Browsing by Author "Pai, Mallesh M."

Now showing 1 - 8 of 8
  • Thumbnail Image
    Item
    (Bad) reputation in relational contracting
    (Wiley, 2022) Deb, Rahul; Mitchell, Matthew; Pai, Mallesh M.
    Motivated by markets for “expertise,” we study a bandit model where a principal chooses between a safe and risky arm. A strategic agent controls the risky arm and privately knows whether its type is high or low. Irrespective of type, the agent wants to maximize duration of experimentation with the risky arm. However, only the high type arm can generate value for the principal. Our main insight is that reputational incentives can be exceedingly strong unless both players coordinate on maximally inefficient strategies on path. We discuss implications for online content markets, term limits for politicians, and experts in organizations.
  • Thumbnail Image
    Item
    Censorship Resistance in On-Chain Auctions
    (Schloss Dagstuhl – Leibniz-Zentrum für Informatik, 2023) Fox, Elijah; Pai, Mallesh M.; Resnick, Max
    Modern blockchains guarantee that submitted transactions will be included eventually; a property formally known as liveness. But financial activity requires transactions to be included in a timely manner. Classical liveness does not guarantee this, particularly in the presence of a motivated adversary who benefits from censoring transactions. We define censorship resistance as the amount it would cost the adversary to censor a transaction for a fixed interval of time as a function of the associated tip. This definition has two advantages, first it captures the fact that transactions with a higher miner tip can be more costly to censor, and therefore are more likely to swiftly make their way onto the chain. Second, it applies to a finite time window, so it can be used to assess whether a blockchain is capable of hosting financial activity that relies on timely inclusion. We apply this definition in the context of auctions. Auctions are a building block for many financial applications, and censoring competing bids offers an easy-to-model motivation for our adversary. Traditional proof-of-stake blockchains have poor enough censorship resistance that it is difficult to retain the integrity of an auction when bids can only be submitted in a single block. As the number of bidders n in a single block auction increases, the probability that the winner is not the adversary, and the economic efficiency of the auction, both decrease faster than 1/n. Running the auction over multiple blocks, each with a different proposer, alleviates the problem only if the number of blocks grows faster than the number of bidders. We argue that blockchains with more than one concurrent proposer can have strong censorship resistance. We achieve this by setting up a prisoner’s dilemma among the proposers using conditional tips.
  • Thumbnail Image
    Item
    Collusive Outcomes via Pricing Algorithms
    (Oxford University Press, 2021) Hansen, Karsten T.; Misra, Kanishka; Pai, Mallesh M.
  • Thumbnail Image
    Item
    Continuous implementation with direct revelation mechanisms
    (Elsevier, 2022) Chen, Yi-Chun; Mueller-Frank, Manuel; Pai, Mallesh M.
    We investigate how a principal's knowledge of agents' higher-order beliefs impacts their ability to robustly implement a given social choice function. We adapt a formulation of Oury and Tercieux (2012): a social choice function is continuously implementable if it is partially implementable for types in an initial model and “nearby” types. We characterize when a social choice function is truthfully continuously implementable, i.e., using game forms corresponding to direct revelation mechanisms for the initial model. Our characterization hinges on how our formalization of the notion of nearby preserves agents' higher order beliefs. If nearby types have similar higher order beliefs, truthful continuous implementation is roughly equivalent to requiring that the social choice function is implementable in strict equilibrium in the initial model, a very permissive solution concept. If they do not, then our notion is equivalent to requiring that the social choice function is implementable in unique rationalizable strategies in the initial model. Truthful continuous implementation is thus very demanding without non-trivial knowledge of agents' higher order beliefs.
  • Thumbnail Image
    Item
    Frontiers: Algorithmic Collusion: Supra-competitive Prices via Independent Algorithms
    (INFORMS, 2021) Hansen, Karsten T.; Misra, Kanishka; Pai, Mallesh M.
    Motivated by their increasing prevalence, we study outcomes when competing sellers use machine learning algorithms to run real-time dynamic price experiments. These algorithms are often misspecified, ignoring the effect of factors outside their control, for example, competitors’ prices. We show that the long-run prices depend on the informational value (or signal-to-noise ratio) of price experiments: if low, the long-run prices are consistent with the static Nash equilibrium of the corresponding full information setting. However, if high, the long-run prices are supra-competitive—the full information joint monopoly outcome is possible. We show that this occurs via a novel channel: competitors’ algorithms’ prices end up running correlated experiments. Therefore, sellers’ misspecified models overestimate the own price sensitivity, resulting in higher prices. We discuss the implications on competition policy.
  • Thumbnail Image
    Item
    Mental health stigma
    (Elsevier, 2017) Bharadwaj, Prashant; Pai, Mallesh M.; Suziedelyte, Agne
    Comparing self-reports to administrative records, we find that survey respondents are significantly more likely to under-report mental illnesses compared to other health conditions. This behavior is consistent with the existence of stigma of mental illnesses. We show that stigma can play a role in determining health-seeking behavior.
  • Thumbnail Image
    Item
    Online Multivalid Learning: Means, Moments, and Prediction Intervals
    (Dagstuhl Publishing, 2022) Gupta, Varun; Jung, Christopher; Noarov, Georgy; Pai, Mallesh M.; Roth, Aaron
    We present a general, efficient technique for providing contextual predictions that are “multivalid” in various senses, against an online sequence of adversarially chosen examples (x, y). This means that the resulting estimates correctly predict various statistics of the labels y not just marginally –as averaged over the sequence of examples – but also conditionally on x ∈ G for any G belonging to an arbitrary intersecting collection of groups G. We provide three instantiations of this framework. The first is mean prediction, which corresponds to an online algorithm satisfying the notion of multicalibration from [5]. The second is variance and higher moment prediction, which corresponds to an online algorithm satisfying the notion of mean-conditioned moment multicalibration from [6]. Finally, we define a new notion of prediction interval multivalidity, and give an algorithm for finding prediction intervals which satisfy it. Because our algorithms handle adversarially chosen examples, they can equally well be used to predict statistics of the residuals of arbitrary point prediction methods, giving rise to very general techniques for quantifying the uncertainty of predictions of black box algorithms, even in an online adversarial setting. When instantiated for prediction intervals, this solves a similar problem as conformal prediction, but in an adversarial environment and with multivalidity guarantees stronger than simple marginal coverage guarantees.
  • Thumbnail Image
    Item
    The Centralizing Effects of Private Order Flow on Proposer-Builder Separation
    (Schloss Dagstuhl – Leibniz-Zentrum für Informatik, 2023) Gupta, Tivas; Pai, Mallesh M.; Resnick, Max
    The current Proposer-Builder Separation (PBS) equilibrium has several builders with different backgrounds winning blocks consistently. This paper considers how that equilibrium will shift when transactions are sold privately via order flow auctions (OFAs) rather than forwarded directly to the public mempool. We discuss a novel model that highlights the augmented value of private order flow for integrated builder searchers. We show that private order flow is complementary to top-of-block opportunities, and therefore integrated builder-searchers are more likely to participate in OFAs and outbid non integrated builders. They will then parlay access to these private transactions into an advantage in the PBS auction, winning blocks more often and extracting higher profits than non-integrated builders. To validate our main assumptions, we construct a novel dataset pairing post-merge PBS outcomes with realized 12-second volatility on a leading CEX (Binance). Our results show that integrated builder-searchers are more likely to win in the PBS auction when realized volatility is high, suggesting that indeed such builders have an advantage in extracting top-of-block opportunities. Our findings suggest that modifying PBS to disentangle the intertwined dynamics between top-of-block extraction and private order flow would pave the way for a fairer and more decentralized Ethereum.