The Effect of Fair Value Versus Historical Cost Reporting Models on Analyst Behavior

• In Event: Analysts and Markets

Abstract

This paper examines how the reporting models¬—the fair value and the historical cost models—affect analyst behavior. Using a sample of US and UK real estate firms reporting their net operating assets at historical cost and fair value, respectively, and exploiting UK firms experiencing IFRS adoption in 2005, we examine how the two reporting models affect the degree of differential opinions regarding future prospects and the length of time for analysts to generate reports. We find that analysts have less differential opinions for UK firms reporting fair value estimates than for US firms reporting historical costs. We further document that a change from the partial fair value reporting model (UK domestic standards) to the full fair value reporting model (IFRS) temporarily increases the degree of dispersion in the earnings forecasts; however, the increase in dispersion disappears several years after adoption. In addition, we find that it takes more time for analysts to issue forecast reports for firms reporting under the fair value model than under the historical cost model. We also provide evidence to suggest that when both the income statement and the balance sheet are reported under the full fair value model, financial statements become more straightforward, reducing analysts’ forecast revision response time. Finally, we find that the increase in forecast error exhibits a decreasing trend in the post-IFRS period. Overall, we find a differential effect of fair value versus historical cost reporting model on forecast dispersion and forecast revision response time and document that the change in accounting standards has a time-varying effect on analyst behavior.

Authors

• Chunlai Ye, Texas A&M International University

• Lin-Hui Yu, National Taiwan University