This preview shows pages 1–7. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentThis preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Omitted Variables Bias Charlie Gibbons Economics 140 October 18, 2009 Outline 1 The problem 2 The bias 3 An example 4 Conclusion The problem Omitted variables bias is the biggest problem in econometrics. Suppose that the true model of a process is Y i = + 1 x i + 2 x i + i , but you run Y i = + 1 x i + i , where i = i + 2 x i . Here, you omitted z i , an important predictor of Y i . We use different Greek letters for the estimated parameters because we are estimating a different (biased) model. The problem Omitted variables arise in two circumstances: 1 You chose the wrong model. 2 The necessary variable isnt measurable ( i.e. , ambition) or isnt in your data set ( i.e. , IQ). The bias Suppose that you have a true model Y = + 1 x 1 i + + k x ki + k +1 z i + i . Now suppose that you omit the variable z i and instead run Y = + 1 x 1 i + + k x ki + i . The bias Imagine that you knew z i and you run this regression: z i = + 1 x 1 i + + k x ki + i , a regression of your omitted variable on all the variables that you did include....
View
Full
Document
 '08
 ABDUS,S.

Click to edit the document details