This document outlines the 7 key assumptions of the classical linear regression model:
1. The regression model is linear, correctly specified, and has an additive error term.
2. The error term has a zero population mean.
3. All explanatory variables are uncorrelated with the error term.
4. Observations of the error term are uncorrelated with each other.
5. The error term has a constant variance.
6. No explanatory variable is a perfect linear function of any other.
7. The error term is normally distributed, though this is not required.
The rest of the course deals with what happens when these assumptions do not hold.
Copyright:
Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online from Scribd
Download as pdf or txt
Assumptions of The Classical Linear Regression Model
This document outlines the 7 key assumptions of the classical linear regression model:
1. The regression model is linear, correctly specified, and has an additive error term.
2. The error term has a zero population mean.
3. All explanatory variables are uncorrelated with the error term.
4. Observations of the error term are uncorrelated with each other.
5. The error term has a constant variance.
6. No explanatory variable is a perfect linear function of any other.
7. The error term is normally distributed, though this is not required.
The rest of the course deals with what happens when these assumptions do not hold.
This document outlines the 7 key assumptions of the classical linear regression model:
1. The regression model is linear, correctly specified, and has an additive error term.
2. The error term has a zero population mean.
3. All explanatory variables are uncorrelated with the error term.
4. Observations of the error term are uncorrelated with each other.
5. The error term has a constant variance.
6. No explanatory variable is a perfect linear function of any other.
7. The error term is normally distributed, though this is not required.
The rest of the course deals with what happens when these assumptions do not hold.
Copyright:
Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online from Scribd
Download as pdf or txt
0 ratings0% found this document useful (0 votes)
130 views7 pages
Assumptions of The Classical Linear Regression Model
This document outlines the 7 key assumptions of the classical linear regression model:
1. The regression model is linear, correctly specified, and has an additive error term.
2. The error term has a zero population mean.
3. All explanatory variables are uncorrelated with the error term.
4. Observations of the error term are uncorrelated with each other.
5. The error term has a constant variance.
6. No explanatory variable is a perfect linear function of any other.
7. The error term is normally distributed, though this is not required.
The rest of the course deals with what happens when these assumptions do not hold.
Copyright:
Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online from Scribd
Download as pdf or txt
You are on page 1/ 7
CHAPTER4:THECLASSICALMODEL
Page1of7
OLS is the best procedure for estimating a linear regression model only under certain assumptions.
The word classical refers to these assumptions that are required to hold.
Assumptions of the Classical Linear Regression Model:
1. The regression model is linear, correctly specified, and has an additive error term. 2. The error term has a zero population mean. 3. All explanatory variables are uncorrelated with the error term 4. Observations of the error term are uncorrelated with each other (no serial correlation). 5. The error term has a constant variance (no heteroskedasticity). 6. No explanatory variable is a perfect linear function of any other explanatory variables (no perfect multicollinearity). 7. The error term is normally distributed (not required).
Essentially, the rest of the course deals with what happens when one or more of these assumptions do not hold and what we can do to remedy the situation.
CHAPTER4:THECLASSICALMODEL
Page2of7
Assumption 1: Linear Model, Correctly Specified, Additive Error
The model with k explanatory variables:
= [ 0 +[ 1 X 1 +[ 2 X 2 + +[ k X k +e
.
The regression model is linear in the coefficients.
The random error is additive.
The model also contains the right independent variables.
CHAPTER4:THECLASSICALMODEL
Page3of7
Assumption 2: Error term has a population mean of zero
= [ 0 +[ 1 X 1 +[ 2 X 2 + +[ k X k +e
In terms of notation: E(e
) = u.
The error term itself cannot be observed. But lets suppose the mean of e
is 10.
So we could just add and subtract 10 to the model to force the error term to have a mean of zero:
= ([ 0 +1u) +[ 1 X 1 +[ 2 X 2 + +[ k X k +(e
-1u).
CHAPTER4:THECLASSICALMODEL
Page4of7
Assumption 3: Explanatory variables uncorrelated with error term
This assumption requires that the observed values of the independent variables be determined independently of the error term.
In terms of notation: Co:(X
, e
) = u.
Lets say that X 1 is correlated with the error term. OLS will not be able to estimate the slope of X 1 separately from movements in the error term. OLS will give credit to X 1 for movement in Y that is due to the error term.
CHAPTER4:THECLASSICALMODEL
Page5of7
Assumption 4: No serial correlation
Error term observations are drawn independently (and therefore not correlated) from each other. When observed errors follow a pattern, they are said to be serially correlated or autocorrelated.
In terms of notation: Co:(e
, e ] ) = u.
Assumption 5: No heteroskedasticity (homoskedasticity)
The error term observations come from the same probability distribution with a constant variance. If the variance of the distribution of the error term changes for each observation or range of observations then we have whats called heteroskedasticity.
In terms of notation: Ior(e
) = o 2 .
CHAPTER4:THECLASSICALMODEL
Page6of7
Assumption 6: No perfect multicollinearity
None of the independent variables have a perfect linear relationship (perfect collinearity or multicollinearity) with any of the other independent variables. If they do, OLS cannot separately estimate the coefficients for them.
Even is the variables are not perfectly collinear, there can still be a problem. If two or more variables are highly correlated, multicollinearity exists and could be problematic. CHAPTER4:THECLASSICALMODEL
Page7of7
Assumption 7: Error term is normally distributed
Have already assumed that the observations of the error terms are drawn independently (Assumption 4) from a distribution that a mean of zero (Assumption 2) has a constant variance (Assumption 5). Now, we specify the shape of the distribution. The normal distribution is assumed.
