Messages
The solution of Question B.8 of the first seminar can be found on the Hendry-Nielsen at page 100 (Ch. 7).
R
As an extra under "Lectures" I have posted David Hendy's slides from his Haavelmo lecture.
R
Have posted the two things I promised in seminar 6.
R
Sorry or the missing upload yesterday. I had misunderstood,
Note that Andres slides are in the folder labeled Tex
R
We will finish talking about the material in the slide set labelled Lecture 10 +11.
Then a short section about What next (Lecture 12, 13 and Sem 6)
R
Sorry for the missing fat greet letters in an already notation heavy lecture.
Have corrected most of them.
Start p. 36 next week.
R
Tomorrow we continue our venture into unit-root econometrics. Hopefully there will be short-run, not only long-run, benefits from attending.
-R
Dear Modellers,
Tomorrow is the fifth computer class, and we have tons of very interesting topics on the agenda; IV-estimation, over identification and tests thereof, FIML, SUR, non-stationarity, spurious regressions and cointegration! We won't have time for all, but we continue with the rest on CC#6!
-André-
There will be an additional lecture on wednesday Oct. 21th at 14:15 in Aud 3 where we will have a walkthrough of the mid-term paper.
I have posted Lecture note 6 with several additional points about IV and GMM. Mostly from Davidson and MacKinnon's book.
R
Lecture 8 is a continuation of Lecture 7. We will talk more about the menu of estimation methods that we can choose from for IDENTIFIED structural equations.
For good measure, I have also posted the first slide set for the next topic (numbered as Lect. 9), where we move into the territory of NON-STATIONARY TIME SERIES (an exciting place, not like Camelot!)
R
Andre has sent a new version of the yesterday's zip file. With handout version included.
R
Computer class, deadline for obligatory..... what more can you hope for?
New lecture tomorrow morning. If you have any energy left: Come to Aud 2 and hear more about the truth of identification and IV estimation!
R
Hi
Sorry about the 11 am/11 pm confusion. The intended deadline was in the evening. The opening hours of the Fronter-room will be changed to 11 pm (23:00).
R
Dear all,
Tomorrow is CC # 4! We will start by covering the topics we didn't have time for at CC # 3. In addition to that, we shall talk about some other very important issues, including identification, simultaneity bias and IV-estimation. We will, among other things, do some MC-simulations to get a grip of the "size" of the simultaneity bias, and we will also demonstrate how a "proper" instrument will fix the problem! This leads up to the next CC (CC # 5), where we will continue along the same lines and talk about weak identification, tests for overidentification and the estimation of simultaneous equation systems by full information maximum likelihood techniques! So, as you understand, very interesting and important topics ahead! Hope to see many of you tomorrow!
-André-
Dear all,
On FRIDAY 8am (NOTE TIME), we shall finish the stuff we didn't have time to cover on CC#2. We will also cover single equation including model typology (See Lecture note 4 for example), Gets modelling and different concepts of exogeneity, reviewing and using the concepts of Lecture 5 and 6 and Seminar exercise 3. The slides (both presentation and handout format), along with the different datasets and batchfiles, we will use are contained in the zip-file CC#3.
See you on Friday morning!
Andre
for tomorrow have been posted. As you can see, they are not so many, so we will start with invariance and super exogeneity from the Lecture 5 slide set.
R
After quite a lot of lecturing about the importance of dynamic stability over the two last weeks, I can now see that the wording in excerise 3 may lead to misunderstandings,
So: In exercise 1ai), what we are after is "stability" in the meaning of "recursive stability of coefficients". Not conditions for dynamic stability and stationarity. Perhaps "constancy of coefficients (parameters)" would have been better.
Same in b) and 1 c) and ii).
Hint: Use the capability of recursive plots of individual coefficients, although that will give a lot of output to digest. Maybe easier to interpret recursive stability "tests": 1- step Chow tests for example, or the recusively estimated residual standard error!
R
Seminar 3 is already next week. It has now been posted!
R
OxMetriics is listed along with other UiO software for students and teachers at:
https://app.uio.no/programvare/index.html
Direct link:
https://app.uio.no/programvare/produkt/6417.html
You can install the program on you own latop. The avaiiable version includes a basic version of the module for Monte Carlo simulation that Andre uses. It includes FULL VERSIONS of the other program features that we use, and that you will need for the obligatory assignment.
R
Lagging a little bit with the answers to the DIYs from last week. But here is the newest slide set.
R