Digital assets are remapping the factors of production: despite comprising far fewer industries, the digital sector now contributes disproportionately to output and productivity growth, implying data, technology and infrastructure should join Smith’s labor–land–capital triad. Economic power is increasingly concentrated in ownership of these digital assets, and economists must rethink their foundational categories to analyze the information economy effectively.

250 years of Smith’s work: How digital platforms bring us back to the core of economic science

Yaroslav Kuzminov, Ekaterina Kruchinskaia · July 08, 2026 · Voprosy Ekonomiki

openalex theoretical low evidence 7/10 relevance Summary only summary available; pdf_status=not_found DOI Source

Arguing from a historical-theoretical perspective and descriptive BEA–BLS industry data (1997–2023), the paper contends that the digital sector’s outsized contribution to output and productivity growth requires expanding Smith’s factors of production to include data, technology, and infrastructure, shifting market power toward digital-asset owners and demanding a rebuild of core economic theory.

The year 2026 marks the 250th anniversary of Adam Smith’s “An inquiry into the nature and causes of the wealth of nations”, the work that had laid the foundations of economic science. The three factors of production Smith implicitly introduced — labor, land, and capital — no longer operate in their pure form: the digital and even non-digital sectors generate no profit without data, technology, and infrastructure, which calls into question the universality of the original system. This gap cannot be ignored: an economy in which the digital sector, comprising three times fewer industries than the physical sector, contributes twice as much labor input to output growth and in some cases leads in productivity growth rates is structured differently from the one Smith described. According to BEA–BLS data for 63 U.S. industries over 1997—2023, this gap is not cyclical: it registers a structural change in the very system of factors of production. The central question of this paper is not historical reconstruction but method: what would Smith have said had he seen digital platforms, and how would he have rethought his own categories? The same question is posed regarding the economists Smith directly influenced and who developed the theory of market organization, the firm, and consumer surplus — Marx, Marshall, Pigou, Schumpeter, Coase. The paper proposes a procedure of returning to the theoretical core that makes it possible to trace the evolution of factors of production from the original triad to a system that includes data, technology, and infrastructure, and to test this evolution through the sequence of these theorists. The analysis shows that the propositions formulated by the classics retain their validity on the whole yet begin to operate under conditions where market boundaries become, in essence, infinite. A transparent market open to small players has emerged, and economic science must reckon with this, and market power is shifting to the ownership of the digital assets that underpin it. Economic analysis of the information society, digital platforms, and artificial intelligence requires rebuilding the “hard core” of science, abandoning its textbook-based learning. Ultimately, such analysis is essential for informed decision-making, but economists must overcome the disease of neglecting the fundamental foundations of their theory.

Summary

Main Finding

The economy has undergone a structural shift in its factors of production: data, technology, and infrastructure have become de facto production factors alongside (and often dominating) Smith’s original triad of labor, land, and capital. Using BEA–BLS industry data for 1997–2023, the paper shows this is not a cyclical phenomenon but a persistent structural change — a digital sector that has fewer industries but contributes disproportionately to labor’s role in output growth and often leads in productivity growth. Classical economic propositions remain broadly useful but must be reinterpreted under conditions where market boundaries and the role of digital assets are fundamentally different.

Key Points

The canonical three factors (labor, land, capital) no longer capture the dominant drivers of output in modern economies; data, technology, and infrastructure function as essential, distinct factors.
The digital sector (≈ one-third the number of industries of the physical sector) contributes roughly twice as much labor input to output growth and frequently leads in measured productivity growth rates.
The observed differences are persistent across 1997–2023 and reflect a structural transformation rather than short-run cycles.
The paper advances a methodological "return to the theoretical core": it re-examines Smith’s categories and traces their evolution through the doctrines of Marx, Marshall, Pigou, Schumpeter, and Coase to test how classical insights map onto the information/digital economy.
Classical theories still provide useful propositions, but their operating conditions have changed — market boundaries are effectively expanded (platform-enabled, cross-market externalities), and ownership of digital assets becomes the main locus of market power.
The rise of transparent, platform-mediated markets alters firm boundaries, competition dynamics, and the allocation of surplus; policy and measurement frameworks based on older factor taxonomies are inadequate.

Data & Methods

Empirical base: BEA–BLS combined data covering 63 U.S. industries over 1997–2023.
Empirical focus: comparative analysis of digital versus physical sectors on:
- the contribution of labor input to output growth, and
- productivity growth rates.
Interpretation: the paper treats the empirical regularities as evidence of a structural change in factors of production rather than a cyclical fluctuation.
Theoretical method: a systematic, historically informed procedure that maps the original factor-concept (Smith) through subsequent theorists (Marx, Marshall, Pigou, Schumpeter, Coase) to identify which core propositions survive, which require reinterpretation, and how to operationalize new factors (data, technology, infrastructure).
Note: the paper emphasizes conceptual and methodological reconstruction rather than only economic historiography — the aim is to provide a core theoretical framework suitable for empirical testing and policy analysis in the digital age.

Implications for AI Economics

Reclassify production inputs: AI, the datasets that train AI, compute platforms, and network/cloud infrastructure should be modeled as distinct production factors (akin to capital) with their own accumulation, depreciation, and ownership dynamics.
Measurement & national accounts: GDP, productivity, and capital stock accounting must adjust to capture intangible digital assets (data inventories, trained models, platform codebases, cloud capacity) to avoid mis-measuring growth and distribution.
Market power & competition policy: ownership and control of data, models, and platform infrastructure are central sources of market power; antitrust frameworks and remedies must incorporate asset-based and access-based approaches (data portability, interoperability, access obligations).
Firm boundaries & organization: platforms and AI services change transaction costs and make traditional make-or-buy decisions different; Coasean firm theory and transaction-cost models need extensions to cover digital asset governance and network externalities.
Distributional effects: returns to owners of digital assets (platforms, model proprietors) may increase relative to labor and traditional capital, requiring rethought tax, social insurance, and labor-market policies.
Innovation dynamics: AI as a general-purpose technology interacts with Schumpeterian creative destruction in novel ways — diffusion depends on data access, infrastructural scale, and modularity of models; policy should consider public goods aspects of foundational models and shared infrastructure.
Research agenda: develop models that treat data/AI/infrastructure as capital with accumulation dynamics; create measurement methodologies for digital asset stocks and services; empirically estimate the contribution of AI and data to productivity and distribution; design policy experiments for data governance, taxation, and competition remedies.
Pedagogy & method: economists should rebuild core theoretical tools (not just deploy textbook variants) to reason about markets with near-global boundaries, pervasive network effects, and asset-based market power; interdisciplinary work (computer science, law, public policy) is essential.

If you want, I can: - Extract specific policy recommendations implied by the paper (antitrust, taxation, national accounts changes). - Draft a short methodological checklist for empirically measuring data/AI/infrastructure as factors of production.

Assessment

Paper Typetheoretical Evidence Strengthlow — The paper relies primarily on conceptual and historical argumentation supported by descriptive industry-level statistics (BEA–BLS for 63 U.S. industries, 1997–2023) but offers no causal identification, quasi-experimental variation, or robustness checks to isolate mechanisms; empirical claims are therefore suggestive rather than robust. Methods Rigormedium — Theoretical reconstruction and literature tracing appear carefully argued, and the use of long-run BEA–BLS industry data adds empirical grounding, but the empirical work is descriptive (aggregation choices, classification of 'digital' vs 'physical' sectors are not fully specified), lacks causal identification and sensitivity analyses, and may omit firm-level heterogeneity and confounders. SampleIndustry-level BEA–BLS data covering 63 U.S. industries for 1997–2023, used to compare contributions to output growth, labour input shares, and productivity growth across a labeled 'digital' vs 'physical' sector; details on variable construction and industry classification for the digital sector are not fully specified in the summary. Themesproductivity innovation governance GeneralizabilityUS-only industry data; findings may not hold in other countries with different digital adoption patterns or market structures, Dependent on the paper's industry classification (digital vs physical), which may be subjective and sensitive to reclassification, Aggregate industry-level analysis may mask firm-level heterogeneity and platform-specific dynamics, 1997–2023 period captures a particular technological epoch; rapid future changes in AI/platforms could alter conclusions, Descriptive evidence cannot establish causal mechanisms linking digital assets to measured productivity gains

Claims (9)

Claim	Direction	Outcome	Confidence & Evidence	Details
The digital sector comprises three times fewer industries than the physical sector. Market Structure	null_result	count of industries in digital vs physical sector	Reading fidelity high Study strength medium	n=63 three times fewer industries 0.12
The digital sector contributes twice as much labor input to output growth (relative to the physical sector). Labor Share	positive	labor input contribution to output growth	Reading fidelity high Study strength medium	n=63 twice as much labor input to output growth 0.12
In some cases the digital sector leads in productivity growth rates. Firm Productivity	positive	productivity growth rate	Reading fidelity high Study strength medium	n=63 0.12
The observed patterns in BEA–BLS data for 63 U.S. industries over 1997–2023 do not reflect cyclical variation but register a structural change in the system of factors of production. Market Structure	null_result	structural change in factors of production	Reading fidelity high Study strength medium	n=63 0.12
Digital and even non-digital sectors generate no profit without data, technology, and infrastructure. Firm Revenue	negative	profit generation dependence on data/technology/infrastructure	Reading fidelity high Study strength low	not reported 0.06
Market power is shifting to the ownership of the digital assets that underpin markets. Market Structure	negative	distribution of market power (ownership of digital assets)	Reading fidelity high Study strength medium	not reported 0.12
A transparent market open to small players has emerged. Market Structure	positive	market openness / ability of small players to participate	Reading fidelity medium Study strength low	not reported 0.04
The propositions formulated by classical economists (Smith, Marx, Marshall, Pigou, Schumpeter, Coase) retain their validity on the whole, but they begin to operate under conditions where market boundaries become, in essence, infinite. Market Structure	mixed	validity and operational conditions of classical economic propositions	Reading fidelity medium Study strength speculative	not reported 0.01
Economic analysis of the information society, digital platforms, and artificial intelligence requires rebuilding the 'hard core' of economic science and abandoning textbook-based learning. Governance And Regulation	negative	adequacy of existing economic methodology for analyzing digital/AI economy	Reading fidelity high Study strength speculative	not reported 0.02