Error with testparm after saving/using estimation results

[Scott Adams]

Hi,
I am getting an error when trying -testparm- after saving and then reloading (using) estimation results from an mlogit model.
The equivalent test with -test- works.
I am not sure if it is a bug or something I need to specify differently with testparm for reloaded estimates. The problem seems to be related to the equation option in testparm.


Version 18, updated 30 Apr 2024. (I am not admin and cannot update to latest -- perhaps this was fixed -- is there a way to find the -whatsnew- lists outside of Stata?)
Thanks,
Scott

clear all
set obs 1000
set seed 99
gen y = runiformint(1,4)
gen x1=runiformint(1,4)
gen x2=runiformint(1,4)

mlogit y ib1.x1##ib1.x2, base(1)
testparm 3.x1#i.x2, equation(2) //works

est store m1
est restore m1
testparm 3.x1#i.x2, equation(2) //usually works

est save "m1", replace
est clear

est use "m1"
testparm 3.x1#i.x2 //works although not interesting since base equation
testparm 3.x1#i.x2, equation(1) //error
testparm 3.x1#i.x2, equation(2) //error "no such variables" altho clearly listed in e(b)

//same test - not see errors so far
test ///
 (   [2]3.x1#2.x2 = 0 ) ///
 (   [2]3.x1#3.x2 = 0 ) ///
 (   [2]3.x1#4.x2 = 0 )

[George Ford]

the testpart equation(1) doesn't work even after the mlogit model.

but i don't use this save approach, so don't know why it doesn't work.