Conservatism Correction in Linear Information Models

Recent research concerned with enhancing conservatism corrections of linearinformation models (LIMs) reports a decrease in bias as compared to the Ohlson (1995)model. However, inaccuracy is not significantly reduced. These findings raise twoquestions: First, are LIMs able to capture unconditional conservatism? Second, ifconservatism can be captured, then why is accuracy not markedly improved? Withregard to the first question, we find that conservatism is captured when theFeltham/Ohlson (1995) model is estimated according to the modification suggested byChoi/O’Hanlon/Pope (2006). On average, one dollar of unrecorded reserves, measuredas the estimated reserve by Penman/Zhang (2002), results in a correction of marketvalue forecasts of approximately one dollar. Furthermore, our results suggest noimprovement of the conservatism corrections for the following cases: (1) disaggregatingbook value into operating and financial assets and (2) estimating the Feltham/Ohlson(1995) model via the valuation function...