3) A new Gretl window will open and you proceed to create a Data
Header File
(4) You type into the window a description of the data and save it as
mydat.hdr .You should save the file to c:\userdata\gretl\user
There are several things to note.
First, the data file mydat.gdt and the header file mydat.hdr BOTH use the base name
“mydat”
Second, the data file must use the file suffix ___.gdt and the header file must use the file
suffix ___.hdr
Third, you should save the files in the c:\userdata\gretl\user location.
Fourth, the data is arranged in columns by observation within the data file mydat.gdt
Fifth, the header file has comment lines ( (* _____*) ), a list of names for the variables (x1
and x2 ending in ; ) and a description of the time series nature of the data (4=quarterly,
1970.11971.1 byobs).
Sixth, the variable names are case sensitive, so X1 is different from x1.
After you have created and saved both files mydat.gdt and mydat.hdr to the folder
c:\userdata\gretl\user you can start Gretl and begin your analysis.
86
BASIC ECONOMETRICS STUDY E MATERIAL
When Gretl starts (after you click on the Gretl icon), the main Gretl window will open
and you can read your data into the window. To do this, you choose FILE and then OPEN
DATA. You, then choose USER FILE and proceed to double left-click on the file mydat
shown. Gretl will then read in your data into the main Gretl window and you will see the
window shown on the following page. Note that both x1 and x2 have been read into Gretl.
A constant has beenautomatically generated also. We can now carry out many types of
statistical analyses on x1 and x2.
Note, also that at the bottom the data frequency and sample range is shown: “Quarterly:
Full range 1970:1 – 1971:1; current sample 1970:1 – 1971:1” Also, at the bottom there are
several convenient shortcuts
These are, respectively: (1) calculator, (2) editor (may not work), (3) interactive Gretl
console, (4) icon view (must have Gretl sessions saved first), (5) Gretl website link, (6)
Gretl Manual (in ___.pdf form, must have free Acrobat Reader from Adobe.com), (7) Gretl
Help, (8) X-Y Graphics, (9) Open Data. There are other ways to read data into Gretl, but
for now, this should be enough since it can be done on virtually any computer.
6.6 Simple Descriptive Statistics in Gretl
If you have successfully downloaded Gretl and have created the example data set
above, then you can proceed to undertake simple statistical analyses of the data.
Finding the sample mean of the variable x1: Choose Data, Summary Statistics,
Selected Variables and get the following
87
BASIC ECONOMETRICS STUDY E MATERIAL
Note that this command gives a number of different summary statistics – not just the
sample mean only. S.D. = standard deviation, C.V. = coefficient of variation =
(S.D./Mean), SKEW = Measure of skewness, EXCSKURT = measure of excess kurtosis.
We can also make a time series graph for X and Y. Choose DATA, GRAPH
SPECIFIED VARS, TIME SERIES PLOT…
Note that x1 and x2 are quarterly variables defined over the sample range 1970:1 –
1971:1. You can copy and paste this graph to a word document. Click on the graph and
follow the options given. This is known as a time series graph. Gretl can also produce X-Y
graphs. Just choose DATA, GRAPH SPECIFIED VARS, X-Y SCATTER…
6.7 Let us sum up
This unit made to introduce the free software, its features, and how to install for
solving the economic problems. After installation of software in desktop or laptop, creation
of data file and how to get descriptive statistics and diagrams are explained with screen.
88
BASIC ECONOMETRICS STUDY E MATERIAL
6.8 Unit -End Exercises
1. What is Gretl? And who had written the code for it?
2. Enumerate the features of Gretl.
3. Explain the steps involved in installing Gretl in your desktop or Laptop?
4. Discuss the way of creating data file and drawing a trend line.
6.9 Answer to Check Your Progress
a. Gretl is an econometrics package, free software including a shared library, a
command-line client program and a graphical user interface.
b. No. Gretl is free Software.
c. Dr.RamuRamanathan has written the code to Gretl Software.
d. Yes. Gretl is able to open in MS Window.
6.10 Suggested Readings
Gretl Used guide , and www.google.com
89
BASIC ECONOMETRICS STUDY E MATERIAL
1. A Cumulative Distribution Function: A mathematical function which allows us to
calculate the probability that a random variable will take on a value equal to or less
than a specified value.
2. Accounting Identity: A relationship which holds exactly as a result of accounting
conventions
3. Adjusted R Square: A measure of goodness of fit which is adjusted for the loss of
degrees of freedom,.i.e. incorporates a penalty for using too many independent
variables.
4. Adaptive Expectation : A model of expectation formation in which expectations
are revised depending upon the discrepancy between current experience and past
expectation
5. A Just Identified Equation: Data and prior information are just enough to uniquely
identify the equation. Indirect least squares feasible.
6. An Elementary Outcome: One of the many possible results of a random
experiment.
7. An Event: A happening of interest in the context of a random experiment. An event
is said to have occurred when any one outcome from a specified subsets of
outcomes occurs
8. An Over-identified equation: Data and prior restrictions apparently contain
redundant information. indirect least squares leads to multiple solutions.
9. An Unidentified Equation:A equation from the system which cannot be
distinguished form similar looking equation based on data and prior information.
indirect least squares not feasible
10. Arithmetic Lags: Lag coefficient exhibit a linear pattern of increase or decline.
11. An ARMA Process: A combination of auto-regressive and moving average
processes
12. Approximate Multicollinearity: Existence of an approximate linear relationship
among the independent variables.
13. Asymptotic Behaviour : Behaviour of an estimator and its distribution as sample
size increases without limit
14. Asymptotic property: A property of a statistic that applies as the sample size grows
large (specifically, as it tends to infinity).
15. Asymptotic Unbiasedness : Bias of the estimator tends to zero as sample size
increases .
16. Attrition bias: Bias caused by unit non-response in panel data. This occurs when
the individuals who drop out of a panel study are systematically different from
those who remain in a panel study.
90
BASIC ECONOMETRICS STUDY E MATERIAL
17. Autoregressive Process: A stochastic process in which the current value of the
disturbance is a function of past values with a random variables super imposed
18. Average effect: A measure of the effect of a binary explanatory variable, x, on the
outcome of interest; based on comparing the outcome when x equals 1 with the
outcome when X equals 0.
19. Average treatment effect (ATE): a measure commonly used in the policy
evaluation literature that gives the expected difference in outcomes between those
who receive a treatment and those who do not, across the whole study population.
Related to the average treatment effect on the treated (ATET) which is the expected
difference for those who would opt for treatment.
20. Backward Elimination: A computational routine in which one variable is dropped
at a time starting from a model which includes all the independent variables.
21. Behavioral Equation: An algebraic relationship based some assumption about the
behaviour of economic agents.
22. Best Estimator : within a given class of estimators the one with the minimum
variance.
23. Best Linear Unbiased Forecast: Within the class of linear unbiased forecasts the
one with minimum error variance.
24. Bias : The difference between the expected value of an estimator and the true value
of the parameter being estimated. A measure of how well the estimator performs
on average
25. Binary variable: A variable that takes only two values, usually coded as zero and
one.
26. Bivariate probit model: A model that combines two binary probit models to deal
with a system of two binary dependent variables.
27. Box-Cox Transformation: A generalised functional form which has as imiting
cases several forms used often in practise. e.g. bilinear, double log, semi-log, etc.
28. Conditional logit: A model for unordered multinomial outcomes in which the
regressors vary across the alternatives (see mixed logitand multinomial logit).
29. Conditional Probability: Probability of a event given that some other specified
event has already occurred.
30. Consistency: The estimator approaches the true parameter as sample size increases.
31. Consistent estimate: An estimate that converges on the true parameter value as the
sample size increases (towards infinity).
32. Contemporaneous Covariance:Across equations covariance between the
disturbances referring to its same time period.
33. Continuous variable: A variable that can take any take the value of any real
number with in an interval.
34. Cox proportional hazard model: A semi parametric model for duration analysis.
35. Cross-section data: Survey data in which each respondent is observed only once,
giving a “snapshot” view of the population at a point in time.
91
BASIC ECONOMETRICS STUDY E MATERIAL
36. Deterministic Model: A collection or set of exact structures, i.e. a set of
autonomous relationships with parameter values unspecified.
37. Diagnostic checking: A set of graphical and formal testing procedures applied to
OLS residuals to detect violations of basic assumptions, outliers etc.
38. Disturbances Related Equations: A set of equations in which disturbances in
different equations are correlated.
39. Dummy variable: Another label for binary variables that take the value zero or one.
A variable used to represent qualitative attribution and or structural breaks in
regression
40. Durbin-Watson Statistic: A test statistic to detect the presence of first order auto-
correlation. using residual from a preliminary OLS regression
41. Distributed Lags: Effect of a change in independent variables is spread over
current and several future periods.
42. Efficiency: Variance of an estimator relative to that of another estimator from
within a specified class of estimators. A measure of precision of an estimator.
43. Embedding: A procedure in which a more general specification is formed such
that the two non-nested hypotheses are special cases of the general model
44. Endogenous Variables: Those variables the values of which are determined within
the system. Otherwise, Variables the values of which are determined by or within
the model.
45. Error components model: A regression model for panel data.
46. Error Sum of Squares(ESS): The sum of squares of discrepancies between observed
and calculated values of the dependent variable.
47. Estimate : Particular numerical value of the estimator for the particular sample at
hand.
48. Estimator : A rule for combining the sample observations to arrive at the “ best
guess” as to the true value of a unknown parameter.
49. Error of Type 1: Rejecting a valid hypothesis.
50. Error of Type 2: Accepting an invalid hypothesis.
51. Extrapolation: A procedure to estimate values outside the observed sample.
52. Expectation of a Random Variable : (also called mean) A measure of central
tendency or “on average “ behaviour of a random variable defined with its density
function.
53. Exogenous Variables: Variables, the value of which have to be supplied from
outside the model.
54. Exact Multicollinearity: Existence of an exact linear relation among the
independent variables.
55. Ex-Ante Forecast: Forecasting the future values before they are actually realised.
56. Excess zeros: A feature of count data, when the number of zeroes observed exceeds
the number that would be expected from the Poisson model.
92
BASIC ECONOMETRICS STUDY E MATERIAL
57. Ex-Post Forecasts: Utilising part of the sample data to forecast values for which
actual values are available.
58. Extraneous Estimators: Estimators of parameters in a model obtained from a
different body of data in possibly a different context
59. Exogenous Variables: Variables which appear in the system but are determined
outside the system.
60. Exogeneity: In the context of regression analysis, the assumption that the
regressors, x, are independent of the error term.
61. FIML: Full-information maximum likelihood (FIML) estimates multiple equation
models using the joint distribution for the equations rather than estimating each
equation separately.
62. Final Form of the Equation System: Each endogenous variable is expressed
exclusively in terms of current and lagged exogenous variables. Lagged
endogenous variables eliminated.
63. First Order Regression: An estimation procedure for missing observations in which
they are estimated from some auxiliary regressions
64. Fixed effects: The fixed effects specification treats the individual effects in panel
data models as parameters to be estimated. This is appropriate when inferences are
to be confined to the effects in the sample only, and the effects themselves are of
substantive interest With individual level survey data fixed effects are best
interpreted as random individual effects that are correlated with the explanatory
variables. This contrasts with randome effects that are assumed to be independent
of the regressors (see random effects).
65. Forward Selection:A computational routine in which one variable is added at a
time starting from a model with one independent variables.
66. Gamma distribution: Probability distribution often used to model individual
heterogeneity, especially in count data regression and duration analysis.
67. Gauss Mark of Theorem; The result that for the standard linear model with scalar
covariance matrix of disturbances the ordinary least squares estimators are BLUE
68. Geometric Lags: Lag coefficients are generated from a geometric distribution and
exhibit exponential decline..
69. Gibbs sampling: a method for drawing samples from a distribution that is used in
MCMC algorithms.
70. GMM:Many of the estimators discussed in this book fall within the unifying
framework of generalised method of moments (GMM) estimation. This replaces
population moment conditions (e.g. based on expected values) with their sample
analogues (e.g. based on sample means).
71. Generalized least squares: A generalization of ordinary least squares which relaxes
the assumption that the error terms are independently and identically distributed
acrossobservations.
93
BASIC ECONOMETRICS STUDY E MATERIAL
72. Hausman test: Tests whether there is a significant difference between two sets of
coefficients: one set that are efficient under the null but inconsistent under the
alternative and another set that are inefficient under the null but still consistent
under the alternative. Commonly used to test the IIA assumpition in multinomial
choice models and as a test of exogeneity (comparing OLS and IV extimates).
73. Hazard function: Defined as the ratio of the density function to the survivor
function for a random variable. The hazard function plays a key role in duration
analysis where it is interpreted as the probability of failing now given survival up
to now.
74. Heckit model: A two-step estimator designed to deal with the sample selection
problem.
75. Heteroskedasticity: When the variance of the error term is not constant across
observations.
76. Heteroskedastic Linear Model: A regression model in which the disturbance
variance can change from observation to observation.
77. Homoskedasticity: The property of constant variance of random disturbance,when
the variance of the error term is constant across observations.
78. Identification: The process of distinguishing a particular structure from a set of
competing structures using data and prior information.
79. Indirect Least Squares: Estimating reduced form parameters by OLS and solving
for structural parameters from these.
80. Independence of Explanatory Variables and Random Disturbances: the
assumption that the explanatory variables, if they are to be treated as stochastic, are
statically independent of the random disturbance.
81. Influential Observation: A measure of influence of an influential observation.
82. Interval Forecast: Analogous to interval estimation. providing an interval which
will bracket the true value with stated probability.
83. Interaction: Joint effect of two attributes. incorporated by including products of
dummy variables
84. Interpolation: A procedure to estimate a missing value lying between two known
values.
85. International Contribution: Increases in the R2 as a result of adding a variable to
the model.
86. Instrumental Variable: A variable which is uncorrelated with the disturbance but
highly correlated with the explanatory variable which it acts as the instrument
87. Intersection of Two Events: Joint occurrence of two events, For more than two
events, the definition is same.
88. Instability of Coefficients: Extreme sensitivity of coefficient magnitudes and signs
to small perturbations of data and/or addition of variables. a consequences of
multicollinearity.
94
BASIC ECONOMETRICS STUDY E MATERIAL
89. Instrumental variables: A method of estimation for models with endogenous
regressors – regressors that are correlated with the error term. It relies on variables
(or “instruments”) that are good predictors of an endogenous regressor, but are not
independently related to the dependent variable. These may be used to purge the
bias caused by endogeneity.
90. Interval regression: A variant on the ordered probit model that can be used when
the threshold values are known.
91. Interval Estimation: Constructing an interval which will bracket the time
parameter value with a specified probability
92. Inverse Mills ratio (IMR): The label given to the hazard rate (ratio of density to
survival functions) for a probit model. The IMR is used in the Heckit correction for
sample selection bias.
93. Inverse probability weights: Used to re-weight sample data to make it
representative of the underlying population. IPWs give more weight to those
observations that are under-represented in the sample.
94. Irrelevant Variables: A specification error. Inclusion of unnecessary variables
leads to inefficient estimates.
95. Item non-response: When a respondent does not provide data for a particular
variable in a survey.
96. Joint Probability: Probability of joint, i.e. simultaneous occurrence of two or more
events.
97. Joint Density and Distribution Function : Function specifying the joint
probabilistic behaviour of one or more random variables conditional on specified
events
98. Kaplan-Meier: A nonparametric estimator for survival curves and hazard
functions.
99. Latent Root Regression: Similar to principal components regression but retain the
predictive multicollinearities
100. Left truncation: A phenomenon that arises with duration data that has been
sampled after the original start of the process. Left truncation occurs when some
observations may have already failed before the data are collected and are therefore
missing from the data.
101. Likelihood of a Sample: The probability that a given sample would have arisen
from a particular population
102. Limited Information Maximum Likelihood (FIML): A joint ML procedure for the
whole system.
103. Linear Estimator: An estimator that is a linear function of the sample values.
104. Linear probability model: A model for binary dependent variables based on the
linear regression model.
105. Non-Linear Model: Models which are linear in parameters, those which are not,
are non -linear models.
95
BASIC ECONOMETRICS STUDY E MATERIAL
106. Logistic distribution: A continuous probability distribution that is the foundation
for the logit model of binary choice.
107. Logit: A model for binary dependent variables based on the logistic distribution.
108. Maintained Hypothesis: A set of assumptions about the phenomenon being
investigated which are accepted on faith.
109. Marginal effect: A measure of the effect of a continuous explanatory variable, x, on
the outcome of interest; based on the derivative of the outcome with respect to x.
110. Marginal Density and Distribution Functions : Given joint distribution these
functions describe the probabilistic behaviour of one of the random variables for all
possible values of the other random variables.
111. Maximum likelihood estimation: A method of estimation that specifies the joint
probability of the observed set of data and finds the parameter values that
maximize it (i.e. that are most likely). Otherwise, An approach to estimation of
parameters which chooses those values for the parameters which would maximise
the likelihood of the sample at hand.
112. Mean Square Error: Expected value of the square of the discrepancy between an
estimator and the true value of the parameter.
113. Mean Square Error of Prediction: Mean square error of the forecast as an estimator
of the actual value.
114. MCMC: A Bayesian method used to form a sample from the posterior density by
constructing a Markov Chain in which each value is drawn conditionally on the
previous iteration.
115. Metropolis-Hastings algorithm: A sampling method used in MCMC techniques
when Gibbs sampling is not possible.
116. Mixed logit: A model for unordered multinomial outcomes in which the regressors
can vary across individuals and across the choices. The label is also applied to the
more general random parameters logit model. (See conditional logitand
multinomial logit).
117. Moving Average Process: A stochastic process in which the current values of the
disturbance term is a weighted sum of current and past values of a random
variable.
118. Multicollinearity: Existence of linear relationship among the explanatory variables.
119. Multinomial logit: A model for unordered multinomial outcomes in which the
regressors vary across individuals (see mixed logitand conditional logit).
120. Negbin: An extension of the Poisson regression model for count data.
121. Nelson-Aalen: A nonparametric estimator for cumulative hazard functions.
122. Nested Hypotheses: A pair of hypotheses in which one is a special case of the
other obtained by imposing suitable restrictions on the parameters
123. Non-Nested Hypotheses: A pair of hypotheses in which neither can be derived as
a special case of the other. also called separate families of hypotheses
96
BASIC ECONOMETRICS STUDY E MATERIAL
124. Normal distribution: A continuous probabilty distribution that has a typical
“bellshape”. Used as the foundation for classical regression and analysis and many
other models such as the probit model and the Heckit model.
125. Null Hypothesis: An assertion, to be tested, that the true parameter values are such
and such.
126. Omitted Variables: A specification error. Exclusion of relevant variables leads to
biased estimates
127. Order Conditions for Identifiability: A necessary condition stated in terms of
number of restrictions imposed on the structural coefficients
128. Ordered probit: A model for ordered multinomial outcomes.
129. Ordinary least squares (OLS): The standard method for fitting the classiical linear
regression model. It is based on finding the parameter values that minimize the
sum of squared errors.
130. Outlier: An observation ( data point ) which appears to depart substantially from
the fitted model.Otherwise, An observation which departs substantially from the
rest of the data.
131. Over-dispersion: When observed count data are more spread out than would be
expected from a Poisson model.
132. Panel data: Survey data in which each respondent is observed repeatedly over
time.
133. Parameters: Unknown, unobservable constants which link up variables into a
relationship.
134. Partial Correlation: A measure of linear association between two variables after
both have been adjusted for their common dependent on a given set of other
variables.
135. Partial effect: Used to measure the impact of a change in a regressor on the
probability of the outcome of interest. Relevant for nonlinear models, such as
binary choice models, where the partial effect is not simply the regression
coefficient.
136. Partial Adjustment: A model of dynamic adjustment in which the current action is
dependent upon the gap between a target and reality.
137. Predetermined Variables: Current exogenous, Lagged exogenous and lagged
endogenous variables.
138. Predictive Multicollinearity: An approximate linear relationship which involves
some independent variables and the dependent variable. such linear combination
have good predictive power.
139. Principal Components Regression: An approach to estimation which uses a few
linear combinations of the original variables known as principal components.
140. Probability : A quantitative measure of uncertainty associated with the occurrence
of an event in the context of a particular experiment.
97
BASIC ECONOMETRICS STUDY E MATERIAL
141. Probability Density Function: A mathematical function which gives the
probability that a random variable will take on a specified value or (loosely
speaking) a value within a small neighbourhood of the specified value.
142. Point estimate: A single number used to estimate an unknown parameter (the “best
guess”). As opposed to an interval estimate, which presents a range of values.
143. Point Estimation: Offering a single estimate as the best guess for a parameter
144. Point Forecast: Analogous to point estimation. giving a single value as the best
guess for the future value of the variable being forecast.
145. Poisson regression: A model for count data.
146. Polynomial Lags: Lag coefficients are generated from a polynomial of a specified
degree.
147. Pooled Data: Data from a time series of cross sections.
148. Probit: A model for binary dependent variables based on the standard normal
distribution.
149. Propensity score: The probability of participating (in a treatment) conditional on a
set of regressors, p(y=1|x). The propensity score is used in matching and sample
selection estimators.
150. Power of a Test: A measure of the ability of a test to distinguish between
alternative hypotheses
151. Qualitative effect: The sign of the effect of one variable on another.
152. Quantitative effect: The magnitude of the effect of one variable on another.
153. Random disturbances: A random variable which represents the discrepancy
between an exact structure and reality
154. Random effects: The random effects specification treats the individual effects in
panel data models as random draws. If individual effects are not of intrinsic
importance in them, and are assumed to be random draws from a population of
individuals, and if inferences concerning population effects and their characteristics
are sought, then a random specification is suitable
155. Random effects probit: A model for binary dependent variables in panel data.
156. Random Experiment: A sequence of action carried out under specified conditions.
157. Random Variable : In the context of a random experiment, it is the variable which
take on (real) numerical values , according to a well-defined rule based on
occurrence of different events
158. Rank Conditions for Identifiability: A necessary and sufficient condition for an
equation to be identified started in terms of rank of a sub matrix of reduced form or
structural form coefficient.
159. Raw Moments R2 (R2m): A measure of goodness of fit for models without intercept.
variation in the dependent variable is measured around zero instead of its mean.
160. Reduced form of a System: A form in which each endogenous variables is
expressed in terms of predetermined variables.
98
BASIC ECONOMETRICS STUDY E MATERIAL
161. Reduced Form: A set of relationship, derived from the structural equations, in
which each endogenous variable is expressed a function of all exogenous variables.
162. RESET: A general test for misspecification of the functional form of a regression
model.
163. Retransformation problem: Highlights the need to use an appropriate
transformation back to the y-scale when regression models are run on transformed
data such as log(y).
164. Ridge Regression: A procedure which attempts to reduce the influence of
Multicollinearities through the introduction of a biasing constant.
165. Ridge Trace: A graphical procedure used for choosing the value of the biasing
constant in ridge regression
166. Right censoring: Occurs when values in the right hand tail of a distribution are cut-
off at some threshold and only the threshold value is known. This often arises in
duration analysis where some spells are incomplete at the time the data are
collected.
167. Risk function: A decision -Theoretic concept. The expected cost of using an
estimator to estimate a parameter. cost arises from the wrong decisions ,i.e. using
an estimate different from true value
168. Sample: A finite collection of values of the random variable actually observed.
169. Sampling Distribution: Probability distribution of an estimator.
170. Sample selection bias: The bias created when non-responders are systematically
different from responders.
171. Sample Space: The collection of all the possible elementary outcomes.
172. Scalar- Covariance Matrix: When the covariance matrix of the random
disturbances is diagonal with identical diagonal elements. Consequence of
homoskedasticity and serial independence.
173. Serial Correlation: Successive disturbance terms in the regression mode are
correlated.
174. Serial Independence: The property of mutual stochastic independence of random
disturbances.
175. Semi parametric: A method that mixes parametric assumptions (e.g. that the
relationship between y and X is linear) and nonparametric assumptions (e.g. that
the distribution of the error term is unknown).
176. Singular- Covariance Matrix: Contemporaneous disturbances are linearly
dependent resulting in a singular covariance matrix of disturbances
177. Specification Error Test: A test designed to detect departure from one or more
assumptions behind a proposed model.
178. Splicing: A procedure to obtain a consistent index series from two series with
different bases but at least one point of overlap.
179. Standardisation: Re-expressing variables with a change of origin and units of
measurement
99
BASIC ECONOMETRICS STUDY E MATERIAL
180. Standard Deviation: The positive square root of the variance.
181. Standard Multiple Correlation( R ): A measure of goodness of fit defined as the
fraction of the variation in the dependent variable around its mean explained by the
model.
182. Stepwise Regression: A procedure which allows addition and deletion of
independent variables one at a time. combination of the above two procedures.
183. Stochastic Independence: tea event are said to be stochastically independent when
the knowledge that one of them has occurred does not affect the probability of
other
184. Stochastic Model: A deterministic model with random disturbances added on.
185. Structure (Exact): A collection of autonomous relationship (i.e.of the above three
types) with specified numerical values for the parameters.
186. Structural Parameters: Parameter which appear in the structural form equations..
187. Structural form of a Simultaneous Equation System : Representation of the system
as a collection of autonomous behavioural, technological and accounting
relationships.
188. Smoothing: Removing trends, cycles on seasonal variations from data.
189. Technological relation: A relationship usually cast aa an algebraic equation which
describes a technological constraint.
190. The t-value of a Coefficient: Estimated coefficient divided by its estimated
standard error. the statistic is used to test the hypothesis that the coefficient equals
zero.
191. The F-value of a Set of Coefficient: The statistic used to test the hypothesis that all
the coefficient in a set are simultaneously zero.
192. Three Stage Least Squares(3SLS): A procedure for joint estimation of all identified
equations using the SUR procedure.
193. Two Stage Least Squares (2SLS): A procedure for the estimation of a single
identified equation using two rounds of OLS.
194. Unbiased Forecast: Expected value of the forecast coincides with the expected
values of the variable being forecast.
195. Unconditional Forecast: Forecasts in the presence of certain knowledge of values of
explanatory variables.
196. Union Of Two Events : Given two events, their union is a third event which is said
to have occurred when either or both of two events occur. union of more than two
events is defined in the same manner.
197. Unit non-response: When a potential respondent does not provide data for any
variables in a survey.
198. Unbalanced panel: A panel dataset that includes all respondents who report data
for atleast one period (wave) of the panel. In contrast to a balanced panel which
only includes those individuals with complete data for all periods.
100
BASIC ECONOMETRICS STUDY E MATERIAL
199. Variables: Entities whose behaviour is being studied. generally they represent
some measurable and observable economic construct.
200. Variance Inflation Factors: Diagonal element of covariance matrix of OLS
estimators. they indicate the impact of multicollinearity on estimator variances.
201. Variance of Random Variable : A measure of dispersion around the mean in the
values of a random variable. defined with respect to its density function.
202. Von Neumann Ratio: A statistic for testing for serial correlation in a series of
random variables.
203. Weibull model: A parametric model for duration analysis.
204. Weighted least squares: Weights (wi) are attached to the values of the dependent
variable (yi) and independent variables (xi) before using least squares regression.
This method can be used to correct for heteroskedasticity.
205. Zero Order Regression: An estimation procedure for missing observations in
which they are replaced by averages of the available observations.
***********
101
BASIC ECONOMETRICS STUDY E MATERIAL
(Question Pattern Based on MKU)
Sample Question Paper-1
TIME : Three hours MARKS: 75
PART-A
Answer the following questions (10X1=10)
1. The term regression was coined by
A. Francis Galton B. Karlpearson
C. Carl Friedrick Gauss.. D. William Sealy Goss
2. Method of ordinary least square is attributed to
A. CarlFriedrick Gauss B.William Sealy Goss
C.Durbin Watson D. Both b and c
3. Locus of the conditional mean of the dependent variable for the fixed values of the
explanatory variable
A. Indifference curve B. Population regression curve
C. Production Possibility curve D. None of these
4. In Yi= β1+β2X+ui, ui
A. Represent the missing values of Y
B. Acts as proxy for all the omitted variables that may affect Y
C. Acts as proxy for important variable that affect Y
D. Represent measurement errors
5. One of the assumption of CLRM is that the number of observations in the sample
must be greater the number of
A. Regressor B. Regressands
C. Dependent variable D. Dependent and independent variable
6. The coefficient of determination shows,
A. Variation in the dependent variable Y is explained by the independent variable X
B. Variation in the independent variable Y is explained by the dependent variable X.
C. Both a and b are correct
D. Both a and b are wrong
7. What is the meaning of the term "heteroscedasticity"?
A. The variance of the errors is not constant
B. The variance of the dependent variable is not constant
C. The errors are not linearly independent of one another
D. The errors have non-zero mean
8. Near multicollinearity occurs when
A. Two or more explanatory variables are perfectly correlated with one another
B. The explanatory variables are highly correlated with the error term
C. The explanatory variables are highly correlated with the dependent variable
D. Two or more explanatory variables are highly correlated with one another
102
BASIC ECONOMETRICS STUDY E MATERIAL
9. If in our regression model, one of the explanatory variables included is the lagged
value of the dependent variable, then the model is referred to as
A. Best fit model B. Dynamic model
C. Autoregressive model D. First-difference form
10. In binary logistic regression:
A. The dependent variable is continuous.
B. The dependent variable is divided into two equal subcategories.
C. The dependent variable consists of two categories.
D. There is no dependent variable.
PART-B
Answer the following questions either (a) Or (b) (5X7=35)
11. (a). Enumerate aim and objectives of Econometrics.
(Or)
(b).State significance of stochastic Error term
^ Σ xi yi ^ Σ xi yi
12. (a).Prove that α = Y - x Σx 2 and β = Σx 2 for a simple regression.
i i
(Or)
(b). List out the properties of OLS estimators.
13. (a). Three related variables take following sets of values: Estimate a regression X1 on
X2 and X3
X1: 123 4 5
X2: 2 15 4 3
X3: 3 14 5 2
(Or)
(b).Illustrate the application of Multiple Regression in day to day life.
14. (a) Write a note on Park test.
(Or)
(b) What are the consequences and remedies of Autocorrelation?
15. (a) What are the reasons for lags in econometrics?
(Or)
(b) Differentiates ANOVA and ANCOVA.
PART-C
Answer any one of the following questions (3X10=30)
16. Illustrate the methodology of Econometrics.
17. What are the assumptions of Classical Linear Regression Model?
18. Derive the formula for 𝛽. of multiple regression in matrix form.
19. Enumerate the causes, consequences and remedies of Multicollinearity.
20. State the consequences of Model Specification Error.
******************
103
BASIC ECONOMETRICS STUDY E MATERIAL
Sample Question Paper-2
Time: 3 hrs Max Marks: 75
Answer all the Questions
PART-A (10x1=10Marks)
1. Econometric is an
A) amalgamation of mathematics and statistics
B) organization of Mathematics and Economics
C) integration of Mathematics, Statistics and Economics
D) integration of Economics and Statistics
2. One of the following assumption is not in OLS
A) E(ui /xi)= 0 B) Cov(uiuj/xixj) = 0
C) No Autocorrelation D) No Perfect Multicollinearity
3. In the simple linear regression model, the regression slope indicates
(A) by how many percent Y increases, given a one percent increase in X.
(B) the explanatory variable will give you the predicted Y.
(C) by how many units Y increases, given a one unit increase in X.
(D) Represents the elasticity of Y on X.
4. Find out which is not the property of a parameter?
A) Linear in Parameter
B) Parameters has the Minimum Variance
C) Parameters are Unbiased
D) Biased
5. The Co-efficient of Determination Measures
A) The correlation between the X and Y
B) Goodness of fit of the model
C) Error
D) TSS
6. The term Multiple regression Stands for
A ) Regressing more than one explanatory variables
B) Regressing no variables
C)Many regression
D)Regressing one explanatory variable
7. Find out which is not the violation of assumption.
A)Autocorrelation C) Multicollinearity
B) Heteroscedasticity D) Dummy variable
8. Heteroscedasticity means that
(A) homogeneity cannot be assumed automatically
(B) the variance of the error term is not constant.
104
BASIC ECONOMETRICS STUDY E MATERIAL
(C) the observed units have different preferences.
(D) agents are not all rational.
9. A binary variable is often called a
(A) dummy variable. (B) dependent variable.
(C) residual. (D) power of a test.
10. The difference between the Autoregressive and Distributed lag model is in its lag of
A) Dependent variable placing among explanatory variable
B) Dependent variable
C) Independent variable
D) Dummy variable.
PART-B (5x7=35)
11. a) Describe the objectives and features of Econometrics.
(OR)
b) Explain the Methodology of Econometrics.
12. a). Estimate the unknown parameter β0 and β1 in OLS estimation.
(OR)
b).Prove that the Property of Parameters are Unbiased.
13. a) What are assumptions of estimation of Multiple Regression?
(OR)
b) Derive the β value in terms of matrix as (X‟X)-1X‟Y
14. a)Explain causes and consequence of Multicollinearity.
(OR)
b) Briefly explain reasons for Heteroscedasticity.
15. a).What are the reasons for the introduction of Lag in Regression?
(OR)
b). Find the difference between ANOVA and ANCOVA
PART-C (3x10=30)
Answer any Three questions:-
16. Illuminate the scope of Econometrics.
17. Enumerate the assumption of Classical Linear Regression model.
18. Three related variables take following sets of values: Estimate a regression X1 on X2
and X3
X1: 1 2 3 4 5
X2: 2 1 5 4 3
X3: 3 1 4 5 2
105
BASIC ECONOMETRICS STUDY E MATERIAL
19. What is Autocorrelation? Explain the source, consequence and detection of
Autocorrelation.
20. What are consequences of Model Specification Error?
******************
106
BASIC ECONOMETRICS STUDY E MATERIAL
Questions for Practice
1. Define Econometrics and State its objectives.
2. Enumerate the scope of Econometrics in daily routine life.
3. Explain the significance of Ui , in econometrics equations.
4. What are the differences between economic model and econometric model?
5. What are the features of Logit model?
6. What are the tests for detecting Heteroscedasticity?
7. What are the difference between analysis of variance and analysis of Co variance?
8. The estimated coefficient of different variables and its standard errors are given in the
following table.. Find the‟t‟ values for each variable.
Standard
Variables Co-efficient
error
A0 6.7213 0.7213
A1 5.3121 0.3451
A2 9.0932 0.7344
A3 6.2436 0.5643
A4 4.1256 0.3642
9. Illuminate the cautions in use of Dummy Variables?
10. Explain the methodology of Econometrics with help of Keynesian theory.
11. State the ten assumptions of Ordinary Least Square Method of estimation.
12. Analyse the causes and consequences of Multicollinearity and explain the remedial
measures to eliminate the effects of Multicollinearity.
13. State and prove the Gauss Markov theorem.
14. Derive the β1coefficient value for a multiple regression by using matrix method.
15. The following table includes the Price and Quantity demanded for Toffees:
Quantity (in numbers): 7 5 6 8 1 2
Price (Rs.): 2 4 3 1 6 5
a) Estimate the demand function for Toffees Yi=β0 +β1Xi + Ui.
b) Calculate the R2
16. Prove that, in a given regression Yi=β0 +β1Xi + Ui, the parameters β0 and β1 are linear
and Unbiased.
17. Elucidate the Sources, Consequences and remedial measures to correct the problem of
Autocorrelation and Heteroscedasticity.
18. Econometrics is a separate discipline. Why?
****$$$***
ALL THE BEST
107
You might also like
- Basic - Econometrics - (Major) Sem 4th - Sol - MaterialNo ratings yetBasic - Econometrics - (Major) Sem 4th - Sol - Material130 pages
- Chapter One: Introduction: 1.1 Definition and Scope of EconometricsNo ratings yetChapter One: Introduction: 1.1 Definition and Scope of Econometrics8 pages
- Introduction and subject matter of EconometricsNo ratings yetIntroduction and subject matter of Econometrics23 pages
- IntroduEconometrics - MBA 525 - FEB2024No ratings yetIntroduEconometrics - MBA 525 - FEB2024266 pages
- Instant ebooks textbook Econometrics 1st Edition Jan Tinbergen download all chaptersNo ratings yetInstant ebooks textbook Econometrics 1st Edition Jan Tinbergen download all chapters71 pages
- VTH Semester Notes For Students For Mathematical Economics & Econometrics VTH SemesterNo ratings yetVTH Semester Notes For Students For Mathematical Economics & Econometrics VTH Semester20 pages
- Econometrics 1st Edition Jan Tinbergen - The latest updated ebook version is ready for download100% (3)Econometrics 1st Edition Jan Tinbergen - The latest updated ebook version is ready for download50 pages
- Econometrics Introduction - Unit 1- III BA - 10-08-2021No ratings yetEconometrics Introduction - Unit 1- III BA - 10-08-202113 pages
- 1-Information Technology, The Internet and YouNo ratings yet1-Information Technology, The Internet and You25 pages
- CHARACTER LIST OF KARNAD'S HAYAVADANA_051311No ratings yetCHARACTER LIST OF KARNAD'S HAYAVADANA_0513116 pages
- Interview With Clients - Best Practices For InterviewsNo ratings yetInterview With Clients - Best Practices For Interviews9 pages
- 3 Ms Annual Syllabus Distribution Roaissat M 2020-2021No ratings yet3 Ms Annual Syllabus Distribution Roaissat M 2020-20212 pages
- RRB 16 Sep 1 Ibps Guide - Ibps Po, Sbi Clerk, RRB, SSC - Online Mock TestNo ratings yetRRB 16 Sep 1 Ibps Guide - Ibps Po, Sbi Clerk, RRB, SSC - Online Mock Test47 pages
- CUBED DDM Ñ Benchmark Ñ Combination BookletNo ratings yetCUBED DDM Ñ Benchmark Ñ Combination Booklet11 pages
- SoSe23-IBS Compulsory-Elective Courses Overview 28thfebruary2023No ratings yetSoSe23-IBS Compulsory-Elective Courses Overview 28thfebruary20232 pages
- 3 Magic Exercises To Practice Guitar Harmony100% (1)3 Magic Exercises To Practice Guitar Harmony4 pages
- Results of Assessment in Science With Comparative Analysis 2023 2024No ratings yetResults of Assessment in Science With Comparative Analysis 2023 20242 pages
