Do subsidized new firms survive longer? Evidence from a counterfactual approach