Overconfident CEOs, Earnings Manipulation, and the Limits of Governance in Concentrated-Ownership Markets

This is the authors' abstract. We don't add key points for this paper.

This study investigates the relationship between chief executive officer (CEO) overconfidence and earnings management in non-financial companies listed on the Saudi Exchange (Tadawul) from 2018 to 2024. Using 686 firm-year observations from 98 firms in markets where measures of executive stock options are unavailable or inappropriate for researchers, we create a formative principal component composite (OC_COMPOSITE) comprising an excess investment residual, a relative compensation ratio, and a 3-year acquisition frequency index. The composite meets pre-specified validation criteria for pairwise correlations, variance inflation factors, eigenvalues, and factor loadings. CEO overconfidence is positively associated with absolute discretionary accruals and aggregate real earnings management in both firm- and year-fixed-effects regressions and System Generalised Method of Moments (GMM). Based on the preferred System GMM specifications, the estimated coefficients associated with a one-standard-deviation increase in OC_COMPOSITE are equivalent to approximately 38% of the sample mean of absolute discretionary accruals and 44% of the sample mean of aggregate real earnings management; these are mean-relative effect-size benchmarks, not literal percentage changes in the dependent variables. Big Four audit engagement attenuates the overconfidence–accrual association but does not significantly moderate the real-activities relationship, whereas ownership concentration strengthens the association across both earnings-management channels, particularly for real activities. The gender interaction is interpreted as suggestive because female-led firm-years are rare in the sample. A pre-/post-2021 comparison further indicates attenuation of the overconfidence–earnings management association in the later period, although the observational design and gradual temporal pattern preclude causal attribution to the 2021 reforms. The study extends behavioural corporate finance and upper echelons theory to a concentrated-ownership emerging-market setting and develops a transferable three-proxy approach to measuring CEO overconfidence in institutional environments where conventional option-based measures are unavailable.