Comparing the Performance of Monthly Poverty Measures
Official poverty estimates for the United States are presented annually and are reported with a considerable lag. The COVID-19 pandemic highlighted the problematic nature of this lag, prompting researchers to develop methods to provide more timely estimates of poverty that could be used to understand economic need, and the impact of the government’s response, in closer to real time. One monthly poverty measure, introduced in Parolin, Curran, Matsudaira, Waldfogel, and Wimer (2022; hereafter PCMWW), provides monthly estimates of poverty based on a family unit’s likely monthly income. A separate measure, introduced in Han, Meyer, and Sullivan (2020; hereafter HMS), produces monthly updates of annual poverty estimates based on a survey question asking individuals about their families’ total cash income over the past 12 months.
Trends in the two measures do not always align, prompting two questions: what explains the divergent trends, and which measure more reliably represents monthly variation in the well-being of households in the U.S.? This paper investigates these questions. We discuss differences between the PCMWW and HMS measures, and we compare the monthly estimates from the two measures to external benchmarks such as monthly levels of food insufficiency, mental health, and employment outcomes. Our analysis produces three primary findings.
First, the PCMWW monthly poverty estimates closely align with monthly levels of eight different material hardship and mental health measures. Monthly estimates from the HMS annual measure, in contrast, are negatively associated with all eight measures of material hardship and mental health challenges. That is, when the PCMWW monthly poverty estimates are lower, so too are levels of material hardship and mental health challenges. When poverty is lower under the HMS measure, by contrast, rates of hardship and mental health challenges are higher. This suggests that monthly movement in the HMS annual measure misrepresents changes in well-being that a poverty measure is meant to track. The HMS measure particularly misrepresents monthly variation in living conditions for families with children and Black individuals of all ages. Tracking monthly employment rates offers a more reliable tool than the HMS estimates for understanding monthly variation in material hardship and mental health. In contrast, the PCMWW estimates generally outperform monthly employment rates in explaining trends in hardship and mental health.
Second, we demonstrate empirically that the differential performance of the two measures is due to the PCMWW’s inclusion of (and HMS exclusion of) benefits from SNAP, the Earned Income Tax Credit, and the Child Tax Credit, three of the largest anti-poverty programs in the U.S. We show that a modified HMS measure that includes these benefits strongly outperforms the original HMS measure in aligning with trends in hardship and well-being.
Third, we identify shared shortcomings in both the PCMWW and HMS measures and recommend several practical steps to improve the measurement of sub-annual poverty estimates moving forward.