This preview shows page 1. Sign up to view the full content.
Unformatted text preview: 7 The TwoVariable Linear Regression Model: Assumptions
Model:
y = b0 + b1x + u Important assumptions:
1. Zero mean of u: The average value of u in the population is 0. That is, E (u ) = 0
o This is not a restrictive assumption since we can always use b0 to
normalize E (u ) to 0
2. Zero conditional mean of u:
o We need to make a crucial assumption about how u and x are related
o We want it to be the case that knowing something about x does not
give us any information about u, so that they are completely unrelated.
That is, that E (u  x ) = 0 , which implies
E (y  x ) = b0 + b1x VER. 9/25/2012. © P. KOLM 8 The TwoVariable Linear Regression Model: Figure
E (y  x ) = b0 + b1x (linear function of x ), where for any x the distribution of y is centered about the regression line represented by E (y  x ) VER. 9/25/2012. © P. KOLM 9 Example (1/2)
A simple savings equation:
savings = b0 + b1income + u Remarks: b0 represents the savings when income = 0 b1 represents the change in wage for a unit change in income when
everything else (u) is held constant How can we think about u?
o u is the unexplained part
o Includes other variables that explain savings such as age, wealth,
savings ethic, etc. VER. 9/25/2012. © P. KOLM 10 Example (2/2) VER. 9/25/2012. © P. KOLM 11 How Do We Fit This Line? (1/2)
Basic idea of linear regression is to estimate the population parameters from a
sample using ordinary least squares
Let {(x i , yi ) : i = 1,...., n } denote a random sample of size n from the population
For each observation in this sample, it will be the case that
yi = b0 + b1x i + ui We call this the data generation process (DGP) VER. 9/25/2012. © P. KOLM 12 How Do We Fit This Line? (2/2) VER. 9/25/2012. © P. KOLM 13 Deriving OLS Estimates
Recall:
y = b0 + b1x + u with the restrictions:
1. E (u ) = 0
2. E (u  x ) = 0
o This implies that u and x are independent, and therefore Cov(x , u ) = E (xu ) = 0
Writing there restrictions in terms of x , y, b0 and b1 , we have
1. E [y  b0  b1x ] = 0
2. E éêëx (y  b0...
View
Full
Document
 Fall '14

Click to edit the document details