Abstract: We show that rational but inattentive agents can become polarized, even in expecta- tion. This is driven by agents’ choice of not only how much information to acquire, but also what type of information. We present how optimal information acquisition, and subsequent belief formation, depends crucially on the agent-specific status quo valuation. Beliefs can systematically update away from the realized truth and even agents with the same initial beliefs might become polarized. We design a laboratory experiment to test the model’s predictions; the results confirm our predictions about the mechanism (rational information acquisition) and its effect on beliefs (systematic polarization).
A Note on Optimal Experimentation under Risk Aversion,
with Godfrey Keller and Tim Willems, 2019. Journal of Economic Theory, 179, 476-487.
Abstract: In a standard two-armed bandit setup, this paper shows – counterintuitively – that a more risk-averse decision maker might be more willing to take risky actions. The reason relates to the fact that pulling the risky arm in bandit models produces information on the environment – thereby reducing the risk that a decision maker will face in the future. This finding gives reason for caution when inferring risk preferences from observed actions: in a bandit setup, observing a greater appetite for risky actions can actually be indicative of more risk aversion, not less.
Scheduling of Multi-class Multi-server Queueing Systems with Abandonments,
with Urtzi Ayesta and Peter Jacko, 2017. Journal of Scheduling, 20(2), 129-145.
Abstract: Many real-world situations involve queueing systems in which customers may abandon if service does not start sufficiently quickly. We study a comprehensive model of multi-class queue scheduling accounting for customer abandonment, with the objective of minimizing the total discounted or time-average sum of linear waiting costs, completion rewards and abandonment penalties of customers in the system. We assume the service times and abandoning times are exponentially distributed. We solve analytically the case in which there is one server and there are one or two customers in the system and obtain an optimal policy. For the general case, we use the framework of restless bandits to analytically design a novel simple index rule with a natural interpretation. We show that the proposed rule achieves near-optimal or asymptotically-optimal performance both in single- and multi-server cases, both in overload and underload regimes, and both in idling and non-idling systems.
Work in progress:
Cognitive Perception and Aspiration Selection
Political Platform Building with Rationally Inattentive Voters
with Sergei Mikhalishchev
Estimating Models with Rational Inattentive Agents
Abstract: We present a likelihood evaluation of a DSGE model with price-setting firms that select properties of their signals subject to a limited attention constraint. We compare the performance of a rational inattention DSGE model (RIM), with an imperfect common knowledge model (ICKM) and a model with price stickiness à la Calvo. We demonstrate that the rational inattention model matches the data better than the Calvo model and reproduces the persistence more easily than the ICKM model. This result occurs because (i) RI firms pay attention to a higher number of lags of fundamentals than is assumed in the ICKM models, and (ii) the full information method selects the different degree of strategic complementarity in various models.
Whittle’s Indexation Approach to and Applications of Bi-objective Two-state Binary-action Markov Decision Processes,
with Peter Jacko, 2015. In A.B. Piunovskiy (Ed.), Modern Trends in Controlled Stochastic Processes: Theory and Applications, Volume II (pp. 140-151). UK:Luniver Press. Invited chapter.
Abstract: In this chapter we present how the Whittle’s indexation approach, originally developed for the restless bandits problem, can be used to address multi-objective stochastic optimization problems. For clarity, we focus on Markov decision processes with two objectives, two states and two actions, and provide an optimal solution in terms of Whittle indices. We then broadly discuss the applications of our solution in optimal scheduling of queueing systems with abandonments and optimal choice of venture capitalists investments. We believe there is a need to develop the Whittle index theory further in order to tackle more general multi-objective problems.