How Does Hedge Fund Activism Reshape Corporate Innovation?

Transcription

1 How Does Hedge Fund Activism Reshape Corporate Innovation? Alon Brav, Duke University Wei Jiang, Columbia University Song Ma, Yale University Xuan Tian, Indiana University December 2016

2 What the existing literature has educated us about hedge fund activism Objectives and tactics. Type of firms targeted. Liquidity and block formation. Short and long-term market reaction. Impact on other firm policies (payout, leverage, governance). Impact on other stakeholders (creditors, labor, etc.). Impact on long-run performance. Impact of the threat of hedge fund activism. More and more on causality. 2

3 Why is innovation important? Innovation is the ultimate metric for investing for the long term. Larry Fink: Delivering Long-Term Value - Letter to Corporates, March 31, in response to the acute pressure, growing with every quarter, for companies to meet short-term financial goals at the expense of building long-term value..., more and more corporate leaders have responded with actions that can deliver immediate returns to shareholders, such as buybacks or dividend increases, while underinvesting in innovation, skilled workforces or essential capital expenditures necessary to sustain long-term growth. Innovation is necessarily affected as a by-product of new business strategies advocated by activists. Innovation has been the main theme in activist targeting increasingly more often: Trian vs. DuPont, Starboard Value vs. AOL, Icahn vs. Motorola, Third Point vs. Amgen. 3

4 Goal of research Change in inputs to and outputs from innovation at target firms. Scope of new internal innovation, such as patenting areas. Patent portfolio adjustment through transactions. Inventor redeployment and inventor-level productivity. Causality. 4

5 Data: Hedge fund activism Section 13(d) of the 1934 Securities Exchange Act requires investors who are beneficial owners of over 5% of any class of publicly traded securities, and who have an intention to influence corporate control, to disclose their ownership and intent within 10 days of crossing the 5% threshold. Information on the identity of the filer, filing date, ownership and its changes, cost of purchase, and the purpose of the investment. Filter out non-hedge funds, and exclude non-activist events Gather information on the hedge fund s motive, the target s response, and the development Supplement the sample with extensive news search for events with < 5% activist stakes. For each year t, classify innovative firms as those with at least one patent filing or positive R&D during years [t-3, t-1]. Event firms are targeted firms; control firms is a match from same year, same 2-digit SIC industry using propensity score incorporating firm size, market-to-book, ROA, etc., and same attrition through t+5. 5

6 Targeted firms over time 6

7 Data: Innovation NBER and USPTO: standard patent information. Obtain annual patent-level information from 1991 to Patent assignee, patent s application and grant year, the number of citations received by the patent, the technology class of the patent. Google Patent: patent assignments and reassignments from 1991 to Name of the patent buyers (assignees), the name of the patent sellers (assignors), unique patent identifiers (patent numbers), and the patents transaction dates. HBS Inventor Database: inventor-level information. Inventor-level information from 1991 to Provides the names of the inventors and their affiliation with the assignees. 7

8 Summary statistics Propensity Score-Matched Targets (N=553) Non-Targets (N=553) Difference Mean Median Mean Median t-statistic Ln(Firm Assets) Ln(MV) (-0.88) Firm Assets MV (-0.08) Firm ROA (-0.88) Firm R&D/Assets Leverage ** Firm Market-to-Book Ratio (-1.39) Ln(1+New Patents) (-0.49) Ln(1+Ave.Citation) (-0.03) Number of New Patents (-0.73) Ave. Citation of New Patents Firm Patent Originality (-0.26) Firm Patent Generality (-0.35) Firm Patenting Explorative (-0.38) Firm Patenting Exploitative (-0.22) 8

9 Main specification Difference-in-difference with industry/firm and year fixed effects: Innovation = b I( Target ) I( Post ) + b I( Post ) + gcontrol + a + a + e i, t 1 i i, t 2 i, t i, t t i i, t I(Post i,t ) = 1 if the firm-year (i, t) observation is within [t + 1, t + 5] years of an activism event or a pseudo-event year (for the control firms). I(Target i ) = 1 if firm i was ever targeted by a hedge fund activist. 9

10 Metrics for innovation Input: R&D, $ amount of as % of assets. Output: Quantity: The number of patent applications in a given year. Quality: Number of subsequent citations; or the right-tail of the distribution of subsequent citations. Patent originality : One minus the Herfindahl index of the two-digit technology class distribution of all the patents it cites. Patent generality : One minus the Herfindahl Index of all the patents that cite it. Explorative and exploitative: Whether the patent cites mostly existing knowledge. Portfolio composition: Diversity: One minus the Herfindahl index of patents filed in the past. The technological distance of a given patent to the firm s portfolio. 10

11 Input and output of Innovation: Basic results R&D/Assets R&D Log Log (%) ($ mil) (1+# New Patents) (1+Ave.Citations) Originality Generality I(Target) I(Post) *** 0.151*** 0.155*** 0.027*** (-1.323) (-3.086) (3.71) (3.07) (2.82) (1.11) I(Post) * *** (0.43) (1.04) (-1.935) (0.18) (-3.973) (-0.279) ln(mv) *** 5.361*** 0.047*** 0.048*** 0.012*** 0.009*** ( ) (4.06) (4.08) (3.31) (3.48) (2.96) ln(age) * (0.11) (-0.677) (-0.747) (-1.506) (-1.888) (0.72) Constant 8.872*** * (7.35) (0.22) (-0.029) (1.06) (1.78) (0.27) Observations 9,817 9,817 9,817 9,817 3,218 2,763 R-squared Year FE Yes Yes Yes Yes Yes Yes Firm FE Yes Yes Yes Yes Yes Yes 11

12 Dynamics of R&D/Assets 12

13 Dynamics of patenting Average Citations New Patents 13

14 More improvement at firms with initially more diverse innovation portfolios ln(1+# New Patents) ln(1+ave.cit) High Low High Low Diversity Diversity F-Test Diversity Diversity F-Test I(Target) I(Post) 0.232*** ** 0.218*** (4.82) (1.20) (0.02) (3.56) (1.63) (0.16) I(Post) ** (-2.152) (-0.828) (-0.177) (0.35) ln(mv) 0.047*** 0.048*** (4.77) (3.73) ln(age) (-0.464) (-1.40) Observations 9,817 9,817 R-squared Year FE Y Y Firm FE Y Y 14

15 More improvement in key tech class Key Technology Class Non-key Technology Class ln(1+# New Patents) ln(1+ave.citation) ln(1+# New Patents) ln(1+ave.citation) I(Target) I(Post) 0.194*** 0.182*** (4.47) (3.44) (-0.525) (0.50) I(Post) (-0.756) (-0.726) (-0.455) (-0.747) ln(mv) 0.053*** 0.038** 0.046*** 0.039** (6.01) (2.34) (3.44) (2.38) ln(age) ** 0.117** ** (-0.211) (-2.218) (2.20) (-2.215) Constant (-0.897) (0.86) (-0.826) (0.86) Observations 9,817 9,817 9,817 9,817 R-squared Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes 15

16 More improvement in explorative area Explorative Key Tech Class Exploitative I(Target) I(Post) 0.040*** (2.67) (-0.751) I(Post) (-0.829) (0.60) ln(mv) 0.009* (1.94) (-0.540) ln(age) ** (-1.203) (-2.060) Constant *** (1.24) (6.16) Observations 3,218 3,218 R-squared Year FE Yes Yes Firm FE Yes Yes 16

17 Mechanisms of changes Resource reallocation: Patent transactions. Investor turnover. Board composition. Selectivity of the transactions. Innovation efficiency subsequent to the reallocation. 17

18 Patents: More selling than buying %Patent Sales/Patents Owned %Patent Purchased/Patents Owned I(Target) I(Post) 0.641** 0.691** (2.17) (2.43) (0.09) (0.63) I(Target) (-1.257) (1.07) I(Post) (0.97) (-0.837) (1.27) (-0.330) ln(mv) *** (0.16) (-0.218) (4.29) (0.62) ln(age) *** *** (-2.711) (1.42) (-3.806) (-1.495) Constant 0.713* (1.91) (-0.743) (1.28) (1.36) Observations 9,374 9,374 9,374 9,374 R-squared Year FE Yes Yes Yes Yes Industry FE Yes No Yes No Firm FE No Yes No Yes 18

19 Selective selling of distant patents Dependent variable Patent selling Distance (Patent to Firm) 0.470*** 0.529*** (7.99) (8.50) Distance After 0.132** 0.283*** (2.25) (4.72) Distance Before *** (-1.601) (-4.444) After 0.443*** 1.082*** Before *** ** (-5.735) (-2.323) Observations 929, ,613 R-squared Year FE Yes Yes Patent Age FE Yes Yes Tech Class FE Yes Yes Firm FE No Yes Measure the distance between a patent and a firm s technology stock following Akcigit, Celik and Greenwood (2013). Based on common citations between two patents. Average over the distance between a given patent and all other patents held by the same firm. 19

20 Citations of patents sold post intervention Patent Sales by Targets of Hedge Fund Activists Patent Sales by Matched Firms + 3 å [ ] Citation = b d t + k + g Control + a + a + e j, t k j, t j, t j t i, t k=- 3 Event year t: The year of patent sale (within two years post intervention). 20

21 Inventor mobility ln(1+leavers) ln(1+leavers) ln(1+new hires) ln(1+new hires) I(Target) I(Post) 0.067* 0.062* 0.081*** 0.086*** (1.831) (1.664) (2.925) (3.184) I(Target) (0.889) (0.266) I(Post) *** ** (-1.365) (-0.812) (-2.791) (-2.399) ln(mv) 0.094*** 0.025*** 0.080*** 0.017*** (9.939) (2.613) (10.090) (2.674) ln(age) (0.943) (1.275) (0.200) (0.144) Constant *** (-2.914) (-0.743) (-1.327) (0.695) Observations 9,817 9,817 9,817 9,817 R-squared Year FE Yes Yes Yes Yes Industry FE Yes No Yes No Firm FE No Yes No Yes Leaver: An inventor who leaves & generates at least one patent prior to and post intervention. New hire: An investor newly hired by the firm during the year. Stayer: An investor who stays with the firm during the year. 21

22 Inventor productivity New Patents (Inventor-level) New Patent Citations (Inventor-level) Stayer Leaver New Hire Stayer Leaver New Hire I(Target) I(Post) 1.088*** 1.121* 0.763** 1.958*** 3.239* (8.096) (1.867) (2.418) (7.380) (1.881) (1.381) I(Target) (1.628) (0.975) (0.397) (-1.045) (-0.892) (-1.202) I(Post) (1.550) (0.998) (-0.673) (-0.643) (-0.729) (-0.651) ln(mv) 0.155** * *** (2.544) (1.258) (1.906) (-1.135) (-2.995) (-0.862) Observations 36,418 1,717 2,836 36,418 1,717 2,836 R-squared Year FE Yes Yes Yes Yes Yes Yes 22

23 Newly appointed board members with more expertise Targets Non-Targets Difference t-statistic Age ** Female Independence Expertise (Yes=1 or No=0) Innovation *** 3.01 Academic Financial *** 2.88 Management Operation * 1.92 Marketing Legal Academic: professor, dean, faculty, etc.2 Accounting: account, actuary, CFO, CPA, tax, treasurer, etc. Financial: Banking, asset management, merger, buyout, etc. Innovation: technology, research, r&d, engineer, product development, software, science, patent, etc. Legal: attorney, compliance, lawyer, counsel, JD, etc. Management: supervisor, management, head, president, CEO, chair, executive, etc. Marketing: sales, marketing, etc. Operation: operations, COO, human resources, manufacturing, etc. 23

24 Causality I: Focus only on hostile events R&D/Assets (%) R&D Expenses ($ mil) ln(1+# New Patents) ln(1+ave. Citations) I(Target) I(Post) * 0.148* 0.135* (-1.072) (-1.930) (1.686) (1.718) I(Post) (1.345) (0.117) (-0.692) (0.412) ln(mv) *** 6.680** 0.077*** 0.086*** (-5.598) (2.523) (3.080) (3.201) ln(age) *** (-0.357) (-2.994) (0.593) (0.447) Constant 7.554*** *** * * (7.668) (3.286) (-1.779) (-1.675) Observations 2,143 2,143 2,143 2,143 R-squared Year FE Yes Yes Yes Yes Firm FE Yes Yes Yes Yes 24

25 Causality II: Stock picking? (1) (2) (3) R&D/Assets (%) ln(1+# New Patents) ln(1+ave.citation) I(13G to 13D) * 0.174** (-0.215) (1.946) (1.968) I(Post) (0.064) (-0.713) (-0.304) Controls Yes Yes Yes Observations 6,756 6,756 6,756 R-squared Year FE Yes Yes Yes Hedge Fund FE Yes Yes Yes Firm FE Yes Yes Yes 25

26 Causality III: Market reaction to patent approval Abnormal Return (in bps) Granting window (months relative to intervention) [t-6,t+6] [t-3,t+3] [t-6,t+6] [t-3,t+3] I(Target) x I(Post) ** * (2.489) (1.712) I(13G to 13D) ** (2.353) (1.253) I(Post) (-0.222) (0.876) (-0.331) (-0.571) Observations 4,885 2,527 3,338 2,384 R-squared Monthly fixed effects of application-approval lag Yes Yes Yes Yes Firm FE Yes Yes Yes Yes 26

27 Conclusion The relation between hedge fund activism and innovation is nuanced. Overall, increase in productivity/efficiency, and refocusing in the scope of innovation. Resource reallocation (patents and personnel) plays an important role. Support for causal effects in addition to stock picking. Short-term investors are not necessarily associated with shorttermism. 27