The current study is underpinned by work/family border theory, which was proposed by Clark (2000). The work/family border theory was developed to address the issues of crossing borders between spheres of life, particularly the spheres of work and family life (Kossek & Lautsch, 2018). On the basis of the work/family border theory, individuals are viewed as border-crossers as they continuously manage and negotiate work and family spheres or the borders between them in order to create and maintain a desired work–family balance. According to Clark (2000), a desired work–family balance is one that provides satisfaction and good functioning at both work and home with the least degree of role conflict (Karassvidou & Glaveli, 2015). According to Clark (2000), each person’s role in society is within a specific sphere of life, and these spheres may be separated by physical or psychological boundaries. Physical boundaries are artificially occurring barriers between the workplace and the family. Psychological boundaries are procedures established by employees that can determine when to think and when behaviour and emotional patterns are suitable for a particular domain and vice versa. According to Clark (2000), the level of integration, the ease with which transitions are made and the level of conflict that exist between these borders are all affected by the flexibility and permeability of the borders. Clark (2000) maintained that work and family spheres are not separated but are connected with flexible and permeable boundaries.

Clark’s (2000) work/family border theory is based on the assumption that work and non-work domains are two separate spheres of life that affect each other. The relationship between these spheres forms a continuum that ranges from segmentation to integration (Karassvidou & Glaveli, 2015). At the pole end of segmentation, the two spheres are mutually exclusive, whereas at the pole end of integration, they are conceived as identical spheres. According to Karassvidou and Glaveli (2015), the more flexible and permeable the line between the two spheres (weak borderline), the higher the integration between them. Here, the individual hardly sees a distinction between what belongs to work and what belongs to home. This blurring condition usually creates work–family balance conflict. As noted by Karassvidou and Glaveli (2015), ‘if the two domains are closely integrated, it is easier for transmission to occur but this has the tendency of creating work-family conflict’ (p. 86). Alternatively, where the line between the two spheres is impermeable and inflexible (strong borderline), segmentation usually results, deterring physical and behavioural elements from passing from one sphere into another (Kossek & Lautsch, 2018). Flexibility and permeability of work and non-work borders can increase a person’s work–life balance (Kossek & Lautsch, 2018). They may also lead to a conflict when people become confused about the role they should be playing at any given time.

The work–family border theory applies to a variety of policies such as job sharing, flexible working hours and home-working that are often used in organisations to assist workers to balance competing demands from work and personal life (Itzkovich et al., 2020). These arrangements may improve the employees’ sense of control over the extent of segmentation or integration they desire (Kossek & Lautsch, 2018). The theory has been criticized for being too simplistic as it exclusively focuses on work and family borders without giving much attention to the borders of the other important non-work roles such as leisure, social, religious and recreational works that happen within a person’s life. Electronic gadgets were not also included in the theory as a component that can affect work–life balance. Electronic gadgets can create permeability by acting as a link between different aspects of life. The use of technological gadgets produces an environment that is neither work nor non-work (Kossek & Lee, 2017).

The work–life border theory is relevant for this study because it applies to a variety of policies such as job sharing, flexible working hours and home-working that are often used by management to assist workers to balance competing demands from work and home.

Three concepts were relevant for this study: (1) life-friendly HR policies, (2) work–life balance and (3) employee engagement.

Life-friendly HR policies are policies that allow employees to balance their work and non-work life roles (Yuile et al., 2012). Life-friendly HR policies are usually grouped into three: (1) flexible work arrangements, (2) leave policy options and (3) caretaking benefits (Yuile et al., 2012). Flexible work arrangements are human resource policies that provide employees with an option to make choices influencing when, where, and for how long they should engage in a work-related task. Maxwell et al. (2007) defined flexible work arrangements as ‘any policy, practice, formal or informal, which permits people to vary when and where work is carried out’ (p. 138). With flexible work arrangements, employees are better able to work outside the traditional confines of the organisation. There are several forms of flexible work arrangements, such as compressed workweek, telecommuting, part-time work, job sharing and shift work (Austin-Egole et al., 2020).

The second broad life-friendly HR policy that helps the employee to ensure an appropriate balance between work and non-work obligations is the leave policy option. Naithani (2010) defined a leave policy as the sum of rules and guidelines that are designed by management to provide regular break periods for their staff. These rules or guidelines include parental leave, paid annual leave, sick leave, sabbatical leave and maternity leave. Organisations that provide leave to employees are committed perceived as being committed to helping their staff to achieve a good work-life balance (Austin-Egole et al., 2020; Maxwell et al., 2007). Caretaking benefits are the third type of life-friendly HR policy, which can help employees to achieve desirable work and family life balance. Carers’ arrangement, as they are often called, include onsite/offsite childcare, eldercare initiatives, family support programmes, counselling services and health programmes (Austin-Egole et al., 2020).

Work–life balance defies any universally accepted definition. Scholars have defined the concept in diverse ways. For instance, Kirchmeyer (2000) defined the concept of work–life balance as having rewarding experiences both at work and at home. Clark (2000) also defined work–life balance as the enjoyment and satisfaction received from work and home with the least amount of disagreement. Again, Greenhaus and Powell (2003) defined work–life balance as a process by which a person is equally interested in and content with his work–family roles. Greenhaus et al. (2006) further provided a revised definition of work–life balance in 2006. According to the authors, work–life balance is the process by which a person’s efficiency and fulfilment in work–family duties are congruent with their life roles at any given moment. Mengistu (2020) defined work–life balance as the accomplishment of role-related expectations that are negotiated and shared between a person and his or her role-related partners in the office and in the family. According to Lakshmi and Gopinath (2013), although the presence of numerous definitions may provide an adequate understanding of the concept, the plethora of definitions may also be a source of misunderstanding among scholars.

Previous studies (John et al., 2020; Mengistu, 2020; Nathani, 2010; Omar et al., 2015) have acknowledged the main factors that influence employees’ work–life balance. Nathani (2010) grouped the factors into three, namely work-related factors, family and personal related factors and other factors. Nathani defined work-related factors as any activity that affect work either positively or negatively. Examples include supervisor support, good workplace policies and good working conditions. Family and personal related factors are the factors that affect family and personal life either positively or negatively (Omar et al., 2015). Examples include family support, parental support and spousal support.

Mengistu (2020) also grouped the determinants of work–life balance into two – demand predictors and resource predictors. Demand predictors are those factors that affect work–life balance negatively. Examples include work and non-work role overloads. According to Mengistu (2020), work overload arises when the entire job demands on time and energy connected with numerous positions are too great to fulfil efficiently and comfortably. Omar et al. (2015) opined that demand predictors such as work overload and role conflict are the dominant issues affecting work–life balance. The resource predictors, on the other hand, are defined by Mengistu (2020) as the factors influencing work–life balance positively, including social support, family support and flexible policies. Resource predictors are those organizational resources or conditions that support employees in accomplishing their task.

Schaufeli and Bakker (2010) defined employee engagement as the active and positive work-related state that is characterised by three main factors: vigor, dedication and absorption. Vigor measures the levels of energy and resilience that employees put in their work while dedication refers to employee’s involvement in the work itself. The concentration and enthusiasm manifested by employees at the workplace is known as absorption. Saks (2006) identified two forms of employee engagement: (1) job engagement and (2) organizational engagement. Job engagement consists of the psychological state of fulfillment with one’s task while organizational engagement consists of employee’s physical, cognitive and emotional investment in the work. Fully engaged workers are physically energised, emotionally connected, mentally focused and feel aligned with the purpose of the organisation (Schaufeli & Bakker, 2010). As noted by Motyka (2018), engaged employees have a sense of personal attachment to their work and the organisation; they are motivated and able to give off their best to help the organisation succeed and from that flows a series of tangible benefits for the organisation and employees alike.

There are four major determinants of employee engagement: (1) perceived organisational support, (1) positive organisational politics, (3) job fit and (4) psychological climate (Motyka, 2018; Saks, 2006). Perceived organisational support is described by Eisenberger et al. (1986) as the employees’ ‘beliefs concerning the extent to which the organization values their contribution and cares about their well-being’ (p. 501). A study by Shaw et al. (2013) has shown that employees who demonstrate high level of perceived organisational support do not usually become stress at work. Another determinant of employee engagement is positive organisational politics. Donald et al. (2016) defined organisational politics as the means by which employees influence their colleagues, subordinates and even superiors in order to obtain personal benefits and or satisfy organisational goals. This definition is based on the traditional view of organisational politics whose research concentrated mostly on examining the attempts made by employees to influence the behaviour of others or the organisation as a whole (Donald et al., 2016). Recent studies (Donald et al., 2016; Motyka, 2018), however, focus more on how people perceive political manoeuvres arguing that the notion of politics rests on the perceptions employees hold about the concept rather than the actual influence tactics.

Person-job-fit is another determinant of employee engagement. According to Hoffman and Woehr (2006), person-job-fit relates to the congruence between employee characteristics such as knowledge, skills, abilities, needs and the demands or attributes of the job. In the opinion of the authors, good job-fit helps employees to be involved in a meaningful work, which in turn affects the development of work-related employee attitudes, such as job satisfaction, job involvement and organizational commitment. It also provides the cognitive stimulus that employees require to engage in positive organisational outcomes, such as improved performance, intention to stay and affective commitment. Thus, by implication, employees with good job fit are more likely to perform their jobs with enthusiasm and energy. Alternatively, low job-fit usually results in decreased productivity, decreased job satisfaction and employee engagement and increased levels of turnover (Hoffman & Woehr, 2006).

The fourth determinant of employee engagement is psychological climate. Kataria et al. (2013) defined psychological climate as an individual attribute, measured in terms of employees’ perception and interpretation of their organisational environments or the policies and practices adopted by management of the organisation. Employees feel safe and available when there is a good psychological climate, and this allows them to invest in their work roles without fear of any adverse criticism or career sabotage. When employees perceive favourable psychological climate, they are likely to be motivated, with the results that they will be prepared to invest their personal energies into their respective work roles. Thus, employees with improved psychological climate are more likely to engage in extra in-role discretionary effort, mediated by engagement in work.

There are three levels of employee engagement in the literature – highly engaged, not-engaged and the actively disengaged (Gallup, 2006). Highly engaged employees perform their work with passion. They perform their task with great enthusiasm and pleasure. They are naturally curious about their organisation and would indulge in productive and innovative ways to direct the organisation in the path of development and success. Not-engaged employees just concentrate only on their tasks and its outcomes without heeding to the organisation and its present position in the market. According to Gallup (2006), actively disengaged employees simply walk through their workday, putting time – but not energy or passion – into their work. Employees who are actively disengaged are not only unhappy at the workplace but can also undermine what their engaged co-workers accomplish.

This section reviews the association between (1) life-friendly HR policies and work–life balance, (2) life-friendly HR policies and employee engagement as well as (3) employee engagement and work–life balance. After each review, a corresponding hypothesis is provided.

The current study was carried out in the CODE, University of Cape Coast, Ghana. The population for this study comprised all the 269 employees in the college. The entire population was used for the study because, given its size, it was realistic and feasible to reach out to all the staff in the college (Kim, 2016).

Data were gathered for this study using the survey questionnaire. The survey method was preferred because according to Kim (2016), it is most appropriate for explanatory research. The instrument was divided into three parts. Part one comprised the items used for collecting data about the sex, age, years of work experience, academic qualification and marital status of respondents. The second part comprised the items for measuring work–life balance, while part three comprised items for measuring life-friendly HR policies. The final part comprised items for measuring employee engagement. Each item was measured on a 7-Likert-scale with 1 = strongly disagree and 7 = strongly agree. The questionnaires were self-administered during week days from July 2021 to September 2021. Out of the total 269 respondents, 232 successfully completed and returned. This gave a response rate of 86.25%.

Data collected for this study were processed using the IBM Statistical Package for Social Sciences (SPSS) Version 25.0 software and analysed using the Partial Least Squares (PLS) structural equation modelling, which allows for simultaneous estimation of co-variation for all variables in a model (Sarstedt et al., 2014). The preparation of the data gathered was in two phasis. Firstly, the raw data were edited, coded and converted into actual variables. Each questionnaire was carefully checked for incompleteness and inconsistencies. The variables were assigned codes to facilitate computer data input. Once entered into the datasheets, the data were carefully screened to minimise entry errors. Frequencies for each variable were checked to detect the out-of-range values. Secondly, the processed data were analysed using SMART PLS 3.2.8. The prepared data file was then converted into ‘comma-delimited’ format ‘CSV’ before the final file was imported into the SMART PLS application for the model configuration (Browne et al., 2019; Kumar et al., 2019; Lew et al., 2019).

The set-up of the PLS tool for the formulation of the model was as follows: Consistent PLS algorithm and consistent bootstrapping were dully marshaled for the analysis with 5000 maximum iterations. Casewise deletion was configured for missing values although there were no missing values in the data (Ringle et al., 2015). A 95% confidence interval with a corresponding 5% level of significance was set for the reflective model. The one-tailed test hypotheses were formulated because of the non-directional nature of the objectives of the study. As a decision rule, indicators that had outer loadings less than 0.7 (not statistically significant) were deleted in order to improve the measurement model. The evaluation of the models began with the measurement model and then the structural model (Fami et al., 2019; Hair et al., 2017; Tabet et al., 2019). Cronbach’s alpha (≥ 0.7) and composite reliability (≥ 0.7) were also computed. Cronbach’s alpha value for all the items exceeded the minimum 0.7 cut-off point (Sarstedt & Ringle, 2019). Cronbach’s alpha and composite reliability are the most common internal consistency measurements (Fami et al., 2019).

The scale’s reliability was measured with the rho_A (≥ 0.7), which is currently the only consistent reliability measure of PLS construct scores (Dijkstra & Henseler, 2015; Henseler, 2017). Convergent validity was measured with the average variance extracted (AVE). Average variance extracted values must be greater than or equal to 0.5 before they can adequately measure convergent validity (Ringle et al., 2015). Discriminant validity was measured with Fornell–Larcker criteria (should follow in descending order). Discriminant validity represents the uniqueness and distinctiveness of each construct relative to other constructs in the model (Afum et al., 2019). As reflective models are susceptible to biases and errors, the authors of the current study thought that it was wise to examine the test of collinearity statistics (Hair et al., 2016). Common method bias was measured with the collinearity statistics (variance inflation factor (VIF) ≤ 5). This was measured with the VIF value, as its usage in this context has been confirmed in reflective models in structural modeling (Kock, 2015). Generally, when collinearity statistics are above 3.3 thresholds, it implies that the model is likely to be affected by common method bias. Alternatively, when the VIF is less than 3.3, the model is deemed to be without common method bias (Afum et al., 2019). Factor loadings for the relevant indicators were measured accordingly, given cognisance to top-values and t-statistics (Ringle et al., 2015).

Path coefficients (unstandardised beta) were used to assess the contributions of both direct and indirect predictors to the variance in the dependent variable (Schberth et al., 2018) while effect size (f²) was used to quantify their contributions to the changes in the dependent variable (Ahrholdt et al., 2019). Effect size values above 0.35, 0.15 and 0.02 were regarded as strong, moderate and weak (Cohen, 2016). This was assessed by the R-square, regarded as the most common effect size measure in path models (Garson, 2016). To this effect, tentative cut-off points have been recommended (Garson, 2016; Hock & Ringle, 2006). Results above 0.67 are described as being ‘substantial’, those above 0.33 are ‘moderate’ and those above 0.19 are ‘weak’. The findings were presented in tables and figures for easy understanding and reporting. Finally, the Q² was analysed using the blindfolding procedure. Q-square is predictive relevance that measures whether a model has predictive relevance or otherwise (> 0 is good). A Q² above 0 shows that the model has predictive relevance.

Three constructs were measured in this study. Work–life balance was measured by using the eight-item scale developed by Valcour (2007). Life-friendly HR policies were measured using the 15-item scale developed by Yuile et al.’s (2012) while employee engagement was measured using the 15-item scale developed by Schaufeli and Bakker (2003).

The University of Cape Coast Institutional Review Board (UCCIRB) has granted Provisional Approval for the implementation of your research titled Workplace Policies, Work Engagement and Work-Life Balance among the Staff of College of Distance Education. This approval is valid from 02 February 2022 to 15 February 2023 (UCCIRB/CHLS/2021/72).

The measurement model evaluation involved confirmatory factor analysis, indicator loadings, reliability and validity. The model was assessed in order to determine the suitability of the indicators used to measure HR policies, employee engagement and work–life balance. According to Kim (2016), an indicator is suitable if its outer loading is more than 0.7.

The indicator variables were subjected to confirmatory factor analysis. This was done in order to identify the indicators that clearly explained the constructs even though the scales used were developed and modified from existing literature with theoretical support (Sarstedt et al., 2014). To increase the overall model ‘reliability, any indicator that loaded below 0.70 was removed. The indicators in Table 1 met the reliability criterion.

Collinearity among constructs such as the coefficient of determination, predictive relevance, effect size, path coefficient and its significance were evaluated in this study. The direct and indirect models were run together (Nitzl et al., 2016). The collinearity test indicated that the VIF of work–life balance was 3.198 (below the threshold of 10) and the tolerance level 0.619 (below the threshold of 1.00). These results mean that no multicollinearity affected the results negatively. The results are presented in Table 4.

The absence of common bias was further confirmed by the variance inflator factor result, which was equal to 3.198. Having a VIF value greater than 3.3 means that the results are tainted by common method bias (Hair et al., 2016). As all the VIF values in the collinearity test were 3.3 or below (3.198), this signifies that the model is devoid of collinearity and common method bias. The endogenous variable’s predictive power was also evaluated, and the coefficient of determination (R²) for the work–life balance (the endogenous latent variable) was 0.835. This means that the two latent constructs accounted for 83.5% variation in staff’s work–life balance as indicated in Table 5 and Figure 2. Workplace policies, on the other hand, explained 22.4% of the variation in employee engagement, as indicated in Figure 2.

The predictive relevance of the model was also checked, using the Stone-Q² Geisser’s statistic (Stone, 1974). The results are presented in Table 6 and Figure 3.

After performing the blindfolding procedure, the values of the Q² in Table 6 and Figure 3 were greater than zero (work–life balance = 0.590, employee engagement 0.142), indicating predictive significance (Hair et al., 2016). Finally, the internal path co-efficient and significance were assessed by running bootstrapping procedure. The results (p = 0.00) indicated that the hypothesised structural paths were significant as presented in Table 7.

The theoretical contribution of this study rests on its extension of the knowledge and application of work–family border theory as reviewed in the literature. By reviewing the original theory as proposed by Clark (2000), and applying the assumptions to an empirical situation in Ghana, this study is expected to increase student’s understanding of the theory and how it is applied in real-life situations. Secondly, by exploring the association between life-friendly HR policies and work–life balance, based on the work–family border theory, this study has succeeded in contributing to the existing body of knowledge on both variables and how they link to enhance the work of the HR department. Finally, the study contributes to theory by investigating the mediation of employee engagement between life-friendly HR policies and work–life balance.

In addition to its theoretical implication, the current study has some important practical implications. Firstly, the application of work–family border theory to explain the relationships among life-friendly HR policies, employee engagement and work–life balance can guide management of the CODE to incorporate appropriate interventions and policies on how the boundaries between work and family responsibilities are controlled to achieve a balance for their employees. This conclusion is based on an aspect of the theory that holds that work and family spheres are not separated but are connected with flexible and permeable boundaries. Secondly, as the synthesised correlation between employee engagement and work–life balance suggests reciprocal relations (Clark, 2000; Kangure, 2014), managers of the CODE would be required to implement HR practices that improve the two variables from a holistic viewpoint rather than as separate variables. In this way, they will be mindful as to when to change and develop their culture, systems and policies to successfully facilitate higher levels of employee engagement and work–life balance.

Finally, management of the CODE could address the issue of work–life conflict by offering employees, tools, practices and learning opportunities that both reduce work–life conflict and improve employee engagement. Furthermore, management could provide support by developing and implementing HR policies and practices such as employee voice and effective communication to reduce work–life conflict in the college.

This study explored the relationship among life-friendly HR policies, employee engagement and work–life balance in a higher institution. The main objective of the study was to examine the mediation of employee engagement between life-friendly HR policies and work–life balance. In this study, it was found that life-friendly HR policies have a positive effect on work–life balance. More importantly, the authors found that employee engagement had a complementary partial mediation between life-friendly HR policies and work–life balance. As the study was restricted to the CODE in the University of Cape Coast, the research framework and hypothesis could be expanded and modified to include other colleges to confirm whether the results of this study can be generalised across other colleges and universities.