This paper analyzes the interplay between early-life conditions and marital status, as determinants of adult mortality. We use individual data from Dutch registers (years 1815-2000), combined with business cycle conditions in childhood as indicators of earlylife conditions. The empirical...

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,...