The key ndings are as follows. Prior to the reform, the eects of a negative health shock
are associated with a dierential decrease in earnings among household members suering from
the shock. Total household income (even after excluding transfer income) and household food consumption are fully insured against the health shock. Taken together, this evidence suggests
that household labor supply is an important insurance mechanism against health shocks. Given
that households were perfectly insured without access to health insurance, one might be tempted
to conclude that the welfare value of additional health insurance is small. However, the second
set of ndings reveals that the benet of health insurance could come from reducing the use of
certain costly smoothing mechanisms. During a negative health shock when there is no access
to health insurance, households reduce investment in children's education and agriculture-related
activities, and increase the use of child labor. Access to health insurance eliminates the use of some
of these costly smoothing mechanisms. Consequently, following a health shock, households with
access to health insurance invest more in children's human capital and reduce the use of child labor,
relative to the levels that they would have been in the absence of the reform. Given that access to
health insurance lessens the nancial burden by reducing out-of-pocket medical expenditure and
the productivity loss following a health shock, the availability of health insurance mitigates some
of the most costly choices households used to make following a health shock.