Insurance increases net costs via added overhead. So, it’s ~80k * ~1.30 * (number receiving treatment) / (people paying insurance). Aka, it’s both more expensive and more affordable.

Health insurance in the US has overhead capped at a maximum of 20%, or 15% in large employer groups.

https://www.healthcare.gov/health-care-law-protections/rate-...

Spending 80% of your premium on direct heathcare cost means ( 1 / .8 - 1) = 25% cost increase, but that’s on top of the heathcare providers insurance related overhead. Insurance only adding 30% to total heathcare costs is a reasonable ballpark though likely an underestimation.

What percentage of the recepients costs are insurance related overhead as well? (Yes, that includes the time doctors spend arguing with insurers).

Surely, there is room for optimization, but in the alternative case of taxpayer funded healthcare, I'm sure the government is going to ask doctors for justifications for the same reasons insurers do. Someone needs to provide a second opinion on whether or not procedures and medications are appropriate for accountability purposes.

Fraud in government programs have much harsher criminal penalties and are more aggressively pursued.

https://www.propublica.org/article/health-insurers-make-it-e...

https://www.propublica.org/article/we-asked-prosecutors-if-h...

https://www.propublica.org/article/how-to-make-health-insure...

Governments add far less overhead in this area because they have more data and can thus use random inspection of bills for validation. Further, when setting rates every insurer has a separate negotiation involving even more overhead.

The government hires health insurance companies to do the inspections. I don't know of any US government that doesn't contract implementation of Medicaid/Medicare to UHG/Anthem/Humana/etc. So all the insurers have all the data anyway, and they are not inspecting each and every bill, they are also using heuristics.

Using heuristics on smaller sample sizes is inherently less efficient. You can get reasonable polling data on 50 million voters using 10,000 samples. That’s 1 sample per 5,000 voters. Drop the election down to a local election with 10,000 voters and the same ratio provides 2 samples telling you almost nothing.

Thus, if 5 insurance companies send patients to the same doctor you simply have more overhead.