A5001211242- Advanced Causal Inference Techniques
- Consumer Market Behavior and Pricing
- Auction Theory and Applications
- Statistical Methods and Inference
- Advanced Bandit Algorithms Research
- Statistical Methods and Bayesian Inference
- Economic theories and models
- Digital Platforms and Economics
- Statistical Methods in Clinical Trials
- Merger and Competition Analysis
- Economic Policies and Impacts
- Game Theory and Applications
- Monetary Policy and Economic Impact
- Economic and Environmental Valuation
- Experimental Behavioral Economics Studies
- Health Systems, Economic Evaluations, Quality of Life
- Machine Learning and Algorithms
- Media Influence and Politics
- ICT Impact and Policies
- Bayesian Modeling and Causal Inference
- Machine Learning and Data Classification
- Misinformation and Its Impacts
- Decision-Making and Behavioral Economics
- School Choice and Performance
- Healthcare Policy and Management
Stanford University
2015-2024
University of Chicago
2021-2024
University of Chile
2024
United States Department of Justice
2022-2023
National Bureau of Economic Research
2013-2023
Analysis Group (United States)
2022
Meta (United States)
2022
Dana-Farber/Harvard Cancer Center
2018-2022
Yaoundé Gynaecology, Obstetrics and Pediatrics Hospital
2021
Université de Yaoundé I
2021
Many scientific and engineering challenges—ranging from personalized medicine to customized marketing recommendations—require an understanding of treatment effect heterogeneity. In this article, we develop a nonparametric causal forest for estimating heterogeneous effects that extends Breiman's widely used random algorithm. the potential outcomes framework with unconfoundedness, show forests are pointwise consistent true have asymptotically Gaussian centered sampling distribution. We also...
We propose generalized random forests, a method for nonparametric statistical estimation based on forests (Breiman [Mach. Learn. 45 (2001) 5–32]) that can be used to fit any quantity of interest identified as the solution set local moment equations. Following literature maximum likelihood estimation, our considers weighted nearby training examples; however, instead using classical kernel weighting functions are prone strong curse dimensionality, we use an adaptive function derived from...
In this paper we propose methods for estimating heterogeneity in causal effects experimental and observational studies conducting hypothesis tests about the magnitude of differences treatment across subsets population. We provide a data-driven approach to partition data into subpopulations that differ their effects. The enables construction valid confidence intervals effects, even with many covariates relative sample size, without "sparsity" assumptions. an "honest" estimation, whereby one...
In empirical work in economics it is common to report standard errors that account for clustering of units.Typically, the motivation given adjustments unobserved components outcomes units within clusters are correlated.However, because correlation may occur across more than one dimension, this makes difficult justify why researchers use some dimensions, such as geographic, but not others, age cohorts or gender.This also explain should cluster with data from a randomized experiment.In paper,...
In this paper, we discuss recent developments in econometrics that view as important for empirical researchers working on policy evaluation questions. We focus three main areas, each case, highlighting recommendations applied work. First, new research identification strategies program evaluation, with particular synthetic control methods, regression discontinuity, external validity, and the causal interpretation of methods. Second, various forms supplementary analyses, including placebo...
This paper develops an alternative approach to the widely used Difference-In-Difference (DID) method for evaluating effects of policy changes.In contrast standard approach, we introduce a nonlinear model that permits changes over time in effect unobservables (e.g., there may be trend level wages as well returns skill labor market).Further, our assumptions are independent scaling outcome.Our provides estimate entire counterfactual distribution outcomes would have been experienced by treatment...
Abstract Clustered standard errors, with clusters defined by factors such as geography, are widespread in empirical research economics and many other disciplines. Formally, clustered errors adjust for the correlations induced sampling outcome variable from a data-generating process unobserved cluster-level components. However, econometric framework clustering leaves important questions unanswered: (i) Why do we some ways but not others, example, state gender, observational studies completely...
We discuss the relevance of recent machine learning (ML) literature for economics and econometrics. First we differences in goals, methods, settings between ML traditional econometrics statistics literatures. Then some specific methods from that view as important empirical researchers economics. These include supervised regression classification, unsupervised matrix completion methods. Finally, highlight newly developed at intersection typically perform better than either off-the-shelf or...
We present a new estimator for causal effects with panel data that builds on insights behind the widely used difference-in-differences and synthetic control methods. Relative to these methods we find, both theoretically empirically, this “synthetic difference-in-differences” has desirable robustness properties, it performs well in settings where conventional estimators are commonly practice. study asymptotic behavior of when systematic part outcome model includes latent unit factors...
This paper derives sufficient conditions for a class of games incomplete information, such as first price auctions, to have pure strategy Nash equilibria (PSNE).The treats between two or more heterogeneous agents, each with private information about his own type (for example, bidder's value an object firm's marginal cost production), and the types are drawn from atomless joint probability distribution which potentially allows correlation types.Agents' utility may depend directly on...
Journal Article Monotone Comparative Statics under Uncertainty Get access Susan Athey Stanford University and National Bureau of Economic Research Search for other works by this author on: Oxford Academic Google Scholar The Quarterly Economics, Volume 117, Issue 1, February 2002, Pages 187–223, https://doi.org/10.1162/003355302753399481 Published: 01 2002
We analyze collusion in an infinitely repeated Bertrand game, where prices are publicly observed and each firm receives a privately observed, i.i.d. cost shock period. Productive efficiency is possible only if high-cost firms relinquish market share. In the most profitable collusive schemes, implement productive efficiency, favored with higher expected share future periods. If types discrete, there exists discount factor strictly less than one above which first-best profits can be attained...
Despite numerous journalistic accounts, systematic quantitative evidence on economic conditions during the ongoing COVID-19 pandemic remains scarce for most low- and middle-income countries, partly due to limitations of official statistics in environments with large informal sectors subsistence agriculture. We assemble from over 30,000 respondents 16 original household surveys nine countries Africa (Burkina Faso, Ghana, Kenya, Rwanda, Sierra Leone), Asia (Bangladesh, Nepal, Philippines),...
We consider an infinitely repeated Bertrand game, in which prices are publicly observed and each firm receives a privately observed, i.i.d. cost shock period. focus on symmetric perfect public equilibria, wherein any "punishments" borne equally by all firms. identify tradeoff that is associated with collusive pricing schemes the price to be charged strictly increasing its level: such "fully sorting" offer efficiency benefits, as they ensure lowest-cost makes current sale, but also imply...
This article examines a model in which advertisers bid for “sponsored-link” positions on search engine. The value derive from each position is endogenized as coming sales to population of consumers who make rational inferences about firm qualities and optimally. Consumer strategies, equilibrium bidding, the welfare benefits auctions are analyzed. Implications reserve prices number other auction design questions discussed.
We apply causal forests to a dataset derived from the National Study of Learning Mindsets, and discusses resulting practical conceptual challenges.This note will appear in an upcoming issue Observational Studies, Empirical Investigation Methods for Heterogeneity, that compiles several analyses same dataset.1.According Study, "A growth mindset is belief intelligence can be developed.Students with understand they get smarter through hard work, use effective strategies, help others when...
Data sharing, research ethics, and incentives must improve
We study entry and bidding patterns in sealed bid open auctions. Using data from the U.S. Forest Service timber auctions, we document a set of systematic effects: auctions attract more small bidders, shift allocation toward these can also generate higher revenue. A private value auction model with endogenous participation account for qualitative effects format. estimate model's parameters show that it explain quantitative as well. then use to assess bidder competitiveness, which has...
Summary There are many settings where researchers interested in estimating average treatment effects and willing to rely on the unconfoundedness assumption, which requires that assignment be as good random conditional pretreatment variables. The assumption is often more plausible if a large number of variables included analysis, but this can worsen performance standard approaches effect estimation. We develop method for debiasing penalized regression adjustments allow sparse methods like...