We use Panel Study of Income Dynamics (PSID) data to compare chronic poverty under the Official Poverty Measure (OPM) and under the new Supplemental Poverty Measure (SPM) as predictors of child outcomes. In particular, we consider 13 outcomes for 1,140 10- to 19-year-olds from the PSID’s 2007 Child Development Supplement, including 4 cognitive development outcomes, 4 physical development outcomes, and 5 social-emotional development outcomes. We compare the OPM and the SPM using five different chronic poverty measures estimated from 1998 to 2006, including a simple sum of the annual normalized poverty gaps over that period and more complex measures developed by Foster (2009); Hoy and Zheng (2011); Bossert, Chakravarty, and D’Ambrosio (2011); and Mendola and Busetta (2012). We rely primarily on J-tests and Cox-Pesaran-Deaton tests for non-nested hypotheses to compare the corresponding OPM and SPM measures, but we also consider R-squared values and Akaike and Bayesian information criteria.

We find that the OPM chronic poverty measures perform at least as well as the corresponding SPM measures in predicting these child outcomes. In our main analyses with demographic controls, the OPM measures are better predictors of 6 outcomes, the SPM measures are better predictors of 5 outcomes, and the two sets of measures are indistinguishable with respect to 2 outcomes. We consider possible reasons for the SPM’s relative underperformance in predicting these outcomes, by focusing on the key differences between the OPM and SPM and conducting supplemental analyses that adjust our SPM calculation to eliminate those differences one at a time. None of the individual adjustments to the SPM that we consider improve the SPM’s performance in predicting these outcomes. However, when we adjust our SPM calculation to both eliminate the deduction of medical out-of-pocket expenses and increase the threshold multiplier, the SPM’s performance improves noticeably, as the SPM measures are better predictors of 6 outcomes, while the OPM measures are better predictors of only 2 outcomes and the two sets of measures are indistinguishable with respect to 5 outcomes. Our results suggest that additional research is needed on the effect of the SPM’s medical out-of-pocket expense deduction on the SPM’s performance as a poverty measure, and that consideration should be given to eliminating or refining that deduction, while simultaneously increasing the SPM thresholds, at least for families with children.