The Complementarity Dividend: How Digital Capability and Institutional Readiness Condition the Productivity Effects of Artificial Intelligence

Abstract

Generative artificial intelligence has moved the economic debate from whether machines can perform cognitive tasks to whether economies can convert technical exposure into broadly shared productivity growth. This paper develops the Capability-Complementarity Dividend framework, which treats the effect of artificial intelligence as a joint function of technological exposure, digital capability, human capital and institutional readiness. Using the 2023 International Monetary Fund Artificial Intelligence Preparedness Index merged with World Bank development and labor indicators, the paper analyzes 173 countries. The evidence is deliberately descriptive rather than causal. AI preparedness has a zero-order correlation of 0.802 with GDP per person employed, 0.760 with internet use and -0.217 with unemployment.

After conditioning on GDP per capita and population, the aggregate preparedness coefficient on log productivity becomes negative, while the component estimates diverge: digital infrastructure and human-capital components are positively associated with productivity, whereas innovation-integration and regulation components enter negatively. An interaction model also finds a negative association between preparedness and productivity at higher internet use levels. These patterns reject a simple linear readiness narrative. They suggest that capability is multidimensional, that readiness components can create institutional frictions and that productivity gains cannot be interpreted as inclusive without observing distributional and labor-market outcomes. The paper contributes a framework for measuring the complementarity problem and a transparent global baseline for future micro-level causal research.