Seyed Iravani--Research

Imagine you want to buy a camera. Scrolling through your phone, you find a wealth of options available at your favorite e-commerce store. Now, imagine narrowing your choices down to two models. They both are respected brands, enjoy similarly positive reviews, and are priced nearly the same. Yet, there’s one difference: one camera’s product page notes it has sold 20 units. The other has sold 4,000 units. Which one would you choose?
While this information does not affect a professional photographer (an informed customer), it certainly affects our decisions. Most of us opt for the camera backed by thousands of previous purchases.

This phenomenon in behavioral economics is called "Peer Effect." According to Peer Effect, when non-informed customers approach a decision with incomplete information, they often make the decision chosen by the majority of people. Hence, by providing (or not providing) the number of products sold, online retailers can impact customers' purchasing decisions. In this project our research questions were: How do rational (uninformed) customers choose a service? How is a customer's purchasing decision affected by the fraction of informed customers (or by advertising that provides more information about the product)? How is the decision affected by the number of products or services offered to them? We used Bayesian Games and Queueing Theory to analyze the problem assuming rational customers. We then developed several lab experiments (using human subjects) and studied the violation of the decision made by human subjects from the optimal decision recommended by our rational model. In the last stage of our research, we incorporated the behavior observed in the lab experiment into our game theoretical model and showed that the resulting model has a much larger ability to predict customer purchasing behavior in the presence of peer effect. (Joint work with L. Debo, J. Chen, and M. Kremer)

Operations management uses optimization models to find the best decision for a variety of management situations. All these models assume that the decision maker is a rational person that can accurately weigh the costs and benefits of his or her decision and identify the best choice. But, is that what really happens? Do managers or consumers always make rational decisions? There are many examples in operations and economics that provide clear evidence that humans are not rational decision makers and are influenced by their emotions and social preferences. For example, when Williams-Sonoma, a high-scale specialty retailer that sells cookware, introduced their home “bread bakery” machine for $275, they were not able to sell it in the market. While the company was thinking about stopping the production of the machine, they hired a marketing research firm for advice. To their surprise, the marketing firm did not recommend removing the product. In fact, their recommendation was that William-Sonoma introduce an additional model, which is a bit larger, but price it at $415, which is about 50% higher than the current machine. This was puzzling: If consumers were not willing to by the current machine, why would they willing to pay 50% more for a machine that is a bit larger? Nevertheless, William-Sonoma followed the marketing firm's suggestion and sales of their smaller machine began to rise. Why?
The answer is because consumers are not rational decision makers. They are influenced by what is known as "decoy effect." With the two choices of bread making machines, consumers find that they can purchase the smaller machine, which is almost as good as the larger $415 machine, for only $275. What a good deal!

Decoys are one of many factors in behavioral decision making. The new field of Behavioral Operations Management brings the two fields of Operations Management and Behavioral Decision Making together and develops models that enhances our understanding of whether, why, when, and how our rational optimization models deviate from the actual observed behavior. It then uses these deviations and designs better predictive models and algorithms, which in turn are used to improve operational decisions. By understanding the role of behavior in operations, we can help businesses to design robust operating systems that take into account the behavioral tendencies of managers, workers, and customers.

Consider the following two pricing strategies for Gas Stations A and B. Gas Station A sells gasoline for $4.00 per gallon but gives a "discount" of $0.10 per gallon if you pay with cash. Gas Station B sells gasoline for $3.90 per gallon but charges a "surcharge" of $0.10 per gallon if you pay with a credit card. While both customers who pay with cash or credit card pay the same amount in both stations, why do we never see Station B’s pricing strategy?

The answer is explained by Prospect Theory. Daniel Kahneman, who received the Noble Prize in Economics for introducing Prospect Theory, explains that people having asymmetric values for gain and loss. According to prospect theory, the disutility (i.e., unhappiness) of losing $100 is about two times larger than the utility (happiness) of gaining $100. So, people who want to pay with a credit card, do not like the "surcharge" of $0.1 in Station B, although they pay the same amount if they go to Station A. There are many such examples in which firms present their services and prices such that they look more like gains than losses.
One such application is in delay announcement in service systems. Airlines overestimate their flight times, so when passengers arrive earlier than the announced time (a gain) they feel satisfied. Restaurants always overestimate the wait time for their tables for the same reason.
In a joint project with an Emergency Department of a local hospital in Chicago, we performed an experiments to study whether announcing the wait time to arriving patients improves patient satisfaction with wait time, and if it does, whether the patients should be given an overestimation of their expected wait time, and by how much. This empirical field experiment provided useful insights into the delay announcement strategies in emergency departments. The methodologies used were empirical research methods, prospect theory and queueing theory. (Joint work with S. Ansari and L. Debo)