Skip to main content

Table 10 The cross-sectional correlation between returns and ICC estimated with the MG model

From: Earnings forecasts: the case for combining analysts’ estimates with a cross-sectional model

Model MG CM AF
Interc. Coeff. Interc. Coeff. Interc. Coeff.
Panel A: FM Regression of ICC based on the MG Model
GLS 4.046 0.665 3.796 0.827 3.440 0.576
[1.612] [1.654] [1.623] [2.354]\(*\) [1.303] [1.380]
CT 4.191 0.613 4.086 0.681 4.742 0.300
[1.776]\(*\) [1.808]\(*\) [1.719]\(*\) [2.976]\(**\) [1.817]\(*\) [1.188]
OJ 4.907 0.257 4.350 0.411 4.852 0.167
[1.943]\(*\) [1.499] [2.095]\(*\) [1.785]\(*\) [1.750]\(*\) [0.679]
MPEG 5.386 0.244 5.603 0.223 5.018 0.123
[2.260]\(*\) [1.569] [2.484]\(**\) [1.368] [1.842]\(*\) [0.677]
Composite 3.911 0.609 4.108 0.659 4.344 0.311
[1.491] [1.866]\(*\) [1.840]\(*\) [2.439]\(*\) [1.558] [0.999]
Model MG CM AF
5 − 1 10 − 1 5 − 1 10 − 1 5 − 1 10 − 1
Panel B: Portfolio sorted on ICC based on the MG Model
GLS 2.68 4.34 4.45 4.98 2.52 4.42
[1.269] [1.526] [1.969]\(*\) [1.656]\(*\) [1.064] [1.313]
CT 3.32 7.42 2.34 6.65 0.18 0.46
[1.370] [2.242]\(*\) [0.863] [2.064]\(*\) [0.065] [0.117]
OJ 2.52 − 0.50 2.38 3.44 − 0.79 1.09
[1.062] [− 0.161] [0.978] [1.129] [− 0.279] [0.298]
MPEG 1.01 0.31 1.97 3.22 0.17 − 0.24
[0.471] [0.106] [0.861] [1.145] [0.060] [− 0.066]
Composite 2.48 2.02 2.33 3.66 0.50 − 0.42
[1.138] [0.719] [0.862] [1.156] [0.177] [− 0.104]
  1. Panel A presents the average coefficients of Fama–Macbeth regressions of the value-weighted excess market return on the excess ICC, both in excess of the risk free asset. We use the yield on the U.S. 10-year government bond as the risk-free asset. The Newey–West t-statistics with three-lag periods are presented in brackets in Panel A. Panel B reports the value-weighted excess portfolio returns sorted on the ICC. We sort stocks into quintiles and deciles based on ICC at the end of June each year from 1985 to 2012. We report results for long–short strategies of 5 − 1 (fifth quintile minus first quintile) and 10 − 1 (tenth decile minus first decile). To compute the excess returns, we use the one-month Treasury bill rate downloaded from Kenneth French’s data library. OLS t-statistics are presented in brackets in Panel B and excess returns are annualized and are expressed in percentages. The ICC models used in this table are estimated with earnings estimated by the Mohanram and Gode (2013) (MG) model, Combined Model (CM), and raw analysts’ forecasts (AF). We estimate the ICC based on Claus and Thomas (2001) (CT), Easton (2004) (MPEG), Gebhardt et al. (2001) (GLS), and Ohlson and Juettner-Nauroth (2005) (OJ) methods. In addition, we include a Composite ICC, which is the average of all of the above-mentioned approaches. \(**\) and \(*\) denote significance based on one-tailed tests at 0.01 and 0.05 levels, respectively. The results from the AF and CM are included in order to facilitate a comparison with the MG model