C.E.O. characteristics and corporate risk-taking: evidence from emerging markets

Abstract Although the Upper Echelon Theory predicts that C.E.O.s play a relevant role in corporate risk-taking, the C.E.O.s’ traits that can be associated with such risk are not well-explored. Our study fills this gap and shows the effect of C.E.O.s’ characteristics on corporate risk-taking of a hand-collected sample of 369 Latin American listed firms. We study six traits: C.E.O.s’ age, tenure, gender, duality (i.e., holding concurrent Chairman and C.E.O. roles), educational background, and career horizon. We find that age increases risk-taking. However, when the C.E.O.'s age reaches a given point, their concern about reputation and retirement results in a negative relationship. We also find that as C.E.O. tenure increases, corporate risk begins to decrease. Nevertheless, there comes a point at which the C.E.O. uses their knowledge and their overconfidence to make risky financial decisions. Female C.E.O.s are negatively related to risk-taking, while C.E.O. duality, C.E.O. educational background, foreign C.E.O.s, and a C.E.O.'s career horizon have the opposite effect. Our study is novel because of the focus on emerging markets and because of the use of different market-based measures of risk-taking. We provide policymakers, investors, and practitioners with fresh evidence about how C.E.O.s’ risk aversion shapes the firm’s risk-taking behaviour.


Introduction
The C.E.O.bears the highest responsibility within the company and is one of its most critical resources.Traditional theoretical approaches consider C.E.O.s to be homogeneous individuals whose personal characteristics are irrelevant, such that corporate decisions are only explained by firm-or industry-level factors (Bertrand & Schoar, 2003;Huang & Kisgen, 2013).However, the Upper Echelons Theory points out that C.E.O.s' experience, values and other attributes play a key role in their decisions and shape organisational outcomes (Hambrick & Mason, 1984).From this point of view, the company's decision-making process can be explained by a C.E.O.'s observable characteristics, such as age, tenure, or duality, as well as their psychological behavioural traits (Graham et al., 2013;Nguyen et al., 2018).Nevertheless, the impact of a C.E.O.'s profile on company outcomes remains an ongoing subject of inquiry and as yet there is no conclusive evidence as to how a C.E.O.'s characteristics impact organisational outcomes.
A growing stream of literature is focusing on the importance of C.E.O.profile in corporate risk-taking.Specifically, several authors indicate that the C.E.O.'s characteristics can affect corporate risk through financial decisions, such as R&D expenditures, diversification, financial leverage, and acquisition, among others (Matta & Beamish, 2008;Serfling, 2014).Yet despite this, the relationship between C.E.O.profile and corporate risk-taking remains relatively unexplored, and the evidence is still unclear.This is due to four main reasons.First, most studies on demographic characteristics usually employ C.E.O.age and C.E.O.tenure measures indistinctly.While C.E.O.age and C.E.O.tenure are highly correlated, the two metrics do not necessarily reflect the same C.E.O.characteristic.Second, the mixed evidence can be explained by the disagreement concerning what is the appropriate measure of corporate risk to be applied.Third, the literature has neglected other C.E.O.features that may have significant explanatory power, such as C.E.O.duality, gender, foreignness, and educational background, among others.Finally, only a few studies address the relationship between C.E.O.s' observable characteristics and corporate risk-taking outside the context of the U.S.A.In this sense, there is a lack of studies on emerging economies.
Considering the above-mentioned gaps, this study aims to empirically test the impact of C.E.O.profile on the firm's risk-taking behaviour.Accordingly, we attempt to address the following research questions: How are the demographic characteristics of C.E.O.sage and tenurerelated to corporate risk-taking?Are there other demographic, cultural or acquired traits (such as gender, power, educational background or foreignness) that are relevant to risk-taking?
To answer these questions, we follow a threefold method.First, we define a set of C.E.O.profile variables to measure C.E.O.risk behaviour, and we also determine the appropriate risk-taking measures.Second, we construct a panel data by matching time-series with cross-section data to deal with unobservable heterogeneity.We also control for different firm-level characteristics that might impact the firm's riskiness.Third, we test the potential explanatory power of a comprehensive set of personal C.E.O.traits on corporate risk-taking by using the two-step Generalised Method of Moments (G.M.M.) to address the endogeneity problem and reverse causality.
We report enlightening results using a sample of 369 Latin American companies between 2005 and 2020.Our initial findings indicate that the relationship between C.E.O.age and risk-taking is non-linear and displays an inverted U-shape.Hence, C.E.O.age at first positively impacts the level of risk-taking.However, the relationship becomes negative when the C.E.O.turns about 45 years old.This indicates that ageing generates a better ability for younger C.E.O.s to learn and adapt quickly to complex environments.For older C.E.O.s, ageing causes a greater preference for the status quo.The second set of results shows that the relationship between C.E.O.tenure and risk-taking is also non-linear but U-shaped.In other words, as tenure increases, the C.E.O.becomes more entrenched and avoids risky decisions.Nevertheless, when C.E.O.tenure reaches around 17 years, the benefits of long tenuresuch as greater trust with other stockholders or experienceoutweigh the costs, and the risk-taking increases.We also test whether C.E.O.gender influences risk-taking.Our results suggest that greater female conservatism leads to lower corporate risk in female-led firms.In contrast, we find that the concentration of power in the hands of a single executive -C.E.O.dualityand a C.E.O.'s education decrease risk aversion.Interestingly, C.E.O.s with a master degree in business sciences increase corporate risk.We also find that foreign C.E.O.s are positively associated with corporate risk-taking.Additional results indicate that the C.E.O.'s career horizon adjusted by the industry, and the uncertainty avoidance of the C.E.O.'s country of nationalityaffect risky financial decisions.
We contribute to the extant literature in several aspects.First, we contribute to the literature addressing the link between a C.E.O.'s characteristics and firm-level issues.Our results align with the Upper Echelon Theory proposed by Hambrick and Mason (1984) and show that the C.E.O.'s observable characteristics have explanatory power on corporate risk-taking.Second, to the best of our knowledge, this is the first work in Latin America to study the impact of C.E.O.s' observable attributes on risk-taking.Latin American countries are characterised by a high concentration of ownershipusually in the hands of familiesand weak investor protection (Gallego & Larrain, 2012;Mellado & Saona, 2020;Santiago-Castro & Brown, 2011).Nevertheless, in terms of corporate governance, there have been significant changes in legislation and regulation to protect investors (Gait an et al., 2018).Previously, Briano-Turrent et al. (2020) reported that long-tenured C.E.O.s pay less dividends, while Gallego and Larrain (2012) studied the impact of C.E.O.s' characteristics on compensation premium.However, none were able to conclude the relationship between C.E.O.profile and risk-taking behaviour.Third, we disentangle the effect of two closely related C.E.O.traits: age and tenure.Longer tenure has usually been associated with older age.However, anecdotal evidence confirms that this situation does not hold for all C.E.O.s.For instance, Mauricio Varela, C.E.O. of the Chilean company Socovesa, had been in that position for over 10 years at the end of 2013, and he was only 40 years old, while the C.E.O. of Cementos B ıo B ıo, had been in the position for only four years at the end of 2008 and yet was over 64 years old.Thus, we identify experience in the company with longer C.E.O.tenure rather than with being an older C.E.O.This is consistent with Cline and Yore (2016), who find that the market undervalues companies with old C.E.O.s but rewards those with long-tenured C.E.O.s.Following this view, we report that C.E.O.age and C.E.O.tenure affect risk-taking in opposite directions.Finally, as a fourth contribution, we use different market-based measures of risk-taking that cannot be managed directly by the C.E.O. but which reflect the market perception of the C.E.O.'s financial decisions, business strategy, and risk-taking behaviour.
The article proceeds as follows.Section 2 discusses the literature and proposes the main hypotheses; Section 3 describes the research model, statistical methodology, sample, and variables; Section 4 presents our empirical results and the discussion; and Section 5 provides the conclusion, contributions, implications, and suggestions for future research.

Literature review and hypotheses
Corporate risk-taking is likely one of the firm-level issues to be most affected by managerial traits.Prior research has identified three main characteristics of managers through which the C.E.O.'s traits can affect corporate risk-taking: the C.E.O.'s social capital, the C.E.O.'s cultural and institutional environment, and the C.E.O.'s demographic characteristics.
As far as the C.E.O.'s social capital is concerned, Ferris et al. (2017Ferris et al. ( , 2019) ) show a positive association between C.E.O.social capital and aggregate corporate risk-taking because social capital alters the risk tolerance of the most connected agents, reinforces the individual sense of power, and enhances opportunities in the labour market.
Another stream of research has explored the cultural and institutional environment in which C.E.O.s take their decisions.For example, Baxamusa and Jalal (2016), D ıez-Esteban et al. (2019), and Jiang et al. (2015) highlight the role played by religious backgrounds in corporate risk-taking.As regards cultural environment, some authors have reported a relationship between risk-taking and a number of cultural dimensions such as power distance, masculinity, individualism, uncertainty avoidance, harmony, and long-term orientation (D ıez-Esteban et al., 2019;Li et al., 2013) or the legal protection of investors rights (Acharya et al., 2011;Boubakri et al., 2013).
In this article, we focus on the relationship between C.E.O.s' demographic characteristics and corporate risk-taking.More specifically, we study the influence of seven traits: age, tenure, gender, power, educational background, foreignness, and career horizon.

C.E.O. age
The literature provides evidence that ageing affects C.E.O.incentives as well as their physiological and psychological characteristics (Croci et al., 2017).Younger C.E.O.s (as opposed to older C.E.O.s) prefer riskier financial decisions to prove their skills (Serfling, 2014).Zhang et al. (2016) called this phenomenon the 'signaling-incentive effect', in which younger C.E.O.s act aggressively and make risky but profitable financial decisions in order to build their reputation.Similarly, younger C.E.O.s possess cognitive skills that can better adapt to strategic changes (Cline & Yore, 2016).Thus, young C.E.O.s acquire better cognitive skills as they get older.The relationship between C.E.O.age and corporate risk-taking could therefore be positive when a young C.E.O.runs the company.However, there comes a time when ageing is associated with less energy and cognitive abilities (Kitchell, 2009).As a result, older C.E.O.s prefer the stability of future returns.Kim et al. (2016) and Yim (2013) report that older C.E.O.s increase their desire for the status quo in their life and reduce R&D expenditures because they are likely to fall short of benefiting from the success of these investments.
From a psychological point of view, older people are resistant to complex problems because ageing generates negative changes in their cognitive capacity (Mata et al., 2011).In this sense, older C.E.O.s increase conflicts of interest with the rest of the stakeholders by disregarding value-enhancing risky projects.Risk-taking partly explains the misalignment of interests between the C.E.O. and shareholders that both parties are willing to assume.Shareholders desire risky projects with high potential for growth opportunities, while older C.E.O.s, unlike shareholders, cannot diversify their risk so easily.
In summary, assuming more risk generates different incentives between older and younger C.E.O.s.For younger C.E.O.s, the benefits of taking on more risk outweigh the costs, while the opposite occurs with older C.E.O.s.Thus, younger C.E.O.s gain more experience over the years, which translates into a greater ability to manage risky projects.However, there comes a time when managers become less risk-tolerant.Taking this into account, our first hypothesis postulates that: Hypothesis 1: C.E.O.age has an inverse U-shaped relationship with risk-taking.

C.E.O. tenure
The literature has failed to report any conclusive results regarding how C.E.O.tenure explains firms' risk behaviour.On the one hand, short-tenured C.E.O.s are concerned about their reputation, career, and about understanding the company as well as the industry (Walters et al., 2007).These short-tenured C.E.O.s need the supervision and guidance of the board of directors or top management team to enhance their skills and to understand how to interact with internal stakeholders (Shen, 2003).They also have little power in the company and are afraid of being misunderstood by the company or the labour market.On the other hand, McClelland et al. (2012) point out that as the years pass, C.E.O.s worry about their working career due to the fear that their specific human capital will not adapt to the other companies in the market.Moreover, long-tenured C.E.O.s have a different agenda than those who are still building and maintaining their careers.The former focus on projects with short-term payoffs rather than risky projects with long-term growth potential.
However, time can also result in an alignment of interests between C.E.O. and stakeholders.Long-tenured C.E.O.s have already earned the trust of the board and other influential stakeholders, so they can better maintain business stability in the midst of risky decisions.The match theory points out that long C.E.O.tenure is due to the fact that their skills and knowledge match the characteristics and needs of the company (Allgood & Farrell, 2003).In contrast, traditional agency theory indicates that C.E.O.s can take advantage of their power, entrench themselves in their position and pursue their interests over the firm's interests.
On the other hand, in emerging economies such as Latin Americacharacterised by a high concentration of ownership and shareholder representatives on the boardthere is a low separation between management and control and a low probability of managerial opportunism (Pombo & Taborda, 2017).While long-tenured C.E.O.s do not risk their career and job stability, they accumulate firm-and industry-specific knowledge and develop specific human capital, which improves the firm's performance (Nguyen et al., 2018).In this sense, the risk behaviour of long-tenured C.E.O.s has been shaped by failure-based learning, and they evaluate risky decisions more objectively and systematically (Simsek, 2007).In addition, companies with long-tenured C.E.O.s have a better alignment of interests between senior management and company culture, which translates into organisational outcomes (Hartnell et al., 2016).
We thus anticipate that as tenure increases, C.E.O.s take fewer risks.However, once they have earned the trust of the rest of the powerful internal stakeholders, they are able to use their experience and knowledge to manage risky decisions.In recent years, some studies have reported that the heterogeneous composition of senior management in terms of gender diversity has a positive impact on firm performance.Khan and Vieito (2013) find that companies led by female C.E.O.s display superior performance but lower corporate risk.Controversially, the literature has stereotyped female C.E.O.s as being conservative and risk-averse individuals (La Rocca et al., 2020).Faccio et al. (2016) and Huang and Kisgen (2013) report that firms run by female C.E.O.s take less risky corporate finance decisions because female C.E.O.s are more risk-averse, less overconfident, and are associated with a higher risk of unemployment.In the same vein, Sah et al. (2022) and Xu et al. (2019) point out that female senior management maintain higher levels of cash holdings for investing in financial decisions that reduce the firm's risk.This view is confirmed by Palvia et al. (2015), who find that female C.E.O.s and chairwomen act conservatively in periods of financial stress, and by Adhikari et al. (2019), who report that female-led companies avoid risky strategies with positive N.P.V. but high litigation probability, such as aggressive R&D investments or intensive advertising.Similar results were also obtained by Elsaid and Ursel (2011), according to whom a change from a male C.E.O. to a female C.E.O. is linked to reduced R&D spending, lower volatility in cash flows, and increased cash holdings.Thus, our third hypothesis is as follows: Hypothesis 3: Female C.E.O.s have a negative relationship with risk-taking.

C.E.O. duality
C.E.O.duality has been widely studied from different perspectives.From an agency theory point of view, Adams et al. (2005) point out that C.E.O.duality generates a concentration of power in a single executive, which allows them to influence the company's decision-making without significant objections from other internal stakeholders.Kim et al. (2009) indicate that C.E.O.duality decreases board vigilance, increases managerial opportunism, and reduces the firm's risk through unrelated diversification.In this way, C.E.O.s maintain the status quo in the firm, preserve their job stability and prestige, and obtain personal benefits at the expense of shareholders.In summary, agency theory supports that these C.E.O.s use their power and discretion in the company's management to avoid risky decisions and to entrench themselves in their position.
In contrast, according to stewardship theory, C.E.O.duality fosters energetic and unified leadership.Peng et al. (2007) note that C.E.O.duality incentivises an efficient leadership unit that streamlines decision-making and has a positive impact on the performance of firms within a dynamic environment.Nguyen et al. (2018) report that C.E.O.duality reduces board conflicts of interest and allows them to respond quickly to high-growth opportunities in complex environments.Similarly, Peni (2014) finds a positive relationship between C.E.O.duality and firm performance.Finally, for a list of Latin American companies, Gait an et al. ( 2018) find no evidence that C.E.O.duality negatively impacts firm outcomes.
Since there is no consensus as to which theory is best suited to emerging economies such as Latin America, we propose two hypotheses: Hypothesis 4a: C.E.O.duality has a negative relationship with risk-taking.
Hypothesis 4b: C.E.O.duality has a positive relationship with risk-taking.et al. (2020) point out that C.E.O.s' educational background promotes efficient financial and investment decision-making.C.E.O.s with a higher level of education are usually more creative, react faster to innovative ideas, and prefer risky and aggressive strategies (Bertrand & Schoar, 2003;Zhou et al., 2021).In a similar vein, C.E.O.s with an M.B.A. exhibit overconfidence and higher risk tolerance (Beber & Fabbri, 2012).Farag and Mallin (2018) find a positive relationship between a C.E.O.'s education and risk-taking measures because C.E.O.s with postgraduate degrees are better informed about new trends and are more likely to pursue innovative ideas.Therefore, highly educated C.E.O.s know to develop risky and more profitable projects (King et al., 2016).

Naseem
We therefore suggest that the firm's financial decisions depend on the C.E.O.'s background.Specifically, we posit that C.E.O.s with a business science master's degree possess more tools and skills to manage risk and take riskier decisions.
Hypothesis 5: Being a C.E.O. with a business science background has a positive relationship with risk-taking.
2.6.Foreign C.E.O.s Shaw (1990) points out that the C.E.O.'s nationality influences their cognitive ability and, thus, decision-making.Nielsen and Nielsen (2011) stress that diversity of nationality within senior management encourages constructive debate and improves the ability to assess risky projects.In fact, Graham et al. (2013) report that U.S. C.E.O.s exhibit different personality traits and attitudessuch as level of optimismto non-U.S.C.E.O.s.Foreign C.E.O.s are more likely to lead companies in risky industries (Hoang et al., 2019).Additionally, foreign C.E.O.s have a larger network of international contacts (Fang et al., 2018) and specific knowledge of their native country's economy (Conyon et al., 2019).
We therefore anticipate that foreign C.E.O.s have better abilities and resources to handle risky projects.Our sixth hypothesis thus reads as follows: Hypothesis 6: Being a foreign C.E.O. has a positive relationship with risk-taking.

C.E.O.s' career horizon
Retirement is the last step of the C.E.O.'s career, in which their track record is judged.In this sense, one common problem is the horizon problem, where the C.E.O.pursues short-term decisions as they approach retirement (Davidson et al., 2007).As a general rule, the further away a C.E.O. is from retirement, the longer their career horizon.C.E.O.s with short career horizons are more likely to display myopic risk aversion because they reject risky projects with expected cash flows after retirement (Aktas et al., 2021).In other words, as they approach retirement, C.E.O.s avoid risky decisions that might jeopardise their reputation or ruin the perception of their recent career (Matta & Beamish, 2008).The C.E.O.'s career horizon therefore influences their decision-making process.For the Italian market, both Wade et al. (2006) and Martino et al. (2020) provide empirical support for the fact that C.E.O.s with longer career horizons make faster and riskier decisions because of their risk preferences.Similar results for U.S. firms have been reported by McClelland et al. (2012).As a result, our seventh hypothesis postulates that: Hypothesis 7: C.E.O.'s career horizon has a positive relationship with risk-taking.

Research model
We capture the impact of C.E.O.profile on corporate risk-taking by using a list of the C.E.O.'s observable characteristics.C.E.O.s' risk aversion is a subjective measure and, in turn, is usually unknown or hard to measure correctly.In contrast, observable characteristicssuch as age, tenure, gender, power, educational background, foreignness, and career horizonare more accessible.These variables thus allow us to proxy the C.E.O.'s risk preference.Figure 1 presents the hypotheses and the expected signs.
We have a panel data set with cross-sectional and longitudinal information, such that we control for unobservable heterogeneity that could affect corporate risk-taking (Baltagi, 2013).We propose Equation (1) to identify the impact of a C.E.O.'s characteristics on corporate risk: where Firm risk it are the two different measures of risk-raking for firm i at time t : Total Risk (T.R.) and Idiosyncratic Risk (I.R.  C.A.P.E.X.We also include a set of fixed effects to control for unobservable industry-specific effects (c i ), unobservable country-fixed effects (g i Þ, and time-variant fixed effects (l t ).We use industry sector dummies based on the 2-digit NAICS Code to control for industry-fixed effects.Finally, e it is the stochastic error in the estimations.
To deal with the endogeneity problem and reverse causality, we use the two-step system G.M.M. proposed by Arellano and Bover (1995) and Blundell and Bond (1998).This econometric technique provides more efficient and consistent estimates than ordinary least squares (O.L.S.) or fixed effects.We run Equation (1) using the two-step system G.M.M. to deal with the potentially endogenous issues of all right-hand-side variables.As instruments we use all the independent variables lagged from t-1 to t-2.To test the validity of the instruments, we run the AR(2) and the Hansen post-estimation test.AR(2) measures the absence of second-order serial autocorrelation in the residuals.We use the Hansen test to test overidentifying restrictions and to assess whether the instruments are exogenously determined.

Sample
Our sample includes 369 companies from six Latin American countries: Argentina, Brazil, Chile, Colombia, Mexico, and Peru.According to the World Bank website, the market capitalisation of listed domestic firms in these countries represents 98% of the whole Latin America region.The period of analysis spans from 2005 to 2020.We exclude financial and non-domestic firms and companies with negative equity because they are subject to specific requirements and accounting standards.To deal with survivorship bias, we include active and non-active firms.We required at least four continuous years to be included in the final sample in order to obtain efficient estimates.We collect data from different sources.We use Refinitiv Eikon to obtain financial and accounting information.C.E.O.s' observable characteristics were hand-collected from annual reports, corporate websites, financial databases, LinkedIn, and business press websites.
The final panel data consists of 3,949 firm-year observations and an average of 10.70 observations per firm.Panel A of Table 1 reports the panel composition.Brazil has the most significant participation in the sample, with 40.62%.Mexico and Chile account for 18.23% and 17.40%, respectively, while Argentina represents 11.57% of observations.The countries with the fewest observations are Peru and Colombia, with 7.42% and 4.76%, respectively.Panel B of Table 1 shows the composition of the sample by country-year.Table 2 reports the distribution of firms by industrial sectors of each country.As can be seen, the largest industrial sector is manufacturing, with 40.92% of the total sample.

Variables
The literature has used different measures to capture the risk of corporate decisions.Whereas some prior studies proxy corporate risk through operating return volatility or some risky investment policies, such as R&D or firm diversification (Faccio et al., 2016;Yim, 2013;Zhang et al., 2016), the dominant stream of research has used market-based measures with stock price data (Aktas et al., 2021;Ferreira & Laux, 2007;Nguyen, 2011;Peltom€ aki et al., 2021;Serfling, 2014).Consequently, and following this line, we use stock return volatility and I.R. as measures of risk-taking.This choice is consistent with the literature on emerging economies (Farag & Mallin, 2018;Sayari & Marcum, 2018).To deal with the common problem of low liquidity, we remove stocks that do not trade on at least 80% of business days (Figlioli & Lima, 2019;Leite et al., 2018). 1 We chose the most liquid series for companies with more than one stock series.T.R. is calculated as the annualised standard deviation of daily stock returns over the last year.Daily returns (r i, t ) are measured as ðP it À P itÀ1 Þ=P itÀ1 , where P it is the stock price for firm i on day t: All prices are denominated in U.S. dollars, and returns are adjusted for dividends and stock splits.I.R. is calculated as the annualised standard deviation of residuals from the Market Model over the last year.We estimated this model as where R it is the stock return for firm i for period t, R mt is the market portfolio return for period t, a i is the constant term, b i is a measure of stock The number of observations is 3,949.Source: Authors.
price sensitivity for firm i to movements in the market, and e it is the error term.As do Figlioli and Lima (2019) and Gonz alez-S anchez (2022) we use the main equity market index as a market portfolio: Argentina (Merval), Brazil (Ibovespa), Chile (IPSA), Colombia (IGBC), Mexico (IPC), and Peru (S&P/BVL Peru General).All C.E.O.characteristic variables have been obtained through a systematic process.First of all, we obtain the C.E.O.'s name from annual reports for each company-year.We then use annual reports, corporate press releases, or corporate websites to determine the C.E.O.'s age, tenure, gender, duality, educational background, and nationality.In cases where the C.E.O. has previously worked at the company in the same position, we measure tenure as the cumulative time in the position.If it is impossible to obtain that information, we turn to Refinitiv Eikon or the Bloomberg website, or we search on business press websites.In this case, we need three different websites that confirm the  2001) Uncertainty Avoidance Index (U.A.I.) to measure the C.E.O.'s country risk aversion.On a scale of 1-100, the U.A.I. measures the society's tolerance of uncertainty and resistance to the unknown.
We include a series of control variables that could influence corporate risk.R.O.A. is the ratio of net income to total assets.We propose a negative relationship, since non-profitable companies have incentives to undertake riskier projects in order to make up for low profitability (Martino et al., 2020).Size is the natural logarithm of total assets.Larger companies have more resources and greater access to better sources of financing to diversify into unrelated sectors, such that we expect a negative relationship with corporate risk-taking (Anderson & Reeb, 2003).L.E.V. is the ratio of total debt to total assets.Boubaker et al. (2016) state that financial leverage is associated with higher volatility and, therefore, a higher level of risk.Age is the natural logarithm of years after the foundation of the company.We expect a negative relationship, since older companies invest more in stable and conservative projects than younger companies (Anderson et al., 2012).Finally, C:A:P:E:X: is the ratio of capital expenditures to total assets.C.A.P.E.X. is usually associated with investments with low volatility.Therefore, we propose a negative relationship.Table 3 summarises the variables included in the estimates.All continuous variables are winsorised at 1% in both tails to mitigate the influence of outliers.

Results
Table 4 presents the mean, standard deviation, minimum and maximum, as well as the 25th, 50th, and 75th percentiles.Our results are comparable with previous evidence for Latin America.Average T:R: is equal to 0.384similar to Poletti-Hughes and Briano-Turrent ( 2019 , 2016;Correa-Garcia et al., 2020;Mardones, 2022;Mellado & Saona, 2020).Table 5 reports the correlation matrix.All control variables note a significant relationship with the two risk-taking measures, which justifies their inclusion in the regression models.We found no high correlations between the independent variables, except for the correlations between C.E.O.Age, C.E.O.Tenure, and C.E.O.Career Horizon.Statistically, older C.E.O.s are more likely to have longer tenures and shorter career horizons.Therefore, we do not jointly include these three measures in the regression models.In turn, multicollinearity should not be a major issue in our estimations and in fact we also estimate the Variance Inflation Factor (V.I.F.) to test for the absence of multicollinearity.All regressions present a V.I.F.post-estimation test of below 2.
We now examine the association between a C.E.O.'s characteristics and corporate risk-taking using the G.M.M. technique to estimate Equation (1).In Table 6, we report the results concerning the C.E.O.'s age and tenure.In Columns 1 and 3, we use the T.R. measure, while in Columns 2 and 4, we use I.R. Columns 1 and 2 show a non-linear, inverse U-shaped relationship between corporate risk-taking and C.E.O.Age.On the one hand, the linear coefficients of C.E.O.Age are positive and  Columns 3 and 4 report the estimates of C.E.O.Tenure.We find that the linear (quadratic) coefficient of C.E.O.Tenure is negative (positive) and statistically significant at 1%, thus supporting a U-shaped relationship between C.E.O.Tenure and corporate risk-taking.According to our estimates, the inflexion point of the relation is around 17 years.The Lind and Mehlum (2010) test confirms the non-monotonic relationship.Based on the mean of C.E.O.tenure (8.678 years), the marginal effect is À0.0078 in Column 3, and À0.0054 in Column 4. These results are economically significant.It means that a one standard deviation change in C.E.O.Tenure implies a 44.95% (36.86%) standard deviation decrease in T.R. (I.R.).These results are consistent with the idea that short-tenured Estimated coefficients (standard errors) from Equation (1) using the two-step system GMM.ÃÃÃ , ÃÃ and Ã for 1%, 5% and 10% significance levels.All the regressions include country, time and industry dummy variables.AR(2) is a test for second-order serial correlation.Hansen is the test of over-identifying restrictions.F-stat is a test for the joint significance of the independent variables.Source: Authors.
Table 7. Female CEO, CEO duality, CEO master, foreign CEO and corporate risk-taking. Variables ( C.E.O.s at first increasingly avoid risk.Nevertheless, long-tenured C.E.O.s have increasingly more specific knowledge and social ties, which leads to greater firm riskiness.Table 7 reports the estimates of the remaining relationships.The estimates for Female C.E.O.s are negative and statistically significant, irrespective of the measure of risk used.In terms of economic significance, companies with a Female C.E.O. are associated with a 67.18% (46.44%) standard deviation decrease in T.R. (I.R.).We thus confirm that female C.E.O.s take less risk than their male counterparts.In Columns 3 and 4 of Table 7,  The coefficients of the control variables are consistent with previous literature.Except for firm age in Columns 1 and Column 3, all the control variables are statistically significant.R.O.A. has a significant negative impact on risk measures since unprofitable firms might invest in risky projects in an effort to improve their financial performance.Size is mainly negatively related to risk because large companies can diversify their activities to reduce risk.As expected, the relationship between risk and financial leverage (L.E.V.) is positive.Finally, C.A.P.E.X. is negatively related to T.R. and I.R. because companies decrease their risk by investing in low-risk projects, such as C.A.P.E.X.investments, instead of projects with uncertain cash flow, such as R&D investments.
All the models report the tests of joint validity of the selected instruments (Hansen test) and the test of second-order autocorrelation of the residuals.The Hansen test reports that instruments are exogenously determined.In addition, we present the pvalue of AR(2), and we confirm the absence of second-order correlation with the error term of the estimates.Finally, all the models report tests of the joint significance of coefficients (F test) at 1% level.
Table 8 presents two additional findings.Columns 1 and 2 test the relationship between C.E.O.Career Horizon and corporate risk-taking.As do Antia et al. (2010), we use the industry-adjusted career horizon measure.The coefficients are positive and statistically significant, which means that C.E.O.s who are about to retire or be dismissed, i.e., with a short career horizon, avoid risky projects for fear of losing their reputation (Chakraborty et al., 2007;McClelland et al., 2012).These results show that a one standard deviation increase in C.E.O.Career Horizon increases the standard deviation of T.R. (I.R.) by 11.18% (14.32%).
We go one step further and examine what impact the culture of the C.E.O.'s nationality has on corporate risk.For this purpose, we focus on the U.A.I. proposed by Hofstede (2001).Managers in countries with high U.A.I. are afraid of the consequences of failure (Mihet, 2013) and do not want to jeopardise their source of wealth (Kanagaretnam et al., 2011).Columns 3 and 4 of Table 8 show that the C.E.O.The U.A.I. is negative and statistically significant.C.E.O.s from countries with high U.A.I. prefer to make less risky decisions.It should be noted that Latin American cultures have high rates of uncertainty avoidance and a low tolerance for the unknown (Boubakri & Saffar, 2016;Sacrist an-Navarro et al., 2022).An increase in C.E.O.U.A.I. from the 25th percentile to the 75th percentile is associated with a 6.41% (5.15%) standard deviation decrease in T.R. (I.R.).

Analysis and discussion
This article aims to disentangle the impact of the most controversial C.E.O.variables on corporate risk-taking.We find evidence to support the idea that a C.E.O.'s characteristics have a strong explanatory power on financial outcomes.Our main results reject prior evidence that C.E.O.age and C.E.O.tenure have a linear relationship with corporate risk-taking.
We first confirm the idea that in the case of younger C.E.O.s, ageing increases risk because they have better cognitive abilities to manage risky financial decisions.In addition, younger C.E.O.s are willing to signal to the market their ability to manage risky projects (Farag & Mallin, 2018;Peltom€ aki et al., 2021;Zhang et al., 2016).However, C.E.O.s may start worrying about their retirement and avoid reputation-ruining risks at  (Naseem et al., 2020).
In addition, we study other C.E.O.characteristics that might impact corporate policy riskiness.The evidence in Table 7 concerning Hypothesis 3 partially confirms the results of Datta et al. (2021), Faccio et al. (2016), andLa Rocca et al. (2020), who point out that female C.E.O.s are more conservative and less overconfident than male C.E.O.s.Faccio et al. (2016) add that female C.E.O.s undertake sub-optimal risky projects because it is more difficult for women to find senior management positions.Similarly, La Rocca et al. (2020) indicate that female C.E.O.s prefer financial decisions which are related to lower risk, such as short-term debt financing.Therefore, we believe there are gender-based decision-making differences which affect corporate risk.We also find evidence regarding the positive relationship between C.E.O.duality and corporate risk-taking (Hypothesis 4), which could be the result of duality endowing the C.E.O. with the freedom and confidence to mitigate suboptimal risk-taking decisions.Our empirical results also add evidence to the scarce literature on C.E.O.s' education and firm outcomes and show that holding a master's degree in business-related topics is positively associated to corporate risk-taking.Farag and Mallin (2018), King et al. (2016), andNaseem et al. (2020) confirm that C.E.O.s with a specialised educational background in business play a key role in risky corporate decision-making because they are more receptive to innovation projects and have accumulated more knowledge.As regards the impact of C.E.O.nationality, we show that foreign C.E.O.sunlike domestic C.E.O.shave more knowledge and better financing conditions that allow them to participate in risky projects.Therefore, we report that C.E.O.nationality significantly influences corporate risk-taking.These results concur with Kim et al. (2020) and Li et al. (2013), who find that a country's cultural factorsand in particular uncertainty aversionimpact risky corporate decision-making.We thus confirm that the culture of the C.E.O.'s country of origin does indeed matter.
We also report some additional results concerning a C.E.O.'s career horizon using an industry-adjusted measure (Hypothesis 7).Our results confirm that the relationship between the C.E.O.'s career horizon and the company's risk-taking is conditional on the dynamism of the company's industrial sector (Antia et al., 2010;McClelland et al., 2012).We also show that myopic C.E.O.s with short careers tend to prefer short-term projects with faster returns.For instance, Matta and Beamish (2008) report that C.E.O.s who are approaching retirement avoid risky decisions such as international acquisitions in order not to jeopardise their reputation.These suboptimal risk-taking decisions are due to the fact that the C.E.O.'s horizon is shorter than the company's life (Antia et al., 2010).

Main findings
We have developed an empirical framework to disentangle the effects of C.E.O.s' observable traits on corporate risk-taking in Latin American companies.Our first finding indicates that the relationship between C.E.O.age and risk-taking is non-linear and inverted U-shaped.This result means that ageing improves the ability of younger C.E.O.s (who are under 45 years of age) to learn and adapt quickly to complex environments but that it causes a greater preference for the status quo among older C.E.O.s.The second set of results shows that the relationship between C.E.O.tenure and risktaking is also non-linear but U-shaped.In other words, as tenure increases, young C.E.O.s become more entrenched and avoid risky decisions.Nevertheless, when C.E.O.tenure reaches a threshold (around 17 years), the benefits of long tenure outweigh the costs, and risk-taking increases.
We also find that C.E.O.gender matters in risk-taking, such that female C.E.O.s take fewer risks than their male counterparts.Furthermore, our results show that when the C.E.O.holds the Chairman of the Board position or has a business-related master's degree, the risk of the firm increases.We document that foreign C.E.O.s take riskier decisions.We further analyse this relationship in the cultural framework and find that C.E.O.s from countries with greater uncertainty avoidance take less risky decisions.Finally, our evidence highlights the relevance of career horizon because it shows that C.E.O.s nearing retirement are more risk-averse.

Contributions
Our research makes a number of contributions.First, on the theoretical side, our results lend support to the Upper Echelon Theory.Taken together, our results challenge the classical theory that C.E.O.s are homogenous individuals who do not have the power to influence the company's financial decisions.In turn, we complement the dominant agency theory approach and show that a C.E.O.'s observable characteristics have explanatory power vis-a-vis corporate risk-taking.
Second, by using a hand-collected dataset of 879 unique C.E.O.s and 369 non-financial firms from six Latin-American countries, we expand the literature, which has mostly focused on other contexts.This region is characterised by a high concentration of ownershipmainly in the hands of familiesand weak investor protection.In turn, our results shed light on a number of emerging countries to which prior research has paid scant attention.Third, we contribute to the debate on the effect of two closely related C.E.O.traits: age and tenure.Whereas longer tenure has usually been associated with older age, we introduce a non-linear specification and find that C.E.O.age and C.E.O.tenure affect risk-taking in opposite directions.Therefore, we refine the measure of C.E.O.traits by showing that C.E.O.age and C.E.O.tenure are not two sides of the same coin but complementary sides of the same person.Fourth, our research also makes a methodological contribution since we use different market-based measures of risk-taking.In turn, we use an external measure of corporate risk-taking that cannot be managed directly by the C.E.O. and is assessed by capital market participants.

Implications
Our work could offer helpful clues to researchers, practitioners, and policymakers alike.Academics and researchers may be interested in understanding the relationship between C.E.O.characteristics and firm risk and in this regard, we provide them with some fresh evidence in the context of emerging economies.In so doing, we shed some light on the little-known debate concerning what impact C.E.O.profile has on organisational outcomes in the Latin American context.We also provide evidence on the profile of the C.E.O., who is most likely to have myopic risk aversion or to generate managerial entrenchment.
At the same time, our results should be welcomed by shareholders, practitioners and participants in capital markets since we uncover some mechanisms which they can use to improve corporate governance and to promote risky but profitable financial decisions.In an environment of concentrated ownership, large dominant shareholders must develop an optimal governance structure in which the interests of all stakeholders are aligned.Our results provide valuable information regarding the profile of the C.E.O. who does not take under-optimal risks.
All of this evidence can also be helpful to policymakers, who can find in our research some guidelines to improve codes of good governance.Most of these codes are inspired by the 'comply-or-explain' principle.Thus, legal and financial authorities could improve the risk appetite or moderate corporate risk-taking by suggesting the optimal profile of firms' decision-makers so as to attain the right level of corporate risk.Moreover, given the importance of C.E.O.s' personal traits, corporate report content could be redefined in order to provide some additional information on top managers.

Directions for future research
Our work is not without limitations, such that further research is needed in the future.Several variables may play a mediating role between C.E.O.traits and risk-taking.For example, compensation can alleviate managerial risk aversion and could be taken into account in future research (Blanes et al., 2020;McClelland et al., 2012;Rehman et al., 2021).Another direction is the interaction between managerial traits and governance structures such as ownership structure, the board of directors, etc.Some studies might therefore address whether the relationship between risk-taking and C.E.O.traits is affected by ownership concentration, large shareholder identity, or the composition of the board of directors.For example, C.E.O.s in family firms are a key issue, such that it is necessary to investigate the impact of C.E.O.profile that encourages corporate risktaking in this type of firm.Furthermore, we only analyse corporate risk-taking, yet this decision must be looked at together with performance.In turn, another promising topic would be to examine whether more risk-taking by certain C.E.O.s also translates into greater corporate profitability.Finally, the international framework could be expanded by considering some legal and institutional characteristics of each country (in addition to uncertainty aversion), such as the protection of investors' rights or the society's longterm orientation.
)while the average I.R. is 0.301.Average C.E.O.Age, C.E.O.Tenure, C.E.O.Master, and Foreign C.E.O. are 54.001years, 8.678 years, 47.8%, and 13.2%, we report a positive and significant relationship between C.E.O.Duality and corporate risk-taking.The magnitude of the relationship is by no means trivial.Firms have a 53.97% (17.50%) standard deviation increase of T.R. (I.R.) when the same person holds the C.E.O. and Chairman of the Board positions.Our evidence suggests that the concentration of power in the hands of a single executive supports risky decision-making.The estimates in Columns 5 and 6 are related to the C.E.O.'s educational background.The coefficients of C.E.O.Master are positive and statistically significant.Having a C.E.O. with a master's degree in business leads to a 45.06% (26.06%) change in the standard deviation of T.R. (I.R.).Therefore, the C.E.O.'s education is positively related to risk because C.E.O.s with a business-related master's degree can better manage risky decisions.The last results of Table 7 are related to C.E.O.culture.The coefficient of Foreign C.E.O.s is positively related to corporate risk-taking.These results are statistically and economically significant.Companies with a foreign C.E.O. are associated with an 88.72% (55.61%) standard deviation increase in T.R. (I.R.).

Table 2 .
Distribution of firms by two-digit NAICS Code., and none that mention otherwise.To obtain the C.E.O.'s master's degree in management (if they have one), we also check LinkedIn.When identifying the C.E.O.'s gender, we rely on their first name.Finally, we drop the observation if we cannot identify this information.Our final sample includes 879 unique C.E.O.s and an average of 4.49 C.E.O.s per company.C.E.O.Age is measured in years.C.E.O.Tenure is the number of years the C.E.O. has served as the firm's C.E.O.Female C.E.O. is a dummy variable that takes the value of 1 if the C.E.O. is female, and 0 otherwise.C.E.O.Duality is a dummy variable that takes the value of 1 if the C.E.O. is also the Chairman of the Board, and 0 otherwise.To measure the level of education -C.E.O.Masterwe use a dummy variable that takes the value of 1 if the C.E.O. has a master's degree in finance, business, economics, or administration, and 0 otherwise. 2Foreign C.E.O. is a dummy variable that takes a value of 1 if the C.E.O. is foreign, and 0 otherwise.In line with Antia et al. (2010), we measure C.E.O.Career Horizon as ðCEO Tenure ind t À CEO Tenure i t Þ þ ðCEO Age ind t À CEO Age i t Þ, where CEO Tenure ind t is the industry median of tenure in year t, CEO Tenure i t is the tenure of the C.E.O.'s company i in year t, CEO Age ind t is the industry median of age in year t, and CEO Age i t is the age of the C.E.O.'s company i in year t: We use the N.A.I.C.S. code to define the industry.Finally, we use Hofstede's ( information

Table 3 .
Definition of variables.

Table 5 .
Lind and Mehlum (2010) are negative and significant at the 1% level.As a result, we confirm the decrease in risk aversion of younger C.E.O.s and the increase in the conservatism of older C.E.O.s as they age, with the inflexion point being between 45 and 46 years.3WeusetheLindandMehlum(2010)test to check the non-monotonic relationship.Furthermore, at C.E.O.Age levels close to the mean (54 years), the marginal effect is around À0.0016 (Column 1) and À0.0009 (Column 2), which is economically significant.4Inother words, a one standard deviation increase in C.E.O.Age is associated with a 9.15% (6.08%) decrease in the standard deviation of T.R. (I.R.). 5 Source: Authors.ECONOMIC RESEARCH-EKONOMSKA ISTRAŽIVANJA significant at the 1% level.On the other hand, the quadratic coefficients of C.E.O.

Table 6 .
CEO age, CEO tenure and corporate risk-taking.
5% and 10% significance levels.All the regressions include country, time and industry dummy variables.AR(2) is a test for second-order serial correlation.Hansen is the test of over-identifying restrictions.F-stat is a test for the joint significance of the independent variables.

Table 8 .
CEO career horizon, CEO UA Index and corporate risk-taking.Estimated coefficients (standard errors) from Equation (1) using the two-step system GMM.ÃÃÃ , ÃÃ and Ã for 1%, 5% and 10% significance levels.All the regressions include country, time and industry dummy variables.AR(2) is a test for second-order serial correlation.Hansen is the test of over-identifying restrictions.F-stat is a test for the joint significance of the independent variables.Source: Authors.a specific time.Therefore, among old C.E.O.s, ageing increases the desire to maintain the status quo in their professional careers.Our second results report a U-shaped quadratic relationship between C.E.O.tenure and corporate risk-taking.Although new C.E.O.s arrive with fresh ideas, they are afraid of jeopardising their main source of wealth and are concerned that their specific knowledge may be obsolete.However, as the years go by, this relationship becomes positive because the C.E.O. has more and more experience and has already earned the trust of the board of directors and shareholders.In this sense, long-tenured C.E.O.s have more specific social networking ties, experience and knowledge that allows them to make riskier decisions