Economics of Digital Markets: An Introductory Course for Data Scientists

Economics of Digital Markets: An Introductory Course for Data Scientists Tetyana Beregovska1 1Department of Computer and Data Sciences, Truman State University, 100 E Normal Ave., Kirksville, MO 63501 Abstract Newly formed digital markets create a multiplicity of jobs for data scientists, statisticians, and other professionals who work with data. All of their work revolves around data collected by businesses operating on digital markets: social media platforms, search engines, streaming services, instant messaging services, online gaming platforms and gaming consoles, credit card markets, and so on. All these markets have a common feature; they bring together different sides of a market to meet and interact. Most of them are two-sided markets because they enable two groups of market participants to interact with each other: players and developers of games, users of computer operating systems and applications developers, holders of bank cards and merchants that accept cards as a method of payment. This course offers students an opportunity to learn how digital markets work, why collect data and how they use said data in their business models, and what data scientists can do to ensure proper data processing. Key Words: digital markets, two-sided markets, digital platforms 1. Why Digital Markets? In the fall of 2024, I was asked to teach a higher-level course for economics majors at Truman State University on topics of my choice. After doing research and gathering information on what might be of interest for economists that would allow me to bring my expertise in Data Science and Statistics, I decided to offer a course that would help students understand newly emerged markets that change not only landscape of the service-oriented economy but also potentials of jobs creation and destruction in the future. After teaching this course and getting positive feedback from the students who appreciated the content of the course, I think that a similar course will be useful for Data Science students, for the material in this course will be exposing them to the information about industries and companies where they may eventually work and markets where they may take upon roles of independent consultants. There is an additional benefit for the Data Scientists to know how these markets function, what their business models are which is a domain knowledge – one of the main 3 components of the skill set for data scientists along with knowledge of math & statistics and knowledge of computer science. There is an occasional mentioning of the models that Data Science students are taught in terms of their application in different industries but I believe that more systematic and broader view of the modeling techniques based on the business models will significantly enhance Data Science students domain knowledge and contribute to better understanding of the work data scientists do within companies. 1.1 Digital Market Definition and Features Two-sided markets are markets that have an intermediary or a platform as a way for the interaction between buyers and sellers, while regular markets do not have anything standing between those sides of the market, i.e., they interact on their own. Digital markets are the two-sided markets with a digital intermediary or platform. Another feature that makes two-sided and digital markets different from the regular markets is the way that the decisions made by one of the market sides affect the outcomes for the other side (market price and quantity bought or sold depending on the side.) In economics this phenomenon is known as an externality. Why is it important? Because it makes strategic decision making for all market participants different and it makes the economic policymaking process different. Some of the market intermediaries offer products that can be related or unrelated to the competing products or competing platforms. These products can be incompatible, compatible, or have some sort of integration. When a customer chooses and able to consume rival products, it is called multihoming. On some markets decision about multi-homing is in the hands of customers. When a customer decides to switch to the consumption of a rival product, sometimes it is an easy thing to do but sometimes it is fairly hard. For example, it is easy to have both VISA and MasterCard in your wallet, but it is much more expensive to have both Xbox and Sony PlayStation at home. N other markets, the openness of a product is in the hands of an intermediary. For example, Microsoft Office Suite file formats are readable on Apple machines with macOS while Windows OS does not allow reading files produced by Apple machines unless a particular software/app is installed. It is important to understand this issue because it affects the pricing level and structure and the market structure, specifically, how prone the market is to monopolization, or ‘tipping’ in econ language. Limited multi-homing affects not only consumers like gamers, it also affect game developers due to differences in operating systems, programming languages and user experience & user interface adopted by different gaming consoles and difficulties of transferring games created for one console to other consoles. 1.2 Digital Markets Examples Two-sided markets have a fairly long history that started with the invention of newspapers selling news to its readers attracting them with exciting stories and selling extra printed space to advertisers who exploited the readers’ attention to the stories to advertizel their products. The newspaper was an intermediary, a platform, but not a digital one yet. Although the market outcomes on this twosided market were being decided upon by the newspaper in a way that took into consideration behavior of readers as well as behavior of advertisers and with realization that sometimes the only way to change the market outcomes for readers is to make the advertisers to change their behavior and vice versa. For example, if the newspaper wanted to charge higher price to advertisers it had to increase its pool of readers, in general or a certain kind, in some ways – either reducing the subscription price for readers or publishing more interesting and exciting stories. Contemporary newspapers with their continued printed-on-paper and ‘newly’ introduced online presence on the market for entertainment added the digital platform to its two-sidedness. Some markets existed for a while and have not seen a change of the intermediary form as much as newspapers. For example, credit cards market is being represented by Visa, Mastercard, American Express and Discover networks. The sides of the market are represented by credit card issuers – large banks such as JPMorgan Chase, Citigroup, Capital One, Bank of America, Wells Fargo, and U.S. Bank – and merchant acquirers that facilitate transactions for merchants – stores, businesses that use credit cards for transactions. Movement of the intermediaries to a digital world did not add new features to the market to the existing ones that characterize this market as a two-sided one but at the same time, new technologies affected accessibility of the credit cards by merchants and pricing level and structure 1 on the market. 1 Pricing on two-sided market is done on both sides, i.e., both buyers and sellers might be charged for participating on the market. The overall level of charges is called pricing level and the division of it between 2 sides or among 3 sides (on three-sided markets) is called pricing structure. The pricing structure may include positive and negative prices. Some markets emerged only because of the digitalization of the economy. The first example of such markets will be ridesharing market with rideshare platforms Uber, Lyft, ARRO, Via, Hitch, Curb, Flywheel, Wingz, Gett, Grab, and others. In general, this market brought in Pareto improvement to the economy (the overall market efficiency increased without negative impact on any economic agents) by utilizing excess capacity of drivers and satisfying excess demand of riders for rides. With this regard, we may consider this market also as an example of the gig economy that digitalization of the economy brought to life. While Uber is an intermediary on the two-sided market, UberEATS is an intermediary on the three-sided market where restaurants are providing food, drivers are delivering it to the consumers who are buying it. Another example of a three-sided market is market for offline gaming that is represented with gamers, game developers, and gaming consoles. The gaming consoles producers – Microsoft Xbox, Sony PlayStation, Nintendo, Sega, Atari -- serve as an intermediary in this market, for they facilitate interaction between gamers and game developers. Other examples of twoand three-sided markets are operating systems plarforms presented by Microsoft Windows, Apple OS, LINUX/UNIX that sell applications to users and buy applications from app developers. There is an interesting case of Adobe Acrobat Software that provides free Reader that allows reading PDF files but sells software that allows creating editing them. Social media platforms are presenting market similar to newspapers, TV and radio stations but rely upon their user to create attractive content and attract advertisers to pay for ads. Figure 1: A structure of three-sided markets with incomplete lists of markets participants on each side. A diagram in Figure 1 summarizes the structure of three-sided markets with incomplete lists of participants on different sides of the markets. 2. Predictive Models on Digital Markets 2.1 Pricing Structure and Predictive Models All companies operating on twoor three-sided markets have a task of ensuring growth of both or all sides of the market to make business viable and profitable. In order to attain this goal, companies use pricing level that allows them to cover expenses and make profits and pricing structure that brings in ‘balanced’ growth of market sides. This pricing structure is very flexible since it allows companies to treat the sides of the market differently, as in the case of offline gaming market where console is a loss leader (consoles are sold at a lower price than cost of production with profit margin on top of it) and gamers and game developers are a profit center. The pricing structure in the credit cards market is very similar -- cards holders are a loss center with negative price being charged in the form of cashback while merchants are a profit center with processing fees being charged that cover risks and handling costs. Other markets such as newspapers, rideshare services, prepared food delivery services, have more traditional pricing structures where both (or all for three-sided markets) sides serve as a profit center. But if we take a closer look at the pricing models for some companies, for example, Uber, we find that the pricing is dynamic, i.e., both pricing structure and pricing levels change as the market conditions change. It is well known that Uber uses predictive models to find appropriate pricing level and pricing structure as parts of its dynamic pricing model. 2.2 Common predictive models used in multi-sided markets All companies nowadays are making data-driven business decisions. The companies that operate on digital markets are doing it based on their needs and available data to help earn profits and to ‘balance’ different sides of the market. The most common predictive models are built for customer attrition and turn prediction, customer segmentation, personalized offers and rewards, and others. On the market for credit and debit cards customer attrition and turn prediction, credit risk scoring, transaction categorization, personalized offers and rewards predictive models are done using survival models, random forests, clustering, decision trees. On the market for ridesharing, churn prediction of riders and drivers is done using logistic regression, random forests, and neural networks. 2.3 Specific needs for predictive models in multi-sided markets Three-sided markets’ intermediaries have more diverse demands, so some of them have more specific tasks to deal with. For example, they use algorithms like regression, survival analysis, anomaly detection, time series analysis, or neural networks to estimate time-to-failure or probability of breakdown. Uber reports that it has approximately 400 active machine learning projects with more than 5K models in production. In earlier days Uber used XGBoost for ETA predictions, risk assessment, and pricing. Other ridesharing platforms also use graph-based routing algorithms for route optimization and traffic prediction, trip duration prediction. 3 Conclusion and Discussion Emergence of new markets presents challenge to those who will be joining job markets as well as all of economic agents that will be participating in economic activities in the future. Understanding how these markets function and what market outcomes are expecting to be realized will be giving an advantage to students in Data Science as well as other programs that prepare students for working with data. This course may start with an overview of what multi-sided markets are and how digitalization of the economy affects them. Then it may proceed with the features of these markets that determine business models that are adapted by companies operating on them. And then students get introduced to collections of predictive models that companies with different business models use. Separate space can be devoted to the development of AI-driven environments to enhance data processing and predictive modeling. References Cramer, J. and Krueger A. (2015) Disruptive Change in the Taxi Business: The Case of Uber. Princeton University, Working Paper #595. Rochet, Jean-Charles and Tirole, Jean. (2003) Platform Competition in Two-Sided Markets. Journal of European Economic Association. Rysman M. (2009) The Economics of Two-Sided Markets. Journal of Economic Perspective. From Predictive to Generative – How Michelangelo Accelerates Uber’s AI Journey. Posted by Kai Wang, Min Cai, Joseph Wang, Eric Chen. https://www.uber.com/blog/from-predictive-togenerative-ai/ (accessed on 10/15/2025)