Re: st: Separability Restrictions Nlsur Quaids

[Jorge Eduardo Pérez Pérez <jorge_perez@brown.edu>]

You are defining g14, g16, g24, g26, g34, g44, g46 and g56 twice in
your code. First you define them along all the other g terms

tempname g21 g22 g23 g24 g25 g26
...

and then you define them again

tempname g14 g16
....

Try to define these only once. To do this, supress the initial
definitions and leave only the definitions due to the restricted
model.

scalar `g11' = `at'[1, 11]
* Do not define g12
scalar `g13' = `at'[1, 13]
scalar `g14' = `at'[1, 14]
scalar `g15' = `at'[1, 15]
* Do not define g16

Do the same for the remaining terms.

Also, you need to make tempname calls for the "fe" and "FE" terms in
your model and add them to your "at" vector

tempname fe1 fe2 ...
scalar `fe1' = `at'[1, ...

tempname FE1 FE2 ...
scalar `FE1' = `at'[1,...

If this doesn't work, send me a piece of your dataset privately
(Attachments are forbidden in Statalist) and I will look again.

Jorge Perez.






--------------------------------------------
Jorge Eduardo Pérez Pérez
Graduate Student
Department of Economics
Brown University


On Mon, Nov 11, 2013 at 11:20 AM, Guilherme Travassos
<travassosgf@hotmail.com> wrote:
> Thank you again for your help, Jorge Perez.
>
> I try to do what you sad, but nothing changes.
>
> If you had some other ideia that could help me, I would be grateful. I don't know what to do anymore.
>
>
>> From: jorge_perez@brown.edu
>> Date: Mon, 11 Nov 2013 09:30:51 -0500
>> Subject: Re: st: Separability Restrictions Nlsur Quaids
>> To: statalist@hsphsun2.harvard.edu
>>
>> - To specify initial values to your parameters, add the following
>> option to your -nlsur- call:
>> initial(a3 1 a5 1 b3 1 b5 1)
>> where you can change the 1's to the values of your choice. Just make
>> sure the initial a3+b3 and a5+b5 are not 0.
>>
>> - Brian Poi's -quaids- command can be installed by writing:
>>
>> net install st0268_1.pkg
>>
>> Then write:
>>
>> help quaids to see how it works.
>>
>> Let me know if changing the initial values works. If not, I will take
>> a closer look at your code.
>>
>>
>> --------------------------------------------
>> Jorge Eduardo Pérez Pérez
>> Graduate Student
>> Department of Economics
>> Brown University
>>
>>
>> On Fri, Nov 8, 2013 at 10:59 AM, Guilherme Travassos
>> <travassosgf@hotmail.com> wrote:
>>> Jorge Perez, thank you for your reply.
>>>
>>> - I checked my independent variables and I don't have missing values on that. Using Brian Poi code, I already have the irrestrict model, so I think that teh problem isn't on the variables but on the restrict model code.
>>>
>>> - I understand when you sad about change the initial values such that they imply that the initial `a3' + `b3' and `a5' + `b5' are not zero, but I don't know how to do this on Brian's code. Can you help me?
>>>
>>> - I need to estimating the restrict model. I intend to use the Likelihood Ratio Test to test the weak separability restrictions on the restrict model against de irrestric model. The literature recommend it. Like I sad, I already have the irrestrict model.
>>>
>>> - I didn't know the Brian Poi's recent quaids - command. Could you show me?
>>>
>>> Anyway, I am sending my code with the specific restrictions. Maybe it could help.
>>>
>>>
>>> program nlsurquaids
>>>
>>> version 10
>>>
>>> syntax varlist(min=33 max=33) if, at(name)
>>> tokenize `varlist'
>>>
>>> args w1 w2 w3 w4 w5 lnp1 lnp2 lnp3 lnp4 lnp5 lnp6 lnexp lnpindex urban norte nordeste sudeste centro_oeste anosestudo mulherchefe crian_adolesc idoso lnMr_est fe1 fe2 fe3 fe4 fe5 F1 F2 F3 F4 F5
>>>
>>>
>>> tempname a1 a2 a3 a4 a5 a6
>>>
>>> scalar `a1' = `at'[1, 1]
>>> scalar `a2' = `at'[1, 2]
>>> scalar `a3' = `at'[1, 3]
>>> scalar `a4' = `at'[1, 4]
>>> scalar `a5' = `at'[1, 5]
>>> scalar `a6' = 1 - `a1' - `a2' - `a3' - `a4' - `a5'
>>>
>>>
>>> tempname b1 b2 b3 b4 b5 b6
>>>
>>> scalar `b1' = `at'[1, 6]
>>> scalar `b2' = `at'[1, 7]
>>> scalar `b3' = `at'[1, 8]
>>> scalar `b4' = `at'[1, 9]
>>> scalar `b5' = `at'[1, 10]
>>> scalar `b6' = - `b1' - `b2' - `b3' - `b4' - `b5'
>>>
>>>
>>> tempname g11 g12 g13 g14 g15 g16
>>>
>>> tempname g21 g22 g23 g24 g25 g26
>>>
>>> tempname g31 g32 g33 g34 g35 g36
>>>
>>> tempname g41 g42 g43 g44 g45 g46
>>>
>>> tempname g51 g52 g53 g54 g55 g56
>>>
>>> tempname g61 g62 g63 g64 g65 g66
>>>
>>>
>>> scalar `g11' = `at'[1, 11]
>>> scalar `g12' = `at'[1, 12]
>>> scalar `g13' = `at'[1, 13]
>>> scalar `g14' = `at'[1, 14]
>>> scalar `g15' = `at'[1, 15]
>>> scalar `g16' = - `g11' - `g12' - `g13' - `g14' - `g15'
>>>
>>>
>>> scalar `g21' = `g12'
>>> scalar `g22' = `at'[1, 16]
>>> scalar `g23' = `at'[1, 17]
>>> scalar `g24' = `at'[1, 18]
>>> scalar `g25' = `at'[1, 19]
>>> scalar `g26' = - `g21' - `g22' - `g23' - `g24' - `g25'
>>>
>>>
>>> scalar `g31' = `g13'
>>> scalar `g32' = `g23'
>>> scalar `g33' = `at'[1, 20]
>>> scalar `g34' = `at'[1, 21]
>>> scalar `g35' = `at'[1, 22]
>>> scalar `g36' = - `g31' - `g32' - `g33' - `g34' - `g35'
>>>
>>>
>>> scalar `g41' = `g14'
>>> scalar `g42' = `g24'
>>> scalar `g43' = `g34'
>>> scalar `g44' = `at'[1, 23]
>>> scalar `g45' = `at'[1, 24]
>>> scalar `g46' = - `g41' - `g42' - `g43' - `g44' - `g45'
>>>
>>>
>>> scalar `g51' = `g15'
>>> scalar `g52' = `g25'
>>> scalar `g53' = `g35'
>>> scalar `g54' = `g45'
>>> scalar `g55' = `at'[1, 25]
>>> scalar `g56' = - `g51' - `g52' - `g53' - `g54' - `g55'
>>>
>>>
>>> scalar `g61' = `g16'
>>> scalar `g62' = `g26'
>>> scalar `g63' = `g36'
>>> scalar `g64' = `g46'
>>> scalar `g65' = `g56'
>>> scalar `g66' = - `g61' - `g62' - `g63' - `g64' - `g65'
>>>
>>>
>>> tempname g14 g16
>>>
>>> scalar `g14' = ((`a4' + `b4')/(`a3' + `b3'))*(`g13' + `a1'*`a3') - `a1'*`a4'
>>> scalar `g16' = ((`a6' + `b6')/(`a5' + `b5'))*(`g15' + `a1'*`a5') - `a1'*`a6'
>>>
>>> tempname g24 g26
>>>
>>> scalar `g24' = ((`a4' + `b4')/(`a3' + `b3'))*(`g23' + `a2'*`a3') - `a2'*`a4'
>>> scalar `g26' = ((`a6' + `b6')/(`a5' + `b5'))*(`g25' + `a2'*`a5') - `a2'*`a6'
>>>
>>> tempname g34 g36
>>>
>>> scalar `g34' = ((`a4' + `b4')/(`a3' + `b3'))*(`g33' + `a3'*`a3') - `a3'*`a4'
>>> scalar `g36' = ((`a6' + `b6')/(`a5' + `b5'))*(`g35' + `a3'*`a5') - `a3'*`a6'
>>>
>>> tempname g44 g46
>>>
>>> scalar `g44' = ((`a4' + `b4')/(`a3' + `b3'))*(`g43' + `a4'*`a3') - `a4'*`a4'
>>> scalar `g46' = ((`a6' + `b6')/(`a5' + `b5'))*(`g45' + `a4'*`a5') - `a4'*`a6'
>>>
>>> tempname g56
>>>
>>> scalar `g56' = ((`a6' + `b6')/(`a5' + `b5'))*(`g55' + `a5'*`a5') - `a5'*`a6'
>>>
>>>
>>> tempname l1 l2 l3 l4 l5 l6
>>>
>>> scalar `l1' = `at'[1, 26]
>>> scalar `l2' = `at'[1, 27]
>>> scalar `l3' = `at'[1, 28]
>>> scalar `l4' = `at'[1, 29]
>>> scalar `l5' = `at'[1, 30]
>>> scalar `l6' = - `l1' - `l2' - `l3' - `l4' - `l5'
>>>
>>>
>>> tempname r11 r12 r13 r14 r15 r16 r17 r18 r19
>>>
>>> tempname r21 r22 r23 r24 r25 r26 r27 r28 r29
>>>
>>> tempname r31 r32 r33 r34 r35 r36 r37 r38 r39
>>>
>>> tempname r41 r42 r43 r44 r45 r46 r47 r48 r49
>>>
>>> tempname r51 r52 r53 r54 r55 r56 r57 r58 r59
>>>
>>> scalar `r11' = `at'[1, 31]
>>> scalar `r12' = `at'[1, 32]
>>> scalar `r13' = `at'[1, 33]
>>> scalar `r14' = `at'[1, 34]
>>> scalar `r15' = `at'[1, 35]
>>> scalar `r16' = `at'[1, 36]
>>> scalar `r17' = `at'[1, 37]
>>> scalar `r18' = `at'[1, 38]
>>> scalar `r19' = `at'[1, 39]
>>>
>>> scalar `r21' = `at'[1, 40]
>>> scalar `r22' = `at'[1, 41]
>>> scalar `r23' = `at'[1, 42]
>>> scalar `r24' = `at'[1, 43]
>>> scalar `r25' = `at'[1, 44]
>>> scalar `r26' = `at'[1, 45]
>>> scalar `r27' = `at'[1, 46]
>>> scalar `r28' = `at'[1, 47]
>>> scalar `r29' = `at'[1, 48]
>>>
>>> scalar `r31' = `at'[1, 49]
>>> scalar `r32' = `at'[1, 50]
>>> scalar `r33' = `at'[1, 51]
>>> scalar `r34' = `at'[1, 52]
>>> scalar `r35' = `at'[1, 53]
>>> scalar `r36' = `at'[1, 54]
>>> scalar `r37' = `at'[1, 55]
>>> scalar `r38' = `at'[1, 56]
>>> scalar `r39' = `at'[1, 57]
>>>
>>> scalar `r41' = `at'[1, 58]
>>> scalar `r42' = `at'[1, 59]
>>> scalar `r43' = `at'[1, 60]
>>> scalar `r44' = `at'[1, 61]
>>> scalar `r45' = `at'[1, 62]
>>> scalar `r46' = `at'[1, 63]
>>> scalar `r47' = `at'[1, 64]
>>> scalar `r48' = `at'[1, 65]
>>> scalar `r49' = `at'[1, 66]
>>>
>>> scalar `r51' = `at'[1, 67]
>>> scalar `r52' = `at'[1, 68]
>>> scalar `r53' = `at'[1, 69]
>>> scalar `r54' = `at'[1, 70]
>>> scalar `r55' = `at'[1, 71]
>>> scalar `r56' = `at'[1, 72]
>>> scalar `r57' = `at'[1, 73]
>>> scalar `r58' = `at'[1, 74]
>>> scalar `r59' = `at'[1, 75]
>>>
>>>
>>> tempname u1 u2 u3 u4 u5
>>>
>>> scalar `u1' = `at'[1, 76]
>>> scalar `u2' = `at'[1, 77]
>>> scalar `u3' = `at'[1, 78]
>>> scalar `u4' = `at'[1, 79]
>>> scalar `u5' = `at'[1, 80]
>>>
>>>
>>> tempname d1 d2 d3 d4 d5
>>>
>>> scalar `d1' = `at'[1, 81]
>>> scalar `d2' = `at'[1, 82]
>>> scalar `d3' = `at'[1, 83]
>>> scalar `d4' = `at'[1, 84]
>>> scalar `d5' = `at'[1, 85]
>>>
>>>
>>>
>>> quietly {
>>> * The b(p) term in the QUAIDS model:
>>> tempvar bofp
>>> gen double `bofp' = 0
>>> forvalues i = 1/6 {
>>> replace `bofp' = `bofp' + `lnp`i''*`b`i''
>>> }
>>> replace `bofp' = exp(`bofp')
>>> // Finally, the expenditure shares for 5 of the 6
>>> // goods (the equation 6 is dropped to avoid singularity)
>>>
>>>
>>> replace `w1' = [`a1' + `g11'*`lnp1' + `g12'*`lnp2' +`g13'*`lnp3' + `g14'*`lnp4' + `g15'*`lnp5' + `g16'*`lnp6' + `b1'*(`lnexp' - `lnpindex') + `l1'/`bofp'*(`lnexp' - `lnpindex')^2 + `r11'*`urban' + `r12'*`norte' + `r13'*`nordeste' + `r14'*`sudeste' + `r15'*`centro_oeste' + `r16'*`anosestudo' + `r17'*`mulherchefe' + `r18'*`crian_adolesc' + `r19'*`idoso' + `u1'*`lnMr_est']*`F1' + `d1'*`fe1'
>>>
>>> replace `w2' = [`a2' + `g21'*`lnp1' + `g22'*`lnp2' +`g23'*`lnp3' + `g24'*`lnp4' + `g25'*`lnp5' + `g26'*`lnp6' + `b2'*(`lnexp' - `lnpindex') + `l2'/`bofp'*(`lnexp' - `lnpindex')^2 + `r21'*`urban' + `r22'*`norte' + `r23'*`nordeste' + `r24'*`sudeste' + `r25'*`centro_oeste' + `r26'*`anosestudo' + `r27'*`mulherchefe' + `r28'*`crian_adolesc' + `r29'*`idoso' + `u2'*`lnMr_est']*`F2' + `d2'*`fe2'
>>>
>>> replace `w3' = [`a3' + `g31'*`lnp1' + `g32'*`lnp2' +`g33'*`lnp3' + `g34'*`lnp4' + `g35'*`lnp5' + `g36'*`lnp6' + `b3'*(`lnexp' - `lnpindex') + `l3'/`bofp'*(`lnexp' - `lnpindex')^2 + `r31'*`urban' + `r32'*`norte' + `r33'*`nordeste' + `r34'*`sudeste' + `r35'*`centro_oeste' + `r36'*`anosestudo' + `r37'*`mulherchefe' + `r38'*`crian_adolesc' + `r39'*`idoso' + `u3'*`lnMr_est']*`F3' + `d3'*`fe3'
>>>
>>> replace `w4' = [`a4' + `g41'*`lnp1' + `g42'*`lnp2' +`g43'*`lnp3' + `g44'*`lnp4' + `g45'*`lnp5' + `g46'*`lnp6' + `b4'*(`lnexp' - `lnpindex') + `l4'/`bofp'*(`lnexp' - `lnpindex')^2 + `r41'*`urban' + `r42'*`norte' + `r43'*`nordeste' + `r44'*`sudeste' + `r45'*`centro_oeste' + `r46'*`anosestudo' + `r47'*`mulherchefe' + `r48'*`crian_adolesc' + `r49'*`idoso' + `u4'*`lnMr_est']*`F4' + `d4'*`fe4'
>>>
>>> replace `w5' = [`a5' + `g51'*`lnp1' + `g52'*`lnp2' +`g53'*`lnp3' + `g54'*`lnp4' + `g55'*`lnp5' + `g56'*`lnp6' + `b5'*(`lnexp' - `lnpindex') + `l5'/`bofp'*(`lnexp' - `lnpindex')^2 + `r51'*`urban' + `r52'*`norte' + `r53'*`nordeste' + `r54'*`sudeste' + `r55'*`centro_oeste' + `r56'*`anosestudo' + `r57'*`mulherchefe' + `r58'*`crian_adolesc' + `r59'*`idoso' + `u5'*`lnMr_est']*`F5' + `d5'*`fe5'
>>>
>>> }
>>> end
>>>
>>>
>>> nlsur quaids @ w1 w2 w3 w4 w5 lnp1 lnp2 lnp3 lnp4 lnp5 lnp6 lnexp lnpindex urban norte nordeste sudeste centro_oeste anosestudo mulherchefe crian_adolesc idoso lnMr_est fe1 fe2 fe3 fe4 fe5 F1 F2 F3 F4 F5, ifgnls nequations(5) param(a1 a2 a3 a4 a5 b1 b2 b3 b4 b5 g11 g12 g13 g15 g22 g23 g25 g33 g35 g45 g55 l1 l2 l3 l4 l5 r11 r12 r13 r14 r15 r16 r17 r18 r19 r21 r22 r23 r24 r25 r26 r27 r28 r29 r31 r32 r33 r34 r35 r36 r37 r38 r39 r41 r42 r43 r44 r45 r46 r47 r48 r49 r51 r52 r53 r54 r55 r56 r57 r58 r59 u1 u2 u3 u4 u5 d1 d2 d3 d4 d5) nolog
>>>
>>>
>>> Thank you again Jorge Perez and Statalisters.
>>>
>>> Guilherme Travassos
>>> ----------------------------------------
>>>> From: jorge_perez@brown.edu
>>>> Date: Thu, 7 Nov 2013 20:27:51 -0500
>>>> Subject: Re: st: Separability Restrictions Nlsur Quaids
>>>> To: statalist@hsphsun2.harvard.edu
>>>>
>>>> - Have you checked your independent variables for missing values? From
>>>> the error message, that may be the problem.
>>>> - If that is not the problem, change your initial values such that
>>>> they imply that the initial `a3' + `b3' and `a5' + `b5' are not zero,
>>>> since you are dividing by those terms.
>>>> - If your purpose is only testing without actual estimation of the
>>>> restricted model, keep in mind that you could just test the
>>>> restrictions from the unrestricted model using -nlcom-
>>>> - Is your estimation something you can not do using Brian Poi's recent
>>>> -quaids- command?
>>>>
>>>> Hope this helps,
>>>>
>>>> Jorge Pérez.
>>>>
>>>>
>>>>
>>>>
>>>> --------------------------------------------
>>>> Jorge Eduardo Pérez Pérez
>>>> Graduate Student
>>>> Department of Economics
>>>> Brown University
>>>>
>>>>
>>>> On Thu, Nov 7, 2013 at 10:27 AM, Guilherme Travassos
>>>> <travassosgf@hotmail.com> wrote:
>>>>> Dear Statalisters,
>>>>>
>>>>> I'm using nlsur Quaids from Brian Poi (2008) to estimate a six group demand system for meat products. But I want to test the system for weak separability too.
>>>>>
>>>>> My routine is:
>>>>>
>>>>> nlsur quaids @ w1 w2 w3 w4 w5 lnp1 lnp2 lnp3 lnp4 lnp5 lnp6 lnexp lnpindex urban norte nordeste sudeste centro_oeste anosestudo mulherchefe crian_adolesc idoso lnMr_est fe1 fe2 fe3 fe4 fe5 F1 F2 F3 F4 F5, ifgnls nequations(5) param(a1 a2 a3 a4 a5 b1 b2 b3 b4 b5 g11 g12 g13 g14 g15 g22 g23 g24 g25 g33 g34 g35 g44 g45 g55 l1 l2 l3 l4 l5 r11 r12 r13 r14 r15 r16 r17 r18 r19 r21 r22 r23 r24 r25 r26 r27 r28 r29 r31 r32 r33 r34 r35 r36 r37 r38 r39 r41 r42 r43 r44 r45 r46 r47 r48 r49 r51 r52 r53 r54 r55 r56 r57 r58 r59 u1 u2 u3 u4 u5 d1 d2 d3 d4 d5) nolog
>>>>>
>>>>> I'm trying to compare two models: unrestricted model and restricted model. The unrestricted model is fine, but I don't know how to alter the Brian's code to test the weak separability restrictions.
>>>>>
>>>>> My restriction is something like this:
>>>>>
>>>>> scalar `g14' = ((`a4' + `b4')/(`a3' + `b3'))*(`g13' + `a1'*`a3') - `a1'*`a4'
>>>>> scalar `g16' = ((`a6' + `b6')/(`a5' + `b5'))*(`g15' + `a1'*`a5') - `a1'*`a6'
>>>>> scalar `g24' = ((`a4' + `b4')/(`a3' + `b3'))*(`g23' + `a2'*`a3') - `a2'*`a4'
>>>>> scalar `g26' = ((`a6' + `b6')/(`a5' + `b5'))*(`g25' + `a2'*`a5') - `a2'*`a6'
>>>>> scalar `g34' = ((`a4' + `b4')/(`a3' + `b3'))*(`g33' + `a3'*`a3') - `a3'*`a4'
>>>>> scalar `g36' = ((`a6' + `b6')/(`a5' + `b5'))*(`g35' + `a3'*`a5') - `a3'*`a6'
>>>>> scalar `g44' = ((`a4' + `b4')/(`a3' + `b3'))*(`g43' + `a4'*`a3') - `a4'*`a4'
>>>>> scalar `g46' = ((`a6' + `b6')/(`a5' + `b5'))*(`g45' + `a4'*`a5') - `a4'*`a6'
>>>>> scalar `g56' = ((`a6' + `b6')/(`a5' + `b5'))*(`g55' + `a5'*`a5') - `a5'*`a6'
>>>>>
>>>>> When I start the model, the stata send me this message:
>>>>>
>>>>> could not evaluate equation 1
>>>>> starting values invalid or some RHS variables have missing values
>>>>> r(480);
>>>>>
>>>>> Can anyone help me?
>>>>> I really appreciate it!
>>>>>
>>>>> *
>>>>> * For searches and help try:
>>>>> * http://www.stata.com/help.cgi?search
>>>>> * http://www.stata.com/support/faqs/resources/statalist-faq/
>>>>> * http://www.ats.ucla.edu/stat/stata/
>>>>
>>>> *
>>>> * For searches and help try:
>>>> * http://www.stata.com/help.cgi?search
>>>> * http://www.stata.com/support/faqs/resources/statalist-faq/
>>>> * http://www.ats.ucla.edu/stat/stata/
>>> *
>>> * For searches and help try:
>>> * http://www.stata.com/help.cgi?search
>>> * http://www.stata.com/support/faqs/resources/statalist-faq/
>>> * http://www.ats.ucla.edu/stat/stata/
>>
>> *
>> * For searches and help try:
>> * http://www.stata.com/help.cgi?search
>> * http://www.stata.com/support/faqs/resources/statalist-faq/
>> * http://www.ats.ucla.edu/stat/stata/
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/