Unhiding the bodies, Part II

Three years ago I covered two examples of repressive regimes hiding the official extent of deaths, but where enterprising researchers deduced the likely truth from other administrative data.  One of those examples was Russia's combat deaths in its war on Ukraine.  While Russian casualties are a state secret, the researchers extrapolated an estimate from a surge in civil probate cases among males of military age.  As of July 2023 the researchers estimated Russian combat dead at "around 47,000".

The same researchers have updated their estimate of Russian combat losses to the end of 2025.  Now their estimate has risen to "352,000 deaths in four years".  As before, the combat losses are extrapolated from probate records. However, a new element is probate cases where the court is asked to declare someone as dead when they have been missing for at least 180 days. This is a familiar concept to actuaries working in more peaceful fields, namely occurred-but-not-yet-reported (OBNR) deaths (Lawless, 1994); I discussed the 'nowcasting' of mortality in the presence of delays in an earlier blog.  Another aspect is where the same individual is the subject of more than one court process, which was also allowed for by the researchers.  Again, actuaries have long been familiar with the need to handle duplicate records.

One question is how accurate this estimation is, and the researchers set out their methodology in quite some detail.  However, the researchers also provided a lower bound for their estimate by collating and verifying the names of deceased soldiers from other, non-administrative sources.  This produced a minimum of around 220,000 dead.

The thoroughness and transparency of the researchers stands in stark contrast to the secrecy of the Russian state.  Whom do you trust?  A state apparatus that says you aren't allowed to know?  Or researchers who publish their sources and open their methodology to scrutiny?