Building Better Prediction Models for Consumer Choices

Caltech News reports that researchers at Caltech and MIT have solved a decades-old puzzle in how economists predict consumer behavior. The random utility model—a mathematical framework used to forecast everything from restaurant choices to car purchases—has long struggled when not all options are observable. Caltech professor Kota Saito and MIT student Alec Sandroni found you can extract reliable information from what people didn't choose, even with incomplete data. Using network flow theory, they showed the traditional workaround actually introduces severe bias. The breakthrough, published in the American Economic Review, could sharpen pricing strategies, antitrust analysis, and policy decisions across industries.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton