Hey Walt, There's no circularity if prices are determined simultaneously based on demand (household preferences, income, and number) and supply (technology, firms) conditions. And if there were circularity involved, it wouldn't be unique to marginal analysis--Sraffian price of production equations would face the same charge, if relevant. And simultaneity is just the worst-case scenario; perhaps there is some additional structure that allows some prices to be determined before others, e.g. through long-term contracting.
So, the price of a good is determined by the utility (or some other property) of the marginal unit of the good. But, obviously, which unit is the marginal one consumed depends on the price.
No functioning mainstream economist would say "the price of a good is determined by the utility of the marginal unit of the good", any more than they'd say that one blade makes a pair of scissors.
This of course could be resolved by postulating a demand curve with all of each price and its corresponding marginal unit, and the corresponding profit maximizing price chosen. This then relies on cost of production.
This isn't a "resolution", just a restatement--commodity demand curves are derived theoretically from the utility maximizing behavior of price-taking households. Supply, in turn, is derived from the profit-maximizing behavior of firms.
You can regress all the way to a point where there is no circularity.
There wasn't any evident circularity to begin with. There is a puzzle about how prices are arrived at, but that would be true for any story of price determination, not just one based on marginal analysis.
But, for more political economy oriented people, doesn't the cost of production of goods, even when you regress to the point where there are no produced inputs, depend on prices? Say at certain prices of goods, workers will effect a slowdown in protest etc...
OK, so now you add a new feedback loop in the simultaneous or sequential determination of prices. Still no necessary circularity.
Is this a circularity or is it resolvable?
Theoretically, at least, resolvable. FWIW, Gil
