The Algorithm's Grip I: How Pricing Bots Could Undermine Fair Markets

Case Studies, Pseudocode, and A Simple Checklist

Image by DallE-3

In the increasingly digital landscape of modern commerce, algorithms have become ubiquitous, silently shaping the prices we pay for everything from groceries to plane tickets. These sophisticated software programs, often referred to as "pricing bots," are designed to dynamically adjust prices based on a myriad of factors, including market demand, competitor pricing, and even our browsing habits. While proponents tout the efficiency and responsiveness of algorithmic pricing, a growing chorus of critics warns of its potential to undermine fair markets and harm consumers.

At the heart of this debate lies the tension between innovation and regulation. Algorithmic pricing, powered by advancements in artificial intelligence and machine learning, promises to optimize pricing strategies and personalize consumer experiences. However, the opacity of these algorithms and the speed at which they operate raise concerns about potential anti-competitive practices, such as price fixing, collusion, and market manipulation.

Let’s have a look into the intricate world of pricing algorithms, exploring their functionalities, benefits, and inherent risks. We will examine prominent regulatory cases where algorithms have been implicated in anti-competitive behavior, analyze the underlying reasons for these breaches, and offer a simple checklist for identifying potential issues.

By understanding the mechanics of these "pricing bots" and their potential impact on market dynamics, we can begin to formulate effective strategies for ensuring that algorithmic pricing serves the interests of both businesses and consumers, fostering fair and competitive markets in the digital age.

Please note that information gleaned and shared is based purely on publicly available data or cases, and is shared in the interest of learning and understanding the issues better. Nothing should be read as prescriptive, and any references to cases in given jurisdictions do not infer that similar issues should arise elsewhere. Also, note some interesting cases that were terminated and grounds for termination.

What are Pricing Algorithms?

Pricing algorithms are sets of rules and calculations that automatically determine the price of a product or service. They operate by analyzing vast amounts of data, including market demand, competitor pricing, inventory levels, customer demographics, and even real-time events like weather patterns or news headlines. Based on this analysis, the algorithms adjust prices dynamically, either in real-time or at predetermined intervals, to optimize revenue, respond to market fluctuations, or achieve other business objectives.

Examples across Industries

• E-commerce: Giants like Amazon and eBay rely heavily on pricing algorithms to adjust prices for millions of products based on competitor pricing, demand fluctuations, and individual customer browsing behavior.

• Travel: Airlines and hotels use sophisticated algorithms to dynamically price flights and hotel rooms, factoring in booking patterns, time of year, competitor pricing, and even the user's search history.

• Ride-sharing: Companies like Uber and Lyft utilize surge pricing algorithms that increase prices during periods of high demand or limited driver availability.

• Retail: Brick-and-mortar stores are increasingly adopting dynamic pricing strategies, using algorithms to adjust prices based on factors like inventory levels, competitor promotions, and even the time of day.

Benefits of Using Pricing Algorithms

From a business perspective, pricing algorithms offer several advantages:

• Revenue Optimization: By analyzing vast datasets and identifying optimal price points, algorithms can help businesses maximize revenue and profitability.

• Market Responsiveness: Algorithms enable businesses to react quickly to changes in market conditions, such as shifts in demand, competitor pricing adjustments, or unexpected events. This agility can provide a competitive edge.

• Personalized Pricing: Algorithms can be used to tailor prices to individual customers based on their purchasing history, browsing behavior, and other factors. This can enhance customer experience and potentially increase sales.

• Efficiency and Automation: Automated pricing can significantly reduce the time and resources required for manual price setting, freeing up staff for other tasks.

Potential Risks and Concerns

Despite their potential benefits, pricing algorithms also raise significant concerns:

• Price Discrimination: Algorithms can be used to charge different prices to different customers based on factors like their location, demographics, or purchasing power. This can raise ethical and fairness concerns, potentially leading to accusations of discriminatory pricing practices.

• Collusion and Price Fixing: Algorithms can facilitate collusion among competitors by allowing them to automatically match or coordinate prices. This can artificially inflate prices, reduce consumer choice, and stifle competition.

• Reduced Market Competition: The use of sophisticated pricing algorithms by dominant companies can create barriers to entry for smaller competitors, making it difficult for them to compete on price. This can lead to a less dynamic and innovative market.

• Lack of Transparency: The opacity of some pricing algorithms makes it difficult for consumers and regulators to understand how prices are being set and whether they are fair and competitive. This lack of transparency can erode trust and make it harder to detect anti-competitive behavior.

Government Action Against Anti-Competitive Pricing Algorithms

As the use of pricing algorithms has proliferated, governments and regulatory bodies around the world have become increasingly concerned about their potential to facilitate anti-competitive practices. Several high-profile cases have highlighted the need for greater scrutiny and potential regulation of algorithmic pricing to ensure fair competition and protect consumers.

Useful Reference: 2023 DOJ/FTC Merger Guidelines

Notable Regulatory Cases

• Apple & Publishers (E-books):

  • Case Details: In 2012, the US Department of Justice (DOJ) filed a lawsuit against Apple and five major publishers, alleging that they had conspired to fix e-book prices, effectively ending Amazon's dominance with its $9.99 pricing strategy.

  • Algorithm/Agreement: The publishers adopted an "agency model" where they set the retail price, and Apple took a 30% commission. This model ensured that e-book prices rose across the board. While not directly involving an algorithm, the agreement itself functioned as a price-setting mechanism, demonstrating how coordinated action can manipulate markets.

  • Outcome: Apple was found guilty of violating antitrust laws and fined $450 million.

  • Reference: DOJ Files Antitrust Suite Against Apple and 5 Publishers

  • Reference: E-Book Retailers Distribute $400Million to Victims of Apple-led Conspiracy

• Google (Online Advertising) EU Antitrust:

  • Case Details: The European Commission found Google guilty of abusing its market dominance in online advertising through its AdSense platform in 2017.

  • Algorithm: Google's AdSense algorithm allegedly favored its own comparison shopping services in search results, demoting rivals. While the exact algorithm is proprietary, it likely involved factors like relevance, bid amount, and advertiser quality score, manipulated to prioritize Google's services. This case highlights the potential for algorithms to be used to unfairly disadvantage competitors.

  • Outcome: Google was fined €2.42 billion (later reduced to €1.49 billion).

  • Update 18/9/24: Google wins EU Antitrust Fine €1.49 billion ($1.66 billion) appeal

  • Reference: European Commission - Press Release: Google Search (Shopping)

• Google (Online Advertising, Search) DOJ Antitrust Case - 2023/24:

  • Case Details: Generally, Google has a monopoly in “general search services” and “general search text advertising.” (Note: The focus is broader, focusing on Google’s overall search and advertising practices vs The EU case which was narrower, focussing on specific contractual practices related to AdSense for Search, whilst the DOJ case is broader, targeting Google overall market behavior)

  • Reference: DOJ Case

  • Reference: DOJ Website for Trial Exhibits

  • Reference: DOJ Complaint (153 Pages)

  • Reference: Judge’s Decision (Aug 2024). Pending decision on what this will mean to Google’s business(es)

  • Reference: DOJ, States win Google Antitrust Case (Attached 286 Page Document). Politico’s Article

  • Another useful Channel to Follow and related updates (Tom Blakely’s Big Tech on Trial):

  • 

    Source: Big Tech On Trial

    The Stakes of the Third Google Antitrust Trial

• Topkins & Others (E-commerce):

  • Case Details: Several online retailers, including Topkins, were accused of using pricing algorithms to coordinate prices for posters and other products on Amazon Marketplace in 2015.

  • Algorithm: The algorithms were designed to monitor competitor prices and automatically adjust their own prices to match or slightly undercut them. This likely involved web scraping techniques and dynamic pricing rules, demonstrating how algorithms can facilitate tacit collusion.

  • Outcome: Fines and settlements were imposed on the involved companies.

  • Reference: New Yorker - When Bots Collude

  • Reference: DOJ - Price Fixing Antitrust

• RealPage (Real Estate):

  • Case Details: The DOJ is suing RealPage, a provider of property management software, alleging that its pricing algorithm facilitates price fixing among landlords.

  • Algorithm: RealPage's YieldStar software uses a complex algorithm that considers various factors like market rents, occupancy rates, and competitor pricing to recommend rent prices. This allegedly allows landlords to coordinate rent increases and avoid price competition.

  • Outcome: The lawsuit is ongoing.

  • Reference: New York Times - RealPage

  • Reference: DOJ Case

• Cendyn (Rainmaker) (Hotel Industry):

  • Case Details: Cendyn, a hotel revenue management software provider, is under investigation by the DOJ and FTC for allegedly facilitating price fixing among hotels.

  • Algorithm: Cendyn's Rainmaker software uses an algorithm that considers factors like demand, competitor pricing, and historical data to recommend room rates. It's alleged that this system enables hotels to coordinate prices and avoid discounting.

  • Outcome: The Price Fixing Claims were Terminated. Please read the grounds for termination which were distinguished from RealPage.

  • Reference: DOJ and FTC Investigating Hotel Pricing Software Provider Cendyn

  • Reference: Algorithmic Price Fixing Claims Terminated

• Express Scripts, CVS Caremark, OptumRx (Pharmaceutical):

  • Case Details: Pharmacy benefit managers (PBMs) like Express Scripts, CVS Caremark, and OptumRx are facing scrutiny for their pricing algorithms that allegedly favor certain drugs based on rebates from pharmaceutical companies.

  • Algorithm: PBMs use complex algorithms to determine which drugs are included in their formularies (lists of covered drugs) and at what cost-sharing levels. These algorithms often prioritize drugs that offer higher rebates, even if they are not the most clinically effective or cost-effective for patients.

  • Outcome: Increased transparency requirements have been implemented, but concerns remain. FTC Action pending.

  • Reference: FTC Pharmacy Benefit Managers: The Powerful Middlemen Inflating Drug Costs and Squeezing Main Street Pharmacies (73 Page Document)

  • Reference: WSJ: FTC to Sue Three Largest PBMs

• Various Airlines (Airline Industry):

  • Case Details: Several airlines have been accused of using revenue management systems (RMS) to coordinate prices and avoid price wars.

  • Algorithm: Airline RMS algorithms analyze demand, historical booking data, competitor pricing, and other factors to dynamically adjust ticket prices. While these systems are generally legal, concerns arise when airlines use them to signal pricing intentions to competitors or engage in tacit collusion.

  • Outcome: Investigations have not found sufficient evidence of a cartel, but concerns about algorithmic collusion persist.

  • Reference: Three Algorithms in a Room (The American Prospect)

• Booking.com (Hotel Industry - EU):

  • Case Details: Booking.com was investigated by EU authorities for allegedly manipulating hotel room prices through its dynamic pricing algorithm.

  • Algorithm: Booking.com's algorithm adjusted prices based on factors like demand, competitor pricing, and the hotel's availability. Concerns arose that this system could lead to inflated prices and reduced consumer choice.

  • Outcome: Booking.com adjusted its practices to address the concerns raised by the investigation.

  • Reference: European Commission Hotel Monitoring Report (43 Pages)

  • Reference: EU Says Booking.com must comply with strict tech rules

• Amazon (E-commerce - EU) - 2019/20:

  • Case Details: The EU investigated Amazon for abusing its market dominance through its "Buy Box" algorithm.

  • Algorithm: The Buy Box algorithm determines which seller's offer is featured prominently on a product page, significantly influencing purchasing decisions. It was alleged that Amazon's algorithm favored its own products and disadvantaged third-party sellers.

  • Outcome: Amazon has settled and is required to adjust its practices to ensure fair competition.

  • Reference: European Commission - Case (49 Pages)

  • Reference: Antitrust Commission accepts Commitments

  • Reference: Amazon Settles EU Allegations

• Amazon (FTC Sues Amazon for Illegally Maintaining a Monopoly) - Sept 2023

  • Case Details: The Federal Trade Commission and 17 state attorneys general is suing Amazon.com, Inc. alleging that the online retail and technology company is a monopolist that uses a set of interlocking anticompetitive and unfair strategies to illegally maintain its monopoly power. The FTC and its state partners say Amazon’s actions allow it to stop rivals and sellers from lowering prices, degrade quality for shoppers, overcharge sellers, stifle innovation, and prevent rivals from fairly competing against Amazon.

  • Outcome: Pending

  • Reference: FTC’s Case (172 Pages)

  • Reference: Lina Khan - Yale Law Journal “Amazon’s Antitrust Paradox” (96 Pages)

  • Reference: Lina Khan - Columbia Law “The Separation of Platforms and Commerce” (127 Pages)

Reasons for Antitrust Breaches

The cases outlined above demonstrate the potential for pricing algorithms to facilitate anti-competitive behavior, leading to antitrust breaches. Several factors contribute to this risk:

Collusion and Price Fixing

Pricing algorithms can be designed, intentionally or unintentionally, to facilitate collusion and price fixing among competitors. This can occur in several ways:

• Explicit Collusion: Competitors may secretly agree to use algorithms that coordinate prices, ensuring that they all charge similar amounts and avoid price wars. This can be achieved through direct communication or the use of a shared algorithm or platform.

• Tacit Collusion: Algorithms can facilitate tacit collusion, where competitors independently adopt similar pricing strategies without explicit communication. For example, algorithms that monitor competitor prices and automatically adjust to match or slightly undercut them can lead to a situation where prices converge across the market.

• Algorithmic Signaling: Algorithms can be used to signal pricing intentions to competitors. For example, an airline might use its algorithm to temporarily increase prices on a particular route, signaling to its competitors that it is willing to raise fares. This can lead to a coordinated price increase across the market without explicit communication.

Market Power Abuse

Dominant companies can leverage their algorithms to unfairly manipulate prices and exclude competitors. This can occur through:

• Predatory Pricing: A dominant company could use its algorithm to set prices below cost to drive out smaller competitors. Once the competition is eliminated, the company can then raise prices to recoup its losses and enjoy monopoly profits.

• Price Discrimination: Algorithms can be used to charge different prices to different customers based on factors like their location, demographics, or purchasing power. This can be used to extract higher prices from consumers with less elastic demand or to target vulnerable populations.

• Personalized Pricing: While personalized pricing can offer benefits to consumers, it can also be used to exploit market power by charging higher prices to consumers who are less price-sensitive or who have fewer alternatives.

Lack of Transparency

The opacity of many pricing algorithms hinders regulatory scrutiny and makes it difficult to detect anti-competitive behavior. This lack of transparency can arise from several sources:

• Proprietary Algorithms: Many companies treat their pricing algorithms as trade secrets, making it difficult for regulators or researchers to access and understand their code.

• Complex Algorithms: Even when access to the code is granted, the complexity of many algorithms can make it challenging to understand how they function and what factors they consider.

• Data-Driven Algorithms: Many algorithms rely on vast datasets that are not publicly available, making it difficult to assess the inputs that are driving price changes.

Inside the Algorithms (Illustrative Examples)

To better understand how pricing algorithms can be used for anti-competitive purposes, let's examine some simplified examples.

Disclaimer: These examples in pseudocode are for illustrative purposes only. Real-world pricing algorithms are significantly more complex and often involve machine learning, artificial intelligence, and vast (streaming) datasets.

1. Competitor-Based Pricing:

This algorithm adjusts prices based on competitor prices, aiming to match or undercut them.

# Input: Competitor prices (list), desired profit margin
# Output: Adjusted price

def competitor_pricing(competitor_prices, profit_margin):
  lowest_competitor_price = min(competitor_prices)
  adjusted_price = lowest_competitor_price * (1 + profit_margin)
  return adjusted_price

# Example usage:
competitor_prices = [10.00, 12.50, 11.00]
profit_margin = 0.15 # 15%
adjusted_price = competitor_pricing(competitor_prices, profit_margin)
print(adjusted_price) # Output: 11.5

Potential Issue: While seemingly benign, if widely adopted across a market, this type of algorithm can lead to tacit collusion, where prices converge and competition is stifled.

2. Demand-Based Pricing:

This algorithm adjusts prices based on real-time demand, increasing prices during periods of high demand and decreasing them during periods of low demand.

# Input: Current demand (integer), base price, demand elasticity
# Output: Adjusted price

def demand_pricing(demand, base_price, elasticity):
  if demand > 100: 
    price_multiplier = 1.1  # Increase price for high demand
  elif demand < 50:
    price_multiplier = 0.9 # Decrease price for low demand
  else:
    price_multiplier = 1.0
  adjusted_price = base_price * price_multiplier
  return adjusted_price

# Example Usage:
demand = 120
base_price = 20.00
elasticity = -0.5  # Assume demand is relatively inelastic
adjusted_price = demand_pricing(demand, base_price, elasticity)
print(adjusted_price) # Output: 22.0

Potential Issue: While demand-based pricing can be a legitimate business strategy, it can also be used to exploit consumers during periods of high demand, such as emergencies or natural disasters.

3. Collusive Pricing (Hypothetical):

This hypothetical algorithm demonstrates how algorithms could be used for explicit collusion, where competitors agree on a price range and adjust their prices accordingly.

# Input: Competitor price (float), agreed price range (tuple)
# Output: Adjusted price

def collusive_pricing(competitor_price, agreed_range):
  min_price, max_price = agreed_range
  if competitor_price < min_price:
    adjusted_price = min_price 
  elif competitor_price > max_price:
    adjusted_price = max_price
  else:
    adjusted_price = competitor_price # Match competitor within the range
  return adjusted_price

# Example usage (illegal):
competitor_price = 9.50
agreed_range = (10.00, 12.00)
adjusted_price = collusive_pricing(competitor_price, agreed_range)
print(adjusted_price) # Output: 10.0

Note: This collusive pricing example is hypothetical and intended to illustrate how algorithms could be used for illegal purposes. It is not intended as a guide or endorsement of such practices.

Note: These simplified examples highlight how algorithms can be designed to achieve specific pricing objectives, including those that may be anti-competitive. Understanding the logic behind these algorithms is crucial for regulators and policymakers to effectively address the potential risks they pose.

Checklist for Identifying Potential Issues

Given the complexity and opacity of pricing algorithms, it can be challenging for regulators, businesses, and consumers to identify potential anti-competitive practices. This checklist provides a framework for assessing the potential risks associated with algorithmic pricing:

Algorithm Design and Functionality

• Collusive Intent: Are algorithms designed to explicitly or implicitly coordinate prices with competitors? Do they include features that facilitate communication or signaling between competitors?

• Competitor Data: Do algorithms analyze competitor pricing data in a way that could facilitate collusion? Do they automatically match or undercut competitor prices?

• Market Power Exploitation: Are algorithms designed to exploit market power by, for example, setting predatory prices, engaging in price discrimination, or creating barriers to entry for smaller competitors?

Market Structure and Competition

• Market Concentration: Is the market concentrated, with a few dominant players? Could the use of algorithms exacerbate existing market power imbalances?

• Barriers to Entry: Do algorithms create barriers to entry for new competitors, for example, by making it difficult for them to compete on price or access necessary data?

• Innovation and Dynamism: Does the use of algorithms stifle innovation and reduce the dynamism of the market?

Transparency and Explainability

• Algorithm Transparency: Are companies transparent about how their algorithms work? Do they provide clear explanations of the factors that influence pricing decisions?

• Explainability: Can the outputs of the algorithms be explained and justified? Can regulators and auditors access and understand the algorithms' code and data inputs?

• Auditability: Are algorithms designed to be auditable, allowing regulators to track pricing decisions and identify potential anti-competitive behavior?

Impact on Consumers

• Price Increases: Are consumers facing artificially inflated prices or reduced choices due to algorithmic pricing?

• Price Discrimination: Are there concerns about price discrimination based on consumer characteristics, such as location, demographics, or purchasing power?

• Consumer Welfare: Is the use of algorithms improving consumer welfare by, for example, offering lower prices, greater convenience, or more personalized products and services?

Note: This checklist provides a framework for evaluating the potential risks associated with pricing algorithms. It is not exhaustive, and the specific factors to consider will vary depending on the industry, market structure, and specific algorithm in question. However, it can serve as a starting point for regulators, businesses, and consumers to engage in a more informed discussion about the responsible use of pricing algorithms.

Conclusion

Pricing algorithms are transforming the way businesses set prices and interact with consumers. While these technologies offer potential benefits in terms of efficiency, responsiveness, and personalization, they also pose significant risks to fair competition and consumer welfare. The cases examined in this document demonstrate the potential for algorithms to facilitate collusion, price fixing, and market power abuse.

The opacity and complexity of many pricing algorithms make it challenging for regulators to detect and address anti-competitive behavior. This necessitates a proactive approach to regulation, focusing on transparency, explainability, and accountability. Businesses must be transparent about how their algorithms work and ensure that they are not designed to facilitate anti-competitive practices. Regulators need to develop tools and expertise to effectively monitor and audit algorithmic pricing, ensuring that markets remain fair and competitive.

Furthermore, ongoing research and public discourse are crucial for fostering a deeper understanding of the implications of algorithmic pricing. This includes exploring the ethical dimensions of algorithmic price discrimination and the potential impact on consumer choice and market dynamism.

Ultimately, the responsible use of pricing algorithms requires a collaborative effort between businesses, regulators, researchers, and consumers. By working together, we can harness the benefits of these technologies while mitigating the risks they pose, ensuring that markets remain fair, competitive, and beneficial to all stakeholders in the digital age.