arXiv subjects

Evgeny Kagan

Publications and source records attributed to Evgeny Kagan.

7 recordsLinked to original sources

arXiv2026

What's in a Queue? An Experimental Study of Job Ordering, Autonomy and Queue Visibility

Problem Definition: How a queue of jobs is arranged and presented to workers is an important design problem in service operations. This includes choosing the order in which jobs are performed, how much say workers have in setting that order, and how much queue and arrival information workers receive. Methodology/Results: To better understand how queue design (job ordering, autonomy, visibility) affects worker performance (speed, quality), we run a series of pre-registered online experiments. We use a new, real-effort task in which workers fulfill order-picking jobs of varying complexity that arrive dynamically over time. Our results are as follows: (1) When workers choose their own picking order, we reproduce the field finding that Easy First (EF) ordering is associated with worse performance than First-in-first-out (FIFO), and show that this is mainly due to worker self-selection rather than due to the ordering itself; (2) Exogenously imposed EF ordering improves work quality (picking accuracy) relative to both FIFO and discretionary ordering; (3) Imposing an ordering may reduce speed for the most capable workers; (4) Seeing a new job arrival leads to a short-term productivity burst; however, removing job arrival and queue information altogether does not affect performance in the long term. Managerial implications: Our results provide guidance on which queue design works best for a given performance goal (speed or quality) and worker ability level. We also identify personality measures that can help managers screen for error-prone workers.

econ.GN↗

arXiv2026

One at a Time? The Personal Productivity Bias in Emergency Department Patient Assignment

Emergency departments (EDs) often use a shared-queue setup in which physicians self-assign cases from a pool of triaged patients. We conduct a multi-method study to examine this self-assignment behavior and its effects on system performance. Using data from five EDs spanning 1.4 million patient visits, we show that batching, i.e., self-assigning multiple patients at once, is common and associated with longer stays for batched patients, even after controlling for clinical acuity, physician fixed effects, and ED congestion. We then develop a continuous-time queueing model that characterizes the optimal self-assignment policy under individual and group throughput incentives. We use the model predictions to test experimentally with 203 healthcare workers and 73 ED physicians whether batching is a rational response to incentives or a deeper behavioral tendency that persists independent of incentives. Indeed, batching is pervasive across both samples, with 94% of healthcare workers and 73% of physicians choosing to batch even when it reduces their own payoffs -- a behavior that we term the personal productivity bias. Together, these results suggest that compensation redesign alone is unlikely to eliminate batching, and suggest changes to the assignment interface in the electronic health record system as a more promising remedy.

econ.GN↗

arXiv2026

Startup Contracting and Entrepreneur-Investor Bargaining (Long Version)

To grow their businesses, entrepreneurs often rely on equity funding. This paper focuses on two elements of entrepreneur-investor equity negotiations: the number of potential investors and the contractual complexity surrounding investor protection. Our approach involves a theoretical model and a series of laboratory experiments that analyze the effects of different bargaining conditions and contractual terms on the equity (ownership) split between entrepreneurs and their investors. We show that the conventional wisdom that entrepreneurs should seek to negotiate with as many investors as possible, while consistent with the theoretical model, is not true in the data. Indeed, negotiating with multiple investors reduces the entrepreneur's profits under most conditions. We also show that investor downside protections may disadvantage early-stage startups, but can be beneficial to later-stage startups. A refinement of belief modeling in multi-party bargaining, as well as a stylized risk allocation framework, reconcile these results with theory predictions. Our findings provide a decision framework for entrepreneurs to optimize their approach to investors and negotiate favorable contractual terms.

econ.GN↗

arXiv2026

On Repeat: Does Iteration Drive Innovation?

Motivated by the widespread adoption of iterative project management techniques, we study the effects of workflow -- iterative or sequential -- on innovative behavior and performance. We conduct a series of laboratory experiments. Our first experiment shows that, in an open-ended creative challenge, iterative task completion leads to better outcomes than sequential task completion. In the second experiment we show that the advantage of iterative workflow further extends to innovation settings that do not involve idea generation. A key mechanism driving the advantage of iterative work is that it leads to frequent task switching, prompting workers to perform a broader search for the best available solution. In the third experiment we delve deeper into the search process and show that sequential work indeed leads to more myopic idea refinement behaviors, often ending in a (suboptimal) local maximum. Our results suggest that iterative workflow improves performance across multiple, structurally distinct innovation settings. We also identify three boundary conditions. First, iterative workflow helps achieve quick gains, but its performance advantage narrows over time. Therefore, workflow effects are stronger when balanced performance across project components is required, but weaker when excellence in one component can offset poor performance in others. Second, workflow has minimal effect on performance in tasks that do not require the worker to perform broad exploration. Third, workflow effects are minimal when workers complete the easier component first.

econ.GN↗

arXiv2026

Customer Service Operations: A Gatekeeper Framework

Customer service has evolved beyond in-person visits and phone calls to include live chat, AI chatbots and social media, among other contact options. Service providers typically refer to these contact modalities as "channels". Within each channel, customer service agents are tasked with managing and resolving a stream of inbound service requests. Each request involves milestones where the agent must decide whether to keep assisting the customer or to transfer them to a more skilled -- and often costlier -- provider. To understand how this request resolution process should be managed, we develop a model in which each channel is represented as a gatekeeper system and characterize the structure of the optimal request resolution policy. We then turn to the broader question of the firm's customer service design, which includes the strategic problem of which channels to deploy, the tactical questions of at what level to staff the live-agent channel and to what extent to train an AI chatbot, and the operational question of how to control the live-agent channel. Examining the interplay between strategic, tactical, and operational decisions through numerical methods, we show, among other insights, that service quality can be improved, rather than diminished, by chatbot implementation.

cs.HC↗

arXiv2026

Chasing Tails: How Do People Respond to Wait Time Distributions?

We use a series of pre-registered, incentive-compatible online experiments to investigate how people evaluate and choose among different waiting time distributions. Our main findings are threefold. First, consistent with prior literature, people show an aversion to both longer expected waits and higher variance. Second, and more surprisingly, moment-based utility models fail to capture preferences when distributions have thick-right tails: indeed, decision-makers strongly prefer distributions with long-right tails (where probability mass is more evenly distributed over a larger support set) relative to tails that exhibit a spike near the maximum possible value, even when controlling for mean, variance, and higher moments. Conditional Value at Risk (CVaR) utility models commonly used in portfolio theory predict these choices well. Third, when given a choice, decision-makers overwhelmingly seek information about right-tail outcomes. These results have practical implications for service operations: (1) service designs that create a spike in long waiting times (such as priority or dedicated queue designs) may be particularly aversive; (2) when informativeness is the goal, providers should prioritize sharing right-tail probabilities or percentiles; and (3) to increase service uptake, providers can strategically disclose (or withhold) distributional information depending on right-tail shape.

econ.GN↗

arXiv2025

Deploying Chatbots in Customer Service: Adoption Hurdles and Simple Remedies

Despite recent advances in Artificial Intelligence, the use of chatbot technology in customer service continues to face adoption hurdles. This paper explores reasons for these adoption hurdles and tests several service design levers to increase chatbot uptake. We use incentivized online experiments to study chatbot uptake in a variety of scenarios. The results of these experiments are threefold. First, people respond positively to improvements in chatbot performance; however, the chatbot channel is utilized less frequently than expected-time minimization would predict. A key driver of this underutilization is the reluctance to engage with a gatekeeper process, i.e., a process with an imperfect initial service stage and possible transfer to a second, expert service stage -- a behavior we term "gatekeeper aversion". We show that gatekeeper aversion can be further amplified by a secondary hurdle, algorithm aversion. Second, chatbot uptake can be increased by providing customers with average waiting times in the chatbot channel, as well as by being more transparent about chatbot capabilities and limitations. Third, methodologically, we show that chatbot adoption can depend on experimental implementation. In particular, chatbot adoption decreases further as (i) stakes are increased, (ii) the human/algorithmic nature of the server is manipulated with more realism. Our results suggest that firms should continue to prioritize investments in chatbot technology. However, less expensive, process-related interventions can also be effective. These may include being more transparent about the types of queries that are (or are not) suitable for chatbots, emphasizing chatbot reliability and quick resolution times, as well as providing faster live agent access to customers who experienced chatbot failure.

cs.HC↗