Fair Value Accounting and Analyst Forecast Properties

• In Event: Analyst Forecasts and Performance

Abstract

This study examines the effect of fair value accounting on the behavior of analysts. Using a measure of firms’ fair value assets holdings intensity, we provide evidence that firms with higher holdings of fair value assets have more accurate analyst earnings forecasts and lower levels of forecast dispersion. Our results are consistent with the value relevance theory of fair value accounting. SFAS No. 157, Fair Value Measurements, establishes a framework of fair value measurement which requires fair values to be disclosed by levels, where Level 1 and Level 2 use either direct or indirect observable inputs while Level 3 use unobservable inputs. Using the disclosure required by SFAS No. 157, we find significant positive associations between analyst forecast accuracy and Level 1 and Level 2 fair value intensity while also finding a significant negative association with Level 3 fair values. This result suggest that all three levels of fair value disclosures impact analyst forecast properties with Level 1 and 2 enhancing analyst ability to make accurate forecasts and Level 3 items diminishing forecast accuracy. We further decompose our analysis into analyst consensus and uncertainty. And we find that firms with higher levels of fair value assets are associated with lower analyst uncertainty. In exploratory analysis, we find that higher levels of fair value asset holdings are associated with higher levels of analyst following, higher levels of analyst sector specialization, and higher frequency of forecasting. Our results contribute to the debate over fair value accounting by showing the impact of fair value accounting upon an important participant in the capital markets.

Authors

• Douglas Ayres, University of Tennessee

• Xuerong Huang, Ball State University

• Mark J Myring, Ball State University