Working Papers

A persistent puzzle throughout the cognitive and social sciences is how people manage to learn social conventions from the sparse and noisy behavioral data of diverse actors, without explicit instruction. Here, we show that the dominant theories…

Recommender systems optimized for short-term engagement often hurt long-term consumer satisfaction and retention. Yet optimizing long-term outcomes is challenging because such signals are sparse and noisy. We study short-form video platforms,…

Problem definition: Accurately modeling consumer behavior in energy operations is challenging due to uncertainty, behavioral heterogeneity, and limited empirical data-particularly in low-frequency, high-impact events. While generative AI trained…

Despite the large common net benefits of climate mitigation, broad-based political consensus for large-scale policy action remains elusive. We hypothesize that financial exposure to energy stocks central to the green transition can induce…

Artificial intelligence (A.I.) will likely be the most important technology we have ever developed. Technologies such as electricity, semiconductors, and the internet have been transformative, reshaping economic activity and dramatically…

Antitrust laws mandate review of mergers and acquisitions (M&As) that exceed an asset size threshold based on accounting standards that exclude most intangible capital. We show that this exclusion leads to thousands of intangible-intensive M…

We discuss the conditions under which global imbalances, such as China being a large foreign creditor and the United States being a large foreign debtor, might also generate power imbalances. We highlight possible theoretical channels and…

We study whether AI methods applied to large-scale portfolio holdings data can improve financial regulation. We build a state-of-the-art, graph-based deep learning model tailored to security-level data on the holdings of financial intermediaries…

Problem Definition: Bangladesh suffers from massive lead emissions from its circular Lead-Acid Battery (LAB)

industry. Informal smelting of scrap lead from Used Lead-Acid Batteries (ULAB) is especially emission-intensive…

Modern recommender systems use machine learning (ML) models to predict consumer preferences based on consumption history. Although these “black-box” models achieve impressive predictive performance, they often suffer from a lack of transparency…

We develop a unified theory of delegated capital management by extending the paradigm of Berk and Green (2004) from mutual funds to alternative assets. With competitive markets and rational investors, we derive the optimal contract and account…

We model the attorney-client relationship as a dynamic moral hazard problem where the client faces a difficult inference challenge. Clients learn about case quality by “learning from silence,” but silence is a noisy signal: it can mean the case…

Developing accurate clinical prediction models is often bottlenecked by the difficulty of generating meaningful predictive features from unstructured data. While electronic health records (EHRs) contain rich narrative information, extracting a…

This paper analyzes interactions between corporate governance and law enforcement practices, focusing on cases where deterrence is weak and harmful misconduct is profitable. We show how managerial compensation contracts, including stock-based…

In a patent privateering strategy, firms sell patents to a non-practicing entity (NPE) with the expectation that the NPE sues the seller’s rivals for patent infringement. We examine whether firms under competitive pressure and facing barriers to…

Policymakers often distort goods markets to effect redistribution—for example, via price controls, differential taxation, or in-kind transfers. We investigate the optimality of such policies alongside the (optimally-designed) income tax. In our…

We examine how patent disclosures influence technology classification decisions by patent examiners and how firms strategically tailor these disclosures to affect classification outcomes. Using a novel machine learning approach, we construct a…

We use a play-by-play dataset from the game show “Jeopardy!” to study the hot hand phenomenon, whereby people appear to exhibit “hot” states of elevated performance in domains with repeat trials. We first demonstrate that Jeopardy contestants…

We study poverty minimization via direct transfers, framing this as a statistical learning problem while retaining the information constraints faced by real-world programs. Using nationally representative household consumption surveys from 23…

Modern recommender systems rely on black-box machine learning models to predict consumer choices. However, because these models do not explicitly represent the underlying data-generating process (DGP), they often struggle to generalize beyond…