In a recent post, Matt Kahn revives a question that I get in every first session of an MBA class. Can demand curves be upward sloping?
Upward Sloping Demand Curves? The Case of the NY Metropolitan Museum of Art https://t.co/TSNElHOsdQ I pose a new structural IO problem .
— Matthew E. Kahn (@mattkahn1966) July 16, 2017
And in his post:
A museum (as I learned at Versaiilles, France recently) suffers from congestion problems. This congestion is especially bad if the price of entrance is $0. (…) BLP should reunite to write a structural IO paper on museum demand and the resulting consumer surplus with endogenous product attributes!
The paper does not per se execute Matt’s idea. The paper shows visitors to the British Museum pictures with different levels of congestion and asks them whether they would prefer free admission or 3 GBP (resp. 8 pounds, etc). There are some significant issues with an approach of this kind (see Do People Mean What They Say? by Bertrand and Mullainathan 2003), but the results are interesting. A more ambitious study would randomize actually paid prices (as in Levitt’s Uber paper) and then estimate demand, while instrumenting for congestion (i.e. using idiosyncratic sources of variation for congestion, weather, or other sources). There are many ways in which IO can push the frontier in the estimation of consumer demand.