The Economics of Non-Rivalrous Data Markets

59 concepts40 papers2 proposals

Traditional economics is built on the concept of scarcity. If you consume a physical commodity—like a barrel of oil or an apple—no one else can consume it. This property is known as rivalry.

Digital data fundamentally breaks this assumption. Data is non-rivalrous: it can be replicated infinitely at near-zero marginal cost, and a seller retains the asset even after a transaction is complete. Because the classical laws of supply and demand assume that items are finite and mutually exclusive, applying traditional economic frameworks to data marketplaces causes them to collapse.

To build functioning data economies, researchers are rethinking valuation, pricing, and market design from the ground up, moving away from "cost-plus" models toward frameworks that account for infinite replication, context-dependent utility, and massive market externalities.

The Replicability Challenge and Zero Marginal Cost

In a standard market, the price of a good is heavily tied to its marginal cost of production. However, as observed by Danmin Qian et al. in Model Research of Database Dynamic Pricing Based on DDLG, digital resources are characterized by high initial development expenses but near-zero replication costs. When supply is effectively infinite, static pricing strategies fail to reliably recover fixed costs or optimize market clearance.

This mismatch invalidates traditional allocation mechanisms. For instance, standard auction formats are designed to allocate scarce, indivisible items to the highest bidder. To address this, Yutao Jiao et al. in Profit Maximization Auction and Data Management in Big Data Markets formulate alternative auction mechanisms specifically tailored for digital goods with unlimited supply.

Furthermore, because data can be shared without the original owner losing access, classic general equilibrium models (such as Arrow-Debreu markets) fail to guarantee stable outcomes. As Hannaneh Akrami et al. demonstrate in On the Theoretical Foundations of Data Exchange Economies, modeling data transactions requires entirely new mathematical foundations that explicitly accommodate non-rivalry and infinite supply.

Instrumental and Context-Dependent Valuation

If data has no intrinsic production cost per unit, how do we determine what it is worth? The answer is that data value is entirely instrumental and context-dependent. Data has no fixed "spot price"; its worth is determined by how much it improves a specific buyer's decision-making relative to what that buyer already knows.

[Buyer's Prior Knowledge] + [New Dataset] ---> [Improved Decision Utility]
                                      ^
                                      |
                       Value = Marginal Increase in Utility

Rui Ai et al. formalize this concept in An Instrumental Value for Data Production and its Application to Data Pricing, proving that data value is defined by the expected increase in a buyer's utility within their unique decision-making context. Because every buyer possesses a different prior information set and different operational goals, a seller can theoretically execute near-perfect price discrimination, tailoring prices to extract the specific surplus generated for each individual user.

This context-dependence makes data an "experience good." As highlighted by Mengxiao Zhang et al. in A Survey of Data Pricing for Data Marketplaces, a buyer often cannot accurately verify the value of information before consuming or integrating it into their workflow. This information asymmetry means that fixed-price models are highly inefficient, necessitating value-based, endogenous pricing models that adapt to consumption patterns (Jian Pei 0001 et al., Data Pricing - From Economics to Data Science).

The Double-Edged Sword of Externalities

Because data is non-rivalrous and can be distributed widely, individual data transactions create powerful ripple effects across the market. These externalities manifest in two distinct ways, both of which severely distort unregulated data markets.

Negative Externalities: Competitive Erosion

In zero-sum or highly competitive environments (such as high-frequency trading or market analytics), the value of data is tied directly to its exclusivity. If multiple competitors acquire the same dataset, any individual competitive advantage is erased.

Anish Agarwal et al. address this in A Marketplace for Data: An Algorithmic Solution, noting that traditional valuation models routinely ignore these negative externalities. When buyers act independently, they fail to account for how their purchases diminish the utility of the data for others. As Safwan Hossain et al. prove in Equilibrium of Data Markets with Externality, this lack of coordination drives unregulated markets toward equilibria that significantly degrade collective social welfare, forcing buyers into defensive purchasing patterns just to avoid falling behind.

Positive Externalities: Free-Riding and Over-Incentivization

Conversely, when multiple buyers attempt to source high-quality data from the same strategic provider (such as an independent sensor network or a crowd-sourced user base), data's non-rivalrous nature turns it into a public good.

In Competitive Statistical Estimation with Strategic Data Sources, Tyler Westenbroek et al. show that if one buyer pays a source to improve its data precision, other buyers automatically benefit from that higher quality without contributing to the cost. This creates a severe free-rider problem.

Paradoxically, in purely competitive settings, this dynamic can flip into aggressive over-incentivization. The authors demonstrate in Statistical Estimation with Strategic Data Sources in Competitive Settings that buyers competing for a relative edge will independently shower a data source with incentive payments. This forces the source to exert effort levels far beyond what is globally efficient, resulting in a high Price of Anarchy and systemic welfare loss.

Designing Non-Rivalrous Market Mechanisms

To circumvent these market failures, researchers have designed specialized trading architectures that embrace, rather than fight, the unique properties of digital goods.

Strategic Partial Sharing

While sharing complete datasets between rivals typically triggers aggressive price competition that erodes profit margins (Ronen Gradwohl et al., Selling Data to a Competitor (Extended Abstract)), total data isolation is also inefficient. To resolve this, Moshe Tennenholtz et al. in Pareto-Improving Data-Sharing✱ demonstrate that carefully calibrated partial data-sharing mechanisms can unlock positive-sum outcomes, simultaneously increasing firm profits and consumer welfare.

Utility Balancing and Flow Networks

Because data can be routed to multiple parties simultaneously without depleting the source, data barter economies cannot be solved using traditional one-to-one matching algorithms. Instead, Aditya Bhaskara et al. in Data Exchange Markets via Utility Balancing model data marketplaces as complex network flow problems. This allows platforms to balance multi-party distribution and optimize collective utility without relying on a uniform global currency.

Non-Monetary "Model Rewards"

In collaborative machine learning, participants pool data to train shared models. However, standard cooperative game theory axioms (like the standard Shapley value) assume a fixed-sum pie, failing to account for the fact that a trained model can be duplicated infinitely to benefit all participants simultaneously.

Traditional Cooperative Game:
[Total Value $100] ---> Split among players: Player A ($60) + Player B ($40)

Non-Rivalrous Model Collaboration:
[High-Quality Model] ---> Replicated at zero cost ---> Sent to Player A (Full Value)
                                                  ---> Sent to Player B (Full Value)

To align incentives without cash friction, Rachael Hwee Ling Sim et al. in Collaborative Machine Learning with Incentive-Aware Model Rewards propose rewarding contributors with personalized, noise-calibrated "model rewards." By utilizing conditional Shapley values, Quoc Phong Nguyen et al. in Trade-off between Payoff and Model Rewards in Shapley-Fair Collaborative Machine Learning prove that marketplaces can successfully trade off monetary payouts with model quality, ensuring fair, stable, and mathematically rigorous collaboration in non-rivalrous environments.

Key Foundational Work

Model Research of Database Dynamic Pricing Based on DDLG (2015) — contributed 1 concept elements to this direction.
Network Non-Neutrality through Preferential Signaling (2013) — contributed 1 concept elements to this direction.
History & citationVersion 1 · 2026-10-04

Current · Permanent link

View frozen version 1

Cite this direction

Latest version (a living page):

@misc{lacuna_non_rivalrous_asset_pricing_a36de265,
  title = {non-rivalrous asset pricing},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/non-rivalrous-asset-pricing/rdr_a36de265cb9847dc93e030a57c48840e},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-20},
}

This exact version (version 1, frozen text):

@misc{lacuna_non_rivalrous_asset_pricing_a36de265_v1,
  title = {non-rivalrous asset pricing},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/non-rivalrous-asset-pricing/rdr_a36de265cb9847dc93e030a57c48840e/v/1},
  note = {Version 1, frozen 2026-10-04. Accessed 2026-11-20},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake