FUNDAMENTALS OF

ECONOMETRICS
DR ABDUL WAHEED
PhD Econometrics
FUNDAMENTALS OF
ECONOMETRICS
Week 3
Lecture 5
 Regression Model
(CLRM)
The Assumptions Underlying the
Method of Least Squares
The regression model was developed first by Gauss in 1821.
It has served as a norm or a standard against which may be compared
the regression models that do not satisfy the Gaussian assumptions.
 Regression Model
(CLRM)
The OLS method is used to estimate and and obtain and .
How close and are to their counterparts in the population or how
close is to the true ?
depends on both and .
Thus, the assumptions made about the variable(s) and the error term
are extremely critical to the valid interpretation of the regression
estimates.
 Regression Model
(CLRM)
The Gaussian, standard, or classical linear
regression model (CLRM), which is the
cornerstone of most econometric theory,
makes 7 assumptions.
 CLRM-Assumptions
ASSUMPTION 1
Linear Regression Model
The regression model is linear in the parameters, though it may or may
not be linear in the variables.
That is the regression model as shown

This model can be extended to include more explanatory variables.

 CLRM-Assumptions
ASSUMPTION 2
Regressors are Fixed or
X Values Independent of the Error Term
Values taken by the regressor X may be considered fixed in repeated
samples.
In the latter case, it is assumed that the and the are independent, that
is, .
 CLRM-Assumptions

 (Unconditional mean)
CLRM-Assumptions
𝒀 𝒊=𝜷 𝟏+ 𝜷𝟐 𝑿 𝒊
 CLRM-Assumptions
ASSUMPTION 3
Zero Mean Value of Disturbance

Given the value of , the mean or expected value of the random

disturbance term is zero.
Symbolically,
 CLRM-Assumptions
Conditional distribution
of the disturbances .
 CLRM-Assumptions
The positive values cancel out the negative values so that their
average or mean effect on Y is zero.
There is no specification bias or specification error in the model
used in empirical analysis. The regression model is correctly
specified.
Specification Error:
Leaving out important explanatory variables,
Including unnecessary variables,
Choosing the wrong functional form of the relationship between
the Y and X variables.
 CLRM-Assumptions
If the conditional mean of one random variable given
another random variable is zero, the covariance between
the two variables is zero and hence the two variables are
uncorrelated.
But if X and u are correlated, it is not possible to assess
their individual effects on Y.
In this situation it is quite possible that the error term
actually includes some variables that should have been
included as additional regressors in the model.
 CLRM-Assumptions
ASSUMPTION 4
Homoscedasticity or Constant Variance of

The variance of the error, or disturbance, term is the same

regardless of the value of X.
Symbolically,
 CLRM-Assumptions
The variance of for each (i.e., the conditional variance of ) is
some positive constant number equal to . Technically, it is
the assumption of homoscedasticity, or equal spread or
equal variance.
Stated differently, that the Y populations corresponding to
various X values have the same variance.
In contrast, the conditional variance of the Y population
varies with X. This situation is known appropriately as
heteroscedasticity, or unequal spread, or unequal variance.
CLRM-Assumptions
Homoscedasticity Heteroscedasticity
 CLRM-Assumptions
ASSUMPTION 5
No Autocorrelation between the Disturbances
Given any two X values, and the correlation between any two
and , is zero.
In short, the observations are sampled independently.
Symbolically,

where and are two different observations.

 CLRM-Assumptions
Simply the disturbances and are uncorrelated. Technically, this
is the assumption of no serial correlation, or no autocorrelation.
This means that, given Xi, the deviations of any two Y values
from their mean value do not exhibit patterns.
If the disturbances (deviations) follow systematic patterns there
is auto or serial correlation.
Autocorrelation: The error term across observations are
correlated with each other. i.e., is correlated with (the
correlation between the successive values of error terms)
CLRM-Assumptions

positive serial negative serial zero correlation

correlation correlation
 CLRM-Assumptions
ASSUMPTION 6
The Number of Observations n Must Be Greater than the
Number of Parameters to Be Estimated and
no Specification Bias

From less observation there is no way to estimate the more

unknowns (parameters).
 CLRM-Assumptions
ASSUMPTION 7
The Nature of X Variables

1. There are no perfect linear relationships among the X

variables. This is the assumption of no multicollinearity.
2. The X values in a given sample must not all be the same.
3. There can be no outliers in the values of the X variable.
 CLRM-Assumptions
No multicollinearity: There is no exact linear relationship
between explanatory variables, is technically known as the
assumption of no collinearity or no multicollinearity.
In short, the assumption of no multicollinearity requires that in
the PRF we include only those variables that are not exact linear
functions of one or more variables in the model.
 CLRM-Assumptions
If all the X values are identical, then and the denominator of
that equation will be zero, making it impossible to estimate and
therefore .
If there is very little variation in X then it is not possible to
explain much of the variation in the Y. The variation in both Y
and X is essential to use regression analysis.
In short, the variables must vary!
 CLRM-Assumptions
The requirement that there are no outliers in the X values is to
avoid the regression results being dominated by such outliers.
If there are a few X values that are, say, 20 times the average of
the X values, the estimated regression lines with or without such
observations might be vastly different.
 CLRM-Assumptions
ASSUMPTION 8
Normality of the distribution
Although it is not a part of the CLRM, it is assumed that the
error term follows the normal distribution with zero mean and
(constant) variance.
Symbolically,
CLRM-Assumptions
Assumption Number Type of Violation
1 Nonlinearity in parameters.
2 Stochastic regressor(s).
3 Nonzero mean of . and
Specification Bias.
4 Heteroscedasticity.
5 Autocorrelated disturbances.
6 Sample observations less than the
number of regressors.
7 Multicollinearity, and
Insufficient variability in regressors.
8 Non-normality of disturbances.
 Theorem
Given the assumptions of the classical linear regression model,
the least-squares estimates possess some ideal or optimum
properties.
These properties are contained in the well-known Gauss–
Markov theorem.
To understand this theorem, consider the best linear
unbiasedness property of an estimator.
 Theorem
On the basis of Assumptions A-1 to A-7, the method of ordinary
least squares (OLS) gives the best linear unbiased estimators
(BLUE).
1. Estimators are linear functions of the dependent variable Y.
The estimators are linear, that is, they are linear functions of the
dependent variable Y. Linear estimators are easy to understand
and deal with compared to nonlinear estimators.
 Theorem
2. The estimators are unbiased.
The estimators are unbiased, that is, in repeated applications of
the method, on average, the estimators are equal to their true
values.
3. In the class of linear estimators, OLS estimators are best.
In the class of linear unbiased estimators, OLS estimators have
minimum variance. As a result, the true parameter values can be
estimated with least possible uncertainty; an unbiased estimator
with the least variance is called an efficient estimator.
 Theorem
In short, under the assumed conditions,
OLS estimators are
BLUE
best linear unbiased estimators
This is the essence of the well-known Gauss–Markov
theorem.