Previous empirical work on adverse consequences of CEO overconfidence raises the question of why firms hire overconfident managers. Theoretical research suggests a reason: overconfidence can benefit shareholders by increasing investment in risky projects. Using options- and press-based proxies for CEO overconfidence, we find that over the 1993–2003 period, firms with overconfident CEOs have greater return volatility, invest more in innovation, obtain more patents and patent citations, and achieve greater innovative success for given research and development expenditures. However, overconfident managers achieve greater innovation only in innovative industries. Our findings suggest that overconfidence helps CEOs exploit innovative growth opportunities.
Link to original paper https://ssrn.com/abstract=1598021.
3. 3
What is overconfidence?
• Tendency to think one is better than
– One actually is
– Average
– One is viewed by others
on some relevant dimension
– Ability
– Judgment
– Prospect for success
• For our purpose
– Overestimate the expected payoffs of uncertain endeavors
4. 4
Variation in Overconfidence
• Strong and stable differences in degree of overconfidence
across individuals
– Biais et al. 2005, Klayman and Burt 1998, Klayman et al. 1999
• Overconfidence also varies with:
– Gender
• Barber and Odean 2001, Biais et al. 2005
– Culture
• Lee et al. 1995, Koellingera, Minnitic and Schaded 2007
– Whether individual is expert in decision domain
• Wagenaar and Keren 1986
• Substantial differences in overconfidence of managers
– CEOs
• Malmendier and Tate, 2005, 2008
– CFOs
• Ben-David, Graham, and Harvey, 2010
5. 5
Link to Innovation
• Greater overconfidence
– In difficult tasks (Hard-Easy Effect)
– In settings with ambiguous and deferred feedback
(Einhorn 1980)
• R&D activities appealing for considering
overconfidence effects
– Highly risky and challenging
– Outcome feedback deferred
– Talent- and vision-sensitive
6. 6
R&D
• Immediate expensing reduces earnings
– Create reputational risk to manager
– Potentially constrains CEOs
– Optimism of overconfident CEOs constrains
them less
7. 7
Research Questions
Do overconfident CEOs (who overestimate expected
project payoffs)
• Take on riskier projects?
• Invest more in R&D?
• Achieve more innovations?
– Greater patent counts
– Greater patent citations
• Achieve more innovative success controlling for
amount of investment in R&D?
Better Innovators?
8. 8
Relation to Literature
• Overconfidence linked to investment
– Capital Expenditures
• Malmendier & Tate JF 2005
– Acquisitions
• Malmendier & Tate JFE 2008
• R&D different type of investment
– Highly risky, challenging, vision-related, deferred
outcome, so especially suited to find
overconfidence effects
– Outcomes are measurable as patents & citations,
so can test whether overconfidence leads to more
effective innovation
9. 9
Literature, contd.
• Firms hire overconfident managers
– Malmendier & Tate (2005, 2008)
– Ben-David, Graham and Harvey (2010)
– Graham, Harvey, and Puri (2010)
• Puzzling if overconfidence implies suboptimal
investments
– More negative acquisition announcement effects
(Malmendier & Tate 2008)
• Our findings suggest a possible solution to the
overconfident manager puzzle
– Overconfident CEOs are better innovators
10. 10
Other Effects of Managerial
Confidence
Overconfident Managers
• Less external finance, less equity issuance
– Malmendier, Tate, and Yan (2010)
• More aggressive financing and other managerial
policies
– Ben-David, Graham, and Harvey (2010)
• Issue more optimistic forecasts, More earnings
management
– Hribar and Yang (2010)
• More earnings management and financial fraud
– Schrand and Zechman (2010)
11. 11
Summary of Results
Overconfident CEOs
• Invest more heavily in R&D
• Take on greater risk (std. dev. of returns)
• Achieve a greater total quantity of innovation as
measured by patenting and citations.
• Achieve greater innovative success after controlling for
the level of R&D expenditure.
• Achieve greater total patents and citations only in
more innovative industries
• Better able to exploit industry growth opportunities
12. 12
Data Sources
• Overconfidence measures
– Execucomp for options compensation data
– Factiva for text of press articles about CEO
• Innovation measures
– Compustat for R&D spending
– NBER Patent Database (1976-2006)
• 2.3 million patents, 23.6 million citations
• CRSP, Compustat, and Execucomp for other
variables
14. 14
Innovation Measures
• R&D spending
• Patent count in application year
• Citation count
– Measure forward citations
– Adjusted to take into account truncation bias (newer
patents have less time to collect citations)
– Qcitation count: Weighting index empirically estimated
• Hall, Jaffe, and Trajtenberg, 2001, 2005
– Ttcitation count: Weighted by the average citation count of
all patents in the same technology class and in the same
year
15. 15
Options-Based Overconfidence Measure
• Malmendier & Tate (2005 and 2008), Campbell, Johnson,
Rutherford, and Stanley (2011)
• Indicator variable
– 1 if overconfident, 0 otherwise
• Overconfidence measured as overexposure of CEOs to
the idiosyncratic risk of their firms through their
holdings of options and stock.
• A CEO is defined as overconfident once she postpones
the exercise of vested options that are at least 67% in
the money.
• Persistent personality trait (Malmendier & Tate 2005)
– Using option measure, once identified as overconfident,
remains so through the sample period.
16. 16
Press-based Overconfidence Measure
• Malmendier & Tate (2005, 2008)
• Factiva search for articles referring to the CEO in New
York Times, Business Week, Financial Times, Wall Street
Journal, The Economist, Fortune, and Forbes
• Confidence-related words
– ‘confident’, ‘optimistic’, and variants
• Non-confidence related words (pessimistic, reliable,
steady, practical, conservative, frugal, cautious, gloomy,
and variants, and negated confidence-related words)
• Overconfident if number of articles with confidence-
related words > number of articles with non-
confidence related words in a given year
• Control for TotalMention
– # of articles mentioning CEO
19. 19
Overconfidence and Risk-Taking
Table 3
• Overconfidence – Higher stock return % volatility
(std. dev. daily stock returns)
• Year & 2-digit SIC fixed effects, Std err cluster at
firm level
• Controls
– Firm characteristics
• Log(sales), Log(PPE/emp), Stock return, Tobin’s Q,Sales
growth, ROA, Book leverage, Cash
– Manager characteristics
• Tenure, Delta (Manager pay sensitivity to firm value) and
Vega (Manager pay sensitivity to firm volatility)
– TotalMention for press-based
20. 20
Overconfidence and Risk-Taking
Table 3
• Options-based Overconfidence
– Increases volatility by 6 bp (daily), 1% (annual)
– Comparison: doubling firm sales (size measure)
decreases volatility by 2.5% per year.
• Press-based Overconfidence
– Increases volatility by 20bp (daily), 3% (annual)
21. 21
Overconfidence and R&D Expenditures
Table 4
• Dependent variable: R&D/Assets
• Overconfident CEOs higher R&D
• Options-based Overconfidence increases R&D by
0.4% to 0.8% of assets.
– Mean R&D/Assets: 2.86%
– An increase of 27% relative to this mean
• Press-based Overconfidence increases R&D by 1%
of assets.
– Mean R&D/Assets: 3.72%
– An increase of 27% relative to this mean
22. 22
Overconfidence and patenting activity
Table 5
• Dependent variable Log (1 + Patent)
• Overconfident CEOs higher Patents
• Options-based Overconfidence increases
patent count by 9% to 11%
• Press-based Overconfidence increases
patent count by 27% to 29%
23. 23
Overconfidence and patent citations
Table 6
• Dependent variable:
– Log(1 + Qcitation Count)
– Log(1 + TTcitation Count)
• Overconfident CEOs – higher citations
• Options-based Overconfidence increases
adjusted citations by 11% to 18%.
• Press-based Overconfidence increases adjusted
citations by 24% to 43%.
24. 24
Industry innovativeness
Table 7
• Innovative industry:
– Adjusted citations per patent for the industry >
median average citations across all industries
• Effect of confidence is only in innovative
industries, stronger effects than in full sample
• Overconfident CEOs – higher patent counts,
higher citations but only in innovative industries
25. 25
Other Robustness Checks
• Control for several lags of stock returns
• Alternative options-based measure, require CEO to
exhibit late exercise twice.
• Press-based, delete articles with words relating to
“project” that are within 20 words of confidence-related
words.
• Press-based, requiring at least one news article
• Exclude self-cites
• Delete firm-years with zero patents (citations)
• Use 2-year lagged overconfidence meaures
• Eliminate tech-boom (1998-2000)
26. Extensions
• Table 8: Overconfident CEOs higher citation
per patent, especially within innovative
industries.
• Table 9: Different degrees of overconfidence.
26
27. 27
Endogeneity
(Matching of Firm to CEO) Table 10
• Firm characteristics (e.g., growth) vary over
time Matching effects maximized when
manager first hired
• Keep firm-years with CEO tenure > 4 years (6
years)
• Similar effects for stock return volatility, R&D,
patent count, and patent citations
• No sign that controlling for matching weakens
the effect of confidence on innovation
28. 28
CEO overconfidence and innovation
effectiveness, Table 11
• Dependent variable:
– Patent count and adjusted citation
– Control for R&D
• Divide into innovative industries vs non-innovative
industries (to maximize power)
• Overconfident CEOs – higher patent counts and citations
within innovative industries
• Smaller than coefficients w/o R&D control, but still
substantial and positive
Overconfident CEOs more effective innovators (for given R&D)
29. 29
CEO overconfidence and innovation
effectiveness
• Overconfident CEOs more effective innovators
• Finding offer possible solution to the overconfident CEO
puzzle:
– Why do firms hire overconfident CEOs, which has major
effects (investment, acquisitions), at least in some contexts
adversely (acquisitions)?
• Impact on firm value? To avoid endogeneity issues, examine a
more specific question: Does overconfidence allow firms to
translate growth opportunities into realized firm value?
30. 30
CEO overconfidence and firm value,
Table 12
• Dependent variable:
– 1-year ahead Tobin’s Q
• Instrument for exogenous growth opportunities
– Industry P/E (Bekaert et al. 2007)
– Interact overconfidence measure with industry P/E
• Stronger association between exogenous growth
opportunities and firm value for overconfident CEOs
– Results mainly within innovative industries
31. 31
Conclusion
Overconfident CEOs
• Take on more risk
• Invest more heavily in R&D
• Achieve a greater total quantity of patents and patent
citations
– Especially in industries where innovations are important
• Better innovators
– Achieve greater innovative success after controlling for the level
of R&D.
– Contrast with past finding of adverse effects of overconfidence
effects in other domains (e.g. acquisitions)
– Suggests possible reason why firms continue to hire
overconfident CEOs– the overconfident CEO puzzle
• Overconfident CEOs are more effective at exploiting
growth opportunities and translating them into firm
value.
Editor's Notes
on some relevant characteristic (ability, …)
Prospect for success – optimism
In psych lit, overconfidence take a variety of forms. Don’t distinguish because tangential to our purpose.
Diff aspects of overconf have similar implications for managers to bear risk and to undertake challenging projects.
‘better than average effect’ vs absolute overestimation of one’s own ability
- induce a belief that one has a high relative position.
“… the women are strong, the men are good-looking, and all of the children are above average.”
Ability to identify good projects (judgmental overconfidence)
Ability to manage challenging projects (overconfidence about implementation ability)
- lead to overestimate the likelihood of success (overoptimism)
Overest expected payoffs because overest belief accuracy, ability to generate action-relevant information (DST 1994), ability to control environment to bring about success
Consonant with self-image of manager who is strongly ego-driven or self-aggrandizing
Patenting may be less relevant for certain industries, either because they are less innovative, or because in these industries innovation does not result in patents.
Innovative industries are where patents are important (measured as citations exceed median count in a given year)
Patent counts measure by applications during the year – only patents that are granted appear in database.
Patents applied in 2004 and 2005 may not appear in the database by latest year 2006.
Hall, Jaffe, and Trajtenberg (2001) suggest restrict period to end in 2003 and include year fixed effects in our regressions.