|
on Discrete Choice Models |
| By: | Roy Allen |
| Abstract: | In a consideration set model, an individual maximizes utility among the considered alternatives. I relate a consideration set additive random utility model to classic discrete choice and the extended additive random utility model, in which utility can be $-\infty$ for infeasible alternatives. When observable utility shifters are bounded, all three models are observationally equivalent. Moreover, they have the same counterfactual bounds and welfare formulas for changes in utility shifters like price. For attention interventions, welfare cannot change in the full consideration model but is completely unbounded in the limited consideration model. The identified set for consideration set probabilities has a minimal width for any bounded support of shifters, but with unbounded support it is a point: identification "towards" infinity does not resemble identification "at" infinity. |
| Date: | 2024–05 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2405.13945&r= |
| By: | Antonio Estache; Renaud Foucart; Konstantinos Georgalos |
| Abstract: | In a laboratory experiment, we find that subjects do not exhibit preference for control when the alternative is a random dictatorship, a lottery implementing either their choice or the choice of someone else with equal probability. In contrast, we replicate Owens et al. (2014)’s result that they do so when the alternative is to have the choice of someone else implemented with certainty. This implies that the introduction of random dictatorships in discrete procedures such as those used for the allocation of some public procurement contracts does not necessarily involve aloss of perceived autonomy. |
| Keywords: | control, lotteries, random dictatorship, procurement |
| Date: | 2024–05 |
| URL: | https://d.repec.org/n?u=RePEc:eca:wpaper:2013/374605&r= |
| By: | Larry G Epstein; Kaushil Patel |
| Abstract: | Given only aggregate choice data and limited information about how menus are distributed across the population, we describe what can be inferred robustly about the distribution of preferences (or more general decision rules). We strengthen and generalize existing results on such identification and provide an alternative analytical approach to study the problem. We show further that our model and results are applicable, after suitable reinterpretation, to other contexts. One application is to the robust identification of the distribution of updating rules given only the population distribution of beliefs and limited information about heterogeneous information sources. |
| Date: | 2024–05 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2405.09500&r= |
| By: | Jian-Qiao Zhu; Haijiang Yan; Thomas L. Griffiths |
| Abstract: | The observed similarities in the behavior of humans and Large Language Models (LLMs) have prompted researchers to consider the potential of using LLMs as models of human cognition. However, several significant challenges must be addressed before LLMs can be legitimately regarded as cognitive models. For instance, LLMs are trained on far more data than humans typically encounter, and may have been directly trained on human data in specific cognitive tasks or aligned with human preferences. Consequently, the origins of these behavioral similarities are not well understood. In this paper, we propose a novel way to enhance the utility of LLMs as cognitive models. This approach involves (i) leveraging computationally equivalent tasks that both an LLM and a rational agent need to master for solving a cognitive problem and (ii) examining the specific task distributions required for an LLM to exhibit human-like behaviors. We apply this approach to decision-making -- specifically risky and intertemporal choice -- where the key computationally equivalent task is the arithmetic of expected value calculations. We show that an LLM pretrained on an ecologically valid arithmetic dataset, which we call Arithmetic-GPT, predicts human behavior better than many traditional cognitive models. Pretraining LLMs on ecologically valid arithmetic datasets is sufficient to produce a strong correspondence between these models and human decision-making. Our results also suggest that LLMs used as cognitive models should be carefully investigated via ablation studies of the pretraining data. |
| Date: | 2024–05 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2405.19313&r= |
| By: | Avi Lichtig; Helene Mass |
| Abstract: | We extend the standard disclosure model between a sender and a receiver by allowing the receiver to independently gather partial information, by means of a test – a signal with at most k realizations. The receiver’s choice of test is observed by the sender and therefore influences his decision of whether to disclose. We characterize the optimal test for the receiver and show how it resolves the trade-off between informativeness and disclosure incentives. If the receiver were aiming at maximizing the informativeness, she would choose a deterministic test. In contrast, the optimal test involves randomization over signal realizations and maintains a simple structure. Such a structure allows us to interpret this randomization as the strategic use of uncertain evaluation standards for disclosure incentives. |
| Keywords: | Disclosure, Information Acquisition |
| JEL: | D82 D83 |
| Date: | 2024–05 |
| URL: | http://d.repec.org/n?u=RePEc:bon:boncrc:crctr224_2024_543&r= |
| By: | Hui Li; Jian Ni; Fangzhu Yang |
| Abstract: | The development of Generative AI enables large-scale automation of product design. However, this automated process usually does not incorporate consumer preference information from a company's internal dataset. Meanwhile, external sources such as social media and user-generated content (UGC) websites often contain rich product design and consumer preference information, but such information is not utilized by companies in design generation. We propose a semi-supervised deep generative framework that integrates consumer preferences and external data into product design, allowing companies to generate consumer-preferred designs in a cost-effective and scalable way. We train a predictor model to learn consumer preferences and use predicted popularity levels as additional input labels to guide the training of a Continuous Conditional Generative Adversarial Network (CcGAN). The CcGAN can be instructed to generate new designs of a certain popularity level, enabling companies to efficiently create consumer-preferred designs and save resources by avoiding developing and testing unpopular designs. The framework also incorporates existing product designs and consumer preference information from external sources, which is particularly helpful for small or start-up companies who have limited internal data and face the "cold-start" problem. We apply the proposed framework to a real business setting by helping a large self-aided photography chain in China design new photo templates. We show that our proposed model performs well in generating appealing template designs for the company. |
| Date: | 2024–05 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2405.15929&r= |
| By: | Mark Whitmeyer; Cole Williams |
| Abstract: | For three natural classes of dynamic decision problems; 1. additively separable problems, 2. discounted problems, and 3. discounted problems for a fixed discount factor; we provide necessary and sufficient conditions for one sequential experiment to dominate another in the sense that the dominant experiment is preferred to the other for any decision problem in the specified class. We use these results to study the timing of information arrival in additively separable problems. |
| Date: | 2024–05 |
| URL: | http://d.repec.org/n?u=RePEc:arx:papers:2405.13709&r= |