Social Media Can Suppress or Fuel Misinformation at Its Source, Depending on Civic Engagement
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.
Ruth: We're looking at the paper: "Social Media Can Suppress or Fuel Misinformation at Its Source, Depending on Civic Engagement".
Warren: Here's the gist of the paper. Misinformation can be suppressed by social media when public scrutiny is high, but it can also exacerbate data manipulation in regions with low oversight <ref:2506.16355#pg3>.
Ruth: This paper investigates this dual potential in the context of local governments’ GDP reporting in China, where data falsifications are widespread <ref:2506.16355#pg3>.
Warren: The Core Finding on Scrutiny The study demonstrates that social media reduces GDP manipulation when paired with active civic engagement, but this effect is stronger in regions with higher political oversight <ref:2506.16355#pg5>.
Ruth: Specifically, the authors find that "social media reduces GDP manipulation in regions with high public scrutiny" <ref:2506.16355#pg5>.
Title and authors: Ruth: So we're looking at this paper titled "Social Media Can Suppress or Fuel Misinformation at Its Source, Depending on Civic Engagement." It’s coming out of a study focusing on how local governments in China report their GDP and the effect social media has on that reporting.
Warren: That sounds like a big topic for data manipulation. What's the main tension they are exploring here, Ruth?
Ruth: Well, it’s this dual potential we mentioned: social media can either suppress misinformation when there's high public scrutiny or it can actually make data manipulation worse in places where there isn't much oversight. It challenges the idea that social media is just a megaphone for falsehoods.
Iris: So, they're looking at how the conditions surrounding public attention determine whether social media helps or hurts this kind of economic reporting fraud. What’s the specific context they chose to study?
Ruth: They focused on local governments in China and their GDP reporting between two thousand eleven and two thousand nineteen <ref:2506.16355#pg2>. The initial finding is that when public scrutiny over political matters is high, using social media actually reduces the extent of misreporting.
Warren: That sounds counterintuitive if you think of social media as just amplifying noise, but they’re suggesting it might increase the cost for bad actors to keep lying if people are watching.
Iris: They find that this reduction in manipulation happens because social media increases the cost of misinformation by facilitating greater information disclosure and bottom-up monitoring. That’s where the mechanism seems to lie.
Ruth: Exactly, it's not about some inherent honesty from the government, but rather mandated transparency through public disclosure that makes it harder to hide things when people are paying attention. They test this by looking at posts that actually disclose specific economic data versus those that don't.
Warren: That’s an interesting way to measure it, focusing on the content itself rather than just the presence of the platform. So, if a post is just pure noise, it doesn't help reduce manipulation?
Iris: Right, they found that only posts disclosing specific economic data are associated with a decrease in misreporting. They even did a placebo test using pollution emission data to make sure this effect wasn't just some general social media activity.
Ruth: And they confirmed that there was no significant association with the pollution data, which reinforces the idea that it’s about disclosure, not some underlying change in government integrity or social media itself. This is a key part of their finding on how to reduce fraud.
Warren: So, what happens when we look at the boundaries for this effect? Does public scrutiny just matter generally, or is there a specific level of oversight that changes everything?
Iris: That’s where they introduce a moderating boundary condition. The authors show that the fraud-reducing effect of social media is stronger in regions with higher levels of political oversight.
Ruth: And conversely, in cities where the public scrutiny falls below the sample mean, using social media is associated with increased manipulation. That suggests that when oversight is weak, local officials might actually exploit social media to boost those inflated economic statistics for their performance narratives.
Warren: So it’s a tipping point; if you have enough watching eyes, social media helps correct things; if you don't have enough watching eyes, it fuels the problem. That’s a pretty clear mechanism for how context matters.
Iris: Precisely. It reframes how we think about digital transparency by showing that the impact of these platforms is entirely dependent on the institutional conditions around them, extending research on public accountability into more nuanced territory.
Ruth: Looking ahead, they suggest that curbing this kind of misinformation shouldn't rely just on reactive strategies like content moderation. Instead, they point toward ex ante mechanisms like mandated transparency and participatory scrutiny to build informational integrity.
Warren: That’s a practical policy implication—focusing on setting the rules before the manipulation starts rather than just cleaning up after it happens online. What about what this means for future research?
Iris: Future work is suggested in several directions, including seeing if this mechanism holds true in other areas like corporate disclosures or electoral contexts. They also suggest a need to explore the micro-foundations of social media accountability, focusing on low-cost, scalable mechanisms for enabling scrutiny.
Ruth: That points toward things like crowdsourced verification systems or expert signaling systems as ways to enhance that public oversight we talked about earlier. It’s about enabling knowledgeable observers.
Warren: So the paper suggests that digital technologies interact with existing institutional structures to produce context-dependent outcomes, which is a broader framework for understanding these issues across different fields.
Iris: That’s right, it provides a pathway through which social media can enhance real-time accountability by linking digital disclosure with measurable reductions in data fraud in systems where top-down oversight might be limited but crowd-based accountability is strong.
Ruth: So to wrap up this paper, the core message of "Social Media Can Suppress or Fuel Misinformation at Its Source, Depending on Civic Engagement" is that the impact hinges entirely on whether public scrutiny is high enough to turn social media into a tool for transparency or an amplifier for fraud.
Warren: It’s a reminder that technology doesn't automatically solve problems; it interacts with the institutional landscape to produce context-dependent results.
Iris: And we have seen how information disclosure, when paired with strong public oversight, creates measurable reductions in data manipulation in these specific economic reporting contexts.
Ruth: So we leave you with this idea that empowering public oversight and mandating transparency are more effective tools than just deplatforming when dealing with digital misinformation.
Warren: That’s it for this paper on "Social Media Can Suppress or Fuel Misinformation at Its Source, Depending on Civic Engagement." Next up, we'll look at how teams can coordinate their investments for better production.
The paper's summary: Ruth: So we've talked about how public scrutiny can either help or hurt data reporting when social media is involved in GDP figures in China, and now we need to look at the actual summary of this paper, "Social Media Can Suppress or Fuel Misinformation at Its Source, Depending on Civic Engagement."
Warren: Right. So what’s the gist of what they actually found about this whole dynamic?
Ruth: They're basically saying that social media isn't a one-size-fits-all tool for fighting misinformation. It depends entirely on whether there's enough public attention or civic engagement around it.
Warren: So, if people are paying attention, does social media help reduce the problem?
Ruth: Yes, but only when that scrutiny is high enough. The paper shows that social media reduces GDP manipulation specifically in regions where public scrutiny is high. It’s not just about having a platform; it’s about what the public is actually looking at on that platform.
Warren: So, the mechanism isn't just some general awareness; there has to be specific action or disclosure happening?
Ruth: Exactly. The study found that this reduction happens because social media only helps when people are posting things that disclose specific economic data. Posts that just spread random noise don't do much to lower the manipulation numbers.
Warren: So they’re prioritizing posts with actual data over general chatter? That sounds like a useful way to think about how we filter information online.
Ruth: It is. The paper argues against the idea that you can just regulate the whole platform and expect it to work. Instead, it suggests that ex ante mechanisms—like making transparency mandatory or encouraging people to actively scrutinize data—are what actually create that fraud reduction effect when public scrutiny is strong.
Warren: So, the real takeaway for us is that if we want social media to be a tool for accountability, we can’t just rely on content moderation alone; we need to build systems that encourage specific types of disclosure.
Ruth: That's it. It shifts the focus from just taking things down online to creating environments where good information becomes visible and scrutinized by the public.
Warren: It makes sense, because if you have low oversight, those local officials can use social media to pump up their numbers without anyone checking them in a meaningful way.
Ruth: Precisely. And that context-sensitive finding—that high oversight amplifies the benefit of social media—is a big piece for anyone looking at how digital technologies actually affect real-world institutions.
Warren: So, if we look at this idea, what does it mean for how we think about regulating these systems in general?
Ruth: It means that building informational integrity isn't just a technical problem; it’s an institutional one. You need to design systems that invite and enable knowledgeable observers to look at the data.
Warren: That leads us right into something else, though, because this whole idea of "enabling scrutiny" is really about making sure the right people are looking at the right things. We’re going to talk next about how we can actually build those systems that encourage that kind of active involvement without it becoming a mess.
The paper's improvements: Tom: We're now moving on to what the authors actually suggest we do next, focusing on how they improve this paper's approach to social media and misinformation control.
Ruth: So, the paper points toward making these interventions more targeted and less reactive. They aren't just saying "moderate this"; they are suggesting ex ante mechanisms that are built into the system from the start.
Warren: What kind of ex ante mechanisms are we talking about here? Because if it’s just reacting to bad posts, we already know that doesn't work in the long run.
Ruth: They want mandates for transparency and participatory scrutiny. Think of it as building a structure where public oversight is required before the data gets reported widely.
Warren: So, they’re pushing for laws or rules that force disclosure rather than just hoping people are good enough to police themselves online?
Ruth: Exactly. The whole point is that when you combine high scrutiny with mandated transparency, you create a much stronger fraud-reducing effect on the GDP numbers. It's about changing the incentives for reporting in the first place.
Warren: That makes sense because it moves the responsibility from policing content to controlling how information gets created and shared in the first place. It’s about design, not just cleanup.
Ruth: Right. And that leads to a practical suggestion: we need low-cost, scalable ways to enable that scrutiny. The authors suggest looking at things like crowdsourced verification or expert signaling systems as tools for this.
Warren: So, instead of needing a massive government agency to police everything, maybe we can leverage the network itself—like getting verified economic experts to signal what data looks legitimate?
Ruth: That’s the direction. The idea is that digital technologies interact with these institutional structures in a specific way, and if we use those technologies to foster real-time accountability, it can help suppress misinformation at its source.
Warren: It's interesting because this gives us a path through which social media can enhance real-time accountability even in systems where top-down control is weak.
Ruth: That’s the big implication for policymakers and platform designers—the need to focus on building systems that invite and enable scrutiny by knowledgeable observers.
Warren: So, the next thing we need to figure out is how we actually build those kinds of crowdsourced verification systems that work reliably in practice.
Conclusion: Tom: So we're wrapping up this paper, "Social Media Can Suppress or Fuel Misinformation at Its Source, Depending on Civic Engagement." Basically, it’s about how public attention determines whether social media helps stop data manipulation or makes it worse in places like China.
Ruth: The core finding is that the impact of social media depends entirely on the level of public scrutiny. High oversight means social media can actually help reduce GDP manipulation because it forces more disclosure when people are watching closely.
Warren: So, if you have weak oversight, like in some cities mentioned in the paper, using social media amplifies those inflated economic statistics because there’s nobody to check them. That’s a clear risk for anyone running an economic strategy where reporting is key.
Iris: I just want to mention that the authors did a good job of making sure their findings were solid by testing it against pollution emission data and using instrumental variables to handle the endogeneity, which is pretty rigorous work for this kind of identification.
Ruth: They did, and what it changes for us is realizing that if we want digital tools to be good for accountability, the focus needs to be on building rules that require disclosure first. Not just waiting for people to self-regulate.
Warren: I agree. It’s about creating incentives where the cost of lying becomes higher than the benefit of manipulating the numbers, provided there’s some public eye on it to enforce those incentives.
Iris: From an econometric standpoint, what this means is that we should look for variations in how much public attention is focused on specific data points as our key driver for the fraud reduction effect.
Ruth: Exactly. And looking ahead, the paper suggests we need to think about designing low-cost ways to enable that scrutiny, like crowdsourced verification systems or expert signaling, instead of just relying on high-level platform policies.
Warren: That moves us into a really interesting technical space—how do you design a system where knowledgeable observers can effectively signal what's true without it becoming too costly to run?
Iris: That’s the next big challenge, because identifying those specific mechanisms that work reliably in the real world is where we need to focus our econometric efforts.
Ruth: Well, that brings us full circle back to how these digital technologies interact with institutional structures, showing that context really dictates the outcome of any digital intervention.
Warren: Yeah, it’s a reminder that technology doesn't just solve problems on its own; it has to be shaped by the environment around it to have any real impact.
Iris: This study provides a pathway through which social media can enhance real-time accountability by linking digital disclosure with measurable reductions in data fraud.
Ruth: That’s the gist of "Social Media Can Suppress or Fuel Misinformation at Its Source, Depending on Civic Engagement." It’s a very context-dependent story.
Gavin Wang, Haofei Qin, Xiao Tang, Lynn Wu
University of Texas at Dallas · Georgia Institute of Technology · Tsinghua University · University of Pennsylvania
econ.GN, q-fin.EC
Submitted: 2025-06-19
Updated: 2026-11-13
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 83/100
The gist: Misinformation can be suppressed by social media when public scrutiny is high, but it can also exacerbate data manipulation in regions with low oversight <ref:2506.16355#pg3>.
Key concepts
- GDP Manipulation
- This refers to the falsification of a region's Gross Domestic Product figures. Researchers measured this by comparing official reported GDP against satellite nighttime light intensity data, which serves as an independent measure of economic activity.
- Public Scrutiny
- This is the level of public attention and oversight applied to local government actions. The study found that high levels of public scrutiny are crucial; social media helps reduce fraud more effectively when citizens are actively watching and engaging with economic data.
- Information Disclosure
- The mechanism behind reducing fraud is not just social media use, but what is shared. Only posts that explicitly disclose specific economic data were linked to a decrease in misreporting, implying transparency and the sharing of raw information are key to accountability.
Terminology
Summary
Misinformation can be suppressed by social media when public scrutiny is high, but it can also exacerbate data manipulation in regions with low oversight <ref:2506. p. 3 We find that social media reduces
GDP manipulation in regions with high public scrutiny, but exacerbates it where scrutiny is low—
highlighting the critical role of public oversight in curbing misinformation. Further, only posts
disclosing GDP data—not general propaganda—are associated with a reduction in misreporting,
consistent with an information disclosure channel. While prior research has often portrayed social media as a channel
for spreading falsehoods1-10
, less is known about the conditions under which it may instead
constrain misinformation by enhancing transparency and accountability. Here we show this dual
potential in the context of local governments’ GDP reporting in China, where data falsifications
are widespread11-15
.Where the paper says it
Where the paper says it
The Core Finding on Scrutiny
The study demonstrates that social media reduces GDP manipulation when paired with active civic engagement, but this effect is stronger in regions with higher political oversight <ref:2506.16355#pg5>. Specifically, the authors find that Further, when
disaggregating article types, we find that only posts disclosing specific economic data reduce
manipulation (Columns 5-6), while non-disclosing posts—typically propaganda-oriented—are
positively associated with misreporting (Columns 7-8). This supports the proposed information
disclosure mechanism: transparency through specific data sharing deters manipulation.social media reduces GDP manipulation in regions with high public scrutiny
<ref:2506.16355#pg5>. The mechanism for this reduction is tied to information disclosure, as only posts disclosing specific economic data are associated with a decrease in misreporting <ref:2506. p. 8Where the paper says it
The Role of Information Disclosure
The empirical results suggest that the reduction in GDP manipulation stems from mandated transparency and public disclosure rather than inherent government honesty <ref:2506.16355#pg12>. The authors test this by decomposing social media activity into articles containing specific economic data versus those without such data, finding that only the former reduces manipulation <ref:2506. p. 6 7
variable indicating whether the government has adopted WeChat to make public announcements,
and social media activity, operationalized as the number of GDP-related articles published on their
official WeChat accounts. We further distinguish between articles that disclose specific economic
data and those that do not (e. Further, only posts
disclosing GDP data—not general propaganda—are associated with a reduction in misreporting,
consistent with an information disclosure channel. As a placebo, using pollution emission data—
also known to be falsified but not mandated for online disclosure13—show no effect, reinforcing
the specificity of the mechanism.Where the paper says it
Where the paper says it
The Moderating Boundary Condition
A key theoretical contribution is identifying the level of public scrutiny as a critical boundary condition for misinformation's impact <ref:2506.16355#pg5>. The authors show that 10
significant, indicating that the fraud-reducing effect of social media is stronger in regions with
higher levels of political oversight. However, in cities where scrutiny falls below the sample mean,
social media use is associated with increased manipulation. Overall, the findings suggest that social media reduces misinformation not only through
information disclosure but also by increasing the cost of deception—but only when public scrutiny
is high. In regions with low oversight, local officials may instead exploit social media to amplify
inflated economic statistics in efforts to attract investment, migrants, or promotions, thus
enhancing their performance narratives.the fraud-reducing effect of social media is stronger in regions with higher levels of political oversight
<ref:2506.16355#pg11>. Conversely, in cities where scrutiny falls below the sample mean, social media use is associated with increased manipulation <ref:2506. p. 9Where the paper says it
Where the paper says it
Methodological Rigor and Robustness
The researchers employed a comprehensive six-step strategy to ensure methodological rigor, beginning with measuring GDP manipulation using an elasticity-based ratio between reported GDP and satellite nighttime light intensity <ref:2506.16355#pg18>. They further validated this measure by using a ranking-based alternative where the 3 percentage points. This result is robust to instrumental variable estimation (Column 2), which addresses potential
endogeneity by using neighboring cities’ social media activity as an instrument. Further, only posts
disclosing GDP data—not general propaganda—are associated with a reduction in misreporting,
consistent with an information disclosure channel. As a placebo, using pollution emission data—
also known to be falsified but not mandated for online disclosure13—show no effect, reinforcing
the specificity of the mechanism.GDP Manipulation Level
is measured by the difference between rankings in nighttime light data and reported GDP data <ref:2506.16355#pg25>. To address endogeneity, they utilized instrumental variable (IV) techniques, such as using provincial average social media activity as an instrument <ref:2506. p. 7Where the paper says it
Where the paper says it
Implications for Policy and Theory
The findings offer practical insights suggesting that Empowering
public oversight, rather than simply content moderation or deplatforming, can enhance the
informational value of social technologies. Theoretically, our results identify a key boundary
condition in the study of misinformation: the level of public scrutiny.social media can help reduce misinformation when paired with active civic engagement
<ref:2506.16355#pg5>. Theoretically, the study reframes social media's role by showing its impact depends on institutional conditions, contributing to more context-sensitive theories of digital transparency and extends research on public accountability
<ref:2506.16355#pg11>. The results imply that curbing misinformation should rely on ex ante mechanisms like mandated transparency and participatory scrutiny rather than solely reactive strategies like content moderation <ref:2506. p. 3Where the paper says it
Future Research Directions
Future research is suggested across several dimensions, including examining the generalizability of this mechanism in other policy domains like corporate disclosures or electoral contexts <ref:2506.16355#pg14>. A deeper exploration into the micro-foundations of social media-based accountability is also needed, focusing on identifying These findings
challenge the prevailing view that social media primarily amplifies misinformation and instead
highlight the importance of civic engagement as a moderating force. Our findings show a boundary
condition for the spread of misinformation and offer insights for platform design and public policy
aimed at promoting accuracy and institutional accountability. This reframing contributes to more context-sensitive
theories of digital transparency and extends research on public accountability, information
disclosure, and the governance of digital platforms. Although our empirical setting is China—where strong bureaucratic incentives for data
manipulation make local governments a fitting context for studying the disciplining effect of
transparency technologies—the challenge of misinformation is global. This mechanism is likely relevant
in contexts where citizen engagement is high but formal oversight is weak, such as municipal
governments in emerging economies, early-stage startups, or non-governmental organizations
under donor pressure. As such, our study offers broader lessons on how social media can promote
institutional accountability beyond the Chinese context. The gray areas with diagonal
stripes show autonomous regions for ethnic minorities. They are excluded due to their differing political systems. Specifically, we
examine how mandated adoption of social media for public announcements influences the
manipulation of GDP data. We construct a comprehensive panel dataset that covers all prefecturelevel governments in China from 2011 to 2019, combining information on social media adoption
and usage for each local government, the officially reported GDP figures, and the real economic
performance inferred from satellite-based night-time light data use
11,34,35. In contrast, in regions with
low public scrutiny, adopting social media can exacerbate data manipulation. These findings
challenge the prevailing view that social media primarily amplifies misinformation and instead
highlight the importance of civic engagement as a moderating force.lowcost, scalable, and robust mechanisms for enabling scrutiny,
such as crowdsourced verification or expert signaling systems <ref:2506.16355#pg15>. This research speaks directly to platform design and public policy by highlighting the need for enabling scrutiny by knowledgeable observers <ref:2506. p. 1Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
- Civic Engagement Moderation for Misinformation Control
The improved system can differentiate between contexts where social media amplifies misinformation and those where it constrains it by incorporating a measure of public scrutiny as a moderating variable. This allows the AI to predict whether social media adoption will reduce GDP manipulation based on the local level of political oversight, as shown by the finding that social media reduces GDP manipulation in regions with high public scrutiny.
- Information Disclosure Mechanism for Data Fraud Detection
The system can be trained to prioritize and verify information based on its type; specifically, it should only act on posts disclosing specific economic data
when assessing potential GDP manipulation, while treating non-disclosing posts—typically propaganda-oriented—as positively associated with misreporting.
- Context-Sensitive Policy Recommendation
The improved system can generate context-sensitive policy recommendations by analyzing the level of public scrutiny. For example, it can advise that empowering public oversight, rather than simply content moderation or deplatforming, can enhance the informational value of social technologies
when public scrutiny is high.
Abstract
Social media is widely viewed as a channel for spreading misinformation, but its effects may begin earlier, by shaping whether misinformation is produced in the first place. We examine whether social media suppresses or amplifies the supply of misinformation and identify civic engagement as a key boundary condition. Using data on the GDP reporting of Chinese local governments from 2011 to 2019, we find that social media adoption reduces GDP misreporting only when civic engagement is high. This reduction is driven by posts that disclose specific economic data rather than promotional content. When civic engagement is low, however, social media adoption instead increases GDP misreporting. These findings reveal a dual upstream role of social media in the misinformation process: it can deter misinformation at its source when disclosures are exposed to engaged citizens, but can amplify its supply when civic engagement is limited.
Related papers
- Huge If True: Learning from the Size of the Surprise
- The Rebundling of Freelance Work in the AI Age
- Dynamic investment and coordination in teams: Theory and experimental evidence
- Does Frontier AI Strengthen Cyber Defense? Evidence from Project Glasswing
- Generative AI at Work
- What is Work For? Labor, Politics, and Meaning in the Age of AI