We study the short-run and long-run economic impact of one of the largest losses that an individual can face; the death of a child. We utilize unique merged registers on the entire Swedish population, combining information on the date and cause of death with parents' labour market outcomes, health...

The matching method for treatment evaluation does not balance selective unobserved differences between treated and non-treated. We derive a simple correction term if there is an instrument that shifts the treatment probability to zero in specific cases. Policies with eligibility restrictions,...