Reexamine ‘Cookie Jar’ and ‘Big Bath’ accounting Using the backing-out method

Prior research documents income-decreasing earnings management in the situations when true earnings exceed targets by a substantial amount; and, true earnings fall far below targets and accounting reserves are not sufficient to reach targets. These two situations are well known as ‘Cookie jar’ and ‘Big bath’ earnings management. True earnings are defined as pre-managed earnings and measured as reported earnings minus adjusted discretionary accruals. However, the use of pre-managed earnings could induce a spurious association between earnings management and pre-managed earnings above or below benchmarks, which is known as the backing-out problem (Lim and Lustgarten, 2002). This study reexamines the ‘Cookie jar’ and ‘Big bath’ type of earnings management and in particularly addresses the issue of backing-out problem. Using an Australian sample of 3,326 observations covers all ASX listed firms for a period from 1999 to 2006, this study suggests that the finding of ‘Cookie jar’ accounting is not simply a consequence of the backing-out problem. The results show that an income-decreasing earnings management occurred when pre-managed earnings are well above targets. This is consistent with the first argument of ‘Cookie jar’ accounting—firms reduce current earnings in order to save some income for the future. However, the results do not support the ‘Big bath’ accounting.