Economics 7393
Time Series I
Spring 2016


 


 

I will post some notes and programs here that I hope will be useful.

 

Bentís Time Series Notes from Brown

 

NOTE, April 14th: We continue with the Kalman Filter today and we then move to Cointegration. (I am not sure if the get to ARCH and GMM.)

 

COURSE REQUIREMENT (unless you have made other arrangements): Download at least 3 time series. (Real GDP, investment, and consumption from the BEA could be it). Test for unit roots, estimate univariate AR and ARMA model (use Kalman Filter), estimate VAR, estimate Johansen Cointegration model. You can do it together, but write up the results on your own.

 

Some GAUSS programs that I have not use for some years,if you invest a little time, I can help figure out how to use them again

 

Kalman Filter (includes a VAR estimator)

Johansen Cointegration

 

(I am sure you can find some more slick programs on the WEB, but for these programs I can tell you what goes on).

 

Homework 1. Download 3 series. Test for unit roots. Estimate univariate AR models.

 

(Note: you can use the software you want, although I prefer GAUSS (MATLAB is very similar), Stata has routines, but you cannot see what it does, but if you want to spend your life doing non-structural micro, you may want to use this. Also, the homework is supposed to part of the course requirement; i.e., you can use the same 3 series and add analysis as we go along.

 

Homework 2. Due April 21. Continue with the variables you downloaded. Estimate a VAR and plot some impulse response functions. (You learn more by doing it using vectors and matrices, but it is also OK to use built-in software.) Estimate ARMA models for each series (experiment with the number of AR and MA terms and argue which is the best specification.)

 

Homework 3. Due May 9. Continue with the variables you downloaded. Estimate co-integration vectors. The best answers uses the Johansen procedure (I have posted the program in Gauss, but I expect it to be readily available for Matlab/Stata etc.) which perform the likelihood ratio test for the number of cointegrating vectors. Extra points for testing a hypothesis about the error correction terms (the alpha matrix).

 

Note: If you want another source for this material, I looked at Davidson and MacKinnonís econometrics text and that is much more to the point than Hamilton.

 

The requirements for the course is the completion of these three homeworks (unless you have made other arrangements).