I study the economics of social media and digital platforms, with a focus on online discourse and on measuring welfare and market power in these environments.
Working papers
Social dynamics of AI adoption
With Leonardo Bursztyn, Alex Imas, Aaron Leonard, Christopher Roth
[
| paper
]
Anxiety about falling behind can drive people to embrace emerging technologies with uncertain consequences. We study how social forces shape demand for AI-based learning tools early in the education pipeline. In incentivized experiments with parents—key gatekeepers for children’s AI adoption—we elicit their demand for unrestricted AI tools for teenagers’ education. Parental demand rises with the share of other teenagers using the technology, with social forces increasing willingness to pay for AI by more than 60%. Providing information about potentially adverse effects of unstructured AI use negatively shifts beliefs about the merits of AI, but does not change individual demand. Instead, this information increases parents’ preference for banning AI in schools. Follow-up experiments show that social information has little effect on beliefs about AI quality, perceived skill priorities, or support for bans, suggesting that effects operate through social pressure rather than social learning. Our evidence highlights social pressure driving individual technology adoption despite widespread support for restricting its use.
Digital news consumption: Evidence from smartphone content in the 2024 US elections
With Guy Aridor, Tevel Dekel, Ro'ee Levy, Lena Song
[
| paper
]
Using novel smartphone content data, we document that exposure to election-related content for the median American is arguably small. Moreover, exposure rarely comes from news apps and instead mostly occurs through non-traditional sources, such as social media and video apps. While the median was low, we find substantial heterogeneity: individuals in the 90th percentile consume over 50 times the content of those in the 10th percentile. A variance decomposition shows that apps play a role in driving exposure gaps (e.g., X versus Facebook), but individual characteristics (e.g., living in a swing state) are the dominant drivers of election-related exposure.
Measuring markets for network goods
With Leonardo Bursztyn, Matthew Gentzkow, Aaron Leonard, Filip Milojević, Christopher Roth
[
| paper
]
Market definition is challenging in settings with network effects, where substitution patterns depend on changes in network size. We study these effects in the context of social media. We conduct an incentivized experiment comparing substitution in response to a proposed U.S. TikTok ban, in which all users simultaneously leave the app, with substitution when only a single user deactivates. Consistent with a simple network model, we find substantially higher valuations of alternative social apps under a collective TikTok ban than under an individual TikTok deactivation. We then show that a collective time limit challenge, where peers jointly reduce TikTok or Instagram use, leads to more time spent on alternative social apps than has been observed in prior individual deactivation experiments. Together, our results suggest that individual-level substitution estimates can be an unreliable guide to market definition for network goods.
Non-user utility and market power: The case of smartphones
With Leonardo Bursztyn, Aaron Leonard, Filip Milojević, Christopher Roth
[
| paper
]
Firms can increase the demand for their products and consolidate their market power by increasing their users' utility but also by strategically decreasing the utility of competing products' users. We study this mechanism in the smartphone market, analyzing Apple’s strategy of differentiating messages sent to Androids with "green bubbles." In surveys with U.S. college students, we show that green bubbles are widely stigmatized and that a majority of both iPhone and Android users would prefer green bubbles to no longer exist. An incentivized deactivation experiment reveals that iPhone users have a significant willingness to pay to prevent their messages from appearing as green bubbles on other iPhones. Finally, we examine the market implications and document that removing green bubbles substantially increases respondents' likelihood of choosing an Android over an iPhone.
Experiments on social media
With Guy Aridor, Ro'ee Levy, Lena Song
Prepared for the Handbook of Experimental Methods in the Social Sciences
[
| paper
]
We provide a practical guide to designing, conducting, and analyzing experiments using social media platforms. First, we discuss the benefits and challenges of using the targeting capabilities of advertisements on social media to recruit participants for a large class of experiments. Next, we outline the different types of interventions and their advantages and disadvantages. Finally, we summarize available compliance and outcome data, as well as the main limitations and challenges involved in the design and analysis of social media experiments. Throughout, we provide technical details that are helpful when implementing these experiments. Overall, we argue that experiments on social media are powerful not only for studying economic issues around social media and online platforms but also for experiments studying economic behavior more broadly.
Toxic content and user engagement on social media: Evidence from a field experiment
With George Beknazar-Yuzbashev, Jesse McCrosky, Mateusz Stalinski
American Economic Review, Revise and resubmit
[
| paper
]
Most social media users have encountered harassment online, but there is scarce evidence of how this type of toxic content impacts engagement. In a pre-registered browser extension field experiment, we randomly hid toxic content for six weeks on Facebook, Twitter, and YouTube. Lowering exposure to toxicity reduced advertising impressions, time spent, and other measures of engagement, and reduced the toxicity of user-generated content. A survey experiment provides evidence that toxicity triggers curiosity and that engagement and welfare are not necessarily aligned. Taken together, our results suggest that platforms face a trade-off between curbing toxicity and increasing engagement.
The effect of content moderation on online and offline hate: Evidence from Germany's NetzDG
With Karsten Müller, Carlo Schwarz
[
| paper
]
Social media companies are under scrutiny for the prevalence of hateful content on their platforms, but there is little empirical evidence on the consequences of moderating such content. We study the online and offline effects of content moderation on social media using the introduction of Germany’s "Network Enforcement Act" (NetzDG), which fines social media platforms for failing to remove hateful posts, as a natural experiment. We show that the NetzDG reshaped social media discourse: posts became less hateful, refugee-related content became less inflammatory, and the use of moderated platforms increased. Notably, the law did not significantly reduce the overall activity of toxic users or alter conversation topics. Offline, the NetzDG caused a 1\% reduction in anti-refugee hate crimes for every standard deviation in far-right social media usage. Using a synthetic control approach, we document similar effects on overall hate crimes in Germany. In terms of mechanisms, we provide evidence that the NetzDG decreased hate crimes by reducing collective action rather than changing attitudes toward refugees.
The economics of content moderation: Evidence from hate speech on Twitter
[
| paper
]
Social media platforms remove posts and ban users to moderate content, but the consequences of such "speech policing" remain poorly understood. I examine the effect of moderating hate speech on user behavior and welfare through two pre-registered field experiments on Twitter. In the first experiment, randomly reporting posts for violating rules against hateful conduct increases their removal probability. While reporting does not deter subsequent hateful behavior by sanctioned users, it increases the engagement of those attacked by the posts, with evidence suggesting they notice the sanctions. The second experiment shows that changing users' perceived content removal does not change their consumer surplus. These findings align with a model of content moderation as a quality decision for platforms that increases engagement with ads. My results imply that content moderation does not necessarily deter users, but it can increase advertising revenue without impacting welfare, providing platforms with an incentive to moderate.
Selected publications
When product markets become collective traps: The case of social media
With Leonardo Bursztyn, Benjamin Handel, Christopher Roth
American Economic Review, 2025
Individuals might experience negative utility from not consuming a popular product. With such externalities to nonusers, standard consumer surplus measures, which take aggregate consumption as given, fail to appropriately capture consumer welfare. We propose an approach to account for these externalities and apply it to estimate consumer welfare from two social media platforms: TikTok and Instagram. Incentivized experiments with college students indicate positive welfare based on the standard measure but negative welfare when accounting for these nonuser externalities. Our findings highlight the existence of product market traps, where active users of a platform prefer it not to exist.
The economics of social media
With Guy Aridor, Ro'ee Levy, Lena Song
Journal of Economic Literature, 2024
We provide a guide to the burgeoning literature on the economics of social media. We first define social media platforms and highlight their unique features. We then synthesize the main lessons from the empirical economics literature and organize them around the three stages of the life cycle of content: (i) production, (ii) distribution, and (iii) consumption. Under production, we discuss how incentives affect content produced on and off social media and how harmful content is moderated. Under distribution, we discuss the social network structure, algorithms, and targeted advertisements. Under consumption, we discuss how social media affects individuals who consume its content and society at large, and explore consumer substitution patterns across platforms. Throughout the guide, we examine case studies on the deterrence of misinformation, segregation, political advertisements, and the effects of social media on political outcomes. We conclude with a brief discussion of the future of social media.
A model of harmful yet engaging content on social media
With George Beknazar-Yuzbashev, Mateusz Stalinski
AEA Papers & Proceedings, 2024
Why do social media users spend so much time consuming content that seemingly harms them? We build a simple model to argue that advertising-driven platforms can find it profitable to display content that harms users when it is complementary to their time spent on the platform. These incentives disappear, absent network effects, in the case of a subscription-based business model because harmful content reduces the willingness to pay for the platform. Our results warn against interpreting increases in engagement on social media as welfare increases.
Estimating the distaste toward price gouging with incentivized consumer reports
With Justin Holz, Eduardo Laguna Müggenburg
American Economic Journal: Applied Economics, 2024
Thirty-four states prohibit price increases during emergencies, and individuals take costly actions to report violators. We measure experimentally the willingness to pay to report sellers who increase prices of personal protective equipment. Over 75 percent of subjects pay to report, even if others are willing to buy at those prices. We argue that reports contain information about a desire to prevent or punish third-party transactions at unfair or illicit prices. Reports are partially driven by a distaste for firm profits or markups, implying that the distribution of surplus between producers and consumers matters for welfare.
Cash: A blessing or a curse?
With Fernando Alvarez, David Argente, Francesco Lippi
Journal of Monetary Economics, 2022
We use two quasi-natural experiments that encouraged the use of debit cards and facilitated the use of ATMs in Mexico to estimate the elasticity of crime and informality to the availability of cash as a means of payment. We then construct a simple model to quantify the private costs of restricting cash usage in the economy. Our model captures the degree of substitution between cash and other payment methods at the intensive and extensive margins. We estimate the welfare effects of restricting cash by means of three key inputs: i) the elasticity of substitution between cash and credit, ii) the share of expenditures in cash by type of good obtained from detailed micro data, and iii) the elasticity of crimes to the availability of cash as means of payment. The social benefits of restricting cash usage are driven by the reduction of some criminal activities. The costs arise from the distortions that the anti-cash regulation imposes on the individual choices regarding the means of payment. We find that the private costs of heavily taxing the use of cash in Mexico outweigh the social benefits that we identify.