Verdi

Management Forecasts, Disclosure Quality, and Market Efficiency : 

Management Forecasts, Disclosure Quality, and Market Efficiency Jeffrey Ng, İrem Tuna, and Rodrigo Verdi

What do we do?: 

What do we do? We examine short-term and long-term market reaction to management forecasts We test the effect of disclosure quality on the long-term market reaction

Motivation: 

Motivation There is substantial evidence of market underreaction to news events e.g., PEAD (Bernard and Thomas, 1989, 1990) Conservatism theory (Barberis et al., 1998) Rational structural uncertainty (Brav and Heaton, 2002) Is there a post-management-forecast drift? Is underreaction to earnings news a function of perceived credibility of the news (e.g., stronger underreaction to good news)?

Motivation: 

Motivation Management forecasts constitute an interesting setting to study reaction to news because of credibility concerns (voluntary disclosure) Stronger short-term market reaction for bad news forecasts despite no difference in bias (Rogers and Stocken, 2005) Management forecasts allow us to examine whether disclosure quality affects the market reaction to news

Overview of Results: 

Overview of Results Figure 1a – Annual earnings forecasts

Anilowski, Feng, and Skinner (JAE 2007): 

Anilowski, Feng, and Skinner (JAE 2007)

Kato, Skinner, and Kunimura (WP 2007): 

Kato, Skinner, and Kunimura (WP 2007)

Hypothesis I: 

Hypothesis I PEAD literature provides evidence of underreaction to earnings surprises Both conservatism and rational structural uncertainty can explain an underreaction phenomenon We hypothesize that credibility concerns related to management forecasts can exacerbate underreaction to earnings news H1: Future stock returns are positively associated with management forecast news

Hypothesis II: 

Hypothesis II Larger stock price reaction to bad news forecasts than to good news forecasts (e.g., Hutton et al., 2003) Typical explanation is that bad news forecasts are more credible than good news forecasts No difference in the forecast bias between bad news and good news forecasts (Rogers and Stocken, 2005) H2: The magnitude of the future returns is larger for firms forecasting good news

Hypothesis III: 

Hypothesis III Better disclosure may help investors understand the future cash flow implications of an information signal Prior literature provides evidence that better disclosure mitigates the magnitude of market inefficiency PEAD (Francis et al., 2005; Kimbrough, 2005) Accruals anomaly (Richardson et al., 2005; Levi, 2007) H3: The magnitude of the hedge portfolio returns from the PMFD trading strategy is lower for firms whose forecasts are of higher quality.

Sample Description: 

Sample Description Follow Anilowski et al. (2007): 17,184 (14,890) forecasts of annual (quarterly) EPS 6,369 (5,859) annual (quarterly) forecasts that do not overlap with earnings announcements Ordinary shares listed on NYSE / AMEX / NASDAQ from 1996 to 2005

Key Variables: 

Key Variables Surprise= (Manag. Forec – Analyst Forec) / Price Abnormal returns: Size-adjusted buy-hold returns Size-BM adjusted buy-hold returns Factor alphas (3-factor, 4-factor, 5-factor) Short-term: 3-day around the forecast Long-term: 12-month subsequent to the forecast month

Short- and long-term reaction: 

Short- and long-term reaction Return = β0 + β1 Good News + β2 Surprise + β3 Surprise x Good News + ∑ βj Surprise x Controls + ∑ βm Controls + ε Return is either: - AbRet3d (3-day return around forecast) - AbRet (future 12-month return) Fama-MacBeth regressions with Newey-West corrected standard errors

Table 3 – FRC Regressions: 

Table 3 – FRC Regressions

Hedge Portfolio Analyses: 

Hedge Portfolio Analyses We sort the observations into quintiles based on the previous year’s distribution of forecast surprises Hedge portfolio strategy: Buy shares of firms in Q5 (extreme good news) Short shares of firms in Q1 (extreme bad news) Abnormal returns: Size-adjusted and size-BM-adjusted buy-hold returns 3-factor, 4-factor, and 5-factor alphas

Hedge Portfolio – Annual Forecasts: 

Hedge Portfolio – Annual Forecasts

Short- and Long-term returns: 

Short- and Long-term returns

Mean Returns by Quarter: 

Mean Returns by Quarter

Median Returns by Quarter: 

Median Returns by Quarter

Factor Alphas – Annual Forecasts: 

Factor Alphas – Annual Forecasts

Conservatism vs. Structural Uncertainty: 

Conservatism vs. Structural Uncertainty Analyst forecast dispersion Proxy for precision of pre-forecast signals Lower dispersion, greater conservatism Intraday return volatility Proxy for the degree of uncertainty generated by the disclosure signal Higher volatility, more structural uncertainty

Research Design: 

Research Design AbRet = β0 + β1 QSurprise + β2 QSurprise * Dispersion + β3 QSurprise * Intraday Vol + β4 Dispersion + β5 Intraday Vol + ∑ βj Risk Controls Prediction: β2 < 0 and β3 > 0 AbRet (future 12-month size-adjusted return) Fama-MacBeth regressions with Newey-West corrected standard errors QSurprise is a quintile variable re-scaled from 0 to 1. Dispersion and Intraday Vol are dummy variables based on median.

Table 6 – Theories: 

Table 6 – Theories

Disclosure Quality (H3): 

Disclosure Quality (H3) Accuracy = -1*|Earnt-1 – Man Forecastt-1| / Price Precision: Point versus range forecasts Horizon: # of days between forecast and Fiscal-year end

Research Design: 

Research Design AbRet = β0 + β1 QSurprise + ∑ βi QSurprise * Disclosure Quality + ∑ βj Disclosure Quality + ∑ βm Risk Controls Prediction: βj < 0 for Accuracy βj < 0 for Precision βj > 0 for Horizon QSurprise is a quintile variable re-scaled from 0 to 1. Disclosure Quality are dummy variables based on median.

Table 7 – Disclosure Quality: 

Table 7 – Disclosure Quality

PMFD vs. PEAD: 

PMFD vs. PEAD

Summary of Findings: 

Summary of Findings The short-term reaction larger for bad news The future long-term reaction is larger for good news A hedge portfolio earns annual abnormal returns above 25% Both analyst forecast dispersion and intraday volatility explain the cross-sectional variation in the PMFD Hedge returns smaller for firms with higher prior forecast accuracy

Implications: 

Implications Significant underreaction to management forecasts. Results consistent with behavioral theories of conservatism (Barberis et al., 1998) and rational structural uncertainty (Brav and Heaton, 2002) Disclosure quality appears to reduce the market underreaction to management forecasts An unresolved question is why the effect is not arbitraged away – As Merton (1987) states: “…an anomaly must uncovered and learned before one can arbitrage…even when learned, there are limits to arbitrage…”