Bessembinder / Zhang (2013): Firm characteristics and long-run stock returns after corporate events. Discussion by Henrik Moser April 24, 2015

Transcription

1 Bessembinder / Zhang (2013): Firm characteristics and long-run stock returns after corporate events Discussion by Henrik Moser April 24, 2015

2 Motivation of the paper 3 Authors review the connection of corporate events and long-run stock returns Previous studies commonly used two approaches to measure long-run abnormal returns following corporate events Calendar time portfolio Buy-and-hold abnormal return (BHAR) However, BHARs often show large abnormal returns, while the calendar time method does not Authors try to reconcile both approaches Four types of corporate events IPOs SEOs merger biddings dividend initiations

3 Related literature 4 Figure 1: Selection of related studies.

4 Calendar time portfolio 5 Focuses on mean abnormal returns to portfolios of event firms (calculated using the Fama-French Three-factor model) R pt R ft = α i + β i (R mt R ft ) + s i SMB t + h i HML t + ɛ it Advantages Method eliminates the problem of cross-sectional dependence among sample firms because the returns on sample firms are aggregated into a single portfolio 1 Disadvantages Approach is misspecified in non-random samples Method is subject to rebalancing bias Approach has low power to detect abnormal returns, because it effectively weights each period equally, while corporate events tend to cluster in certain time periods 1 Lyon, John D., Brad M. Barber, and Chih-Ling Tsai (1999): Improved Methods for Tests of Long-Run Abnormal Stock Returns, The Journal of Finance 54 (1):

5 Buy-and-hold abnormal return (BHAR) 6 Based on the difference between buy-and-hold returns to event firms (e) as compared to matched firms (m) after event date t = 0 T T BHAR et = (1 + r et ) (1 + r mt ) t=1 t=1 [ T ] [ T ] = exp ln(1 + r et ) exp ln(1 + r mt ) t=1 t=1 Alternative (but equivalent) approach: Wealth relative [ T t=1 WR et = (1 r T et) T t=1 (1 r mt) = exp [ ] ] ln(1 + r et ) ln(1 + r mt ) t=1

6 Evaluation of BHARs 7 Advantages Free of rebalancing bias Resembles investors actual experience, as opposed to periodic rebalancing required by other methods Disadvantages Matching based on particular firm characteristics, such as size or book-to-market ratio Ignoring other firm characteristics, such as, e.g., liquidity, beta, momentum, and capital investment Matching is typically done at a particular point in time, while the matching criteria might actually move apart over longer time periods

7 Difference between sample firms and their matches Median beta 600 Median size 0.55 Median BM Beta Size 400 BM Month Month Month Bidder Match Bidder Match Bidder Match 0.2 Median momentum 0.38 Median idiosyncratic volatility.06 Median illiquidity Momentum Idiosyncratic volatility Illiquidity Month Month Month Bidder Match Bidder Match Bidder Match Figure 2: Characteristics of bidding firms and their size- and book-to-market matched comparable firms.

8 Sample selection 9 U.S. public companies from Thomson Financial s SDC database (IPO, SEO, and M&A samples) or CRSP (dividend sample) Time span from 1980 to 2005, allowing for a five-year period to measure post-event stock returns Excluding certain categories of events (e.g., issue of depository receipts, minority and non-control mergers) Special filters of the merger sample Transaction size (in terms of market value of the bidder) greater than 5% Transaction value greater than $5m

9 Matching samples 10 SEO, M&A, and dividend samples Matched firm is the company with the closest book-to-market ratio among firms with market capitalization between 70% and 130% at the end of year t 1 Match must not be in original sample during then ten years around the event IPO sample Matched firm is the company with the closest but greater market capitalization at the end of December following the IPO Match must have been publicly traded for more than five years

10 Sample 11 Table 2 Number of event rms. This table presents the number of mergers and acquisitions(m&as), seasoned equity offerings (SEOs), initial public offerings(ipos), and dividend initiations in our sample,by year. Year M&A SEO IPO Dividend Total 3,972 5,131 8, Figure 3: Sample sizes by year.

11 Variables 12 Market beta Estimated using the market model with monthly stock returns during years t 5 to t + 5 Firm size Market capitalization Book-to-market ratio Ratio of the book value to the market value of common equity Momentum Cumulative return over months 12 to 2

12 Variables (cont d) 13 Illiquidity Average ratio of daily absolute stock return to dollar trading volume, relative to market average illiquidity during the same period Idiosyncratic volatility Annualized standard deviation of the residuals of a Fama/French three factor regression implemented in daily returns in month 2 Investment Annual change in gross PPE plus annual change in inventory, divided by assets

13 Regression approach 14 ln(1 + r et ) ln(1 + r mt ) = α β 1 Beta et + β 2 Size et + β 3 BM et + β 4 Mom et + β 5 Illiquidity et + β 6 IdioVol et + β 7 Investment et + ɛ et where denotes the normalized difference in the associated firm characteristics across the event firm and the matching firm (in order to make coefficients comparable across characteristics)

14 General idea 15 In any regression, the intercept measures the mean of the dependent variable, conditional on outcomes of zero for each independent variable In this particular case, the intercept estimates the mean abnormal log return to event firms, conditional on no difference in firm characteristics across event and control firms Testing the hypothesis that the intercept is zero is equivalent to testing whether the BHAR is zero

15 Regression output: SEO sample 16 Figure 4: Regression outputs for the SEO sample.

16 Regression output: IPO sample 17 Figure 5: Regression outputs for the IPO sample.

17 Comparison to calendar time portfolio 18 Figure 6: Regression outputs for the calendar time portfolio method. Outcomes of the calendar time portfolio method show no abnormal returns (α insignificant) No conflict of BHARs and calendar time portfolios found (in this sample)

18 Concluding remarks 19 Many previous studies have found significant abnormal returns The abnormal returns are attributed to the events themselves There might be a different explanation of the abnormal returns rooted in known return regularities The authors show that typical matching algorithms are imperfect in addressing a number of company characteristics Variation in the characteristics completely explains abnormal returns Results reconcile confounding results of BHAR and calendar time portfolio studies Apparently abnormal returns reflect characteristics of the firms rather than event-specific effects