Identifying proxies for discretionary accruals can be a challenge. Often, studies regress
total accruals on only the nondiscretionary variables and assume the residual is discretionary
(e.g., applications of the Jones model). Of course, failure to identify fully the nondiscretionary
component implies the regression residual contains both discretionary and nondiscretionary
components, and the researcher has measured the estimated discretionary and
nondiscretionary components with error. Typically, the explicit conditioning variables for
discretionary accruals, such as earnings or leverage, are generic. These generic variables
can be proxies for many firm characteristics, which make the interpretation of coefficients
of the discretionary accruals in valuation equations challenging.