South Africa has a rich history of environmentalism, dating back as far as the mid-1800s when the Forest and Herbage Preservation Act (No 18 of 1856) was passed (McCormick 1989). This act was later modified and renamed the Forestry Act (No. 22 of 1888), which aimed at preserving the indigenous forests of specific areas within South Africa (Burgess 2001). It, however, wasn’t until the 1980s when environmental concern accelerated amongst South Africans. Steyn and Wessels (2000) speculate that the reasons for this may be attributed to the establishment of Earthlife Africa in 1988, the changing political atmosphere of the time, a lack of governmental action concerning environmental degradation and the shift towards broader, more inclusive environmental agendas amongst existing Environmental Non-government Organisations (ENGOs). After the fall of the apartheid regime in South Africa, environmental law gained a more significant position within the legislation in that environmental health itself became a fundamental human right. According to Section 24 of the Constitution (Republic of South Africa [RSA] 1996), the law states that everyone has the right:

For South Africa, preserving the natural environment is thus, not merely a suggestion or a ‘nice-to-have’, but is seen as a necessity for survival. Shapshay and Tenen (2018) even went as far as stating that the degradation of earth’s natural environment robs humanity of instrumental goods and a moral right. If the global crisis is to be avoided, then ‘business-as-usual’ practices will have to change.

Since 1992, several Acts regarding environmental management and regulation have been passed that have implications for organisations operating within South Africa. The National Environmental Act (NEMA), for example, permits legal action to be taken against any organisation that does not comply with the existing legal requirements (Department of Environmental Affairs 2019). More recently, the Carbon Tax Act (No. 15 of 2019) and the Good Green Deeds programme were passed, respectively, to lower greenhouse gas emissions. This was done by imposing a tax on polluters and encourage South Africa’s social and economic sectors to be more environmentally responsible (RSA 2019).

Further implications for South African business regarding environmental sustainability can be seen as stated by the King IV report for corporate governance that organisations are expected to act in a way that is responsible, ethical and sustainable towards society and the natural environment, thus ensuring that present needs are not met at the expense of future generation’s needs (IoDSA 2016). In addition to this, the Companies Act (No. 71 of 2008) Section 7 urges organisations to comply with the Bill of Rights as set out by the Constitution, which includes the preservation of individuals’ right to an ‘environment that isn’t harmful to their health or well-being’ and encourages efficient and responsible management of organisations (RSA 2009).

Apart from an organisations legal expectations to be sustainable, organisations have been pressured by stakeholders, consumers, competitors and regulators to reduce their carbon footprint by having more environmentally friendly production processes (Yenipazarli 2019). Doran and Ryan (2016) suggest that integrating environmental innovation into business strategies could be leveraged as a means of gaining a competitive advantage whilst simultaneously addressing environmental problems. Sustainability and greening the business sector can thus be viewed as a possible business opportunity.

A single cross-sectional, descriptive, quantitative research approach was followed to collect the study’s required data. This was deemed the most appropriate means to quantify the data collected from a relatively large sample numerically.

A mix of convenience and judgement, non-probability sampling techniques, was incorporated to draw the study’s sample population. The original sample was made up of 600 university students of which the final usable number amounted to 522. The sample consisted of both male and females, between the ages of 18 and 33 from three registered public South African higher education institutions (HEIs). Considering the sample size for this study, the historical reference approach was used. This approach involved looking at similar studies and using its sample sizes as a guideline. Similar studies referred to and relating to entrepreneurial intention and environmental concern include Marire (2015) with a sample size of 400 students; Synodinos (2014) with a sample size of 500 students; Cheah and Phau (2011) with a sample size of 600 and Minton and Rose (1997) with a sample size of 500. The three HEIs comprised one technical university, one comprehensive university and one traditional university.

A structured, self-administered survey questionnaire was developed and utilised to gather information from sample participants. The questionnaire comprised three sections: Section A, which required participants to provide demographic information relevant to the study; Section B containing previously validated scales adapted from Lee (2008) and Mostafa (2007), which aimed to collect information regarding participants’ environmental concern, environmental behaviour, perceived effectiveness of pro-environmental behaviour and perceived environmental knowledge. Section C comprised previously validated scales adapted from Liñán and Chen (2009) and Fielding et al. (2008), which measured participants’ attitude towards ecopreneurship, subjective norm, perceived behavioural control and intention towards ecopreneurship. The validated scales that were used in both Section B and Section C can be found in Appendix 1.

The captured data was analysed using Statistical Package for Social Sciences (SPSS) version 26.0. The empirical data analysis methods included a reliability and validity analysis, a descriptive analysis, significance tests, internal consistency reliability, discriminant validity, nomological validity, factor analysis and regression.

Approval to conduct the study was obtained from the North-West University, Economic and Management Sciences Research Ethics Committee (EMS-REC), clearance number NWU-0114-19-A4. Before participating, students were informed of the study’s purpose and were assured that they would remain anonymous. Students completed the questionnaires voluntarily and were given the option to opt-out at any given point with no consequences.

Of the 600 questionnaires, approximately 560 were returned, indicating a response rate of 93%. However, once screening was undertaken, the actual response rate was 87% or 522 responses. That meant 38 questionnaires were disregarded in the editing and cleaning process as these questionnaires did not meet the study’ judgement criteria. These requirements included any participant that had not completed a minimum of 90% of the questionnaire, did not pass the screening question or was not a Generation Y cohort member. The demographic information of the participants is summarised in Table 1.

From Table 1, it can be noted that the majority of students indicated that their field of study is in economics and management sciences (44%), they are in the process of obtaining an undergraduate degree (91%) and they are under the age of 25 years (93%) and primarily speak English (21%) or Afrikaans (20%). The gender distribution was relatively equally dispersed between females (51%) and males (46%), with 3% of respondents not disclosing their gender. The demographic information relates to the demographics of the larger South African population in that each of the 11 national languages of the country was represented in the study. Once the questionnaires had been screened and demographics computed, the data set was subjected to an exploratory factor analysis to assert the data’s factorability.

The study implemented an exploratory principal components analysis using Promax rotation to determine whether any item questions were cross-loaded or loaded onto an incorrect factor. The first step entailed performing the Kaizer-Meyer-Olkin (KMO) test of sampling adequacy and Bartlett’s test of sphericity. The KMO test yielded a result of 0.951, and Bartlett’s test chi-square of 11 311.56 with df = 496 and significance at p = 0.000 < 0.05. The outcome of these statistics was deemed satisfactory, thus establishing the data’s factorability (Pallant 2010).

The next step was to assess the communalities originating from the exploratory principal components analysis. According to Hair et al. (2014), items yielding values below 0.50 are excluded from the final factor analysis. Based on the extraction results, seven factors explained 70.2% of the total variance in the data set. Problematic cross-loadings were observed for item B4 in the environmental concern construct ‘I often think about how the environmental quality can be improved’ (Lee 2008). Upon further investigation, it was determined that the item could be deleted without changing the construct’s intended purpose. As a result, the environmental concern construct only constituted items B1–B3, and item B4 was not included in further data analysis.

Moreover, problematic cross-loadings were observed for the perceived effectiveness of the environmental behaviour construct (B13–RB16). Two of the items in the construct loaded on their own factor, and two of the items cross-loaded onto other constructs. A preliminary Cronbach alpha test of the construct was conducted, which yielded unsatisfactory results below the recommended cut-off level. Since the perceived effectiveness of environmental behaviour construct constitutes a supporting rather than an essential role in determining Generation Y students’ intentions towards becoming ecopreneurs, it was omitted from further data analysis.

A rerun of the exploratory factor analysis was performed, excluding items B4 and B13 to RB16. The new factor analysis produced a KMO of 0.949 and Bartlett’s test score of 10 947.1 with df = 465 and significance at p = 0.000 < 0.05. Based on the new extraction results, seven factors explained 70.7% of the total variance. Table 2 summarises the rotated factors from the pattern matrix and factor communalities and extracted percentage variance.

As seen in Table 2, factors aligned with the relevant constructs showed that the variables loaded as expected, and that all items achieved values above the recommended 0.50 cut-off mark. Based on the new extraction results, the seven factors explained 70.7% of total variance within the accepted range, higher than 60% (Hair et al. 2014). The data set was then subjected to reliability analysis with the conclusion of satisfactory exploratory factor analysis.

It is essential for a measurement to be trustworthy and to measure accurately that which it sets out. For this reason, the reliability of the measurement scales was analysed both in the pilot test and the main study. The Cronbach alpha values recorded for each construct, based on this study’s survey data, are outlined in Table 3.

As depicted in Table 3, the construct with the highest Cronbach alpha value is the intention towards ecopreneurship construct (0.939), which is similar to the reported value (0.943) in the study from which the intention towards entrepreneurship construct was adapted from (Liñán & Chen 2009). Compared to their original sources, all the constructs in this study had lower Cronbach alpha values, except the perceived behavioural control and the environmental behaviour constructs. In this study, the perceived behavioural control construct yielded a value of 0.907, whereas Liñán and Chen (2009) reported a value of 0.847. In the case of the environmental behaviour construct, a value of 0.847 was recorded for this study, whereas Lee’s (2008) study recorded a value of 0.7. Interestingly, the higher Cronbach alpha values are both related to respondents’ behaviour.

Conversely, the construct that yielded the lowest Cronbach alpha value (0.532) was the perceived effectiveness of environmental behaviour. This is much lower than the value (0.710) recorded for this construct by Lee (2008). Considering the low item-to-total correlation value of this construct (0.304), excluding this construct from the main study results was advisable. Further investigation proved that merely removing certain items from the construct did not improve its overall internal consistency reliability. One explanation for the low value of the perceived effectiveness of environmental behaviour construct could be that reverse-scored items, or items stated in the negative form, often yield significantly lower Cronbach alpha values than regular statements (Schriesheim et al. 1991). Satisfactory Cronbach alpha values were recorded for the remaining constructs (Hair et al. 2014).

Participants were asked to respond to statements using a six-point Likert scale in Sections B and C of the questionnaire. A response of one indicated that the participant strongly disagreed, whereas six suggested that the participant strongly agreed with the statement. Reported mean values higher than (Mean = 3.00) would indicate a positive attitude towards the construct, whilst mean values that were lower than (Mean = 3.00) would indicate a negative attitude towards the construct. Therefore, a higher mean value signifies a higher level of agreement to the statements in the construct, and a lower mean value indicates a lower level of agreement. Table 4 reports the mean statistics, standard deviation, skewness and kurtosis values of each scale used in the measuring instrument.

From Table 4, it can be derived that participants generally displayed high levels of agreeability towards each construct as all constructs yielded mean values greater than 3. The degree to which participants agreed varied, however. The environmental concern construct yielded the highest mean value (Mean = 4.36), followed by the environmental knowledge construct (Mean = 4.03). Conversely, the lowest mean value was recorded for the perceived behavioural control construct (Mean = 3.34), closely followed by the intention towards ecopreneurship construct (Mean = 3.30). Generation Y students’ knowledge regarding environment and concern for the environment is thus relatively high compared to the other constructs. Their intention towards ecopreneurship and perceived behavioural control is relatively low compared to the other constructs.

Thus, it can be deduced that Generation Y students in South Africa display high levels of knowledge related to the environment and seem to be concerned about the deteriorating state of the environment and indicate that they advocate pro-environmental behaviour to some extent. Furthermore, Generation Y students can also be said to have positive attitudes towards ecopreneurship, consider ecopreneurship as something their peers or influencers would approve of, perceive themselves to have reasonable behavioural control to some degree and appear to show positive intentions towards becoming ecopreneurs. However, when it comes to their perceived behavioural control and intention towards ecopreneurship, Generation Y students display lower levels of agreeability than they do for the other constructs. Ntanos et al. (2018) and Williams (2017) have reported similar findings regarding Generation Y students displaying high environmental concern and knowledge levels. Thondhlana and Hlatshwayo (2018) also found that many university students within the South African context do not always behave in a way that could be considered pro-environmental. The main reasons for this, as indicated by the students, were laziness, forgetfulness, comfort and convenience, in that particular order.

With reference to the normality of the data set, no irregularities were identified as all kurtosis and skewness values fall in the valid range of -2 and +2, as such the data can be deemed as normally dispersed (Berndt & Petzer 2011). The negatively skewed values indicate high scores in the distribution; in other words, the mean fell within the agreement area of scale responses. The standard deviation values recorded for each construct indicate variability in participants’ answers to items in that scale. The lowest standard deviation value (standard deviation [SD] = 0.95) was recorded for the environmental knowledge construct, exhibiting low variability of responses. Conversely, the highest standard deviation value (SD = 1.32) was recorded for the intention towards ecopreneurship construct, indicating larger variability of responses (McDaniel & Gates 2013:458).

A correlation analysis serves to determine whether a linear relationship exists between variables. Constructing a correlation matrix is also a useful method of determining the nomological validity of constructs (Hair et al. 2010). This study utilised Pearson’s Product-Moment correlation coefficients to conduct the correlation analysis. The significance level was set at p ≤ 0.01. Table 5 outlines the correlation matrix, which reports on the correlation between each of the constructs presented.

As is seen from Table 5, all corresponding constructs yielded a positive significant correlation coefficient amongst one another (p ≤ 0.01). According to the results, nomological validity is thus established as all measurement constructs are positively correlated (Berndt & Petzer 2011). Some constructs, however, display stronger relationships with each other. The two weakest correlation coefficients were both in relation to the environmental concern construct, the weakest being perceived behavioural control (r = 0.348, p < 0.01) and the second weakest, intention towards ecopreneurship (r = 0.389, p < 0.01).

Contrastingly, the strongest two correlation coefficients were both in relation to intention towards ecopreneurship, the strongest being perceived behavioural control (r = 0.751, p < 0.01) followed by attitude towards ecopreneurship (r = 0.691, p < 0.01). Liñán and Chen (2009) report similar findings; however, in their study, the strongest correlation was between intention towards entrepreneurship and personal attitude, and the second strongest was between the former and perceived behavioural control. A positive relationship, therefore, exists between Generation Y students’ concern for the environment, environmental knowledge, pro-environmental behaviour, attitude towards ecopreneurship, subjective norm, perceived behavioural control and their intention towards ecopreneurship.

The findings of this study mainly contribute to the limited available literature in the field of ecopreneurship, as this area is still in its infancy and lacks a broad empirical foundation (Farinelli et al. 2011). Furthermore, this study’s results also contribute to the broader fields of the Generation Y cohort, that is, higher education students, environmentalism and entrepreneurship as the empirical objectives that were achieved and provide valuable findings and information on each of these topics. Educational institutions can utilise the results of this study to expand the study field of entrepreneurship to include the niche field of ecopreneurship.

The findings of this study provide insight into students’ intentions towards becoming a solution to environmental problems through ecopreneurship. Therefore, the results are valuable to all who seek to develop entrepreneurs who wish to solve environmental issues through innovation or simply manage an organisation in an environmentally sustainable way. This study contributes explicitly to the literature within a South African context; however, the findings could also be used for cross-cultural or international type studies.

Six main recommendations are proposed based on the results from this study.

Great care was taken to ensure that the sample would be adequately represented in the study. However, limitations such as sampling bias could still arise when generalising convenience sampling results to the population. In addition to this, using a single-cross sectional research design could also be limiting as it only represents a single point in time instead of multiple observations made over a more extended period.

It would be feasible to use qualitative, mixed-method, longitudinal designs, panel data analyses or probability sampling techniques for future research on the topic of Generation Y students’ intentions towards ecopreneurship. Moreover, the study could be repeated with another sample, and the data analysis could be extended to include further Confirmatory Factor Analysis (CFA) and Structural Equation Modelling (SEM). Possible future studies on the subject could also include developing a process to categorise and define different types of ecopreneurs based on their position on the ecological orientation continuum and exploring the incentives and barriers for ecopreneurship within a South African context.