impact of media on corporate social responsibility: evidence from corporate philanthropic...
TRANSCRIPT
Impact of media on corporate social responsibility: Evidence from
corporate philanthropic contribution
Jingoo Kang Y.Han (Andy) Kim
Korea University Nanyang Business School
IKEA founder pledges £1bn to charity following Nazi past revelations
• He instructed the IKEA foundation to more than double its charitable spending.
• The move follows last month’s revelations by Swedish journalist Elisabeth Åsbrink that Mr Kamprad at the age of 17 had been an active recruiter and a registered member of the Svensk Socialistisk Samling (SSS), the successor to the Swedish Nazi party. – 18th Sep 2011, The Telegraph
Nov 8th, 2009
Negativity in public opinion about Goldman Sachs
Goldman’s Philanthropic Contribution (PC, hereafter) jumped to rank #2 in 2010 [$70m315m!]
Source: The Chronicle of Philanthropy + Compustat
Research questions
• Should a firm spend shareholders’ money on PC in response to negative public opinion about the firm in the media?
• Is spending more on PC in response to media negativity (especially CSR news) a symptom of agency problem? Or is it consistent with stakeholder value maximization?
Goldman Sachs
• They belong to stronger governance group – Low E index group (the index value=2)– High independent directors group (82%)– Warren Buffet, the mastermind of long-term
investor, bought perpetual preferred stock of GS for $5 billion in Nov.2008, and has been monitoring the bank.
Stakeholder value maximization story
• Firm is a nexus of contracts between shareholders and other stakeholders (Coase, 1939).
• CSR is a signal that the firm is committed to its implicit contracts with the various stakeholders (Deng, Kang, & Low 2013).
• Media is an outlet of the opinions of the stakeholders (Dyck & Zingales, 2003).
• PC in response to negative media coverage would not only improve CEO’s reputation but also improve the firm’s relationship with important stakeholders.
Stakeholder value story (cont.)
• In the long run, PC spending would signal the commitment to stakeholder value, and contribute to the firm’s long-term performance.
• Also, PC spending would enhance the reputation of the CEO in the outside labor market, and increase the probability of being appointed as an independent director.
• Stock price response to the announcement of PC spending would be mixed or unclear because:– PC spending could be anticipated as the investors read negative
media coverage in CSR beforehand– Difficult to price in the stock market (Edmans 2011)
Shareholder value maximization (Agency problem) story
• The only way a corporation can do good in society is to maximize shareholder value (Friedman, 1970).
• Media is not a producer of information but that of entertainment (Jensen, 1979).
• Market is efficient (Fama, 1970), so no need to care about the tone in the media.
• PC spending is a waste of shareholders’ money (Masulis & Reza, 2013).
• Weakly governed firms would spend more on PC• Investors’ response would be negative• Unless there is a friction in the director labor market, CEOs who spend
$ on PC would have smaller chance of being appointed as independent directors.
Summary of findings• Supporting stakeholder value maximization hypotheses.• Firms spend more on PC when the tone of the CSR news is more negative
with extensive coverage.• PC spending is not a function of governance strength
(E-index/%ind.dir./small board).• Rather, firms with high ownership by long-term oriented institutions spend
more on PC (as well as R&D).• Consistent result when using ∆CSR score of KLD instead of PC spending.• Insignificant price response to PC spending ann.• CEOs with PC spending (t) face better chance of being appointed as
independent directors (t+1).• Negative CSR news tone (t) market share declines (t+1)• PC spending is effective in reducing the negativity of media and in
strengthening Tobin’s Q.
Empirical design 1
• Firms that (1) contact more retail customers (2) are heavily regulated (3) are known to be environmentally hazardous would spend more on PC. industry fixed effects
• Firms would spend more on PC when natural disasters take place. year fixed effects
• Significant persistence in PC due to multi-year commitments: “$1 billion commitment over the next 4 years” = 250 million per year control for current year’s PC spending
Empirical design 2
• Measure of negativity about the firm in the media:– Tetlock’s (2007, 2008) measure [untabulated, but
consistent results]– Loughran & McDonald’s (2011) measure refined for
finance and accounting [V].– Aggregate all the news articles about the firm (i) in a given
fiscal year (t).
%FinNegativ e¿=¿negative words¿− ¿ positivewords¿
total word count ¿
Empirical design 3
• Measure of negativity about the firm in the media:– Predicts future earnings and Q (Gurun & Butler,
2012)– Concern: linguistic negativity measure could be a
proxy for soft information that is otherwise hard to quantify about the future fundamentals (Tetlock et al., 2008)
– Solution: • Filter out the news about fundamentals & focus on CSR
news: use Factiva News codes
Data 1
• The Chronicle of Philanthropy– Top 300 out of Fortune 500 firms, 2000~2010– Survey data augmented by hand collected data based on tax
return (Forms 990-PF) for firms with foundations.– 202 unique firms
• 1.7m (0.5m) Factiva news articles of Execucomp firms (sample firms) – top sources: Dow Jones News Wires, Reuters Newswires, WSJ, FT,
NYT, etc.– Use tags of 841 unique news codes in the XML format to filter:
• Non financial (580 codes)• CSR (66 codes)
Data 2
• Risk Metrics: E-index• Thomson Financial 13F – institutional
ownership• Brian Bushee’s website – classification of
dedicated/quasi-indexer/transient institutions
Table 1. primary result
• Other controls: lag(ln(PC /assets)), size, ad margin, R&D margin, CEO tenure, %TRA, %QIX, %ind.directors, leverage ratio, industry FE, year dummies.
Table 2. PC spending, regulated industry, and “sin stocks”
Dependent variable: ln(PC/assets)t+1
News article category: CSRIndustry Group: All Regulated Non Regulated
%FinNegt [1] 2.05 5.573 ** 1.895 0.183(1.18) (2.10) (0.81) (0.06)
ln(1+#articles)t [2] 0.026 ** 0.048 * 0.023 * 0.028(2.35) (1.97) (1.71) (1.53)
[1] * [2] 2.612 ** 3.983 ** 2.453 * 3.001 *(2.42) (2.29) (1.66) (1.72)
1{Regulated Industry} -0.112(-1.56)
1{Sin Stocks} -0.152 *(-1.66)
Other controls Yes Yes Yes YesIndustry FE No No No Yes
Year FE Yes Yes Yes YesN 798 205 593 593Adj.R2 0.893 0.778 0.846 0.849
Do better governed firms respond more sensitively to media negativity through PC spending?
• Table 3. not significantly!
• Other controls: the same as Table 1.
Concern about the PC data
• Some of you may be concerned about the nature of the PC data we use: survey data augmented by additional hand collection
• Under profit maximization hypothesis, change in PC is a part of firm effort on CSR.
• KLD data gives us CSR score for larger number of firms over longer period of time.
• Free us from selection bias concern.• Replace PC with the change in CSR score to see if we get
consistent results• CSR score excluding governance score• Kang (2013) shows that PC is significantly positively correlated
with change in CSR.
Replacing PC with Chg.KLD score ex. Governance (T.4)
Other controls: ROA, R&D margin, AD margin, size, 1{chairman=CEO}, tenure, E-index, %DED, %QIX, %TRA, %ind. Directors, E-index, tax rate, leverage, and constant.
Stock market event study
• If PC maximizes profit, the price response to the announcements of PC should be positive under the conditions of negative media coverage and better governance.
• Search news articles in Factiva over the sample period for the sample firms
• Keyword “(charitable or philanthropic) and (donation or contribution)”
• Source: all news outlets in Factiva• Restriction: only in the title and the first paragraph• First news announcements for each event
Cleaning the sample if contaminated by major corporate events [-1,1](T.5)
Panel A. Identification of uncontaminated announcements of philanthropic contributionAll announcements 584 Less: Earnings announcements 15 Less: M&A announcements 2 Less: Restatement 0 Less: Lawsuit 0 Less: CEO turnover 0
Uncontaminated announcements: 567
CAR regression(T.6)
• Investors welcome the PC spending announcement that was preceded by media negativity & combined with stronger governance
Sample: amount of cash donation by the firm is at least 1,000,000Dependent variable: CAR[0] CAR[0,1]%Fin Negative in CSR news [1] -0.475 -0.245
(-1.63) (-0.70)1{High %ind.director} [2] -0.002 -0.002
(-0.51) (-0.35)[1] * [2] 0.675 * 0.8 *
(1.67) (1.66) ln(#firm news about CSR) 0 -0.001
(0.30) (-0.37)1{Employee or customer matched donation} -0.024 -0.045 *
(-1.21) (-1.67)1{with non-cash donation} 0.01 0.007
(1.29) (0.49) ROA -0.016 -0.054
(-0.56) (-1.28)Tobin's Q 0.002 0.004
(0.93) (1.32) E-index 0 -0.002
(-0.14) (-0.73)constant -0.002 0.003
(-0.27) (0.43) Disaster dummies Yes YesN 141 141Adj.R2 0.006 0.026
We Masulis & Reza (2013)Follow Masulis, Wang, Xie (2012)? Do Don't Not use proxy filing date due to serious contamination concern
Follow Don't
Remove the contaminated events Follow Don'tEvent window short:[0],[0,1] long:[-1,3]N= 141 53
CARCAR>0 for good
governance & negative CSR tone'
CAR<0
supporting story Profit Maximization Agency problem
Why opposite result with Masulis and Reza (2013) in event study?
Stakeholder value
PC and CEO reputation in director labor market: probit model (T.7)
Other controls: 1yr stock perf. ROA, size, idiosyncratic risk, 1{chairman=CEO}, year fixed effects, constant.4.26% of unconditional probability. One standard deviation increase in PC increases the probability by 2.4%
Dependent variable: 1{appointed as independent director of another firm}t+1
Coeff Mfx Coeff Mfx
ln(Philanthropic Contribution/Total Assets)t 0.196 ** 0.015 * 0.19 * 0.024 *(2.10) (1.94) (1.68) (1.77)
%FinNeg in CSR news -16.604 -1.236 -14.645 -1.811(-1.39) (-1.19) (-1.07) (-1.01)
Tenure 0.127 0.009 * 0.074 0.009(1.31) (1.71) (0.83) (0.90)
Tenure^2 -0.012 -0.001 ** -0.005 -0.001(-1.58) (-2.29) (-1.34) (-1.52)
Firm age 0.003 0 0.008 0.001(0.33) (0.32) (0.63) (0.63)
CEO age 0.077 *** 0.006 ** 0.048 0.006(3.03) (2.52) (1.42) (1.38)
Industry FE Yes(SIC2) Yes(SIC3)N 324 257Pseudo R2 0.219 0.205
PC spending in response to media negativity and firm value (Q)(T.10)
• Other controls: ln(#articles), 1yr stock perf., ROA, ad margin, R&D margin, leverage, investment, size, %inst.own, idiosyncratic risk, constant, industry FE, year dummies.
Does negativity in CSR news chase away the customers? (T.8)
Dependent variable: (Ranking of market share t+1)*(-1)
News Category: Non Financial CSR CSR
%FinNegt [1] -0.035 -9.607 * -21.769 **
ln(#articles)t [2] -0.019 0.025 -0.025[1] * [2] 6.0961yr stock performance 0.262 0.263 0.271
ROAt -1.32 -1.442 -1.384
ROAt-1 0.829 0.829 0.898
Advertisement expense/Salest -2.951 -2.771 -3.448
Advertisement expense/Salest-1 23.556 ** 23.906 ** 23.997 **
R&D/Salest -1.658 -1.729 -1.655
Leveraget -0.373 -0.383 -0.434
Ranking of market sharet 0.805 *** 0.803 *** 0.803 ***
Size:ln(Total Assets)t 0.482 0.494 * 0.467
Herfirndahl indext 0.31 ** 0.413 ** 0.481 **constant -5.93 * -6.163 * -5.786 *Industry & year FE Yes Yes YesN 672 672 672Adj.R2 0.94 0.941 0.941P-value of Wald Test:[1]+[1]*[2]=0 0.0346
Conclusion• Corporations spend shareholders money on PC (or CSR, more generally)
to maximize stakeholder value• Firms spend more on PC when the tone of the CSR news is more
negative with extensive coverage.• PC spending is not a function of governance strength • Rather, firms with high ownership by long-term oriented institutions
spend more on PC (as well as R&D).• Insignificant price response to PC spending ann.• CEOs with PC spending (t) face better chance of being appointed as
independent directors (t+1).• Negative CSR news tone (t) market share declines (t+1)• PC spending is effective in reducing the negativity of media and in
strengthening Tobin’s Q.
Implication for current issues:
Negative shocks to CSR news take place in the media. Contribution to philanthropy is a valid way of coping.
Thank you!