Giacomo Mantegazza


Hello, and welcome to my website! I am a PhD candidate in Operations, Information & Technology Stanford Graduate School of Business, where I am advised by Kostas Bimpikis.

Broadly, my research investigates the interface between competition on markets and the strategies agents adopt to seek, distribute, and use information. In particular, I am interested in understanding how the design and operations of online and decentralized markets influence information acquisition, pricing and the welfare of consumers and the environment.

I was born in Milano, Italy. Prior to Stanford I got a Bachelor and Master in Economics from Bocconi University.

I am on the 2023-2024 academic job market.

You can find my full CV here.

Email: giacomo.mantegazza 'at' stanford 'dot' edu

Published papers

Two-sided platforms play an important role in reducing frictions and facilitating trade, and in doing so they increasingly engage in collecting and processing data about supply and demand. This paper establishes that platforms have an incentive to strategically disclose (coarse) information about demand to the supply side, as this can considerably boost their profits. However, this practice may also adversely affect the welfare of consumers. By optimally designing its information disclosure policy, a platform can influence the entry and pricing decisions of its potential suppliers. In general, it is optimal for the platform to disclose its information only partially to either “nudge” entry when it is a priori costly for suppliers to join or, conversely, discourage it when suppliers do not have access to attractive outside options. On the other hand, consumers may end up being worse off, as they have access to fewer trading options and/or face higher prices compared with when the platform refrains from sharing any demand information to its potential suppliers.

Working papers

Artificial Intelligence and Spontaneous Collusion, with M. Banchio

Extended abstract in the Proceedings of the 24th Conference on Economics and Computation (EC '23)

We develop a tractable model for studying strategic interactions between learning algorithms. We uncover a mechanism responsible for the emergence of algorithmic collusion. We observe that algorithms periodically coordinate on actions that are more profitable than static Nash equilibria. This novel collusive channel relies on an endogenous statistical linkage in the algorithms' estimates which we call spontaneous coupling. The model's parameters predict whether the statistical linkage will appear, and what  market structures facilitate algorithmic collusion. We show that spontaneous coupling can sustain collusion in prices and market shares, complementing experimental findings in the literature. Finally, we apply our results to design algorithmic markets.

Maritime transportation is essential for global supply chains, but the issue of ballasting—vessels traveling without cargo—imposes significant economic and environmental costs. This paper focuses on the oil transportation industry, where about half of the total traveled miles are sailed empty, and reveals that fragmentation is the most important cause to ballasting after demand imbalances, accounting for 17-20% of the total. We find that it is possible to reduce carbon emissions associated with ballasting by as much as 13%  by consolidating the market into small shipping pools, which avoids concerns about excessive market power. Consolidation improves utilization because larger pools better coordinate and diversify the set of ports they serve, which reduces the need for vessel relocations. At a higher level, this work shows the extent of the sustainability gains that can be obtained solely by organizing more efficiently the resources available in today’s supply chains.

Work in progress

The Adverse Effects of Algorithmic Dynamic Pricing, with M. Banchio

Sellers generally benefit from the ability to credibly commit to a sequence of prices, for example because this allows them to earn the most profits from patient customers. In this work we ask whether online learning algorithms can earn such credibility. In a durable-good monopoly problem, we argue that reinforcement learning procedures need to be able to assess the value of reputation in order to sustain time-inconsistent policies, and we show how well-known design choices lead to effectively learning reputation. We consider a monopolist learning to price an inventory of durable goods for patient, strategic, and heterogeneous consumers. Sustaining high prices guarantees maximal revenues, but it is irrational when facing only low-value consumers. We show that an algorithm with sufficiently long time-horizon learns to attribute the revenues earned early in the game to high prices in later periods, thus effectively learning to commit. Thus, as a consequence of dynamic pricing implemented via learning algorithms, consumers are worse off.