Dear all,

I am working with a cross-sectional data set for the years 2005 to 2015. I am trying to examine whether a change in tax legislation (loss carryforward rules) had an influence on investments in tangible fixed assets. Therefore, my first basic regression looks like this:
Code:
reg lnTFAS lnlcf lnSales lnAge lnEmployees i.year#c.lnlcf i.year, vce(robust) coeflegend noheader
The new tax legislation was introduced in 2009, therefore my implementation year is 2009. I have read in the literature that one should omit the interaction of the implementation year dummy and normalize it to zero in order to avoid perfect collinearity. I am struggling with normalizing it to zero in order to interpret all the other interaction coefficients relative to 2009.
After running the regression, I want to plot the point estimates of the interaction dummies using the coefplot command like this:
Code:
coefplot, keep(2006.year#c.lnlcf 2007.year#c.lnlcf 2008.year#c.lnlcf 2009.year#c.lnlcf 2010.year#c.lnlcf 2011.year#c.lnlcf 2012.year#c.lnlcf 2013.year#c.lnlcf 2014.year#c.lnlcf 2015.year#c.lnlcf) /*
*/ vertical yline(0) rename(2006.year#c.lnlcf = 2006 2007.year#c.lnlcf = 2007 2008.year#c.lnlcf = 2008 2009.year#c.lnlcf = 2009 2010.year#c.lnlcf = 2010 2011.year#c.lnlcf = 2011 /*
*/ 2012.year#c.lnlcf = 2012 2013.year#c.lnlcf = 2013 2014.year#c.lnlcf = 2014 2015.year#c.lnlcf = 2015) xtitle("Interaction LCF and years") ciopts(recast(rcap))
Do you have any suggestions or ideas how to normalize the 2009 interaction coefficient to zero so that all the other interaction coefficients can be interpreted relative to the implementation of the tax legislation in 2009?

Thanks a lot in advance for your time and your help!
Chris