AI innovation is propelling China and India ahead while Brazil, Russia and South Africa fall behind; patents and R&D move productivity frontiers, robots raise efficiency but deepen a two-tier BRICS productivity hierarchy.

AI-driven productivity dynamics in BRICS economies: Evidence from a Malmquist Total Factor Productivity Analysis

Adil Azizoğlu · April 16, 2026 · International Journal of Business and Management (IJBM)

openalex correlational medium evidence 7/10 relevance DOI Source PDF

Across BRICS (2005–2023), AI-related innovation (patents, R&D) is strongly associated with frontier-shifting technological change in China and India, while adoption-oriented measures (robot density) correlate more with efficiency gains, leaving Brazil, Russia and South Africa lagging and reinforcing a two-tier productivity split.

This study examines the impact of artificial intelligence (AI) on productivity dynamics in BRICS economies (Brazil, Russia, India, China, and South Africa) over the period 2005–2023. Using a two-stage empirical approach, productivity growth is first measured through the Malmquist Total Factor Productivity (TFP) index, decomposing changes into efficiency change (EC) and technological change (TC). In the second stage, panel regression analysis evaluates the relationship between these components and key AI penetration indicators, including AI patents, investment, robot density, and digital infrastructure. The results reveal significant divergence across BRICS economies. China and India exhibit sustained productivity growth driven primarily by technological progress, whereas Brazil, Russia, and South Africa experience stagnation or decline in both efficiency and technological advancement. The decomposition analysis shows that innovation-oriented AI activities, such as patents and research investment, are strongly associated with frontier-shifting technological change, while adoption-oriented indicators, including robot density, contribute to efficiency improvements. Digital infrastructure emerges as a critical complementary factor influencing both channels of productivity growth. Overall, the findings indicate that AI adoption is reinforcing existing structural disparities within the BRICS bloc, creating a two-tier productivity hierarchy. The study contributes to the literature by providing a comparative, frontier-based assessment of AI-driven productivity in emerging economies and by distinguishing between innovation and diffusion effects of AI. Policy implications highlight the importance of strengthening digital infrastructure, human capital, and innovation capacity to ensure inclusive productivity gains from the AI revolution.

Summary

Main Finding

Using a Malmquist TFP decomposition (2005–2023) and a second-stage panel regression on AI indicators, the paper finds a clear divergence within BRICS: China and India show sustained productivity growth driven mainly by technological change (frontier shifts), while Brazil, Russia, and South Africa exhibit stagnation or decline in both efficiency and frontier advancement. Innovation-oriented AI activity (patents, R&D/investment) is strongly associated with frontier-moving technological change; adoption-oriented measures (robot density, diffusion) are linked to catch-up (efficiency) gains. Digital infrastructure is an important complement that strengthens both channels. Overall, AI adoption appears to be reinforcing pre-existing structural disparities, producing a two-tier productivity hierarchy across BRICS.

Key Points

Scope and period: BRICS (Brazil, Russia, India, China, South Africa), 2005–2023.
Two-stage empirical strategy:
- Stage 1: Malmquist Total Factor Productivity (TFP) index via output-oriented DEA (VRS) to decompose TFP growth into Efficiency Change (EC) and Technological Change (TC).
- Stage 2: Panel regression linking EC and TC to AI penetration proxies.
AI proxies used: AI patents, AI-related investment, robot density (diffusion/adoption), and digital infrastructure indicators.
Core empirical findings:
- China and India: sustained TFP growth driven primarily by TC — evidence of frontier-shifting innovation related to AI.
- Brazil, Russia, South Africa: weak/negative EC and TC — limited catching-up and limited frontier movement.
- Innovation-oriented indicators (patents, R&D/investment) predict TC; adoption-oriented indicators (robots) predict EC.
- Digital infrastructure amplifies both TC and EC effects.
Contribution: provides a frontier-based comparative assessment that distinguishes innovation (frontier shifts) vs diffusion (catch-up) effects of AI in emerging economies.
Methodological caveats noted: DEA/Malmquist is non-parametric and deterministic (sensitive to measurement error and outliers); cross-country comparability and potential endogeneity concerns around AI measures and productivity.

Data & Methods

Sample: Five BRICS countries across 2005–2023.
Productivity measurement:
- Malmquist TFP index (output-oriented) computed using Data Envelopment Analysis under Variable Returns to Scale.
- Decomposition into Efficiency Change (EC = movement relative to the frontier) and Technological Change (TC = outward shift of the frontier).
Second-stage analysis:
- Panel regressions where EC and TC (dependent variables) are regressed on AI penetration indicators (AI patents, AI investment, robot density, digital infrastructure).
- Interpretation: AI patents and research investment → TC (innovation/frontier shifts); robot density and diffusion metrics → EC (catch-up).
Robustness and limitations:
- MPI/DEA attributes all deviations from frontier to inefficiency (no stochastic error term).
- Sensitive to variable selection and outliers; measurement and comparability of AI proxies across countries remain challenging.
- Potential two-way causality (higher TFP may also attract AI investment); causality is suggestive rather than definitively established.

Implications for AI Economics

Theoretical implications:
- Empirical support for endogenous-growth style mechanisms: AI as a GPT can both shift production frontiers (via innovation) and enable catch-up (via diffusion), but effects depend on absorptive capacity and complementary assets.
- Demonstrates how AI’s dual role (innovation vs diffusion) yields heterogeneous macro outcomes across countries with different institutional and infrastructural endowments.
Policy implications (for emerging economies and multilateral policy makers):
- Strengthen digital infrastructure (broadband, cloud, data ecosystems) to amplify both frontier and catch-up effects.
- Invest in innovation capacity: R&D funding, AI research centers, IP protection, and tech transfer to promote frontier-moving TC.
- Build absorptive capacity: education/skill development (STEM and AI-related skills), managerial capabilities, and incentives for organizational co‑invention.
- Promote diffusion policies: support SMEs in adopting off-the-shelf AI/robotics, subsidies or financing for adoption, and standards/interoperability to reduce barriers.
- Address distributional/structural risks: targeted policies to avoid reinforcing a two-tier hierarchy—e.g., regionally balanced investments, support for inclusion of smaller firms.
Directions for future research:
- Causal identification using quasi-experimental or instrumental-variable approaches to disentangle endogeneity between AI and productivity.
- Micro-level (firm/establishment) and sectoral studies to trace channels from AI investment to firm performance and reallocation effects.
- Improved measurement of AI penetration (distinguishing narrow AI applications, foundation models, datasets, and human capital).
- Examination of distributional impacts (labor market, regional inequality) and trade/firm-entry responses associated with AI-driven frontier shifts.

Summary judgement: The paper delivers a useful frontier-based decomposition showing that AI’s macro productivity impacts in emerging economies are heterogeneous and contingent on whether countries generate frontier innovations or mainly adopt existing technologies. The policy message is that without investments in innovation capacity and complementary assets, AI risks amplifying existing global and within‑bloc divergences.

Assessment

Paper Typecorrelational Evidence Strengthmedium — The study uses a clear frontier-based productivity metric and links it to multiple AI indicators across a long panel, which provides informative patterns; however, the analysis is observational with likely endogeneity (reverse causality, omitted variables), measurement challenges in AI proxies, and a small set of countries (BRICS), limiting confidence in causal claims. Methods Rigormedium — Methodologically sound elements (Malmquist TFP decomposition, panel regressions, multiple AI measures and decomposition into EC/TC) indicate careful empirical work, but rigor is constrained by potential identification threats (no instrumental/quasi-experimental strategy described), limited cross-sectional variation (five countries), and possible sensitivity to specification and measurement choices. SampleCountry-level panel of the five BRICS economies (Brazil, Russia, India, China, South Africa) for 2005–2023; Malmquist TFP indices constructed from country input–output data (labor, capital, output) and decomposed into efficiency and technological change; AI-related variables include counts of AI patents, measures of AI/R&D investment, industrial robot density, and digital infrastructure indicators (e.g., broadband, mobile penetration); sample size roughly equals the number of country-year observations across 2005–2023, subject to missing data. Themesproductivity innovation adoption inequality IdentificationTwo-stage empirical approach: (1) compute Malmquist Total Factor Productivity (TFP) indices and decompose changes into efficiency change (EC) and technological change (TC); (2) estimate panel regressions of EC and TC on AI penetration indicators (AI patents, AI investment, robot density, digital infrastructure) using country-year panel data and standard controls (time and/or country effects and other covariates as reported). Causal identification is associative rather than based on exogenous variation or quasi-experimental variation. GeneralizabilityLimited to BRICS countries — results may not apply to advanced economies or smaller/emerging nations, Small cross-sectional sample (only five countries) limits heterogeneity and external validity, Aggregate country-level analysis masks important within-country and sectoral heterogeneity (services vs manufacturing), AI proxies (patents, robot density, investment) are imperfect and may not capture the full scope of AI diffusion/use, Observational design limits causal generalization beyond associations — findings may reflect reverse causality or omitted confounders, Results may be sensitive to the Malmquist decomposition inputs, measurement choices, and major shocks in the period (e.g., 2008 crisis, COVID-19)

Claims (8)

Claim	Direction	Confidence	Outcome	Details
China and India exhibit sustained productivity growth over 2005–2023 driven primarily by technological progress. Firm Productivity	positive	high	Total Factor Productivity (Malmquist TFP index) and its Technological Change (TC) component	n=5 0.3
Brazil, Russia, and South Africa experience stagnation or decline in both efficiency and technological advancement over 2005–2023. Firm Productivity	negative	high	Efficiency Change (EC) and Technological Change (TC) components of the Malmquist TFP index	n=3 0.3
Innovation-oriented AI activities (AI patents and research investment) are strongly associated with frontier‑shifting technological change (TC). Firm Productivity	positive	high	Technological Change (TC) component of the Malmquist TFP index	n=95 0.3
Adoption-oriented AI indicators, including robot density, contribute to efficiency improvements (EC). Firm Productivity	positive	high	Efficiency Change (EC) component of the Malmquist TFP index	n=95 0.3
Digital infrastructure is a critical complementary factor influencing both efficiency improvements and frontier‑shifting technological change. Firm Productivity	positive	high	Efficiency Change (EC) and Technological Change (TC) components of the Malmquist TFP index	n=95 0.3
AI adoption is reinforcing existing structural disparities within the BRICS bloc, creating a two‑tier productivity hierarchy (China & India vs. Brazil, Russia & South Africa). Inequality	negative	high	Cross-country divergence in Total Factor Productivity (TFP) growth and its components	n=5 0.3
The study contributes methodologically by providing a comparative, frontier‑based assessment of AI-driven productivity in emerging economies and by distinguishing innovation (frontier-shifting) and diffusion (efficiency) effects of AI. Research Productivity	positive	high	Research/methodological contribution (comparative frontier-based assessment and decomposition of AI effects)	0.5
Policy implications: strengthening digital infrastructure, human capital, and innovation capacity is important to ensure inclusive productivity gains from the AI revolution in BRICS economies. Governance And Regulation	positive	high	Policy levers for inclusive productivity gains (digital infrastructure, human capital, innovation capacity)	0.05