Discussion of ‘‘Evaluating cross-sectional forecasting
models for implied cost of capital’’
Published online: 23 May 2014
Ó Springer Science+Business Media New York 2014
An estimate of the implied cost of capital (ICC) is useful in valuation, investment,
and capital budgeting. The computation of ICC requires earnings forecasts, for
which prior studies generally use analyst forecasts to proxy. Hou et al. (2012)
generate earnings forecasts using a cross-sectional model and thus estimate ICC for
a large sample of ﬁrms, including those not covered by analysts.
Li and Mohanram (2014) extend Hou et al. by considering two other cross-
sectional earnings forecast models (RI and EP models). Li and Mohanram compare
their models with the one in Hou et al. in two ways. First, they show that the
earnings forecasts generated from their models outperform those from the Hou et al.
model on forecast accuracy, forecast bias, and earnings response coefﬁcients.
Second, the ICC computed based on the earnings forecasts generated from the Hou
et al. model exhibits lower correlations with future returns and more abnormal
correlations with risk factors than the ICCs from the RI and EP models. The
improvement in performance is particularly signiﬁcant for small ﬁrms and ﬁrms
without analyst following, where the cross-sectional earnings forecasts are most
needed. Li and Mohanram contribute to the literature by providing improved cross-
sectional earnings prediction models for ICC computation.
My discussion is organized as follows. I ﬁrst discuss the relation between
earnings forecasts and ICC estimates, followed by the two earnings forecast models
considered in LM. I then move to the implication of the ICC estimates based on
these earnings forecast models.
M. Feng (&)
University of Pittsburgh, Pittsburgh, PA, USA
Cheung Kong Graduate School of Business, Beijing, China
Rev Account Stud (2014) 19:1186–1190