This folder includes the following:
(updated 5/27/21)

**********************************************
1. Raw data figures by adopter group
**********************************************
Figures with the raw data per Editor's suggestion:
split the counties into early adopters, middle adopters, and late adopters,
then plot long-term outcomes against birth cohorts, separately for the three
types of counties

These figures are located in subfolder /figures/ and titled
"z_composite_all_by_adopter_group.pdf",
"ess_7_all_by_adopter_group.pdf", etc.

***New: The excel sheet with raw data means is in subfolder /figures/
titled adopter_group_means.xls

**********************************************
2. Residuals Plot
**********************************************
Regress the conposite index on county FE, state x year
FE, and the linear trends interacted with 1960 county characteristics,
then plot the residuals for each of the adopter groups 

This figure is titled "residual_z_composite_all_by_adopter_group.pdf" and are
located in subfolder /figures/.

**********************************************
3. Exposure Model by early and late adopters
**********************************************
Run the exposure model for Share FS %IU-5 separately by early and late
adopters (columns (1) and (2)). 

***New: Alternative to stratifying, we run the exposure
model for Share FSP %IU-5 including an interaction term for share IU-5 x early
(column (3)).

These are located in subfolder /exposure/, titled "z_composite_adopter.xls",
etc.

**********************************************
Construction Note
**********************************************
We split the counties into early adopters, middle adopters, and late adopters
as follows:
a) Limit observations to the analysis sample for z_composite_all (observations
which are nonmissing for this outcome and our controls)
b) Keep one birth cohort per county (choose the birth cohort closest to when
food stamps was introduced in that county)
c) Find the year in which 1/3 of this population has been exposed to food
stamps (1966). Counties with food stamp this year or earlier
are "early adopters"
d) Find the year in which 2/3 of the population has been exposed to food stamps
(1969). Counties which introduced food stamps between this year and the early
year are "middle adopters". Counties which introduce food stamps after this year
are "late adopters" .

For (3), we only define two groups: early and late. We define groups similarly
to the description above, but use the year (1968) in which the median share of
the population was exposed as the cutoff for early and late (early includes
cutoff year).