Using Stata's esttab, add dollar sign to cell format, export to Latex

Question

My advisor wants me to add dollar signs to my table of summary statistics. I generate this table and export it to Latex using Stata's esttab command.

I need to 1) Add dollar signs to some of the results cells (not all) and 2) Make sure that Latex can handle the dollar signs.

I think that 2 might be accomplished using the substitute option, but I can't figure out how to do 1. Here is some minimal code that I am trying to use to solve this problem.

sysuse auto, clear

estpost summarize price mpg weight length if foreign==0
est store A
estpost summarize price mpg weight length if foreign==1
est store B

esttab A B  using $root/Outputs/test.tex, replace ///  //a file path on my machine
cells("mean (fmt(%9.0fc %9.2fc %9.0fc))" "sd(par fmt(%9.0fc %9.2fc %9.0fc))") ///
mtitle("Domestic" "Foreign") ///
mgroups("Type", pattern(1 0) prefix(\multicolumn{@span}{c}{) suffix(}) span erepeat( \cmidrule(lr){@span})) ///
nonumber booktabs f label  collabels(none)

eststo clear

This produces:

                    &\multicolumn{2}{c}{Type} \\\cmidrule(lr){2-3}
                    &\multicolumn{1}{c}{Domestic}&\multicolumn{1}{c}{Foreign}\\
\midrule
Price               &       6,072&       6,385\\
                    &     (3,097)&     (2,622)\\
Mileage (mpg)       &       19.83&       24.77\\
                    &      (4.74)&      (6.61)\\
Weight (lbs.)       &       3,317&       2,316\\
                    &       (695)&       (433)\\
Length (in.)        &         196&         169\\
                    &        (20)&        (14)\\
\midrule
Observations        &          52&          22\\

I'd like to get it so the output would have \$ in front of the 6,072 and the 6,385

I see some discussion on the Statalist regarding workarounds for graphs, but nothing for esttab. Someone also mentions creating "custom formats" but I can't seem to find documentation on that anywhere.

Answer 1

I had a similar problem once: I wanted to color cells differently depending on the significance level. In the end, the easiest automated solution I could come up with was modifying the esttab code... That is easier done than it sounds, in fact.

Look for the following code in estout.ado:

if `:length local thevalue'<245 {
    local thevalue: di `fmt_m' `"`macval(thevalue)'"'
}

Just after that you can insert, e.g.

local thevalue `"\$`macval(thevalue)'\$"'

That would produce:

                    &\multicolumn{2}{c}{Type} \\\cmidrule(lr){2-3}
                    &\multicolumn{1}{c}{Domestic}&\multicolumn{1}{c}{Foreign}\\
\midrule
Price               &$       6,072$&$       6,385$\\
                    &$     (3,097)$&$     (2,622)$\\
Mileage (mpg)       &$       19.83$&$       24.77$\\
                    &$      (4.74)$&$      (6.61)$\\
Weight (lbs.)       &$       3,317$&$       2,316$\\
                    &$       (695)$&$       (433)$\\
Length (in.)        &$         196$&$         169$\\
                    &$        (20)$&$        (14)$\\
\midrule
Observations        &          52&          22\\

(Don't forget to program drop estout before exporting, so that the .ado reloads)

So, all the numeric values in the main table are encapsulated in $ signs. If you want specific values only, you can do a simple regex condition. E.g., if you care capturing only those values where there is a comma (for whatever reason), you can do:

if strpos("`macval(thevalue)'", ",") {
    local thevalue `"\$`macval(thevalue)'\$"'
}

And you can also add you own option (just in the beginning of estout.ado), so that the modified behaviour is not triggered all the time.