(3) The data quality is so bad that there is no way to reliably know if a case is fraudulent or not without an investigation so expensive the ROI is negative, so everything remains unclear and unresolved forever.
That's usually how it goes with cases like these. Sometimes there are gangs who organize large scale fraud against the system and those might attract the attention of prosecutors, but people not reporting a death or double payment or similar isn't worth it. There might never be a clear answer to how the database got into this state. But the basic point stands that data quality for a critical dataset is really low in obvious ways, so what about all the non-obvious ways?
I mentioned blue zones because there are a lot of people on this thread who are really having a hard time believing there can be problems as obvious as people who have died but continue receiving payments in social security schemes. Blue zones is just an easily searchable keyword to learn more about other times when it's happened at scale, e.g.
In 2010, the Japanese government announced that 82 percent of its citizens reported to be over 100 had already died.
In 2012, Greece announced that it had discovered that 72 percent of its centenarians claiming pensions – some 9,000 people – were already dead.
Puerto Rico’s government said in 2010 that it would replace all existing birth certificates due to concerns about widespread fraud and identity theft.
https://www.aljazeera.com/news/2024/9/26/the-secret-of-blue-...
Obviously nobody is claiming the database reflects reality. Governments often have multiple data sources that are badly out of alignment. A census can give more accurate data, but that doesn't mean SS is synced to it. For instance, in the UK during COVID, more people came forward in some age ranges for a COVID vaccine than theoretically existed in the country at all. The UK's population data is so badly screwed up that people started using the quantity of NHS numbers issued instead to try and estimate it.