In contrast to earlier weeks, the most recent “excess fatality” count is solidly in the positive region, despite the severe under-reporting bias in the most recent observations. To see this, consider the most recent estimates for each of the previous vintages of “excess fatalities” calculated as actual-expected.
Figure 1: Excess fatalities, 10/7 vintage (purple red), 9/30 vintage (violet), 9/23 vintage (chartreuse), 9/16 vintage (red), 9/9 vintage (green), 9/2 vintage (orange), 8/25 vintage (blue). Note excess fatalities differ from CDC series which are bounded below at zero. Source: CDC , various vintages, and author’s calculations.
This pattern suggests to me we should take with circumspection (1) the most recent counts of excess fatalities as they are likely to be revised substantially upward; and (2) administrative counts, either from CDC or from alternative compilations, as they are possibly missing many actual Covid-19 related deaths. Extending point (1), it is unclear to me whether excess fatalities are indeed falling, as they seem to have been pretty constant from the week ending 7/25 through week ending 8/15. (The Economist has a discussion of excess fatalities around the world; the Excess Death Tracker is here.)
Here are the various series of interest, latest available.
Figure 2: Weekly fatalities due to Covid-19 as reported to CDC for weeks ending on indicated dates (black), excess fatalities calculated as actual minus expected (teal), fatalities as tabulated by Our World in Data (dark red). Note excess fatalities differ from CDC series which are bounded below at zero. Light green shading denotes CDC data that are likely to be revised. Source: CDC 10/7/2020 vintage, OurWorldinData version of 10/8 accessed 10/9/2020 and author’s calculations.
One conclusion that seems obvious: Cumulative excess fatalities through week ending 8/25 are substantially higher than administratively defined Covid-19 cumulative fatalities.