Earnings manipulation in Bangladeshi listed firms: Sectoral patterns, Covid-19 anomalies, and discriminant reliability of the Beneish M-Score

Abstract

This study examines the sectoral prevalence of earnings manipulation and evaluates the discriminant reliability of the Beneish M-Score model in Bangladeshi listed firms. Using financial statement data from 127 companies across 12 sectors over the period 2018 to 2022, the Beneish M Score is applied to identify potential manipulation, producing 635 firm-year observations. The results show that earnings manipulation is unevenly distributed across sectors, with an overall prevalence of 20%. Engineering and Miscellaneous sectors exhibit the highest susceptibility, while Pharmaceuticals and Ceramics display relatively lower manipulation levels. Volatile patterns in Travel & Leisure and Food & Allied sectors indicate sensitivity to economic shocks and industry-specific pressures. The study further identifies a methodological shift in manipulation tactics during the COVID-19 pandemic (2020–2022). While the market-wide frequency of manipulation remained statistically stable, the pandemic period introduced distinctive anomalies characterized by a significant rise in the Days' Sales in Receivables Index (DSRI) and a significant decline in Total Accruals to Total Assets (TATA), indicating a tactical pivot toward aggressive revenue recognition alongside increased use of income-deferral accrual adjustments to obscure the true impact of economic shocks. Linear Discriminant Analysis identifies TATA (Total Accruals to Total Assets), GMI (Gross Margin Index), DEPI (Depreciation Index), SGI (Sales Growth Index), and DSRI (Days’ Sales in Receivables Index) as significant variables separating manipulation and non-manipulation firms, with Total Accruals to Total Assets emerging as the most influential indicator. Classification evaluation shows that the model correctly identifies 98.8% of non-manipulation firms and 71.9% of manipulation firms, with validation accuracy consistently between 91% and 93% across cross-validation and temporal holdout testing. The findings suggest that the Beneish M Score functions as a reliable screening tool in the Bangladeshi context, while sectoral differences highlight the need for targeted regulatory monitoring