AI Adoption and Labour Displacement in Developing Countries

Background to the Study

The economic history of developing countries has generally followed a structural transformation pathway in which labour shifts from low-productivity agriculture into labour-intensive manufacturing and services. This process historically depended on comparative advantage that abundant low-cost labour affords developing economies relative to capital-abundant advanced economies.

The accelerating diffusion of AI technologies, encompassing robotic process automation, AI-assisted customer service and back-office functions, predictive analytics, and generative AI capable of performing tasks requiring human judgement, has raised concern that this technology wave may erode labour-cost advantage. Nigerian manufacturing and service firms have, over past several years, begun adopting AI-enabled tools at modest but accelerating pace since 2023 onward with accessible generative AI platforms.

A further dimension concerns firm-size heterogeneity. Larger firms, possessing diversified internal task structures, dedicated human resources functions, and greater financial capacity for retraining, may be better positioned to redeploy workers whose tasks are automated into other roles within same firm, a form of internal reinstatement effect theorized by Acemoglu and Restrepo. Smaller firms, with narrower task structures and limited retraining budgets, may be less able to replicate this. Anecdotal reports from Lagos-based firms in customer service, logistics, financial services, and light manufacturing point to reduced hiring attributed partly to AI, though rigorous firm-level empirical evidence quantifying this relationship and its firm-size heterogeneity in Nigerian context has been limited, as highlighted in World Bank and ILO reports on AI and future of work.

Statement of the Problem

Nigerian labour market policy discussions around AI and automation have drawn substantially on international, predominantly advanced-economy evidence and projections, which may not accurately characterise pattern, magnitude, or firm-size distribution of AI-related labour displacement in a developing-country context dominated by SMEs with limited internal redeployment capacity.

Without firm-level Nigerian evidence disaggregated by firm size and routine-task composition, policymakers lack clear empirical basis for calibrating skills reorientation and social protection responses to specific pattern of AI adoption actually occurring among Nigerian firms of differing scale. This study addresses this gap by providing Lagos-based evidence with explicit interaction and stratified tests for firm-size-dependent displacement.

Aim and Objectives of the Study

The aim of this study is to examine the effect of artificial intelligence adoption on labour market displacement among manufacturing and service firms in Lagos, Nigeria.

·         Examine the level and pattern of AI adoption among sampled manufacturing and service firms

·         Assess the trend in net employment change among sampled firms over the two years preceding the survey

·         Determine the effect of AI adoption intensity on net employment change, controlling for other firm characteristics

·         Examine whether the employment effect of AI adoption varies by the routine/non-routine task composition of firm employment

·         Test whether the AI-employment relationship differs significantly between large and small firms

·         Draw policy implications for skills reorientation and social protection in response to AI-related labour displacement

Research Questions

·         What is the level and pattern of AI adoption among manufacturing and service firms in the study area?

·         What has been the trend in net employment change among sampled firms over the two years preceding the survey?

·         What effect does AI adoption intensity have on net employment change among sampled firms?

·         Does the employment effect of AI adoption vary by the routine/non-routine task composition of firm employment?

·         Does the AI-employment relationship differ significantly between large and small firms?

Research Hypotheses

·         H₀1: AI adoption intensity has no significant effect on net employment change among sampled firms.

·         H₀2: The employment effect of AI adoption does not significantly differ between firms with a high versus low share of routine-task employment.

·         H₀3: Firm size has no significant effect on net employment change among sampled firms.

·         H₀4: The effect of AI adoption intensity on net employment change does not differ significantly between large and small firms.

Significance of the Study

This study is of value to Federal Ministry of Labour and Employment, National Bureau of Statistics, and National Universities Commission in designing skills reorientation curricula and social protection responses calibrated to specific pattern of AI-related displacement observed among Nigerian firms. Firm-size heterogeneity component is of particular practical value, since it speaks directly to whether support should be uniformly available or prioritised toward workers in smaller firms, which face most acute displacement risk given limited internal redeployment capacity. It also contributes to developing-country AI and labour literature by extending task-based automation theory to specific question of firm-size-dependent internal reinstatement capacity, building on work by Acemoglu & Autor and recent World Bank analysis of automation in Sub-Saharan Africa.

Scope of the Study

The study is limited to registered manufacturing and service firms operating within Lagos metropolis, drawing on cross-sectional primary data collected Q1 2026, covering reported employment change over two years preceding survey (2024-2026). Firm-size heterogeneity analysis compares firms above and below 50-employee threshold, aligning with National Policy on Micro, Small and Medium Enterprises classification boundaries.

Operational Definition of Terms

Artificial Intelligence Adoption: Use by firm of software or systems performing tasks conventionally requiring human cognitive input, including machine learning analytics, RPA, AI-assisted customer service, generative AI.

AI Adoption Intensity Index: Composite score capturing breadth and depth of AI tool adoption across firm's functional areas, 0 (no adoption) to 10 (extensive multi-functional).

Labour Market Displacement: Reduction in firm employment or growth relative to output growth, attributable at least partly to substitution of AI systems for tasks previously performed by humans.

Routine-Task Employment: Employment involving well-defined repetitive procedures, conventionally identified as most susceptible to automation.

Internal Reinstatement Capacity: Firm's ability to redeploy workers whose specific tasks have been automated into other roles within same firm, theorised to vary with firm size.

Large Firm: For this study, firm with 50+ employees, consistent with MSME policy medium-enterprise threshold.

Conclusion

This study finds AI adoption is beginning to exert measurable, task-composition and firm-size-dependent displacement effect on formal sector employment in urban Nigeria. OLS results show negative significant association between AI intensity and net employment change, concentrated among high routine-task share firms. Interaction and stratified regressions reveal displacement effect significantly smaller among larger firms, consistent with greater internal redeployment and retraining capacity, robust to propensity-score weighting and alternative thresholds. Findings suggest developing-country structural transformation pathway may face new pressure as accessible AI erodes labour-cost advantage, particularly for smaller firms. Recommended responses include skills reorientation toward non-routine interpersonal and technical tasks, targeted social protection for displaced routine-task workers, and monitoring of AI adoption, prioritising support for smaller firms with limited redeployment capacity.

Frequently Asked Questions (FAQs)

1. Does AI adoption cause job losses in Nigeria?

Survey of 260 Lagos firms shows AI adoption intensity is negatively and significantly associated with net employment change 2024-2026, particularly where routine-task employment share was initially high. Effect is measurable but task-dependent, not uniform.

2. Are small firms more affected than large firms?

Yes. Interaction results show displacement effect is significantly smaller among firms with 50+ employees. Larger firms' diversified tasks and HR capacity enable internal redeployment, while smaller firms lack this buffer.

3. What types of jobs are most at risk?

Routine, repetitive, low-skill tasks: data entry, basic customer service, standard back-office processing – identified as most susceptible to robotic process automation and generative AI.

4. What is internal reinstatement capacity?

Ability to move workers whose specific tasks are automated into other roles within same firm instead of terminating employment. Theorised and empirically shown to vary with firm size.

5. What methods were used in this study?

OLS with net employment change as dependent variable, logit model for probability of job losses, firm-size interaction specification, and fully stratified sub-sample regressions, plus propensity-score-weighted robustness check.

6. Does this mean AI is bad for developing countries?

No. AI raises productivity but changes task composition. Challenge is managing transition. Findings support skills reorientation and social protection rather than restricting adoption.

7. What policies are recommended?

Skills reorientation curricula toward non-routine tasks, targeted social protection for routine-task workers, retraining subsidies prioritized for SMEs, and continued monitoring by Federal Ministry of Labour and National Bureau of Statistics.

8. Is Lagos representative of Nigeria?

Study is limited to Lagos formal sector and may not generalise to other regions with different industrial structures, noted as limitation. Lagos provides leading indicator as commercial hub with early AI adoption.

9. How was AI adoption measured?

Composite AI Adoption Intensity Index (0-10) capturing breadth and depth across functional areas: analytics, RPA, customer service, generative AI tools.

10. Where can I download the full project?

Download complete original material with survey instrument, regression tables, and robustness checks from SCHOLARNESTHUB as formatted academic document.