arXiv subjects

Ran Spiegler

Publications and source records attributed to Ran Spiegler.

15 recordsLinked to original sources

arXiv2026

Cross-Validation Equilibrium

We study strategic interaction when players delegate belief formation to predictive machine learning (ML). In a static Bayesian game, each player's ML agent predicts a payoff-relevant outcome variable as a function of the player's type. The ML agent's training sample is endogenous: it is drawn from the outcome distribution generated by players' ML-guided behavior. In Cross-Validation Equilibrium (CVE), each player's ML agent selects a predictive model to minimize expected out-of-sample squared error, given its realized training sample, and each player best-replies to the belief generated by the model her ML agent selected. We analyze CVE and relate it to other equilibrium concepts. We apply CVE to jury voting, speculative betting, and games with linear-quadratic payoffs. E.g., in a team-effort game, endogenous model selection can give rise to multiple equilibria.

econ.TH↗

arXiv2026

Limited belief propagation and contingent thinking

An agent updates her beliefs over a set of variables after observing some of them. We provide a representation of updated beliefs that captures limited propagation of her observation's implications through the directed acyclic graph that represents the relations between all variables. Failure of contingent thinking occurs when she performs fewer inference steps from unobserved variables than observed ones, leading to correlation neglect and violations of iterated expectations. Our framework offers a new perspective on existing experiments about contingent thinking and suggests new directions. We characterize the model's relationship with familiar Bayesian and non-Bayesian benchmarks, and illustrate it with applications to public-good provision and social learning games.

econ.TH↗

arXiv2026

Whataboutism

We propose a model of whataboutism -- a rhetorical strategy that deflects criticism by citing similar misconduct that goes uncriticized on the critic's side -- and study its implications for social norms governing offensive speech. In an infinite-horizon psychological game with two rival camps, agents weigh the intrinsic benefit of offensive speech against the risk of condemnation. External criticism can be deflected via an equilibrium-based whataboutism rebuttal. We characterize the unique dynamically stable Psychological Subgame Perfect Equilibrium and show that the availability of whataboutism exacerbates offensive speech, to the extent that civility norms can break down entirely, especially in polarized societies.

econ.TH↗

arXiv2025

Monopolistic Data Dumping

A profit-maximizing monopolist curates a database for users seeking to learn a parameter. There are two user types: "Nowcasters" wish to learn the parameter's current value, while "forecasters" target its long-run value. Data storage involves a constant marginal cost. The monopolist designs a menu of contracts described by fees and data-access levels. The profit-maximizing menu offers full access to historical data, while current data is fully provided to nowcasters but may be withheld from forecasters. Compared to the social optimum, the monopolist keeps too much historical data, too little current data, and may store too much data overall.

econ.TH↗

arXiv2025

Machine-Learning to Trust

Can players sustain long-run trust when their equilibrium beliefs are shaped by machine-learning methods that penalize complexity? I study a game in which an infinite sequence of agents with one-period recall decides whether to place trust in their immediate successor. The cost of trusting is state-dependent. Each player's best response is based on a belief about others' behavior, which is a coarse fit of the true population strategy with respect to a partition of relevant contingencies. In equilibrium, this partition minimizes the sum of the mean squared prediction error and a complexity penalty proportional to its size. Relative to symmetric mixed-strategy Nash equilibrium, this solution concept significantly narrows the scope for trust.

econ.TH↗

arXiv2024

News Media as Suppliers of Narratives (and Information)

We present a model of news media that shape consumer beliefs by providing information (signals about an exogenous state) and narratives (models of what determines outcomes). To amplify consumers' engagement, media maximize consumers' anticipatory utility. Focusing on a class of separable consumer preferences, we show that a monopolistic media platform facing homogenous consumers provides a false "empowering" narrative coupled with an optimistically biased signal. Consumer heterogeneity gives rise to a novel menu-design problem due to a "data externality" among consumers. The optimal menu features multiple narratives and creates polarized beliefs. These effects also arise in a competitive media market model.

econ.TH↗

arXiv2024

Identifying Assumptions and Research Dynamics

A representative researcher has repeated opportunities for empirical research. To process findings, she must impose an "identifying assumption." She conducts research when the assumption is sufficiently plausible (taking into account both current beliefs and the quality of the opportunity), and updates beliefs as if the assumption were perfectly valid. We study the dynamics of this learning process. While the rate of research cannot always increase over time, research slowdown is possible. We characterize environments in which the rate is constant. Long-run beliefs can exhibit history-dependence and "false certitude." We apply the model to stylized examples of empirical methodologies: experiments, various causal-inference techniques, and "calibration."

econ.TH↗

arXiv2023

Behavioral Causal Inference

When inferring the causal effect of one variable on another from correlational data, a common practice by professional researchers as well as lay decision makers is to control for some set of exogenous confounding variables. Choosing an inappropriate set of control variables can lead to erroneous causal inferences. This paper presents a model of lay decision makers who use long-run observational data to learn the causal effect of their actions on a payoff-relevant outcome. Different types of decision makers use different sets of control variables. I obtain upper bounds on the equilibrium welfare loss due to wrong causal inferences, for various families of data-generating processes. The bounds depend on the structure of the type space. When types are "ordered" in a certain sense, the equilibrium condition greatly reduces the cost of wrong causal inference due to poor controls.

econ.TH↗

arXiv2022

False Narratives and Political Mobilization

We present an equilibrium model of politics in which political platforms compete over public opinion. A platform consists of a policy, a coalition of social groups with diverse intrinsic attitudes to policies, and a narrative. We conceptualize narratives as subjective models that attribute a commonly valued outcome to (potentially spurious) postulated causes. When quantified against empirical observations, these models generate a shared belief among coalition members over the outcome as a function of its postulated causes. The intensity of this belief and the members' intrinsic attitudes to the policy determine the strength of the coalition's mobilization. Only platforms that generate maximal mobilization prevail in equilibrium. Our equilibrium characterization demonstrates how false narratives can be detrimental for the common good, and how political fragmentation leads to their proliferation. The false narratives that emerge in equilibrium attribute good outcomes to the exclusion of social groups from ruling coalitions.

econ.TH↗

arXiv2021

On the Behavioral Consequences of Reverse Causality

Reverse causality is a common causal misperception that distorts the evaluation of private actions and public policies. This paper explores the implications of this error when a decision maker acts on it and therefore affects the very statistical regularities from which he draws faulty inferences. Using a quadratic-normal parameterization and applying the Bayesian-network approach of Spiegler (2016), I demonstrate the subtle equilibrium effects of a certain class of reverse-causality errors, with illustrations in diverse areas: development psychology, social policy, monetary economics and IO. In particular, the decision context may protect the decision maker from his own reverse-causality causal error. That is, the cost of reverse-causality errors can be lower for everyday decision makers than for an outside observer who evaluates their choices.

econ.TH↗

arXiv2021

A Simple Model of Monetary Policy under Phillips-Curve Causal Disagreements

I study a static textbook model of monetary policy and relax the conventional assumption that the private sector has rational expectations. Instead, the private sector forms inflation forecasts according to a misspecified subjective model that disagrees with the central bank's (true) model over the causal underpinnings of the Phillips Curve. Following the AI/Statistics literature on Bayesian Networks, I represent the private sector's model by a direct acyclic graph (DAG). I show that when the private sector's model reverses the direction of causality between inflation and output, the central bank's optimal policy can exhibit an attenuation effect that is sensitive to the noisiness of the true inflation-output equations.

econ.TH↗

arXiv2021

Anabolic Persuasion

We present a model of optimal training of a rational, sluggish agent. A trainer commits to a discrete-time, finite-state Markov process that governs the evolution of training intensity. Subsequently, the agent monitors the state and adjusts his capacity at every period. Adjustments are incremental: the agent's capacity can only change by one unit at a time. The trainer's objective is to maximize the agent's capacity - evaluated according to its lowest value under the invariant distribution - subject to an upper bound on average training intensity. We characterize the trainer's optimal policy, and show how stochastic, time-varying training intensity can dramatically increase the long-run capacity of a rational, sluggish agent. We relate our theoretical findings to "periodization" training techniques in exercise physiology.

econ.TH↗

arXiv2019

Cheating with (Recursive) Models

To what extent can agents with misspecified subjective models predict false correlations? We study an "analyst" who utilizes models that take the form of a recursive system of linear regression equations. The analyst fits each equation to minimize the sum of squared errors against an arbitrarily large sample. We characterize the maximal pairwise correlation that the analyst can predict given a generic objective covariance matrix, subject to the constraint that the estimated model does not distort the mean and variance of individual variables. We show that as the number of variables in the model grows, the false pairwise correlation can become arbitrarily close to one, regardless of the true correlation.

econ.TH↗

arXiv2018

A Model of Competing Narratives

We formalize the argument that political disagreements can be traced to a "clash of narratives". Drawing on the "Bayesian Networks" literature, we model a narrative as a causal model that maps actions into consequences, weaving a selection of other random variables into the story. An equilibrium is defined as a probability distribution over narrative-policy pairs that maximizes a representative agent's anticipatory utility, capturing the idea that public opinion favors hopeful narratives. Our equilibrium analysis sheds light on the structure of prevailing narratives, the variables they involve, the policies they sustain and their contribution to political polarization.

econ.TH↗

arXiv2018

The Model Selection Curse

A "statistician" takes an action on behalf of an agent, based on the agent's self-reported personal data and a sample involving other people. The action that he takes is an estimated function of the agent's report. The estimation procedure involves model selection. We ask the following question: Is truth-telling optimal for the agent given the statistician's procedure? We analyze this question in the context of a simple example that highlights the role of model selection. We suggest that our simple exercise may have implications for the broader issue of human interaction with "machine learning" algorithms.

econ.TH↗