Entrepreneurship, age, and social value creation: A constraint-based individual perspective

ABSTRACT

We advance research on social entrepreneurship by offering a constraint-based individual perspective of “who” (gender, education) chooses to create social value “when” in their life course (proxied by age). Integrating predictions from situational strength theory in psychology and the life course perspective in sociology, we theorize that resource constraints determine at what age entrepreneurs are likely to prioritize social relative to economic value creation goals when starting their enterprise. We examine the intersection of entrepreneur age with gender and education to account for distinct patterns of resource constraints over the life course. Multilevel analyses of nationally representative samples of 5,251 new entrepreneurs from 44 countries reveal a robust curvilinear (U-shaped) relationship between age and social value creation and a steeper U-curve for more highly educated women. Our study offers a springboard for future entrepreneurship research considering individuals’ constraints on prosocial value expression by applying intersectional analyses.

KEYWORDS:

• Social entrepreneurship
• social value creation
• life course perspective

Acknowledgments

The authors are grateful for the valuable contributions and involvement of Jan de Kok on earlier versions of this paper. We also acknowledge funding for one of the authors from European Commission’s FP7 Framework Programmes for Research Technological Development, and Demonstration under the grant agreement no 613500 (SEFORIS project).

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary Material

Supplemental data for this article is available online at https://doi.org/10.1080/00472778.2022.2133128.

Notes

¹ While somewhat overlapping with Brieger et al. (2021) in finding a U-shaped relationship between age and social entrepreneurship, our study extends their research by considering the interplay of individual characteristics of age, gender, and education. By contrast, drawing on institutional theory, Brieger et al. (2021) focus on country-level institutions as a boundary condition of the age-social entrepreneurship relationship.

² While our independent variable is age, we refer to three specific age ranges in line with life course research to introduce our rationale and explain and discuss our findings. We refer to early adulthood as 18–35 years of age, midlife as 35–50 years of age, and mature adulthood starting at 50 years of age, drawing from the 12 stages proposed by Armstrong (2007). We focus on the working population with an age range of 18–64, thus only examining mature adulthood to 64 years of age, whereas Armstrong also covers older adults up to 80 years of age in the mature adulthood classification.

³ By contrast, personality traits are typically defined in terms of relatively stable patterns of behavior, thoughts, and emotions (Parks-Leduc et al., 2015). However, research suggests that personality traits, such as the Big Five, may well shift over time (Specht et al., 2011, 2014).

⁴ While the 2015 GEM study includes social entrepreneurship as a topic, it does not provide the necessary data to measure social value creation for new entrepreneurs.

⁵ Table S1, available in the online supplementary information on the journal’s website, presents the 44 countries included in the estimation sample and the number of observations per country.

⁶ We check for sample selection bias by incorporating an additional selection equation, that is, we define a variable S taking a value of 1 if the individual is a new entrepreneur, and a value of 0 if the individual is not involved in new entrepreneurship. To identify the sample selection model, we need to include a variable that is related to S but not to social value creation. Based on earlier research (Estrin et al., 2016), we select the variable indicating whether an individual personally knows someone who started a business in the past two years. We find nonsignificant correlations between the selection and outcome equations, and hence, no evidence of sample selection bias.

⁷ We constructed a nascent entrepreneur subsample according to the GEM definitions. We first considered respondents who answered one of the following two questions in the affirmative: (1) Are you, alone or with others, currently trying to start a new business, including any self-employment or selling any goods or services to others? (2) Are you, alone or with others, currently trying to start any kind of activity, organization, or initiative that has a particularly social, environmental, or community objective? In addition, respondents had to satisfy the following criteria: having taken concrete steps to start the business in the past 12 months, being the owner (partly or fully) of the business, and making any financial payments to the enterprise for less than three months.

⁸ This means that in quadratic specifications the coefficients of the linear age terms now indicate the slope of the U-shape at age 18, which is relevant in the context of one of the Haans et al. (2016) tests that we conducted. The alternative mean-centered data (centered on the mean age in the sample) confirms that the results lead to identical conclusions about the U-shape and hypotheses. This is to be expected, as “ … results obtained with centered data and raw data are mathematically equivalent and mean-centering does not increase the power to detect quadratic or interaction effects” (Haans et al., 2016, p. 1184).

⁹ We excluded household income as a control variable from the main analysis due to the high and nonrandom pattern of missing values. Nevertheless, household income was included in one of the robustness checks.

¹⁰ Using GDP per capita in purchasing power parity does not lead to qualitatively different conclusions.

¹¹ To verify the normal distribution of the residuals in a model including all individual-level and country-level (control) variables, we calculated the standardized residuals at the individual level and plotted these against their normal scores; the resulting graph indicates normality. Second, we plotted the residuals against the predicted values of the outcome variable using the fixed part of the model for the prediction. We do not find an indication of violation of homoscedasticity in this scatter plot.

¹² Figure S1, available on the journal’s website as part of the online supplementary information, shows the empirical distribution of the dependent variable categorized into 10 intervals because multiples of five (0, 5, 10, …) occurred more frequently than the remaining values.

¹³ Naturally, the VIF values for age squared and for the interaction terms (not included in Table 1) are higher. Note that we performed additional analyses using mean centered data, and although this substantially reduces the VIF values, our conclusions about the U-shapes and hypotheses remain identical.

¹⁴ The difference in social value creation between individuals can be substantial. For example, the difference in points allocated to social value creation between a 45-year-old and a 55-year-old new female entrepreneur, based on Figure 4, is 3.6 points (42.4–38.8), which amounts to an increase in social value creation of about 9%. The difference between a 55-year-old and a 64-year-old female entrepreneur is as high as 6.1 points (48.5–42.4), a 14% increase.

¹⁵ The predicted value is based on the estimated coefficients of the fixed part of the model (random effects set equal to their expected mean value zero).

¹⁶ Table S2 and Figure S2, are available in our online supplementary information on the journal’s website, and present full results for the three levels of education.

¹⁷ Table S3 and Figure S3 are available in our online supplementary information on the journal’s website..

¹⁸ Table S4, controlling for household income, including imputed values, is available in the online supplementary information on the journal’s website.

¹⁹ Table S5, replacing the random effects model with a country fixed effects model, is available in the online supplementary information on the journal’s website.

²⁰ Tables S6 and S7, which substitute GDP with ln(GDP and the linear and quadratic terms for GDP, respectively, are available in the online supplementary information on the journal’s website.

²¹ Brieger et al., (2021) suggest that cohort differences in postmaterialism (Inglehart, 1977), explain the higher priority of social value creation in early adulthood while drawing on gerontology research of prosocial value shifts in extreme old age (Baltes et al., 2007) to explain the rise again in mature adulthood.

²² SHARE: http://www.share-project.org/home0.html and HRS: https://hrs.isr.umich.edu/about.

²³ https://ec.europa.eu/growth/industry/strategy/innovation/social/competition_en.

²⁴ https://www.ashoka.org/en-aaw.

Additional information

Funding

The work of the last author on this paper was supported by the Seventh Framework Programme [613500 (SEFORIS project)].