The BLS's issues with CPI data collection appear to be getting worse. The share of uncollected items subject to "different cell" imputation (meaning they're based on different items or items from different geographies) hit another record last month. (Data goes back to 2019.)

Frustratingly, BLS provides data on the share of uncollected items that are subject to different cell vs home cell imputation, but not on the share of total items that are uncollected. So hard to gauge just how big a deal this is. But certainly not encouraging.