Replies: 7 comments
-
@nikhilwoodruff and I looked into this and it seems to result from benefit data, especially SNAP which the ASEC understates: in 2022, the ASEC totals $45 billion, we projected $184 billion, and actuals were $114 billion. Our 2023 projection is closer: $118 billion vs. $115 billion forecasted. Beyond survey data inaccuracy, we're over because we don't model asset limits or nonparticipation. The ASEC is similarly low on SSI: $49 billion vs $70 billion, to which we calibrate. These gaps seem sufficiently large to explain the gap. On balance our numbers are probably closer to the truth than the ASEC, but we still have room to improve via enhancements already on our roadmap. |
Beta Was this translation helpful? Give feedback.
-
Likely due to #3686 |
Beta Was this translation helpful? Give feedback.
-
After merging #3686, which added broadband subsidies, medical expenses, and capped childcare and work expenses to the baseline, and which inflation-indexed some poverty-relevant values including the threshold, the baseline poverty rates rose a bit (notebook).
|
Beta Was this translation helpful? Give feedback.
-
Moving @nikhilwoodruff's comment from #4146, with regard to the ECPS: Unclear why the child poverty rate is 4.7% in 2022. The values between 2022 and 23 are not substantially different, the targets are mostly similar (no deviations of more than 15%), but the share of weight from PUF records is ~10% in 23 and 24 but ~30% in 22. |
Beta Was this translation helpful? Give feedback.
-
Couple ideas:
|
Beta Was this translation helpful? Give feedback.
-
Agree we should get around all of these but they seem very unlikely to reveal the root cause (if there is an intuitive cause) since they are also present in 2023+. |
Beta Was this translation helpful? Give feedback.
-
Numbers from the latest ECPS, which adds filing status calibration:
|
Beta Was this translation helpful? Give feedback.
-
Our poverty rates in 2022, 2023, and 2024 are broadly about a third lower than the actual 2022 SPM:
We should be calculating taxes and benefits in a generally similar way. We may estimate higher amounts for programs like SSI and WIC that we calibrate to match administrative aggregates, but I don't think this would cause such a large gap.
Beta Was this translation helpful? Give feedback.
All reactions