I'm trying to use the group-time average treatment effect estimator proposed by Callaway & Sant'Anna (Journal of Econometrics, 2021) to do event study analysis. Here's the problem I've encountered:
We have data only in years 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2014 2016 2018: that is, before 2012, the survey occurs every year; and after 2012, the survey only occurs every 2 years.
However, we have treatment in 0 (never treated) 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Thus, when we aggregate the treatment effects, the estimates jump up and down, like this (this figure is generated by did package):
Does anyone have any suggestions regarding this matter? What should I do?
