Messages
If you are be interested in for example 14 days work, mainly working with building up new data-input framework for one of Statistics Norway's operstional models, contact me with a mail, or at the office.
Ragnar N
It has been a pleasure working with you during a very hectic autumn semester.
As I said in the last lecture, one good advice for what to review is:Seminar exercises, not forgetting the obligatory assignement, and past exams (all of which are relevant, but I think that you can make a sensible “ordering” in the light what we have been working with).
Some of you have asked about answer suggestions to the postponed exam in 2016, which is not on the Department's official pages.
I have found an annotated (“kommentert”) version on my system, which is not very “clean”, but which should be of help , for example when it comes to Exercise C (which by the way is inspired by an example in Engle's and Granger's famous Econometrica paper (top notch!) so that you pride yourself at being at very high level here !).
To save time, I have posted it directly on the semester page (last posting in the Seminars section).
Ragnar
The time has come for the last lecture.
Today I will finish cointegration (Lect 11+12 slide), with an emphasis on the very important point about doing a valid statistical test of the null hypothesis of absence of cointegration (avoid the pitfall of spurious regression), and within that topic I will focus on the simplest case of a single cointegration vector.
Then a few remarks about multiple cointegration (identification and a super quick look at the practical side of testing rank > 1 in the program).
After that I will answer questions.
If there is time left, I will then mention a hew highlights from the theory and practice of forecasting economic time series. (Lecture 13 slide set).
Ragnar
Since there was a lot of algebra in parts of Seminar 5, and not much time, we have prepared some answer notes that hopefully can be an aid when you review this.
R
Just posted slide set about cointegration for lect 11 and 12.
Today we will do a litte more of ADF (agumented Dickey-Fuller) testing of the unit root hypothesis, before moving on to cointegration theory.
R
1. Demonstrate SVAR and recursive model for the Fulton data
2. Show equivalence of GIVE and 2SLS (for the Fulton data)
3. Practice testing the unit-root hypthesis. Reference will then be last part of Lecture 10/11 slide set..
R
After super-brief remark about recursive models being equivalent to Cholesky decomposition of VAR residuals (Lecture 9), we will start doing non stationary econometrics.
Rember that we have been placed in Aud 4 tomorrow (don't go on a Random Walk)
R
Hi all
I worked long and hard to give you all individual comments, so you should learn how to access them:
Go to Fronter and enter the folder where you handed in your term paper (It is in your seminar room under "Innlevering")
Next to your file (your .zip if you handed in as .zip, or on your .pdf or .docx if you used those formats), there is a small downward-pointing arrow. Click that one, and you will get a small menu. On the bottom of the menu, there is an option called "Kommentar" - click that one.
You should be redirected to a new page, where you will find an empty comment section and it says "approved". There is another file there, which has the same name as your hand-in with the extension _corrected (or _rettet if you wrote in Norwegian). That is a commented version of your term paper that you should definitely look at.
Herman.
Just posted an annoted version of exercise set with answer notes.
R
Tomorrow there will be an extra lecture to go through the obligatory exercise.
See the schedule for more information
Tomorrow: Quick repeat of OLS simultaneous equation bias from CC5 on Monday. Then estimation methods for identified equations in SEMs. Then a brief encounter with another identification problem: for VAR impulse respones (Lect 9 slides), and how the solution to the problem is closely related to recursive models (also know as "conditional +"marginal models" ).
R
You will find the result in Fronter. If your paper is approved you are not going to hand in a new paper.
If your paper is not approved you will get an e-mail to your UiO-email at 24 October 2017 with information regarding the second attempt.
If you have questions, please contact the Department of Economics.
We will first do some more Autometrics work with the two datasets from CC4.
Then work with SEMs using the Fish market data set from the book (on our semester page for several weeks ) and a new data set that I have prepared for you!
Ragnar
is about the identification issue for SEMs, and IV/2SLS estimation of identified SEM equations. Lecture 9 is about structural VARs. The common ground between the two is recursive models, which we mention in lecture 8.
R
Finish presentation of exogeniety concepts (Lecture 6 slide set), and then follow up Monday's computer class: machine learning and automatic variable selection (Lecture 7 slide set).
Ragnar
should be OK now. Use it as one of the examples on Thursday.
R
We will tackle two practical modelling tasks today. Using what we have learned so far.
R
Because the obligatory assignment replaces "Seminar 3" there will be no exercise seminars next week. The scedule on the semester page has been updated to fit this fact.
However, there will be a joint "seminar" where Ragnar will "go through" the obligatory assignment with you. This will take place after the grading of your work has been finished ,and information about time and place will therefore come later.
The CC on Monday, and the Lecture on Thursday will go ahead as normal (but is there a normal?).
Meanwhile: Trust all is going well with your obligatory work.
Herman and Ragnar
You will find the compulsory term paper on the semester page.
Program for CC
1. Expand a little on "MC Experiment 1" from CC2
2. MC-2: Hurwitz bias
3. VAR and ADL modelling, with reference to Lecture 5, and using the fish market data set as example
Ragnar
Hi!
My apologies. I made a few mistakes in the calculation of the delta method for the first seminar group. A correction note has been added under the seminar-section. As it turns out, we even find exactly the same variance as the program gives!
Herman
PS: All corrections you need to make in your notes are marked in red.