Current Assignments for Econ 623

  1. Read The Free Installment Puzzle by John Rust and Sungjin Cho
  2. Finish reading Chapter 16 in Greene, on Maximum Likelihood.
  3. Do Problem set 4 Due Tuesday November 29th at start of class
    To do the problems you will need
    regression.out (for problem II-I)
    adaptreg.out (for problem III)
    x.dat (for problem V)
    Also the Matlab code below might be helpful (though there is no requirement to use Matlab, you can use Excel, Stata, or any software you want to do the calculations requested in Problem set 1)
    kdentest.m (test program that compares MLE and non-parametric kernel density estimates of a normal density)
    kdensity.m (computes kernel density estimator at a point x)
    denplot.m (computes kernel density estimator at 1000 points over an interval [a,b] and plots the resulting estimated density at these points)

  4. Solutions to Problem Set 4

  5. It may help in doing problem set 4 to see the Answers to Problem Set 2 (2007) There a maximum likelihood problem in this problem set and also a proof of the delta theorem.
  6. It may help in preparing for the final exam to look at the Answers to 2008 Final Exam There are some time series questions including a question about autorgression, and a maximum likelihood question and a discussion of block diagonality of the Information matrix that may help in doing problem set 4.
  7. It may help in doing problem set 4 to see the Answers to Problem Set 1 (2010) There are questions on convergence of non-stochastic and stochastic sequences and additional maximum likelihood problems here
  8. I have also posted answers to Answers to Problem Set 2 (2010) This problem set was on Bayesian inference which we may not have time to cover this semester in Econ 623.

  9. Code I wrote in class today to compute a parametric versus adaptive maximum likelihood estimates of the mean (location) parameter of a distribution when the data are really normally distributed.
    lln.m (Gaussian log-likelihood function)
    nlln.m (adaptive log-likelihood function based on a density estimated from 1st stage regression residual and a kernel density estimator)
    data_gen.m Run this program to generate a simulated data set and then plot the Gaussian and adaptive non-parametric log-likelihood functions and to estimate the unknown parameter mu by a) sample mean, b) maximization of Gaussian log-likelihood function, and c) maximization of the adaptive log-likelihood function computed from the first stage regression residuals
    illustrative plot of the Gaussian vs adaptive log likelihood functions

  10. Take Home Final Exam (due in my mailbox in 3105 Tydings Hall by 5pmm Tuesday, December, 20th)
  11. dice-draws.txt data file needed to answer 1-F of the final exam.

  12. Solution to Q1 on the Final
  13. Solution to Q3 and Q4 on the Final

  14. Distribution of Econ 623 total course scores (100 point scale) PDF   CDF
  15. Distribution of Econ 623 first half course scores (50 point scale) PDF   CDF
  16. Distribution of Econ 623 second half course scores (50 point scale) PDF   CDF
  17. Distribution of Econ 623 final exam scores (600 point scale) PDF   CDF
  18. Distribution of Econ 623 Problem Set 4 scores (400 point scale) PDF   CDF


Send questions/comments to: jrust@gemini.econ.umd.edu