It is also important to mention that the health shock is based on the respondent's health condition
in the past four weeks. Outcome variables related to income and certain variables for household
investment are based on values over the past year. For these outcome variables, we would ideally
also like to observe health shocks from the past year, but this is not possible without additional
data. The assumption underlying the regressions when outcome variables are measured annually is
that the percentage of days wi th severe illness in the past four weeks is a proxy for the percentage
of days with severe illness over the past year.