Frank Schorfheide, "Loss Function Based Evaluation of DSGE Models",
Journal of Applied Econometrics, Vol. 15, No. 6, 2000, pp. 645-670.

Special Issue on ECONOMETRIC INTEGRATION OF INFERENCE AND DECISION-MAKING

Contact Information:

Frank Schorfheide
University of Pennsylvania
Department of Economics
3718 Locust Walk
Philadelphia, PA 19104-6297

E-mail: schorf@ssc.upenn.edu
URL   : www.econ.upenn.edu/~schorf

Dataset: 

   Source:           schorfheide-data.zip
   Target Directory: ../dsgesel/data
   Filename:         data.prn (ASCII, DOS format)
              
   Data format:
   COLUMN 1: GDPD
   COLUMN 2: GDPQ
   COLUMN 3: GPOP
   ROW   1: 1949:IV
   ROW 193: 1997:IV

   Time series, extracted from DRI database 05/04/00
   GDPD = GROSS DOMESTIC PRODUCT:IMPLICIT PRICE DEFLATOR (INDEX,92=100)(T7.1)
   GDPQ = GROSS DOMESTIC PRODUCT
   GPOP = POPULATION, NIPA basis (THOUS.,NSA)
   Quarterly Data from 1949:IV to 1997:IV

GAUSS Libraries:

   Source          : libzip.zip
   Target Directory: c:/gauss/usersrc/
   Filename:  ciamod.src  =   solution of cash-in-advance models
              cialh.src   =   evaluate likelihood for CIA models
              dsgesel.src =   various procedures for evaluation of DSGE models,
                              including a generic minimization routine, originally
                              written by Chris Sims in Matlab

   Instructions:

   1) copy files in c:\gauss\usersrc
   2) At the GAUSS prompt, type:
      lib user    c:\gauss\usersrc\dsgesel.src
      lib cialib  c:\gauss\usersrc\ciamod.src
      lib cialib  c:\gauss\usersrc\cialh.src
  
   NOTE: you have to make sure that the names of the library are consistent
         with the names that appear in the GAUSS programs!

Prior, Hessian, Mode in ASCII format:

   Source          : parazip.zip
   Target Directory: ../dsgesel/gauss/para
   Filename:  m11pri.out Prior for parameters of standard CIA Model
              m21pri.out Prior for parameters of PAC Model
              m11mod.out Posterior Mode, standard CIA model
              m21mod.out Posterior Mode, PAC model
              m11hes.out Hessian at Mode, standard CIA model
              m21hes.out Hessian at Mode, PAC model

GAUSS Procedures:

   To reproduce the DSGE model evaluation, you have to run
   the following GAUSS procedures.

   Source          : gausszip.zip
   Target Directory: ../dsgesel/gauss/
   Filenames:

  loaddata.g =  reads the ASCII data set into GAUSS

  ciatest.g  =  tests the ciamod procedure, computes steady states, IRFs
    
  ciapm.g    =  finds the posterior mode for the DSGE models,
                the prior is obtained from the ASCII file mXXpri.out

  ciahess.g  =  evaluates the Hessian at mode, computes penalty for 
                Laplace approximation
                this file reads the posterior mode from the file
                mXXmod.out and the prior from the file mXXpri.out

  ciamh.g    =  Metropolis-Hastings Algorithm for DSGE model
                This file read prior, posterior mode and hessian from the
                files mXXpri.out, mXXmod.out, mXXhes.out

  ciadd.g    =  Computes marginal data density via modified harmonic 
                mean estimator based on output from ciamh.g

  ciappara.g =  Computes posterior means and S.E. for DSGE model parameters
                based on output from ciamh.g

  ciapred.g  =  Convert DSGE parameters into IRFS, correlations and covariances
                based on output from ciamh.g

  ciapmom.g  =  Posterior mean and S.E. for correlations and covariances
  ciapirf.g  =  Posterior mean and S.E. for IRFs
  cialoss.g  =  Loss function estimation of DSGE model

  varpost1.g =  Marginal Data Density for VAR (various estimators)

  varps.g    =  Posterior simulator for VAR, output can be used to 
                compute marginal data density via modified harmonic mean
		estimator (ciadd.g)

  varbq.q    =  Draws from posterior of VAR parameters, conversion into
                IRFs, correlations and covariances
     
  irfconv.g  =  Normalizes Impulse response functions