Thursday, December 14, 2006
Starburst Considered As a Giffen Good??
by Ken Houghton
One of the more facile arguments against economics-as-science is that "it doesn't work that way in the real world."* But what we have here is not so much a real-world failure as a failure of the presumption of monotonicity at or near p=0.
In science, it is relatively easy to revise the model, and so it is in economics. The difference may well be that the teaching doesn't change.
To be clear, one of the studies cited by Shampan’er and Ariely was used as the first chapter (iirc) of Freakonomics—charging a nominal amount for picking up children late caused more people to be late, since they placed more value on their time than the cost of the "late fee."
It is not difficult to make clear that the standard "Econ 101" models (e.g., the minimum wage will cause people to lose jobs if everyone has the same power—that is, there is not such thing as monopsony) have underlying assumptions, and that there are multiple known areas where the assumptions may not apply.
It's probably heresy to say it here, but economics won't be a science until we start teaching it like one. And declaring that there are areas where our underlying mathematical assumptions may not hold is a lot easier for people to believe than the idea that Starburst is a Giffen Good.
Via Mark Thoma comes one of those "yes, this is obvious, and we should talk about it" articles from the FRB Boston(PDF link here).
Related to these findings on motivation and incomplete contracts, it has also been shown that when prices are mentioned, individuals apply market norms, while when prices are not mentioned (that is, when the price is effectively zero), individuals apply social norms to determine their choices and effort (Heyman and Ariely, 2004). As an illustration of this idea, Ariely, Gneezy, and Haruvy (2005) have shown that when Starburst candy is offered at a cost of 1¢ per piece, an average MIT student is likely to take about four pieces, whereas when the price is zero, more students will take some, but almost no‐one will take more than one (effectively creating a decreased demand when prices are reduced).
One of the more facile arguments against economics-as-science is that "it doesn't work that way in the real world."* But what we have here is not so much a real-world failure as a failure of the presumption of monotonicity at or near p=0.
In science, it is relatively easy to revise the model, and so it is in economics. The difference may well be that the teaching doesn't change.
To be clear, one of the studies cited by Shampan’er and Ariely was used as the first chapter (iirc) of Freakonomics—charging a nominal amount for picking up children late caused more people to be late, since they placed more value on their time than the cost of the "late fee."
It is not difficult to make clear that the standard "Econ 101" models (e.g., the minimum wage will cause people to lose jobs if everyone has the same power—that is, there is not such thing as monopsony) have underlying assumptions, and that there are multiple known areas where the assumptions may not apply.
It's probably heresy to say it here, but economics won't be a science until we start teaching it like one. And declaring that there are areas where our underlying mathematical assumptions may not hold is a lot easier for people to believe than the idea that Starburst is a Giffen Good.
