I don’t have time to fix this one before the exam tomorrow, but someone just pointed out to me an error in Exercise 2 of lesson E.2.2. In that exercise, I cancelled a term of (1-B) from either side of a time series model. However, since the solution of 1-B=0 is not strictly greater than 1, the inverse operator for 1-B does not exist. Therefore the (1-B) factors should not be cancelled, and the series is not stationary.
I believe that this issue is treated correctly at other points in the seminar where it appears, but somehow this error in E.2.2 never came to my attention before.