at each level before adding the next lag - the tests for model correctness - testing the auto-correlation and testing for significance is done. If after doing this, and finding that the autocorrelation significance is now 0 - once more a lag is added - the AR(P+1) model now is a misspecified model. you found autocorrelation to be significant at p=3. added 4th variable. autocorrelation at p=4 was not significant any more, t-statistic was close to 0. now you still added p=5 --> this is a misspecification.
^ good one cp. Q7. True or False Unit root is solved by using the Dickey Fuller Engle Granger t-statistics to make the series convariance stationary by deriving a meaningful mean reverting level for the series.
false that test is for cointegration…
Q8. Judge if any of the below 2 graphs covariance stationary? Graph1: http://ecoclub.com/news/053/seasonality.gif (exhibits seasonality) Graph2: http://www.statcan.gc.ca/edu/image/ExponentialIcon.gif (exponential curve)
