Scribd
Upload
16 views3 pages

Week 6 Session 2

The document discusses the implications of non-zero covariance between variables and error terms in econometric models, particularly affecting Maximum Likelihood Estimation (MLE) and Ordinary Least Squares (OLS). It highlights the potential biases in parameter estimates and emphasizes the importance of addressing endogeneity issues in economic studies. The document also outlines theoretical reasons for covariance being non-zero, including reverse causality, omitted variable bias, and inaccurate proxy variables.

Uploaded by

Mostafa Allam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
16 views3 pages

Week 6 Session 2

The document discusses the implications of non-zero covariance between variables and error terms in econometric models, particularly affecting Maximum Likelihood Estimation (MLE) and Ordinary Least Squares (OLS). It highlights the potential biases in parameter estimates and emphasizes the importance of addressing endogeneity issues in economic studies. The document also outlines theoretical reasons for covariance being non-zero, including reverse causality, omitted variable bias, and inaccurate proxy variables.

Uploaded by

Mostafa Allam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 3

Week 6- Session 2

Generalized Method of Moments


Motivation
1. Sometimes Cov ( X , ε ) ≠ 0. Implications for MLE and OLS?
 Information matrix is misleading as off-diagonals will not be
zero. Therefore, trinity tests are not valid. This suggests that the
parameters are not credible (biased).

^β=( X ' X )−1 X ' Y →(1)

Y = Xβ+ ε → (2)

Introduce (2) inside (1)

^β=( X ' X )−1 X ' ( Xβ+ ε )

^β=( X ' X )−1 ( X ' X ) β + ( X ' X )−1 X ' ε

^β=β + ( X ' X )−1 X ' ε


If the Covariance term X ε ≠ 0, ^β ≠ β . This suggests that the parameters
'

are biased!

{
−1
Assume ^β> 0 , Biasness= Upward if ( X X ) −1X ε >0
' '

Downward if ( X X ) X ε <0
' '

Note. In Econometric books, when they writeCov ( X , ε )=X ' ε =0, they mean
no dependence even in higher order moments (nonlinear forms).
2. Do we have a Statistical indicator to capture the non-linear
relationship?
Statistical Point of View
 We have some higher order co-moments (Financial Economics).
 More complex dependence between any two variables might not
be captured.
Econometrics Point of View
 Test Durbin-Hausman-Wu (Journal of Econometrics).

3. How do we know (capture) that Cov ( X , ε ) ≠ 0 ?


Majority of Economics/Finance Studies suspect it through literature and
logical discussions!
Every Papers in Business Field should elaborate a separate discussion
about the Endogeneity issues!
GDP=α+ β Trade+ ε

In the model above: if we suspect that X is affected by Y. therefore the


X is assumed to be affected by ε =(Y −Y^ ).

^ ^
GDP Trade X Exog X Endog
(Y ) (X ) unrelated ¿ Y due ¿Y
2 5 10 -5
3 6 0 6
4 2 1 1
5 1 -5 6

4. Why theoretically Cov ( X , ε ) ≠ 0 ?

1. Reverse Causality (Endogeneity): X values are pre-determined by


the Y values.
2. Omitted Variable Bias. (Ramsey Test): If we did not include an
essential variable in our regression that is already correlated with
the X, it will be captured in the error term. This inflates the error
term and create correlation with this error term (that includes this
variable) and the X.
3. Inaccurate Proxy Variables.
4. Non-Random Sampling techniques.

You might also like