> Are the punishments for crimes committed by police officers scaled based on the likelihood of punishment?
No, but they should be.
> I would not pay 30% of my salary for statins.
Do you pay 30% of your salary for policing?
> The question is not what percentage of an individual's income goes to police, but whether, from the perspective of the city, the overall allocation is an effective use of public funds.
That's an.. ahem.. interesting way to look at it. So if the city wastes twice as much money elsewhere, policing becomes a better deal?
Or if the city gets 30 pennies, and uses 33% of it to stop a fraction of crime, that's a bad deal?
It might be better to look at the services that residents receive in exchange for the amount they pay in taxes, instead of using some ridiculous metric based on fraction of one political entity's tax income.
Otherwise we come up with ridiculous silly comparisons. The sheriff is a tiny proportion of my county's expenditures, and police are a big proportion of my city's expenditures... by your metric, the sheriff is more cost effective, despite costing nearly the exact same amount per capita for the people they're first responders for. (Because the county gets a much bigger share of tax income and spends a bunch on other services that my city doesn't have to pay for).
Or if we're going with the health analogy-- we spend the overwhelming majority of our public health "income" on things with very high NNTs.