1 Introduction

Why do some firms grow faster than others? This question touches upon a key feature of market economies, which has implications for macroeconomic performance, the evolution of industries, business strategy, and policy-making. From a theoretical perspective, firm growth has been approached from different angles (Geroski, 1999). Starting with stochastic models dating back to the seminal work of Gibrat (1931), much attention has been paid to the question, even in recent times by Hart and Oulton (1996), Bottazzi and Secchi (2006) and Arata (2019), because of the consistent properties of growth rate distributions across countries, industries, and over time (Dosi et al., 2015). Dynamic competitive equilibrium models (Jovanovic, 1982; Ericson & Pakes, 1995; Cabral & Mata, 2003; Cooley & Quadrini, 2004) explain businesses’ growth through the structural characteristics of firms, such as newer or younger and smaller sized firms, a higher level of efficiency, investment, and more effective access to financial resources.Footnote 1 Evolutionary models (Dosi et al., 1995; Nelson & Winter, 1982), which are grounded in the Schumpeterian idea of “creative destruction,” have examined the role of innovation in firm dynamics.Footnote 2

While all these strands of the literature have focused on the role of firms’ characteristics in explaining their episodes of size expansion (e.g., Arrighetti, 1994; Capasso et al., 2015; Yang & Tsou, 2020), the role exerted by managers’ targets and characteristics in firm growth has been much less explored, with few exceptions such as the literature on the impact of the characteristics of the firm founders on the growth of new firms (e.g., Colombo & Grilli, 2005, 2010). This is unfortunate, especially as top executives are the main agents responsible for corporate strategies (Bertrand & Schoar, 2003; Foss & Pedersen, 2016; Hambrick & Mason, 1984).

The principal-agent (P-A) framework, which is traditionally adopted in industrial organization to model the relationship between shareholders and managers’ interests within the firm, sheds light on how firm growth may depend on executives’ choices. Indeed, while the owners of the firm (principals) are interested in the expected present value of the firm’s economic profits, managers (agents) aim at maximizing their utility, which relates more to firm size than to profit (see Baumol, 1959; Marris, 1964; Williamson, 1964). This is because of the positive relationship between managers’ compensation and firm size (Jensen, 1986),Footnote 3 which has been empirically confirmed by several studies (see Brunello et al., 2001; Cosh & Hughes, 1997; Gabaix et al., 2013; Kato, 1997). Based on the P-A framework, managers are expected to push firm size (through growth) beyond the “optimal” size envisaged by shareholders, thus possibly deviating from profit maximization and creating a conflict of interest within the firm.

This study adopts the P-A framework to show how managers’ targets and characteristics help explain that part of growth rate variability among firms, which cannot be ascribed to the stochastic, structural, financial, and technological characteristics of the firm. We investigate whether firms’ higher propensity to grow may be ascribed to the specific incentives of young chief executive officers (CEOs). We hypothesize that the objective of signaling their managerial talent in the market, and the possibility of receiving a better stream of future compensation benefits, leads younger CEOs to seek higher firm growth in contrast to their older counterparts. This may well create a divergence of interests with shareholders. Consistent with this framework, we expect the incentives for firm growth given to younger CEOs to be hindered in firms with more concentrated ownership, where shareholder monitoring is more stringent (Chaigneau & Sahuguet, 2018; Shleifer & Vishny, 1986).

This study focuses on organic growth, which is usually proxied by the rate of growth of sales or assets. Even if firms achieve their growth through mergers and acquisitions (M&As), we believe that the interest in organic growth is grounded in three facts. First, the majority of actual growth episodes are due to the internal expansion of firms in terms of capacity and output. Second, M&As are usually considered “risky” strategies:Footnote 4 these may be preferred by young CEOs simply because of their lower risk aversion due to their youth (Falk et al., 2018). In this study, we include proxies for CEOs’ risk tolerance and attempt to estimate the role of young CEOs’ incentives and targets in firm growth, “net of” their attitude toward risk.Footnote 5 Third, economic theory has shown that organic growth can generate significant agency problems.Footnote 6

We analyze the role of CEO age in organic growth from a large sample of manufacturing firms from seven European countries (Austria, France, Germany, Hungary, Italy, Spain, and the United Kingdom) in the period 2009–2014. The database we employ results from merging Bureau Van Dijk’s (BvD) Amadeus database with the European Firms in a Global Economy (EFIGE) survey. The EFIGE survey has several unique features. First, it is a sample built to represent the manufacturing sectors of the countries covered. In particular, the sampling design follows a stratification by industry, region, and firm size. A relevant implication of this design of this study is that we can consider a large number of mostly privately held firms (only about 2% of firms in the sample are listed) including over 90% small- and medium-sized enterprises (SMEs; with less than €50 million turnover). This allows us to highlight that agency problems in the relationship between shareholders and managers are not an exclusive feature of large and widely held firms. Indeed, while a firm where one person wields both the ownership and management roles can be considered the base case of zero-agency costs (Jensen & Meckling, 1976), the initial separation between ownership and control takes place precisely when a firm is small (Danielson & Scott, 2007). Second, data from EFIGE are fully comparable across countries, since they are derived from responses to the same questionnaire, administered over the same time span (January to May 2010). Third, EFIGE includes a wide range of questions that allow us to control for a rich set of important firm and CEO characteristics, which is key to limiting omitted variable bias issues. Our final sample includes about 7200 manufacturing firms, across twenty-four 2-digit industries.

Our econometric analysis shows three key results. First, when controlled for industry-country unobserved heterogeneity and a relevant set of firm and CEO characteristics, firms managed by CEOs younger than 45 grow faster (over the period 2009–2014) in terms of both sales and total assets than those managed by older CEOs at the median (50th percentile) of the conditional growth rate distribution. Through quantile regressions, we also show that the effect is asymmetric along the distribution of growth rates: firms managed by younger CEOs do grow more than their counterparts managed by older CEOs, and this difference is stronger in the highest (75th and 90th) conditional percentiles. This result is consistent across all the countries considered. Second, our results show no significant statistical relationship between the growth in operating profits and CEO age. We consider this second result as evidence of a potential divergence of interests between CEOs and shareholders. Third, and consistently with agency theory, we find that in firms characterized by more effective monitoring of managers' decisions (proxied by more concentrated ownership), the relationship between CEO age and firm organic growth is much weaker.

As EFIGE is a cross-sectional survey, we cannot employ an identification strategy based on changes in the longitudinal dimension of the CEO age variable. Nonetheless, we control for different time horizons over which growth rates are calculated, to minimize the risk of biased results due to the (unobserved) turnover of CEOs. Moreover, we take the potential endogeneity issue stemming from CEO selection based on firms’ growth trajectories and other characteristics into account by employing a propensity score matched (PSM) sample analysis. Finally, we show that our results are consistent even with the inclusion of a proxy for the level of education of managers. We conduct several robustness checks on the linearity of the CEO age effect, the role of CEO tenure, and the existence of country specificities in the relationship between CEO age and firm organic growth. The main results are robust to all these checks.

Our study contributes to modern literature on industrial dynamics, by offering some interesting insights on the role played by the age of managers in firm growth. While the growth effect of firm characteristics such as age, size, efficiency, innovative activities, and demand factors have been intensively examined (El Shoubaki et al., 2020; Esteve-Pérez et al., 2021; Foster et al., 2016; Heshmati, 2001), less evidence has been provided regarding the role played by the characteristics of the individuals who lead the firm.

This study also relates to the literature that inquires about the role of managerial characteristics in firms’ strategies (see Cronqvist et al., 2012; Faccio et al., 2016; Malmendier & Tate, 2005, 2008). From this perspective, our study is related to Li et al.’s (2017) work which provided evidence that career concerns explain why younger CEOs take investment decisions that boost firm growth. It is also related to Belenzon et al.’s (2019) work which examines the relationship between firm growth and CEO age in a sample of European companies from BvD’s Amadeus. However, two novel perspectives are contributed by this work. First, unlike Li et al. (2017) who rely on a sample of only large US firms (with an average plant size of about 300 employees), our analysis covers a set of diverse EU countries and is based on a sample that includes over 90% SMEs (with 87 employees per firm on average, and a median size of 26 employees). This is key to highlighting that agency problems are also at work in firms of smaller size. Second, while Belenzon et al. (2019) only focus on owner-managed firms to “[…] eliminate the potential agency problems […], which could affect managerial actions and hence firm growth […]” (p. 924), we focus precisely on this agency problem by comparing the relationship between CEO age and firm growth in firms with different degrees of ownership concentration. Indeed, our empirical design delivers enough variability in governance structures, which allows us to test for the role of concentrated ownership as a monitoring device that can align CEOs’ strategies with shareholders’ interests.

The rest of this paper is structured as follows. Section 2 draws the theoretical underpinnings of the relationship between CEO age and firm growth. Section 3 describes the data. Section 4 presents the econometric results, and the robustness checks and Section 5 concludes.

2 Literature review and hypotheses development

2.1 CEO age, risk aversion, and corporate strategies

Younger CEOs may be more risk-tolerant, thus pursuing riskier and incidentally size-enhancing strategies (like M&As, R&D investments, and internationalization). In a representative sample of 80,000 individuals from 76 countries, Falk et al. (2018) have demonstrated a positive relationship between risk aversion and individual age. Existing empirical studies largely support a negative relationship between CEO age and managerial risk-taking. Serfling (2014) finds a negative relationship between CEO age and firms’ risky strategies, i.e., R&D investments, un-diversified acquisitions and operations, and higher operating leverage. Elia et al. (2021) confirm that older CEOs are less likely to engage in cross-border M&As in unrelated industries, although they find an inverted-U relationship, suggesting that inexperience and limited track record of the youngest CEOs make their positions more uncertain and volatile, thus reducing their propensity towards riskier strategies. Yim (2013) empirically confirms that firms managed by younger CEOs demonstrate a higher probability of pursuing M&As in the period 1992–2007. Li et al. (2017) find that firms managed by younger CEOs are more likely to significantly invest and divest than firms managed by older CEOs.

However, not all episodes of firm growth are the result of risky strategies: for example, managers may devote effort toward organic growth via aggressive pricing strategies that can boost sales volumes and increase market share, possibly at the expense of profitability (Aghion & Stein, 2008).

This then begs the question: what factors, other than their inherently lower risk aversion, would motivate young CEOs to boost their firms’ organic growth?

2.2 Signaling, compensation, and career concerns

First, there may be a managerial signaling factor (Prendergast & Stole, 1996). A young CEO may prefer rapid expansion of firm size—rather than to achieve the optimal firm size (for shareholders)—to signal their talent or capabilities to the market for managers. Conversely, older CEOs may be more reluctant to change their investment behavior frequently because this may be a sign of previous unsuccessful decisions (conservatism). Second, a compensation incentive factor may also be at work. To the extent that firm size is a primary determinant of the CEO’s remuneration (see Brunello et al., 2001; Cosh & Hughes, 1997; Gabaix et al., 2013; Kato, 1997) and given that an optimal contracting scheme cannot be signed,Footnote 7 CEOs may be incentivized to pursue size expansion early on in their careers to access longer streams of future compensation benefits (Yim, 2013). Third, the upper echelons theory (Hambrick & Mason, 1984) would also predict that older CEOs attach more value to career and financial security and exhibit a greater commitment to the status quo of the firm. Indeed, older CEOs may see their time of retirement as a moment of imminent assessment of their jobs and roles in the organization and may prefer legacy conservation and wealth preservation.

While the factors discussed above predict a negative relationship between CEO age and firm growth, career concerns may work in the opposite direction. Indeed, younger CEOs may face a higher probability of being fired due to a yet unestablished or fledgling reputationFootnote 8 and thus suffer greater market scrutiny (Holmstrom, 1999; Scharfstein & Stein, 1990). If that is the case, young CEOs may pursue less size-enhancing investments with respect to their older counterparts, especially if these are also risky strategies (Hong et al., 2000). However, career concerns should be less relevant in the context of the present work, given that organic growth may well be pursued via non-risky strategies, such as reduction in the price-cost margin. A firm may devote efforts toward growth and the expansion of its market share by lowering the price-cost margin, and this may create a divergence in interests between firm growth (managers) and profit maximization (shareholders).

Among a variety of solutions proposed to reconcile this divergence, monitoring of executives by more concentrated ownership is certainly one of the most common, due to several reasons (Chaigneau & Sahuguet, 2018; Shleifer & Vishny, 1986; Shleifer & Vishny, 1997). First, large blockholder owners have the incentive, given the magnitude of their ownership stake, to collect information on executives’ actions. Second, they have high voting power to put pressure on the managers and possibly threaten removal (e.g., via a takeover). Third, when large shareholders own more than half of the firm’s equity shares, they have full control over the firm and its management.

The role of ownership concentration seems relevant in the context of our empirical analysis of European firms. The evidence suggests that in continental Europe, high reliance is placed on large investors and their monitoring effectiveness (with respect to alternative mechanisms, such as superior legal protection of investors or incentive contracts) in aligning the interests of the owners and their managers (La Porta et al., 1998). The empirical evidence has shown that ownership concentration can be an effective tool in aligning shareholders’ interests with executives’ utility.Footnote 9

To the extent that the relationship between CEO age and firm organic growth is due to divergent interests between shareholders (more interested in profit growth) and CEOs (more interested in firm growth), one could expect this relation to be weaker when agency problems are reduced via effective monitoring, which may be achieved through more concentrated ownership. The discussion above leads us to hypothesize that:

H1. Firms with younger CEOs show higher rates of organic growth, even conditional on risk-taking.

H2. The negative relation between CEO age and firm organic growth is weaker in firms with more concentrated ownership.

3 Data and descriptive analysis

In this study, we rely on a unique source of data which results from merging Bureau Van Dijk (BvD)’s Amadeus with the EU-EFIGE/Bruegel-UniCredit dataset (EFIGE). Amadeus is a well-known and widely used source of economic and financial information on European companies. For the purposes of this study, we were able to gather data on the period from 2001 to 2014. EFIGE is the result of a survey administered within the project European Firms in a Global Economy: internal policies for external competitiveness, which was supported by the Directorate General Research of the European Commission through its Seventh Framework ProgrammeFootnote 10. In a first for Europe, EFIGE combines measures of firms’ international activities (e.g., exports, imports) with quantitative and qualitative information on about 150 items ranging from R&D and innovation, labor organization, financing, and organizational activities. Data consist of a representative sample (at the country level for the manufacturing industry) of almost 15,000 surveyed firms (above ten employees) in seven European economies (Germany, France, Italy, Spain, United Kingdom, Austria, and Hungary). Data were collected in 2010, and most questions refer to the year 2008. In some cases, respondents (i.e., individuals who were occupying leading positions in their firms) were asked about the activities of their firms during the years 2007–2009. Considering that the information on the age of CEOs is available for the year 2009, we use the information available in Amadeus, to calculate our dependent variable as the growth rate per unit of time (following Evans, 1987; Variyam & Kraybill, 1992; Moschella et al., 2019):

$$ {\overline{gr}}_{i,2009-2014}=\frac{\ln \left({SIZE}_{i,2014}\right)-\ln \left({SIZE}_{i,2008}\right)}{6} $$

where SIZEi, t is equal to the operating revenue (sales) of firm i at the end of year t.Footnote 11 This approach is better suited to analyze medium-run growth profiles of firms, which are more likely to result from CEOs’ strategies and less affected by noise and measurement errors, which can be severe in the case of yearly growth rates (Pieri, 2018). However, growth rates over longer periods, such as our 6-year period, may be more prone to the fallacy of not capturing a change (turnover) in CEO, which cannot be directly controlled in the data. To partially cope with this issue, we will replicate our analysis over shorter periods of time (see Section A.3.1 in the Online Appendix for a discussion on this issue). It is worth mentioning that the information on sales growth derives from unconsolidated accounts, which excludes growth through M&As, hence accurately reflecting organic growth. Since data on firm turnover is not available for all firms for the relevant years, our initial sample is effectively reduced to about 7200 firms.Footnote 12 About 93% of our sample firms have turnover values of less than €50 million and are thus SMEs.

The EFIGE survey provides information on the age of the CEO as a categorical variable with seven items (< 25 years old; 25–34; 35–44; 45–54; 55–64; 65–74; >= 75). The frequency distribution of the sample firms across these CEO age bins is shown in Fig. 1.

Fig. 1
figure 1

Histogram on the percentage of firms by CEO age category. This histogram shows on the x-axis, the seven categories of age of the CEO, as they are available in the EFIGE survey. Overall, 13533 manufacturing firms included in the Amadeus-EFIGE sample have information on the age category of their CEO. The y-axis shows the percentages of firms corresponding to each CEO age class

The histogram shows that the modal value of CEO age corresponds to the category 45–54 years. While it is impossible to have a precise figure for the median and mean age of the CEOs in our sample, an inspection of Fig. 1 suggests that it could be slightly above 50, which is in line with the evidence provided by other academic studies and scientific reportsFootnote 13. For example, Faccio et al. (2016) report a mean (median) age of 50.3 (51) for the CEOs of the firms included in their analysis, while in the sample of firms used by Li et al. (2017) and Belenzon et al. (2019) the mean age of CEOs is at 53.6 and 50.6, respectively.

To clearly separate young and old CEOs, and in line with previous studies on the role of CEOs in firms’ strategies, we group the seven categories into two macro-categories: CEOs younger than 45 versus the rest.Footnote 14 While about 76% of our sample firms are managed by CEOs who are 45 or older, the share of CEOs younger than 45 is not negligible (24%).Footnote 15

The EFIGE survey allows us to control for several firm and CEO characteristics that could confound the relationship between firm growth and CEO age, thus limiting the omitted variable bias issue. In particular, we have information on firm age (based on the year of establishment), which is typically associated with firm growth (Barba Navaretti et al., 2014). We also control for the propensity to innovate (Geroski, 1999) and the degree of internationalization (Serra et al., 2012), which are typically associated with risk-taking, hence could result in higher firm growth. Moreover, we can account for firms that have undergone any form of quality certification and widened their product range, as strategies for building a customer base (Foster et al., 2016). We also control for several proxies of economic and financial characteristics of firms, such as firm size (Hall, 1987; Lotti et al., 2009) profitability (Coad, 2007), debt to asset ratio and past sales growth rate, all calculated from BvD’s Amadeus over the period 2001–2008. As for CEOs, we can control for several characteristics associated with their risk-tolerance, including their gender, international experience, and over-confidence. Young CEOs may be more overconfident than their older counterparts (Citci & Inci, 2016): over-confident CEOs systematically overestimate their ability to create value and, in the case of abundant internal financing, tend to overinvest (Malmendier & Tate, 2005, 2008). This may affect firm growth, especially at the early stages of startups (Szerb & Vörös, 2019).Footnote 16 A detailed description of the variables used in the empirical model is contained in Table 2.

Table 1 provides some summary statistics on the sales growth rates in the period 2009–2014.Footnote 17

Table 1 Descriptive statistics of the sales growth rates (per unit of time) in the period 2009–2014

Bearing in mind that the 2009–2014 period embraces the years of the Great Recession, which impacted Europe heavily, it is not surprising that the sales growth rates per unit of time at both the 50th percentile and the mean of the distribution have been negative (approximately -0.008 and -0.031 log changes, respectively, as shown in Panel A). In line with our expectations, the median firm managed by a CEO younger than 45 has experienced a negative growth rate but lower (in absolute value) than the median firm managed by an older CEO. To account for possible differences across industries and countries, we also computed the growth rates per unit of time adjusted by industry (2-digit NACE rev.2), country, and year:Footnote 18 these are shown in Panel B of Table 1. The growth rate of the group of firms managed by CEOs younger than 45 is around 0.7 points higher than the growth rate of the group of firms managed by CEOs who are 45 or older, at the 50th percentile of the growth rate distribution. This difference is statistically significant as shown in the bottom panel of Table 1 (as per the p value associated with the Wilcoxon; Mann-Whitney test). Moreover, the lower (in absolute value) skewness of the growth rate distribution for the firms managed by CEOs younger than 45 points to fewer episodes of heavy size contraction for this group of firms in the period 2009–2014. Figure 2 confirms these results.

Fig. 2
figure 2

Distributions of sales growth rates (per unit of time) in the period 2009–2014 by CEO age category. This figure plots the distributions of sales growth rates (per unit of time) calculated over the period 2009–2014 by CEO age category for those firms in the Amadeus-EFIGE database with information on both CEO age and growth in the period 2009–2014. The x-axis indicates the sales growth rates (only values within the interval from −0.4 to 0.4 have been reported in the plot to make it more readable), while the y-axis reports the percentage of firms. Firms’ sales have been normalized by industry (2-digit), country and year averages to remove common trends, such as inflation and business cycles effects

The distribution of sales growth rates for the firms managed by younger CEOs (grey line) shows not only a higher median but also a higher number of episodes of positive and fast growth and fewer episodes of heavy and negative growth (contraction) than the distribution of growth rates of firms managed by CEOs who are 45 or older. It is relevant to underline that the distribution of growth rates is not normal, while better approximated by a Laplace (symmetric exponential) distribution (Bottazzi et al., 2001; Bottazzi & Secchi, 2006). The characteristic tent-shape of this distribution points to the existence of “fat tails”, i.e., the higher incidence of episodes of fast growth and heavy contraction with respect to what a Gaussian (normal) distribution would predict.Footnote 19 The presence of fat tails justifies two choices made in the econometric analysis. First, the adoption of a least absolute deviations (LAD) estimator instead of the more common ordinary least squares (OLS) to get insights into the central tendency growth rate of firms managed by CEOs of different ages. Indeed, the first estimator is more robust in the case of a non-Gaussian distribution of the dependent variable (Wooldridge, 2010), like that of sales growth rates. Second, the attention paid to differences in growth between firms managed by CEOs of different ages, which may especially be in the tails of the distribution. To uncover them, we will use a generalization of the LAD estimator, i.e., quantile regressions.

Table 2 shows some descriptive statistics regarding firms’ and CEOs’ characteristics of the European manufacturing firms contained in the Amadeus-EFIGE database.

Table 2 Descriptive statistics by CEO age category

Firms managed by younger and older CEOs are different in several dimensions, and the last column of Table 2 confirms that these differences are usually statistically significant. The median firm in our sample, managed by a young CEO, is smaller, younger, more profitable, and more indebtedFootnote 20 than its counterpart that is run by an older CEO and more likely to have introduced process innovations, while being less internationalized (in terms of the number of markets served via exports). Moreover, a higher share of females is observed among younger CEOs, who are also found to be more overconfident. Finally, firms managed by young CEOs grew more than their counterparts did in the period 2001–2008. With the two groups of firms being different in several dimensions, we will conduct a multivariate analysis, controlling for firms and CEOs’ characteristics which may well be correlated with both CEO age and firm organic growth.

4 Econometric analysis

4.1 Baseline estimates

Our baseline empirical model is a cross-sectional regression model of firm growth rates per unit of time calculated over a 6-year period as a function of CEO and firm characteristics at the beginning of the period. It may be written as:

$$ {\overline{gr}}_{i,2009-2014}={\alpha}_{\theta }+{\beta}_{\theta } CEO\kern.2em {\left(<45\kern.2em y.o.\right)}_{i,2009}+{\gamma_1^{\prime}}_{\theta }{\overline{Z}}_{i,2001-2008}+{\gamma_2^{\prime}}_{\theta }{W}_{i,2008}+{\mu}_{\theta j}\cdot {\tau}_{\theta c}+{\varepsilon}_{\theta i,2009-2014} $$

\( {\overline{gr}}_{i,2009-2014} \) is the growth rate per unit of time of the firm i in the period 2009–2014, and CEO (<45 y. o.)i, 2009 is a dummy variable taking value one if the CEO is younger than 45 and zero otherwise. The vector \( {\overline{Z}}_{i,2001-2008} \) contains proxies (calculated as averages over the period 2001–2008) for relevant characteristics of firm i, which may be related to both CEO age and firm growth, i.e., firm size and firm age, firm profitability, and debt to assets ratio.Footnote 21 The vector Wi, 2008 contains proxies of both firm and CEO/executive characteristics obtained from the EFIGE survey, which mostly refer to 2008. Variables included in the \( {\overline{Z}}_{i,2001-2008} \) and Wi, 2008 vectors have been described in Section 3. In all specifications, we include a vector of country-industry fixed effects (indicated by the subscripts c and j, respectively), where manufacturing industries are defined at the 2-digit NACE rev.2 level of classificationFootnote 22 to control for systematic differences in firm growth and CEO age across countries and industries. We also report the heteroskedastic robust standard errors (as suggested by Machado et al., 2011) in parentheses.Footnote 23 It is worth mentioning that our analysis is conditional on firm survival until the end of the period over which the growth rate is calculated, which is either 2014 or earlier because in several robustness checks, the growth rate per unit of time is calculated over shorter time horizons (see Section A.3.1 and Table A.3 in the Online Appendix). This is in line with several studies in the literature on firm growth, such as Coad (2007), Coad and Rao (2008), Barba Navaretti et al. (2014), Bianchini et al. (2017), Pieri (2018), and Moschella et al. (2019).

As described in Section 3.1, we employ the quantile regression estimator (Koenker & Basset, 1978), which is the vector of parameters δ that solves the following operation:Footnote 24

$$ {\displaystyle \begin{array}{c}\mathit{\min}\\ {}\delta \end{array}}\frac{1}{n}\left\{\sum \limits_{i:{\overline{gr}}_{i,2009-2014}\ge {\delta}^{\prime }{X}_i}\theta \left|{\overline{gr}}_{i,2009-2014}-{\delta}^{\prime }{X}_i\right|+\sum \limits_{i:{\overline{gr}}_{i,2009-2014}<{\delta}^{\prime }{X}_i}\left(1-\theta \right)\left|{\overline{gr}}_{i,2009-2014}-{\delta}^{\prime }{X}_i\right|\right\}, $$

where, for notational simplicity, we denote the vector of the “right-hand-side” variables in Eq.2 as Xi and the respective vector of parameters to be estimated, as δ. For the time being, we will focus on θ=0.5, the conditional median, or what is also referred to as the LAD estimator, but slope parameters may vary at different quantiles of the conditional growth rate distribution. This is particularly relevant for us to assess the role of CEO age in organic growth across the entire spectrum of growth rates, and because as shown in the tails of the distribution of Figure 2, there could be appreciable differences in growth rates among firms with CEOs of different ages.

Table 3 shows the estimates for Eq. 2 by means of LAD. In col. (1), we introduce only the main explanatory variable together with the vector country-industry dummies. From col. (2) to col. (8), we add the other firm characteristics hierarchically, giving col. (8) the richer specification. Once controlled for a large set of firm characteristics, the growth rate of firms managed by CEOs younger than 45 is around 0.7 log points higher than the growth rate of firms managed by older CEOs, at the median of the conditional growth rate distribution. The magnitude of this coefficient is stable across specifications.

Table 3 LAD (median) regressions of sales growth rates (per unit of time) in the period 2009–2014 on firm characteristics.

As for the coefficients associated with the control variables, it is worth mentioning that firm size and age show the expected negative sign, that is, younger and smaller firms have grown more than their older and larger counterparts in the period 2009–2014 (see Barba Navaretti et al., 2014; Haltiwanger et al., 2013). Most notably, since smaller and younger firms are more likely to appoint younger CEOs, in col. (2) we notice that when controlling for firm size and age, the coefficient on CEO age drops slightly. However, our results show that if we compare two firms of the same size and age, the firm with a CEO younger than 45 will have 0.59 log points higher sales growth. While the debt-to-assets ratio is never significant in explaining the growth of the firm at the median of the conditional distribution of growth rates, more profitable firms (higher ROE) show higher growth rates. This result may concurrently be explained by models that analyze the role of financing constraints for firms’ investment (Fazzari et al., 1988), and by the evolutionary models of the “growth of the fitter” (Coad, 2007). Ceteris paribus, more innovative (especially in terms of process innovation) and internationalized firms (in terms of the number of countries to which the firm exports, and the status of the importer), have grown faster in the period 2009–2014 (see Coad & Rao, 2008; Grazzi & Moschella, 2018). Finally, while the share of graduate workers is not associated with higher growth, both quality certification and product range widening strategies are.

As discussed in Section 2, CEO age may be correlated with other characteristics of top executives that relate to higher firm growth. Thus, to limit any omitted variable bias problem, we further enrich our empirical specification with a vector of CEO characteristics observed at the beginning of the period. Specifically, we include the information on CEO gender and proxies of risk-tolerance, over-confidence, and international experience. Table 4 shows that while over-confidence is positively correlated with firm growth, risk-tolerance and CEO gender are not, and the international experience of the executives is only marginally significant. Simultaneously, while a small decrease in the CEO age coefficient is appreciable, the growth rate of firms managed by a CEO younger than 45 is around 0.66 log points higher than the growth rate of firms managed by an older CEO, at the median of the conditional distribution of growth rates.

Table 4 LAD (median) regressions of sales growth rates (per unit of time) in the period 2009–2014 on firm and CEO characteristics

These results support the view that organic growth is the outcome of investing in high-growth segments of activity, creating new products, services, and business models (Ahuja et al., 2017). These results are also consistent with the idea that younger CEOs may not necessarily boost organic growth because of their higher risk tolerance. Indeed, among our control variables, we have several measures that allow partial control for the higher-risk tolerance of younger CEOs. In particular, we include the propensity not to hedge against exchange rate risk, a measure of over-confidence and CEO characteristics that are traditionally associated with more risk-taking, such as CEO gender (male) and the international experience of the executives. In Section A.3.5 in the Online Appendix, we control for another proxy for risk, by including the coefficient of variation in turnover (i.e., the ratio of the standard deviation of sales to the average sales figure) over the period 2009–2014.

From the first set of econometric results, we obtain evidence that CEO age is negatively correlated with organic growth in the period 2009–2014. We submit that even conditional on younger CEOs being more prone to risk-taking, these results are consistent with the arguments developed in Section 2 and, thus, lead to the formulation of H1. Young CEOs have the incentive to boost organic growth not necessarily because of a lower risk aversion, as growth can also be achieved by relatively low-risk strategies, like aggressive pricing, which allows market share gain and boost firm size. This will signal the value of the CEO (managerial signaling) to the market and provide the justification for an increase in compensation due to the higher complexity associated with managing a larger company (compensation incentive). Simultaneously, since this strategy may lower the firm’s profit margins, firm growth may occur at the expense of profitability maximization, and can thus create a wedge between the young CEO and the shareholders’ incentives.

When the quantile regression estimator in Eq. 3 is applied at different percentiles (p10, p25, p50, p75, p90) of the conditional growth rates distribution, we uncover that the relationship is statistically relevant for the episodes located on the right tail of the distribution, while it is not significant for the episodes located on the left tail (Table 5).

Table 5 Quantile regression of the sales growth rates (per unit of time) in the period 2009–2014 on firm and CEO characteristics

This result is consistent with earlier evidence showing that CEO compensation is more sensitive to episodes of firm expansion than to episodes of firm contraction (Bebchuck & Grinstein, 2005). In other words, to the extent that young CEOs’ compensation grows when their firm grows in size, but does not decline when their firm shrinks, they have an incentive to boost growth, but not to prevent downsizing. This may map into larger differences in growth between firms managed by younger CEOs and those with older CEOs, at the higher percentiles of the conditional distribution (i.e., above the median, which is close to zero). Instead, these differences disappear among firms whose growth rate is below the median. Moreover, the effect is stronger at the highest percentiles (p90) of the right tail of the distribution, i.e., for episodes of high growth. This result is coherent with the results shown by Yim (2013) on the negative relationship between CEO age and acquisitions, which the author found significant only for episodes of relevant acquisitions that is a deal value exceeding 5% of the firm’s market capitalization, which would more likely affect CEO compensation.

In Section A.3 in the online appendix, we conduct a battery of robustness checks and explore (i) different time horizons over which the growth rate is calculated, to lessen the issue related to a possible (unobserved) turnover of the CEO (Section A.3.1 and Table A.3); (ii) issues relating to the timing of CEO appointment, by controlling for firm past-growth and by running the baseline model on the sub-sample of firms established within the last five to ten years before 2009 (Section A.3.2 and Tables A.4, A.5); (iii) a possible reverse-causality bias, from firms with certain characteristics and growth trajectories which choose a young CEO, by means of a propensity score matched (PSM) sample analysis (Section A.3.3 and Tables A.6, A.7, A.8); (iv) the validity of the age threshold set at 45 (Section A.3.4 and Table A.9); (v) the possible role of higher risk-tolerance of younger CEOs (Section A.3.5 and Table A.10), by controlling for the volatility in sales (as proxied by the coefficient of variation in firm turnover over the period 2009–2014); (vi) the possible confounding effect of the education level of the younger cohorts of managers (Section A.3.6 and Table A.11); (vii) the robustness of our findings with the exclusion of firms engaged in M&As (Section A.3.7 and Table A.12); and (viii) the existence of country specificities in the relationship between CEO age and firm growth (Section A.3.8 and Table A.13). Overall, the correlation between CEO age and firm organic growth is robust to all these tests. In Section 4.2, we inquire, through the lens of the P-A framework, into the mechanisms that may enhance or moderate the relationship between CEO age and organic growth.

4.2 Identifying the mechanism in a P-A framework

Section 2 highlighted that the higher growth achieved by firms managed by younger CEOs could be the result of a specific set of incentives for top managers, which may diverge from shareholders’ targets. We have argued that younger CEOs may be motivated to boost firm growth to signal their talent and to increase their compensation. However, shareholders are normally more interested in profit maximization, than the growth of sales and market share. In Table 6 we provide evidence of this potential divergence of interests. In fact, while firms managed by younger CEOs show higher growth of total assets (panel A), they do not show any difference in the growth of operating profits (panel B).

Table 6 Quantile regression of total assets growth (panel A) and profit growth (panel B) (per unit of time) in the period 2009–2014 on firm-level and CEO characteristics.

This suggests that strategies pursued by younger CEOs may place a wedge between their own targets and the interests of their shareholders. As discussed in Section 2, this divergence should be moderated by the ability of shareholders to monitor CEOs’ decisions. It is established in the literature that in firms with more concentrated ownership, the shareholders can monitor CEOs more effectively; therefore, we use this to test the arguments leading to H2. In Tables 7 and 8, we estimate two variants of Eq. 4 as below:

$$ {\overline{gr}}_i={\alpha}_{\theta }+{\beta}_{\theta } CEO\kern.2em {\left(<45\kern.2em y.o.\right)}_i+{\eta}_{\theta }{OWN}_i+{\omega}_{\theta }{OWN}_i\ast CEO\kern.2em {\left(<45\kern.2em y.o.\right)}_i+{\delta}^{\prime }{X}_i+{\mu}_{\theta j}\cdot {\tau}_{\theta c}+{\varepsilon}_{\theta i} $$
Table 7 The role of monitoring (concentrated ownership) for the relationship between CEO age and firm growth
Table 8 The role of monitoring (concentrated ownership) for the relationship between CEO age and firm growth.

where OWNi is a measure of ownership concentration. The notation has been simplified by dropping the time period indicator and collapsing the control variables in vector Xi.

In the first variant of Eq. 6 (Table 7), we interact the CEO age dummy with a dummy that takes a value of one for firms where the first shareholder owns 50% or more of the firm’s equity and zero otherwise. Results show that this interaction indeed negatively moderates the effect of CEO age on firm growth, although the coefficient is significantly different from zero only at the 90th percentile. At the 90th percentile, firms with relatively more dispersed ownership and young CEOs grow faster than their counterparts with young CEOs but more concentrated shareholders (H0: ηθ + ωθ = 0, rejected). This suggests that concentrated ownership curbs the incentives of young CEOs to achieve more growth in firm size. Conversely, no difference in growth is found between firms with younger or older CEOs, if ownership is concentrated (H0: βθ + ωθ = 0, not rejected).

In Table 8, we estimate a variant of Eq. 6, where the measure of ownership concentration is continuous. In particular, we simply introduce the share of equity owned by the first (largest) shareholder into the regression. The results indicate that at the 90th percentile, the degree of ownership concentration negatively moderates the effect of CEO age on firm growth.

5 Concluding remarks

We show that the age of CEOs is significantly associated with firm organic growth. In a large sample of mostly small and medium privately held European manufacturing firms, we find that, at the median of the conditional growth rates distribution, the growth in sales (and total assets) per unit of time over the period 2009–2014 was 0.66 (0.52) log points higher in firms managed by CEOs younger than 45. This difference is twice as large if observed at the 90th percentile, suggesting that firms with younger CEOs are more likely to be very fast-growing firms. Our results are robust to the inclusion of a large vector of firm and CEO characteristics, industry-country unobserved heterogeneity, and a wide set of robustness checks, including issues relating to the time horizon and the appointment of CEOs, the educational attainment of younger cohorts of managers, and endogeneity.

The relationship between CEO age and growth could be explained by the higher risk-taking propensity of young CEOs (Falk et al., 2018)—which has already been associated with firms’ risky strategies, such as R&D investments and M&As (Serfling, 2014; Yim, 2013)—or by other individual characteristics such as over-confidence, stamina, higher brain-process capacity, and a lesser rule-of-thumb approach to decision making (Besedeš et al., 2012). Indeed, our findings reveal that some CEO and firm characteristics that are typically associated with risk-taking and over-confidence are also correlated with firm organic growth. However, we argue that, conditional on these characteristics, the negative relation between CEO age and firm organic growth can be explained by the incentive for younger CEOs to boost firm sales and assets, possibly at the expense of profit margins. Indeed, young CEOs want to signal their value and increase firm size, so that an increase in their compensation can be justified. Clearly, this is unlikely to be in the interests of shareholders. Consistent with this interpretation, we find no significant statistical relationship between CEO age and the growth in operating profits. We also provide evidence that in firms characterized by more effective monitoring governance (more concentrated ownership), the relationship between CEO age and firm organic growth is much weaker. We contend that if our results were driven by CEOs’ individual characteristics, other than their greater incentive to achieve fast growth, there would be no reason to find weaker results in more closely monitored firms.

This study contributes to the modern literature on industrial dynamics, by offering insights on the role played by the age of the top executive in firm growth, exclusive of the growth effects ensured by the stochastic, structural, financial, and technological firm-level characteristics. Indeed, while the growth effect of firm characteristics such as age, size, efficiency, innovative activities, and those associated with demand factors has been deeply studied, less evidence has been provided on the role played by the characteristics of the individuals who lead the firm.

These results bear many implications both for policy and governance. First, in many European countries, the management, including board members, is old. In our sample, around 80% of the firms have CEOs older than 45. This could be a sign of insufficient turnover in management. This gerontocratic approach to firm management, combined with the evidence that firms run by younger CEOs have been growing more (especially in the complex times of the aftermath of the financial crisis of 2009), could explain why European firms are less dynamic than their US counterparts (Bravo-Biosca et al., 2016). Given that the small size of European firms is closely related to the slow path of productivity growth in Europe, faster turnover in management could help reduce allocative inefficiencies and improve gross domestic product (GDP) growth in the aggregate. Furthermore, as firm growth among incumbent firms is the most important source of job creation among OECD countries (Criscuolo et al., 2014), our results suggest that rejuvenating the management can have important aggregate employment and welfare effects.

Second, our evidence supports the view that governance with independent management from shareholders (and looser supervision by the latter) improves firm performance and should be fostered and implemented as a best practice. Of course, policy-makers do not have the power to rejuvenate management, yet a gradual spread of a corporate culture that increasingly emphasizes management turnover, and independence in management will certainly benefit firm performance and the growth of firms (and SMEs in particular) in the long run.

It is worth laying out a few caveats about our analysis. First, even if the main results of our work are robust to checks and provide the readers with a specific channel of causation, the identification of the causal effect of CEO age on firm growth could be sharpened if we had more precise information on the time of CEO appointment and turnover. Indeed, longitudinal information on CEOs may prove useful for future research on this topic. Second, even if the quantile regression model allows us to appreciate the heterogeneity of the relationship between CEO age and firm growth along the entire distribution of growth rates, the results provided only hold for the group of firms that “survive” (do not exit the market) during the entire period (2009–2014). The field of research about quantile regression models with sample selection is very promising (Arellano & Bonhomme, 2017), but their application is still not fully developed and it is certainly out of the scope of this study. Third, our study would have benefitted from more information on the psychological and cognitive traits of CEOs. This would have allowed us to better disentangle the role of age, from other traits that may be correlated with age and may impact how managers lead their firms.