Sample profile Methodology 1 Data sources

1230 ABRES denotes the accrual quality as measured by the absolute residual from the regression of total current accrual on current, last and next period cash flows. ABSDATCAABSDATA denotes the discretionary current accruals and discretionary total accruals. Table 3.1 Sample Number of companies with estimated earnings forecasts 213 Eliminated due to change in accounting year end and insufficient data to calculate cost of equity 10 203 Table 3.2 Sample size based on available data for the calculation of earnings quality variables Market Capitalization ABRES 141 31 ABSDATCA ABSDATA 151 32 3.3 Variable definition and measurement 3.3.1 Accruals quality There are two measures of accrual quality as operationalized in Aboody, Hughes and Liu 2005 based on models developed by Jones 1991, Dechow, Sloan and Sweeney 1995 and Dechow and Dichev 2002 . The first measure of accruals quality is defined as the amount of discretionary accruals DA or abnormal accruals. Large DA is associated with low quality. First non-discretionary accruals NDA is measured from a model developed by Jones 1991 and subsequently modified by Dechow et al 1995. This modified version is used in many earnings management studies. The estimation is done in the following way time t refers to year 2004: NDA t A t-1 = 1A t-1 +  REV t - REC t A t-1 + PPE t A t-1 Where, NDA t - Non-discretionary accruals at time t 1231 A t-1 - Total assets at t-1 REV t - Change in revenues REC t - Change in receivables PPE t - Plant,property and equipment at t The coefficients ,  and  are estimated from coefficients a, b and c of the following cross-sectional regression by industry : TA t A t-1 = a 1A t-1 + b REV t A t-1 + PPE t A t-1 +  t Then, DA = TA t A t-1 - NDA t A t-1 Where, TA = Total accruals = Change in current assets – Change in current liabilities – Change in cash + Change in short term debt – Depreciation In the following analysis the absolute DA is taken and the acronym ABSDATA absolute discretionary total accruals is assigned to the variable. The following analysis will also use the current accrual variation of the above model as given below. NCA t A t-1 = 1A t-1 +  REV t - REC t A t-1 Where, NCA t - Non-discretionary current accruals at time t The coefficients ,  and  are estimated from coefficients a, b and c of the following cross-sectional regression by industry : TCA t A t-1 = a 1A t-1 + b REV t A t-1 +  t Then discretionary current accruals DCA, DCA t = TCA t A t-1 - NCA t A t-1 Where, TCA t = Total current accruals