The biggest issue I see with Pigouvian taxes is that they’re computationally tricky to estimate. Who knows what downstream effect in the big chain of causality this particular person/action had! Pollution/carbon taxes are an easy exception.
Aside, a bit off topic: Even if we could compute Shapley values, Shapley values suffer from combinatorial explosion. (Things other than Shapley values or approximating them might work.)
The biggest issue I see with Pigouvian taxes is that they’re computationally tricky to estimate. Who knows what downstream effect in the big chain of causality this particular person/action had! Pollution/carbon taxes are an easy exception.
Aside, a bit off topic: Even if we could compute Shapley values, Shapley values suffer from combinatorial explosion. (Things other than Shapley values or approximating them might work.)