The current official poverty statistics published by the Census Bureau (DeNavas-Walt et al., 2004) are based on money income data collected on the Annual Social and Economic Supplement (ASEC) to the Current Population Survey (CPS), as compared to an absolute poverty standard (the official poverty thresholds). Citro and Michael (1995), among others, have suggested both that the appropriate measure of resources to use in such a poverty measure is broader than money income — more of a disposable income concept that takes account of noncash benefits and work expenses (including taxes) — and that the poverty thresholds ought to be revised as well. Rector et al. (1999), among others, have suggested, based on comparisons to the National Income and Product Accounts (NIPAs), that income is underreported on the CPS ASEC. Such underreporting would suggest that the estimated poverty rate is too high.

Whether these attempts to change the way poverty is measured are informative will ultimately depend on the ability of the available data sources to measure economic well-being. This paper focuses on the quality of one of those data sources — the CPS ASEC. The examination is organized in three parts that mirror the survey process — questionnaire design, data collection and preparation (including edits and imputation), and post-collection data processing (to enhance the dataset). Finally, the paper proposes a set of research projects that could remedy many of the deficiencies identified.