Earnings autocorrelation and accounting restatements

Abstract

Purpose This study aims to examine whether earnings autocorrelation affects the risk of an accounting restatement. Design/methodology/approach This paper uses logistic regression and identifies restatements between 2004 and 2016. Following prior research, (Dao et al., 2012; Francis and Michas, 2013; Francis et al., 2013; Lennox and Li, 2014; Lobo and Zhao, 2013; Paterson and Valencia, 2011), this study allows time between the end of our sample period and the date that this study obtained the restatement data because it takes time for material misstatements to be identified. Findings Bryan et al. (2018) report a negative association between autocorrelation and audit fees, suggesting that auditors view lower autocorrelation as increasing inherent risk. This study finds that autocorrelation is negatively related to accounting restatements, implying that although auditors react to lower autocorrelation by increasing their risk assessments (Bryan et al., 2018), their risk response is not sufficient. This study finds that autocorrelation has a fairly large effect: a shift from the 75th to the 25th percentile of autocorrelation is associated with a 9.38% increase in the likelihood of a restatement. Originality/value This study contributes to the stream of research that investigates the determinants of restatements. Not only do this study identifies autocorrelation as a factor that contributes to restatements, but importantly, this study’s results reveal a fairly substantial effect size: a shift from the 75th to the 25th percentile of autocorrelation is associated with a 9.38% increase in the likelihood of a restatement. While Bryan et al. (2018) find that autocorrelation affects audit fees, this study links autocorrelation to a more drastic consequence: accounting restatements.