arXiv subjects

Susan Athey

Publications and source records attributed to Susan Athey.

At least 19 recordsLinked to original sources

arXiv2026

Estimating Causal Effects from Data Generated by Stochastic Algorithms

Recommendation systems and chatbots present content to users, typically using stochastic algorithms that select the content based on user characteristics or context. Examples of content include chat responses, videos, or items available for purchase. Scientists and application developers are often interested in whether characteristics of content increase outcomes such as user engagement. Estimates of such causal effects may guide content providers to generate content that emphasize desirable features. However, in settings with a large content library or where content is generated uniquely for a given user, it can be difficult to use observational data to learn the causal effect of content features, because the content a user sees is tailored to that user, and because content varies in many dimensions. This paper proposes a new method for estimating the impact of content features using observational data, when the algorithm that determines user exposure incorporates some randomization, and when two additional data elements are logged for each user: $(i)$ the identity of at least one item that could have been exposed to the user, but was not (the unexposed item); $(ii)$ an estimate of the ratio of the probability that the unexposed item would have been shown to the probability that the exposed item was shown. We show that causal effects of features are identified in this setting, even in the presence of unobserved confounders that affect both user preferences and the identity of the considered pair of items (exposed and unexposed). Our estimator differs from prior approaches in terms of what data is used and how the estimator is constructed.

stat.ME↗

arXiv2026

TERMS-Bench: Diagnosing LLM Negotiation Agents Beyond Deal Rate

Negotiation is a central mechanism of economic exchange, shaping markets, procurement, labor agreements, and resource allocation. It is also a canonical testbed for agentic language models, requiring multi-turn interaction under hidden preferences, strategic communication, and binding constraints. These properties make negotiation hard to evaluate: unlike math or code, it has no intrinsic verifier. Existing LLM negotiation evaluations rely on LLM-vs.-LLM interaction or aggregate outcomes such as deal rate, leaving failures opaque. We introduce Terms-Bench, short for Testbed for Economic Reasoning in Multi-turn Strategy, a Bayesian-game framework that makes the environment itself the verifier by specifying the counterpart's latent type, policy, and payoff structure. We instantiate it in bilateral price negotiation, where the counterpart's private state and simulator policy are hidden from the agent but observable to the evaluator. This turns the counterpart from a black-box opponent into a diagnostic instrument, enabling agent-attributable failure analysis and oracle-reference optimality gaps. Evaluating 13 LLM agents spanning frontier systems from major providers, Terms-Bench turns negotiation evaluation from aggregate ranking into actionable diagnosis: where agents fail, why they fail, and what to strengthen. Empirically, frontier models saturate deal rate yet diverge in surplus extraction, cue use, belief calibration, and compliance, revealing agent-specific bargaining bottlenecks masked by prior benchmarks.

cs.GT↗

arXiv2025

Cross-Validated Causal Inference: a Modern Method to Combine Experimental and Observational Data

We develop new methods to integrate experimental and observational data in causal inference. While randomized controlled trials offer strong internal validity, they are often costly and therefore limited in sample size. Observational data, though cheaper and often with larger sample sizes, are prone to biases due to unmeasured confounders. To harness their complementary strengths, we propose a systematic framework that formulates causal estimation as an empirical risk minimization (ERM) problem. A full model containing the causal parameter is obtained by minimizing a weighted combination of experimental and observational losses--capturing the causal parameter's validity and the full model's fit, respectively. The weight is chosen through cross-validation on the causal parameter across experimental folds. Our experiments on real and synthetic data show the efficacy and reliability of our method. We also provide theoretical non-asymptotic error bounds.

econ.EM↗

arXiv2025

Estimating Variances for Causal Panel Data Estimators

There has been a recent surge in research on causal panel data models, leading to many new estimators for average causal effects. However, researchers have paid less attention to quantifying the precision of these estimators. This paper addresses that gap by studying the problem of variance estimation in causal panel settings. We develop a unified framework for comparing the three main variance estimators used in these settings: regression-based, Unit-Placebo, and Time-Placebo estimators. We show that each relies on a distinct exchangeability assumption and, correspondingly, each targets a different conditional variance. We find that, under some assumptions, all three estimators are all valid, but that their statistical power differs substantially depending on the heteroskedasticity present in the data. Building on these insights, we propose a new variance estimator that flexibly accounts for heteroskedasticity across the unit and time dimensions, and delivers superior statistical power in realistic panel data settings.

econ.EM↗

arXiv2025

Triply Robust Panel Estimators

This paper studies estimation of causal effects in a panel data setting. We introduce a new estimator, the Triply RObust Panel (TROP) estimator, that combines (i) a flexible model for the potential outcomes based on a low-rank factor structure on top of a two-way-fixed effect specification, with (ii) unit weights intended to upweight units similar to the treated units and (iii) time weights intended to upweight time periods close to the treated time periods. We study the performance of the estimator in a set of simulations designed to closely match several commonly studied real data sets. We find that there is substantial variation in the performance of the estimators across the settings considered. The proposed estimator outperforms two-way-fixed-effect/difference-in-differences, synthetic control, matrix completion and synthetic-difference-in-differences estimators. We investigate what features of the data generating process lead to this performance, and assess the relative importance of the three components of the proposed estimator. We have two recommendations. Our preferred strategy is that researchers use simulations closely matched to the data they are interested in, along the lines discussed in this paper, to investigate which estimators work well in their particular setting. A simpler approach is to use more robust estimators such as synthetic difference-in-differences or the new triply robust panel estimator which we find to substantially outperform two-way fixed effect estimators in many empirically relevant settings.

stat.ME↗

arXiv2025

Identification of Average Treatment Effects in Nonparametric Panel Models

This paper studies identification of average treatment effects in a panel data setting. It introduces a novel nonparametric factor model and proves identification of average treatment effects. The identification proof is based on the introduction of a consistent estimator. Underlying the proof is a result that there is a consistent estimator for the expected outcome in the absence of the treatment for each unit and time period; this result can be applied more broadly, for example in problems of decompositions of group-level differences in outcomes, such as the much-studied gender wage gap.

econ.EM↗

arXiv2024

Robust Offline Policy Learning with Observational Data from Multiple Sources

We consider the problem of using observational bandit feedback data from multiple heterogeneous data sources to learn a personalized decision policy that robustly generalizes across diverse target settings. To achieve this, we propose a minimax regret optimization objective to ensure uniformly low regret under general mixtures of the source distributions. We develop a policy learning algorithm tailored to this objective, combining doubly robust offline policy evaluation techniques and no-regret learning algorithms for minimax optimization. Our regret analysis shows that this approach achieves the minimal worst-case mixture regret up to a moderated vanishing rate of the total data across all sources. Our analysis, extensions, and experimental results demonstrate the benefits of this approach for learning robust decision policies from multiple data sources.

cs.LG↗

arXiv2024

Estimating Wage Disparities Using Foundation Models

The rise of foundation models marks a paradigm shift in machine learning: instead of training specialized models from scratch, foundation models are first trained on massive datasets before being adapted or fine-tuned to make predictions on smaller datasets. Initially developed for text, foundation models have also excelled at making predictions about social science data. However, while many estimation problems in the social sciences use prediction as an intermediate step, they ultimately require different criteria for success. In this paper, we develop methods for fine-tuning foundation models to perform these estimation problems. We first characterize an omitted variable bias that can arise when a foundation model is only fine-tuned to maximize predictive accuracy. We then provide a novel set of conditions for fine-tuning under which estimates derived from a foundation model are root-n-consistent. Based on this theory, we develop new fine-tuning algorithms that empirically mitigate this omitted variable bias. To demonstrate our ideas, we study gender wage decomposition. This is a statistical estimation problem from econometrics where the goal is to decompose the gender wage gap into components that can and cannot be explained by career histories of workers. Classical methods for decomposing the wage gap employ simple predictive models of wages which condition on coarse summaries of career history that may omit factors that are important for explaining the gap. Instead, we use a custom-built foundation model to decompose the gender wage gap, which captures a richer representation of career history. Using data from the Panel Study of Income Dynamics, we find that career history explains more of the gender wage gap than standard econometric models can measure, and we identify elements of career history that are omitted by standard models but are important for explaining the wage gap.

cs.LG↗

arXiv2024

LABOR-LLM: Language-Based Occupational Representations with Large Language Models

This paper builds an empirical model that predicts a worker's next occupation as a function of the worker's occupational history. Because histories are sequences of occupations, the covariate space is high-dimensional, and further, the outcome (the next occupation) is a discrete choice that can take on many values. To estimate the parameters of the model, we leverage an approach from generative artificial intelligence. Estimation begins from a ``foundation model'' trained on non-representative data and then ``fine-tunes'' the estimation using data about careers from a representative survey. We convert tabular data from the survey into text files that resemble resumes and fine-tune the parameters of the foundation model, a large language model (LLM), using these text files with the objective of predicting the next token (word). The resulting fine-tuned LLM is used to calculate estimates of worker transition probabilities. Its predictive performance surpasses all prior models, both for the task of granularly predicting the next occupation as well as for specific tasks such as predicting whether the worker changes occupations or stays in the labor force. We quantify the value of fine-tuning and further show that by adding more career data from a different population, fine-tuning smaller LLMs (fewer parameters) surpasses the performance of fine-tuning larger models. When we omit the English language occupational title and replace it with a unique code, predictive performance declines.

cs.LG↗

arXiv2024

Data-driven Error Estimation: Excess Risk Bounds without Class Complexity as Input

Constructing confidence intervals that are simultaneously valid across a class of estimates is central to tasks such as multiple mean estimation, generalization guarantees, and adaptive experimental design. We frame this as an ``error estimation problem," where the goal is to determine a high-probability upper bound on the maximum error for a class of estimates. We propose an entirely data-driven approach that derives such bounds for both finite and infinite class settings, naturally adapting to a potentially unknown correlation structure of random errors. Notably, our method does not require class complexity as an input, overcoming a major limitation of existing approaches. We present our simple yet general solution and demonstrate applications to simultaneous confidence intervals, excess-risk control and optimizing exploration in contextual bandit algorithms.

cs.LG↗

arXiv2024

The Value of Non-Traditional Credentials in the Labor Market

Workers without formal credentials experience substantially lower employment rates than their credentialed counterparts, but the extent to which information frictions contribute to these disparities remains unclear. We conducted a randomized experiment with over 800,000 online certificate earners from developing countries who lack college degrees, encouraging credential sharing on LinkedIn through reduced friction and reminders. We study credential visibility for the full sample and track employment outcomes for 40,000 learners. The intervention increased new employment by 6% (1.0 percentage point), with larger effects of 9% (1.2 percentage points) for jobs related to certificates. Treatment effects concentrate among learners with weak baseline employability: those in the bottom tercile experience employment gains of 12% while those in the top tercile see negligible effects. Those most responsive to the intervention benefit most from sharing, with a local average treatment effect of 11 percentage points for compliers. Counterfactual resume scoring shows that credentials improve perceived candidate quality by 8.5 points on average (on a scale from 1 to 100), with effects of 15-20 points for weak resumes but near-zero for strong resumes. Our findings demonstrate that information frictions substantially constrain employment for workers without traditional credentials, and that minimal-cost interventions targeting credential visibility can generate employment gains comparable to intensive training programs.

econ.GN↗

arXiv2023

Digital interventions and habit formation in educational technology

As online educational technology products have become increasingly prevalent, rich evidence indicates that learners often find it challenging to establish regular learning habits and complete their programs. Concurrently, online products geared towards entertainment and social interactions are sometimes so effective in increasing user engagement and creating frequent usage habits that they inadvertently lead to digital addiction, especially among youth. In this project, we carry out a contest-based intervention, common in the entertainment context, on an educational app for Indian children learning English. Approximately ten thousand randomly selected learners entered a 100-day reading contest. They would win a set of physical books if they ranked sufficiently high on a leaderboard based on the amount of educational content consumed. Twelve weeks after the end of the contest, when the treatment group had no additional incentives to use the app, they continued their engagement with it at a rate 75\% higher than the control group, indicating a successful formation of a reading habit. In addition, we observed a 6\% increase in retention within the treatment group. These results underscore the potential of digital interventions in fostering positive engagement habits with educational technology products, ultimately enhancing users' long-term learning outcomes.

econ.GN↗

arXiv2023

Machine Learning Who to Nudge: Causal vs Predictive Targeting in a Field Experiment on Student Financial Aid Renewal

In many settings, interventions may be more effective for some individuals than others, so that targeting interventions may be beneficial. We analyze the value of targeting in the context of a large-scale field experiment with over 53,000 college students, where the goal was to use "nudges" to encourage students to renew their financial-aid applications before a non-binding deadline. We begin with baseline approaches to targeting. First, we target based on a causal forest that estimates heterogeneous treatment effects and then assigns students to treatment according to those estimated to have the highest treatment effects. Next, we evaluate two alternative targeting policies, one targeting students with low predicted probability of renewing financial aid in the absence of the treatment, the other targeting those with high probability. The predicted baseline outcome is not the ideal criterion for targeting, nor is it a priori clear whether to prioritize low, high, or intermediate predicted probability. Nonetheless, targeting on low baseline outcomes is common in practice, for example because the relationship between individual characteristics and treatment effects is often difficult or impossible to estimate with historical data. We propose hybrid approaches that incorporate the strengths of both predictive approaches (accurate estimation) and causal approaches (correct criterion); we show that targeting intermediate baseline outcomes is most effective in our specific application, while targeting based on low baseline outcomes is detrimental. In one year of the experiment, nudging all students improved early filing by an average of 6.4 percentage points over a baseline average of 37% filing, and we estimate that targeting half of the students using our preferred policy attains around 75% of this benefit.

econ.EM↗

arXiv2023

The Heterogeneous Earnings Impact of Job Loss Across Workers, Establishments, and Markets

Using rich Swedish administrative data, we apply causal machine learning methods to study how earnings losses after job displacement vary with observable characteristics that may be relevant for targeting policy interventions for workers. Heterogeneity in effects is as large within as across worker groups defined by age and schooling, and as large within as across establishments. A substantial portion of cross-establishment heterogeneity can be explained by industry and local labor market characteristics, suggesting a role for place- and industry-based targeting. The largest losses are concentrated among already vulnerable workers, indicating that well-designed targeting policies can improve both efficiency and equity.

econ.GN↗

arXiv2023

Proportional Response: Contextual Bandits for Simple and Cumulative Regret Minimization

In many applications, e.g. in healthcare and e-commerce, the goal of a contextual bandit may be to learn an optimal treatment assignment policy at the end of the experiment. That is, to minimize simple regret. However, this objective remains understudied. We propose a new family of computationally efficient bandit algorithms for the stochastic contextual bandit setting, where a tuning parameter determines the weight placed on cumulative regret minimization (where we establish near-optimal minimax guarantees) versus simple regret minimization (where we establish state-of-the-art guarantees). Our algorithms work with any function class, are robust to model misspecification, and can be used in continuous arm settings. This flexibility comes from constructing and relying on "conformal arm sets" (CASs). CASs provide a set of arms for every context, encompassing the context-specific optimal arm with a certain probability across the context distribution. Our positive results on simple and cumulative regret guarantees are contrasted with a negative result, which shows that no algorithm can achieve instance-dependent simple regret guarantees while simultaneously achieving minimax optimal cumulative regret guarantees.

cs.LG↗

arXiv2023

Qini Curves for Multi-Armed Treatment Rules

Qini curves have emerged as an attractive and popular approach for evaluating the benefit of data-driven targeting rules for treatment allocation. We propose a generalization of the Qini curve to multiple costly treatment arms, that quantifies the value of optimally selecting among both units and treatment arms at different budget levels. We develop an efficient algorithm for computing these curves and propose bootstrap-based confidence intervals that are exact in large samples for any point on the curve. These confidence intervals can be used to conduct hypothesis tests comparing the value of treatment targeting using an optimal combination of arms with using just a subset of arms, or with a non-targeting assignment rule ignoring covariates, at different budget levels. We demonstrate the statistical performance in a simulation experiment and an application to treatment targeting for election turnout.

stat.ME↗

arXiv2023

Federated Offline Policy Learning

We consider the problem of learning personalized decision policies from observational bandit feedback data across multiple heterogeneous data sources. In our approach, we introduce a novel regret analysis that establishes finite-sample upper bounds on distinguishing notions of global regret for all data sources on aggregate and of local regret for any given data source. We characterize these regret bounds by expressions of source heterogeneity and distribution shift. Moreover, we examine the practical considerations of this problem in the federated setting where a central server aims to train a policy on data distributed across the heterogeneous sources without collecting any of their raw data. We present a policy learning algorithm amenable to federation based on the aggregation of local policies trained with doubly robust offline policy evaluation strategies. Our analysis and supporting experimental results provide insights into tradeoffs in the participation of heterogeneous data sources in offline policy learning.

cs.LG↗

arXiv2023

Torch-Choice: A PyTorch Package for Large-Scale Choice Modeling with Python

The $\texttt{torch-choice}$ is an open-source library for flexible, fast choice modeling with Python and PyTorch. $\texttt{torch-choice}$ provides a $\texttt{ChoiceDataset}$ data structure to manage databases flexibly and memory-efficiently. The paper demonstrates constructing a $\texttt{ChoiceDataset}$ from databases of various formats and functionalities of $\texttt{ChoiceDataset}$. The package implements two widely used models, namely the multinomial logit and nested logit models, and supports regularization during model estimation. The package incorporates the option to take advantage of GPUs for estimation, allowing it to scale to massive datasets while being computationally efficient. Models can be initialized using either R-style formula strings or Python dictionaries. We conclude with a comparison of the computational efficiencies of $\texttt{torch-choice}$ and $\texttt{mlogit}$ in R as (1) the number of observations increases, (2) the number of covariates increases, and (3) the expansion of item sets. Finally, we demonstrate the scalability of $\texttt{torch-choice}$ on large-scale datasets.

cs.LG↗