(Short et al., 1999) indicated that the overall poverty rate using the actual expense approach in 1997 was 15.4 percent, or about 0.5 percentage points lower than when using the expected expense approach (15.9 percent). As expected, the difference produced by alternative methods is larger for the poverty rates of groups who are more likely to incur such expenses. For example, among full-time working families with children, the poverty rate using the actual expense approach was about 1.2 percentage points lower than the rate subtracting expected expenses (Iceland, 2000: Detailed Table 1).

Although many workshop participants supported the idea that work-related expenses should be taken into account in a new poverty measure, Douglas Besharov (American Enterprise Institute) expressed concern about the quality of data on child care expenses in the SIPP. Because of these data quality concerns, he favored taking the simpler approach to accounting for expenses—subtracting a flat amount based on a few characteristics (or expected expenses)—rather using a measure that claims to accurately capture actual expenditures. Diana Pearce (University of Washington) and others favored the expected expense approach for conceptual reasons, viewing these expenses as a basic expenditure, or need, for working families. It appeared that many of the workshop participants supported this approach—incorporating expected work-related expenses rather than estimating actual expenses in a new poverty measure. Rebecca Blank (University of Michigan) offered her summary of the discussion: “… [T]here was a strong belief [that] we should indeed account for child care work expenses in assessing poverty, and the idea was rather than trying to distribute that to individuals, that we should focus on more aggregate calculations where we assign fixed amounts to specific groups.”