How Does Income Inequality Emerge in the Age of AI?: From the Market to Internal Distribution Structures

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

This paper reframes income inequality in the age of AI not as a direct consequence of AI as a technology, but as the outcome of a process through which value formed with the contribution of AI is attributed to different parties through distributional decisions within the firm. AI can contribute to value formation, but it is not a party that receives the resulting gains as income. Whether those gains are allocated to wages, executive compensation, dividends, retained earnings, or other uses is determined through internal distributional decisions under an institutionally defined structure of distributional decision rights. The same AI can therefore give rise to different degrees of income inequality. The paper also shows that, while the labor share is useful as an aggregate indicator, it combines executive compensation and employee wages and therefore cannot sufficiently distinguish differences in internal value attribution. Industry-level data show that plastic and cosmetic surgery and daycare centers have nearly identical labor shares, yet their executive compensation distribution ratios are 30.1% and 9.2%, respectively. On this basis, the paper proposes a multi-stage measurement framework that decomposes the labor share into the executive compensation distribution ratio and the wage distribution ratio, and then uses the control pay ratio to measure the distributional relationship between the controlling subject and employees. When AI adoption involves workforce reductions, however, changes in employee numbers and in the income of workers who leave the firm must also be observed.