FIRM CHARACTERISTICS AND ACCRUAL-BASED EARNINGS MANAGEMENT IN PAKISTAN: A DYNAMIC PANEL GMM ANALYSIS OF LISTED NON-FINANCIAL FIRMS

Abstract

This study examines the association between firm-level characteristics and accrual-based earnings management among non-financial firms listed on the Pakistan Stock Exchange (PSX) for the period covered 2019–2025. Discretionary accruals are estimated using the Modified Jones Model and analyzed in both signed (DA) and absolute (|DA|) form to distinguish the direction from the magnitude of discretionary accrual activity. The final analytical sample comprises 150 continuously listed non-financial firms observed annually from 2019 to 2025, yielding 1,050 firm-year observations. The primary empirical analysis employs a two-step system-GMM specification with Windmeijer-corrected standard errors and collapsed instruments. For signed DA, profitability (ROA) is positive and statistically significant (β = 0.422, p = 0.007), while firm size, leverage and sales growth are not statistically significant; the lagged dependent variable is negative and marginally significant (ρ = −0.100, p = 0.067). For absolute DA, sales growth is positive and statistically significant (β = 0.004, p = 0.039), while firm size is negative and marginally significant (β = −0.007, p = 0.065); leverage, profitability and the lagged dependent variable are not statistically significant. Neither model rejects the absence of second-order serial correlation at the 5% level (AR (2) p = 0.090 and 0.935, respectively). The Hansen test does not reject the instrument-validity null for signed DA (p = 0.055), whereas the baseline |DA| specification is marginally below the 5% threshold (p = 0.049), indicating specification sensitivity. Sensitivity analyses using alternative instrument windows provide partial reassurance but also reveal changes in coefficient significance. Overall, profitability shows a consistent positive association with the direction of discretionary accruals, while sales growth shows a positive association with their magnitude. Firm size provides weaker, specification-sensitive evidence for magnitude, whereas leverage is not statistically significant. The study contributes to the Pakistani earnings-management literature by distinguishing directional and magnitude-based discretionary accrual measures and by providing dynamic-panel estimates together with key specification diagnostics and robustness checks.