“code1_data cleaning” - construct the difference-in-difference samples from the initial datasets including company fundamentals and ESG ratings of the incumbents. Three samples are constructed: one for the rating disagreement test, one for the news predictability test, and one for the rating metrics test. The corresponding entropy balanced samples are also constructed.

"code2_main DiD" - after running the first part, this part conducts the basic Difference-in-Difference analyses, where only the initial coverage by Sustainalytics is used as the shock.

"code3_additional" - this part covers all the other tests, including staggered and stacked DID, cross-sectional tests, and other additional tests.