We describe a measure of welfare, "vulnerability", which measures the difference between the highest feasible average level of utility in a population given aggregate resources, and the actual average level of utility. This measure can be decomposed into two components, related to inequality and to risk. We provide methods for computing vulnerability, inequality, and risk using only data on expenditures from repeated cross-sections of household data, and relate these to Atkinson's family of inequality measures. Using methods developed here and household-level Ecuadorean data from 1995 and 2006, we estimate the vulnerability and risk of different population groups. Taking the population altogether, we find that the crisis of the late nineties was not only a large shock for the country as a whole, but also greatly increased the risk faced by individual households in the Sierra, risk which was subsequently translated into greater inequality. After 1999, overall risk borne by the average household fell dramatically, with the consequence that inequality remained nearly constant from 1999-2006. Levels of rural risk are considerably greater than are urban; further, rural risks tend to be the consequence of spatial shocks, while urban risks are much more idiosyncratic in nature.