The development of an indicator for measuring information quality of discretionary accruals

Abstract

This article proposes a new model, referred to as Information Quality Indicator of Discretionary Accrual (IQIDA), to study earnings management. A limitation of prior studies on earnings management is examined using the two dimensions of decision validity. Based on the notion of ‘information quantity’ and ‘information quality,’ using the original values or absolute values of discretionary accruals (DAs) as the measurement for earnings management could lead to suboptimal judgments and decisions. A new model is proposed as an improvement over those models using DAs. The new model attends to both central tendencies and dispersion degrees of variable sample distributions, which requires a two-step transformation process to achieve.

Keywords:

• earnings management
• discretionary accrual
• information quantity
• information quality

JEL Classification:

• M00
• G30

Notes

¹ The informative effect implies that earnings management has incremental information about future cash flows (Barth et al., 1999).

² The opportunistic effect asserts that opportunistic managers will take advantage of their positions to inflate earnings via earnings management (Healy and Wahlen, 1999).

³ Included among these models are the original Jones Model (1991), the Modified Jones Model by Dechow et al. (1995) and the performance-matched model by Kothari et al. (2005) which have taken DAs as a proxy to measure the earnings management where the DA is measured with the residual terms resulting from a regression process.

⁴ In our research, k is a character that reflects the character of investment decisions, we assume k = 1.

⁵ DA can be calculated by following Jones (1991) method in conjunction with Teoh et al. (1998) cross-section estimates.

⁶ In the first step, DAs are obtained by applying Jones (1991) method in conjunction with Teoh et al. (1998) cross-section estimates. The second step deals with the transformation of DAs from step one into information quality of DAs by the application of Equation 4.

⁷ Following the approach of Dechow and Dichev (2002), we do not hypothesize each relationship; rather we use ‘expectation’ to express it.

⁸ In order to translate our independent variables from firm-year level to firm level, the method by Dechow and Dichev (2002) is adopted. An average is taken for DEBT, SIZE and SD is used for sales revenue and ROA.

⁹ According to Horngren et al. (2009), a correlation coefficient is greater than 0.7 reveals multicollinearity problems.

¹⁰ The IQIDA model composes the central tendency and dispersion degree; therefore, we can identify various propensities of earnings management, such as conservative or aggressive, by central tendency (i.e. the mean of DA). Therefore, we first multiply the IQIDA by negative one (−1) if a manager belongs to a conservative type and the sample is classified into four groups. Second, we examine how IQIDA is affected by each firm character under different earnings management behaviours by managers.