Working papers/Under revision

1. ‘Robust Technology Regulation’
   (with Sivakorn Sanguanmoo; latest draft: March 25)
   [pdf] [ssrn]
   Talks: AEA 2025 meeting session on “Policy Implications of Transformative AI”

   • Regulatory sandboxes for risky R&D are (i) robust when the researcher’s learning process is chosen adversarially; (ii) dominant among robust mechanisms; and (iii) important — without robustness, the worst-case learning process induces weak optimism

2. `Balanced Social Learning’ — preliminary, comments welcome!
   (with Ricky Li; latest draft: Apr 25)
   [pdf]

   • Balance is essential for efficiency under multidimensional social learning. There’s a tradeoff between speed (learning quickly) and breadth (learning broadly). Complete characterization of when and how efficiency can be guaranteed under different information regimes.

3. ‘Attention Capture’
   (with Sivakorn Sanguanmoo; latest draft: Sep 24)
   [pdf] [ssrn] [arxiv]
   Revise & resubmit, Journal of Political Economy
   Talks: AEA 2025 meeting session on “Attention and Engagement on Platforms”, National University of Singapore, Singapore Management University

   • A unified analysis of how information captures attention.

4. `Informational Puts’
   Extended abstract in EC’24 (titled ‘Full Dynamic Implementation’)
   (with Sivakorn Sanguanmoo & Kei Uzui; latest draft: Dec 24)
   [pdf] [arxiv] [ssrn]
   Talks: Nuffield Oxford, Cambridge, EC’24, various Econometric Society meetings

   • Complete characterization of full implementation in dynamic supermodular games. Whatever can be partially implemented under commitment can also be implemented fully without commitment!

5. `Inertial Coordination Games’ — submitted
   (with Ricky Li & Kei Uzui; draft date: March 25)
   [pdf] [arxiv]

   • Learning speeds determine whether shocks persist vs fizzle out: risk dominance iff posterior precisions grow sub-quadratically; otherwise shocks can propagate and generate self-fulfilling spirals.

6. ‘Market Segmentation through Information’ — submitted
   (with Matt Elliott, Andrea Galeotti, & Wenhao Li; latest draft: Feb 25)
   [pdf]

   • Information is a powerful tool for shaping price competition + characterization of all outcomes when a platform controls both access and information.

7. ‘Delaying the Deviation’ (with Anna Merotto) — new!!
   [write us for a draft]

   • Randomizing small transfers across agents and across time uniquely implements regime survival even when attacking is both payoff- and risk-dominant. Both kinds of randomization are necessary.

8. `Flexible Demand Manipulation’ — new!!
   (with Yifan Dai; draft date: Oct 24)
   Talks: 2024 Berkeley/Columbia/Duke/MIT/Northwestern IO Theory Conference (short session)
   [ssrn] [arxiv]

   • How does targeted advertising interact with market power? We fully characterize the form and value of consumer- and producer-optimal advertising plans under both ex-ante and ex-post measures of welfare.

9. `Persuasion and Optimal Stopping’
   (with Sivakorn Sanguanmoo & Weijie Zhong; latest draft: Dec 24)
   [pdf]
   Talks: Harvard EconCS seminar, SITE 2024

   • A unified analysis of optimal dynamic persuasion to implement joint distributions over actions, states, and stopping times. No intertemporal commitment is necessary.

10. `Prices and Symmetries’ — new!!
    (with Pedro Martinez-Bruera; latest draft: Feb 25)
    [pdf] [ssrn] [slides]
    Talks: Network Science in Econ (Stanford)

    • Price interventions dominate quantities when shocks to the network of spillovers are sufficiently symmetric. Quantities dominate if they are sufficiently antisymmetric.

11. ‘Speed vs Resilience in Contagion’ — new version with a generalization to sparse continuous graphs + heterogeneous thresholds coming soon!!
    (with Stephen Morris; draft date: Dec 22)
    [pdf] [ssrn] [slides]

    • Threshold contagion is faster in networks where it is harder to initiate contagion. Intermediate links are important bulwarks against contagion.

Work-in-progress (drafts coming soon!)

1. `The Supply of Rationalizations’

   • How do we offer flow information to agents who have directional demand for beliefs and seek justifications for beliefs they want to hold for non-epistemic reasons?

2. `Feasible Adapted Beliefs’ (with Sivakorn Sanguanmoo)

   • General principles for how information should be delivered when (i) the principal is learning over time; and (ii) the agent is learning from outside information.

3. `Manipulated Competition’ (with Yifan Dai)

   • A complete characterization of optimal advertising plans to maximize producer or consumer surplus when firms compete on prices.