An Inverse Problem in Economics
Wednesday, April 24, 2002 - 3:00pm - 4:00pm
Suzhou Huang (Ford Research Laboratory)
A consumer's behavior on a market should reflect his/her underlying preference. Given a rational decision making process for each consumer, extracting consumer preference from observed market data is, therefore, a typical inverse problem. I will first briefly describe how this kind of problems is mathematically formulated in economics. Then, I will review what are typically practiced in recent econometric studies. There are two types of uncertainties that could impede the accuracy of the inversion: measurement error and model misspecification. The latter is a harder problem to deal with and can potentially give rise to misleading results. Concrete illustrations will be given to exemplify the problem and demonstrate its seriousness. The purpose of this presentation is to stimulate exchange of ideas and methodologies across various disciplines and to solicit inputs from experts.