Research Highlights Article
June 19, 2024
Product variety and congestion effects online
The growth of mobile apps illustrates the importance of congestion externalities for online markets.
Source: Bloomicon
In 2008, Apple launched its App Store. By 2023, the platform had grown to support a trillion dollar industry for developers and offered millions of apps and games. While this explosion of online apps is a boon to consumers, it’s not without costs, according to a paper in the American Economic Journal: Microeconomics.
Author Daniel Ershov presents the first empirical evidence on the existence and magnitude of congestion effects online. He says that more apps in online markets directly reduces per app usage and downloads. His findings are based on a study of a redesign to Google Play, an online app store with over 100 million US consumers.
In the last 30 years, the internet has unlocked a new digital economy that economists are working to better understand. Previous research has highlighted the benefits of the enormous variety found in online markets, but has tended to overlook potential frictions.
“One of the main things that digitalization does is allow for the aggregation of information. Producers can put their products on a single platform and compete globally, and then consumers potentially have access to thousands or tens of thousands or millions of products,” Ershov told the AEA in an interview. “But when you're actually searching for something online, it can take you forever to do it.”
Mobile app usage in 2013
The chart below shows monthly mobile app usage by Android and iPhone users in the United States, just before the Google Play redesign, by age.
In spite of personalized recommendations and better search algorithms, consumers still have to spend substantial time and effort to find the right products. And as the number of products increases, the less visible new products become, creating what’s known as congestion externalities.
To test the size of these frictions, Ershov used data from AppMonsta.com consisting of daily snapshots of all apps on the US Google Play Store from January 2012 to December 2014.
App stores tend to group apps into game and nongame sections. In early 2014, Google Play reorganized its 6 game categories into 18, reducing the number of apps in each category. The number of categories for nongame apps was not changed. Comparing the downloads of game apps to nongame apps before and after the redesign shows the impact of product visibility on app usage.
In the short run, downloads increased for games relative to nongames. But more importantly, downloads increased the most for games in less populous categories, which suggests that having fewer choices in a category improved the ability of consumers to find apps they wanted to use.
In response, developers boosted the number of games in the store, which in turn increased product variety. Overall, the number of game apps increased by roughly 34 percent after the redesign.
The high-level takeaway is that congestion is real. But platforms can obviously do things to influence that. One of them is including more categories or changing the categorization structure. The other thing is using machine learning tools to try to predict what consumers really want and serve it to them more effectively, either when they search for stuff or when they're just browsing.
Daniel Ershov
Applying these findings to an economic model of app demand revealed how much consumers were losing to congestion externalities. Without congestion, US Google Play consumers as a whole gained nearly $70 million per year from the recategorization and increased product variety. However, congestion externalities nearly cut these gains in half.
While better platform design has likely brought the cost of the congestion externalities down in recent years, Ershov’s estimates help make explicit their potentially large costs. With the digital economy continually expanding, platforms will need to grapple with product visibility and limited attention among consumers.
“The high-level takeaway is that congestion is real. But platforms can obviously do things to influence that,” Ershov said. “One of them is including more categories or changing the categorization structure. The other thing is using machine learning tools to try to predict what consumers really want and serve it to them more effectively, either when they search for stuff or when they're just browsing.”
♦
“Variety-Based Congestion in Online Markets: Evidence from Mobile Apps” appears in the May 2024 issue of the American Economic Journal: Microeconomics.