positive serial correlation and lagged dependent variable

The textbook is extremely confusing when it comes to this topic.

Under what circumstances do we have to readjust the regression coefficients in the mode in regards to serial correlationl? I am specifically confused on how this related to having a lagged dependent variable in the regression.


For AR, ARMA or MA models you can incur in model misspecification, so looking at serial correlation indicators (probably up to 30+ levels of lag differences) will tell us if the model requires adjustments. Those adjustments are very simple like adding or removing lagged dependent variables as explicators (i.e. independent variables).