3. (a) (b) (c) Explain what is meant by serial correlation in the context of eco
ID: 1143969 • Letter: 3
Question
3. (a) (b) (c) Explain what is meant by serial correlation in the context of econometric models. 5 What are implications of serial correlation for the OLS estimators? 5 Consider the following model, estimated by OLS D, =-0.63 + 0.32P, + 0.57P-i +0.1 ID-1 R2=0.97 DW= 1.48 where D is dividends and P is profits annually for a ABC company in Emillion for 1913-1946 Apply the Durbin Watson test for autocorrelation, clearly stating the null and the alternative hypotheses and comment on your results. 2.5 (i) The LM test for serial correlation for this equation yields 2(1)-2.45 . Interpret the results, explaining application of the LM test and clearly stating the null and the alternative hypotheses in this context. Also explain which tests in (i) and (ii) you prefer, giving your reasons. 2.5 (ii)Explanation / Answer
(a) Serial correlation in OLS models essentially occur when the error terms between subsequent observations are correlated. This causes the problem of serial autocorrelation.
(b) In case of autocorrelation this will mean that standard errors are inflated and so this has the impact of insignificant t stats and so the slope coeffcients are insignificant. Autocorrelation will also increase the R squared artifically.
(c) (i)The Durbin Watson D here is given by 1.48. The null hypothesis here is given by H0: 0 autocorrelation and H1: positive autocorrelation. Thus the formula for the Durbin Watson test D=2(1-P) where P is the autcorrelation. So in this case there is weak positive autocorrelation given the value of the D.
(ii) The lagrange multiplier test assumes a null hypothesis of no correlation and a alternative of existence of correlation. The chi squared value of 2.45 means that we reject the null hypothesis and so there is autocorrelation. Of the 2 tests the Durbin Watson test is always preferred.
Related Questions
drjack9650@gmail.com
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.