Jeeva Somasundaram

Assistant Professor of Decision Sciences (Operations & Technology area)

IE Business School, Madrid, Spain

Published Papers

[1] Regret Theory: A New Foundation

We present a new behavioral foundation for regret theory. The central axiom of this foundation — trade-off consistency — renders regret theory observable at the individual level and makes our foundation consistent with the existing measurement method. For the first time, our behavioral foundation allows deriving a continuous regret theory representation and separating utility from regret. Finally, the axioms in the behavioral foundation clarify that regret theory minimally deviates from expected utility by relaxing transitivity only.

Awards: Finalist, 2014, INFORMS Decision Analysis Society Student Paper Award

[2] Regret Theory and Risk Attitudes

We examine risk attitudes under regret theory and derive analytical expressions for two components - the resolution and regret premiums - of the risk premium under regret theory. We posit that regret-averse decision makers are risk seeking (resp., risk averse) for low (resp., high) probabilities of gains and that feedback concerning the forgone option reinforces risk attitudes. We test these hypotheses experimentally and estimate empirically both the resolution premium and the regret premium. Our results confirm the predominance of regret aversion but not the risk attitudes predicted by regret theory; they also clarify how feedback affects attitudes toward both risk and regret.

[3] Risk and time preferences interaction: An experimental measurement

(with Vincent Eli, Forthcoming at Journal of Risk and Uncertainty )

We experimentally characterize and measure the interaction between risk and time preferences. Our results indicate that risk and time preferences are intertwined. We find that decision makers are insensitive to time delay for small probabilities of gains, but become progressively more sensitive to time delay as the probability of gain increases. We compare the fit of existing decision models that capture risk and time preferences. Our results indicate that the models which allow for probabilitytime interaction and capture magnitude effect fit the data better. We also show that accounting for risk-time preferences interaction leads to lower estimated discount rates.

Working papers

[4] Raising the Temperature in the Tropics: Gradual Targets Promote Energy Conservation Habits

(with Teck-Hua Ho, Noah Lim, and Ingrid Koch)

In tropical countries, ACs account for a significant fraction of energy consumption. Increasing the AC temperature can lead to substantial reduction in energy use. We conduct a randomized control trial (RCT) in which people are financially rewarded for setting a higher AC temperature than usual and investigate whether people can adapt to the higher temperature. Along with a no-incentive control condition, two treatments were conducted over two time periods. In treatment 1, subjects were incentivized to raise the temperature gradually, by 1°C in period 1, and an additional 1°C in period 2. In treatment 2, subjects were only incentivized to abruptly increase the temperature by 2°C in period 2. The treatment with gradual temperature targets worked better in maintaining the higher AC temperature during the intervention and post-intervention periods. Energy usage data confirmed that these higher AC temperatures translated into lower energy consumption.

[5] The Reassurance effect in information acquisition

To be effective, marketers of information need insights into the motivational bases of information acquisition. Conventionally, information acquisition is motivated by its instrumental value—the value derived from adapting action to new knowledge. In this paper, we analyse a model of individual preferences that anticipate elation and disappointment—the emotional responses to good and bad news. We assume loss aversion (bad news looms larger than good news) and diminishing sensitivity. While emotions are generally presumed to cause information avoidance, our key finding is that a consumer faced with a large potential loss of low probability may seek non-instrumental information for the purpose of reassurance. When information is weakly instrumental, the pursuit of reassurance can cause a consumer who is less likely to face a loss to value information more than a consumer who is more likely to face the same loss. This paradox disappears at higher levels of information instrumentality. We provide empirical support for this effect, first in the context of an incentivized controlled experiment involving a compound lottery choice (N = 403), and then in the context of a survey (N = 1349) about the desire to undertake COVID-19 testing, carried out at the early stages of the pandemic.

Awards: Best Paper Award in the theme “Marketing, Wellbeing, and Healthcare” at the 2018 American Marketing Association Winter Academic Conference.

[6] Rational addiction in Mobile consumption

(with Laura Zimmermann and Pham Quang Duc, Reject & Resubmit, Management Science )

Understanding how to incentivize consumers to change a behavior depending on the nature of inter-temporal consumption is important for marketers. We develop a model that examines how a (forward-looking or myopic) consumer responds to temporary monetary incentives offered to reduce the consumption of a good depending on the nature of inter-temporal consumption (habit forming or satiating). We empirically test our predictions in the domain of mobile usage since many consumers would like to reduce their smartphone usage but fail to do so. In two pre-registered randomized control trials, we provide temporary targets and monetary incentives to reduce smartphone usage. When future targets and incentives are pre-announced, subjects behave in a forward-looking (rather than myopic) manner and reduce their mobile usage as predicted by habit formation models even before they are actually incentivized to do so (rather than increase consumption as predicted by satiation models). Subjects maintain a lower usage during the incentivized period and sustain a lower usage even after incentives are removed. Consistent with predictions of habit formation, we show that the reduction in post-treatment usage is driven mainly by subjects who reduced their usage during the treatment periods. Our manuscript thus provides empirical support for the theoretical framework of rational addiction (or forward-looking habit formation) in the domain of mobile usage and shows that pre-announcing future targets and incentives can be a cost-effective intervention to kickstart behavioral change in this domain.

[7] Adoption of New technology Vaccines

(with Laura Zimmermann and Barsha Saha, Revise & Resubmit, Journal of Marketing )

We explore aversion towards new technology and ways to overcome this aversion with a simple nudge in the context of vaccine decision making. A plethora of research has shown that consumers are hesitant to receive a potentially life-saving vaccination. But are consumers more hesitant towards certain vaccines (i.e., those that are based on a new technology such as mRNA) even if they are more effective? We present four experiments (Ntotal = 478) which test the propositions of a formal model that incorporates ambiguity and other people’s choices into the decision to vaccinate. We show that consumers are unduly averse towards new technology vaccines compared to traditional technology vaccines due to higher perceived uncertainty of side effects. We test a simple nudge based on herd behavior to overcome this new technology aversion. By communicating increasing population vaccination rates, we effectively increase uptake of new technology vaccines at a higher rate than uptake of traditional vaccines. Rather than being driven by social conformity or social learning, information about herd behavior seems to alleviate perceived uncertainty (or ambiguity) by narrowing the confidence interval of risk estimates. We do not find evidence of free-riding due to herd immunity.

[8] Algorithmic support and newsvendor risk attitudes

(with Matthias Seifert and Prana Narayan )

We study experimentally how algorithmic support influences newsvendor risk attitudes. Using the anchoring paradigm, we derive predictions when decision-makers (DMs) observe risk averse algorithmic recommendations and test the predictions in two experimental studies involving single and multi-item newsvendor tasks (studies 1 and 2, respectively). Across both studies, we find that risk averse algorithms lead to “sticky” order decisions: subjects who receive risk averse algorithmic recommendations become more risk averse and continue to remain so even after the algorithm is no longer available. This higher level of risk aversion significantly reduces the “pull-to-center” bias. Furthermore, we find that subjects tend to anchor on and appreciate algorithmic recommendation more compared to equivalent human advice (study 3). Our findings demonstrate the mutability of newsvendor risk preferences when temporarily exposed to algorithmic aids.

[9] Incentivizing contest participation

(with Konstantinos Stouras )

This experiment compares the performance of three key contest designs on attracting participants: the number of potential participants (pool size 3 vs. pool size 7), the prize type (Proportional-Prize (PP) vs. Winner-Takes-All (WTA) contest), and the degree of diversity in participant ability (homogeneous vs. diverse). Contrary to Nash predictions, we find that participation increases in pool size and all contests of pool size 7 exhibit over-participation. Yet, as predicted by Nash equilibrium, participation decreases in ability diversity. Further, prize type does not influence participation except for homogeneous contests with pool size 7. Overall, we find that the homogeneous WTA contest of pool size 7 maximizes participation, but the pool size 3 diverse contests maximize subjects' total payoff. These findings could inform the design of competitive markets with endogenous supply such as on-demand services provided by independent contractors, platforms to crowdsource tasks, and philanthropic campaigns or marketing promotions that aim to attract participants and reward achievement.

Work in Progress

[11] Replacing screen time with step count: Evidence from a field experiment

(with Laura Zimmermann and Pham Quang Duc)

[12] Over-Diagnosis Equilibria: The Willful Marketing Of False Positives

(with Luc Wathieu, Preliminary model and results available)

In many markets, the decision to buy a product is preceded by taking a diagnostic test (or, more generally, gathering information to assess relevance). Testing is itself a proliferating industry in many domains, and critics have argued that, particularly in the medical context (but the problem is broadly relevant for many classes of consumer searches), consumers tend to engage in unnecessary amounts of testing. The purpose of this paper is to analyze over-testing as a market equilibrium. We start with a rational analysis of a doctor’s decision to test for the presence of a medical condition. Then, we take the perspective of a monopolistic firm that develops and sells diagnostic tests, and we identify this firm's decisions in terms of optimal test coarseness and pricing. We find that it is optimal for firms to offer a sequence of tests of increasing diagnosticity and price, such that a significant pool of consumers will be over-diagnosed initially, as compared to the socially optimal outcome. We show how free entry competition on the testing market will not necessarily resolve the over-diagnosis problem.

Other Research Papers & Projects

[1] Behavioral Game theory

Standard game theory has been criticized for requiring unreasonable assumptions about human cognitive abilities, and for failing to predict actual choices in experiments and in the field. Recent developments in behavioural game theory address these shortcomings by allowing people to be boundedly rational. We discuss two such developments: (1) the cognitive hierarchy (CH) model, which has successfully predicted non-equilibrium behaviour in one-shot games; and (2) the experience weighted attraction (EWA) model, which accurately tracks how choices converge to equilibrium over time. We use the famous p-beauty contest to demonstrate how both models work. Both have successfully been applied to many games and in the field.