
  This document was last modified: 27 April 2006. 


  GVAR.zip contains the following folders and files:


  readme.txt		    this file

  gvar.prg		    gauss program file

  gvarlb.lib		    gauss library file
 
  gvarsc.src		    gauss source code file
 
  ImportFiles.xls           folder that contains the input excel files that the gauss
                            program reads from

  TemplatesGraphs.xls       folder that contains excel templates for viewing girf and
                            gfevd graphs  
   
  CountryRegionalInfo.xls   folder that contains excel files with information on the
                            country and regional codes,# domestic variables etc 



  The program gvar.prg is set up to replicate the empirical results in:

  Stephane Dees, Filippo di Mauro, M. Hashem Pesaran and L. Vanessa Smith, (2006),   
  "Exploring the International Linkages of the Euro Area: A Global VAR Analysis", 


 ***********************************************************************************  
  Note: The gvar program was written using gauss6.0 for windows. Its use with later
        versions should be compatible, while its use with earlier gauss versions may
        be problematic.       
 ***********************************************************************************  



 
     Contents:

    I.   INSTALLATION OF THE GVAR PROGRAM
    II.  DESCRIPTION OF INPUT FILES
    III. DESCRIPTION OF OUTPUT FILES
    IV.  DESCRIPTION OF OTHER FOLDERS    
    V.   NOTATION & ADDITIONAL INFORMATION   



   I. INSTALLATION OF THE GVAR PROGRAM



  1. Create a directory and extract (or copy) all of the folders and files within
     GVAR.zip to that directory.


  2. Add gvarsc.src, the source code file to the src folder within the gauss6.0
     folder. Add gvarlb.lib, the library file to the lib folder within the gauss6.0
     folder.


  3. Create the path c:\\gauss6.0\\gvar

        i.  Place the folder ImportFiles in the gvar folder

        ii. Within the gvar folder create a folder named output and within the 
            output folder create another folder named data.

        The following paths should then exist:
        
         c:\\gauss6.0\\gvar\\ImportFiles
         c:\\gauss6.0\\gvar\\output
         c:\\gauss6.0\\gvar\\output\\data

   *****************************************************************************   
   NOTE: If the path that defines the location of the gauss program on your pc
         is different than c:\\gauss6.0, then this path needs to be replaced
         by the appropriate one, both in the gvar program and in the source
         code file gvarsc. Tip: Search for c:\ this should cover all locations
         where the path needs to be replaced.    
   *****************************************************************************  

       
    
    II. DESCRIPTION OF INPUT FILES
     

      All the input files that the gvar program reads from are located in the folder
      named ImportFiles. This folder consists of the following files:                       

       
      1.countrydata33.xls       : Excel folder that contains 33 worksheets, one for each country,
                                 containing quarterly data (1979Q1-2003Q4) for all variables
                                 except for real GDP. The real GDP variable for all countries
                                 is dealt with seperately in the gdp_interpl_normalz file. 

                                 The variables included in this workbook are the following:

                       
                                 CPI                   : Consumer Price Index

                                 Q                     : Real Equity Price Index
 
                                 E_{superscript_index} : Exchange Rate of country i at time t
                                                         expressed in US dollars, tranformed to
                                                         an index for the countries that belong
                                                         to the euroarea

                                 R_{superscript_S}     : Short rate of interest per annum, in per
                                                         cent (typically a three month rate)
 
                                 R_{superscript_L}     : Long rate of interest per annum, in per
                                                         cent (typically a ten year rate) 

                                 POIL                  : Nominal price of oil


                          Notes: i.  The exchange rate data for the countries that are included
                                     in the euroarea has been converted to an index using 2000
                                     as the base year.

                                 ii. Where data was not available the corresponding excel cells
                                     have been replaced by the value 8934567.


     2.countrycodes.xls        : Country names and country codes for the 33 countries. The
                                 ordering of the countries should follow that of the excel
                                 sheets as they appear in the countrydata33 file. 


     3.regcountry.xls          : Names and numbers of countries/regions as well as country
                                 codes. The country/regional ordering as appears in this
                                 file is that adopted throughout the program once any
                                 regions such as the euroarea have been created.


     4.tradematrix8003.xls     : Annual trade figures for the period 1980-2003.


     5.EuroDollarRate2000.xls  : Quartery Euro Dollar Rate for the year 2000. 


     6.pppgdp9901.xls          : PPP-GDP figures for all 33 countries for the period 1999
                                 -2001.
                               

     7.pppgdp9901_reg.xls      : PPP-GDP figures for all countries/regions for the period
                                 1999-2001 (25 countries plus the Euroarea).


     8.idpppgdp9901_reg.xls    : Country names and country codes for the 26 country/regions, 
                                (8 countries are grouped into the Euro Area). The ordering
                                 should follow that in the pppgdp9901_reg file.


     9.indxregcir_reg.xls      : Region Number and Country/Regional Code for the regions as
                                 defined for aggregation of the impulse responses. The
                                 ordering of the  countries/regions should follow that in
                                 the regcountry file. 


     10.gdp_interpl_normalz.xls: This excel workbook consists of three worksheets. The first
                                 worksheet contains the seasonally adjusted data (seasonal
                                 adjustment was performed in Eviews). The second worksheet
                                 deals with interpolation of the real gdp data for countries
                                 that only have annual data for this variable, over any given
                                 period or throughout the whole period. The third worksheet 
                                 provides information on the relevent base years required
                                 to transform any GDP figures that are not an index into one.
                                       
                           Note: Some countries like Australia have the base year extending
                                 over two years. These two years are reported for BY1 and BY2.
                                 For countries that have only one base year, the same year is
                                 used for BY1 and BY2.


     11.OveridRestrictions.xls: Overidentifying-restrictions imposed on the cointegrating
                                vectors of the GVAR (Not relevant for DdPS(2006)). 

    



     III. DESCRIPTION OF OUTPUT FILES 
   

     *************************************************************************************      

      Notes:

       a. To control some of the output that is exported and printed to screen, and thus to
          control the speed of the program, please read carefully the notes at the beginning
          of the gvar program.


       b. Upon every run of the gvar program, all output files in the output folder need to
          be deleted or moved to another location.
 


       c. Templates are provided to view the graphs for the GIRFs, GFEVDs and bootstrap
          persistence profiles. A description of these can be found below. 

       d. The existing templates are not appropriate in the case where countries and/or
          variables are shuffled around, such as for example in the case of structural 
          impulse response analysis. To use the existing templates in this case, the
          ordering of the countries and/or variables in the underlying files that the
          templates read from need to be re-ordered to match the existing ordering.
          Alternatively, new templates will need to be made. 
   
       e. To change the titles of the graphs when using the templates go the worksheet
          named titles and change the title to that required. This will update all graph
          titles automatically.
      
        
      ************************************************************************************* 

        
        All the files described below are extracted to the output folder:

           
      1. out_DataFiles.xls: When this is variable is set to one, which is controlled at
                            the beginning of the gvar program, data files for all countries
                            with the corresponding
 country name are exported to the folder
                            named data created under the output folder.


      2. Info.out         : Gives summary information corresponding to the most recent run
                            of the gvar program, such as the type of weights used, the type
                            of identification of the long run parameters for cointegration,
                            the type of shock selected for girf etc.    


      3. rgshmtx.xls      : Regional Share Matrix exported in such a way that the columns
                            not the rows sum to one.    



      4. Unit Root Tests         
 
        If out_UnitRoot at the beginning of the gvar program is set to 1 (default is zero)
        the following files are produced:

           unitroot.out: Detailed results for unit root tests ADF, GLS_ADF, MAX_ADF,
                         WS_ADF using AIC.
        
           urendvar.xls: Summary unit root test results for the endogenous variables
                         based on AIC.

          urforvar.xls: Summary unit root test results for the foreign (star)
                         variables based on AIC.



     5. Selection of the VAR order for the individual country models
 

          VARXord.out  : Lag order of the endogenous and star variables in the country-
                         specific VARX* models chosen using AIC or SBC. 

                   Note: For the country-specific VAR models (i.e in the case of no
                         foreign variables) the lag order for the endogenous variables
                         is that used for the VARX* models.


           VARXSC.out  : Detailed results of a 4th order LM test for residual serial
                         correlation of the individual equations in thecountry-
                         specific VARX* models.
 



      6. Exogeneity Tests

           exogen.out         : Lag order of endogenous and foreign (star) variables used in
                                the exogeneity tests. See relevant section in DdPS(2006) for
                                a description of this test.


          exogeneity_test.xls : Exogeneity test results
 for the 26 country/regions.  



      7. Cointegration

           ncntVARX.out            : Number of cointegrating relations for the country-specific
                                     VARX* models
     


           coint_max&traceVARX.xls : Detailed cointegration results for the maximum eigenvalue
                                     and trace statistics for the country-specific VARX* models 


           max&tr_cv90_VARX.xls    : Upper 10% small sample critical values of the maximum
                                     eigenvalue and trace statistics for the country-specific
                                     VARX* models. This file is exported to the output folder
                                     only if the option small sample critical values is selected. 


           max&tr_cv95_VARX.xls    : Upper 5% small sample critical values of the maximum
                                     eigenvalue and trace statistics for the country-specific
                                    VARX* models. This file is exported to the output folder
                                     only if the option small sample critical values is selected. 


           print_cointVARX         : Prints to screen the detailed cointegration results for the
                                     country- specific VARX* models with the corresponding
                                     asymptotic
 critical values.  


           nbcntVAR.out            : The same as above for the country-specific VAR models 


           coint_max&traceVAR.xls  : The same as above for the country-specific VAR models


           max&tr_cv90_VAR.xls     : The same as above for the country-specific VAR models


           max&tr_cv95_VAR.xls     : The same as above for the country-specific VAR models


           print_cointVAR          : The same as above for the country-specific VAR models  




     8. Results for the Individual Country Models

           
           ECMs_RsqAIC&SBC_VARX.xls: When out_ECMVARX is set to 1 at the beginning of the gvar
                                     program, individual country VARX* error-correction estimates
                                     are exported to the output folder along with values for the
                                     R_bar square, AICand SBC of the individual equations. 


           ECMs_RsqAIC&SBC_VAR.xls : When out_ECMVAR is set to 1 at the beginning of the gvar
                                     program, individual country VAR error- correction
 estimates
                                     are exported to the output folder along with values for the
                                     R_bar square, AIC and SBC of the individual equations.  


           ContmpCorr.xls          : Contemporaneous effects of the foreign variables on their
                                     domestic counterparts for the country-specific VARX* models
                                     along with White's heteroskedastic robust standard errors
                                     and t-ratios.  

 
           avgcorr_vall_VARX.xls   : Average pair-wise cross-section correlations of the level
                                     and first difference of all variables and associated individual
                                     country VARX* residuals. In addition to the average pair-wise
                                     correlations, when the variable out_corrVARX is set to 1 at the
                                     beginning of the gvar program,pair-wise cross-section correlations
                                     (not averaged) of the individual country VARX* residuals for all
                                     variables are exported to the output folder under the names
                                     corr_v1_VARX, corr_v2_VARX etc.  
                                   


           avgcorr_vall_VAR.xls    : Average pair-wise cross-section correlations of the level and
                                     first difference of all variables and associated individual
                                     country VAR residuals. In addition to the average pair-wise
                                     correlations, when the variable out_corrVARX is set to 1 at the
                                     beginning of the gvar program,pair-wise cross-section correlations
                                     (not averaged) of the individual country VAR residuals for all
                                     variables are exported to the output folder under the names
                                     corr_v1_VAR, corr_v2_VAR etc.



      9. Standard Errors, Eigenvalues and Chi-Square Results for Overidentifying Restrictions
          on the Cointegrated Vectors


            Standard_Errors.xls: Standard Errors for the first-difference of the data, the
                                 residuals of the country-specific VAR models, the residuals
                                 of the country-specific VARX* models and the residuals of
                                 the GVAR model.


            Eigenval.csv       : Eigenvalues and moduli of eigevalues of the GVAR.          


            ChiSqForOR.xls     : Chi-square tests and bootstrap critical values for the over-
                                 identifying restrictions on the cointegrating vectors (not
                                 valid for DdPS(2006)).     



      10. Weights used for Aggregation of Impulse Responses, Error Variance Decomposition etc  


            cwgts.xls :  Country weights calculated using PPP-GDP figures over the period
                         1999-2001 for all variables     


            rgwgts.xls:  Regional weights calculated using PPP-GDP figures over the period
                         1999-2001 for all variables     



      11. Point Estimates for Persistence Profiles, Impulse Responses and Variance Decompositions   


         i. Effect of Shocks on Cointegrating Relations   

          effect_cr.xls :  Effect of system wide shocks (persistence profiles) or variable-
                            specific shocks (girfs) on the cointegrating relations. The effect
                            of shocks on cointegrating relations is expected to die out.

                     Note: The use of an excel template to view the graph is not required in
                           this case. Simply select the relevant cells and construct the graph.      


        ii. Generalised Impulse Response Functions (GIRF) 


         a. Country Results
  
     

           cvnamcallvindgirf.xls   :  girfs arranged in the order all countries, individual
                                      variables        

           cvnamcindvallgirf.xls   :  girfs arranged in the order individual countries, all
                                      variables 


         b. Regional Results 
     

           cvnamcallvindgirfrg.xls : girfs arranged in the order all regions, individual
                                     variables        

           cvnamcindvallgirfrg.xls : girfs arranged in the order individual regions, all
                                     variables    



      iii. Generalised Forecast Error Variance Decomposition (GFEVD)  
 

         a. Country Results
 
    

           cvnamcallvindgevd.xls   :
 gfevds arranged in the order all countries, individual
                                       variables
  

           cvnamcindvallgevd.xls   :
 gfevds arranged in the order individual countries, all
                                       variables
 

          
         b. Regional Results  
     

           cvnamcallvindgevdrg.xls :
 gfevds arranged in the order all regions, individual
                                     variables
            

           cvnamcindvallgevdrg.xls :
 gfevds arranged in the order individual regions, all
                                     variables
  
 



     12. Bootstrap Mean and Median Estimates and Confidence Intervals for Persistence Profiles,
         Impulse Responses and Variance Decompositions  


        #Bootstraps.xls   : An excel file that allows one to check the progress of the bootstrap.
                            Exports the number of bootstrap replications using a 20 step increment.
         

  
      i. Effect of shocks on cointegrating relations (CR) 
     

          Mean_CR.xls  : Bootstrap mean estimates of the effect of a system-wide shock
                         (persistence profiles) or a variable-specific shock (girfs) on
                         the cointegrating relations  
  
          Median_CR.xls: Bootstrap median estimates of the effect of a system-wide shock
                         (persistence profiles) or a variable-specific shock (girfs) on
                         the cointegrating relations  


          Lband_CR.xls : Lower 5% confidence bound for Mean_CR
 and Median_CR     


          Uband_CR.xls : Upper 95% confidence bound for Mean_CR
 and Median_CR    
   
 

     ii. Generalised Impulse Response Functions (GIRF)   
  

         a. Country Results
      


             Mean_girf.xls  : Bootstrap mean estimates of the effect of a country or
                              global shock to the error in the country-specific models
                              or in the global model
  


             Median_girf.xls: Bootstrap median estimates of the effect of a country or
                              global shock to the error in the country-specific models
                              or in the global model
  
   

             LBand_girf.xls : Lower 5% confidence bound for Mean_girf
 and Median_girf   


             UBand_girf.xls : Upper 95% confidence bound for Mean_girf
 and Median_girf  

       

         b. Regional Results
     


             Mean_girfrg.xls  : Bootstrap mean estimates of the effect of a country or
                                global shock to the error in the country-specific models
                                or in the global model having aggregated the countries
                                into regions
  

             Median_girfrg.xls: Bootstrap median estimates of the effect of a country or
                                global shock to the error in the country-specific models
                                or in the global model having aggregated the countries
                                into regions
  

             Lband_girfrg.xls : Lower 5% confidence bound for Mean_girfrg
 and Median_girfrg  

 
             Uband_girfrg.xls : Upper 95% confidence bound for Mean_girfrg
 and Median_girfrg
  
  

     iii. Generalised Forecast Error Variance Decomposition (GFEVD) 
      

        a. Country Results
  


            Mean_gevd.xls  : Bootstrap mean estimates of the proportion of the N-step
                             ahead forecast error variance of the variable of interest
                             explained by conditioning on contemporaneous and future
                             innovations of the individual country equations

            Median_gevd.xls: Bootstrap median estimates of the proportion of the N-step
                             ahead forecast error variance of the variable of interest
                             explained by conditioning on contemporaneous and future
                             innovations of the individual country equations
                                         

            LBand_gevd.xls : Lower 5% Confidence bound for Mean_gevd
 and Median_gevd
   

            UBand_gevd.xls : Upper 95% Confidence bound for Mean_gevd
 and Median_gevd 
    


        b. Regional Results
 
      

            Mean_gevdrg.xls  : Bootstrap mean estimates of the proportion of the N-step
                               ahead forecast error variance of the variable of interest
                               explained by conditioning on contemporaneous and future
                               innovations of the individual country equations having
                               aggregated the countries into regions
   

            Median_gevdrg.xls: Bootstrap median estimates of the proportion of the N-step
                               ahead forecast error variance of the variable of interest
                               explained by conditioning on contemporaneous and future
                               innovations of the individual country equations having
                               aggregated the countries into regions


            Lband_gevdrg.xls : Lower 5% Confidence bound for Mean_gevdrg and Median_gevdrg
    


            Uband_gevdrg.xls : Upper 95% Confidence bound for Mean_gevdrg and Median_gevdrg    
 




      13. Structural Break Tests   


          The following files are produced when conducting structural break tests with the
          corresponding critical values calculated via the bootstrap:
   



           #Bootstraps_sb.xls   : An excel file that allows one to check the progress of the
                                  bootstrap of the structural break tests. Exports the number
                                  of bootstrap replications using a 20 step increment.


           SBtests_detailed.xls : Detailed results for the individual structural break tests.
                                  The value 1 indicates that the tests reject the null of 
                                  structural stability and 0 otherwise, using the bootstrapped
                                  critical values. 
    

           SBtests_summary.xls  : Summary results for structural break tests. The number of
                                  rejections is reported per variable and in total together
                                  with the corresponding percentage in parenthesis.    


           BreakDate_supchow.xls: Break dates are reported only for the cases where the null
                                  hypothesis is rejected and are applicable for the sup Chow
                                  test (non-robust version) only.
   


              Note: _01,_05,_10 refer to the 1%,5% and 10% significance levels.
      





        IV. DESCRIPTION OF OTHER FOLDERS 

   
           i. CountryRegionalInfo.xls


          The above folder located in GVAR.zip contains the following files:


          Codes for Countries and Regions.xls:  Lists all country and regional codes used in
                                                the gvar program


          Non-existing Variables.xls         :  Lists the non-existing domestic variables 


          # Domestic Variables.xls           :  Number of domestic variables for each country




          ii. TemplatesGraphs.xls

          *************************************************************************************  
           Notes: 
           
              1. All excel template files need be placed in the same folder as the underlying 
                 files they read from. If asked what sheet to read from select sheet 1.

              2. To adjust the title of the graphs 
          *************************************************************************************  



          The TemplatesGraphs.xls folder located in GVAR.zip contains the following subfolders:


           a. PointEstimates.xls:

           
         Template_callvindgevd_11c.xls: Reads from the underlying output file cvnamcallvindgevd.
                                        Shows graphs of the gfevds for the individual variables
                                        for 11 countries: US, EA, China, Japan, UK, Australia,
                                        Canada, NZ, Norway, Sweden, Switzerland.


         Template_callvindgevd.xls    : Reads from the underlying output file cvnamcallvindgevd.
                                        Shows graphs of the gfevds for the individual variables
                                        for all countries.                              
     

         Template_callvindgevdrg.xls  : Reads from the underlying output file cvnamcallvindgevdrg.
                                        Shows graphs of the gfevds for the individual variables
                                        for the 10 country/regions: US, EA, China, Japan, UK, LA,
                                        ODC, Rest Asia, Rest World, Rest W. Europe. 
                                         
   
         Template_callvindgirf_11c.xls: Reads from the underlying output file cvnamcallvindgirf.
                                        Shows graphs of the girfs for the individual variables
                                        for 11 countries: US, EA, China, Japan, UK, Australia,
                                        Canada, NZ, Norway, Sweden, Switzerland.


         Template_callvindgirf.xls    : Reads from the underlying output file cvnamcallvindgirf.
                                        Shows graphs of the girfs for the individual variables
                                        for all countries.                                

         Template_callvindgirfrg.xls  : Reads from the underlying output file cvnamcallvindgirfrg.
                                        Shows graphs of the girfs for the individual variables for
                                        the 10 country/regions: US, EA, China, Japan, UK, LA, ODC,
                                        Rest Asia, Rest World, Rest W. Europe. 


         Template_cindvallgirf.xls    : Reads from the underlying output file cvnamcindvallgirf.
                                        Shows graphs of the girfs for the individual countries
                                        for all variables.       

 
         Template_cindvallgirfrg.xls  : Reads from the underlying output file cvnamcindvallgirfrg.
                                        Shows graphs of the girfs for the 10 country/regions as
                                        mentioned above for all variables.



             b. Bootstrap.xls:

           *************************************************************************************  
            Note: To view the Bootstrap Median Estimates the following templates need to be 
                  adjusted accordingly.
           *************************************************************************************  

          Template_CR_EXACT_B.xls   :  Reads from the underlying output files Mean_CR, Lband_CR 
                                       and Uband_CR. Shows graphs of the effect of system-wide 
                                       shocks (persistence profiles) or variable specific shocks 
                                       (girfs) on the cointegrating relations together with lower 
                                       5% and upper 95% bootstrap confidence bounds.
 
                                       This template should be used together with the option exact-
                                       identifying restrictions on the long run parameters when
                                       performing cointegration analysis of the VARX* models. This
                                       template is applicable in the case of DdPS(2006).
        
            
          Template_OVERepepsus_B.xls : Reads from the underlying output files Mean_CR,Lband_CR and
                                       Uband_CR. Shows graphs of the effect of individual variable
                                       shocks or system-wide shocks (persistence profiles) on the
                                       cointegrating relations together with lower 5% and upper 95%
                                       bootstrap confidence bounds.

                                       This template should be used together with the option over-
                                       identifying restrictions using the exchange rate expressed
                                       viz a viz the US dollar. This template is not applicable in
                                       the case of DdPS(2006).
         

          Template_OVERepeps_B.xls   : Reads from the underlying output files Mean_CR,Lband_CR and
                                       Uband_CR. Shows graphs of the effect of individual variable
                                       shocks or system-wide shocks (persistence profiles) on the
                                       cointegrating relations together with lower 5% and upper 95%
                                       bootstrap confidence bounds.

                                       This template should be used together with the option over-
                                       identifying restrictions using the effective exchange rate.
                                       This template is not applicable in the case of DdPS(2006).         


          Template_callvindgirf_B.xls: Reads from the underlying output files Mean_girf,Lband_girf
                                       and Uband_girf. Shows graphs of the girfs together with lower
                                       5% and upper 95% bootstrap confidence bounds for the individual
                                       variables for all countries.          


          Template_callvindgevd_B.xls: Reads from the underlying output files Mean_gevd,Lband_gevd
                                       and Uband_gevd. Shows graphs of the gfevds together with
                                       lower 5% and upper 95% bootstrap confidence bounds for the
                                       individual variables for all countries.    




        V. NOTATION & ADDITIONAL INFORMATION
   


       a. NOTATION 

         ep      = e-p   
         eps     = es-ps   
         epeps   = ep-eps  
         epepsus = ep+pus

      
         where e is the log of the nominal exchange rate in terms of US dollars, p is the
         log of the price level, pus is the log of the US price level, and es (e*), ps (p*) 
         is the corresponding foreign exchange rate and foreign price level, respectively.  
         See DdPS(2006) for the defnition of e* and p*.

         
         AIC                  : Akaike Information Criterion         


         SBC                  : Schwarz Bayesian Criterion
       

         NA                   : Not Applicable



         NC                   : Not computable. This abbreviation could be found in the
                                output for the structural break tests, indicating that
                                a particular structural break test was not calculated
                                due to a variance covariance matrix not being invertible.
    



       b. ADDITIONAL INFORMATION        

  
         i. Construction of the Euroarea


         The time series data for the euro area was constructed by cross section
         weighted averages of the variables, over Germany, France, Italy, Spain,
         Netherlands, Belgium, Austria and Finland, using the average Purchasing
         Power Parity GDP weights over the 1999-2001 period.
  

         The real equity price index for Germany was created using the price index 
         with base year (1995). 

         The price index for Germany was subsequently rebased to match the base year
         (2000) of the rest of the countries.    

         For the construction of the euro area exchange rate, each of the country
         members' exchange rate was converted to an index using
 2000 as the base
         year and premultiplied by the euro/dollar rate of that year.
 
   


         ii. Cointegration
         
         The following options are available:

        1. no intercepts and no trends    
        2. restricted intercepts and no trends  
        3. unrestricted intercepts and no trends    
        4. unrestricted intercepts and restricted trends   
        5. unrestricted intercepts and trends       
    

         All cointegration results reported in DdPS(2006) are for case 4 (cs=4, 
         as set in the gvar program), which applies to all countries and variables. 
         Asymptotic 5% critical values are used taken from MacKinnon, Haug and
         Michelis (1999), "Numerical distribution functions of likelihood ratio
         tests
 for cointegration", Journal of Applied Econometrics, 14,563-577.
               

         
At present, asymptotic critical values are provided for case 4 only.
  
         If one would like to use any of the other cases, the variable cs in  
         the gvar  program needs to be set to the appropriate value and the  
         option small sample critical values given during the execution of the 
         program needs to be selected. Small sample critical values are obtained 
         via
 simulation and will take some time to run. The number of replications
         is set to 10000. Again this value can be adjusted within the gvar.prg. 


        iii. Structural identification of shocks

          DdPS(2006) focus on identification of shocks to the US economy as described
          in the relevant section, which requires including the US model as the first
          country model and selecting the desired variable ordering. The shocks to
          the US are structurally identified and the usual GIRFs are then computed.
          Note that once the option of structural indentification is selected within
          the program, all results following the solution of the GVAR such as the
          effect of shocks on the cointegrating relations as well as the variance
          decompositions are based on structural identification of the US. 


        iv. Bootstrapping

          The number of bootstraps can be controlled in the main body of the gvar program
          by adjusting the values for the following variables:

           maxnb    : Maximum number of bootstraps permitted for GIRFs, GFEVDs and PPs 
           nb       : Number of Bootstraps for GIRFs, GFEVDs and PPs

           maxnb_sb : Maximum number of bootstraps permitted for Structural Break Tests 
           nb_sb    : Number of Bootstraps for Structural Break Tests                  

          The values for (maxnb, maxnb_sb) and (nb,nb_sb) are currently set to 4000 and
          2000 respectively.
            

         A number of conditions need to be satisfied before a replication can be included 
         in the counter for nb and nb_sb. If these are not satisfied, a new bootstrap sample
         is generated. The higher the value set for maxnb and  maxnb_sb, the greater the
         probability that 2000 replications will have been made by the time the maximum
         bootstrap values (given by maxnb and maxnb_sb) are attained. If this is not the
         case, which typically happens with some of the structural break tests, then the
         reported statistics will be based on a value lower than that specified by nb and
         nb_sb. These final values are reported in the info file. When the counter for 
         nb and nb_sb reaches 2000, the bootstrapping will stop,regardless of the value
         specified for maxnb and maxnb_sb.  
  
          

         Note:
         
         1. The minimum value that the nb variable can be set to is 20. 

         2. maxnb>=nb. In some cases the equality might not hold. You will
            be informed by the program in this case .
   
         3. maxnb_sb>=nb_sb. In some cases the equality might not hold. You
            will be informed by the program in this case.

            

         v. Miscellaneous

         The solution of the GVAR model using the code in its present form can only
         deal with individual country models that have a VAR order no greater than
         two for both the domestic and foreign variables.    

    

     L.Vanessa Smith
     lvs21@cam.ac.uk
    