Following propositions of theories from the domain of lifespan psychology (Baltes and Baltes, 1990; Carstensen et al., 1999), we argue that entrepreneurial motivation is likely to change when people get older. Importantly, motivational changes are not directly caused by age, but by other age-related factors such as the remaining time and opportunities people perceive in their lives (Carstensen, 2006; Cate and John, 2007; Henry et al., 2017).

The lifespan perspective suggests that this change in people’s perception of opportunities is related to several age-related individual (i.e., internal) and contextual (i.e., external) mechanisms. Regarding internal mechanisms, research has shown an age-related decline of resources such as information processing, perceived time left, and physical stamina (Schulz and Heckhausen, 1996; Baltes and Lang, 1997). As these resources are important to maintain a focus on opportunities, the perception of opportunities (i.e., how many new goals, plans, options, and opportunities people believe to have in their personal future; Cate and John, 2007; Zacher and Frese, 2009) is likely to decrease with age. External mechanisms are, for example, age-related norms and environmental constraints that may lower perceived opportunities for entrepreneurship. Norms in Western societies often depict retirement as the better alternative over pursuing new professional opportunities once a certain age is reached (Neugarten et al., 1965; Hershey et al., 2002; Kautonen et al., 2010). Based on these arguments, we hypothesize that the perception of opportunities for entrepreneurship decreases with age.

To act upon a business opportunity, it is first necessary to perceive it (Baron, 2004). Once an opportunity is detected, its actual pursuit often depends on individual goal selection and persistence during goal pursuit (Seijts, 1998; Aspinwall, 2005). According to lifespan psychology, perceiving that many opportunities remain in life is likely to lead to the pursuit of long-term goals. Regarding entrepreneurial activity, those long-term goals may be venture creation or business foundation. In contrast, people perceiving fewer opportunities are likely to pursue short-term goals, such as emotional well-being (Lang and Carstensen, 2002; Carstensen, 2006). Individuals who perceive many remaining opportunities are likely to set more challenging goals, and ultimately apply higher standards to evaluate their goal-accomplishment (Markus and Nurius, 1986; Cross and Markus, 1994). Based on these findings, researchers have concluded that perceiving opportunities fosters the amount of effort and persistence invested in goal pursuit, leading to higher work engagement and performance (Zacher et al., 2010; Schmitt et al., 2013). Consequently, the perception of opportunities for entrepreneurship should positively impact entrepreneurial activity, as people set more ambitious long-term goals and are more likely to pursue these.

An example of the importance of opportunity perception for entrepreneurial activity is entrepreneurship training. In those programs, people are often trained to recognize opportunities. Research by DeTienne and Chandler (2004), for example, showed that after the training, students were better able to identify opportunities, while generating more ideas with higher innovativeness. Increased opportunity recognition, in turn, may heighten entrepreneurial behavior. We therefore assume that perceiving opportunities is positively related to entrepreneurial activity.

Taking Hypotheses 2 and 3 together, we argue that perceived opportunities mediate the proposed negative relationship between age and entrepreneurial activity. More specifically, due to lower perceptions of entrepreneurial opportunities, older adults are likely to see themselves as less prepared for entrepreneurial activity. They view their time to achieve long-term goals as limited, and thus prefer to maximize present outcomes such as immediate financial returns (Lévesque and Minniti, 2006). In addition, older people have usually achieved their most important personal and business goals, such as desired income (Smallbone and Wyer, 2006), and therefore might not focus on opportunities as much as their younger counterparts. This corresponds to research findings by Curran and Blackburn (2001), which show that the main reasons against entrepreneurial activity for employees aged 50 to 75 are the uncertainty of income, feeling old, and missing job security. These changes in opportunity perception at higher ages are, in turn, likely to negatively impact entrepreneurial activity. In other words, older people should favor salaried employment yielding instant payoffs over long-term returns from business ideas that would yet need to be implemented.

The lifespan perspective (Baltes, 1987) proposes that various individual capabilities both increase and decrease at different rates over the course of time. One of the attributes that increases with age is crystallized intelligence (Salthouse, 2012). In regard to the work context, this increase in skills, knowledge, and experience might be attributed to higher job tenure and increased human, social, and financial capital. Thus, higher ages should relate to higher perceptions of entrepreneurial skills. The existence of agreement between broad actual skills and perceived skills was shown by Ackerman et al. (2002). They found that individuals generally tend to know their strengths and weaknesses in regard to different domains, including business and management.

Skills for entrepreneurship, however, do not solely comprise individuals’ crystallized cognitive abilities, as starting a venture requires the ability to recognize and exploit new opportunities, persistence to overcome potential obstacles, as well as physical strength and endurance to deal with likely stress in the starting phase and beyond. Based on the lifespan perspective, these other personal resources such as processing speed, memory, and physical endurance tend to decrease with age (Schulz and Heckhausen, 1996). As a consequence, people are less likely to be equipped with the necessary means to start new, future-oriented plans involving uncertainty as they age, leading to lower indications of perceived skills for entrepreneurship.

Moreover, with age people usually accumulate occupational, job, and organizational tenure, and therefore also higher task-related human capital, which could compensate for decreased physical abilities. However, when it comes to skills for entrepreneurship, physical abilities seem crucial. Thus, older adults might perceive that even though they have sufficient task-related human capital to work independently in their respective field of work, they may be relatively unequipped in regard to entrepreneurship (e.g., managing a business, handling stress, dealing with formalities). With increasing age, individuals might therefore perceive that they lack the skills relevant for entrepreneurship.

Conversely, younger adults might be biased regarding their own perceived skills and capabilities for entrepreneurship. As younger adults are more likely to be dissatisfied with the status quo (Charles, 2010), they are also more likely to suffer from self-enhancement bias (i.e., overconfidence), which builds on a personal need to increase personal satisfaction and self-worth (Lee and Im, 2007). We therefore assume that younger individuals are more likely to perceive themselves as having the necessary skills for entrepreneurship.

As mentioned before, one of the recognitions that change across the lifespan may be the perception of own skills for entrepreneurship. The perception of own skills itself is strongly related to self-efficacy, which describes an individual’s belief to be capable of performing a given task (Gist, 1987). As self-efficacy builds on an individual’s assessment of own resources (Ajzen, 1987; Gist and Mitchell, 1992), it also relates to beliefs about goal-attainment. These beliefs, in turn, play a crucial role in the development of intentions and actions (Boyd and Vozikis, 1994). More specifically, only if individuals believe in having the necessary skills to attain a goal will they act upon it (Bandura, 1991).

Previous research has demonstrated the importance of self-efficacy, and thus the perception of skills, for entrepreneurial intention (e.g., Boyd and Vozikis, 1994; Chen et al., 1998; Zhao et al., 2005; Wilson et al., 2007). In a study by Chen et al. (1998), for example, an individual’s confidence in the ability to master entrepreneurial roles and tasks related positively to start-up intentions. Similarly, Zhao et al. (2005) found that the desire to become an entrepreneur is grounded in high entrepreneurial self-efficacy. As perceived skills for entrepreneurship resemble entrepreneurial self-efficacy (Bandura, 1993), we build on the aforementioned theorizing by proposing that perceived skills increase entrepreneurial activity itself, rather than merely intentions.

Taking the rationales for Hypotheses 5 and 6 together, we propose that perceived skills also mediate the relationship between age and entrepreneurial activity. First, age should be negatively associated with perceived skills due to certain age-related cognitive and physical declines. Second, this perception of having the necessary skills for entrepreneurship, or self-efficacy, should relate positively to entrepreneurial activity, as these perceptions are the foundation for goal pursuit.

Data for this study were based on the Global Entrepreneurship Monitor (GEM) from 2013, as it is the most recent publicly available dataset that is most likely to reflect current developments in entrepreneurship. The GEM is a global, standardized survey study. In 2013, it was administered to a representative sample of adults aged 16–98 (M = 40.55, SD = 14.19) in each participating country¹ (N = 70), yielding a cross-country total of 244,471 participants. In each country, data collection was completed by professional firms, which were supervised by an academic or research institution. The project is coordinated by the Global Entrepreneurship Research Association (GERA), which monitors the data collection and secures standardization and international comparability of the data collection. The GEM data is frequently used in academic research, as it provides a “major database for internationally comparative entrepreneurship” (Bergmann et al., 2014, p. 1). The benefits of GEM data are its universality and comparability, which are based on the large number of national level observations which are comparable across countries (Ho and Wong, 2007; Langowitz and Minniti, 2007).

All procedures performed in this study were in accordance with the ethical standards of the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Because this study used existing data, which was collected by “National Teams” led by academic or research institutions that collected and reported the data, the GEM project was subject to their respective, ethical standards. Moreover, the data collection included a confidentiality note, so that no university ethical approval was required.

For this research, we excluded 3,661 participants (1.5%) due to missing values in some of the core variables, leading to a final sample size of 240,810 participants. Thereby, the sample sizes within the countries ranged from 33,287 in Turkey to 2,000 in Greece, Hungary, Norway, Poland, Malaysia, Singapore, Japan, South Korea, Lithuania, Latvia, Croatia, Macedonia, and Puerto Rico.

Except for chronological age, all of the measures yielded binary answers (gender: 0 = female, 1 = male; all remaining variables: 0 = no, 1 = yes). Moreover, for reasons of practicability, most measures consisted of a single item. Yet, we argue that they are still valid measures, as the constructs they measure are rather clear and homogeneous. For example, Fisher et al. (2016) concluded that many single item measures can be used to assess several psychological constructs (e.g., work centrality, job control, life satisfaction) in a reliable and valid way.

We used logistic path analysis in Mplus (Muthén and Muthén, 2012) to test our hypotheses. Due to the large amount of data coming from different countries, we first checked for random slopes to determine the type of analysis using Mplus. As there was not sufficient variation between countries, we continued with the analysis using fixed slopes. Specifically, we proceeded with the main analysis on the within-level (i.e., person-level) only. Even though we hypothesized linear relations, we included age-squared² in our analysis to check for curvilinearity to account for potential non-linear relationships, as recommended by Bohlmann et al. (2017).

As the dependent variables were binary, the WLSMV estimator was used to estimate both main and indirect effects. Its robustness with non-normally distributed variables has been shown previously, and makes it the best option for modeling ordered or categorical data (Brown, 2006). This model builds on the principles of general linear models, but allows for a better account of dichotomous dependent variables. Moreover, probit models extend the standard log-linear model, thereby allowing for mixtures of both categorical and continuous independent variables and their relation to a categorical outcome. Another advantage of probit models is that the resulting regression coefficients can be interpreted as the change in probabilities when the binary variable changes from 0 to 1 (Langowitz and Minniti, 2007).

Differences between countries were accounted for by using “TYPE = COMPLEX” to ensure the validity and reliability of the results. In the first model, we examined the direct effect of age and age-squared on entrepreneurial activity while including gender as a control variable. In the second model, we added the two mediator variables to examine mediation effects (i.e., ran a single model with multiple mediators). To further investigate the mediation effects, we checked whether the significant, direct effects of age on entrepreneurial activity became insignificant after including the mediators to the model (i.e., full mediation), or whether direct, indirect, and total effects were all significant (i.e., partial mediation).

In this article, we adopted a lifespan perspective to uncover how age relates to entrepreneurial activity through individuals’ perceptions of entrepreneurial opportunities and skills. In accordance with the hypotheses, results showed that there is a negative relation between age and entrepreneurial activity. Moreover, this relation was partially mediated by perceived opportunities and perceived skills for entrepreneurship. Specifically, age related negatively to both perceived opportunities and skills. Perceived opportunities and skills, in turn, related positively to entrepreneurial activity.

The relations between age and entrepreneurial activity, as well as between age and both perceived opportunities and perceived skills, were weakly curvilinear (i.e., inversely U-shaped). While this was not part of our hypotheses, the decline in entrepreneurial activity with age might be less steep in middle adulthood, as these individuals are less likely to be aware of their future time perspective due to higher commitments in regard to both career and family, as compared to younger or older adults. Moreover, the steeper decline in perceived skills in later years might be due to the fact that the decline in both cognitive (Verhaeghen and Salthouse, 1997) and physical abilities (Spirduso et al., 1995) accelerates in late adulthood.

While the results are of statistical significance, they are practically relevant as well. More specifically, the pseudo R-square indicated that 40% of the variance in entrepreneurial activity can be explained by the probit analysis (McKelvey and Zavoina, 1975). Therefore, the results are a first step toward a more comprehensive conceptualization of the entrepreneurial process from a lifespan perspective. They are especially important as previous research in entrepreneurship has assumed an influence of age, but mostly controlled for it, instead of adding it as a substantive predictor of entrepreneurial activity (Schjoedt and Bird, 2014). Consequently, the findings have important implications for entrepreneurship theories. First, major theories have conceptualized entrepreneurial activity as timeless. For example, according to the theory of planned behavior (Ajzen and Fishbein, 1977), attitudes toward a given behavior, subjective norms about the appropriateness and desirability of the behavior, as well as perceived control about the behavior determine the exertion of a given behavior, such as entrepreneurial activity. Yet, the current study confirmed the importance of age by showing that entrepreneurial activity changes over time and at different rates. Second, by including perceived opportunities and skills as mediators, the present study hints at the ways through which age exerts its influence. Current theories used in entrepreneurship (e.g., the theory of planned behavior; Ajzen and Fishbein, 1977) can therefore not accurately predict entrepreneurial activity, as they do not account for changes in individual skills and opportunities that are due to age and age-related characteristics. It is thus crucial to amend current models of entrepreneurship by adding age as a predictor. However, it has to be noted that age itself cannot be directly related to entrepreneurial activity, as it is mostly a proxy variables that helps to measure time-related changes as people age (Wohlwill, 1970). Thus, it is important to investigate mediators, such as perceived opportunities and perceived skills for entrepreneurship.

Next to the theoretical implications, the results have practical implications as well. Findings suggest that younger and older adults need to be supported differently throughout the entrepreneurial process. While younger adults have less difficulty to perceive opportunities and acknowledge that they are skilled enough to act entrepreneurially, older adults may be constrained by a lower future time-perspective and declining physical and fluid cognitive abilities, and thus less likely to perceive opportunities and skills. It might therefore be important to help younger adults acquire the relevant skills to act on their identified opportunity, while older adults might need help finding opportunities that require less time to reach the desired goal. This knowledge is especially important for governments and educational institutions, as these usually employ training programs to foster entrepreneurship (Obschonka et al., 2010; Halvorson and Morrow-Howell, 2016). These programs currently do not accurately take the needs and motivations of different ages into account.

A central limitation of this study is the cross-sectional nature of the data, which does not allow for a separation of cohort effects and age-related change (Schaie, 1965). Moreover, there may be reverse effects, as the definite direction of the hypothesized relations can only be confirmed in a longitudinal study. It might thus be that entrepreneurial activity increases perceived opportunities (i.e., because people are in the entrepreneurial mindset once they engaged in entrepreneurship and therefore constantly perceive new opportunities) and perceived skills (i.e., because performance can enhance self-efficacy perceptions; Sitzmann and Yeo, 2013). However, cross-sectional studies still provide valuable information on systematic age differences (Lévesque and Minniti, 2006; Kautonen et al., 2010; Gielnik et al., 2012; Caliendo et al., 2014; Heim, 2015) that can be amended and replicated in other studies employing different designs.

Another limitation is the binary nature of most study variables, which might restrict a continuous assessment of the respective variables. However, the aim of the global research project (GEM) is to collect representative data in as many countries as possible, which requires keeping the questionnaire relatively short and avoiding answer options that might lead to translation errors or cultural biases. Moreover, the validity of measures used has been established in previous studies (Bergmann et al., 2014). Thus, results can still yield preliminary insights on what influences the changing relation between age and entrepreneurial activity.

To validate results of the few studies that have yet investigated relationships between age and entrepreneurship, future research should employ different designs that can help to determine both developments over time as well as causality (e.g., by using longitudinal or experimental designs; Bohlmann et al., 2017). While opportunities and skills are some of the concepts through which age and age-related characteristics impact the entrepreneurial process, future studies need to research other individual and contextual factors that play a role for entrepreneurship while being influenced by age. For example, investigating whether age-differences are the same or different across cultures can help to isolate developmental change across the lifespan from culture-related cohort effects (McCrae et al., 1999). Other examples are push factors of entrepreneurship, such as unemployment in older age (e.g., Faria et al., 2010).

Moreover, the time remaining relative to one’s age is not only shaped by individual, chronological age, but also tied to the average life expectancy of a given country (Seijts, 1998; Griffin et al., 2016). It is therefore important to consider variations in life expectancies across countries as an influence of an individual’s future-time perspective when investigating the impact of age on perceived opportunities and, in turn, entrepreneurship. More specifically, if the life expectancy in a given country is relatively short, the future time perspective of a respondent from this country is likely to be lower compared to a respondent from a country with the same age and a higher average life expectancy. In regard to an individual’s context, cultural variables that merge characteristics of regional infrastructure as well as the social and economic environment are likely to shape entrepreneurial activity as well. Thereby, factors such as institutional support or aging stereotypes seem to be especially impactful (e.g., Delmar and Davidson, 2000; Thomas and Muller, 2000; Kennedy et al., 2003).