> There's pretty good evidence for logarithmic marginal utility of money
1) This is, at best, controversial. Measuring cardinal utility is an unsolved, perhaps unsolvable problem. The kinds of experiments that lead to these claims are things like "what more would you buy if you had $X" or "how much, out of 10, would your happiness increase if you had $X more". Needless to say these are full of problems with self reporting, perception, lack of skin in the game, etc.
The only utility-adjacent concept that can be tested rigorously is preference ordering in situations where the subject actually bears the costs and benefits of the choice. This makes most experiments you'd want to do very costly, and potentially unethical.
2) Even if we accept that everyone has a logarithmic curve describing their own utility for money, this doesn't mean you can make intersubjective comparisons. To give an analogy, it's a "units" problem. Bob's utility is measured in Bobutils, and Alice's utility is measured in Aliceutils. Even if Bobutils(dollars) and Aliceutils(dollars) are both logarithmic, this doesn't get you any closer to aggregating their utility or evaluating tradeoffs (can't add/subtract things with different units).
Edit: playing minor devil's advocate to myself on point 1, it's certainly true that there's good evidence of diminishing marginal utility of money (or of anything). I was specifically responding to the kinds of experiments that lead to the claim of a logarithmic form, which I consider to have many problems.
1) This is, at best, controversial. Measuring cardinal utility is an unsolved, perhaps unsolvable problem. The kinds of experiments that lead to these claims are things like "what more would you buy if you had $X" or "how much, out of 10, would your happiness increase if you had $X more". Needless to say these are full of problems with self reporting, perception, lack of skin in the game, etc.
The only utility-adjacent concept that can be tested rigorously is preference ordering in situations where the subject actually bears the costs and benefits of the choice. This makes most experiments you'd want to do very costly, and potentially unethical.
2) Even if we accept that everyone has a logarithmic curve describing their own utility for money, this doesn't mean you can make intersubjective comparisons. To give an analogy, it's a "units" problem. Bob's utility is measured in Bobutils, and Alice's utility is measured in Aliceutils. Even if Bobutils(dollars) and Aliceutils(dollars) are both logarithmic, this doesn't get you any closer to aggregating their utility or evaluating tradeoffs (can't add/subtract things with different units).
Edit: playing minor devil's advocate to myself on point 1, it's certainly true that there's good evidence of diminishing marginal utility of money (or of anything). I was specifically responding to the kinds of experiments that lead to the claim of a logarithmic form, which I consider to have many problems.