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: The Nature of Serial Correlation The Consequences of Serial Correlation Testing for autocorrelation Remedies of Serial Correlatio Chapter 9 Serial Correlation Masao Ogaki Department of Economics, Ohio State University November 24, 2007 Masao Ogaki Department of Economics, Ohio State University Chapter 9 Serial Correlation The Nature of Serial Correlation The Consequences of Serial Correlation Testing for autocorrelation Remedies of Serial Correlatio The Nature of Serial Correlation Serial correlation is a time series problem when the Classical Assumption IV is violated. Typical pattern for yearly time series: persisting patterns model used mostly: regression error i satisfies i = ρ i 1 + u i u i : no autocorrelation; regular error term ρ > 0: positive serial correlation (See Figure 9.1) ρ = 0: no serial correlation (See Figure 9.2) ρ < 0: negative serial correlation (See Figure 9.3) Masao Ogaki Department of Economics, Ohio State University Chapter 9 Serial Correlation The Nature of Serial Correlation The Consequences of Serial Correlation Testing for autocorrelation Remedies of Serial Correlatio  ρ  < 1: stable;  ρ  > 1: explosive Here, we rule out  ρ  > 1 Other forms of serial correlation: i = ρ 1 i 1 + ρ 2 i 2 + u i secondorder serial correlation i = ρ i 4 + u i It the data are quarterly, then this is a form of seasonality....
View
Full Document
 Fall '07
 OGAKI
 Economics, Econometrics, Regression Analysis, Null hypothesis, Autocorrelation, Statistical hypothesis testing, Durbin–Watson statistic

Click to edit the document details