This paper develops a measure or rural household vulnerability, based on the risk of
consumption falling below a poverty line, which combines existing approaches to
estimating idiosyncratic risks with an approach to measuring covariate risk arising
from crop production. The methodology is applied to rural households in Tanzania,
using household surveys in a relatively wealthy and a relatively poorer region. The
results suggest that covariate risk faced by rural households that arises from crop
production is substantial, but quite different in the two regions. Consumption is
estimated to depend significantly on crop income, and the variability of that induces
considerable overall consumption risk. Covariate risk is found to constitute smaller
shares of total consumption risk in the wealthier region. The share of covariate risk in
total household consumption risk is found to be similar among poor and non-poor.
Vulnerability is quite different in the two regions, with rural households in the poorer
region exhibiting considerably higher vulnerability. It is found that non-poor
households are quite vulnerable and that there is a considerable degree of hard core
poverty. It is demonstrated that observable indicators can be identified to discriminate
vulnerable from non-vulnerable households