Artificial Intelligence-driven Audit Risk Reduction in Thin Capitalisation Engagements: Implications for External Audit Effectiveness among South African Multinational Corporations

Abstract

This study examined the effect of artificial intelligence-driven audit techniques on audit risk reduction associated with thin capitalisation engagements. It also examined their implications for external audit effectiveness among South African multinational corporations. Panel regression and System Generalised Method of Moments (GMM) models were used to analyse the data gathered from 14 listed multinational corporations in Johannesburg Stock Exchange. The adoption of artificial intelligence significantly reduced audit risk at β = −0.371, p < 0.001. It improved external audit effectiveness (β = 0.123, p < 0.05) after controlling for endogeneity through GMM instrumentation. Thin capitalisation and audit risk mitigation activity were positively related (β = 0.576, p < 0.001), contrary to conventional expectations. Artificial intelligence adoption did not moderate the relationship between thin capitalisation and audit risk reduction (β = 0.231, p < 0.10). Audit risk reduction partially mediated the AI–effectiveness relationship (indirect effect = 0.094). The lagged dependent variable in GMM (α = 0.446, p < 0.05) confirmed that audit effectiveness is path-dependent. The study contributes to the limited literature on the causal interrelationship between artificial intelligence and auditing in Africa and extends institutional theory within the context of tax compliance and audit quality governance in emerging markets. The findings underscore the need to enhance audit effectiveness and mitigate audit risks that can be traced to thin capitalisation engagements.