External accountant links, portfolio risk and SME failure prediction: evidence from Norwegian registry data

Purpose This study aims to examine whether external-accountant affiliation improves small and medium-sized enterprise failure prediction and whether its value reflects reporting continuity, direct co-client exposure or persistent portfolio heterogeneity. Design/methodology/approach Norwegian registry data comprise 2,259,629 firm–year observations from 454,837 firms. Logistic regression and extreme gradient boosting are evaluated for 12- and 36-month bankruptcy and adverse-exit outcomes using temporal validation, block-wise ablation, bootstrap uncertainty and an independent portfolio test. Findings Accountant variables improve average precision and top-tail capture. For 12-month bankruptcy, average precision rises from 0.045 to 0.063 and top 1% capture from 20.2% to 27.4%. Status, transition and portfolio blocks largely reproduce the gain because they overlap in encoding reporting continuity. Direct co-client exposure contributes little and no consistent incremental value remains among firms with a current accountant. An earlier residual portfolio score predicts future events (IRR 1.17; 95% CI 1.08–1.27). The estimate remains similar after accountant financial controls, although precision declines (IRR 1.13; 95% CI 0.99–1.29). Research limitations/implications The predictive design and non-random matching do not identify contagion, accountant quality or causal protection. Practical implications Registry-observable accountant information can support capacity-constrained, human-led monitoring by banks and supervisors, but not automatic adverse action. Originality/value The study introduces external accountants as a reporting-infrastructure link and separates overlapping continuity signals from direct co-client exposure and out-of-time portfolio persistence.