|
on Discrete Choice Models |
| By: | Easton Huch; Michael Keane |
| Abstract: | Discrete choice models are fundamental tools in management science, economics, and marketing for understanding and predicting decision-making. Logit-based models are dominant in applied work, largely due to their convenient closed-form expressions for choice probabilities. However, these models entail restrictive assumptions on the stochastic utility component, constraining our ability to capture realistic and theoretically grounded choice behavior$-$most notably, substitution patterns. In this work, we propose an amortized inference approach using a neural network emulator to approximate choice probabilities for general error distributions, including those with correlated errors. Our proposal includes a specialized neural network architecture and accompanying training procedures designed to respect the invariance properties of discrete choice models. We provide group-theoretic foundations for the architecture, including a proof of universal approximation given a minimal set of invariant features. Once trained, the emulator enables rapid likelihood evaluation and gradient computation. We use Sobolev training, augmenting the likelihood loss with a gradient-matching penalty so that the emulator learns both choice probabilities and their derivatives. We show that emulator-based maximum likelihood estimators are consistent and asymptotically normal under mild approximation conditions, and we provide sandwich standard errors that remain valid even with imperfect likelihood approximation. Simulations show significant gains over the GHK simulator in accuracy and speed. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.24705 |
| By: | Ting Ji; Laura Liu; Yulong Wang; Jiahe Xing |
| Abstract: | This paper proposes a specification test for the conventional distributional assumptions of error terms in binary choice models, focusing on its tail properties. Based on extreme value theory, we first establish that the tail index of the unobserved error can be recovered by that of the observed covariates. The null hypothesis of the index being zero essentially covers the widely used probit and logit models. We then construct a simple and powerful statistical test for both cross-sectional and panel data, requiring no model estimation and no parametric assumptions. Monte Carlo simulations demonstrate that our test performs well in size and power, and applications to three empirical examples on firm export and innovation decisions and female labor force participation illustrate its general applicability. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.27881 |
| By: | Keaton Ellis; Wanying Huang |
| Abstract: | As AI agents become more autonomous, properly aligning their objectives with human preferences becomes increasingly important. We study how effectively an AI agent learns a human principal's preference in choice under risk via stated versus revealed preferences. We conduct an online experiment in which subjects state their preferences through written instructions ("prompts") and reveal them through choices in a series of binary lottery questions ("data"). We find that on average, an AI agent given revealed-preference data predicts subjects' choices more accurately than an AI agent given stated-preference prompts. Further analysis suggests that the gap is driven by subjects' difficulty in translating their own preferences into written instructions. When given a choice between which information source to give to an AI agent, a large portion of subjects fail to select the more informative one. Moreover, when predictions from the two sources conflict, we find that the AI agent aligns more frequently with the prompt, despite its lower accuracy. Overall, these results highlight the revealed preference approach as a powerful mechanism for communicating human preferences to AI agents, but its success depends on careful implementation. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.29317 |
| By: | Varun Bansal; Mihir Bhattacharya; Ojasvi Khare |
| Abstract: | We study the existence of stable matchings when agents have choice correspondences instead of preference relations. We extend the framework of Chambers and Yenmez (2017) by weakening the Path Independence assumption. For many-to-many markets, we show that stable matchings exist when choice correspondences satisfy Substitutability and a new General Acyclicity condition. We provide a constructive proof using a Grow or Discard Algorithm that iteratively expands or eliminates contracts until a strongly maximal Individually Rational set is reached. We provide an algorithm to obtain stable matchings in which rejected contracts are not permanently discarded, distinguishing our approach significantly from standard DAA-type algorithms. For one-to-one markets, we show that Path Independence alone does not guarantee stability. We introduce a replacement-based notion of stability and provide an algorithm that constructs stable matchings when choice correspondences satisfy Binary Acyclicity. JEL classification: C62, C78, D01, D47 Keywords: choice correspondences, substitutability, general acyclicity, many-to-many matching, matching with contracts, Grow or Discard algorithm, replacement stability, binary acyclicity. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.23038 |
| By: | Jieun Lee; Anil K. Bera |
| Abstract: | We study Cressie Read power divergence (CRPD) estimation for moment based models, focusing on finite sample behavior. While generalized empirical likelihood estimators, dual to CRPD, are known to outperform generalized method of moments estimators in small to moderate samples, the power parameter is typically chosen arbitrarily by the researcher, serving mainly as an index. We interpret it as a hyperparameter that determines the loss function and governs the learning procedure, shaping the curvature of the objective and influencing finite sample performance. Using second order asymptotics, we show that it affects both the structural estimator and the associated Lagrange multipliers, governing robustness, bias, and sensitivity to sampling variation. Monte Carlo simulations illustrate how estimator performance varies with the choice of the power parameter and underlying distributional features, with implications for second order bias and coverage distortion. An empirical illustration based on Owen (2001)s classical example highlights the practical relevance of tuning the power parameter. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.22599 |
| By: | Elchin Suleymanov |
| Abstract: | Human decision makers increasingly delegate choices to AI agents, raising a natural question: does the AI implement the human principal's preferences or pursue its own? To study this question using revealed preference techniques, I introduce the Luce Alignment Model, where the AI's choices are a mixture of two Luce rules, one reflecting the human's preferences and the other the AI's. I show that the AI's alignment (similarity of human and AI preferences) can be generically identified in two settings: the laboratory setting, where both human and AI choices are observed, and the field setting, where only AI choices are observed. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.27868 |
| By: | Alexander Ugarov; Arya Gaduh; Peter McGee |
| Abstract: | We use a laboratory experiment to study preferences over false-positive and false-negative rates of warning signals for an adverse event with a known prior. We find that subjects decrease their demand with signal quality, but less than predicted by our theory. There is asymmetric under-responsiveness by prior: for a low (high) prior, their willingness-to-pay does not fully adjust for the increase in the false-positive (false-negative) costs. We show that neither risk preference nor Bayesian updating skills can fully explain our results. Our results are most consistent with a decision-making heuristic in which subjects do not distinguish between false-positive and false-negative errors. |
| JEL: | C91 D81 D84 D91 |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:34992 |
| By: | Fernando Rios-Avila (Universidad Privada Boliviana); Andrey Ramos (Bank of Spain); Gustavo Canavire-Bacarreza (World Bank and Universidad Privada Boliviana); Leonardo Siles (Universidad de Chile) |
| Abstract: | This paper proposes a method to estimate quantile regression models with multiple fixed effects. We extend the quantile–via–moments estimator of Machado and Santos Silva (2019) and suggest a computationally efficient Frisch–Waugh–Lovell residualization to partial out additive fixed effects in both the location and scale equations. A unified influence-function inference framework is derived, accommodating heteroskedasticity-robust, clustered, and feasible GLS standard errors. Monte Carlo simulations provide strong support for the validity of the proposed procedure in applications with multi-way unobserved heterogeneity and intra-cluster correlated disturbances. An empirical application to Climate Growth-at-Risk illustrates how temperature shocks affect the conditional distribution of macroeconomic outcomes in a panel of 194 countries. Our findings suggest that in low income countries, downside risks to growth are more strongly linked to temperature shocks than the central tendency or upside risks. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:ays:ispwps:paper2615 |
| By: | Diaz Tautiva, Julián Andrés (Universidad Católica De Temuco); Vasquez-Lavin, Felipe; Oliva, Roberto Ponce; Reyes, Carolina Martinez; Gelcich, Stefan |
| Abstract: | This article examines how climate-induced environmental changes, including beach erosion and temporary closures, affect tourist welfare in Chile’s main sun-and-beach destinations. Using the Travel Cost Method, we estimate conservative lower bounds for tourists’ welfare measures to mitigate beach retreat, conserve nearby ecosystems, and maintain beach access. Results show a positive willingness to pay of CLP 868.17 per additional meter of beach width (USD 1.00) and significant welfare losses from potential one-day closures, ranging from CLP 374.39 million (USD 0.43 million) in Coquimbo to CLP 325.24 million (USD 0.37 million) in Valparaíso. Tourists value natural features but react negatively to visible infrastructure such as drainage systems. These findings offer practical insights for climate adaptation and sustainable coastal management in tourism-dependent regions. |
| Date: | 2026–03–25 |
| URL: | https://d.repec.org/n?u=RePEc:osf:socarx:78dfg_v1 |
| By: | Anke Becker; Christina Borner; Thomas Dohmen; Armin Falk; David B. Huffman; Uwe Sunde |
| Abstract: | A growing body of empirical research has developed measures of economic preferences related to risk taking and intertemporal choice. This research has documented pronounced heterogeneity in preferences across and within societies, and also provided evidence that these differences are culturally transmitted. This chapter discusses existing data sets that allow for a comparable measurement across the globe, takes stock of commonalities and differences in approaches, and presents an extended synthetic cross-country data set that combines information from existing data sets. The analysis then establishes various empirical regularities, such as broadly similar patterns of heterogeneity across the globe, revealed by the different datasets, but also some systematic divergences by measurement approach, and substantial correlations of economic preferences with country-aggregate and individual-level outcomes and traits. We also briefly discuss international data sets measuring social preferences, and end with an outlook on avenues for future research. |
| Keywords: | willingness to take risks, patience |
| JEL: | D1 |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_12589 |
| By: | Ezzaaime Youness (Université Hassan 1er [Settat]); Maizzou Said (Université Hassan 1er [Settat]); Abdeljabbar Abdouni (Université Hassan 1er [Settat]) |
| Abstract: | Youth employment constitutes one of the most persistent structural challenges confronting North African economies. Against this backdrop, the present study aims to identify and quantify the socioeconomic and demographic factors shaping the labor market position of individuals aged 15 to 35 in Algeria, Egypt, and Morocco. The dataset consists of a population-weighted aggregate contingency table constructed for the year 2026, covering a population in excess of 65 million individuals. Since the dependent variable takes four mutually exclusive values employed, unemployed, inactive, and student, a Multinomial Logit model provides the most appropriate estimation framework for capturing the full complexity of youth labor market outcomes. Drawing on this econometric framework, the analysis yields several key findings. Most prominently, gender emerges as the dominant predictor of labor market status: women are substantially overrepresented among inactive individuals and students, reflecting the enduring influence of sociocultural norms across these economies. Beyond gender, age produces a markedly nonlinear effect, with employment transitions becoming significantly more pronounced beyond the age of 25, thereby pointing to a critical threshold in youth professional trajectories. These individual-level determinants are further compounded by significant sectoral disparities with service industries absorbing the vast majority of formal employment as well as by country fixed effects that reveal meaningful structural differences across the three economies under study. The quantification of these effects lends greater analytical precision to the above findings. Average Marginal Effects indicate that being female reduces the probability of employment by 35.2 percentage points (pp), all else equal, while simultaneously increasing the probability of inactivity by 18.4 pp and of being a student by 22.7 pp. Regarding the age effect, the inflection point of the employment transition is statistically located around ages 22–24. In terms of cross-country disparities, Egypt exhibits a significantly more favorable employment profile than Algeria, with a conditional employment probability 15.4 pp higher. On the robustness front, the Hausman-McFadden test validates the Independence of Irrelevant Alternatives assumption (χ² (6) = 8.34; p = 0.214), thereby confirming the reliability of the model specification. Taken together, these results carry direct implications for the design of gender-sensitive and age-targeted employment policies across the region. |
| Abstract: | La question de l'accès à l'emploi pour les nouvelles générations représente l'un des enjeux structurels les plus aigus des économies nord-africaines. C'est dans ce contexte que la présente étude se propose d'identifier et de quantifier les facteurs socioéconomiques et démographiques qui conditionnent le positionnement des jeunes âgés de 15 à 35 ans sur le marché du travail en Algérie, en Égypte et au Maroc. Les données mobilisées consistent en une table de contingence agrégée pondérée, construite pour l'année 2026 et représentant une population de plus de 65 millions d'individus. La variable à expliquer étant catégorielle à quatre modalités emploi, chômage, inactivité et scolarisation, la régression logistique polytomique (Logit Multinomial) constitue le cadre d'estimation le mieux adapté à cette structure de données. L'exploitation de ce cadre économétrique permet de dégager plusieurs résultats saillants. En premier lieu, l'appartenance au genre féminin s'impose comme le principal facteur prédictif du statut d'activité : les femmes sont fortement surreprésentées parmi les inactifs et les étudiants, témoignant de la prégnance de normes sociales persistantes dans ces économies. En second lieu, l'âge exerce un effet de nature non linéaire, la probabilité d'accès à l'emploi augmentant sensiblement après le cap des 25 ans, ce qui signale l'existence d'un seuil critique dans les trajectoires d'insertion professionnelle des jeunes. À ces deux déterminants individuels s'ajoutent des asymétries sectorielles significatives le tertiaire concentrant la grande majorité des emplois formels ainsi que des effets fixes pays qui révèlent des spécificités structurelles marquées entre les trois économies étudiées. La quantification précise de ces effets renforce la portée analytique de ces constats. Les effets marginaux moyens indiquent ainsi qu'être une femme réduit la probabilité d'emploi de 35, 2 points de pourcentage (pp), toutes choses égales par ailleurs, et augmente simultanément la probabilité d'inactivité de 18, 4 pp et de scolarisation de 22, 7 pp. S'agissant de l'effet d'âge, le point d'inflexion de la transition vers l'emploi est statistiquement localisé aux alentours de 22–24 ans. Du point de vue des disparités entre pays, l'Égypte présente un profil d'insertion significativement plus favorable que l'Algérie, avec une probabilité d'emploi supérieure de 15, 4 pp. Sur le plan de la robustesse, le test de Hausman-McFadden valide l'hypothèse d'Indépendance des Alternatives Non Pertinentes (χ²(6) = 8, 34 ; p = 0, 214), confirmant la solidité du modèle retenu. Pris ensemble, ces résultats offrent des implications directes pour la conception de politiques d'emploi ciblées sur le genre et les cohortes d'âge critique dans la région. |
| Keywords: | North Africa, gender, inactivity, Labor market segmentation, Microeconometrics, youth employment, Multinomial Logit, microeconometrics, labor market segmentation, Multinomial Logit youth employment North Africa gender inactivity labor market segmentation microeconometrics, microéconométrie, segmentation du marché du travail, inactivité, genre, Afrique du Nord, emploi des jeunes, Logit Multinomial |
| Date: | 2026 |
| URL: | https://d.repec.org/n?u=RePEc:hal:journl:hal-05553120 |
| By: | Victor H. Aguiar; Nail Kashaev |
| Abstract: | Modern pretrained time-series foundation models can forecast without task-specific training, but they do not fully incorporate economic behavior. We show that teaching them basic economic logic improves how they predict demand using an experimental panel. We fine-tune Amazon Chronos-2, a transformer-based probabilistic time-series model, on synthetic data generated from utility-maximizing agents. We exploit Afriat's theorem, which guarantees that demand satisfies the Generalized Axiom of Revealed Preference (GARP) if and only if it can be generated by maximizing some utility function subject to a budget constraint. GARP is a simple condition to check that allows us to generate time series from a large class of utilities efficiently. The fine-tuned model serves as a rationality-constrained forecasting prior: it learns price-quantity relations from GARP-consistent synthetic histories and then uses those relations to predict the choices of real consumers. We find that fine-tuning on GARP-consistent synthetic data substantially improves prediction relative to zero-shot Chronos-2 at all forecast horizons we study. Our results show that economic theory can be used to generate structured synthetic data that improves foundation-model predictions when the theory implies observable patterns in the data. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.23993 |
| By: | Wayne Gao |
| Abstract: | Normalization is ubiquitous in economics, and a growing literature shows that ``normalizations'' can matter for interpretation, counterfactual analysis, misspecification, and inference. This paper provides a general framework for these issues, based on the formalized notion of modeling equivalence that partitions the space of unknowns into equivalence classes, and defines normalization as a WLOG selection of one representative from each class. A counterfactual parameter is normalization-free if and only if it is constant on equivalence classes; otherwise any point identification is created by the normalization rather than by the model. Applications to discrete choice, demand estimation, and network formation illustrate the insights made explicit through this criterion. We then study two further sources of fragility: an extension trilemma establishes that fidelity, invariance, and regularity cannot simultaneously hold at a boundary singularity, while a normalization can itself introduce a coordinate singularity that distorts the topological and metric structures of the parameter space, with consequences for estimation and inference. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.27762 |
| By: | Joseph Marshall |
| Abstract: | This paper studies how to estimate an individual's taste for forming a connection with another individual in a network. It compares the difficulty of estimation with and without the assumption that utility is transferable between individuals, and with and without the assumption that regressors are symmetric across individuals in the pair. I show that when pair-specific regressors are symmetric, the sufficient conditions for consistency and asymptotic normality of the maximum likelihood estimator that assumes transferable utility (TU-MLE) are also sufficient for the maximum likelihood estimator that does not assume transferable utility (NTU-MLE). When regressors are asymmetric, I provide sufficient conditions for the consistency and asymptotic normality of the NTU-MLE. I also provide a specification test to assess the validity of the transferable utility assumption. Two applications from different fields of economics demonstrate the value of my results. I find evidence of researchers using the TU-MLE when the transferable utility assumption is violated, and evidence of researchers using NTU-model-based estimators when the validity of the transferable utility assumption cannot be rejected. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.25641 |
| By: | Fally, Thibault; Ligon, Ethan |
| Keywords: | Social and Behavioral Sciences |
| Date: | 2026–03–27 |
| URL: | https://d.repec.org/n?u=RePEc:cdl:agrebk:qt6kr0q483 |
| By: | Chihiro Inoue; Yusuke Ishihata; Suguru Otani |
| Abstract: | We study marital sorting using a novel dataset from a marriage matching platform, which uniquely records a rich set of pre-marital attributes, including preferences for children and for the division of housework and childcare. Unlike census or post-marital surveys, all characteristics are collected prior to matching and validated using official documents, yielding clean measures of preferences uncontaminated by post-marital coordination. Applying a multidimensional matching framework to twelve attributes, we find strong positive assortative matching across all dimensions. Age is the most salient trait, but preferences for children are the second most important - exceeding education - a pattern largely invisible in standard data. Preference measures play a distinct role in the matching process: they exhibit limited cross-attribute interactions with sociodemographic and anthropometric characteristics, in contrast to the pervasive interactions among those attributes. A low-dimensional factor representation shows that preferences for children constitute a separate and salient margin of sorting. Using the staged structure of the platform, we further show that assortative matching along different dimensions emerges at distinct points in the dating process: sorting by age and income is already present at the initial Application stage, whereas sorting by preferences for children becomes robust only at later stages of relationship formation, reflecting selective continuation rather than sorting at the point of final agreement. A simple theoretical exercise demonstrates that ignoring preference-based sorting and assuming homogeneous preferences across couples leads to biased estimates of policy effects on subsequent household decisions. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.25372 |
| By: | Alaina Barca; Evan Mast |
| Abstract: | We construct a new neighborhood geography using a revealed preference intuition: If people disproportionately move within neighborhoods, their boundaries can be backed out from migration flows. Our “districts, ” which consist of about nine census tracts each, correspond to recognizable local areas, as their boundaries align with physical barriers, sharp demographic changes, and local government borders. To illustrate applications, we first show that tract-level analyses of neighborhood sorting miss important broader patterns. Second, aggregating tract-level intergenerational mobility estimates to the district level increases precision threefold while introducing little aggregation bias, resulting in improved predictive power in a hold-out sample. |
| Keywords: | Neighborhood definition; residential mobility; residential sorting |
| JEL: | R23 |
| Date: | 2026–03–24 |
| URL: | https://d.repec.org/n?u=RePEc:fip:fedpwp:102934 |
| By: | Edvard Bakhitov |
| Abstract: | This paper develops a penalized GMM (PGMM) framework for automatic debiased inference on functionals of nonparametric instrumental variable estimators. We derive convergence rates for the PGMM estimator and provide conditions for root-n consistency and asymptotic normality of debiased functional estimates, covering both linear and nonlinear functionals. Monte Carlo experiments on average derivative show that the PGMM-based debiased estimator performs on par with the analytical debiased estimator that uses the known closed-form Riesz representer, achieving 90-96% coverage while the plug-in estimator falls below 5%. We apply our procedure to estimate mean own-price elasticities in a semiparametric demand model for differentiated products. Simulations confirm near-nominal coverage while the plug-in severely undercovers. Applied to IRI scanner data on carbonated beverages, debiased semiparametric estimates are approximately 20% more elastic compared to the logit benchmark, and debiasing corrections are heterogeneous across products, ranging from negligible to several times the standard error. |
| Date: | 2026–03 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2603.29889 |