I am trying to obtain signficance levels for a synthetic control method (SCM) result that I had gotten before. To do so, I am hoping to use difference-in-difference by using the observed unit as the treated and the synthetic unit from SCM as the control. I have no covariates involved in my regression.
Below is my dataset where _Y_treated is treated (observed) and _Y_synthetic the control from SCM. The treated dummy variable is to show that _Y_treated is treated and similarly, time is a dummy to signify treatment period. DiD is the interaction term. The code is below the data. Although I have used the diff package below, I have previously tried some different approaches with regression method and the hashtag method too.
The cause of the collinearity is glaringly obvious however, I do not know how to overcome it to get the desired results. I have a feeling my query is not particularly difficult but for some really silly reason I cannot seem to solve my issue.
Any help is massively appreciated!
Code:
* Example generated by -dataex-. For more info, type help dataex clear input int year double(_Y_treated _Y_synthetic) float(treated time DiD) 1990 13075.236033310537 13064.229969791773 1 0 0 1991 11865.82340535107 12809.582282848382 1 0 0 1992 11948.485091255892 12200.496798981847 1 0 0 1993 12666.07765362244 12294.224792439196 1 0 0 1994 13328.105203227276 12723.206068804231 1 0 0 1995 13468.770647527726 13291.301199237227 1 0 0 1996 13723.506351783068 13417.546589916237 1 0 0 1997 13868.65473862296 14008.040285347932 1 0 0 1998 14207.79715012313 14494.286904203862 1 0 0 1999 14546.888269048695 14930.615285279106 1 0 0 2000 15599.383609345408 15959.718825180997 1 0 0 2001 16598.865359506875 16418.796944581212 1 0 0 2002 17603.698180723404 17152.63916879498 1 0 0 2003 18257.585818030788 17757.063504177408 1 0 0 2004 18209.756581653233 18481.478522490313 1 0 0 2005 19494.49515696955 19472.45801531873 1 0 0 2006 20235.30166315405 20230.225567799378 1 0 0 2007 20563.460050324473 21239.95197525975 1 0 0 2008 21609.551610825085 21900.13207177752 1 0 0 2009 21736.021288448075 21787.424136898848 1 0 0 2010 22831.748867021175 22602.803870210126 1 0 0 2011 23099.73739803632 22567.855970629993 1 0 0 2012 22811.263414202935 23099.542375920508 1 1 1 2013 23116.146653782016 23098.763351945272 1 1 1 2014 23701.170722645234 23598.176443746775 1 1 1 2015 25009.652164882053 25743.755791811032 1 1 1 2016 25443.407972951845 27046.947356093417 1 1 1 2017 26908.965224752807 28837.484831334223 1 1 1 2018 27928.041764413265 29354.435507782386 1 1 1 2019 29385.537713743535 30739.800278648032 1 1 1 end
Code:
gen time = (year>=2012) & !missing(year) . gen treated = (_Y_treated) & !missing(_Y_treated) . gen DiD = treated*time . diff _Y_treated, t(treated) p(time) DIFFERENCE-IN-DIFFERENCES ESTIMATION RESULTS Number of observations in the DIFF-IN-DIFF: 30 Before After Control: 0 0 0 Treated: 22 8 30 22 8 -------------------------------------------------------- Outcome var. | _Y_tr~d | S. Err. | |t| | P>|t| ----------------+---------+---------+---------+--------- Before | | | | Control | 1.7e+04| | | Treated | 1.7e+04| | | Diff (T-C) | 0.000 | . | . | . After | | | | Control | 2.6e+04| | | Treated | 2.6e+04| | | Diff (T-C) | 0.000 | . | . | . | | | | Diff-in-Diff | 0.000 | . | . | . -------------------------------------------------------- R-square: 0.58 * Means and Standard Errors are estimated by linear regression **Inference: *** p<0.01; ** p<0.05; * p<0.1
0 Response to Collinearity issue for difference-in-difference
Post a Comment