Page 156 - 35-2

The Effect of the Fair Value Reporting Model on Analyst Forecast Properties: Evidence from Real Estate
Firms

(Markov and Tamayo, 2006; Mikhail et al., 2003). The results also show that even for
sophisticated financial statement users—who should be more capable of reconciling
differences between balance sheets and income statements and may not be greatly affected
by new standards—improved consistency between statements leads to taking significantly

less time to analyze reports and makes the forecasting task more straightforward.

6.2 Re-Examination of Forecast Error
In this subsection, we revisit Liang and Riedl (2014), documenting that there is

an increase in EPS forecast error following the implementation of the requirement that
unrealized holding gains and losses pass through net income. More specifically, we
examine whether their results also exhibit a temporal trend and whether their conclusion
holds after 2010—the final year of their sample. We estimate Equation (1) with the
dependent variable ABS_ERROR, which is the absolute value of the difference between

the I/B/E/S actual earnings and the mean of analyst forecasts for firm i and quarter t in the
final consensus preceding the earnings announcement, scaled by the absolute value of the
mean EPS forecast.

Table 5 presents the results of the regression with the dependent variable ABS_
ERROR. We first replicate Liang and Riedl (2014), the results of which are shown in
Column (1). The coefficients on UK (-0.182, t-statistic = 0.49) and POST (-1.168, t-statistic
= -1.33) are both insignificant. The coefficient on UK×POST (0.892, t-statistic = 2.84) is
significant and positive, suggesting that forecast error is higher among UK firms reporting

under IFRS than it is among US firms reporting under GAAP. Overall, the results thus far
12
are similar to those reported by Liang and Riedl (2014), aligning with their argument that
the net income of UK firms includes non-serially correlated and transient items, increasing

the difficulty of predicting earnings.
Next, we partition our sample by year to examine whether the effect of the change in
accounting standards on forecast error also varies over time. The results of this partitioned

12 We also examine whether our findings on forecast error are robust to an alternative scaler. Similarly,
we refine ABS_ERROR as the absolute value of the difference between the I/B/E/S actual earnings
and the mean of analyst forecasts for firm i and quarter t in the last consensus immediately preceding
the earnings announcement, scaled by the market price at the end of year t-1. Untabulated results
show that our findings are similar to those reported in Table 5.

148