How Do Firm

1
How Do Firms Increases in RD Affect Long-Run
Performance of Intra-Industry Competitors?
Weifeng Hung Feng Chia Universty Sheng-Syan
Chen National Taiwan University Yanzhi Wang Yuan
Ze University
2
RD investment is a favorable strategy

• When a firm increases its RD outlay, the firm
  earns positive abnormal return both in the short
  run (Chan, Martin and Kenisnger, 1990 JFE
  Szewczyk, Tsetsekos and Zantout, 1996 FM) and in
  the long run (Chan, Lakonishok, and Sougiannis,
  2001 JF Eberhart, Maxwell and Siddique, 2004 JF
  2008 JAR).

3
RD spillover effect

• According to Bernstein and Nadiri (1989),
• A feature of RD investment that distinguishes
  it from other forms of investment is that firms
  which do the investing are often not able to
  exclude others from freely obtaining the benefits
  from the RD projects.

4
Short-run results are not consistent with the RD
spillover hypothesis

• Zantout and Tsetsekos (1994) document that the
  rivals of firms that make announcements of
  increases in RD expenditures suffer a
  statistically significant negative abnormal
  return.
• Sundaram, John and John (1996) find that the
  market reaction to competitors varies depending
  on their competitive strategy measure.

5
Motivations

• The long-term impact of RD increase on
  intra-industry competitors remains unknown.
  Particularly, no study explores the long-run
  market reactions to spillover effect.
• Lev and Sougiannis (1996) and Chan, Lakonishok
  and Sougiannis (2001) suggest that because the
  RD valuation is hardly realized and not easily
  evaluated in a short horizon, long-term study is
  more adequate to capture the intangible
  information of the RD investment.
• Managers seldom announce RD increases formally.
  There might be a large time elapses between
  firms investment and markets perception. As a
  result, the market might take time to fully
  reflect managers investment decision.

6
Motivations

• If the market underreacts to the direct future
  benefits of the RD increases, it might also
  underreact to any indirect future benefits, if
  any, which a firms rivals might gain from that
  firms RD increase.
• Fama (1998) argues that the abnormal returns
  might reflect normal random variations that occur
  in efficient markets, the long-term results can
  be viewed as an important challenge to the
  efficient market hypothesis.

7
The benefits of RD spillover effect

• Bernstein and Nadiri (1988) have indicated that
  the RD investment by a firm reduces its own
  production cost and, as a result of spillovers,
  costs of other firms are also reduced.
• If spillovers do lower rivals production cost,
  then we would expect this effect to show up in
  the operating performance of rivals.
• We use changes in operating performance and
  analyst forecast revisions to proxy for
  improvements in profitability.

8
Spillover Hypothesis

• Firms undertaking RD investment are often not
  able to appropriate the RD benefits that is,
  the benefits from RD investment may extend to
  other firms in the same industry and/or economy.
• It is possible for rival firms to abstain from
  RD investment, and yet to take advantage of the
  knowledge generated by a firm that does invest in
  RD. (Jeffe, 1986 Bernstein, 1989 Bernstein and
  Nadiri, 1988 Goto and Suzuki, 1989 Nadiri and
  Kim, 1996 Srinivasan, 1995).
• Intra-industry rivals earn positive long-term
  abnormal return and experience improvements in
  operating performance and analysts forecast
  revisions

9
Strategic Reaction Hypothesis

• A firm that increases its RD spending might gain
  unfriendly attention from its rivals.
• RD increasing behaviors might be taken as that
  firm is moving ahead in the race to be the first
  to innovate to exploit the future benefits.
• Sundaram, John, and John (1996) suggest that
  firms adopt RD announcement as for a means of
  strategic interaction.
• Massa, Rehman, and Vermaelen (2007) suggest that
  a repurchase firm conveys a valuable signal about
  firm undervaluation, which threatens competitors.
  To undo this effect, the rival may mimic and
  repurchase shares.
• Massa, Rehman, and Vermaelen (2007) suggest that
  most strategic reacting behavior occurs in
  concentrated markets.
• Do managers use a non-announcement channel, such
  as RD investment racing, to strategically react
  to their rivals?

10
Sample selection

• Source
• The sample includes listed stocks in
  NYSE/AMEX/NASDAQ during the period 1974 to 2006.
  Data on stock price and number of shares
  outstanding to compute market value of equity are
  obtained from the CRSP database.
• Sample selection criterion
• (1) RD intensity (measured by the ratio of
  RD-to-assets (RDA, data46/data6) and
  RD-to-sales (RDS, data46/data12) of at least 5,
  
• (2) increases in dollar RD by at least 5 (RD
  growth rate, or RDI),
• (3) increases in ratio of RD-to-assets (RDAI) by
  at least 5.

11
Sample selection criterion

• Further, we exclude the non-common stock ADRs,
  SBIs, unit trusts, closed-end funds, REITs, and
  financial firms, as the work done by Fama and
  French (1992, 1993).
• Sample stocks are also excluded if they have the
  following conditions
• (1) non-positive book equity, (2) without sales,
  operating income before depreciation (data13),
  earnings before interest and taxes (data178),
  total assets, or market value (3) without
  industry concentration measures (4) a firm have
  not appeared in COMPUSTAT for more than two years
  (Banz and Breen, 1986).
• The final sample consists of 10,452 firm-year
  observations, in which the sample includes 3,646
  RD increasing firms.

12
Definition of industry rivals

• Throughout the paper, we use CRSP four-digit SIC
  classification to define industry membership.
• We measure the industry concentration using
  Herfindahl-Hirschman Index (HHI).
• HHI is a commonly accepted as the measure of the
  product market concentration. HHI is the sum of
  squared market share of each firm in the
  industry.
• Each year, we classy three groups based on HHI,
  where low concentration portfolio corresponds to
  the 30 of industries with the lowest
  concentration, while high concentration portfolio
  corresponds to the 30 of industries with the
  highest concentration.

13
Definition of industry rivals cont.

• For each sample firm, we construct its
  corresponding industry portfolio as all stocks,
  except the sample firm itself, in the same
  four-digit SIC industry as the sample stock.
• The returns on industry portfolio are equally and
  value weighted.
• That is, if we have 10,452 firm-year
  observations, then 10,452 industry portfolios
  will be obtained.

14
Methodologies

• Calendar time abnormal returns
• For each calendar month t in our sample period,
  we form a portfolio of all sample firms that have
  significantly increased their RD investment in
  the previous five years (60 months).
• We then run the Fama and French three-factor
  model and Carhart four-factor model for long-term
  abnormal stock returns shown in the following
  equation
• Both equally- and value-weighted portfolio
  returns are calculated.

15

• Rolling-over method
• A firms risk may change in response to its RD
  change (Berk, Green, and Naik, 1999 Chan,
  Lakonishok, and Sougiannis, 2001)
• We use the first 60 monthly returns (e.g., from
  April 1975 to March 1980) of the portfolio to
  estimate its factor loadings, and calculate the
  expected portfolio return in month 61 (e.g.,
  April 1980) based on these factor loadings
  estimated over the previous 60 months multiplied
  by their corresponding factor returns in month
  61.
• The abnormal return in month 61 is the difference
  between the actual portfolio return and expected
  portfolio return.

16

• Delisted-adjusted returns
• To mitigate survival-ship bias in returns for
  firms delisted from CRSP for performance reasons,
  we follow the procedure of Shumway (1997) and
  Shumway and Warther (1999).
• Specifically, for firms delisted for performance
  reasons, we substitute -30 as the last return
  for NYSE and Amex stocks and -55 for Nsadaq
  firms.
• Cumulative benchmark adjusted returns
• Our procedure for calculating benchmark-adjusted
  returns follows the methodology outlined in the
  Daniel, Grinblatt, Titman, and Wermers (1997, JF)
  study that developed benchmarks to evaluate
  mutual fund performance.
• Specifically, we form 25 benchmark portfolios
  that capture three stock characteristics namely
  book-to-market equity and size which are
  significantly related to the cross-sectional
  variation in returns.

17

• Each stock, in each year, is assigned to a
  benchmark portfolio according to its rank based
  on SZ and BM. Excess monthly returns of a
  particular stock are then calculated by
  subtracting the stocks corresponding benchmark
  portfolios returns from the stocks returns.
  Specifically, the characteristics-adjusted return
  is defined as
• where and are the return on security i
  and the return on a SZ-BM-matched portfolio in
  month t, respectively.
• Each month, we use characteristics-adjusted
  return to calculate portfolios abnormal returns,
  then the abnormal monthly returns after formation
  period are cumulated as cumulative abnormal
  returns.

18

• RATS approach
• Sock excess returns are regressed on the Carhart
  (1997) four-factor for each month in event time,
  and the estimated intercept represents the
  monthly average abnormal return for each event
  month.
• The long-run abnormal returns between 1 month and
  60 months (j) after a large increase in RD at a
  sample firm are adopted.
• The following regression is run each event month
  j
• ri,t are the equally- and value-weighted
  portfolio returns on industry portfolios in
  calendar month t that corresponds to the event
  month j, with j 0 being the month of the
  beginning of the fourth month following fiscal
  year-end in which there is a large increase in
  RD at a sample firm.

19
Summary statistics

• The statistics reported in Table 1 are very
  similar to the those reported in EMS.
• The average (median) HHI is 0.245 (0.176),
  suggesting that the most of sample firms are
  within less concentrated industries.
• The average (median) number of rival firms in
  each industry portfolio is around 91 (58).

20
Table 1 Summary statistics
21
Spillover effect

• Consistent with EMS, Panel A of Table 2 shows
  that both equally and value-weighted long-run
  abnormal returns on sample stocks are
  significantly positive. The abnormal returns are
  0.86 and 0.34 for equally- and value-weighted
  method. The results are quantitatively similar to
  EMS.
• There are significantly positive abnormal returns
  for the rival portfolio.

22
Table 2 Long-Term Abnormal Return for Large
RD-Increase Firms and Rival Portfolios
23
The influence of strategic reaction

• Table 3 shows that the coefficient of the
  Concentration x RD increase wave term is 0.261
  (Model 3), which is significant at 1 confident
  level.
• This indicates that the higher the concentration
  of the industry and the higher total number of
  firms that largely increase RD expense over past
  five years in the industry, the more likely that
  the firms located in that industry will increase
  their RD expense.

24
Table 3. Probit Regression of Indicator for Large
Increases in RD
25
The influence of strategic reaction

• Table 4 shows significant positive abnormal
  returns for the rival portfolio in less
  concentrated industries.
• Instead, in the concentrated industries, the
  abnormal returns for the rival portfolios are not
  significant, and some rival portfolios even earn
  negative abnormal returns.

26
Table 4 Long-Term Abnormal Return for Rival
Portfolio Sorted by Industry Concentrations
27
The influence of strategic reaction

• Fig. 1 shows that the long-term return of the
  rival portfolio in low concentration industry
  experiences high return. In particular, the rival
  portfolio in high concentration industry earns
  negative BHARs.
• Table 5 demonstrates that over 12 (24, 36, 48,
  60) months, for the full sample, the cumulative
  equally-weighted average abnormal returns of
  10.05 (22.28, 34.93, 46.54, 58.50), are all
  significant at the 1 level. The results of the
  subsample indicate that for the low industry
  concentration group, the CARs are all significant
  at the 1 level.
• Therefore, these results further provide supports
  for the strategic reaction hypothesis

28
Figure 1 Cumulative Abnormal Return for Rival
Portfolios
29
Table 5. Long-Term Cumulative Abnormal for Rival
Portfolios
30
Cross-sectional regression analysis

• The dependent variable is 60-month buy-and-hold
  abnormal returns (BHAR) of each industry
  portfolio, in which the buy-and-hold abnormal
  return is controlled for the size, B/M matching
  portfolio return.

31
Further spillover evidence

• In Model 1 and 2, the results show that the BHAR
  of sample firm term is positive and highly
  significantly across all the models indicating
  that the higher the buy-and-hold abnormal returns
  to sample firms occur following the RD
  increases, the greater the buy-and-hold abnormal
  returns to rival portfolios will earn.
• The long-run abnormal returns of industry
  portfolio are also positively associated with the
  level of RD increases by largely RD-increase
  firm.
• This clearly suggests that the RD increases has
  spillover effect on rival firms, and is
  consistent with the spillover hypothesis.

32
Further strategic reaction evidence

• The coefficient estimate of Concentration is
  significantly negative. Thus, the higher the
  concentration of the industry, the lower the
  long-run abnormal returns to industry portfolios
  will be.
• The coefficients of the interacting terms are
  negative.

33
Table 6 Cross-Sectional Analysis of Long-Run
Abnormal Returns to Rival Portfolios
34
Changes in operating performance

• First, the operating performance of the rival
  portfolios deteriorates prior to the event year
  and increased subsequent to the event year.
• Second, the improvements in post-event operating
  performance are the higher for rivals with low
  concentration and lower for rivals with high
  concentration.

35
Figure 2 Changes in Return on Assets (ROA) of
Rival Portfolios
36
Figure 3 Changes in Profit Margins (PM) of Rival
Portfolios
37
Cross-sectional regression analysis

• The dependent variable is five-year average
  post-event changes in operating performance (ROA
  and PM) of each industry portfolio.
• First, for all models, the intercept indicates
  that industry portfolio experiences positive
  changes in ROA (PM) post to the RD increasing
  year.
• The long-run post-event changes in ROA of
  industry portfolio are positively associated with
  the level of RD increases by sample firm.
• Second, the coefficient estimate of Concentration
  is significantly negative.

38
Table 7 Cross-Sectional Analysis of Changes in
Long-run Operating Performance of Rivals
39
Analysts forecast revisions

• The dependent variable is the post-event 60-month
  average of abnormal analysts EPS forecast
  revisions of industry portfolios.
• The evidence indicates that the long-run averages
  of abnormal analysts forecast revisions of
  industry portfolio are positively associated with
  the level of RD increases by largely
  RD-increase firm.
• On the other hand, the coefficient estimate of
  Concentration is significantly negative.

40
Table 8 Cross-Sectional Analysis of Changes in
Analysts EPS Forecast Revisions of Rivals
41
Institutional Trading Surrounding Share
Repurchase Announcements (SRA)

• Weifeng Hung (???)
• Department of Finance, Feng Chia University,
  Taiwan

42
Agenda

• Motivations
• Contributions
• Data and methodologies
• Empirical Results
• Conclusions

43
Motivation SRA attracts institutions?

• Allen, Bernardo, and Welch (2000) argue that
  undervalued firms who want to signal their worth
  would like to attract institutions because
  institutions are better at assessing the firms
  true worth.
• Several studies indicate that SRA attracts
  institutions (Grinstein and Michaely, 2005
  Shleifer and Vishny, 1986 Allen, Bernardo, and
  Welch, 2000).
• On the other hand, unlike individual investors,
  institutions are expected to be less prone to
  attention-driven trading behavior (Barber and
  Odean, 2008).

44
Motivation Institutional response

• Institutional investors are expected to have
  ability to move prices directly through their own
  trading, as well as indirectly, by influencing
  the trading decisions of other market
  participants who may follow their actions.
• Institutional trading affects stock
  returns(Bannet et al., 2003 Gompers and Metrick,
  2001).
• Understanding of whether SRA attracts
  institutions is of great importance for firms
  announcing share repurchases.

45
Motivation Superior information?

• Institutions would be expected as sophisticated
  investors in processing information to arbitrage
  the repurchases anomaly to earn superior returns.
• Prior studies indicate that institutional
  investors are able to correctly identify
  corporate events, such as IPO and SEO.
• Why SRA?
• The buyback anomaly has persisted for 25 years in
  U.S. stock market (Peyer and Vermalen, 2008).

46
Motivation Superior information cont.

• SRA in Taiwan
• On average, firms announcing share buyback earn
  significantly positive abnormal returns.
• However, about 45 of events in Taiwan experience
  negative cumulative abnormal returns in the 30
  days following SRA.
• If institutional investors do have informational
  advantages in processing corporate activities, it
  is intuitively credible that individual investors
  can profit from the buy-sell information by
  imitating institutional trades surrounding the
  corporate announcements.

47
Motivation Unique datasets

• Since daily institutional trading data is not
  easily assessed, most empirical studies of
  institutional trading have focused on quarterly
  or annual data, such as 13(F) database.
• Few studies have explored the relationship
  between institutional daily trading behavior and
  SRA.
• Puckett and Yan (2010) suggest that trading
  performance shown by prior studies using
  quarterly data are biased downwards because of
  inability of publicly accessing interim trades.

48
Motivation Unique datasets, cont.

• We argue that the quarterly holdings data cannot
  capture the intra-quarter institutional trading,
  such as the exact timing of institutional trading
  surrounding the share repurchases announcements.
• Particularly, we show that institutional trading
  occurs very near to the SRA date, about 10 days
  before SRA and a month after.
• Daily institutional trading data in Taiwan allows
  us, for the first time, to contribute to the
  literature by examining the daily institutional
  trading behavior in response to SRA.

49
Contributions

• 1. SRA significantly attracts institutions,
  switching their trading behavior from net selling
  to net buying. This finding is consistent with
  the argument that SRA attracts institutions
  (Grinstein and Michaely, 2005 Shleifer and
  Vishny, 1986 Allen, Bernardo, and Welch, 2000).
• 2. There is an institutional price impact before
  and after SRA.
• 3. Institutional trading seems to have predictive
  ability for the post-SRA stock performance.
• 4. However, this trading skill disappears after
  controlling for their post-SRA price impact.

50
Data

• We obtain daily data from Taiwan Economic Journal
  (TEJ), including stock repurchases announcement
  events (for the interests of shareholders),
  market index returns (including dividends), and
  institutional trading volumes.
• Annual accounting data, such as book equity, are
  also retrieved from TEJ. This paper includes 610
  repurchasing samples from October 13, 2000
  through December 31, 2006.
• We exclude events without institutional trading,
  stock returns, market value, and accounting
  variables at announcement date. The stocks with
  less than 130 trading days prior to the share
  repurchase announcement are also dropped.

51
Institutional trade imbalance

• A positive (negative) institutional trade
  imbalance for a stock stems from institutional
  net buying (net selling) activities and increases
  (decreases) in institutional ownership for the
  stock.
• We use the mean institutional trade imbalance of
  period from day -130 through day -31 (relative to
  the initial announcement day 0) to estimate the
  expected institutional trade imbalance.
• The daily abnormal institutional trade imbalance
  is calculated as the difference between the
  actual trade imbalance and expected trade
  imbalance across stocks for each day.

52
Operating performance

• We define the unexpected change in performance as
  the change in performance of the repurchasing
  firm minus the change in performance of a
  matching firm.

53

• Empirical results

54
Favorable information

• The mean (median) CAR for the announcement period
  (-2,2) of 1.20 (1.18) is positive and
  significant at the 1 level.
• The positive mean (median) CAR of 2.96 (2.23)
  for the post-announcement (3,30) period
  indicates a significant reversal for firms
  announcing stock repurchases.

55
Table 1 Summary Statistics
56
S.R.A. attracts institutions

• SRA attracts institutions
• The SRA significantly affects institutional
  trading behavior, i.e., from net selling behavior
  to net buying behavior.
• Price impact
• There is a positive concurrent relationship
  between institutional trading and stock returns
  around SRA window.

57
Table 1 Summary Statistics
58
Table 2 Short-Run Price Reactions and
Institutional Reactions of Each Year
59
Table 3 Summary Statistics of Each Industry
60
(No Transcript)
61
Predictive ability

• Institutional trading has predictive ability for
  the stock performance following SRA.
• Specifically, it appears that institutional
  investors are able to identify stocks with good
  (underpriced) or bad (overpriced) SRA.

62
Table 5 Abnormal Institutional Trade Imbalances
and Price Behavior Surrounding S.R.A.
63
(No Transcript)
64
Institutional trading following S.R.A.

• Institutional investors are not feedback traders
  following SRA, i.e., their post-SRA trading
  behavior is not driven by prior returns.
• The decisions of institutional trading following
  share repurchase announcements seem to be
  consistent with the institutional herding
  hypothesis, i.e., they trade by following
  themselves or others trades.

65
(No Transcript)
66
(No Transcript)
67
Price impact

• Post-event institutional trading indeed impacts
  stock price.
• For the net sell and net buy groups, the
  CARs(3,30) are significantly negative and
  positive at the 1 level, respectively.
• However, the impact accounts for partial market
  reactions to SRA.
• The CAR(3,30) of the neutral group, 4.15, is
  significant at the 1 level.

68
Table 10 Portfolios based on CATI(3,30)
69
Short-run predictive ability V.S. price impact

• It thus seems natural to ask whether the
  short-run trading skill of institutional trading
  surrounding SRA is due to the price impact caused
  by their trading persistence.
• If institutional trading skill mainly results
  from the price impact caused by their persistent
  trading, we should see an insignificant
  cumulative abnormal return following SRA.
• For groups without persistent trading, their
  market reactions are not significantly different
  from zero, implying that the predictive ability
  of institutional trading surrounding SRA mainly
  results from their post-eve nt price impact.

70
Table 11 Independent Double Classifications based
on CATI(-2,2) and CATI(3,30)
71
(No Transcript)
72
Conclusions

• We use daily data to study institutions in
  response to buyback announcements.
• Buyback announcement attracts institutions.
  Institutions are net sellers before buyback
  announcement and net buyers after.
• Institutions trading skill is driven by their
  post-event price impact.
• The evidence does not support institutional
  informed trading.

73
Can Institutional and Individual Trading Drive
Value and Size Premiums in Japan?

• Weifeng Hung
• Associate Professor, Department of Finance, Feng
  Chia University, Taichung, Taiwan

74
Agenda

• Motivations
• Data descriptions and variable definitions
• Empirical results
• Conclusions

75
The possible explanations of value and size
premiums

• 1. Rational compensation for risks (Fama and
  French, 1993 1996)
• 2. Behavioral bias overreaction (Lakonishok,
  Shleifer, and Vishny, 1994)
• 3. Data snooping (Lo and MacKinlay, 1990)

76
Price impact by institutional trading

• Institutional trading has a dynamic relation with
  stock returns .
• Two possible effects
• 1. Destabilization If trading results from fads,
  reputational concerns, or preference for certain
  firm characteristics, such trading may drive
  asset prices away from fundamental values and
  create return reversals in the subsequent period.
  (DeLong et al., 1990 Choe et al., 1999 Wermers,
  1999 Sias, 2004)
• 2. Stabilization Institutional buying (selling)
  may stabilize the stock market when prices are
  undervalued (overvalued).
• Which one does institutional trading have?

77
Trading preference by institutional investor

• Frazzini and Lamont (2008) and Sharma, Hur, and
  Lee (2006) indicate that institutions tend to buy
  growth stocks and sell value stocks.
• However, seldom studies explicitly show evidence
  that trading preference by institutional investor
  drives or mitigates the value and size premiums.

78
Trading preference by individual investor

• Kaniel, Saar, and Titman (2008) examine NYSE
  trading data and find that individual investor
  tend to be contrarian traders in the short-run,
  i.e., they buy stocks after prices decrease and
  sell stocks after prices increases.
• However, there is less agreement about the
  long-run trading preference by individual
  investors.
• Particularly, the long-run relation between
  individual trading and future stock returns has
  received little attention.

79
Purposes

• 1. Do institutional (individual) investors buy
  (sell) growth stocks and sell (buy) value stock
  in Japan?
• 2. What is the dynamic relation between
  institutional (individual) trading and stock
  returns?
• 3. Does institutional trading and/or individual
  trading drive value and size premiums?
• 4. Can the strategy based on trading preferences
  by institutional investors or individual
  investors enhance value and/or size strategy?

80
Data

• From Pacific Basin Capital Market Research Center
  (PACAP)
• 2. The sample period from 1975 to 2005
• 3. The risk-free interest rate 30-day Gensaki
  rate
• 4. 36,233 firm-year observations

81
Variables

• We compute BE/ME as the ratio of book value of
  equity (as Fama and French, 1992) at the end of
  March (the end of the fiscal year) divided by the
  market value of equity at the end of March from
  1975 to 2005.
• 2. We compute market capitalization (ME) using
  market equity at the end of June in the calendar
  year t.

82
Variables

• 3.We calculate institutional trading (DITH) as
  changes in institutional ownership between fiscal
  year end t-2 and fiscal year end t-1.
• To control for systematic component, we compute
  industry adjusted change in institutional
  ownership (AdjDITH) as DITH subtracts median
  value of industry DITH, where industry DITH is
  measured by two-digits SIC industry.
• Adjusted change in individual ownership (AdjDIND)
  is defined similarly.

83
(No Transcript)
84
Relations among BE/ME, size, AdjDITH, and AdjDIND

• BE/ME is negatively associated with institutional
  trading and positively associated with individual
  trading.
• Size is negatively associated with individual
  trading, however, unrelated with institutional
  trading.

85
(No Transcript)
86
(No Transcript)
87
Table 3. Characteristics for quintile portfolios
formed on book-to-market equity ratio or size
88
Table 4. Average parameter values from
cross-sectional regressions of annual
book-to-market ratio and size on changes in
institutional and individual ownership
89
AdjDITH, AdjDIND, and the cross-section of
average stock returns

• The institutional trading is significantly and
  negatively related to future stock returns.
• Individual trading has no significant influence
  on current price. Particularly, their trading is
  also unrelated to future stock returns.

90
Table 5. Average monthly percent returns and
characteristics for decile portfolios formed on
AdjDITH
91
Table 6. Average monthly percent returns and
characteristics for decile portfolios formed on
the AdjDIND
92
Table 7. Average parameter values from
cross-sectional regressions of monthly returns on
size, book-to-market ratio, and AdjDITH and
AdjDIND
93
Institutional and individual trading behavior and
BE/ME and size premiums

• The relation between AdjDITH (or AdjDIND) and
  BE/ME (or size) seems to be weak.
• After purging the premiums associated with
  AdjDITH and AdjDIND, the BE/ME and size premiums
  are still significantly positive.

94
Table 8. Portfolio returns based on two-way
independent sorts
95
Table 9. Portfolio returns based on dependent
double sorts
96
Table 10. Average returns of BE/ME decile
portfolios
97
Table 11. Average returns of size decile
portfolios
98
Size strategies with institutional and individual
trading preference

• Neither the institutional trading nor individual
  trading can significantly improve the size
  strategy!

99
Table 12. Investing strategies based on
independent double sorts
100
Value strategies with institutional and
individual trading preference

• The strategy (3) is the highest profits among
  strategies (1) to (4).
• This suggests that by including the information
  about institutional trading preference, i.e., buy
  growth stock and sell value stock, one can
  improve the profitability of the value strategy.
• Information about individual trading preference
  has limited ability in improving value strategy.

101
Table 12. Investing strategies based on
independent double sorts
102
Conclusions

• A significantly and economically negative
  relation between institutional trading and future
  stocks returns exists.
• There is a negative association between
  institutional trading and book-to-market ratio
  (BE/ME). However, insignificant relation between
  institutional trading and size has been found.
• Although institutional and individual trading
  seem to be associated with BE/ME and size, their
  impacts appear to be limited on BE/ME and size
  premiums.
• Incorporating information about the institutional
  trading preference can significantly enhance the
  value strategy.

103

• The End
• Thanks