By Alex Bloedel
Having good information is key to making good decisions, such as purchasing a high-quality product or selecting a profitable investment. Economists, therefore, have long viewed information as a scarce and valuable resource. However, in the modern world—since the advent of online commerce, social media, and real-time newsfeeds—information is arguably so abundant that the real constraint on good decision-making is people’s ability to absorb the information available to them. Simply put, information is plentiful while people’s attention is scarce.
What, then, is the optimal way to provide information to people who cannot absorb all of it? Consider the problem of an online retailer displaying information about its products to consumers (e.g., Amazon listing product recommendations and user reviews). Consumers find it costly to pay attention to the website’s contents and therefore may not fully absorb all the displayed information. Instead, they optimally allocate their limited attention (e.g., by choosing for how long and in what order to visually scan the page) with the goal of optimizing their purchase decisions. By choosing what information about product characteristics to display, the platform influences consumers’ optimal allocation of attention, which in turn determines the consumers’ purchase decisions and the platform’s revenues.
In their paper “Persuading a Rationally Inattentive Agent”, Professors Alex Bloedel (UCLA) and llya Segal (Stanford) develop a theoretical framework to study such information design problems. In their model, a Sender (e.g., online retailer) chooses what information to disclose about a payoff-relevant state (e.g., product quality) to a “rationally inattentive” Receiver (e.g., consumer), who optimally decides how much and which pieces of Sender’s information to attend to before taking an action (e.g., buying a product or not). Receiver finds paying attention costly, while Sender only cares about the quality of Receiver’s decision.
The authors characterize the form of Sender’s optimal disclosure. For example, when the agents’ preferences over actions are perfectly aligned (e.g., the online retailer maximizes social surplus), providing full information would be optimal if Receiver were fully attentive. But with costly attention, Sender optimally withholds information about “extreme” states in which the stakes of taking the correct action are greatest, while providing full information about “intermediate” states. Intuitively, Sender provides a definitive recommendation when the correct action is clear-cut, and lets Receiver learn on his own otherwise. Somewhat subtly, this improves upon full disclosure because it induces Receiver to pay closer attention to “medium-stakes” states, thereby increasing the total amount of attention he pays and the overall quality of his decisions.
The paper delivers a number of additional insights. For one example, when Sender’s and Receiver’s preferences are misaligned (e.g., the online retailer maximizes revenue), it is optimal to disclose detailed information that “distracts” Receiver and increases the likelihood that she makes mistakes (e.g., purchases low-quality products). For another example, depending on the form of Receiver’s attention cost, it can be optimal for Sender to simplify or complexify the “language” in which she presents information to Receiver—even when doing so does not change the actual information conveyed (e.g., the online retailer merely changes the color or font of a product description).
This research contributes to an exciting new area of study aimed at designing good economic institutions for the modern “information-rich” world, in which managing peoples’ scarce attention is a first-order concern.