Rescaling y-axis of gapfigure

[Harina Peternella]

Hi everyone,


Currently, I am running a synthetic control with the command allsynth whereby I am using the option "gap figure". According to the help function one can use two-way_options.
I want to change the scale of the y-axis since it is way too large to show the discrepancy between the synthetic control and the trend of the treatment country (see picture 1).

When adding the option ylab(-250000 250000) does change the labels but the graph is not rescaled (see picture 2).

I thought the following command (looking at the added option) should have fixed this, however, Stata returns the same graph as picture 1 in which the range of the y axis is not defined...

allsynth road_tkm lag_road_tkm gdp_capita road_investment(1996(1)2013) share_road(1994(1)2013) FDI(1994(1)2013) emp(1990(1)2013) cn(1990(1)2013) pl_x(1990(1)2013) pl_m(1990(1)2013), trunit(70) trperiod(2003) unitnames(country) keep(synth1.dta, replace) gapfigure(classic lineback, ysc(r(-250000 250000))) nested allopt

Does anyone know what I did wrong?

Thank you in advance.

[Andrew Musau]

You are at the mercy of those who use that command which you should identify as community-contributed (FAQ Advice #12). In general, Stata will not truncate data, and it is not apparent what is the underlying data that the command uses. I have a pair of suggestions for generating the graph using twoway:

1. Obtain the gap variable and plot it, or
2. Retrieve data from the graph and plot.

An example of #2 is #10 of https://www.statalist.org/forums/for...s-in-one-graph

[Harina Peternella]

Thank you for your reply, Andrew Musau, I will generate the gap variable by myself and plot it with a normal graph.

[Justin Wiltshire]

Two things you should note. First, you have specified

gapfigure(classic lineback, ysc(r(-250000 250000))

, which tells allsynth that you want to plot the gaps (the differences between the treated-unit outcome values and the synthetic control outcome values). If you want to plot the treated-unit outcome values and the synthetic control outcome values, I recommend you tell Stata to load your synth1.dta file, where you saved the results of the synthetic control estimation, and then simply plot the two series using e.g.

twoway (line _Y_treated _time, lp(solid)) (line _Y_s _time, lp(dash)) [, twoway_opts]

I wanted to retain all of the functionality and syntax of synth with allsynth, so I did not change the figure option to make that figure customizable.

Second, try

gapfigure(classic lineback, ysc(r(-250000 250000) ylabel(-250000(125000)250000))