Python code for Aggregate Group-Time Average Treatment Effects

This code implements the Aggregate Group-Time Average Treatment Effects from the DiD with multiple time periods methodology of Callaway and Sant’Anna (2021).

Description

I built the Python function for aggte function of the R package (and c++) from Callaway and Sant’Anna (2021). This code is a function to take group-time average treatment effects and aggregate them into a smaller number of parameters. There are several possible aggregations including "simple", "dynamic", "group", and "calendar.".

Roadmap

• Aggregate Group-Time Average Treatment Effects function (aggte)
• Child function (compute_aggte)
• Multipler bootstrapping (mboot)
  • Clean and call for bootstrapping (mboot)
  • Call for parallel computing or single computing for multiplier_bootstrap (run_multiplier_bootstrap)
• Bmisc
  • Execute bootstraiping (multiplier_bootstrap)
  • Clean arrays for logical vectors (TorF)
• Utils
  • Compute influence function matrix(wif)
  • Apply influence function for parameters (get_agg_inf_func)
  • Return standard errors (get_se)
  • Print results (AGGTEobj)

Getting Started

Executing program

• Download all the folder
• Paste the path in folder_path
• Select the type of aggregation
  • out = aggte(out,'simple')
  • out = aggte(out,'group')
  • out = aggte(out,'dynamic')
  • out = aggte(out,'calendar')
• Run the following script "tryal_runing_att_gt_object" that contains the following code:

import os
import pickle

folder_path = r"" 
os.chdir(folder_path)

from aggte import *

pickle_file = "att_gt_object.pkl"
with open(pickle_file, "rb") as f:
    out = pickle.load(f)
    
out =  aggte(out)

Outcome

Group aggregation (deafult)

Overall summary of ATT's based on group/cohort aggregation:
   ATT Std. Error  [95.0%  Conf. Int.]
-0.031     0.0124 -0.0552      -0.0068 *

Group Effects:
   Group  Estimate  Std. Error   [95.0% Simult.   Conf. Band]
0   2004   -0.0797      0.0291      -0.1437     -0.0158  *
1   2006   -0.0229      0.0171      -0.0605      0.0147
2   2007   -0.0261      0.0165      -0.0624      0.0102
---
Signif. codes: `*' confidence band does not cover 0
Control Group:  Never Treated ,
Anticipation Periods:  0.0
Estimation Method:  Outcome Regression

Simple aggregation

  ATT Std. Error  [95.0%  Conf. Int.]  
-0.04     0.0122 -0.0638      -0.0161 *

---
Signif. codes: `*' confidence band does not cover 0
Control Group:  Never Treated , 
Anticipation Periods:  0.0
Estimation Method:  Outcome Regression

Dynamic aggregation

Overall summary of ATT's based on event-study/dynamic aggregation:
    ATT Std. Error  [95.0%  Conf. Int.]  
-0.0772     0.0204 -0.1173      -0.0372 *

Dynamic Effects:
   Event time  Estimate  Std. Error  [95.0% Simult.   Conf. Band]   
0          -3    0.0305      0.0152     -0.0080      0.0690   
1          -2   -0.0006      0.0128     -0.0329      0.0318   
2          -1   -0.0245      0.0141     -0.0600      0.0111   
3           0   -0.0199      0.0127     -0.0519      0.0120   
4           1   -0.0510      0.0167     -0.0931     -0.0088  *
5           2   -0.1373      0.0393     -0.2364     -0.0381  *
6           3   -0.1008      0.0339     -0.1864     -0.0152  *
---
Signif. codes: `*' confidence band does not cover 0
Control Group:  Never Treated , 
Anticipation Periods:  0.0
Estimation Method:  Outcome Regression

References

Callaway, Brantly and Pedro H.C. Sant’Anna. “Difference-in-Differences with Multiple Time Periods.” Journal of Econometrics, Vol. 225, No. 2, pp. 200-230, 2021.