Data Pricing and the Economics of Information Markets

They say that knowledge is power.  As the world becomes increasingly digitized, data has become a powerful tool, especially with the emergence of generative AI.  AI allows companies to rapidly comb through vast amounts of data and create quick, simple answers on how to improve functions and processes.  However, not all data is of equal value.  With almost everyone online for hours a day, we collectively generate massive tranches of data.  Some of it has immediate benefits for profit-seeking firms, such as our order histories at online retailers, while some may have little benefit without lots of complementary context.

Economics of Information Markets: Data as Highly Productive

Companies will pay for data that helps boost revenue.  This can be data that helps improve productive efficiency and reduce production costs, such as information on better use of equipment and resources, information on employee training techniques, or information on the best available prices for supplies.  It can also be data that helps boost revenue, such as information on current consumers tastes and preferences, information on potential bundling and networking deals with firms that produce complementary goods or services, and information on how to capitalize on the weaknesses of competitors’ substitute goods.

Limited Supply:  Data as a Non-Rival but Excludable Good

Because data is digital, firms can consume massive quantities of it and do not have to choose between Data Set A and Data Set B - they can purchase both if they have enough money!  This makes data essentially non-rivalrous.  Unlike deciding between which truck or drill press to purchase, a firm is not limited to one choice of data.  While some might believe that this reduces the value of data due to high supply, data can be limited to the first purchaser through exclusivity agreements.  A firm that buys a data set on consumer trends often wants its rivals to not have access to that same beneficial information.  To make that data set an excludable good, the seller charges a premium since it cannot make repeat revenue.  This reduces the supply of data and raises the price of data sets.

Complementary Goods: Bundling of Data Sets

Data often complements other data by providing important context.  For firms, information on consumer purchase histories is valuable…but additional information about those same consumers’ income levels and net worth makes it even more valuable!  Sellers of data sets can thus charge a premium by bundling two data sets that complement each other.  They can also make revenue from seemingly unimportant data by including it in a bundle, giving purchasing firms the chance to use powerful AI tools to make use of it through context.

Low Marginal Cost Makes it Easy to Consume More

The digital nature of data makes it easy for firms to gather many data sets and use software, including AI, to crunch them in search of potential operational improvements.  Once firms have the hardware to process data, such as computer servers, the marginal cost of adding another data set is very low.  Because firms don’t have to spend a lot of time or resources to manage an additional data set, demand for data is high due to convenience.  Therefore, they are willing to pay more for data than otherwise; it can be used quickly and easily.

Economics of Information Markets: Data Risks

Data can be valuable, but it can also be risky.  Buying data sets and crunching them to glean information about consumers, suppliers, and competitors is no guarantee of market success.  Companies that compile and sell data sets may be prone to puffery, or perhaps even outright fraud, when marketing their wares.  They may claim that a data set provides deep insights into consumer tastes and preferences, when the reality is far weaker.  

Subjectivity of Resource Value

Many economic resources can be valued objectively based on the value of output they generate.  Marginal productivity theory states that a resource’s value is determined by its increase in the firm’s revenue.  For example, the annual cost for a farmer to rent a combine would be just under the amount of additional income the farmer could make by using said combine.  If the combine can process $100,000 worth of crops per year, then the farmer is willing to rent and pay to operate it for up to that price; there is some profit to be gained.  

Using data as an informational tool is tougher to quantify.  Changes in revenue after using new data sets can be measured, but there are likely multiple factors vying for credit for the increased income: harder-working employees, newer capital goods, savvier marketers.  Or, perhaps, the increased income enjoyed by a firm is simply due to rising consumer incomes.  The difficulty in objectively valuing data sets as a business resource can soften demand for them, especially for smaller firms that have less disposable money for research-and-development.

Property Rights and Data Ownership

Who owns data?  Many firms, especially large retailers, generate lots of their own usable data about consumers.  Other firms wanting to know about consumer trends may have to purchase data sets…and run the risk of this data later being declared illegal.  Did the company selling the data sets have the appropriate rights to package and market this data?  Companies may worry that buying data sets could result in legal headaches as the firms and consumers whose data points are included declare “cease and desist.”  They may claim privacy violations and say that the data seller inappropriately accessed their information in the first place.

Minimum Efficient Scale of Data?

When new firms enter a market, they have to make lots of decisions quickly: how to operate, what supplies to purchase, what prices to charge, and to which demographics to market.  If high-quality data is not available, these firms run the risk of making costly errors that could drive them out of the market.  Therefore, not having enough money to rapidly accumulate a minimum efficient scale of data on how to operate effectively could be seen as a barrier to entry to new firms.  The need to have such large amounts of high-quality data could dissuade many entrepreneurs from entering a market, especially given the above risks of not being able to use the data with a reasonable guarantee of success.  With the stakes so high, the market becomes stagnant as fewer and fewer start-ups take the risk.