Positive own price elasticity using QUAIDS

[Amlan Das Gupta]

Dear Statalisters,

I am trying to estimate the price and expenditure elasticity using QUAIDS command of Brian Poi (http://www.stata-journal.com/sjpdf.h...iclenum=st0268). This is the command I use:

quaids w_beef w_milk ex_share_all, ///
anot(5) prices(price_beef price_milk p) expenditure(CE)  ///
demographics(mil_beef mil_mlk)

While running post-estimation commands, the elasticities evaluated at means using:

estat expenditure, atmeans stderrs

and

estat compensated, atmeans stderrs

are giving me perfect results. However when I try to find own price elasticity for each observation using:

estat compensated e_p*

I invariably get more than 10% of the sample with a positive own price elasticity of milk which should not be the case. I am wondering if I am making some obvious error.

I would appreciate any help that leads me to some resolution of this mystery.

I am using Stata 13.1. My data is household level survey data of about a 1,00,000 observations.

[Abdul Adam]

I can't see issues with the command. But, two things come to mind: firstly, I would check if the prices and expenditures that I entered have issues: do they make sense? Secondly, though this is rare, your actual data may be against the law of demand. Can you think of reasons why people actually might buy more milk while the price is raised. Could it also be the case that something else happened, in parallel to a price increase, that can explain what you observe?

[Henni Burk]

Dear Amlan,

I am also using the quaids command QUAIDS command by Brian Poi (http://www.stata-journal.com/sjpdf.h...iclenum=st0268). And consequently also the same commands for the estimation of own and corss price elasticities.

I now encounter the problem that I struggle to incorporate asterisks (*) that are based on the p-values (0.05, 0.01, 0.001) in the output tables of the elasticities. Since you already worked with the command quaids, I wonder if you have a solution to that named problem?


For simplification I demonstrate my problem with the help of the food.dta file (after downloading the respective ado-file)

*installing quaids.ado
help st0268_1


webuse food
generate nkids = int(runiform()*4)
generate rural = (runiform() >0.7)

quaids w1-w4, anot(10) prices (p1-p4) expenditure(expfd) demographics (nkids rural) nolog

estat expenditure, atmeans stderrs
matrix elas =r(expelas)
matrix list elas
matrix list r(sd)

-> now I would like to have the elas matrix with significance asterisks. , but by typing:

estimates table, star(.05 .01 .001) // only works for first table, but not matrix (elasticity table "elas"

* install needed package: estout
ssc install estout, replace
estimates store runcompelas
estout runcompelas, cells(sham(star))

-> asterisks are being put on the quaids parameter table not elasticity table.

the same issue occurs when trying the folloing commands


*trying another version: parmest
ssc describe parmest
ssc install parmest
help parmest
parmest, list(,) stars(0.05 0.01 0.001) // only works for parameter table not for matrix


(Stata 16 on Windows 7 used)


I appreciate any help.

Best,
Henni