Page 139 - 35-2

NTU Management Review Vol. 35 No. 2 Oct. 2025

well-developed, increasing the comparability of the fair value and historical cost models. 2
We first examine whether the transition from a partial fair value model under UK
domestic standards to a full fair value model under IFRS leads to any difference in forecast
dispersion among UK firms. We find an immediate but temporary increase in forecast

dispersion following IFRS adoption due to earnings reported under the full fair value
model being exposed to temporal changes in the market to a greater extent and including
a higher number of less-correlated items. While there is an adjustment period following

IFRS adoption during which firms have higher forecast dispersion, analysts have less
divergent opinions on earnings per share (hereafter EPS) forecasts on average under the

fair value reporting model. This difference in dispersion is economically meaningful. In
2013-2014, UK firms exhibited forecast dispersion approximately 43% lower than that of
US firms.

Moreover, upon income statements and balance sheets becoming more consistent
with each other following IFRS adoption, we find that analysts spend less time reconciling
the differences between the two, leading to a reduction in overall forecast revision response

time. As with the time-varying effect on forecast dispersion, we find that this change in
forecast duration is not static throughout the post-IFRS period. In fact, the difference in
forecast duration is not evident within the first six years following IFRS adoption; still,
the difference emerges in later years, with the reduction in forecast revision response
time becoming more pronounced over time. We note, however, that while the improved

consistency shortens the overall forecast revision time, the average time is generally longer
among UK firms under the fair value model than it is among US firms under the historical
cost model.

Finally, motivated by our evidence that the effect of IFRS adoption is not static
throughout the sample period, we re-examine whether the shift from the partial fair value
model to the full fair value model also exhibits a time-varying effect on forecast error.
Liang and Riedl (2014) find that, in their sample period of 2002-2010, forecast error
is higher under the UK full fair value model than it is under either the UK partial fair

value model or the US historical cost model. Therefore, the change in forecast dispersion

2 For more detailed discussion, please refer Liang and Riedl (2014), p. 1152.

131