Econolimerick #6
“In good theory utility is ordinal.
Whenever I hear the word ‘cardinal’
I think of blue unicorns
and three fates called Norns
and insist on the word ‘pseudo-cardinal.’”
Note: there are three prominent examples of pseudo-cardinality, all in cases where additive separability is convenient:
the utility function for expected utility,
the period utility function for an additively time-separable overall utility function,
the contribution to social welfare from each individual’s welfare in an additively-separable social welfare function.
In each of these cases (even in 3, where there is interpersonal comparison going on), I consider it only pseudo-cardinality since one can do any monotonic transformation of overall utility (in these cases, the sum) without changing the preferences that are implied.
Cardinality would be if there were some way of measuring utility that was independent of looking at the choices people make. People claim that happiness data does that, but it isn’t so. (See “My Experiences with Gary Becker.”) And even if there were an independent way of measuring utility besides looking at the choices they made, why wouldn’t we be OK also with monotonic transformations of that independent way of measuring?
Don’t miss my other econolimericks: