European Financial Management Symposium 2009 Judge Business School, University of Cambridge, UK - PowerPoint PPT Presentation

Loading...

PPT – European Financial Management Symposium 2009 Judge Business School, University of Cambridge, UK PowerPoint presentation | free to download - id: 24c9aa-YzgzO



Loading


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation
Title:

European Financial Management Symposium 2009 Judge Business School, University of Cambridge, UK

Description:

The Chinese stock market (with its inadequate information environment and ... In China's emerging stock market, the financial mass media are just 'opening up' ... – PowerPoint PPT presentation

Number of Views:42
Avg rating:3.0/5.0
Slides: 30
Provided by: aa70103
Learn more at: http://efmaefm.org
Category:

less

Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: European Financial Management Symposium 2009 Judge Business School, University of Cambridge, UK


1
European Financial Management Symposium 2009
Judge Business School, University of Cambridge,
UK
  • Media Coverage, Stock Price Informativeness and
    Trading Activity Evidence from China
  • Stephen X.H. Gong and Ferdinand A. Gul
  • Hong Kong Polytechnic University

2

Presentation outline
  • Research objectives
  • Background literature review
  • Research hypotheses
  • Sample selection, variable measurement and
    testing method
  • Empirical results
  • Further analyses
  • Conclusion

3

Research objectives
  • Primary objective examine the effects of media
    coverage on stock price performance in Chinas
    emerging market context
  • Specifically, how does media coverage affect
    stock price informativeness and trading activity?
  • What drives the observed relationship between
    media coverage and stock price performance?
  • Parallel objectives (1) What determines the
    level of media coverage a company gets? (2)The
    relative descriptive validity and explanatory
    power of alternative measures of media coverage
    (quantity versus quality)?
  • The Chinese stock market (with its inadequate
    information environment and speculative investors
    who tend to herd) provides a fertile testing
    ground for these research objectives.

4

Background and literature review
  • No role for information aggregators in
    economic and finance models that assume perfect
    markets.
  • Two conflicting schools one school of thought
    questions the validity of the press as an
    important information intermediary in the economy
    (Jensen, 1979 DeAngelo, DeAngelo and Gilson,
    1994, 1996).
  • Some research (based in US) finds that
    substantial movements in share prices do not seem
    to correspond to changes in economic fundamentals
    (Cutler, Poterba and Summers, 1989), and that
    stories reported in the financial press have
    little impact on stock returns and trading volume
    activity over short-event horizons (Mitchell and
    Mulherin, 1994 Berry and Howe, 1994)
  • Such results do not rule out the possibility
    that media coverage may affect individual firms
    or the broader market in ways other than
    immediate price or volume changes.
  • Shiller (2005) concludes that while the news
    stories that broke around the time of the stock
    market crashes of 1929 and 1987 were not the
    essential cause of the crashes, news events have
    the effect of causing an attention cascade among
    investors by heightening their fixations on
    market moves, and in so doing, the news media
    become fundamental propagators of speculative
    price movements.

5

Background and literature review (2)
  • In the real world, people obtain much of their
    information from the media, which play an
    important role in selecting which pieces of
    information to communicate to the public and in
    adding credibility to information collected
    through other sources (Dyck and Zingales, 2002).
  • The impact of media coverage on stock prices is
    formalized in Merton (1987), who posits an asset
    pricing theory that deviates from the
    Sharpe-Lintner capital asset pricing model in
    that investors do not have full information about
    the available securities and thus choose to
    invest in the shares of companies they know
    about.
  • Merton (1987) notes that a newspaper or other
    mass media story about the firm or its industry
    that reaches a large number of investors who are
    not currently shareholders could increase the
    number of investors in the firm, and hence reduce
    the firms cost of capital and increase the
    market value of the firm.
  • This incomplete information hypothesis is
    empirically supported in Amihud, Mendelson and
    Uno (1999) in the Japanese market context and
    Fang and Peress (2009) in the U.S. market
    context.
  • The availability of low-cost (perhaps new,
    substantive) information through the mass media
    improves the cost-benefit tradeoff on information
    collection, leading to more informed trading and
    more informative pricing (Grossman and Stiglitz,
    1980).

6

Background and literature review (3)
  • Media coverage increases the reputation cost of
    regulators, intermediaries and managers, and thus
    has the potential to improve corporate governance
    (Dyck, Volchkova and Zingales, 2008). Better
    corporate governance, in turn, leads to more
    informative stock prices by encouraging
    collection of, and trading on, private
    information (Ferreira and Laux, 2007).
  • Apart from being an information intermediary,
    the mass media have the potential to breed
    familiarity and hence affect overall stock
    performance (e.g. propensity to invest/trade in
    stocks)
  • There is a growing body of research on
    investors tendency to invest with a company they
    know or think they know (e.g. Huberman, 2001 Li,
    2004 Lehavy and Sloan, 2008).
  • Grullon, Kanatas and Weston (2004) find that
    investors degree of familiarity with a firm
    affects its cost of capital and consequently its
    value.
  • Individual investors prefer holding stocks with
    high recognition (Frieder and Subrahmanyam,
    2005).
  • Barber and Odean (2008) report that individual
    investors are net buyers of attention-grabbing
    stocks, e.g. stocks in the news, stocks
    experiencing high abnormal trading volume, and
    stocks with extreme one-day stock returns.

7

Background and literature review (4)
  • Thus, both finance theory and practice predicts
    that media coverage affects stock performance.
  • In countries with under-developed institutions
    (e.g. poor information environment and investor
    protection), mass media has the potential to play
    an important role, either in providing the
    general public with (perhaps new, substantive)
    information, and/or in breeding familiarity or
    grabbing investors attention.
  • Such potential may however be adversely affected
    by the credibility/independence of the media and
    investors behavioral patterns (e.g. are they
    fixated on (swayed by) the mere appearance of a
    company in the media, or do they also care about
    the quality of coverage?)

8

Background and literature review (5)
  • In Chinas emerging stock market, the financial
    mass media are just opening up for relatively
    free reporting, and there is concern for the
    independence and credibility of the media sector.
  • Both institutional investors and (even more so)
    individual investors engage in active short-term
    trading, with the latter (which account for the
    lions share of stock turnover) tending to hold
    and trade in stocks characterized by smaller
    capitalization, lower price, poorer performance,
    and higher price-earnings ratio (CRSC, 2008).
  • A high level of stock synchronicity in China
    (Morck, Yeung Yu, 2000)Is the mass media a
    driving force behind such documented co-movement?

9

Research hypotheses
  • The presence and extent of media effects in
    emerging markets is an empirical question.
  • We focus on two key aspects of the media effect
    stock price informativeness and stock market
    participation.
  • Stock price informativeness (idiosyncratic
    volatility, IV) the extent to which stock prices
    incorporate firm-specific relative to marketwide
    information.
  • Prior research (e.g. Morck et. al, 2000 Chan
    and Hameed, 2006 Jin and Myers, 2006) suggests
    that if the firms information or institutional
    environment causes stock prices to incorporate
    more firm-specific information, market factors
    should explain a smaller proportion of the
    variation in stock returns, i.e. the return
    synchronicity or R-square from a market model
    regression should be lower.
  • Verrecchias (1979) hypothesizes that the
    relative degree of pricing efficiency of a
    security is predicated on the number of traders
    who actively participate in a market for that
    security.
  • H1 Is stock price informativeness enhanced by
    media coverage? The literature review (as well as
    intuition) suggests so.

10

Research hypotheses (2)
  • Stock turnover how frequently do investors
    buy/sell stocks.
  • Given the short-term trading behavior of Chinese
    investors, media coverage may fuel trading volume
    without necessarily increasing the number of
    registered shareholders. Hence our focus on stock
    turnover instead of shareholder base as a measure
    of the level of market participation.
  • H2 Is stock turnover enhanced by media
    coverage?
  • There is a need to control for possible
    endogeneity media coverage may follow from high
    stock turnover (an attention-grabbing
    event--Barber and Odean, 2008)
  • Hence our parallel interest in the determinants
    of media coverage, H3. There are few existing
    studies on this. Our tests are exploratory in
    nature.
  • H4, the source of media effects Does the media
    effect arise from media reports providing new,
    substantive information, or does it arise from
    media coverage breeding familiarity?

11

Research hypotheses (3)
  • Designing strong tests to discriminate between
    these competing hypotheses is difficult. We made
    a first attempt at this important question by
    devising and comparing two measures of media
    coverage, one based primarily on the quantity
    (times mentioned) of coverage, the other on the
    quality of coverage (i.e. the content of
    newspaper reports). The result on the relative
    descriptive validity and explanatory power of
    these measures may shed light on the sources of
    the media effect, and offer possible reasons for
    the conflicting results in prior studies with
    respect to the existence of media effects (e.g.
    Core, Guay and Larcker, 2008).
  • Although we propose H1 and H2 separately, there
    is a close link between stock price
    informativeness and stock turnover.
  • Trading is theoretically linked to the quality
    or extent of private information (e.g. Blume,
    Easley and OHara, 1994).
  • Ferreira and Laux (2007) consider stock turnover
    as one alternative to stock price informativeness
    in proxying for the intensity of private
    information flowing to a stocks market (they
    also use PIN etc as alternative measures).
  • Active short-term trading is prevalent among
    Chinese investors. Such trading may reflect
    factors unrelated to the quality or extent of
    private information (it can be fuelled by the
    quantity of media coverage).

12


Sample selection, variable measurement and
testing method
  • All China-listed A-shares during the period
    2000-2006.
  • Stock price informativeness (idiosyncratic
    volatility, IV)
  • IV captures the effects on stock returns of
    company-specific information relative to
    marketwide factors (Fama, 1976). This measure has
    been used in many papers (e.g. Morck et al. 2000
    Ferreira and Laux, 2007)
  • Recent research (e.g. Durnev et al. 2003)
    suggests that higher firm-specific return
    variation as a fraction of total variation
    signals more information-laden stock prices and
    therefore more efficient markets.
  • Chen, Goldstein and Jiang (2007) use stock
    return nonsynchronicity as a measure of private
    information incorporated into stock prices and
    find that investment responds more to stock
    prices when the stock return synchronicity is
    lower.
  • There is a continuing debate on the relationship
    between market model R-square and the quality of
    firm-specific information (e.g. Teoh, Yang and
    Zhang, 2008).


13


Sample selection, variable measurement and
testing method (2)
  • Stock turnover Annual average of total number
    of shares traded per day over number of tradable
    shares per day (monthly alternatives, as well as
    annual total number of shares and total values of
    shares traded, are also used as a check on
    robustness)
  • Media coverage (1) A simple frequency count
    (i.e. how many times is a company mentioned in
    the newspaper reports? Multiple mentions in the
    same article is counted once only). (2) A
    content-analysis-based measure (i.e. each
    newspaper report is read and scored on the basis
    of its correspondence with key information
    categories, e.g. discussions relating to
    current/future operations, earnings forecast,
    risk-return, corporate governance, management
    turnover, litigation, regulatory investigation,
    MA and spin-offs, related party transactions,
    etc). We adopt the content analysis methodology
    that is well-established in social sciences and
    communication research (Neuendorf, 2002
    Krippendorff, 2004 Riffe, Lacy and Fico, 2005) .
    See Appendix 1 for details.

14


Sample selection, variable measurement and
testing method (3)
  • Explanatory variables our focus is on media
    coverage as the key explanatory variable.
    Following Ferreira and Laux (2007), Chen,
    Goldstein and Jiang (2007), Wei and Zhang (2006),
    and Gaspar and Massa (2006) amongst others, we
    include the following control variables in the
    stock price informativeness and stock trading
    tests profitability (ROE), profits volatility
    (VROE), financial leverage (LEVERAGE),
    market-to-book ratio (MTB), natural log of market
    capitalization of equity (SIZE), number of
    analysts following (ANALYST), percentage
    ownership by institutional investors
    (INSTITUTION), a dividend payer dummy (DD), firm
    age since listing (AGE), year dummies, industry
    dummies, and a stock exchange dummy. In addition,
    we add systematic risk (BETA) as an additional
    control variable in the stock price
    informativeness tests following Dasgupta, Gan and
    Gao (2008). Institutional controls specific to
    China (e.g. ST, dual listing, real identity of
    beneficial controller) are also included. See
    Table 1 for definitions.

15


Sample selection, variable measurement and
testing method (4)
  • Pooled cross-sectional OLS regression
  • Typically used in prior research, but there is
    concern that the sample observations are not
    random (independent) drawings from the
    population.
  • System of equations method (2SLS and 3SLS)
  • Media coverage and stock turnover may be jointly
    determined (i.e. endogeneity)2SLS. There may be
    cross-equation correlation in the error
    terms3SLS.
  • Fixed effects panel data model
  • We use first differencing to control for
    unobserved (but time-constant) firm effects and
    strengthen the inference of causality.

16


Empirical results
17


Table 3. Correlation matrix
18

Brief note on summary and descriptive stats
  • Average R-squared is 0.42.
  • HITS has a wide dispersion. In contrast, CONTENT
    for many firms is zero. This is largely the
    result of our use of a random sampling procedure
    in order to make the task manageable (we first
    randomly select one day per month during the
    period 2000-2006. All news articles for each
    company on these days are then identified and
    content analyzed).
  • Correlation between HITS and CONTENT is
    0.44they are positively but imperfectly
    correlated.
  • IV and stock turnover are each correlated with
    media coverage in the expected direction
    (significantly ve).

19


Table 4 (partial). Distribution of news coverage
over time and across stock exchanges (a positive
time trend, with no significant difference
between the 2 stock exchanges)
20

Table 5. Determinants of media coverage (Panel
A) HITS
21

Table 5. Determinants of media coverage (Panel
B) CONTENT
22


Table 5. Determinants of media coverage (summary
of key results)
  • As a first control for possible endogeneity, we
    use lagged values of stock turnover in all model
    specifications here. Later a simultaneous
    equation is estimated.
  • Key results
  • A much better fit when measuring media
    coverage by HITS (which captures mainly the
    quantity of media coverage), with R-squared
    ranging from 65 to 70.
  • Size, (lagged) turnover, institutional
    ownership, analyst following, ST companies, and
    those controlled by private investors (relative
    to those controlled by SAMB) have a consistent
    strong positive association with media coverage
  • Shareholding concentration and companies with
    both A and B shares have a consistent strong
    negative association with media coverage
  • When media coverage is measured by CONTENT (with
    emphasis on the quality of media reports),
    goodness of fit ranges from 22 to 27.
  • Size, analyst following, and residual volatility
    have a consistent strong positive association
    with media coverage
  • Companies with both A and B shares has a
    consistent strong negative association with media
    coverage
  • The results here in part results from the lack
    of sufficient dispersion in CONTENT.

23


Table 6. Relationship between share price
informativeness and media coverage
24


Table 6. Relationship between share price
informativeness and media coverage (summary of
key results)
  • The model specifications have good fit, with
    R-squared ranging from 55 to 72 (mostly over
    70).
  • Media coverage (no matter if HITS or CONTENT is
    used), MTB, leverage, age, absolute past return,
    and ST firms are positively associated with stock
    price informativeness.
  • Beta, No. of shareholders, ROE, and DY are
    negatively associated with stock price
    informativeness.
  • Relative to firms controlled by private parties,
    firms controlled by the central government (firms
    controlled by local governments) have greater
    (lower) stock price informativeness.
  • Thus, greater media coverage is associated with
    higher stock price informativeness, even after
    controlling for other relevant factors.

25


Table 7. Relationship between share turnover and
media coverage (Panel A) HITS
26


Table 7. Relationship between share turnover and
media coverage (summary of key results)
  • The model specifications have good fit, with
    R-squared ranging from 55 to 65 (similar no
    matter how media coverage is measured).
  • Media coverage (no matter if HITS or CONTENT is
    used), beta, dividend yield, and those with both
    H-share and A-shares are positively associated
    with stock turnover.
  • No. of shareholders, shareholding concentration,
    age, residual volatility and absolute past return
    are negatively associated with stock turnover.
  • Thus, greater media coverage is associated with
    higher levels of stock turnover, even after
    controlling for other relevant factors.

27


Further analyses
  • To address the possible endogeneity problem
    involving media coverage and stock turnover, we
    run 2SLS (assuming no cross-equation correlation
    in the error terms) and 3SLS (appropriate if
    there is cross-equation correlation in the error
    terms).
  • The key results remain qualitatively the same as
    those obtained when running OLS for the separate
    equations.
  • To further support the inference of causality,
    we are in the process of estimating a fixed
    effects first-differencing panel data model.

28


Further analyses (2)
  • An on-going effort looks at the possible sources
    of the documented media effects (e.g. the media
    breeds familiarity vs the media conveys
    substantive information). This involves
    constructing more refined measures of media
    coverage, one that emphasizes mainly the quantity
    of media, the other emphasizing the quality (e.g.
    how much the newspaper reports relate to
    fundamental corporate information).
  • Some preliminary analyses divide media coverage
    into those that have high quantity but low
    quality, and those that have high quality but low
    quantity. Although the number of cases with
    divergent quality-quantity is relatively small,
    the preliminary results (see online version of
    paper) suggest that the quantity of media
    coverage has a stronger and more consistent
    impact on stock price informativeness and stock
    turnover. However, in many cases the quality of
    media coverage has incremental power beyond the
    quantity dimension of media coverage.
  • Work is in progress using other measures of
    information flow or information risk (PIN,
    bid-ask spread, etc).

29


Conclusion
  • Media coverage is positively associated with
    greater stock price informativeness and stock
    turnover (so yes to both H1 and H2). This holds
    no matter how media coverage is measured
    (quantity or quality).
  • Thus in the Chinese emerging stock market, media
    coverage plays a positive role in causing more
    firm-specific relative to market-wide information
    to be impounded into stock prices. Given that
    greater media coverage is also associated with
    higher levels of stock trading, a trading link
    may be established as the mechanism through which
    the media exerts its influence.
  • More work remains in order to answer H4 Does
    the media effect arise from media reports
    providing new, substantive information, or does
    it arise from media coverage breeding
    familiarity?
  • The preliminary evidence suggest the quantity of
    media coverage has a stronger and more consistent
    impact than does the quality of media coverage,
    although the latter has incremental power in some
    cases. Better empirical measures of the quality
    of media coverage are necessary and there lies
    the challenge.
  • Suggestions/comments are welcome.
About PowerShow.com