The attention economy earns $600 billion annually. The externalities it generates may cost more. Here is what a complete social cost accounting requires — and what it reveals.
Social cost accounting applies the methods of environmental economics to non-environmental harms: it attempts to monetize all costs and benefits imposed on third parties by a given economic activity, including those not captured in market prices. For environmental applications, this methodology is well-developed — the social cost of carbon, for example, is a government-maintained estimate of the economic costs of one additional ton of CO₂ emissions, used to evaluate climate policy. Applying analogous methods to the attention economy requires specifying the relevant cost categories, identifying appropriate monetization approaches for each, and aggregating with appropriate uncertainty characterization.
This paper presents the structure of a social cost accounting for the attention economy — the categories, the monetization approaches, and the resulting order-of-magnitude estimates — rather than a single-number definitive valuation. Single-number valuations of complex social phenomena mislead by implying precision that is not available. What the accounting structure reveals is the order of magnitude and the direction of the finding: at best available estimates, the externalized social costs of the attention economy appear to substantially exceed its private revenues, suggesting a net negative social value even before accounting for positive externalities.
The healthcare costs attributable to platform-associated mental health conditions — anxiety disorders, depression, eating disorders, sleep disruption, and suicide — estimated at $50–150 billion annually in the US, with substantial uncertainty in the attribution fraction (the share of total mental health costs attributable to platform use versus other contributing factors). Global estimate: $200–600 billion annually at proportional scaling, with the largest concentrations in high-income countries with high social media penetration.
The productivity costs from attention fragmentation in knowledge work, estimated at $650 billion or more annually in the US alone from digital distraction effects, combined with educational outcome reductions estimated at $100–200 billion in reduced lifetime earnings from measurable test score decrements attributable to device presence during schooling. Combined US estimate: $750 billion–$1 trillion annually. Global estimate: substantially larger given the larger working-age population affected.
The costs of local journalism collapse — measured in higher municipal borrowing costs, reduced local government accountability, and measurable corruption increases in news-desert jurisdictions — are estimated at tens of billions annually in the US. The costs of misinformation spread through platform recommendation systems — measured in vaccine hesitancy, public health compliance reductions, and election administration burdens — are harder to monetize but have been estimated in the range of $50–100 billion annually for health misinformation alone. Political polarization costs — measured in reduced legislative productivity, increased political violence risk, and reduced institutional trust — are the most speculative but potentially the largest democratic cost category.
The cumulative human capital loss from attentional capacity impairment during developmental years is the most speculative externality category — it requires long-horizon projections of lifetime income differences between cohorts with different developmental digital exposure levels. Conservative estimates based on measured test score differences and established relationships between educational outcomes and lifetime earnings suggest the cohort-level losses may be in the hundreds of billions cumulatively, though the annual flow equivalent is smaller.
A complete social cost accounting must include platform positive externalities, which are real and substantial:
Social connection value: The welfare value of maintaining relationships with geographically dispersed family, friends, and communities — quantifiable through studies of relationship satisfaction and its determinants, and the documented welfare costs of social isolation — is significant and represents a genuine positive externality of platform social connection infrastructure.
Information access value: Access to search engines and encyclopedic content provides welfare value estimated in studies of willingness to pay for internet access. The informational productivity gains from rapid information retrieval are real.
Economic opportunity value: Platform infrastructure enables creator economy employment (estimated 50 million people globally earning some income from content creation), small business advertising access, and marketplace participation that reduces friction in economic exchange.
Communication infrastructure value: Platform communication tools — messaging, video calling, group coordination — provide genuine social infrastructure value, particularly in lower-connectivity regions where alternative communication infrastructure is limited.
Estimating positive externalities in comparable monetary terms suggests they may range from $200 billion to $1 trillion annually in the US, with the largest component being social connection value (which is large but methodologically difficult to monetize) and information access value (which is large but substantially captured in GDP already through productivity measurements).
These externality estimates are speculative and politically motivated. Social cost accounting produces whatever number its commissioners need. The methodology is too uncertain to support policy conclusions.
The uncertainty critique is methodologically serious and deserves a direct response. Social cost accounting is indeed uncertain — all estimates in this paper carry wide confidence intervals, and the true values could be substantially higher or lower. The methodological critique does not, however, justify ignoring the accounting exercise. Environmental policy would be paralyzed if it required the precision the objection demands — the social cost of carbon has wide uncertainty bounds and is nonetheless appropriately used in policy evaluation. The relevant question is not "is the estimate precisely correct?" but "does the estimate, given its uncertainty, support a policy conclusion?" At the lower bounds of defensible externality estimates and the upper bounds of defensible positive externality estimates, the net finding — that externalized costs may exceed private revenues — is sufficiently robust to support the conclusion that market failure is occurring and that internalization interventions are economically justified.
Combining the cost categories — healthcare externalities ($50–150B US), human capital externalities ($750B–$1T US), democratic/information costs ($100–200B US), and developmental costs (smaller annual flow) — against the positive externalities ($200–$500B US for the most monetizable components) produces a net externalized cost estimate in the range of $700 billion to $1.5 trillion annually in the US.
Global attention economy revenues are approximately $600 billion. US attention economy revenues — the US being approximately 40% of the global digital advertising market — are approximately $250 billion. The comparison: at lower-bound externality estimates, US attention economy external costs substantially exceed US attention economy revenues. At mid-range estimates, they exceed global revenues. This is not a finding that the attention economy produces no value — it is a finding that the private revenues capture a small fraction of the total social welfare effects, with a large negative externality gap that constitutes a market failure of substantial scale.
The wide uncertainty ranges in all externality estimates reflect a fundamental data gap: platform companies hold the behavioral and outcome data that would be necessary for precise externality estimation and do not make it available to researchers. Platform-specific research on user mental health outcomes, attention impacts, and behavioral consequences would allow substantially more precise attribution than is currently possible from the general population studies available to researchers. The data opacity is not accidental — detailed externality data would strengthen the policy case for internalization measures that reduce platform revenues.
Mandatory platform data access for independent research — proposed in various forms in the EU Digital Services Act, US legislative proposals, and academic research frameworks — would reduce the data gap and enable more precise externality estimation. The political obstacles to mandatory data access are documented in Series III (PE). The consequence of the data gap is policy design based on estimates substantially more uncertain than they need to be — which introduces systematic bias toward policy inaction, because uncertain estimates are easier to contest than precise ones.
The True Cost Estimate has specific implications for the internalization interventions described in EX-005. The ratio of externalized costs to private revenues determines the appropriate magnitude of internalization: if externalized costs are 3–6x private revenues, the internalization mechanism must impose costs proportional to that ratio to fully internalize the externality. A Pigouvian tax sufficient to internalize medical and human capital externalities alone would, at mid-range estimates, need to be substantial — potentially in the range of 30–60% of platform revenue — to create financial incentives proportional to the documented social costs.
This order of magnitude underscores why the political economy of internalization is so challenging: the financial stakes for platform companies are existential, creating maximum resistance to any effective internalization mechanism. Understanding the scale of the market failure is prerequisite to understanding why the political obstacles to addressing it are so formidable — and why partial measures that reduce the internalization requirement below the range justified by the evidence are politically easier but economically insufficient.
Internal: This paper is part of The Externality Record (EX series), Saga VIII. It draws on and contributes to the argument documented across 55 papers in 12 series.
External references for this paper are in development. The Institute’s reference program is adding formal academic citations across the corpus. Priority papers (P0/P1) have complete references sections.