AI Economic Impact in Emerging Markets: BIS Analysis of the Growing Digital Divide

How AI Is Reshaping Economic Growth in Emerging Market Economies

Artificial intelligence stands poised to become the defining economic force of the coming decade, but its impact will not be distributed equally across the global economy. The Bank for International Settlements (BIS), in Bulletin No. 121, presents a comprehensive analysis of how AI’s economic impact will differ dramatically between advanced economies and emerging market economies (EMEs) — with potentially far-reaching consequences for global income convergence, financial stability, and development trajectories.

The central finding is both striking and sobering: while AI promises substantial productivity gains at the task and firm level, the aggregate economic benefits for emerging markets will be significantly smaller than for advanced economies. A standardized increase in AI preparedness raises real value-added growth by 0.6 percentage points in advanced economies versus only 0.45 percentage points in EMEs. This gap, if sustained over a decade, could reverse decades of gradual income convergence between developed and developing nations.

Understanding why this divergence occurs — and what policymakers in emerging markets can do about it — is essential for anyone involved in global finance, development policy, or technology strategy. The BIS analysis draws on extensive micro-level evidence, cross-country modeling across 70 economies, and the IMF’s AI Preparedness Index covering 174 countries to build a nuanced picture of AI’s uneven economic impact.

AI Productivity Gains: What Evidence Tells Us About Emerging Markets

The micro-level evidence on AI productivity gains is impressive but must be interpreted carefully when projecting aggregate economic impact for emerging markets. Task-level studies across multiple domains demonstrate that generative AI can boost individual worker productivity by 10 to 65 percent depending on the task, skill level, and implementation context.

In software development, junior developers using AI coding assistants see productivity improvements of 21 to 67 percent, while senior developers gain a more modest 7 to 26 percent. In consulting, management professionals using AI tools complete analysis tasks 20 to 40 percent faster with measurable quality improvements. In writing and content creation, AI assistance reduces production time by 30 to 50 percent while maintaining or improving output quality.

Crucially, the BIS highlights an equalizing effect within tasks: lower-skilled workers consistently capture larger productivity gains from AI than their more experienced counterparts. This suggests that AI could narrow performance gaps within occupations, potentially reducing within-firm inequality. However, this positive effect is counterbalanced by a less optimistic dynamic: the workers who benefit most from AI productivity gains are also the most vulnerable to outright task automation, as AI systems become capable enough to replace rather than augment their work.

Translating these task-level gains into aggregate economic impact reveals a wide range of estimates. Acemoglu’s conservative projection suggests AI will add only 0.07 percent to annual total factor productivity (TFP), while more optimistic models from Aghion, Bunel, and Bergeaud project gains of 0.3 to 0.9 percent annually. The gap reflects fundamental disagreements about how rapidly AI capabilities diffuse across sectors, how effectively firms reorganize around AI, and how labor markets adapt to displacement.

Why Sectoral Composition Limits AI Economic Impact in Emerging Markets

Perhaps the most illuminating finding in the BIS analysis concerns the role of sectoral composition in determining a country’s capacity to benefit from AI. Using the Felten et al. AI Exposure Measure — which quantifies how extensively generative AI can augment workplace abilities across occupations — the BIS maps AI exposure by economic sector and then weights these by each country’s production structure.

The results reveal a stark structural disadvantage for many emerging markets. Finance, education, and information technology — the sectors with the highest AI exposure and therefore the greatest potential for AI-driven productivity gains — account for a much larger share of GDP in advanced economies. Conversely, agriculture, construction, and transport — sectors with the lowest AI exposure — remain dominant in many emerging market economies.

This means that even if an emerging market achieves identical AI adoption rates to an advanced economy at the firm level, the aggregate economic impact will be smaller because a larger proportion of its economy operates in sectors where AI delivers limited productivity benefits. A country where agriculture represents 25 percent of GDP will capture far less aggregate AI benefit than one where finance and professional services represent the same share, regardless of how aggressively either country pursues AI adoption.

The BIS invokes the Baumol cost disease framework to explain why aggregate productivity gains may be further dampened: as AI boosts productivity in exposed sectors, labor and resources may shift toward less productive, non-exposed sectors, diluting the economy-wide impact. For emerging markets with large agricultural and informal sectors, this structural drag could be particularly severe.

AI Preparedness Index: How Ready Are Emerging Economies?

The IMF’s AI Preparedness Index (AIPI) provides the most comprehensive cross-country measure of structural readiness for AI adoption, covering 174 countries across four equally weighted pillars: digital infrastructure, human capital and labor market policies, innovation and economic integration, and regulation and ethics.

The AIPI data reveals systematic gaps between advanced economies and emerging markets across all four dimensions. In digital infrastructure, advanced economies benefit from universal broadband connectivity, abundant data center capacity, reliable electricity, and mature cloud computing ecosystems. Many emerging markets face persistent deficits in each area — limited broadband penetration, insufficient data center investment, unreliable power grids, and high costs for cloud services relative to income levels.

In human capital, the gap encompasses both the supply of AI-specialized talent (data scientists, ML engineers, AI researchers) and the broader workforce’s digital literacy. Advanced economies produce significantly more AI-related graduates per capita and invest more in continuous professional development. Emerging markets often face a brain drain dynamic where their most talented AI professionals emigrate to better-compensated positions in advanced economies.

The innovation and economic integration pillar captures research output, venture capital availability, patent activity, and integration into global technology supply chains. Here, the variance within the EME group is substantial: economies like South Korea, Singapore, and Israel score comparably to many advanced economies, while countries in Sub-Saharan Africa and parts of South Asia score at the bottom of the global distribution.

The BIS emphasizes that AIPI scores are not destiny — they are a policy variable. Countries that invest strategically in closing preparedness gaps can capture significantly larger shares of AI-driven growth. The modeling shows that the economic returns to preparedness improvement are substantial: each standardized increase in AIPI translates directly into higher growth trajectories.

The AI Divide: Growth Effects in Advanced vs Emerging Economies

The BIS quantifies the growth differential between advanced and emerging economies using a multi-country general equilibrium model calibrated to 70 countries (23 advanced economies and 47 emerging market and developing economies). The model simulates a perfectly anticipated positive TFP supply shock from generative AI, benchmarked at 0.5 percent annual TFP growth for the United States over a decade.

The results confirm what sectoral analysis and preparedness indicators suggest: advanced economies capture disproportionately larger growth benefits from AI. A standardized increase in AI preparedness translates into 0.6 percentage points of additional real value-added growth in advanced economies versus 0.45 percentage points in emerging markets — a 33 percent differential that compounds significantly over time.

The drivers of this gap are multifaceted. Beyond sectoral composition and preparedness scores, advanced economies benefit from deeper capital markets that fund AI investment, more flexible labor markets that facilitate workforce reallocation, stronger intellectual property protections that incentivize AI innovation, and more effective regulatory frameworks that balance innovation with responsible AI governance.

Importantly, the BIS model accounts for global trade and capital flow spillovers. Emerging markets integrated into AI supply chains — semiconductor exporters like Taiwan and Vietnam, for example — may capture near-term benefits from surging demand for AI hardware. However, these gains are concentrated in specific EMEs and do not offset the broader structural disadvantage facing the group as a whole.

Long-Term GDP Impact: Will Emerging Markets Fall Further Behind?

The most consequential finding in the BIS analysis concerns the long-term GDP divergence scenarios. Under a no-convergence scenario — where current preparedness gaps persist unchanged over a decade — a sustained 0.5 percent annual TFP shock from AI would lift advanced economy GDP by more than 2 percentage points above emerging markets. This represents a significant reversal of the gradual income convergence that characterized the 2000–2020 period, when emerging markets were closing the GDP-per-capita gap with advanced economies.

Under a partial convergence scenario — where emerging markets close half their preparedness gap with the United States over the same decade — the GDP divergence shrinks to less than 1 percentage point. This represents a dramatically different trajectory, one where AI narrows rather than widens global inequality, and where the benefits of technological progress are more equitably shared across the global economy.

The gap between these two scenarios is entirely driven by policy choices. The BIS emphasizes that the divergence is not inevitable — it is a function of investment decisions in digital infrastructure, education, regulatory capacity, and international cooperation that governments in emerging markets can influence directly. The economic returns to closing the preparedness gap are high enough to justify substantial public investment, particularly when considering the compounding nature of productivity gains over decades.

For the global development community, these findings argue for prioritizing AI preparedness alongside traditional development priorities. Infrastructure investment, education reform, and regulatory capacity building are not new agendas — but the AI dimension adds urgency and specificity to each, with clear evidence that underinvestment will carry compounding costs over time.

Labor Market Risks of AI in Emerging Markets

The labor market implications of AI in emerging markets are particularly concerning because of a structural vulnerability that advanced economies largely do not share: the prevalence of informal employment. In many EMEs, 40 to 70 percent of the workforce operates in the informal sector — without employment contracts, social insurance, or access to employer-provided training and reskilling programs.

As AI automates routine cognitive tasks — data entry, basic analysis, customer service, document processing — the workers displaced in emerging markets face steeper barriers to re-employment than their counterparts in advanced economies. Formal sector workers in developed countries typically have access to unemployment insurance, government-funded retraining programs, and institutional support for career transitions. Informal workers in emerging markets have none of these safety nets.

The BIS estimates that up to 60 percent of occupations could face substantial task reallocation by 2030, with 25 to 50 percent of workloads in AI-exposed jobs potentially replaceable. While not all task reallocation leads to job displacement — many roles will be restructured rather than eliminated — the transition creates significant adjustment costs. In emerging markets where social protection systems are weak, these costs fall disproportionately on the most vulnerable workers.

The youth demographic adds another dimension. Many emerging markets have large, young populations entering the labor force at a time when the entry-level jobs they would traditionally fill — clerical work, basic financial services, customer support — are being automated. Without proactive investment in AI-relevant skills and digital literacy, these economies face a potential mismatch between labor supply and demand that could fuel social unrest and political instability.

Closing the AI Preparedness Gap: Policy Priorities

The BIS analysis identifies five policy priority areas that emerging market governments should pursue to close the AI preparedness gap and capture a larger share of AI-driven economic growth. These are not theoretical recommendations — each is grounded in empirical evidence about which factors most strongly predict cross-country variation in AI economic impact.

First, invest massively in digital infrastructure. Broadband connectivity, data center capacity, reliable electricity supply, and affordable cloud computing access are the foundational enablers of AI adoption. Without these basics, no amount of AI strategy or innovation policy will deliver economic results. The BIS data shows that digital infrastructure gaps account for the single largest component of the preparedness differential between advanced and emerging economies.

Second, expand AI-related education at every level. This means integrating computational thinking into primary and secondary curricula, scaling university programs in data science and machine learning, and investing in continuous professional development for the existing workforce. Critically, education investments must be complemented by labor market policies that retain AI talent domestically rather than losing it to brain drain.

Third, strengthen social protection systems to support workers through the AI-driven economic transition. This includes unemployment insurance expansion, portable benefits for informal workers, targeted reskilling programs for displaced populations, and transitional income support. The absence of these safety nets amplifies the social costs of AI adoption and can generate political backlash that slows beneficial technology deployment.

Fourth, develop sound AI regulatory and ethical frameworks that build trust and enable responsible deployment. Emerging markets that establish clear, predictable regulatory environments for AI will attract more foreign investment, encourage domestic innovation, and build consumer confidence in AI-powered services. The BIS emphasizes that regulation should be proportionate and risk-based, avoiding both the extremes of regulatory absence and over-prescriptive rules that stifle innovation.

Fifth, leverage trade networks and global value chain positions to capture near-term economic benefits from the AI boom. Countries positioned in the AI hardware supply chain (semiconductor manufacturing, rare earth minerals), AI services (software development, data labeling), or AI-intensive industries (financial services, healthcare IT) can accelerate economic benefits through strategic trade and industrial policies.

Trade and Global Value Chains Shape AI Economic Impact

The BIS analysis recognizes that AI’s economic impact does not occur in national isolation — it flows through trade channels, global value chains, and international technology transfer mechanisms that create both opportunities and vulnerabilities for emerging markets.

Countries positioned as AI hardware suppliers — particularly those involved in semiconductor manufacturing (Taiwan, South Korea, Vietnam), rare earth mineral extraction (Democratic Republic of Congo, Chile, Indonesia), and data center construction — stand to benefit from surging capital expenditure by major technology companies. Global AI-related investment has grown exponentially, and emerging markets that participate in this supply chain can capture significant economic rents in the near term.

However, this benefit is highly concentrated. The vast majority of emerging markets are AI technology consumers rather than producers, importing AI services and platforms from a handful of advanced-economy providers. This creates dependency dynamics where the technology rents flow outward, and the balance of AI-related trade consistently favors advanced economies. The BIS warns that without deliberate policies to develop domestic AI capabilities, this trade pattern will reinforce rather than reduce the economic divergence.

Technology transfer mechanisms — foreign direct investment, international research collaborations, open-source AI ecosystems, and development assistance — can help bridge the gap. The BIS highlights the potential of open-source AI models and platforms, which lower the barriers to AI adoption by reducing licensing costs and enabling local customization. Emerging markets that invest in the technical capacity to leverage open-source AI may achieve adoption levels comparable to advanced economies at a fraction of the proprietary technology cost.

Digital Infrastructure: The Key to AI-Driven Growth

The BIS analysis returns repeatedly to a central conclusion: digital infrastructure is the single most important determinant of whether an emerging market captures meaningful economic benefit from AI. Without reliable broadband, adequate computing capacity, and affordable data services, AI remains an academic abstraction rather than a practical economic tool.

The infrastructure gap is quantifiable and consequential. Advanced economies offer near-universal fixed broadband coverage with average speeds exceeding 100 Mbps. Many emerging markets have fixed broadband penetration rates below 30 percent, with average speeds of 10–30 Mbps — insufficient for real-time AI applications, cloud-based model inference, or the data transfer requirements of modern machine learning workflows.

Data center capacity presents an equally stark disparity. The United States alone hosts more data center capacity than all emerging markets combined. As AI workloads increasingly require low-latency access to computing resources, the physical proximity of data centers to users and businesses becomes a competitive factor. Emerging markets without domestic data center infrastructure face latency penalties that degrade AI application performance and increase costs.

Power infrastructure is a frequently overlooked constraint. AI model training and inference are energy-intensive operations. Data centers require reliable, continuous electricity supply — something that many emerging markets cannot guarantee. Countries with frequent power outages, unreliable grid infrastructure, or high electricity costs face a fundamental barrier to AI deployment that no software innovation can overcome.

The BIS recommends that emerging market governments treat digital infrastructure investment with the same strategic urgency they applied to physical infrastructure (roads, ports, railways) in previous development eras. The returns to digital infrastructure investment in the AI era are potentially even higher than historical returns to physical infrastructure, given the multiplier effects of AI-driven productivity across the entire economy.