Determinants of Managerial Earnings Guidance Prior to Regulation Fair Disclosure and Bias in Analysts' Earnings Forecasts
53 Pages Posted: 12 Aug 2004
Date Written: February 2002
Prior to Regulation Fair Disclosure (Reg FD) some management spent considerable time and effort guiding analyst earnings estimates, often through detailed reviews of analysts' earnings models. In this paper I use proprietary survey data from the National Investor Relations Institute to identify firms that reviewed analysts' earnings models prior to Reg FD and those that did not. Under the maintained assumption that firms conducting reviews implicitly or explicitly guided analysts' earnings forecasts, I document firm characteristics associated with the decision to provide private managerial earnings guidance. Then, I document the characteristics of 'guided' versus 'unguided' analyst earnings forecasts. Findings demonstrate an association between several firm characteristics and guidance practices: managers are more likely to review analyst earnings models when the firm's stock is highly followed by analysts and largely held by institutions, when the firm's market-to-book ratio is high, and its earnings are important to valuation (high Industry-ERC R2), but hard to predict because its business is complex (high # of Segments). A comparison of guided and unguided quarterly forecasts indicates that guided analyst estimates are more accurate, but also more frequently pessimistic. An examination of analysts' annual earnings forecasts over the fiscal year does not distinguish between guidance and no guidance firms; both experience a "walk down" in annual estimates. To distinguishing between guidance and no guidance firms, one must examine quarterly earnings news: unguided analysts walk down their annual estimates when the majority of the quarterly earnings news is negative, guided analysts walk down their annual estimates even though the majority of the quarterly earnings news is positive.
JEL Classification: M41, M45, G29, K22
Suggested Citation: Suggested Citation