Outline _ Simple Regression
Outline _ Simple Regression
Third Year
Economics Department
2024/2025
Pre- requisites
4
Outline
Topic Reference
1. Simple regression model Gujarati (chapters 1-6)
2. Multiple regression model Gujarati (chapters 7,8)
3. Dummy variable regression model Gujarati (chapter 9)
Yi = β1 + β2Xi + ui
ˆui denotes the (sample) residual term. Conceptually ˆui is analogous to ui and can
be regarded as an estimate of ui. It is introduced in the SRF for the same reasons
as ui was introduced in the PRF.
Least squares method minimizes the sum of squares of
residuals (deviations of individual data points from the
regression line)
ˆui = (Yi − Yˆi)
ˆu2i = (Yi − Yˆi)2
= (Yi − βˆ1 − βˆ2Xi)2
THE ASSUMPTIONS UNDERLYING THE
METHOD OF LEAST SQUARES
1- It contains the relative variables according to the economic theory , for example
the demand function must contains the price of the commodity and the price of other
goods. for example, if we ignored the variable of the price of the other goods, so the
model will not be correctly specified.
2- It uses the correct functional form, for example
Suppose the “true’’ or correct model in a cost-output study is as
follows:
Marginal costi = β1 + β2 outputi + β3 output2i + ui (1)
but we fit the following model:
Marginal costi = α1 + α2 outputi + ui (2)
So, model (2) uses the wrong functional form, so model (2) is not
correctly specified, but model (1) is correctly specified.
THE CLASSICAL LINEAR REGRESSION MODEL: THE
ASSUMPTIONS UNDERLYING THE METHOD OF LEAST
SQUARES
• If there is perfect linear relationship between X2 & X3 in this
case, we can not estimate the regression model.
• So, the classical linear regression model assumes that there is no
perfect multicollinearity or no perfect linear relationship
between the explanatory variables to be able to estimate the
regression model or to estimate the parameters
Note that: this assumption related to the linear relationship bet.
explanatory variables
So, if there is nonlinear relationship we can estimate the regression
model and there is no any problem.
PROPERTIES OF LEAST-SQUARES ESTIMATORS:
THE GAUSS–MARKOV THEOREM