When are Fair Value Estimates Comparable? Evidence from the SFAS No. 157 Fair Value Hierarchy

Abstract

One motivation for expanding fair value accounting is to improve financial statement comparability by matching the timing of valuation across companies. However, the effect of fair value accounting on comparability also relies on consistent measurement methods and inputs. As the quantity of items reported using fair value estimation expands it is likely that firms will use a larger variety of measurement techniques, which could impede the comparability of fair value estimates across companies. Thus, we investigate whether the difference in exposure to fair value estimates across firms reduces the comparability of these estimates. Due to the increasing discretion in estimates across the SFAS 157 hierarchy of fair value assets, we also predict that inconsistent measurement will reduce comparability more for level 2 and 3 assets relative to level 1 assets. Results from firm-pair regressions provide evidence that firm-pairs with more diverse proportions of assets valued at fair value have less comparable fair value estimates. Furthermore, we find that differences in more discretionary estimates (Level 2 and 3) are more strongly associated with decreases in comparability than differences in less discretionary Level 1 estimates. We extend our main results to two cross-sectional hypotheses, providing evidence that comparability is reduced even further for subsamples where the firm-pairs have the incentive and ability to introduce discretion into fair value estimates. Taken together, these results suggest that standardization of fair value measurement matters for effectively improving comparability through expanded use of fair value estimates.