Algorithmic Inflation: Dynamic Pricing and the End of Transparent Pricing
Inflation is no longer just a matter of monetary policy, but of algorithms. Dynamic pricing bots, by processing immense volumes of data, estimate each individua
Inflation, in its traditional macroeconomic sense, is a phenomenon linked to the expansion of the money supply, supply chain shocks, or geopolitical crises. AI-driven pricing bots do not possess the levers to "print money" or cause systemic inflation. However, the most recent economic and empirical literature unequivocally demonstrates that algorithms can amplify and redistribute inflation with surgical precision. By making prices hyper-reactive, granularly personalized, and structurally difficult to compare, dynamic pricing software manages to increase corporate profit margins, silently transferring the cost of economic uncertainty onto those who have less time, less data, or less bargaining power. The critical point that legislators and antitrust authorities must address today is not the existence of dynamic pricing itself, but the total lack of transparency, the risk of collusion, and the vacuum of protections for the end consumer.
The Invisible Mechanism: From the Storefront to Granular Discrimination
For decades, retail economics was based on an implicit assumption: the cost of changing a price tag (the so-called "menu cost") forced companies to keep prices stable for prolonged periods. As noted in a note by the Norwegian delegation to the OECD (2023), algorithms have eliminated modification costs, allowing for extremely high-frequency adjustments: prices rise instantly when demand grows and fall just as rapidly when it declines [2956].
But artificial intelligence goes far beyond simple adaptation to aggregate demand. A key 2025 OECD document on downstream markets argues that AI vastly expands the ability to estimate the so-called "willingness to pay" of the individual consumer. By integrating enormous volumes of behavioral data, browsing history, and geolocation, algorithms enable highly granular price discrimination and hyper-specialized offers [2954]. In this new paradigm, the exact same product or service is sold at profoundly different prices: it will systematically cost more to those who demonstrate they have fewer alternatives, less time for comparison, or a greater level of urgency.
The Shadow of Cartels: Tacit Collusion and Hub-and-Spoke Models
The most alarming impact of algorithmic pricing concerns market competition. The most solid institutional document on the matter, produced by the OECD for G7 jurisdictions (Algorithmic pricing and competition in G7 jurisdictions, 2025), distinguishes three scenarios of increasing risk: machine-facilitated explicit collusion, insidious hub-and-spoke schemes, and autonomous tacit collusion [2952].
The hub-and-spoke scheme currently represents the most complex antitrust threat: competing companies do not enter into any secret illegal agreement among themselves (the spokes), but all outsource the management of their prices to the same algorithmic software provider (the hub). The central algorithm, trained to maximize the margins of the entire industry, stabilizes prices upward without any human handshake ever having taken place.
Even more sophisticated is autonomous tacit collusion. As already highlighted in a historic 2017 OECD roundtable, machine learning models that observe and react to each other can learn, through trial and error, that triggering price wars is suboptimal, spontaneously converging toward a collusive equilibrium and artificially high prices [2957, 2959]. The OECD's Global Forum on Competition (2025) added a fundamental technical piece to understanding this phenomenon: artificial intelligence reduces demand uncertainty. In a traditional market, a company experiencing a sales decline struggles to understand whether it is a general contraction in demand or a competitor secretly lowering prices. AI dissolves this fog. By enabling companies to immediately distinguish a market shock from a competitive deviation, it makes tacit "agreements" of non-belligerence much more sustainable and algorithmic punishments against those who lower prices extremely rapid [2955].
The empirical effect of this hyper-rationalization has been documented with precision. A study by the Federal Reserve Bank of Richmond (2026) analyzed the German gas station market: after the massive introduction of algorithmic pricing software in 2017, there was no explosion of price wars. On the contrary, average prices and margins for gas stations increased by about 1.3 cents per liter, a net increase in profitability estimated at around 15% [2961]. The algorithm does not create inflation out of thin air, but it is extraordinarily efficient at capturing consumer surplus and transforming it into profit.
The Empirical Counterpoint: Efficiency Is Not a Myth
However, it would be a mistake to fall into absolute technological determinism, labeling every digital price variation as fraud. The empirical literature imposes a nuanced reading, demonstrating that the effect of dynamic pricing depends intimately on market structure, the type of data, and the rules of engagement.
A large-scale study conducted by UC San Diego (2025), based on the analysis of over 180 million product observations, categorically refuted fears of surge pricing (congestion-based price increases) in physical supermarkets. The research found no evidence that the adoption of electronic shelf labels produced opportunistic real-time price spikes, not even during periods of high general inflation. Price increases in supermarkets closely followed wholesale costs, nearby competitor moves, and manufacturers' promotional cycles [2962]. It is essential, therefore, not to confuse the mundane transmission of logistics chain inflation with an algorithmic conspiracy.
There are also sectors where high-frequency dynamic pricing generates unequivocal systemic benefits. In the food delivery sector, research from the University of Arizona (2024) found that algorithmic pricing significantly reduced demand volatility, spreading orders more rationally, increasing overall kitchen output, and even managing to lower, on average, final prices for users [2965].
These empirical data support the defensive arguments of the Mercatus Center (2025), which argues that dynamic pricing ensures a vastly more efficient allocation of resources compared to fixed prices, rewarding with substantial discounts consumers willing to buy during low-demand periods. In ideal scenarios of high competition, price personalization could even benefit low-income consumers, offering them rates calibrated to a lower spending capacity [2964].
Information Asymmetry and the Urgency Tax
The real sociological and political question hidden behind algorithmic inflation is not "do algorithms make prices rise in absolute terms?", but rather: "who pays more when the price changes in real time, and who is subsidized?".
Artificial intelligence, stripped of moral intent, merely maps transactional vulnerability. An algorithm can very well lower the price of an intercontinental flight booked six months in advance by a flexible tourist, but it will ruthlessly raise the last-minute price purchased by a worker in a family emergency. It can offer a targeted discount to someone with a low income and plenty of free time to compare offers online, while simultaneously extracting the maximum possible value from the credit card of someone who cannot postpone the purchase and has no time for comparative research [2961, 2964, 2954].
In this dynamic, price ceases to be a transparent indicator of the intrinsic value of a good or its production costs, transforming into an opaque mirror of the buyer's momentary need. Algorithmic inflation, ultimately, is not merely a cold statistical indicator that appears in the consumer price indices compiled by central banks; it is the constant, pervasive sensation, experienced daily by millions of citizens, that the shopping cart, the plane ticket, or the rent change amount a moment before one can even understand the real reason why [2952, 2961, 2962].
Bibliographic References and Sources
- [Algorithmic pricing and competition in G7 jurisdictions — OECD, 2025] [2952]
- [AI-related competition concerns in downstream markets — OECD, 2025] [2954]
- [Understanding Algorithmic Pricing — Federal Reserve Bank of Richmond, 2026] [2961]
- [Algorithms and collusion — OECD, 2017] [2957, 2959]
- [Artificial Intelligence and Competition — OECD Global Forum on Competition, 2025] [2955]
- [Algorithmic competition — Note by Norway — OECD, 2023] [2956]
- [New Research Debunks Fears of Supermarket Surge Pricing — UC San Diego, 2025] [2962]
- [Dynamic Pricing, Intertemporal Spillovers, and Efficiency — University of Arizona, 2024] [2965]
- [The Case for Algorithmic Pricing — Mercatus Center, 2025] [2964]
Article by the Editorial Staff of La Bussola dell’IA