INFORMS Podcast Series Features Chris Dellarocas on Double Marginalization in Performance-Based Advertising
“The Science of Better” Spotlights Dellarocas’s Management Science Article
In his June 2012 Management Science article “Double Marginalization in Performance-based Advertising: Implications and Solutions,” Chrysanthos Dellarocas explores an unexpected consequence of online pay-per-action systems (PPAs), such as the pay-per-click model—research that was recently highlighted by the Institute for Operations Research and Management Science (INFORMS) podcast series “The Science of Better.”
Dellarocas is Professor (effective September 2012) and Chair of Information Systems at Boston University School of Management. He is also a widely-cited expert in online reputation, social media, and information economics.
Although the pay-per-click model and other PPA systems in online markets were intended to reduce advertising costs and boost efficiency, Dellarocas reports in his Management Science paper that they “tempt firms to increase the prices of their goods so they generate fewer clicks/sales but make more profit per click and, most importantly, pay fewer commissions to pay-per-click publishers, such as Google.” Such behavior reduces consumer surplus. It can also “reduce publisher revenues relative to pay-per-exposure methods,” creating what the author calls “a form of double marginalization.”
“It’s possible to calculate a system that induces sellers to maintain product price at the levels that would maximize the profits if they bought advertising the traditional way. Ultimately this is to the benefit of the advertisers as well.”
In the podcast posted by INFORMS spotlighting this research, Dellarocas discusses his findings, explaining the phenomenon of double marginalization and offering solutions to improve pricing equilibrium and consumer surplus, as well as to improve publishers’ expected profits in PPA systems.
“One idea is a system of penalties and rewards on top of standard pay-per-clicks,” Dellarocas suggests. “If you buy an ad but nobody clicks on it—because, for example, your price is too high—then the next time you buy an ad from the same place, your cost-per-click would be just a little higher.”
Comparing this model to that of auto insurance premiums, Dellarocas explains, “The objective is to discourage sellers from increasing the prices of their products so they can receive fewer clicks (and, therefore, pay the publishers fewer times) but make more profits when they do.”
This works, Dellarocas argues, because, as he shows in his Management Science paper, “It’s possible to calculate a system that induces sellers to maintain product price at the level that would maximize the profits if they bought advertising the traditional way. Ultimately this is to the benefit of the advertisers as well.”
Listen to the full podcast at the INFORMS site The Science of Better.