Welcome to our course web site for RECO 777 "Bayesian Data Analysis".

Administrative Stuff

• Login on to the CABNR network / Accessing your student folders
   
• Link to the library server (to use Matlab)
  Use your UNR login name and password. Then double-click on the "Matlab 2008r" icon.
   
• Syllabus & Schedule
   
• Textbook links
   
  • Publisher's web site for Christian Robert's "The Bayesian Choice"
  • Web materials for Gary Koop's "Bayesian Econometrics"
  • Web materials for Koop/Poirier/Tobias' "Bayesian Econometric Methods"
     

Matlab Links

• Tutorials for Matlab and Matrix Algebra, plus Matlab related web links

Lecture Material

• Principles of Bayesian Analysis / Normal linear regression with conjugate priors
  • Lecture Notes
  • Matlab scripts and functions
    • normal_data (simulation script to generate data for the normal regression model)
    • normal_ols (script to estimate normal regression model via OLS)
    • ols (function; derives OLS estimator)
    • normal_mle (script to estimate normal regression model via MLE)
    • normal_llf (function; defines likelihood function)
    • normal_Bayes_conjugate (script to estimate normal regression model via Bayes with conjugate priors)
    • normal_Bayes_conjugate_plots (script; generates prior and posterior plots)
    • normal_Bayes_conjugate_constrained (script; estimates "flawed" normal regression model)
    • normal_Bayes_conjugate_compare (script; compares correct and flawed models via marginal likelihoods and Bayes Factor)
    • epanech_klaus (function; estimates densities using the Epanechnikov kernel estimator)
       
• Normal linear regression with independent priors / Gibbs Sampler / Sampler Diagnostics
  • Lecture Notes
  • Matlab scripts and functions
    • normal_Bayes_independent (script to estimate normal regression model via Bayes with independent priors)
    • gs_normal_independent (function with Gibbs Sampler to go with the above script)
    • inverse_gamma_plots (script to examine the properties of the ig density)
    • klausdiagnostics (function with diagnostic tools for MCMC samplers)
    • wage_example (script using the normal linear regression with independent priors to estimate Mroz's (1987) wage data)
      • wage data (Excel format - see Matlab tutorial for loading instructions)
    • wage_example2 (same script as wage example, except splitting betas into 2 blocks to show effect on Gibbs Sampler efficiency)
    • gs_normal_independent2 (GS to go with wage_example2);
    • cholcov (utility function to avoid errors due to Matlab's overly stringent symmetry check for matrices - place in your "functions" folder)

       

• Importance Sampling, Accept-Reject Sampling
  • Lecture Notes
  • Matlab scripts and functions
    • IS1 (main script to illustrate importance sampling, based on KPT Ex. 11.4)
    • AR1 (main script to illustrate Accept-reject sampling)
    • trnd_klaus (function to draw from the univariate t-density)
    • tpdf_klaus (function to evaluate the univariate t-density)
      
       
• Posterior Predictive Densities, Posterior Predictive P-values
  • Lecture notes
  • Matlab scripts and functions
    • predictions (main script to illustrate PPDs)
    • ppp (main script to illustrate ppp's)
       

• SUR Model, Highest Posterior Density Intervals
  • Lecture notes
  • Matlab scripts and functions
    • sur_data (script to generate simulated data for the SUR model)
    • sur (main script to illustrate SUR model)
    • hdpi (script to illustrate HPDIs)
    • gs_sur (function; Gibbs Sampler for the SUR model)
    • var_vec (utility function - turns a symmetric matrix into a vector with unique elements)
    • vec_var (reverses above process)
    • klaus_hpdi (function; estimates HPDI bounds for unimodal densities)
    • outage (data set for outage cost example, Excel format)
    • sur_outage (SUR script for outage example)
    • hpdi_outage (hpdi script for outage example)
      
       
• Model Comparison: Savage-Dickey, Gelfand-Dey, and the Chib Method of Computing the Marginal Likelihood
  • Lecture Notes
  • Matlab scripts and functions
    • sddr (script illustrating the SDDR method using simulated SUR data)
    • sddr_outage (same script using outage data)
    • sur_chib (script to start Chib's method to compute the mLH for the SUR model)
    • gs_sur_chib (function executing Chib's method for the SUR model)
    • iwishpdf (utility function, evaluates the ordinate of the IW density)
      
       
• Hierarchical Normal Regression Model, Data Augmentation
  • Lecture Notes
  • Matlab scripts and functions
    • HNRM_data (simulated data for the hierarchical normal regression model)
    • HNRM_v1 (main script for the fully augmented hierarchical normal regression model)
    • gs_HNRM_v1 (Gibbs Sampler for the fully augmented hierarchical normal regression model)
    • HNRM_v2 (main script for the partially augmented hierarchical normal regression model)
    • gs_HNRM (Gibbs Sampler for the partially augmented hierarchical normal regression model)
    • HNRM_chib (starter script for the Chib method for the partially augmented HNRM)
    • gs_HNRM_chib (Gibbs Sampler implementing the Chib method for the partially augmented HNRM)
    • outage_50firms (re-configured outage data, excel format)
    • HNRM_outage (application of the HNRM using outage data)
    • HNRM_outage_chib (starter script for the Chib method using outage data)
    • HNRM_outage_nocovs (script for HNRM outage model without covariances)
    • gs_HNRM_nocovs (GS to go with that script)
    • HNRM_outage_nocovs_chib (starter script for no-cov. model, Chib method)
    • gs_HNRM_nocovs_chib (GS to implement Chib method)
    • HNRM_outage_singleRE (script for HNRM outage model with single random effect)
    • HNRM_outage_singleRE_chib (starter script for single-RE model, Chib method)
    • HNRM_predict (script generating posterior predictive densities for the HNRM model using outage data)
    • HNRM_naive (HNRM ignoring random effects - essentially a basic normal regression model)
    • HNRM_naive_predict (prediction script for the naive model)
      
       

• Nonlinear regression model / Metropolis-Hastings algorithm

  • Lecture Notes

  • Matlab scripts and functions

    • CES_data (script generating simulated data for the CES model)

    • CES_mle (Estimation of CES model via MLE)

    • CES_llf (likleihood function for MLE estimation)

    • CES_rwc_prep (prep script for RWMH approach)

    • CES_rwc (RWMH script with refined tuner variance)

    • gs_CES_rwc ( Gibbs Sampler function to go with the RWMH scripts)

    • CES_ic_v1 (script for Independence Chain MH, basic version)

    • gs_CES_ic_v1 (Gibbs Sampler to go with the above)

    • CES_ic_v2 (script for Independence Chain MH, reflected version)

    • gs_CES_ic_v2 (Gibbs Sampler to go with the above)

    • CES_mode (sub-routine to estimate the mode of the conditional posterior kernel)

    • mvtpdf_klaus (utility function to evaluate the ordinate of a mvt density; used in the IMH)

    • mvtrnd (utility function to draw from the mvt density)

  • Referencences

    • Chib, S. and E. Greenberg. 1995. Understanding the Metropolis-Hastings algorithm. The American Statistician 49: 327-335.

    • Chib, S. and I. Jeliazkov. 2001. Marginal likelihood from the Metropolis-Hastings output. Journal of the American Statistical Association 95: 270-281.
       

• Working with latent Dependent Variables: The Probit and Tobit Models

  • Lecture Notes

  • Matlab scripts and functions

    • probit_data (data generating script for the probit model)

    • probit (main script for the probit)

    • probit_Fair (script for probit model with Fair's (1978) affairs data)

    • probit_Fair_predict (prediction script for the Fair example)

    • gs_probit (Gibbs Sampler for the probit)

    • tobit_data (data generating script for the tobit model)

    • tobit (main script for the tobit)

    • tobit_adoption (script for tobit model with horse adoption data)

    • tobit_adoption_predict (prediction script for the horse example)

    • gs_tobit (Gibbs Sampler for the tobit)

    • tnormrnd (utility function to draw form the truncated normal)

    • tnormpdf (utility function to evaluate the truncated normal)

  • Data

    • Fair_data (Fair's data on extramarital affairs, Excel)

    • adoption (Data for horse adoption, Excel2003)
      
       

• Efficiency gains through re-parameterization: The Ordered Probit Model

  • Lecture Notes

  • Matlab scripts and functions

    • OP_data (data generating script for the OP model)

    • OP (main script)

    • gs_OP (Gibbs Sampler)

    • OP_paddlers (script for Tahoe paddling example)

    • OP_predict (prediction script for Tahoe paddling example)
       

• Marginal Likelihood Estimation for Models with MH Components

  • Lecture Notes

  • Matlab scripts and functions

    • OP_chib (prep script for the C/J method)

    • gs_OP_chib (Gibbs Samplers for the C/J method)

    • OP_paddlers_constrained (constrained Tahoe paddlers model)

    • OP_chib_paddlers (prep script for unconstrained model)

    • OP_chib_paddlers_constrained (prep script for constrained model)
       

• Selection and Treatment Models

  • Lecture Notes

  • Matlab scripts and functions

    • select_data (data generation script)

    • select (main script for sample selection model)

    • select_door2door (sample selection script for door-to-door application)

    • select_door2door_predict (prediction script for the door-to-door application)

    • gs_select (Gibbs Sampler)

    • iw_sig11 (Nobile's (2000) algorithm to draw from a constrained IW)

    • treat_data (simulated data for the treatment effect model)

    • treat (main script for the treatment effect model)

    • gs_treat (Gibbs Sampler for the treatment effect model)

    • treat_regress (naive regression model using simulated treatment data)

    • switch_data (data generation for the switching regression model)

    • switchreg (main script for the SRM)

    • gs_switch (Gibbs Sampler for the SRM)

    • switch_dropout (main script for SRM, dropout application)

    • switch_dropout_predict (prediction script for dropout application)

    • dropout (data set, courtesy of Li et al and JoAE's data repository)
       

  • Paper for switching model & data:

    Mingliang Li, Dale J. Poirier and Justin L. Tobias , "Do Dropouts Suffer from Dropping Out? Estimation and Prediction of Outcome Gains in Generalized Selection Models", Journal of Applied Econometrics, Vol. 19, No. 2, 2004, pp. 203-225.

Problem Sets

• PS 1 (due Sept. 10, 2008)
  • Instructions
     
• PS 2 (due Sept. 29, 2008)
  • Instructions
  • ps2_data (data for q1 in Excel format)
  • Moeltner and Woodward (paper to illustrate application and refinement of priors for Q1b)
     
• PS3 (due Oct.20, 2008)
  • Instructions
  • invgampdf (utility function, evaluates the ordinate of the ig density; needed for Q4)
     
• PS4 (due Nov.12, 2008)
  • Instructions
  • pacific( data for Q1, Excel2003 format)
  • production (data for Q2, Excel2003 format)
     
• PS5 (due Dec.8,2008)
  • Instructions
  • door2door (data for fundraising application)
  • log file for "select_door2door"