ANOMALIES AND NEWS JOEY ENGELBERG (UCSD) R. DAVID MCLEAN (GEORGETOWN) JEFFREY PONTIFF (BOSTON COLLEGE)

Transcription

1 ANOMALIES AND NEWS JOEY ENGELBERG (UCSD) R. DAVID MCLEAN (GEORGETOWN) JEFFREY PONTIFF (BOSTON COLLEGE) 3 RD ANNUAL NEWS & FINANCE CONFERENCE COLUMBIA UNIVERSITY MARCH 8, 2018

2 Background and Motivation 2 Academic research has uncovered many predictors of cross-sectional stock returns E.g., long-term reversal, size, momentum, book-to-market, accruals, and post-earnings drift. This anomalies research goes back to at least Blume and Husick (1973) Yet 43 years later, academics still cannot agree on what causes this return predictability Important Question: What explains cross-sectional return predictability?

3 Theories of Stock Return Predictability 3 Three popular explanations for cross-sectional predictability Differences in discount rates, e.g., Fama (1991, 1998) Mispricing, e.g., Barberis and Thaler (2003) Data-mining, e.g., Fama (1998) This Paper: Uses 97 anomalies along with firm-specific news and earnings announcements to differentiate between the three explanations

4 The Discount Rate Story 4 Cross-sectional return predictability is expected The predictability may be surprising to academics, but it is not to other market participants Ex-post return differences reflect ex-ante differences in discount rates There are no surprises here Ex-post returns were completely expected by rational investors ex-ante E.g., Fama and French (1992, 1996)

5 Discount Rates and News Anomaly Returns around an Earnings Announcement Long Short

6 Mispricing Biased Expectations 6 Investors have systematically biased expectations of cash flows and cash flow growth Expectations are too high for some stocks, too low for others The anomaly variables are correlated with such expectations New information causes investors to update their beliefs, which corrects prices, and creates the return-predictability. Goes to back to at least (Basu, 1977)

7 Mispricing and News Anomaly Returns around an Earnings Announcement Long Short

8 Data Mining 8 As Fama (1991) suggests, academics have likely tested thousands of variables It s not surprising to find that some predict returns in-sample Realization of a multiple testing bias in empirical research dates at least back to Bonferroni (1935) This is stressed more recently in the finance literature by Harvey, Lin, and Zhu (2015).

9 Mispricing vs. Data Mining 9 Most anomalies focus on monthly returns Stocks with high (low) monthly returns likely had good (bad) news during the month A spurious anomaly would therefore likely perform better insample on earnings days and news days Do anomaly strategies still have high returns on news and earnings days after controlling for this?

10 Our Findings 10 Anomaly returns are higher by 7x on earnings announcement days 2x on corporate news days

11 11 Returns in Event Time (3-day window)

12 Financial Analysts 12 We also examine financial analysts forecasts errors For stocks in long portfolios, forecasts are too low For stocks in the short portfolios, forecasts are too high

13 Interpretation Difficult to Reconcile with Risk 13 Hard to tie stock-price reactions to firm-specific news to systematic risk Anomalies do worse on days when macroeconomic news is announced Anomalies do worse when market returns are higher, i.e., anomalies have a negative market beta Risk cannot explain the analyst forecast error results

14 Interpretation Not (just) Data Mining 14 A spurious anomaly would likely perform better insample on earnings days and news days However, controlling for contemporaneous monthly return, anomalies still perform better on news days Out-of-sample anomalies perform better on news days and have the forecast error results The relation between anomalies and news is stronger in small stocks

15 Interpretation Consistent with Mispricing 15 The results are easy to explain with a simple behavioral theory of biased expectations Expectations are too high for some stocks, too low for others The anomaly variables are correlated with such expectations New information causes investors to update their beliefs, which corrects prices, and creates the return-predictability. The analyst forecast error results fit this framework too

16 Our Place in the Literature 16 We build on previous studies showing anomalies predict returns on earnings announcement days E.g., Chopra Lakonishok and Ritter (1992), La Porta et al. (1994), and Sloan (1996) Edelen, Kadlec, and Ince (2015) anomalies and institutions Our paper: Investigates 6 million news days that are not earnings announcements Uses 97 anomalies compare across anomaly types Relates a large sample of anomalies to analyst forecast errors Develops new data-mining tests

17 The Anomalies 17 Choosing the Anomalies The list is from McLean and Pontiff (2016) The anomaly has to be documented in an academic study Primarily top 3 finance journals Can be constructed with COMPUSTAT, CRSP, and IBES data Cross-sectional predictors only

18 The Anomalies in Anomalies in Total Oldest: Blume and Husic (1973) Stocks sorted each month into long and short quintiles 16 of the 97 variables are binary Can be replicated with CRSP, COMPUSTAT and I/B/E/S Average pairwise correlation of anomaly returns is low (.05)

19 The Sample 19 Earnings announcements from COMPUSTAT Corporate news from the Dow Jones Archive Used in Tetlock (2010) Sample period is ,220,437 firm-day observations in total

20 20 The Sample

21 Aggregate Anomaly Variables 21 We construct 3 aggregate anomaly variables The variables are the sum of the number of stock i s anomaly portfolio memberships in month t Long, Short, and Net Net = Long - Short

22 Aggregate Anomaly Variables 22 Variable Mean Std. Dev. Min Max Long Short Net

23 23 The Main Specification

24 24 Main Specification

25 Economic Magnitudes 25 Net = 10 Daily Basis Points Annualized Buy and Hold Return No Earnings Day % Earnings Day %

26 26 Long and Short Separately

27 Economic Magnitudes 27 Long = 10 Daily Basis Points Annualized Buy and Hold Return No Earnings Day % Earnings Day % Short = 10 Daily Basis Points Annualized Buy and Hold Return No Earnings Day % Earnings Day %

28 Robustness 28 Are the results related to a day of the week effect (Birru, 2016)? Controlling for day-of-week does not alter our findings Macroeconomic news (Savor and Wilson, 2016)? Perhaps firm-specific news reflects systematic risk? No, anomalies do worse on macro announcement days Endogeneity of news? Stock return volatility causes news? We control for daily volatility and nothing changes

29 Anomaly Types 29 The effects are robust across anomaly types 1. Event Corporate events, changes in performance, downgrades 2. Fundamental constructed only with accounting data 3. Market Constructed only with market data and no accounting data 4. Valuation Ratios of market values to fundamentals

30 Analyst Forecast Errors 30 Biased expectations suggests biases in analysts earnings forecasts, risk does not Forecasts should be too low for stocks on the long side of the anomaly portfolios. Forecasts should be too high for stocks on the short side of the predictor portfolios.

31 31 Analysts Forecast Error

32 Data Mining Tests 32 A spurious anomaly would likely perform better in-sample on earnings days and news days Stocks with high (low) monthly returns likely had good (bad) news during the month Do anomaly strategies still have high returns on news and earnings days after controlling for this?

33 33 Data Mining Tests

34 34 Data Mining Tests Analyst Forecast Errors

35 Conclusions 35 Evidence of cross-sectional return-predictability goes back at least 43 years to Blume and Husick (1973) still disagreement over why In this paper we provide evidence that the cross-section of stock returns is best explained by a cross-section of biased expectations. Anomaly returns 9x on info days Anomaly signal predicts analyst forecast errors Difficult to explain the results with risk Harder to rule out data mining, but it does not seem to explain the full effects