Long Run Stock Returns Following R

1
Long Run Stock Returns Following RD Increases-
A Test of RD Spillover Effect
By Sheng-Syan Chen, Wei-Ju Huang and Yanzhi Wang

• Yanzhi Wang
• Yuan Ze University
• Taiwan

2
RD and Firm Valuation

• What is research and development (RD)?
• The investments on innovations and patents.
• In accounting treatment, RD is expensed.
• Related to intangible assets
• RD has the externality (spillover effect)
• ASKEY (??) obtained the technologies via patents
  from ATEONIX NETWORKS in 2004.
• These patents are about
• Method and system for providing remote storage
  for an internet appliance
• Method and system for providing a modulized
  server on board
• ASUS (??) obtained the technologies via patents
  from ASKEy via acquisition in 2008. Then, ASUS
  sued IBM.

3
RD and Stock Return

• RD level is positively related to future stock
  return (Lev and Sougiannis, 1996 Chan,
  Lakonishok, and Sougiannis, 2001).
• Given that RD generates intangible assets which
  is difficult to be valued, the RD is normally
  underreacted. RD level affects future long term
  stock return.
• RD increase is also positively related to future
  stock return (Chan, Martin and Kensinger, 1990
  Eberhart, Maxwell and Siddique 2004)
• RD increase indicates the improvement on
  profitability that investors slowly react to.
• RD outlay reduces the current earnings while RD
  is beneficial to future profitability. Investors
  extrapolate RD increase firms future earnings
  too low, and this biased estimation causes
  positive return.

4
Ebarhart, Maxwell and Siddique (2004)

• The firms with RD intensity over 5 and RD
  increase over 5 are investigated. The RD
  information is obtained from the annual report.
• The five-year long run abnormal return is about
  0.74 and 0.53 per month based on equal- and
  value-weighted Carhart (1997) four-factor model,
  respectively.
• The operating performance improves in consequent
  of the RD increase.

5
Some Comparisons
Event Article Methods Abnormal Return per Month
IPO Ritter and Welch (2002) Three-factor EW -0.21 (t -1.23)
SEO Loughran and Ritter (2000) Three-factor EW Three-factor VW -0.47 (t -5.42) -0.32 (t -3.00)
Repurchase Chan, Ikenberry and Lee (2007) Four-factor EW Four-factor VW 0.28 (t 4.24) 0.25 (t 3.35)
MA Mitchell and Stafford (2000) Three-factor EW Three-factor VW -0.20 (t -3.70) -0.03 (t -0.48)
RD increase Eberhart, Maxwell and Siddique (2004) Four-factor EW Four-factor VW 0.74 (P-value0.001) 0.53 (P-value0.001)
6
The Issue

• Why the RD increase that is a non-timing event
  experiences abnormal return so high?
• There could be other factor affecting the long
  run abnormal return of the RD increase.
• Economics literature has widely discussed the
  spillover effect of the RD for past decades
  (Arrow, 1962 Griliches, 1979 Bernstein and
  Nadiri, 1988 Hanel and St-Pierre, 2002 Agarwal,
  Echambadi, Franco and Sarkar, 2004 Hunt, 2006),
  but few papers mention this in finance
  literature. This could be the factor.

7
RD Spillover Effect

• RD spillover describes the fact that privately
  owned firm does not (or cannot) appropriate the
  outcome of its RD investment.
• Due to the industrial competition, the rival
  firms may follow to increase the RD after the
  EMS sample firm increases RD.

8
Hypothesis

• Eberhart, Maxwell and Siddique (2004) find
  significantly high abnormal return for RD
  increase firms.
• We hypothesize that this result is related to the
  RD spillover effect.
• For the Eberhart, Maxwell and Siddique (2004)
  sample firm with more RD followers, then the
  abnormal return of the sample firm should be
  higher.

9
Sample Collection

• Our sample is collected from U.S companies listed
  on the NYSE/Amex/Nasdaq during January 1974 to
  December 2006.
• We start our RD sample collection from 1974
  because the requirement of reporting RD became
  effective from 1974.
• We mainly follow EMS and set up five criterions
  for our sample that includes firm-year
  observations with significant increases in RD
  (i) the ratio of RD expenditures divided by
  sales over 5, (ii) the ratio of RD expenditures
  divided by average total assets over 5, (iii)
  the change of the ratio of RD expenditures
  divided by sales over 5, (iv) the ratio of
  change of RD expenditures divided by average
  total assets over 5, and (v) the ratio of change
  of RD expenditure and than divided by RD
  expenditures over 5.
• As a result, the sample with significant increase
  in RD expenditures includes 10,280 U.S firm-year
  observations.

10
Methodology- Return

• From July 1976 to December 2006, each RD
  increase portfolio p is formed by including
  sample firms which were increase firm within past
  60 months. For example, the RD increase
  calendar-time portfolio is composed of firms
  classified as the RD increase firm in any of the
  past 5 years. As a result, the RD increase
  portfolio monthly returns are regressed on
  Carhart (1997) four-factors as follows
• Coefficient tests are adjusted with Newey-West
  autocorrelation-heteroskedasicity estimation.

Holding period
Compute ?RD
?RD is publicly available
Year t-1/Dec
Year t /Dec
Year t1 /July
Year t6/June end
11
Methodology- Operating Performance

• Using EBITDA/Assets as ROA and as well as the
  measure of operating performance (Barber, Lyon,
  1996)
• We look at the median of changes of ROAs.
• We compute abnormal operating performance by
  matching firm approach.
• Minimize OPt-OPt around 80120 of OPt with
  the same 2-digit SIC code.
• Minimize OPt-OPt around 80120 of OPt with
  the same 1-digit SIC code if its not found at
  step 1.
• Minimize OPt-OPt around 80120 of OPt without
  industry requirement if its still not found at
  step 2.
• Minimize OPt-OPt without industry and filter
  requirement for all remaining sample.

12
Summary Statistics

• RD increase firms are small growth firms
• RD increase firms are high RD firms
• About 30 of RD increasing firms are followed by
  their industry peers.
• RD increasing firms cluster in two industries
  manufacturing and pharmaceutical.
• Table 1
  

13
Abnormal Return of RD Increase

• RD followed ratio is the percentage of rival
  firms that follow the EMS sample firms to
  increase RD over at least 1
• High RD followed ratio implies higher RD
  spillover effects.
• Get a closed look at result

14
Some Robust Checks

• Fama and French (1993) three-factor model
• Control the time-varying risk betas in factor
  model
• Control the delisting return in factor model
• Remove the repeating RD increase events
• Change the definitions of the RD increase
  follower
• All these approaches appear consistent result.

15
Operating Performance

• We use Fama and French (2000) earnings regression

Abnormal ROAt5 -Abnormal ROAtß0ß1RD followed
ranktß2RD followed ratiot (?1?2NDFEDt?3NDFEDt
DFEt?4PDFEDtDFEt)DFEt
(?1?2NCEDt?3NCEDtCEt?4PCEDtCEt)CEtet
16
RD Mimicking

• Rival firms engage in RD mimicking to undo the
  negative effect of the sample firms RD
  investment.
• In a concentrated industry, the strategic
  reactions are more active, thus the RD
  increasing benefit could be offset by rivals
  following RD increases.
• The RD increasing firms earn lower return in a
  more concentrated industry.

17
Stock Return and Industry Concentration

• Firms in high-concentration industry earn lower
  return
• This confirms the mimicking hypothesis.

18
Fama and MacBeth Regression

• For monthly stock returns during July at year t1
  to June year t2, we include the monthly stocks
  with significant RD increases in any of past
  five years (t to t4) in the regression model. We
  regress the monthly returns on independent
  variables including the RD-followed ratio.
• The Fama-MacBeth estimates are obtained by the
  time-series average and tested by time-series
  volatility.

19
Fama-Macbeth (1973) Regression
20
Industry RD Growth as Spillover Proxy

• The RD spillover describes the impact of rivals
  RD inputs on sample firms outputs.
• So the aggregative industrial RD inputs
  (excluding sample firm) could be an alternative.
• We use the industry-wide RD growth.
• Link to Table 9

21
Conclusion

• The long-term positive abnormal stock return
  following a firms RD increase is argued by
  Eberhart, Maxwell and Siddique (2004).
• In this paper, we turn to propose an economical
  hypothesis, the RD spillover effect, to account
  for the long-term stock return post to the RD
  increase.
• As a firm increases the RD investment, its
  industry peers may follow and increase their RD
  investments under a competitive industry. Given
  that RD investment has spillover effect, the
  followers RD investment is beneficial to the
  firm that has significantly invested in RD
  projects.
• Hence we argue and find that the firm with
  significant RD increases and with sufficient RD
  investment followers tends to outperform that
  with few RD followers.
• This economic explanation also helps to answer
  the puzzle why the RD increase, which is a
  non-timing event, is followed by significant
  abnormal stock return.