Direct measures of poverty and well-being: A theoretical framework and an application to housing poverty in the United States

Economists have traditionally used income as their exclusive informational basis for the construction of poverty indices. This has unduly limited our understanding of poverty and constrained our policy prescriptions. In this dissertation, I use direct measures of well-being to broaden our understanding of poverty and expand our policy prescriptions.; In any analysis of poverty the first step is identifying the profile of poor persons. Here I solve the identification problem in a multi-dimensional framework with a three-step procedure. First, axioms, chosen to reflect underlying beliefs about well-being, are used to construct a Well-Being Evaluation Function (WBEF). Second, a poverty line is set for each indicator of well-being yielding a poverty-line vector. Persons are identified as poor if and only if their well-being is less than the value of the well-being associated with the poverty-line vector, where both are defined by the WBEF. Third, a distance function defines a poor person's degree of impoverishment.; I employ this theoretical structure to study housing poverty in the United States from 1985 to 1993 using data from the American Housing Survey, a bi-annual survey of U.S. housing conditions. I use three direct indicators--a housing unit's adequacy, comfort, and neighborhood--to measure housing quality. The study uncovers three important aspects of poverty obscured by income-based measures. These are (1) higher housing-poverty rates for Latino primary households than for African-American primary households despite lower income-poverty rates; (2) substantially higher housing-poverty rates for income-poor Latino and African-American primary households than for income-poor white primary households; and (3) lower housing-poverty rates for Senior primary households than the entire population of primary households despite higher income-poverty rates.; I use a logit model (with the dependent variable of housing poor/not housing poor) to explore other explanations besides current income for these differences. In this model independent variables include a broad array of income variables for the primary household, the presence of non-primary household members, a proxy for average housing prices in an area, and demographic characteristics of the primary household. Using this model, I analyze the entire population of primary households and income-poor primary households. Among other findings, I find the insignificance of current income in explaining the housing-poverty of income-poor primary households. With an interactive logit model, I find different characteristics, rather than differing returns to characteristics generally explain differences between groups in housing-poverty rates. I conclude with avenues for future research and a wide array of policy implications.