J

In the empirical literature, various attempts have been made to assess the likelihood or extent of earnings management. Th e statistical procedures that have been put forward can be classifi ed into cash fl ow management models, for example, the detection of discretionary R&D or advertising outlays, accruals models, and combined approaches, for example, distributional tests and rankings (see Goncharov, 2005, for an overview). Th e Jones Model, proposed by Jennifer J. Jones (1991), falls in the category of accruals models and builds on and extends previous work by Healy (1985), DeAngelo (1986), and McNichols and Wilson (1988). All these studies jointly test a model of discretionary accruals and the existence of earnings management. In her paper, Jones addresses the question whether fi rms that are subject to import relief investigations by the U.S. International Trade Commission (ITC) manage their accruals to show lower earnings

fi gures. Even though the fi rms under scrutiny would benefi t from managing earnings and performance downward, the ITC traditionally did, unlike other addressees of fi nancial reports, neither address nor adjust for such behavior. Th e study thus uses a unique research setting in which clear earnings management incentives prevail. Th e basic idea of the model applied for analysis is that the accrual component of earnings, that is, total accruals, can be understood as being composed both of nondiscretionary and discretionary accruals where the latter proxy for the extent of earnings. Both accrual components are not directly observable from fi nancial statements. Following Healy (1985) nondiscretionary accruals can be understood as accounting adjustments to the fi rm’s cash fl ows authorized by the accounting standard-setting organizations while discretionary accruals are adjustments to cash fl ows opportunistically chosen by the manager “from an opportunity set of generally accepted procedures defi ned by accounting standard-setting bodies” (p. 89). Th e Jones Model provides a structure for empirically estimating both components of total accruals. Total accruals, the dependent variable, are measured as changes in noncash current assets less nonfi nancial current liabilities minus depreciation and amortization expense. Th e independent variables are intended to pick up the major drivers of nondiscretionary accruals resulting in the error term grasping the discretionary accrual component. According to this structure, Jones argues that changes in revenues should be included in the model as an independent variable to control for a fi rm’s economic environment as they are, even though not being completely exogenous, a measure for the fi rm’s operations before managers’ manipulations. Likewise,

a variable measuring gross property, plant and equipment, is included to control for nondiscretionary depreciation expenses. All variables in the model are scaled by lagged total assets for statistical reasons and the model is fi nally estimated using an ordinary least square approach. Compared to previous models, the main advantages of the Jones Model are that it (1) is more sophisticated than simple models in which discretionary accruals are supposed to equal total accruals; (2) considers total accruals, that is, a comprehensive measure, instead of a single accrual component to detect earnings management behavior; and (3) neither supposes nondiscretionary accruals to be equal over time nor to have a mean of zero in the estimation period (therefore, the intercept is dropped in the estimation procedure). Hence, the Jones Model allows the nondiscretionary and the discretionary parts of accruals to vary with the economic circumstances a fi rm faces in each reporting period. However, the Jones Model has also been criticized for various reasons. For example, it was argued that the model does not capture earnings management related to the revenue recognition process. Moreover, it is said to overestimate the level of discretionary accruals within periods of extreme fi nancial performance. Th ese kinds of problems have led to

modifi cations of the original Jones Model, where the most prominent is the so-called Modifi ed Jones Model, which was proposed by Dechow et al. (1995). Due to the problems of accruals models in detecting earnings management in general, distributional tests and rankings have more oft en been used in recent research, particularly in crosscountry studies (see Leuz et al., 2003).