DiD
Aguiar28D
Part 1. Package DiD exploration
1.Dataset and package
library(did)
data(mpdta)2. To estimate group-time average treatment effects
lemp This is the log of county-level teen employment. It is the outcome variable
first.treat This is the period when a state first increases its minimum wage. It can be 2004, 2006, or 2007. It is the variable that defines group in this application
year This is the year and is the time variable
countyreal This is an id number for each county and provides the individual identifier in this panel data context
out <- att_gt(yname = "lemp",
gname = "first.treat",
idname = "countyreal",
tname = "year",
xformla = ~1,
data = mpdta,
est_method = "reg"
)
summary(out)Call: att_gt(yname = “lemp”, tname = “year”, idname = “countyreal”, gname = “first.treat”, xformla = ~1, data = mpdta, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2004 2004 -0.0105 0.0237 -0.0740 0.0530
2004 2005 -0.0704 0.0330 -0.1588 0.0180
2004 2006 -0.1373 0.0404 -0.2455 -0.0291 2004 2007 -0.1008 0.0364 -0.1982 -0.0034 2006 2004 0.0065 0.0243 -0.0584 0.0715
2006 2005 -0.0028 0.0205 -0.0575 0.0520
2006 2006 -0.0046 0.0162 -0.0481 0.0389
2006 2007 -0.0412 0.0203 -0.0955 0.0130
2007 2004 0.0305 0.0142 -0.0074 0.0684
2007 2005 -0.0027 0.0162 -0.0461 0.0406
2007 2006 -0.0311 0.0178 -0.0787 0.0166
2007 2007 -0.0261 0.0174 -0.0725 0.0204
— Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0.16812 Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression
3. Group-time average treatment effects
ggdid(out, ylim = c(-.25,.1))
4. Event Studies
A main type of aggregation is into an event study plot.
es <- aggte(out, type = "dynamic")
#summary(es)
ggdid(es)
The figure here is very similar to the group-time average treatment effects. Red dots are pre-treatment periods, blue dots are post-treatment periods. The difference is that the x-axis is in event time.
5. Overall Effect of Participating in the Treatment
The event study above reported an overall effect of participating in the treatment. This was computed by averaging the average effects computed at each length of exposure.
In many cases, a more general purpose overall treatment effect parameter is given by computing the average treatment effect for each group, and then averaging across groups. This sort of procedure provides an average treatment effect parameter with a very similar interpretation to the Average Treatment Effect on the Treated (ATT) in the two period and two group case.
To compute this overall average treatment effect parameter, use
group_effects <- aggte(out, type = "group")
summary(group_effects)Call: aggte(MP = out, type = “group”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Overall ATT:
ATT Std. Error [95% Conf. Int.]
-0.031 0.0128 -0.0561 -0.006 *
Group Effects: group ATT Std. Error [95% Simult. Conf. Band]
2004 -0.0797 0.0281 -0.1429 -0.0166 2006 -0.0229 0.0172 -0.0616 0.0158
2007 -0.0261 0.0155 -0.0610 0.0088
— Signif. codes: `’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression
Of particular interest is the Overall ATT in the results. Here, we estimate that increasing the minimum wage decreased teen employment by 3.1% and the effect is marginally statistically significant.
6. Set control group to not yet treated
out06 <- att_gt(yname = "lemp",
gname = "first.treat",
idname = "countyreal",
tname = "year",
xformla = ~1,
data = mpdta,
est_method = "reg",
control_group = "notyettreated"
)
tab06<-summary(out06)Call: att_gt(yname = “lemp”, tname = “year”, idname = “countyreal”, gname = “first.treat”, xformla = ~1, data = mpdta, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2004 2004 -0.0194 0.0240 -0.0821 0.0434
2004 2005 -0.0783 0.0317 -0.1613 0.0047
2004 2006 -0.1363 0.0353 -0.2286 -0.0439 2004 2007 -0.1008 0.0353 -0.1932 -0.0084 2006 2004 -0.0026 0.0232 -0.0633 0.0582
2006 2005 -0.0019 0.0189 -0.0514 0.0476
2006 2006 0.0047 0.0173 -0.0407 0.0500
2006 2007 -0.0412 0.0209 -0.0959 0.0134
2007 2004 0.0298 0.0141 -0.0070 0.0666
2007 2005 -0.0024 0.0163 -0.0451 0.0403
2007 2006 -0.0311 0.0189 -0.0805 0.0183
2007 2007 -0.0261 0.0176 -0.0722 0.0201
— Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0.16814 Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression
7. Drop the never treated and then set control group to never treated
#Here I filter the non-treated by excluding those that have a first treatment valur of 0
mpdta07<-filter(mpdta, first.treat > 0)
#Here I run the same function but in the filtered dataset
out07 <- att_gt(yname = "lemp",
gname = "first.treat",
idname = "countyreal",
tname = "year",
xformla = ~1,
data = mpdta07,
est_method = "reg",
control_group = "notyettreated"
)
tab07<-summary(out07)Call: att_gt(yname = “lemp”, tname = “year”, idname = “countyreal”, gname = “first.treat”, xformla = ~1, data = mpdta07, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2004 2004 -0.0354 0.0243 -0.0959 0.0251
2004 2005 -0.0926 0.0345 -0.1784 -0.0068 2004 2006 -0.1340 0.0425 -0.2398 -0.0281 2006 2004 -0.0240 0.0240 -0.0838 0.0358
2006 2005 0.0000 0.0222 -0.0552 0.0551
2006 2006 0.0265 0.0194 -0.0218 0.0747
— Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0.60722 Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression ## 8. Check if I can write two types at the same time (my guess is not)
I will not use the non-yet-treated control for this. The type options mentioned are: simple, dynamic, group, calendar. *I am going to try dynamic and group (instead of event)
#agg_08 <- aggte(out06, type = "dynamic", type = "group")
#summary(agg_08)This is the error that I got: *Error in aggte(out06, type = “dynamic”, type = “group”) : formal argument “type” matched by multiple actual arguments
8.a type = dynamic
agg_08a <- aggte(out06, type = "dynamic")
tab08a<-summary(agg_08a)Call: aggte(MP = out06, type = “dynamic”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Overall ATT:
ATT Std. Error [95% Conf. Int.]
-0.0774 0.0222 -0.121 -0.0338 *
Dynamic Effects: event time ATT Std. Error [95% Simult. Conf. Band]
-3 0.0298 0.0136 -0.0055 0.0650
-2 -0.0024 0.0124 -0.0347 0.0298
-1 -0.0243 0.0140 -0.0605 0.0119
0 -0.0189 0.0121 -0.0503 0.0125
1 -0.0536 0.0167 -0.0969 -0.0103 2 -0.1363 0.0399 -0.2396 -0.0329 3 -0.1008 0.0365 -0.1954 -0.0062 — Signif. codes: `’ confidence band does not cover 0
Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression
8.b type = dynamic
agg_08b <- aggte(out06, type = "group")
tab08b<-summary(agg_08b)Call: aggte(MP = out06, type = “group”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Overall ATT:
ATT Std. Error [95% Conf. Int.]
-0.0305 0.0127 -0.0553 -0.0056 *
Group Effects: group ATT Std. Error [95% Simult. Conf. Band]
2004 -0.0837 0.0288 -0.1460 -0.0214 2006 -0.0183 0.0154 -0.0516 0.0150
2007 -0.0261 0.0180 -0.0648 0.0127
— Signif. codes: `’ confidence band does not cover 0
Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression
9 Run the package with weights
#Lets begin by creating a new variable that assigns random weights: (I know that this is not exactly what I was asked to do, but there are too many counties on this dataset, I can modify this later):
mpdta09<-mutate(mpdta,
weights_var = sample(x = 1:2, size = 2500, replace = TRUE))
tab09<-head(mpdta09, n=10)
kable(tab09)| year | countyreal | lpop | lemp | first.treat | treat | weights_var | |
|---|---|---|---|---|---|---|---|
| 866 | 2003 | 8001 | 5.896761 | 8.461469 | 2007 | 1 | 1 |
| 841 | 2004 | 8001 | 5.896761 | 8.336870 | 2007 | 1 | 2 |
| 842 | 2005 | 8001 | 5.896761 | 8.340217 | 2007 | 1 | 1 |
| 819 | 2006 | 8001 | 5.896761 | 8.378161 | 2007 | 1 | 2 |
| 827 | 2007 | 8001 | 5.896761 | 8.487352 | 2007 | 1 | 2 |
| 937 | 2003 | 8019 | 2.232377 | 4.997212 | 2007 | 1 | 2 |
| 938 | 2004 | 8019 | 2.232377 | 5.081404 | 2007 | 1 | 1 |
| 939 | 2005 | 8019 | 2.232377 | 4.787492 | 2007 | 1 | 1 |
| 940 | 2006 | 8019 | 2.232377 | 4.990433 | 2007 | 1 | 1 |
| 941 | 2007 | 8019 | 2.232377 | 5.036953 | 2007 | 1 | 1 |
Part 2. Trying the package in the subsample dataset
2.1. Identifying the variables
This are the main variables in the main dataset:
lemp This is the log of county-level teen employment. It is the outcome variable
first.treat This is the period when a state first increases its minimum wage. It can be 2004, 2006, or 2007. It is the variable that defines group in this application
year This is the year and is the time variable
countyreal This is an id number for each county and provides the individual identifier in this panel data context
Now let’s identify them in the sample ataset:
meantax mean amount of taxes
vat_year1 This is the period when a state first implemented the VAT. It is the variable that defines group in this application
year This is the year and is the time variable
state_id This is an id number for each state and provides the individual identifier in this panel data context
2.2. Running the model: attempt 1
IMPORTANT!!! I was getting this message: missing value where TRUE/FALSE needed
my_data_02<-na.exclude(my_data)
tab2.2<-head(my_data_02)
kable(tab2.2)| year | state | meantax | maxtax | mintax | mediantax | modetax | stdevtax | residual_tax_rate | number_items | number_tax_rates | state_id | igrowth_rate_gsdp_constantprices | logitotal_foodgrains_ag_year | logipop | logihighways | logitotalipccrimes | vat_year1 | j_salegross00 | j_salegross02 | state_election_year | bjp_state_coalition | inc_state_coalition | state_part_of_center | state_support_center | state_aligned_with_center | bjp_vote_share | inc_vote_share | bjp_seat_share | inc_seat_share | election_happening | _merge |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2000 | Andhra Pradesh | 8.254247 | 70 | 0 | 8 | 8 | 5.621067 | 12.0 | 478 | 12 | 28 | 8.16 | 10.546257 | 18.13041 | 9.098739 | 13.05252 | 2005 | 735678437301 | 915436398533 | 1 | 1 | 0 | 1 | 0 | 1 | 3.67 | 40.61 | 4.0816326 | 30.95238 | 1999 | 3 |
| 2002 | Andhra Pradesh | 9.857958 | 70 | 0 | 9 | 4 | 6.117981 | 12.0 | 475 | 17 | 28 | 2.73 | 9.966801 | 18.15465 | 9.098739 | 13.26115 | 2005 | 735678437301 | 915436398533 | 0 | 1 | 0 | 1 | 0 | 1 | 3.67 | 40.61 | 4.0816326 | 30.95238 | 1999 | 3 |
| 2005 | Andhra Pradesh | 6.694359 | 90 | 0 | 4 | 4 | 12.161378 | 12.5 | 511 | 8 | 28 | 9.57 | 10.431230 | 18.18977 | 9.098739 | 13.35108 | 2005 | 735678437301 | 915436398533 | 1 | 0 | 1 | 1 | 0 | 1 | 2.63 | 38.56 | 0.6802721 | 62.92517 | 2004 | 3 |
| 2006 | Andhra Pradesh | 6.300332 | 70 | 0 | 4 | 4 | 10.968633 | 12.5 | 530 | 6 | 28 | 11.18 | 10.387702 | 18.20120 | 9.098739 | 13.45258 | 2005 | 735678437301 | 915436398533 | 0 | 0 | 1 | 1 | 0 | 1 | 2.63 | 38.56 | 0.6802721 | 62.92517 | 2004 | 3 |
| 2007 | Andhra Pradesh | 5.443012 | 70 | 0 | 4 | 4 | 10.748420 | 12.5 | 587 | 7 | 28 | 12.02 | 10.561163 | 18.21251 | 9.098739 | 13.45933 | 2005 | 735678437301 | 915436398533 | 0 | 0 | 1 | 1 | 0 | 1 | 2.63 | 38.56 | 0.6802721 | 62.92517 | 2004 | 3 |
| 2008 | Andhra Pradesh | 5.333195 | 70 | 0 | 4 | 4 | 10.716602 | 12.5 | 591 | 8 | 28 | 6.88 | 10.617466 | 18.22369 | 9.098739 | 13.48297 | 2005 | 735678437301 | 915436398533 | 0 | 0 | 1 | 1 | 0 | 1 | 2.63 | 38.56 | 0.6802721 | 62.92517 | 2004 | 3 |
#Problem I end up with 256 observations.2.3. Running the model: attempt 2
IMPORTANT!!! Once I remove the NA observations I end up with this error. There is no available never-treated group.
#I am going to change the control group.
out_2.1 <- att_gt(yname = "meantax",
gname = "vat_year1",
idname = "state_id",
tname = "year",
xformla = ~1,
data = my_data_02,
est_method = "reg",
control_group = "notyettreated"
)2.4. Running the model: attempt 3
New error: No pre-treatment periods to test
List of errors (only first 10):
1: In pre_process_did(yname = yname, tname = tname, idname = idname, … :
Be aware that there are some small groups in your dataset.
Check groups: 0,2003,2006,2007.
2: In compute.att_gt(dp) :
Not enough control units for group 2003 in time period 2002 to run specified regression
3: In compute.att_gt(dp) :
Not enough control units for group 2003 in time period 2003 to run specified regression
4: In compute.att_gt(dp) :
Not enough control units for group 2003 in time period 2004 to run specified regression
5: In compute.att_gt(dp) :
Not enough control units for group 2003 in time period 2005 to run specified regression
6: In compute.att_gt(dp) :
Not enough control units for group 2003 in time period 2006 to run specified regression
7: In compute.att_gt(dp) :
Not enough control units for group 2003 in time period 2007 to run specified regression
8: In max(abs(b/bSigma)) : no non-missing arguments to max; returning -Inf
9: In max(abs(b/bSigma)) : no non-missing arguments to max; returning -Inf
10: In max(abs(b/bSigma)) : no non-missing arguments to max; returning -Inf
#I know this is not ideal but I a going to exclude the group 2003
my_data_03 <- filter(my_data_02, vat_year1>2003)
#out_2.1 <- att_gt(yname = "meantax",
# gname = "vat_year1",
# idname = "state_id",
# tname = "year",
# xformla = ~1,
# data = my_data_03,
# est_method = "reg",
# control_group = "notyettreated"
# )
#summary(out_2.1)2.5. Running the model: attempt 4
New error: No pre-treatment periods to test
List of errors (only first 10):
1: In pre_process_did(yname = yname, tname = tname, idname = idname, … :
Be aware that there are some small groups in your dataset.
Check groups: 0,2006,2007.
2: In compute.att_gt(dp) :
Not enough control units for group 2005 in time period 2002 to run specified regression
3: In compute.att_gt(dp) :
Not enough control units for group 2005 in time period 2005 to run specified regression
4: In compute.att_gt(dp) :
Not enough control units for group 2005 in time period 2006 to run specified regression
5: In compute.att_gt(dp) :
Not enough control units for group 2005 in time period 2007 to run specified regression
6: In max(abs(b/bSigma)) : no non-missing arguments to max; returning -Inf
7: In max(abs(b/bSigma)) : no non-missing arguments to max; returning -Inf
8: In max(abs(b/bSigma)) : no non-missing arguments to max; returning -Inf
9: In max(abs(b/bSigma)) : no non-missing arguments to max; returning -Inf
10: In max(abs(b/bSigma)) : no non-missing arguments to max; returning -Inf
3. Let’s start again
3.1. Cleaning the database
I will try to clean the database in a different way
#I am going to select the columns that I am going to use
selected_vars <- c("meantax", "vat_year1", "state_id", "year")
my_data_3.1<- my_data
my_data_3.1<-my_data_3.1[selected_vars]
#out_2.1 <- att_gt(yname = "meantax",
# gname = "vat_year1",
# idname = "state_id",
# tname = "year",
# xformla = ~1,
# data = my_data_3.1,
# est_method = "reg",
# control_group = "notyettreated"
# )
#summary(out_2.1)Error in if ((glist[g] <= tlist[(t + 1)])) { :
missing value where TRUE/FALSE needed
In addition: Warning messages:
1: In pre_process_did(yname = yname, tname = tname, idname = idname, :
Be aware that there are some small groups in your dataset.
Check groups: 0,2003,2007,2008.
2: In pre_process_did(yname = yname, tname = tname, idname = idname, :
Dropped 1 observations that had missing data.
3.2. Take out the NA
I will try to clean the database in a different way
my_data_3.2<-na.exclude(my_data_3.1)
#out_3.2 <- att_gt(yname = "meantax",
# gname = "vat_year1",
# idname = "state_id",
# tname = "year",
# xformla = ~1, #If we wanted to include controls it would be here
# data = my_data_3.2,
# est_method = "reg",
# control_group = "notyettreated",
# allow_unbalanced_panel = TRUE
# )
#summary(my_data_3.2)
#summary(my_data_3.2$vat_year1)
#summary(out_3.2)Error in if ((glist[g] <= tlist[(t + 1)])) { :
missing value where TRUE/FALSE needed
In addition: Warning message:
In pre_process_did(yname = yname, tname = tname, idname = idname, :
Be aware that there are some small groups in your dataset.
Check groups: 0,2003,2006,2007,2008.
3.3. Try to duplicate observations
my_data_3.3<-my_data_3.2[rep(seq_len(nrow(my_data_3.2)), each = 5), ]
#out_3.3 <- att_gt(yname = "meantax",
# gname = "vat_year1",
# idname = "state_id",
# tname = "year",
# xformla = ~1, #If we wanted to include controls it would be here
# data = my_data_3.3,
# est_method = "reg",
# control_group = "notyettreated"
# )
#summary(my_data_3.2)
#summary(out_3.2)3.4. The first observation had more years than the others, that was the problem
my_data_3.4<-filter(my_data_3.2, year!=2001, year!=2003, year!=2004)
out_3.4 <- att_gt(yname = "meantax",
gname = "vat_year1",
idname = "state_id",
tname = "year",
xformla = ~1, #If we wanted to include controls it would be here
data = my_data_3.4,
est_method = "reg",
control_group = "notyettreated"
)
#summary(out_3.4)
Results
3.5. Graph: event study
es_3.5 <- aggte(out_3.4, type = "dynamic")
ggdid(es_3.5)
3.6. Group time-average treatment effects
ggdid(out_3.4)
4.1 Trying different models - changing the dependent variable
y_vars <-c("meantax", "maxtax", "mintax", "mediantax", "modetax", "stdevtax", "residual_tax_rate", "number_items", "number_tax_rates")
controls <- c("igrowth_rate_gsdp_constantprices", "state_aligned_with_center")
selected_vars <- c(y_vars, "vat_year1", "state_id", "year", controls)
my_data_4.1<- my_data
my_data_4.1<-my_data_4.1[selected_vars]
my_data_4.1<-na.exclude(my_data_4.1)
my_data_4.1<-filter(my_data_4.1, year!=2001, year!=2003, year!=2004)
for (i in 1:9){
iy<- y_vars[i]
out_4.1 <- att_gt(yname = iy,
gname = "vat_year1",
idname = "state_id",
tname = "year",
xformla = ~1,
data = my_data_4.1,
est_method = "reg",
control_group = "notyettreated"
)
summary(out_4.1)
es_4.1 <- aggte(out_4.1, type = "dynamic")
Event_study<-ggdid(es_4.1)
print(Event_study)
Plots<-ggdid(out_4.1)
print(Plots)
}Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 -0.1154 0.1539 NA NA NA 2003 2005 -1.3586 0.0000 NA NA NA 2003 2006 -1.4644 0.0000 NA NA NA 2003 2007 -1.7988 0.0000 NA NA NA 2003 2008 -1.8313 0.0000 NA NA NA 2003 2009 -1.9815 0.0000 NA NA NA 2003 2010 -1.6711 0.0000 NA NA NA 2003 2011 -1.6874 0.0000 NA NA NA 2003 2012 -1.6945 0.0000 NA NA NA 2005 2002 0.1333 0.1647 NA NA NA 2005 2005 -1.3607 0.4160 NA NA NA 2005 2006 -1.8718 0.4048 NA NA NA 2005 2007 -2.0140 0.4378 NA NA NA 2005 2008 -2.1120 0.4467 NA NA NA 2005 2009 -2.1671 0.4474 NA NA NA 2005 2010 -1.8857 0.4564 NA NA NA 2005 2011 -1.6684 0.5007 NA NA NA 2005 2012 -1.5571 0.4771 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0.53574 Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 -4.1667 2.9515 NA NA NA 2003 2005 0.0000 0.0000 NA NA NA 2003 2006 0.0000 0.0000 NA NA NA 2003 2007 0.0000 0.0000 NA NA NA 2003 2008 0.0000 0.0000 NA NA NA 2003 2009 5.0000 0.0000 NA NA NA 2003 2010 5.0000 0.0000 NA NA NA 2003 2011 5.0000 0.0000 NA NA NA 2003 2012 5.0000 0.0000 NA NA NA 2005 2002 4.4118 3.0011 NA NA NA 2005 2005 -12.0294 7.5413 NA NA NA 2005 2006 -13.2059 7.4156 NA NA NA 2005 2007 -13.1912 7.8286 NA NA NA 2005 2008 -13.1176 7.8645 NA NA NA 2005 2009 -13.1176 7.8645 NA NA NA 2005 2010 -13.0294 7.8491 NA NA NA 2005 2011 -11.4118 7.9132 NA NA NA 2005 2012 -9.7941 7.5259 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0.20854 Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0 0 NA NA NA 2003 2005 0 0 NA NA NA 2003 2006 0 0 NA NA NA 2003 2007 0 0 NA NA NA 2003 2008 0 0 NA NA NA 2003 2009 0 0 NA NA NA 2003 2010 0 0 NA NA NA 2003 2011 0 0 NA NA NA 2003 2012 0 0 NA NA NA 2005 2002 0 0 NA NA NA 2005 2005 0 0 NA NA NA 2005 2006 0 0 NA NA NA 2005 2007 0 0 NA NA NA 2005 2008 0 0 NA NA NA 2005 2009 0 0 NA NA NA 2005 2010 0 0 NA NA NA 2005 2011 0 0 NA NA NA 2005 2012 0 0 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 -1.2222 0.3112 NA NA NA 2003 2005 0.0000 0.0000 NA NA NA 2003 2006 0.0000 0.0000 NA NA NA 2003 2007 0.0000 0.0000 NA NA NA 2003 2008 0.0000 0.0000 NA NA NA 2003 2009 0.0000 0.0000 NA NA NA 2003 2010 1.0000 0.0000 NA NA NA 2003 2011 1.0000 0.0000 NA NA NA 2003 2012 1.0000 0.0000 NA NA NA 2005 2002 0.7353 0.4117 NA NA NA 2005 2005 -1.3235 0.7131 NA NA NA 2005 2006 -1.8235 0.6567 NA NA NA 2005 2007 -1.8235 0.6567 NA NA NA 2005 2008 -1.8235 0.6567 NA NA NA 2005 2009 -1.8235 0.6567 NA NA NA 2005 2010 -1.4118 0.6259 NA NA NA 2005 2011 -1.1324 0.6746 NA NA NA 2005 2012 -0.9559 0.6785 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0 Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 -0.4444 0.9152 NA NA NA 2003 2005 -8.0000 0.0000 NA NA NA 2003 2006 -8.0000 0.0000 NA NA NA 2003 2007 -8.0000 0.0000 NA NA NA 2003 2008 -8.0000 0.0000 NA NA NA 2003 2009 -8.0000 0.0000 NA NA NA 2003 2010 -7.0000 0.0000 NA NA NA 2003 2011 -7.0000 0.0000 NA NA NA 2003 2012 -7.0000 0.0000 NA NA NA 2005 2002 0.4706 1.0055 NA NA NA 2005 2005 -0.2059 1.1928 NA NA NA 2005 2006 -0.7059 1.1081 NA NA NA 2005 2007 -0.7059 1.1081 NA NA NA 2005 2008 -0.7059 1.1081 NA NA NA 2005 2009 -0.7059 1.1081 NA NA NA 2005 2010 -0.2941 1.1081 NA NA NA 2005 2011 -0.0147 1.1132 NA NA NA 2005 2012 0.1618 1.1030 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0.81479 Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 -0.0121 0.1190 NA NA NA 2003 2005 -1.5256 0.0000 NA NA NA 2003 2006 -1.3224 0.0000 NA NA NA 2003 2007 -1.5418 0.0000 NA NA NA 2003 2008 -1.5545 0.0000 NA NA NA 2003 2009 -1.6230 0.0000 NA NA NA 2003 2010 -1.6568 0.0000 NA NA NA 2003 2011 -1.6627 0.0000 NA NA NA 2003 2012 -1.6531 0.0000 NA NA NA 2005 2002 0.0964 0.1385 NA NA NA 2005 2005 -1.2068 0.6497 NA NA NA 2005 2006 -1.4898 0.5703 NA NA NA 2005 2007 -1.5635 0.5715 NA NA NA 2005 2008 -1.6612 0.5617 NA NA NA 2005 2009 -1.7180 0.5728 NA NA NA 2005 2010 -1.7211 0.5842 NA NA NA 2005 2011 -1.6611 0.5796 NA NA NA 2005 2012 -1.5797 0.5658 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0.37132 Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0.0000 0.0000 NA NA NA 2003 2005 0.0000 0.0000 NA NA NA 2003 2006 2.5000 0.0000 NA NA NA 2003 2007 2.5000 0.0000 NA NA NA 2003 2008 2.5000 0.0000 NA NA NA 2003 2009 2.5000 0.0000 NA NA NA 2003 2010 2.5000 0.0000 NA NA NA 2003 2011 2.5000 0.0000 NA NA NA 2003 2012 2.5000 0.0000 NA NA NA 2005 2002 0.0000 0.0000 NA NA NA 2005 2005 3.8529 0.5797 NA NA NA 2005 2006 3.8529 0.5797 NA NA NA 2005 2007 3.8529 0.5797 NA NA NA 2005 2008 3.8529 0.5797 NA NA NA 2005 2009 3.8529 0.5797 NA NA NA 2005 2010 4.1471 0.6002 NA NA NA 2005 2011 4.5294 0.6143 NA NA NA 2005 2012 4.5882 0.6118 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 1.8889 2.4802 NA NA NA 2003 2005 350.0000 0.0000 NA NA NA 2003 2006 388.0000 0.0000 NA NA NA 2003 2007 415.0000 0.0000 NA NA NA 2003 2008 417.0000 0.0000 NA NA NA 2003 2009 428.0000 0.0000 NA NA NA 2003 2010 434.0000 0.0000 NA NA NA 2003 2011 435.0000 0.0000 NA NA NA 2003 2012 436.0000 0.0000 NA NA NA 2005 2002 -2.0000 2.6164 NA NA NA 2005 2005 57.0000 110.1486 NA NA NA 2005 2006 95.7059 113.1138 NA NA NA 2005 2007 105.1765 113.3344 NA NA NA 2005 2008 111.1765 113.6063 NA NA NA 2005 2009 114.5882 113.0625 NA NA NA 2005 2010 118.7647 113.9398 NA NA NA 2005 2011 121.1765 114.1296 NA NA NA 2005 2012 124.4118 113.7756 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0.54972 Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, control_group = “notyettreated”, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0.3889 0.6086 NA NA NA 2003 2005 -2.0000 0.0000 NA NA NA 2003 2006 -2.0000 0.0000 NA NA NA 2003 2007 -2.0000 0.0000 NA NA NA 2003 2008 -2.0000 0.0000 NA NA NA 2003 2009 0.0000 0.0000 NA NA NA 2003 2010 1.0000 0.0000 NA NA NA 2003 2011 1.0000 0.0000 NA NA NA 2003 2012 1.0000 0.0000 NA NA NA 2005 2002 -0.4118 0.6464 NA NA NA 2005 2005 -4.3529 1.4826 NA NA NA 2005 2006 -4.5294 1.4877 NA NA NA 2005 2007 -4.5882 1.4877 NA NA NA 2005 2008 -4.5882 1.4672 NA NA NA 2005 2009 -4.4118 1.4826 NA NA NA 2005 2010 -3.8235 1.5288 NA NA NA 2005 2011 -3.4118 1.5390 NA NA NA 2005 2012 -2.7647 1.6775 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
P-value for pre-test of parallel trends assumption: 0.79724 Control Group: Not Yet Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
4.2 Trying different models - adding controls
#for (i in 1:9){
# iy<- y_vars[i]
# out_4.2 <- att_gt(yname = iy,
# gname = "vat_year1",
# idname = "state_id",
# tname = "year",
# xformla = ~state_aligned_with_center,
# data = my_data_4.1,
# est_method = "reg",
# control_group = "notyettreated"
# )
# summary(out_4.2)
# es_4.2 <- aggte(out_4.2, type = "dynamic")
# Event_study<-ggdid(es_4.2)
# print(Event_study)
# Plots<-ggdid(out_4.2)
# print(Plots)
#}4.3 Changing the control group
for (i in 1:9){
iy<- y_vars[i]
out_4.3 <- att_gt(yname = iy,
gname = "vat_year1",
idname = "state_id",
tname = "year",
xformla = ~1,
data = my_data_4.1,
est_method = "reg"
)
summary(out_4.3)
es_4.3 <- aggte(out_4.3, type = "dynamic")
Event_study<-ggdid(es_4.3)
print(Event_study)
Plots<-ggdid(out_4.3)
print(Plots)
}Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0.0199 0.0000 NA NA NA 2003 2005 -1.3586 0.0000 NA NA NA 2003 2006 -1.4644 0.0000 NA NA NA 2003 2007 -1.7988 0.0000 NA NA NA 2003 2008 -1.8313 0.0000 NA NA NA 2003 2009 -1.9815 0.0000 NA NA NA 2003 2010 -1.6711 0.0000 NA NA NA 2003 2011 -1.6874 0.0000 NA NA NA 2003 2012 -1.6945 0.0000 NA NA NA 2005 2002 0.1432 0.1764 NA NA NA 2005 2005 -1.3607 0.4191 NA NA NA 2005 2006 -1.8718 0.4094 NA NA NA 2005 2007 -2.0140 0.4402 NA NA NA 2005 2008 -2.1120 0.4461 NA NA NA 2005 2009 -2.1671 0.4511 NA NA NA 2005 2010 -1.8857 0.4615 NA NA NA 2005 2011 -1.6684 0.4939 NA NA NA 2005 2012 -1.5571 0.4560 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0.0000 0.0000 NA NA NA 2003 2005 0.0000 0.0000 NA NA NA 2003 2006 0.0000 0.0000 NA NA NA 2003 2007 0.0000 0.0000 NA NA NA 2003 2008 0.0000 0.0000 NA NA NA 2003 2009 5.0000 0.0000 NA NA NA 2003 2010 5.0000 0.0000 NA NA NA 2003 2011 5.0000 0.0000 NA NA NA 2003 2012 5.0000 0.0000 NA NA NA 2005 2002 4.4118 3.0011 NA NA NA 2005 2005 -12.0294 7.4541 NA NA NA 2005 2006 -13.2059 7.3771 NA NA NA 2005 2007 -13.1912 7.4181 NA NA NA 2005 2008 -13.1176 7.4156 NA NA NA 2005 2009 -13.1176 7.4156 NA NA NA 2005 2010 -13.0294 7.4541 NA NA NA 2005 2011 -11.4118 7.4002 NA NA NA 2005 2012 -9.7941 7.5874 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0 0 NA NA NA 2003 2005 0 0 NA NA NA 2003 2006 0 0 NA NA NA 2003 2007 0 0 NA NA NA 2003 2008 0 0 NA NA NA 2003 2009 0 0 NA NA NA 2003 2010 0 0 NA NA NA 2003 2011 0 0 NA NA NA 2003 2012 0 0 NA NA NA 2005 2002 0 0 NA NA NA 2005 2005 0 0 NA NA NA 2005 2006 0 0 NA NA NA 2005 2007 0 0 NA NA NA 2005 2008 0 0 NA NA NA 2005 2009 0 0 NA NA NA 2005 2010 0 0 NA NA NA 2005 2011 0 0 NA NA NA 2005 2012 0 0 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 -1.0000 0.0000 NA NA NA 2003 2005 0.0000 0.0000 NA NA NA 2003 2006 0.0000 0.0000 NA NA NA 2003 2007 0.0000 0.0000 NA NA NA 2003 2008 0.0000 0.0000 NA NA NA 2003 2009 0.0000 0.0000 NA NA NA 2003 2010 1.0000 0.0000 NA NA NA 2003 2011 1.0000 0.0000 NA NA NA 2003 2012 1.0000 0.0000 NA NA NA 2005 2002 0.2353 0.3488 NA NA NA 2005 2005 -1.3235 0.7285 NA NA NA 2005 2006 -1.8235 0.6515 NA NA NA 2005 2007 -1.8235 0.6515 NA NA NA 2005 2008 -1.8235 0.6515 NA NA NA 2005 2009 -1.8235 0.6515 NA NA NA 2005 2010 -1.4118 0.6464 NA NA NA 2005 2011 -1.1324 0.6862 NA NA NA 2005 2012 -0.9559 0.6400 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0.0000 0.0000 NA NA NA 2003 2005 -8.0000 0.0000 NA NA NA 2003 2006 -8.0000 0.0000 NA NA NA 2003 2007 -8.0000 0.0000 NA NA NA 2003 2008 -8.0000 0.0000 NA NA NA 2003 2009 -8.0000 0.0000 NA NA NA 2003 2010 -7.0000 0.0000 NA NA NA 2003 2011 -7.0000 0.0000 NA NA NA 2003 2012 -7.0000 0.0000 NA NA NA 2005 2002 0.4706 0.9029 NA NA NA 2005 2005 -0.2059 1.2056 NA NA NA 2005 2006 -0.7059 1.1491 NA NA NA 2005 2007 -0.7059 1.1491 NA NA NA 2005 2008 -0.7059 1.1491 NA NA NA 2005 2009 -0.7059 1.1491 NA NA NA 2005 2010 -0.2941 1.1184 NA NA NA 2005 2011 -0.0147 1.1389 NA NA NA 2005 2012 0.1618 1.1338 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0.1495 0.0000 NA NA NA 2003 2005 -1.5256 0.0000 NA NA NA 2003 2006 -1.3224 0.0000 NA NA NA 2003 2007 -1.5418 0.0000 NA NA NA 2003 2008 -1.5545 0.0000 NA NA NA 2003 2009 -1.6230 0.0000 NA NA NA 2003 2010 -1.6568 0.0000 NA NA NA 2003 2011 -1.6627 0.0000 NA NA NA 2003 2012 -1.6531 0.0000 NA NA NA 2005 2002 0.1712 0.1292 NA NA NA 2005 2005 -1.2068 0.5893 NA NA NA 2005 2006 -1.4898 0.5407 NA NA NA 2005 2007 -1.5635 0.5309 NA NA NA 2005 2008 -1.6612 0.5419 NA NA NA 2005 2009 -1.7180 0.5492 NA NA NA 2005 2010 -1.7211 0.5524 NA NA NA 2005 2011 -1.6611 0.5585 NA NA NA 2005 2012 -1.5797 0.5511 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0.0000 0.0000 NA NA NA 2003 2005 0.0000 0.0000 NA NA NA 2003 2006 2.5000 0.0000 NA NA NA 2003 2007 2.5000 0.0000 NA NA NA 2003 2008 2.5000 0.0000 NA NA NA 2003 2009 2.5000 0.0000 NA NA NA 2003 2010 2.5000 0.0000 NA NA NA 2003 2011 2.5000 0.0000 NA NA NA 2003 2012 2.5000 0.0000 NA NA NA 2005 2002 0.0000 0.0000 NA NA NA 2005 2005 3.8529 0.4925 NA NA NA 2005 2006 3.8529 0.4925 NA NA NA 2005 2007 3.8529 0.4925 NA NA NA 2005 2008 3.8529 0.4925 NA NA NA 2005 2009 3.8529 0.4925 NA NA NA 2005 2010 4.1471 0.5233 NA NA NA 2005 2011 4.5294 0.5399 NA NA NA 2005 2012 4.5882 0.5476 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0.0000 0.0000 NA NA NA 2003 2005 350.0000 0.0000 NA NA NA 2003 2006 388.0000 0.0000 NA NA NA 2003 2007 415.0000 0.0000 NA NA NA 2003 2008 417.0000 0.0000 NA NA NA 2003 2009 428.0000 0.0000 NA NA NA 2003 2010 434.0000 0.0000 NA NA NA 2003 2011 435.0000 0.0000 NA NA NA 2003 2012 436.0000 0.0000 NA NA NA 2005 2002 -2.0000 2.6164 NA NA NA 2005 2005 57.0000 107.4451 NA NA NA 2005 2006 95.7059 106.7987 NA NA NA 2005 2007 105.1765 107.1116 NA NA NA 2005 2008 111.1765 106.7063 NA NA NA 2005 2009 114.5882 107.4964 NA NA NA 2005 2010 118.7647 107.7683 NA NA NA 2005 2011 121.1765 107.5220 NA NA NA 2005 2012 124.4118 107.3989 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
Call: att_gt(yname = iy, tname = “year”, idname = “state_id”, gname = “vat_year1”, xformla = ~1, data = my_data_4.1, est_method = “reg”)
Reference: Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Forthcoming at the Journal of Econometrics https://arxiv.org/abs/1803.09015, 2020.
Group-Time Average Treatment Effects: Group Time ATT(g,t) Std. Error [95% Simult. Conf. Band]
2003 2002 0.0000 0.0000 NA NA NA 2003 2005 -2.0000 0.0000 NA NA NA 2003 2006 -2.0000 0.0000 NA NA NA 2003 2007 -2.0000 0.0000 NA NA NA 2003 2008 -2.0000 0.0000 NA NA NA 2003 2009 0.0000 0.0000 NA NA NA 2003 2010 1.0000 0.0000 NA NA NA 2003 2011 1.0000 0.0000 NA NA NA 2003 2012 1.0000 0.0000 NA NA NA 2005 2002 -0.4118 0.6413 NA NA NA 2005 2005 -4.3529 1.5442 NA NA NA 2005 2006 -4.5294 1.5749 NA NA NA 2005 2007 -4.5882 1.5339 NA NA NA 2005 2008 -4.5882 1.5236 NA NA NA 2005 2009 -4.4118 1.5544 NA NA NA 2005 2010 -3.8235 1.5698 NA NA NA 2005 2011 -3.4118 1.6160 NA NA NA 2005 2012 -2.7647 1.7699 NA NA NA — Signif. codes: `*’ confidence band does not cover 0
Control Group: Never Treated, Anticipation Periods: 0 Estimation Method: Outcome Regression 
