I am examining whether a policy resulted in increased profitabiltiy. The new policy was adopted in 2005 but announced in 2002. Is there a way to mitigate the no anticipation assumption? Is matching a correct approach?
Many thanks!
Asked
Active
Viewed 146 times
0
-
Does it make sense to perform a robustness test where we include leads to control for the anticipatory effect? – Henrik Dalriksson May 11 '17 at 15:26
-
Hi, I think using the difference in difference approach once using 2002 and once using 2005 as your treatment year will help ot give you a better idea. Moreover, https://stats.stackexchange.com/questions/160359/difference-in-difference-method-how-to-test-for-assumption-of-common-trend-betw as this post shows you, you can test whether pretrends held in 2002 or not. If not then I presume you are better of using 2002 as your treatment year. – karsha Sep 13 '17 at 07:12