Updated 4 May 2006 -- a work in progress!

                        The Linear Regressions Models Estimator
Abstract:

    REG is a "GRBL2" GAUSS program  that will estimate linear regressions.
    These include OLS, Panel Models, and TOBIT models.


I. Introduction

  REGT is a GRBL2 program -- please see GRBL2.TXT for a general description. In particular,
  REG uses "command file" to select models, variables, etc.

  REG currently supports the following estimators.
     OLS : Linear, semi-log, full-log
     TOBIT : Censored, Truncated, and Heckit
     PANEL : Fixed Effect, Random Effect, several variations of serial correlated FE and RE

     And more will be added.

   -----------------------------


II. Description of Commands


REG is built using the GRBL2 interface. GRBL2 supports a number of sophisticated
keywords for use in "commands files".

The following keywords are those that are either unique to REG, or are sufficiently
important that they bare repeating. Note that keywords described in GRBL2_BATCH.TXT,
but not mentioned here, WILL work in REG command files.

In other words, if you want to get fancy, please see GRBL2_BATCH.TXT

IIa. List of keywords (with examples):

  MODEL :  Specify model
  X     : specify independent variables
  Y     : specify dependent variable
  WEIGHT : specify weight (typically, a population weight)
  ID    : Specify Panel identifier

Details on REG keywords.

MODEL: Specify model.

   Two models are supported: POISSON and NEGBIN (Negative Binomial).
   Actually, two variants of the Negative Binomial are supported NEGBINI and NEGBINII.

   Syntax: MODEL model_name modifiers ;

    Model_name can be:
        OLS  :  Ordinary least squares
        TOBIT : Censored and truncated least squares
        PANEL : Fixed and Random effects models


   Several modifiers can be included, depending on the model chosen.

   Modifiers for OLS:

     LOG=xx  : Semi or full log model
         NO    : (the default) Do NOT log indepedent or dependent variable.
         SEMI  : Semi-log OLS (log of Y used as dependent variable)
         FULL  : Full-log OLS (log of Y and X used)

     LOG0=aval  : If LOG= is used, and a value (of Y or X) is 0, use aval

     RATIO=rvar : Compute a ratio -- the dependent variable is Y/rvar.
                  If any RVAR has a value of 0, an error occurs.
                  Note: WEIGHT causes all variables to be divided by the WEIGHT,
                         RATIO applies ONLY to the dependent (Y) variable.

     BOOT=nnn   : Compute bootstrap estimates. nnn is the number of bootstrap simulations.
             If nnn=0, bootstrapping  is NOT done.

             NOTE: bootstrapping is NOT supported for ratio or log models.

     CI=aci     : If BOOT=nnn specified, aci is the "confidence interval" to report estimates of.
                  aci should be between 0.50 and 1.0 (the default value is 0.95)

     Examples:
         MODEL OLS ;
         MODEL OLS LOG=SEMI ;
         MODEL OLS RATIO=POP05 ;
         MODEL OLS LOG=FULL RATIO=ACRES ;
         MODEL OLS BOOT=100 CI=0.80  ;

     Note: if RATIO and LOG are specified, the log is taken AFTER the ratio is computed.

  Modifiers for TOBIT:

      -TRUNC   : If specified, a truncated TOBIT (i.e.; all dependent variable values
                 strictly greater than 0)
                 Otherwise, a censored TOBIT (i.e.; lower bound of 0) TOBIT is estimated.

     -NORM   : Do estimation with normalized variables.
                Warning: normalization does not work well with truncated model.


     -UPPER  : Compute upper bounded model (otherwise, compute lower bounded model). Sorry, upper & lower bounded model
               is not currently supported.

      LIMIT=N or LIMIT=avar : Specify a lower bound. For standard TOBIT, use LIMIT=0 (the default).
                 N should be a value (such as 0); or avar should be a variable (in the dataset) that contains
                 a row specific lower limit

     -MIN    : Use the minimum value (of the dependent variable) as the lower bound
     -MAX    : Use the maximum value (of the dependent variable) as the lower bound (used if -UPPER is specified)

     Note: -MIN, -MAX, and LIMIT=n should NOT be used together

X  : Specify independent variables.

   Syntax:  X var_list;
   Example:
      X -CONST PRICE INCOME AGE COLLEGE ;



X  : Specify dependent variable.

   Syntax:  X varname;
   Example:
      Y TRIPS ;



WEIGHT : Specify weight variable

   Syntax: WEIGHT varname ;
   Example:
       WEIGHT POP2000 ;
       WEIGHT NO ;      (no weighting)

  In the OLS model, the WEIGHT is used as a GLS style correction -- all values divided by the observation's WEIGHT.

  In the TOBIT model, the WEIGHT modifies the variance: the variance of observation i = SIGMA*SIGMA/WEIGHT.

  Note that the weighted truncated model gives unusual results, and may not be trustworthy.



      -------------------------------------------

Technical details.

