Abstract
Type of the article: Research ArticleAbstractThe performance-related associations of circular practices may differ because waste recovery, recycled-input use, and circularity-related investment involve distinct costs, investments, and organizational mechanisms. This study examines heterogeneous associations between organizational circularity components and economic performance across 25 industrial enterprises in the Ukrainian operating and European comparison subsamples during 2013–2023. Fixed-effects regressions, decomposition, lagged and subgroup models, a non-imputed subsample, a balanced common-support window, pooled interactions, wild-cluster-bootstrap inference, and Holm adjustment were applied.The aggregate circularity index shows no statistically significant short-term association with EBITDA margin (β = +0.0222; p = 0.894).The decomposed components show different coefficient profiles, but formal heterogeneity is sample-sensitive: coefficient equality is not rejected in the full-panel decomposed model (M2) (p = 0.101) but is rejected in the non-imputed model (M3) (p = 0.002). Recycled-input use is predominantly negative and nominally significant in the one-period lag-only (M4) and Ukrainian operating subsample (M6) models, although neither coefficient survives Holm adjustment and the lagged estimate is not reproduced in the combined model. Circularity-related eco-capex is positive in M3 and the European comparison subsample model (M5), where it is supported by unadjusted wild-cluster-bootstrap (p = 0.0198) and Holm-adjusted CR1 inference (p = 0.0273). The country-context interaction block is jointly significant (p = 0.042), with RI × UA as the only individually significant interaction (β = –1.566; p = 0.038).The findings indicate that aggregate measures may conceal component-specific patterns, but most associations remain sample- or specification-sensitive and should not be interpreted causally. Enterprise assessment should distinguish recovery, material-substitution, and investment-modernization channels.