This preview shows pages 1–4. 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 Document
Unformatted text preview: 1 Lecture Notes 6 1. Autocorrelation (or Serial Correlation) 1. Definition A problem usually for time series data The error terms, u t , are supposed to be random However, they are related by time The error terms are correlated Using the covariance term, then j i for u u Cov j i The correlation between u t and u tr is called an autocorrelation of order r. Usually econometricians use k, but then it is confusing Could result from the impact of a missing x variable Parameters estimates are unbiased, but estimated standard errors are biased. F and tstatistics are invalid t kt k t t t u X X X Y 3 3 2 2 1 If you plot the residuals, then 2 Autocorrelation of degree 1 is very common Denoted AR(1) t kt k t t t u X X X Y 3 3 2 2 1 Where t t t v u u 1 Rho, , is between 1 and 1 v is error term Covariance function shows how two variables vary together Correlation comes from this function Y E Y X E X E Y X Cov , (i) If X = Y, then this becomes the variance for X X Var X E X E X E X X E X E X X Cov 2 , (ii) Remember, E(X) is the expected value and average (iii) Remember, E(u t ) = E(u t1 ) = 0 Autocorrelated error terms are still equal to zero on average 1 1 1 1 , t t t t t t t t u u E u E u u E u E u u Cov 3 I can pull out (i.e. factor) the rho from the expression and t t t v u u...
View Full
Document
 Fall '11
 BLAIR

Click to edit the document details