Hi,

I am still new with Gauss and it came out with this error. Please help me with this. Thank you.

new; cls;

library lr; lrset;

format 8,8;

/*name of data file*/

dataset = "firmx1";

/*dependent variables*/

depvar = { Y1, Y2 };

pp = rows(depvar);

/*independent variables*/

veq1 = { const, X11 };

veq2 = { const, X21 };

invars = veq1|veq2;

/*number of parameters*/

np1 = rows(veq1);

np2 = rows(veq2);

numvar = np1|np2;

Q = LSUR( dataset, pp, invars, numvar );

end;

## 1 Answer

0

Hello,

I have looked over your code and have a few suggestions. I was unable to run your code because I do not have your data. However, I have noted a couple of things. First in defining your independent variables, *veq1* and *veq2*, you use the variable *const*. However, it appears that you have not defined *const* prior to using it to define your independent variables.

The second issue, and more likely reason for your error code, is that you have four inputs into the ** lsur** procedure. The

**procedure requires 5 inputs. The proper usage is:**

*lsur*Q =LSUR(dataset,LHS_vars,RHS_vars,NUM_var,Restrict);

You are currently missing the *Restrict* input. The source code (or user manual) provides the following description of *Restrict*:

** Restrict -- String, constrained information on parameters

** to perform restricted estimation. The syntax

** of Restrict is as follows:

**

** Restrict="rest1, rest2,...., restN";

**

** More than one restriction is allowed provided

** each is separated by commas. Each restriction

** must be written as a linear equation with all

** variables in the left-hand side and the constant

** in the right-hand side (i.e., x1:1+x1:2=1).

** Variables shown in each restriction must be

** variables in the regression model. Note that

** the numeric value following the (:) signifies

** which equation the variable comes from (i.e.,

** X4:10 indicates the X4 variable comes from the

** 10th equation). Restrictions in the RESTRICT

** argument must be consistent and not redundant

** otherwise error messages will be given. Users

** should note that only the parameters associated

** with the variables are restricted, and not the

** variables in the model.

**

** Examples of some restrict arguments:

**

** 1) Restrict="x1:1 + x1:2 + x1:3 = 1";

** 2) Restrict="const:1 + const:2 + const:3 = 1,

** trend:1 = 0,

** trend:2 = 0,

** trend:3 = 0";

Below you will find a a complete example using

*lsur***:**

/*---------------------------------------------------------------------------

Program file: lr15

Data file: t11_3

Last changed: May 13, 1992

Written by: Gene LeungNote........: This example is from Judge et. al. (1988, p.460). The use

of seemingly unrelated regression with restrictions imposed

is illustrated.

-----------------------------------------------------------------------------*/library lr;

lrset;

output file = lr15.out reset;dataset = "t11_3";

lhs = { q1,q2,q3 };

rhs = { const,p1,y,

const,p2,y,

const,p3,y };// NO. OF RHS VARIABLES IN EACH EQN.

novars = { 3,3,3 };//Set restrict

restrict = "p1:1-p2:2=0,

p1:1-p3:3=0";//USING NORMAL DIVISOR

_lrdv = 0;//LSUR ESTIMATION WITHOUT RESTRICTION IMPOSED

result1 = lsur(dataset,lhs,rhs,novars,0);//LSUR ESTIMATION WITH RESTRICTION IMPOSED

result2 = lsur(dataset,lhs,rhs,novars,restrict);///* RESULT1 AND RESULT2 ARE VECTORS THAT STORE

ALL STATISTICS AND ESTIMATED COEFFICIENTS. SEE VLIST AND VREAD TO LIST CONTENTS

OF THESE VECTORS AND EXTRACT INFORMATION FROM THEM. */print;

print "-------------------------------------------";

print "Variables stored in result2 are as follows: ";

print "-------------------------------------------";

call vlist(result2); /* VLIST IS A FUNCTION TO LIST ALL

VARIABLES INSIDE THE VECTOR */format /rd 8,4;

print;

print "-----------------------------------------------------";

print "The coeff. with restrictions imposed are as follows: ";

print "-----------------------------------------------------";

let mask[1,2] = 0 1;

let fmt[2,3] = "-*.*s " 10 8

"*.*lf " 14 8;

answer = rhs~vread(result2,"b"); /* VREAD GETS b FROM RESULT2 */

call printfm(answer,mask,fmt);output off;

## Your Answer

## 1 Answer

Hello,

I have looked over your code and have a few suggestions. I was unable to run your code because I do not have your data. However, I have noted a couple of things. First in defining your independent variables, *veq1* and *veq2*, you use the variable *const*. However, it appears that you have not defined *const* prior to using it to define your independent variables.

The second issue, and more likely reason for your error code, is that you have four inputs into the ** lsur** procedure. The

**procedure requires 5 inputs. The proper usage is:**

*lsur*Q =LSUR(dataset,LHS_vars,RHS_vars,NUM_var,Restrict);

You are currently missing the *Restrict* input. The source code (or user manual) provides the following description of *Restrict*:

** Restrict -- String, constrained information on parameters

** to perform restricted estimation. The syntax

** of Restrict is as follows:

**

** Restrict="rest1, rest2,...., restN";

**

** More than one restriction is allowed provided

** each is separated by commas. Each restriction

** must be written as a linear equation with all

** variables in the left-hand side and the constant

** in the right-hand side (i.e., x1:1+x1:2=1).

** Variables shown in each restriction must be

** variables in the regression model. Note that

** the numeric value following the (:) signifies

** which equation the variable comes from (i.e.,

** X4:10 indicates the X4 variable comes from the

** 10th equation). Restrictions in the RESTRICT

** argument must be consistent and not redundant

** otherwise error messages will be given. Users

** should note that only the parameters associated

** with the variables are restricted, and not the

** variables in the model.

**

** Examples of some restrict arguments:

**

** 1) Restrict="x1:1 + x1:2 + x1:3 = 1";

** 2) Restrict="const:1 + const:2 + const:3 = 1,

** trend:1 = 0,

** trend:2 = 0,

** trend:3 = 0";

Below you will find a a complete example using

*lsur***:**

/*---------------------------------------------------------------------------

Program file: lr15

Data file: t11_3

Last changed: May 13, 1992

Written by: Gene LeungNote........: This example is from Judge et. al. (1988, p.460). The use

of seemingly unrelated regression with restrictions imposed

is illustrated.

-----------------------------------------------------------------------------*/library lr;

lrset;

output file = lr15.out reset;dataset = "t11_3";

lhs = { q1,q2,q3 };

rhs = { const,p1,y,

const,p2,y,

const,p3,y };// NO. OF RHS VARIABLES IN EACH EQN.

novars = { 3,3,3 };//Set restrict

restrict = "p1:1-p2:2=0,

p1:1-p3:3=0";//USING NORMAL DIVISOR

_lrdv = 0;//LSUR ESTIMATION WITHOUT RESTRICTION IMPOSED

result1 = lsur(dataset,lhs,rhs,novars,0);//LSUR ESTIMATION WITH RESTRICTION IMPOSED

result2 = lsur(dataset,lhs,rhs,novars,restrict);///* RESULT1 AND RESULT2 ARE VECTORS THAT STORE

ALL STATISTICS AND ESTIMATED COEFFICIENTS. SEE VLIST AND VREAD TO LIST CONTENTS

OF THESE VECTORS AND EXTRACT INFORMATION FROM THEM. */print;

print "-------------------------------------------";

print "Variables stored in result2 are as follows: ";

print "-------------------------------------------";

call vlist(result2); /* VLIST IS A FUNCTION TO LIST ALL

VARIABLES INSIDE THE VECTOR */format /rd 8,4;

print;

print "-----------------------------------------------------";

print "The coeff. with restrictions imposed are as follows: ";

print "-----------------------------------------------------";

let mask[1,2] = 0 1;

let fmt[2,3] = "-*.*s " 10 8

"*.*lf " 14 8;

answer = rhs~vread(result2,"b"); /* VREAD GETS b FROM RESULT2 */

call printfm(answer,mask,fmt);output off;