This is great. If you update or do more like this in the future it would be great to have 2 additional things: n counts, and variance/sd. I might worry that for

This is great. If you update or do more like this in the future it would be great to have 2 additional things: n counts, and variance/sd. I might worry that for example the high water mark in 2020-2021 had a much smaller n count and if the stats were more people weighted the decline would be less meaningful. In other words if only one tenant rented at the peak year in some very outlier luxury unit, and then next year 1000 tenants came along and rented more typical lower priced units, it would look like a decline. But it’s not a real decline, if that makes sense? I also wonder if (strongly suspect) the temporary eviction ban could be considered a temporary price increase that has now been removed, so the decline you note is not a real decline for that reason, but a reversion of public policy that directly impacted prices for a year or two then vanished. Thank you for doing this!