Birth/Death employment estimates

I think it is well established by now that I am not a fan of the birth/death employment estimates in the establishment survey. That is where the BLS estimates the number of new jobs created by the birth or death of new businesses. It is often a significant portion of the jobs survey and it is a seasonally adjusted guess. There really is no alternative but to make this estimate, but at the beginnings of recessions it always overestimates the number of jobs, and at the beginnings of recoveries it will underestimate them.