Reliability of R&D capitalization: Evidence from ex post impairment in China

Unlike prior studies that investigate research and development (R&D) accounting as a dichotomous choice between capitalizing vs. expensing, this study identifies low-reliability R&D capitalization by the occurrence of ex post impairment of capitalized R&D costs. I find that low-reliability capitalization is associated with higher discretionary accruals but fails to signal future innovation, whereas normal capitalization without subsequent impairment lacks earnings aggressiveness and predicts future innovation positively, compared to expensing firms. Next, this study shows that Big 4 and industry specialist auditors improve reliability by notably decreasing the likelihood of low-reliability R&D capitalization. The results remain robust after controlling for R&D investment intensity and potential endogeneity in the capitalization decision. Additional tests show that managers strategically time the recognition of impairment for big-bath and earnings-smoothing purposes, and that analyst coverage does not help differentiate between low-reliability and normal R&D capitalization. Collectively, this paper increases our understanding of R&D accounting and auditing and contributes to the debate on the reliability of R&D capitalization.