Wouter J. den Haan - CFM-LSE Numerical Summer School


Numerical Summer School - Preparation

Matlab
  • Past experience has shown that students get a lot more out of the course, if they understand Matlab well and have already already quite a bit of experience. The following may help to get you started or improve your Matlab skills.
  • The best way to learn a programming language is to use it. So in preparation to the course, try the following.
    • Write a program (without using any Matlab modules) that
      • (i) simulates data for a least-squares regression problem with multiple explanatory variables,
      • (ii) does the regression and calculates standard errors,
      • (iii) creates a figure that plots what the 1st regressand explains, the 1st & 2nd, the 1st, 2nd, & 3rd, etc.
    • Write a program (again without using any Matlab modules) that
      • (i) simulates data for a VAR,
      • (ii) estimates the VAR (write the program such that you can easily adjust the number of lags), and
      • (iii) calculates the impulse response functions if you use Cholesky decomposition (or just the IRFs associated with the reduced-form shocks)

The essentials: Solving and estimating DSGE models
  1. It is important that you are familiar with the basics of DSGE models. There are many ways to do this, but one reference would be dynamic optimization or equilibrium models .
  2. The Dynare user guide available at dynare user guide
  3. Make sure you know the basics of Bayesian statistics. There are many good online tutoriols (just search for Bayesian statistics and introduction).
  4. Slides on topics related to the course and other topics can be found at teaching notes

Advanced tools: Solving and estimating advanced models
  1. It will be very helpful if you are familiar with perturbation techniques (Dynare) and projection methods. Info can be found at slides on Dynare , official Dynare user guide, and projection methods
  2. Kalman filter (See, e.g., the textbook by Ljungqvist and Sargent)
  3. Some Knowledge of models with heterogeneous agents. Useful articles are