Lect 15 posted (RNy)
Unfortunately, I made an error in the handout for seminar 9 (my deepest apologies). I forgot to add the desired range in the dummy variable tests of structural breakdown in gamma! This leads to, as you will see in the revised handout, not rejecting the null hypothesis in one of the cases. We now get qualitatively the same outcome from the Chow test and the dummy test (which is great news).
Please print out and check the new handout on page 3 and 4, where the dummy tests have been updated.
Herman
PS: To all those (and there are quite a few) who still freak out when they see a Stata printout, you are most welcome to email me and set up a (brief) meeting where I can explain how to read a regression table. My email is herman.kruse@econ.uio.no
PPS: Please come in packs rather than singles (I have exams too).
For seminar 9: Sorry about late message, but there's a typo in the seminar set. The duplicate beta3 is supposed to be a beta4, and then re-enumerate the rest of the dummies such that the last one becomes beta8
For seminar 10: Please note that except for seminar group 3 (Thursdays at 14:15-16:00 with Erling)ALL SEMINARS have one week off, meaning next week there are no seminars. EXCEPT Erling's at Thursday. Check the schedule!
Herman.
Note that there will be no handout for seminar 8 (No Stata-problems). Herman.
After a long break we begin the final batch of lectures on Thursday. I will Lecture on dynamic regression models (remaining Lect 15 stuff) then two separate topics: Forecasting from dynamic models and simultaneity bias of OLS (Lect 16 slide set). The reason for putting them together is that they can be explained with the same example: a small macro model
(RNY).
I am sorry about the confusion I made on the interpretation where I (at best) was imprecise. Here's a clarification.
Basically, what I meant is we are allowed to accept a higher chance of committing a type I (rejection) error. This means, as I wrote, a smaller alpha. So bottom line; we are SO sure the test of significance is rejected (meaning the coefficient IS significantly different from 0) from the evidence we have in this data. In fact, since the p-value is 0.0000, we can claim significance at very low alpha-levels.
Sorry about creating unecessary confusion, hope this clears up everything.
Herman.
In HGL 5.14f, one is supposed to use the delta method (three times!) in order to solve it completely. However, we have decided we simplify this by assuming PRICE=120 000 instead. Then pretend PRICE is constant, so that var(alpha_3*PRICE) = PRICE^2var(alpha_3) etc.
My pardons if you have attempted to reproduce this and got syntax error. In the loop, the ending bracket } should have been on it's own line. Please correct this in your handouts, or print out a new one! Thanks to Sindre for pointing that out to me. The do-file, however, should be correct as is.
If you find any other mistakes in the handouts (or do-files), please don't hesitate to contact me on herman.kruse@econ.uio.no
Handout and do-file for seminar 5 have been posted. Herman
Handout and do-file for seminar 4 have been posted. Note the minor edit on page 1 from the ones handed out: I added the missing } to the loop. Herman
Also about seminar 5, Exercise A: I have re-phrased the text in question A4 to make it possible to understand which data transformations to use. Please just consult the updated Seminar exercise set on that point! (RNy)
Important message for seminar 5: In the data set oilvol.dta, the date variable was not correctly formatted when I transfered the data from excel to stata. Thus, I had to reformat the date-variable. This means you should use the variable "edate" and NOT the variable "date" (as it would return only errors!) Herman
Handout and do-file for seminar 3 have been posted. H
Tomorrow: Wrap up multivariate model and extend it to dummy variables as indicated in slides. Also posted a short note with some important algebra for Frisch-Waugh theorem
Next lecture is already tomorrow morning. Some of the material is posted in the usual way.
Handout and do-file for seminar 2 have been posted. H
Slides to Lect 10 posted (and a note with some algebra that we will go trough)RN
I have posted a note with the OLS normal equations and estimates that we did in class today (Lect 9): We derived eq (1), (2) and (3). RNy
Harald has written a lecture note based on, and extending, his two lectures (#7 and #8 in our list). It is posted under Lecture 8 (RNy)
Answ. suggestion to that extra exercise as well (rny)
The handout and .do-file for seminar 1 have both been added. Herman.