MANAGEMENT SCIENCE
Vol.57,No.8,August2011,pp.1469–1484
issn0025-1909 eissn1526-5501 11 5708 1469doi10.1287/mnsc.1110.1374
©2011INFORMS CEO Overconfidence and Innovation
Alberto Galasso
Joseph L.Rotman School of Management,University of Toronto,Toronto,Ontario M5S3E6,Canada,
alberto.galasso@rotman.utoronto.ca
Timothy S.Simcoe
Boston University School of Management,Boston,Massachusetts02215,tsimcoe@bu.edu
A re the attitudes and beliefs of chief executive officers(CEOs)linked to theirfirms’innovative performance?
This paper uses a measure of overconfidence,based on CEO stock-option exercise,to study the relationship between a CEO’s“revealed beliefs”about future performance and standard measures of corporate innovation.
We begin by developing a career concern model where CEOs innovate to provide evidence of their ability.The model predicts that overconfident CEOs,who underestimate the probability of failure,are more likely to pursue innovation,and that this effect is larger in more competitive industries.We test these predictions on a panel of large publicly tradedfirms for the years from1980to1994.Wefind a robust positive association between overconfidence and citation-weighted patent counts in both cross-sectional andfixed-effect models.This effect is larger in more competitive industries.Our results suggest that overconfident CEOs are more likely to take theirfirms in a new technological direction.
Key words:innovation;R&D;CEO overconfidence;managerial biases
History:Received May25,2010;accepted April8,2011,by Kamalini Ramdas,entrepreneurship and
innovation.Published online in Articles in Advance June24,2011.
1.Introduction
Economic models typically assume that beliefs are correct on average.However,a large body of evi-dence from applied psychology shows that individu-als routinely overestimate their ability(Svenson1981). Although much of this evidence comes from sur-veys and lab experiments,there is growing interest in measuring the impact of overconfidence in thefield (DellaVigna2008).This paper uses a novel measure of chief executive officer(CEO)overconfidence devel-oped by Malmendier and Tate(2005a,b;2008)to study the relationship between managerial overconfi-dence and corporate innovation.1
Prior research on managerial overconfidencefinds that overconfident CEOs tend to destroy value through unprofitable mergers and suboptimal invest-ment behavior(Malmandier and Tate2005a,b;2008). These results raise the question of why companies hire and retain overconfident CEOs.Although sev-eral theories suggest that overconfidence may gener-ate value through greater exploration and risk taking (Bernardo and Welch2001,Goel and Thakor2008), 1Throughout this paper we follow the convention adopted by Malmendier and Tate(2008),and refer to self-serving attribution ,overestimating outcomes related to one’s own abilities) as“overconfidence.”We use the term“optimism”to refer to over-estimation of exogenous positive ,growth in gross domestic product).
there is no empirical evidence of this effect from out-side the lab.
We argue that for large-firm CEOs,overconfidence is associated with an increased propensity to inno-vate.To make this idea precise,we propose a sim-ple career concern model where CEOs decide whether or not to innovate.Successful innovation is rewarded because it reveals new information about managerial ability.Conversely,failure causes the market to infer that a CEO lacks talent.Overconfident CEOs underes-timate the likelihood of failure,and are therefore more likely to pursue innovation.This effect is larger in more competitive industries,where successful inno-vation reveals more information about CEO ability, leading to a large payoff that overconfident CEOs are eager to capture.This simple theory provides one answer to the puzzle posed by previous research:If the rewards from successful innovation are large,an overconfident CEO’s increased propensity to innovate may offset the negative impacts of unprofitable merg-ers and suboptimal investment behavior.
To test these predictions,we combine standard measures of innovation,based on U.S.patent data, with a measure of CEO overconfidence developed in
a series of papers by Malmendier and Tate(2005a,b;
2008).The measure is constructed by using CEOs’personal investments to capture“revealed beliefs”
about theirfirms’future performance.Specifically, CEOs are classified as overconfident if they hold 1469
Galasso and Simcoe:CEO Overconfidence and Innovation 1470Management Science57(8),pp.1469–1484,©2011INFORMS
highly in-the-money stock options after they are fully vested.We run panel data regressions on a sample of290firms and627CEOs during the period1980 to1994.This sample consists of largefirms,primarily from manufacturing and technology industries,where we observe significant patenting.
Our main result shows that the arrival of an over-confident CEO is correlated with a25%to35% increase in citation-weighted patent ,for-ward citations received by patentsfiled in a given year).The effect is larger if we assume that a CEO only becomes overconfident after failing to exercise in-the-money option grants,instead of treat-ing overconfidence as a permanent trait.We con-sider several additional outcome variables andfind that overconfidence produces similar-sized effects for unweighted patent counts,research and development (R&D)expenditures and citations per issued patent. Interacting overconfidence with industry-level mea-sures of competition reveals that the overconfidence effect is larger when product market competition is more intense.These interactions pr
ovide support for our analytical framework and also illustrate how industry characteristics may amplify(or mute)the impact of behavioral biases.
We extend these baseline results in several direc-tions.First,we examine the link between overconfi-dence and two measures of innovative direction based on the Hall et al.(2001)measure of patent originality and a new measure based on self-citation rates.The results suggest that overconfidence leads to a change in direction and not just an increase in R&D spend-ing and productivity.Second,we show that the link between overconfidence and innovation is stronger for CEOs who are less constrained.Specifically,the overconfidence effect is larger when a CEO also holds the titles of chairman and president,or thefirm has greater cashflows.These twofindings strengthen our preferred interpretation of the main results by show-ing that overconfidence is more salient when a CEO has greaterflexibility to make changes in theirfirm’s strategic direction.Finally,we examine trends in var-ious innovation measures prior to hiring an overcon-fident CEO.Although the match betweenfirm and CEO remains(potentially)endogenous,these mod-els reassuringlyfind no evidence thatfirms“treated”with an overconfident executive behave differently from controls beforehand.
Because we rely on indirect measures of patent value and have no information on the opportunity cost of R&D,our results do not reveal whether inno-vations by overconfident CEOs actually create value. In
particular,these CEOs may overinvest in risky projects.Nevertheless,wefind that overconfident CEOs have greater R&D ,citations per dollar of R&D invested).Although this result is not conclusive,it strongly suggests that executive overconfidence can,under the right circumstances, provide benefits that offset the negative effects docu-mented in previous research.
1.1.Related Literature
Psychologists provide a wealth of evidence that indi-viduals overestimate their own abilities.For exam-ple,most people believe they have above average driving skills(Svenson1981)and ability to remember trivia(Moore and Cain2007).CEOs and other high-ranking executives may be particularly susceptible to this bias,because overconfidence is stronger among highly skilled individuals(Camerer and Lovallo1999) and when the link between actions and outcomes is complex(Moore and Kim2003).
Given the uncertainty and complexity associated with research and development,we might expect overconfidence to play an important role in the inno-vation process.In fact,there have been many stud-ies of entrepreneurial overconfidence(for a review, see Shane2003,p.12).Much of this research invokes overconfidence to explain persistence in the face of long odds.For instance,Astebro(2003)and Lowe and Ziedonis(2006)ask whether overconfidence is needed to rationalize entrepreneurial behavior.
Our study departs from this tradition in two important ways.First,we consider the role of overconfidence at the opposite end of thefirm-size distribution,among CEOs of large publicly traded companies.And sec-ond,instead of asking whether latent overconfidence is required to rationalize observed behavior,we exam-ine the correlation between a novel measure of over-confidence andfirm-level innovative performance. This paper contributes to an emerging literature at the intersection of industrial organization and behav-ioral economics(see Camerer and Malmendier2007 for a survey)and builds upon three broad streams of prior research.First,our data and measure of over-confidence come from Malmendier and Tate(2005a,b; 2008),who use it to study corporatefinance.Their key insight is that a CEO’s personalfinancial decisions—specifically,whether they exercise fully vested stock options that are highly in-the-money—can be used to infer beliefs about future performance.As described below,Malmendier and Tate do extensive work to validate this measure,and use it to show that over-confident CEOs are more sensitive to cashflows (Malmendier and Tate2005a)and more likely to do mergers and acquisitions(M&As)(Malmendier and Tate2008).Closer to our work is the study by Hirshleifer et al.(2010),who independently look at the correlation between options-and press-based mea-sures of overconfidence and various measures of risk taking,including patenting and stock-return volatility. We also build on a long line of research that uses patents to measure corporate innovation.Pakes and
Galasso and Simcoe:CEO Overconfidence and Innovation
Management Science57(8),pp.1469–1484,©2011INFORMS1471 Griliches(1980)were thefirst to estimate a patent pro-
duction function,and their model was extended by
and Hausman et al.(1984)and Blundell et al.(1999).
Within this literature,our work is closely related
to papers that emphasize corporate governance and
stock-based compensation,such as those by Lerner
and Wulf(2007),who study the link between innova-
tion and incentive compensation for R&D managers,
or Aghion et al.(2009),who examine the link between
institutional shareholding and innovation.
Finally,our paper adds to a small literature that
uses asymmetric beliefs to model the innovation pro-
cess.For instance,in Klepper and Thompson(2007,
2010),asymmetric beliefs about the potential of a new
technology lead to spinouts,whereby entrepreneurs
leave incumbentfirms to work on a new idea.
2.A Model of Overconfidence and
Innovation
Aghion et al.(2009)extend the Holmström(1999)
career concern model by allowing innovation to serve
as an indicator of managerial ability.This section
builds on their framework by introducing managerial
overconfidence.
Consider a CEO whose ability is ∈ 0 ¯  ,where
¯ >0,is unknown to the market(M)and to the CEO
(C).The market and CEO have different prior beliefs
about CEO ability,denoted by Pr M and Pr C respec-
tively.Specifically,we assume that
1 2=Pr M  =¯  <Pr C  =¯  =1
2
1+o
where o∈ 0 1 is a parameter that measures CEO overconfidence.Because o>0,the overconfident CE
O believes that the market underestimates his expected talent.This belief structure is common knowledge.2 The model has two periods,t=1 2 In period1, the CEO decides whether to stay the present course or try for an innovation i∈ 0 1 .One might think of this as an observable choice between taking thefirm in a new direction,which leads to a broad increase in exploration,and sticking with an established strat-egy.If the CEO does not innovate(i=0),thefirm’s second-period revenue is y0=0,and no information is revealed about the CEO’s ability.If the CEO does innovate(i=1),he incurs an innovation cost I,and the second-period revenue y1∈ 0 1 provides infor-mation about .Specifically,we assume that
E y1  =¯  =p>E y1  =0 = p
where ≡1− ,and ∈ 0 1 is an index of prod-uct market competition.Thus,high-ability CEOs are 2See Galasso(2010)for a discussion of the role of common priors in likely to succeed,and the link between CEO ability and performance is stronger when competition is more intense.The term can be interpreted as a reduced form of an unmodeled race in which a patent is awarded to the best idea in a technologyfield.The greater the degree of competition,the less likely are low-ability CEOs tofind innovations that distinguish them from competitors.3
The timing of the game is as follows:(i)the CEO chooses whether to pay I and innovate;(ii)y is realized,and the market updates its assessment of the CEO’s talent;(iii)the CEO decides whether to leave thefirm based on the comparison between his expected period2income and his outside option. Following Holmström(1982),we assume the CEO operates in a fully competitive labor market,so his second-period income if he stays with thefirm is equal to the market’s perception of his ability(condi-tional on the information acquired in period1).The CEO’s outside option is to reallocate to another sector. As in Aghion et al.(2009),we assume that CEO abil-ity is sector specific,so compensation after relocating is independent of ability and equal to w=¯ /2− , where is a switching cost.
We solve the model by backward induction.If the CEO decides to innovate,market beliefs follow Bayes’rule.Thus,the CEO’s period2income within thefirm equals
w2 y1 =¯ Pr M  =¯  y1 =¯
y
1
1+
+
1−y1  1−p
2−p− p
(1) We assume that w2 1 >w>w2 0 ,so a CEO will leave thefirm in period2if their attempted innova-tion was unsuccessful.In period1,a CEO will try to innovate if the expected net benefits E w2 y1  o −I exceed their known second-period payoff without innovation w2 y0 =¯ /2.The expected benefits of innovation are given by
E w2 y1  o =
1
2
1+o p+
1
2
1−o  p
¯
1+
+
1
2
1+o  1−p +
1
2
1−o  1− p
w
(2) Thefirst term in(2)is w2 1 times the probability of success,and the second term is the outside payoff 3Consider this simple rent-seeking game that Baye and Hoppe (2003)show is equivalent to a classic patent race model.Two play-ers H(high ability)and L(low ability)exert effort,x,sustain-ing marginal costs c H and c L with c H<c L.The probability that each player obtains the patent is x i/ x i+x j .It is possible to show that the presence of a third player with marginal cost c M∈ c H c L reduces the probability of winning the race for both players but has a stronger impact on the low-ability player than on the high-ability player.
Galasso and Simcoe:CEO Overconfidence and Innovation 1472Management Science57(8),pp.1469–1484,©2011INFORMS
w times the probability of failure.Both probabilities
reflect the CEO’s optimistic beliefs.In equilibrium,the
CEO will choose to innovate if and only if costs are
not too large,specifically,I≤ˆI≡E w2 y1  o −¯ /2.
This condition yields two testable implications.The
first prediction relates to the direct effect of CEO over-
confidence.Because
ˆI  o =
p
2
1−
¯
1+
−w
>0
innovation takes place at higher cost ,the probability of innovation is higher)when the CEO is overconfident.We can write this result as follows:
documented evidence
Implication1.Overconfident CEOs are more likely to innovate than nonoverconfident CEOs.
Second,the model suggests a link between product market competition and innovation.The cross-partial derivative
2ˆI  o  =−
p
2
¯
1+
−w
−
p
2
¯  1−
1+  2
<0
and the fact that =1− imply that overconfidence
and competition are complements in the CEO’s deci-sion to innovate,or formally, 2ˆI/ o  >0. Implication2.The impact of CEO overconfidence on innovation increases with the level of product market
competition.
Intuitively,competition reduces the chance of suc-
cess and hence the expected benefits of innovation,
but less so for overconfident CEOs who believe
they have high ability.4More generally,Implication2
shows that industry characteristics(in this case,the
level of competition)may amplify or mute the impact
of managerial biases.
2.1.Discussion
This section discusses several of our simplifying
assumptions.First,we assume that CEOs can influ-
ence corporate innovation.Though we do not empha-
size specific mechanisms,a CEO might influence
patent-based innovation measures in several ways.
For example,CEOs could change the compensation
schemes of R&D executives,which can significantly
alter innovation output,as documented in Lerner and
4Increasing has two effects:thefirst-order impact is a reduced probability of successful innovation,which outweighs a second-order increase in the CEO’s expected payoff conditional on success. The size of thefirst effect declines with overconfidence(and in the limit where o=1,vanishes completely),whereas the second effect is not influenced by overconfidence.Thefirst effect is consistent with the results of Carmer and Lovallo(1999)showing that overconfi-dent subjects discount the negative threat created by an increase in the level of competition.Wulf(2007).CEOs with strong beliefs about inno-vation may attract employees with similar beliefs through labour market sorting,as in Van den Steen (2005).CEOs could also instigate a broad shift in the technology strategy.For instance,Lou Gerstner, the CEO of IBM from1993to2002,generated a large increase in patenting with three strategic moves(Kanellos and Spooner2002).First,Gerstner expanded IBM’s research from a narrow focus on hardware to exploration in areas such as software and supercomputers.Second,he placed greaterfis-cal responsibility on the R&D department by setting precise goals,cutting waste,and rewarding succes
s-ful innovators.And third,Gerstner increased efforts to exploit the company’s patents,generating a sub-stantial increase in licensing revenue.
In our stylized model,the players’beliefs should converge over longer time horizons,thereby weaken-ing the link between overconfidence and innovation. However,Yildiz(2004)shows that the conditions for complete learning become extremely onerous when the signal space—in our model,the binary variable y1—grows more complex.In practice,CEOs can send a large variety of signals to the market,making it rea-sonable to assume that beliefs about ability need not converge during a typical CEOs tenure.5
We consider several extensions to our baseline model in the online appendix to this paper(avail-able at people.bu.edu/tsimcoe/documents/ published/Overconfidence-Final.pdf).First,building on Malmendier and Tate(2005),we add a budget constraint to the model and assume that a CEO mustfinance innovation efforts using either internal funds(cash)or externalfinance(debt).This extended model shows that innovation by overconfident CEOs is more sensitive to cashflow than innovation by non-overconfident CEOs.Intuitively,because overcon-fident CEOs accept a larger range of projects,their budget constraints are more likely to be binding. Therefore,a marginal increase in cashflow tends to have a greater impact on the investment decisions of biased CEOs.A second set of extensions show that our results do not depend on specific assumptions about how to model the CEO’
s outside ,we consider non-sector-specific ability,as in Aghion et al. (2009),and also drop the switching cost ).Finally, we generalize our setup and show how our main results still hold under alternative assumptions about the interplay between competition and innovation. Of course,our stylized career concerns model omits many factors that might influence either afirm’s 5One obvious way to expand the signal space of our model is to assume that the CEO’s initial beliefs are unknown to the market. This leads to a two-dimensional career concern model,similar to the one studied by Koszegi and Li(2008).Their results suggest that overconfidence may increase innovation in this more complicated setting.
Galasso and Simcoe:CEO Overconfidence and Innovation
Management Science57(8),pp.1469–1484,©2011INFORMS1473
innovation strategy or the market’s assessment of CEO ability.Because CEOs may respond to all of these forces,we now turn to the data and use a novel mea-sure of CEO overconfidence to evaluate the main pre-dictions of the model.
3.Data and Methods
We begin with a panel of450large publicly traded U.S.firms between1980and1994.These data are described by Hall and Liebman(1998)and Malmendier and Tate (2008).Eachfirm in the sample appeared at least four times on a Forbes magazine list of the largest U.S-panies.These data provide a very detailed picture of CEOs’stock-option holdings,which Malmendier and Tate(2008)use to construct the measure of CEO over-confidence described below.
We use the Compustatfirm identifier(GVKEY)to merge this panel of large publicly tradedfirms to the National Bureau of Economic Research(NBER)U.S. patent datafile.The NBER patent data are described by Hall et al.(2001)and provide detailed information on all U.S.patents during our sample period,includ-ing application and grant years,citations to other patents,and assignee codes that can be used to iden-tify the owners.To match U.S.patent assignee codes with Compustatfirms,we started with the name-matching tool of Bessen(2009)and then searched by hand for variations on the names in our panel. After droppingfirms in thefinance,insurance,and real estate(FIRE)sector(one-digit Standard Industrial Classification(SIC)code6),which has a very low rate of patenting,we arrive at an estimation sample with 290firms,3,648firm-years,and627individual CEOs, covering the period1980to1994.6Table1provides summary statistics for this sample.
Our primary measure of innovation is a citation-weighted count of U.S.patents.Patent citations iden-tify
prior knowledge upon which a patent builds and delimit the scope of the property rights awarded to the inventor.A patent examiner who is an expert in the technology area is responsible for insuring that all appropriate patents have been cited.Because of this important legal function of patent citations,the eco-nomics of innovation literature has often employed the number of forward citations received by a patent as an indirect measure of patent value(for example, Pakes and Griliches1980,Hall et al.2005,Harhoff et al.1999,Aghion et al.2009).7Because citation counts are inherently truncated,we weight each 6Retainingfirms from the FIRE sector does not change the main results.
7A number of studies have been conducted to validate the use of patent citations.Harhoff et al.(1999)show that the price at which a patentee was willing to sell a patent is highly correlated with the citations received by the patent.Highly cited patents have been Table1Summary Statistics
Mean Median Min Max SD Obs. Total Citations489 016 000 0032,5091,7473 648 Total Patents27 791 000 001,22181 293 648 Citations per Patent8 624 000 0024013 323 648 log(R&D Expense)3 803 920 008 731 941 864 Overconfidence0 581 000 001 000 492 441 Holder670 490 000 001 000 501 533 Lerner Index0 110 090 030 220 053 648 CEO-Chairman0 380 000 001 000 493 640 log(Cash Flow)5 315 33−5 4513 921 513 624 log(Sales)7 857 752 9511 811 123 641 log(Employees)2 682 72−2 236 781 293 627 log(Capital/Labor)4 294 010 097 471 353 637 Stock Ownership0 020 000 000 950 073 6
48 Vested Options0 260 000 00786 0013 103 648 Totalfirms290
Total CEOs627
Overconfident168
Not overconfident136
Unclassified323
Notes.Holder67is a dummy equal to1for all CEO years after the CEO fails to exercise an option67%in-the-money withfive years remaining duration. Overconfidence is the maximum value for Holder67for a given CEO.Lerner Index is the median gross profit margin of all Compustatfirms in a two-digit SIC code.Cash Flow equals Compustat earnings before extraordinary items (item18)plus depreciation(item14).CEO-Chairman is a dummy equal to1 if a CEO also holds the titles of chairman and president.
patent by the truncation-adjusted citation countfield contained in the NBER patent datafile(see Hall et al. 2001for details).
We match patents tofirm-year observations accord-ing to theirfiling date,which approximates the date of invention,because the patent system provides incen-tives tofile quickly.Becausefiling dates may antic-ipate a CEO’s ultimate payoff from innovation,we view this as a reasonable choice given the evidence that lags from application to grant are short on aver-age(1.89years in our data),citations peak roughly two years after a patent is granted(Mehta et al.2010),and Tobin’s q is correlated with the future citations of a firm’s current patent stock(Hall et al.2005),suggest-ing that citations are a lagging indicator of value.8 Ideally,we would like to identify patents that are attributable the actions of a particular CEO.Unfor-tunately,we have no information on when specific research investments were initiated,and we know of no research that tries to measure the average time lag between starting a research project and issuance of the shown to be more frequently litigated(Lanjouw and Schankerman, 2001),to be more frequently traded amongfirms(Serrano2010), and to boost market value(Hall et al.2005).
8Although it may take a long time to for a patented innovation to be embedded into products and to generate profits,there is evidence that markets react very quickly to patent awards(Hall et al.2005).
Galasso and Simcoe:CEO Overconfidence and Innovation 1474Management Science57(8),pp.1469–1484,©2011INFORMS
first related patent.9Moreover,if R&D investments are staged,it may be impossible to partition“respon-sibility”for a given patent between CEOs who make complementary investment decisions at differ-ent stages of the research process.Nevertheless,it is reasonable to think that a new CEO might quickly influence research at all stages of the development process,through both investment decisions and by changing thefirm’s strategic priorities.And a virtue of linking patents to CEOs based on the patent’sfil-ing date is that we know the overconfident CEO was responsible for thefinal decision to incur the applica-tion costs.
We also consider several additional innovation metrics.First,we decompose our primary citation-weghted patent measure into an unweighted patent count and the average number of citations per patent (excluding self-citations).Second,we use the research and development expenditures(Compustat item46) as a measure of innovation inputs.Becausefirms are not required to account for their R&D expenditures, this variable has many missing values,even after we interpolate over any gaps of three years or less. Finally,in a series of extensions,we examine changes in ,dispersion of citations across tech-nology areas)and in the share of self-citations.Table1 shows that the distribution of innovative activity in our sample is highly skewed.Whereas the median firm-year observation consists of a single patent that receives six citations,the sample mean is much higher, at28patents and489citations.
To measure competition,we use a Lerner index, as in Aghion et al.(2009).Specifically,we calculate the median gross margin of allfirms in the Compu-stat database with the same two-digit SIC code as a focalfirm.Our baseline model allows this competition measure to vary over time.However,we also consider robustness tests that use a time-invariant Lerner index or a dummy forfirms whose average gross margin over the entire sample period falls above the median of allfirms in the estimation sample.
In our regressions,we condition on size and the capital–labor ratio,as suggested by literature on patent production functions(see inter alia Aghion et al.2009,Hall and Ziedonis2001).At the CEO level,we control for age and tenure(and their squared terms)to capture experience and career con-cern incentives of the top executives.We also include 9We expect that in some ,software)the time from initial idea to patent application may be quite short,and in others (e.g.,pharmaceuticals)quite long.However,such measurements would be quite difficult if the research process was cumulative and chaotic,with overlapping projects and opportunistic patent appli-cations,rather than an orderly“linear”process that proceeds from investment l variables for the effects of CEO stock owner-ship and options holding.Our main Compustat items are sales(item1);a capital–labor ratio constructed from the book value of total assets(item6)and the number of employees(item29);and a deflated R&D stock.To construct the R&D stock,we follow the method described by Hall(1990),depre
ciating all reported R&D activity at a rate of15%over a10-year period.As in Malmedier and Tate(2005a,b;2008), we construct a measure of cashflow adding Com-pustat earnings before extraordinary items(item18) and depreciation(item14).We also have several CEO-level control variables used in Malmendier and Tate (2008),including age,job tenure,and a set of dum-mies categorizing their educational background as finance or technical.CEOs with a“finance”back-ground received a degree in accounting,finance,busi-ness(including MBA),or economics.CEOs with a “technical”background received a degree in engi-neering,physics,chemistry,mathematics,operations research,biology,or applied sciences.
3.1.Measuring Overconfidence
Our measures of CEO overconfidence build on a series of papers by Malmendier and Tate(2005a,b;2008). These papers use CEOs’personal investment decisions to construct a proxy for overconfidence or systematic overestimation of the returns to holding stock in their ownfirm.The key idea behind this measurement strat-egy is to focus on the decision to exercise executive stock options.These options give the holder a right to purchase stock in their own company,usually at the prevailing price on the date of the option grant.They typically have a10-year life,and are fully exercisable after a4-year vesting period.At exercise,the shares are almost always immediately sold.
Although investors may hold ordinary options because the right to delay a stock purchase has positive value,executive stock options have sev-eral unique features that create strong incentives for exercise,so long as they are fully vested(and in-the-money).In particular,executive stock options are non-tradable,and CEOs cannot legally hedge their risk by shortselling shares in their ownfirm.Moreover, most CEOs are highly exposed to idiosyncratic risk associated with their ownfirm through equity com-pensation,stock holdings,andfirm-specific human capital.Consequently,standard models of decision making under ,Hall and Murphy 2002)indicate that a risk-averse CEO should exer-cise vested executive options before expiration as long as the stock price is sufficiently high.Nevertheless, many of the CEOs in our sample fail to exercise their executive options,often repeatedly.Malmendier and Tate(2005a,b;2008)use this behavior as an indicator of CEO overconfidence,or systematic overestimation of expected returns from holding the stock.

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。