To test earnings-smoothing, the level around which income is smoothed must be
specified. As in prior empirical research on earnings smoothing (see, e.g., Archibald
1967; White 1970), I assume that the previous year's EPS is the target.' White (1970,
263-64), for example, advocates the use of this definition for two reasons: (1) it is relatively
simple; and (2) it appears more realistic than other definitions, which require managers
to attain constant annual growth rates. In the context of the timing of asset sales,
these considerations provide the following hypotbesis