Adding a custom column to esttab

[Abhishek Nagaraj]

Hello, could someone help me figure out how to achieve the following output using estout/esttab?

Basically, I would like the first column to include "regular" results from a regression model, and I'd like the second column to have "custom" values that I would like to define.

I tried a few different approaches. The one that came closest was to use use return repost and replace values in e(b) with my custom values (see below). The trouble is, I dont know how to show the std. errors and stars for the first column, but not for the second.

Any help? Thanks in advance!
------

Code so far:

est clear
eststo: reg price weight mpg

program define alterations, eclass
estimates restore est1
matrix m = e(b)
matrix m[1,1]=30
matrix m[1,2]=26
matrix m[1,3]=35
ereturn repost b=m
end

// change coeffs
alterations

eststo est2

eststo est3: reg price weight mpg

esttab est3 est2

[Liu Qiang]

what' the purpose of doing so?

[Abhishek Nagaraj]

I'm interested in specifying implied percent effects for each coefficient in that additional column ...

[Andrew Musau]

If you can live with blank parentheses, then just modify the variance matrix

matrix v = J(3,3,0)
ereturn repost V=v

Specifying multiple rules for columns in the same table is not desirable because it ends up confusing the reader. Therefore, I do not think that there is a direct solution for your problem.

[Liu Qiang]

If you can live with blank parentheses, then just modify the variance matrix

matrix v = J(3,3,0)
ereturn repost V=v

Specifying multiple rules for columns in the same table is not desirable because it ends up confusing the reader. Therefore, I do not think that there is a direct solution for your problem.

totally agree! Although he can replace the coefficient with the desired or so-called custom numbers, he still cannot estimate the standard error. It would be confounding if he uses the previous estimation results. Thus it's meaningless to do so.