Job crafting , proactive personality and meaningful work : Implications for employee engagement and turnover intention

Research has shown that employees perform optimally in challenging, resourceful work environments, as these environments foster employee engagement (Bakker & Demerouti, 2014). Organisations in the financial services industry should, therefore, develop and implement human resource practices and interventions that influence the work conditions (i.e. job demands and job resources) of their employees. In addition, the researchers argue that it is just as important for employees themselves to adjust their work conditions proactively. This can be achieved through crafting behaviour (i.e. job crafting) (Wrzesniewski & Dutton, 2001).


Introduction
Research has shown that employees perform optimally in challenging, resourceful work environments, as these environments foster employee engagement (Bakker & Demerouti, 2014). Organisations in the financial services industry should, therefore, develop and implement human resource practices and interventions that influence the work conditions (i.e. job demands and job resources) of their employees. In addition, the researchers argue that it is just as important for employees themselves to adjust their work conditions proactively. This can be achieved through crafting behaviour (i.e. job crafting) (Wrzesniewski & Dutton, 2001). work environment. Research has shown that employees who have a disposition towards proactive behaviour (i.e. proactive personality) are more inclined to exhibit proactive work behaviours than their counterparts (McCormick, Guay, Colbert, & Stewart, 2018). It is, therefore, proposed that job crafting, proactive personality and meaningful work are salient antecedents of variance in prominent work-related outcomes, such as employee engagement and turnover intention.

Research purpose and objectives
Against this background, the purpose of the study was to examine the role of job crafting, proactive personality and meaningful work in predicting employee engagement and turnover intention among employees in the financial services industry within the framework of the Job Demands-Resources (JD-R) theory. More specifically, the study aimed to: • determine the level of employee engagement and turnover intention among employees in service-providing firms that operate in the financial services industry; • propose and empirically test an explanatory employee engagement and turnover intention structural model among employees in service-providing firms that operate in the financial services industry; • highlight the managerial implications of results, and recommend specific human resource practices and interventions to address challenges pertaining to employee engagement and turnover intention among employees in service-providing firms that operate in the financial services industry.

Literature review Job Demands-Resources theory
The JD-R theory is a meta-theoretical framework that can be applied to diverse occupational settings. Even though each occupation has a unique set of work characteristics that are associated with employee wellness and effectiveness, the JD-R theory suggests that these characteristics can be categorised as job demands or job resources (Demerouti, Bakker, Nachreiner, & Schaufeli, 2001).
Job demands are the physical, psychological, social or organisational features of a job that require sustained physical, mental and/or psychological (i.e. cognitive and emotional) effort from employees (e.g. job insecurity or role conflict). Notably, job demands do not necessarily incur physiological and/or psychological costs. These characteristics evoke strain only when they exceed the adaptive capability of employees. Job resources, on the other hand, are the physical, psychological, social or organisational features of a job that enable employees to reduce job demands, achieve work goals, and stimulate development, learning and personal growth (Tims & Bakker, 2010). These characteristics are found at an organisational level (e.g. opportunities for advancement), an interpersonal level (e.g. team processes), a task level (e.g. feedback) or the level of work design (e.g. autonomy) (Demerouti & Bakker, 2011).
The JD-R theory suggests that two different underlying psychological processes influence the development of job strain (i.e. burnout) on the one hand and motivation (i.e. employee engagement) on the other. These processes reflect a health impairment process and a motivational process. The health impairment process (or effort-driven process) assumes that demanding jobs or jobs with chronic demands incur physiological and/or psychological costs by exhausting employees' physical and mental resources. This, in turn, may lead to health problems and the depletion of their energy. The motivational process (or motivation-driven process), on the other hand, assumes that job resources are not only used to deal with job demands but may be valued in their own right (i.e. intrinsic motivation) or may help employees achieve or protect other valued resources (i.e. extrinsic motivation) as well. By implication, these characteristics possess motivation potential that may foster employee engagement, improve job performance and discourage cynicism. Empirical research has demonstrated support for both processes (Bakker & Demerouti, 2014;Demerouti & Bakker, 2011).
Even though the JD-R theory was initially restricted to work conditions (i.e. job demands and job resources), Xanthopoulou, Bakker, Demerouti and Schaufeli (2007) highlighted that personal resources (e.g. optimism or selfefficacy) are also important determinants of employees' adaptive capability. These researchers defined personal resources as 'aspects of the self that are generally linked to resiliency and refer to individuals' sense of their ability to control and influence their environment successfully' (p. 123). Their cross-sectional study among 714 Dutch employees from an electrical engineering and electronics organisation showed support for this notion and demonstrated that personal resources mediate the positive association between job resources and employee engagement (i.e. motivational process) and influence employees' perception of job resources. Parker and Ohly (2008) proposed that employees could actively shape their job design from the bottom up. This employee-driven process, known as job crafting, was initially defined as cognitive and physical changes employees make to the relational or task boundaries of their job (Wrzesniewski & Dutton, 2001). However, this definition of job crafting offered a limited perspective of the construct. To address concerns with the above-mentioned definition of job crafting, Tims, Bakker and Derks (2012) proposed that it involves self-initiated changes that employees make to their work conditions (i.e. job demands and job resources) in accordance with the JD-R theory. More specifically, these researchers proposed that job crafting involves increasing social or structural job resources, as well as challenging (positive) job demands, and/or decreasing hindrance stressors that interfere with work performance.

Job crafting
By drawing on the central precepts of the JD-R theory, the researchers operationalised proactive personality as a personal resource and meaningful work as a job resource among employees in the financial services industry.

Proactive personality
All humans have a unique personality characterised by a set of distinct personality traits. These traits signify the dimensions along which humans' personalities vary in relatively stable ways (Roberts et al., 2017). According to Barrick, Mount and Li (2013), personality traits are distal motivational forces that can be used to make generalisations about human nature and explore between-person differences, as they determine humans' behavioural responses. The researchers argue that proactive personality represents an invaluable personal resource among employees in the financial services industry.
In other words, it denotes a dispositional tendency to engage in proactive behaviour. While some employees adapt to, react to and are shaped by their work environment, employees who have a proactive personality recognise opportunities, take personal initiative and persevere until they have brought about meaningful change in their work environment .
By drawing on the theory of proactive personality (Bateman & Crant, 1993), the researchers propose that employees who have a proactive personality, would be more inclined to exhibit crafting behaviour. More specifically, employees who have a proactive personality are more inclined to increase social or structural job resources and challenging job demands, and decrease hindering job demands. Research has shown direct support for the positive association between proactive personality and job crafting. For example, a cross-sectional study conducted by Bakker et al. (2012) among 95 dyads (n = 190) in the Netherlands found that employees who have a proactive personality were more inclined to craft their jobs (i.e. increasing structural or social job resources and challenging job demands

Turnover intention
Employee turnover involves the voluntary movement of employees across the membership boundaries of organisations (Price, 2001). It occurs when employees decide to depart from their current organisation despite having the opportunity of continued employment. Notably, employee turnover is a multistage process. Turnover intention is the last sequence of withdrawal cognitions in the turnover process (Parker & Martin, 2009). Although turnover intentions may not necessarily lead to employee turnover, research has indicated that it represents an important outcome variable (Chang, Wang, & Huang, 2013).
Research has revealed that employees who have a proactive personality strive for congruence between themselves and their work environment in terms of their needs and abilities (Parker & Collins, 2010;Tims & Bakker, 2010). The researchers argue that this might be compounded by their inclination to adopt proactive work behaviours. If their need for congruence with their work environment is not met, employees with a proactive personality may be more inclined, because of their inclination to adopt proactive work behaviours, to seek alternative employment opportunities. As opposed to tolerating a frustrating work environment, these employees may respond to the lack of congruence between themselves and their work environment by developing turnover intention. For this reason, the researchers formulated the following hypothesis: H2: Proactive personality has a significant positive influence on turnover intention.

Meaningful work
Adlerian theory proposes that all humans live in the realm of meanings (Alderfer, 1972). They experience their reality by attaching meaning to experiences through the process of meaning-making. According to Mezirow (1981, p. 394), 'at its simplest meaning-making refers to a lifelong process of understanding the world and our relationships with it'. Through the interpretation of their experiences, humans spin webs of meaning known as 'meaning systems' (Molden & Dweck, 2006, p. 201). These integrated mental representations of interrelations between objects, events and relationships encompass everything humans know.
In light of the preceding discussion, it is to be expected that employees seek meaning in their work experiences. It is, however, necessary to recognise that considerable differences exist in the way employees perceive their work (Wrzesniewski, McCauley, Rozin, & Schwartz, 1997). Job design alone does not determine the meaningfulness of work (Steger, Dik, & Duffy, 2012;Tims, Derks, & Bakker, 2016). Inherent in the existence of meaningful work are job design (e.g. task significance and comprehensibility), person-job fit (e.g. values and mission), work relationships (e.g. colleagues and supervisors) and work beliefs (e.g. calling orientation) (Pratt & Ashforth, 2003). Wrzesniewski (2003) proposed that employees can improve the meaningfulness of their work by aligning it with their perception of work through job crafting.
A three-wave study conducted by Tims et al. (2016) among a heterogeneous group of 114 employees in the Netherlands has examined whether job crafting and meaningful work are positively related. These researchers demonstrated that employees who engaged in job crafting in the first week (i.e. increasing social or structural job resources and challenging job demands, and decreasing hindering job demands) reported higher levels of person-job fit the following week and, in turn, found their work more meaningful in the final week. In line with this, Wrzesniewski, LoBuglio, Dutton and Berg (2013) suggested that job crafting is related to work meaning and identity. This suggests that job crafting optimises person-job fit and, in turn, enhances the meaningfulness of work. In line with the above, the researchers formulated the following hypothesis: H3: Job crafting (i.e. increasing social or structural job resources and challenging job demands, and decreasing hindering job demands) has a significant positive influence on meaningful work.

Employee engagement
Employee engagement has attracted much attention from researchers and practitioners alike. Perhaps this is to be expected given that it is considered a key determinant of organisational success (Macey, Schneider, Barbera, & Young, 2009;Rich, LePine, & Crawford, 2010). Despite being one of the most popular constructs in the research field of industrial and organisational psychology, there is still a lack of agreement about its definition and meaning. In this study, employee engagement is defined in accordance with Schaufeli, Salanova, González-Romá and Bakker (2002). Schaufeli et al. (2002, p. 74) define employee engagement as 'a positive, fulfilling, work-related state of mind that is characterised by vigour, dedication, and absorption'. These researchers describe it as a relatively persistent and pervasive affective-cognitive state that does not depend on specific objects, events, individuals or behaviours. They suggest, in contrast to early studies (e.g. Maslach & Leiter, 1997) which conceptualised burnout and engagement as opposite poles of a continuum, that burnout and engagement are independent concepts (that should be measured with different instruments).
Apart from the positive association between job crafting and meaningful work, the study propose that employees who consider their work meaningful are more inclined to invest themselves in their work and engage more fully. Because of the intrinsic motivational force associated with meaningful work, these employees may find it easier to approach their work-related activities with vigour, dedication and absorption. Research has shown direct support for the positive association between meaningful work and employee engagement (Shuck, 2019). As an illustration, a cross-sectional study conducted by Jung and Yoon (2016)  In line with hypothesis 4, the researchers propose that employees who consider their work meaningful are less inclined to quit their job in the near future. Because of the presence of the intrinsic motivational force associated with meaningful work, these employees may be less inclined to seek alternative employment opportunities.
Research has revealed support for the negative association between meaningful work and turnover intention. As an illustration, a cross-sectional study conducted by Janik and Rothmann (2015) demonstrated that meaningful work and turnover intention were negatively related among 502 secondary school teachers in Namibia. Comparable results were reported by Arnoux-Nicolas, Sovet, Lhotellier, Di Fabio and Bernaud (2016)  Meta-analytic evidence reported by Bakker, Hakanen, Demerouti and Xanthopoulou (2007) showed that job resources and employee engagement were positively related, especially when job demands were high. By implication, when employees, who face high job demands, have sufficient job resources they are still able to flourish in their work. With this in mind, the researchers proffered that employees who exhibit crafting behaviour are more inclined to be engaged in their work as they adjust their work conditions (i.e. job demands and job resources) proactively. Job crafting, therefore, represents a practical way in which emplyees themselves can enhance their own work engagement.
Research has shown direct and indirect support for the above-mentioned process of 'self-engagement' (Bakker et al., , p. 1363 Notably, Bakker (2011) suggested that the positive association between employee engagement and job crafting may be dynamic. He proposed that apart from the positive association between job crafting and employee engagement, it is plausible that a recerse causal association also exists between these constructs. Drawing on the Broaden-and-Build theory of positive emotions (Fredrickson, 2004), the researchers support this notion. The Broaden-and-Build theory proposes that positive emotions broaden employees' momentary thought-action repertoire (i.e. broaden hypothesis). As their momentary thought-action repertoire broadens, employees build personal resources that range from psychological and social resources to intellectual and physical resources (i.e. build hypothesis). In contrast, negative emotions narrow the momentary thought-action repertoire of employees (Fredrickson, 2004). A study conducted by Schaufeli and Van Rhenen (2006) among 815 managers from a telecommunications organisation in the Netherlands showed that engaged employees experienced positive emotions. In view of this, the current researchers support the notion that the positive emotions (including vigour, dedication and absorption) that engaged employees experience broaden their momentary thought-action repertoire and, in turn, encourage job crafting.
Research has revealed direct and indirect support for the reverse causal association between job crafting and employee engagement. For example, a diary study conducted by Sonnentag (2003) among 147 employees from public service organisations in Germany demonstrated that day-level employee engagement was positively associated with day-level proactive behaviour (i.e. the pursuit of learning and personal initiative) during the following work day. In accordance with this, Fritz and Sonnentag (2009) found that positive affective experiences were related to proactive behaviours (i.e. personal initiative) among 172 employees from civil service organisations in Germany. Harju et al. (2016) demonstrated direct support for this notion among 1630 educated employees in Finland. More specifically, they found that engaged employees were more inclined to exhibit crafting behaviour (i.e. increasing social or structural job resources). Comparable results were reported by Hakanen, Peeters and Schaufeli (2018) among 1877 dentists in Finland. The following hypothesis was consequently formulated in the study: H7: Employee engagement has a significant positive influence on job crafting (i.e. increasing social or structural job resources and challenging job demands, and decreasing hindering job demands).
Apart from encouraging job crafting, the positive emotions (specifically vigour and dedication) that engaged employees experience may cultivate a sense of organisational commitment and, in turn, discourage the development of turnover intention. The researchers offer the Conversion of Resources theory (Hobfoll, 1989) to support this notion. The Conversion of Resources theory proposes that employees invest resources in their work environment and seek to minimize the loss of resources. Engaged employees willingly commit themselves to their work (i.e. dedication) and have difficulty detaching themselves after tasks have been completed (i.e. absorption) thereby investing resources in their current organisation (Schaufeli et al., 2002). In consideration of this, the researchers advocated that engaged employees are less inclined to seek alternative employment opportunities. The loss of resources that is associated with starting work in a new organisation discourages these employees to develop turnover intentions.
Research has shown direct support for the negative association between employee engagement and turnover intention. Meta-analytic evidence reported by Halbesleben (2010) and Saks (2019) found that employees who exhibit vigour and dedication were less inclined to develop turnover intention. In accordance with this, Takawira, Coetzee and Schreuder (2014) demonstrated that dedication and turnover intention were negatively related among 153 academic and non-academic employees from a higher education institution in South Africa. Comparable results were reported by Gabel Shemueli, Dolan, Suárez Ceretti and Nuñez Del Prado (2016) among 818 nurses in Uruguay (n = 316) and Spain (n = 502) and by Agarwal and Gupta (2018) among 1302 heterogenous managers in India. For this reason, the researchers formulated the following hypothesis: H8: Employee engagement has a significant negative influence on turnover intention.

Research approach
The researchers adopted a quantitative research approach based on correlational analysis to achieve the research objectives. More specifically, primary data were collected through a cross-sectional survey of job crafting, proactive personality, meaningful work, employee engagement and turnover intention.

Research participants
The setting for the study comprised organisations in the financial services industry. Three service-providing firms that operate in the financial services industry were invited to participate in the study. Each firm, with regional offices across South Africa, was part of an international service provider offering primarily auditing and accounting services.
The researchers used non-probability purposive sampling to select a sample of approximately 800 employees from participating regional offices. Three hundred and ninety-one employees participated in the study (49% response rate). Table 1 offers a description of the demographic and employment characteristics of participants.
Data showed that women comprised 61% of the sample. The majority of participants had completed an Honour's degree (highest level of education; 70%), were employed less than a year at their current service-providing firm (organisational tenure; 34%) and worked in the auditing department (department; 82%). Furthermore, the age of participants varied from 18 to 57 years (mean = 26.53 years).

Measuring instruments
The web-based survey comprising two sections was developed specifically for the study. The first section requested the demographic and employment information of participants and the second section measured the latent variables of interest.

Proactive Personality Scale
Proactive personality was measured with the abbreviated Proactive Personality Scale (Claes, Beheydt, & Lemmens, 2005). This self-report measure consists of six items. An example item is 'If I see something I don't like, I fix it'. A cross-cultural analysis across three countries (i.e. Belgium, Finland and Spain) showed support for the internal consistency reliability of the abbreviated Proactive Personality Scale (α = 0.78 to 0.86) (Claes et al., 2005). The abbreviated Proactive Personality Scale measured the degree to which participants had a proactive personality. A seven-point Likert-type scale varying from 1 (certainly not agree) to 7 (certainly agree) was used to score responses. No items of the abbreviated Proactive Personality Scale were reverse-scored.

Job Crafting Scale
The Job Crafting Scale  was used to measure job crafting. This self-report measure, which consists of 21 items, assesses the four dimensions of job crafting, namely increasing social job resources (five items, e.g. 'I ask others for feedback on my job performance'), increasing structural job resources (five items, e.g. 'I try to develop myself professionally'), increasing challenging job demands (five items, e.g. 'when there is not much to do at work, I see it as a chance to start new projects') and decreasing hindering job demands (six items, e.g. 'I make sure that my work is mentally less intense'). A validation study among a random sample of 375 employees in the Netherlands  demonstrated support for the internal consistency reliability of the subscales of the Job Crafting Scale (increasing social job resources: α = 0.77; increasing structural job resources: α = 0.82; increasing challenging job demands: α = 0.75; decreasing hindering job demands: α = 0.79). The four subscales of the Job Crafting Scale were combined in the study. The composite measurement indicator measured the degree to which participants engaged in crafting behaviour. A five-point Likert-type scale varying from 1 (never) to 5 (often) was used to score responses. No items of the Job Crafting Scale were reverse-scored.

Psychological Meaningfulness Scale
Meaningful work was measured with the Psychological Meaningfulness Scale (May, Gilson, & Harter, 2004). This self-report measure consists of six items. An example item is 'I feel that the work I do on my job is valuable'. A validation study among 213 employees from an insurance firm in the United States of America (US) demonstrated support for the internal consistency reliability of the Psychological Meaningfulness Scale (α = 0.90) (May et al., 2004). The Psychological Meaningfulness Scale measured the degree of meaning that participants in the study found in their work-related activities. A fivepoint Likert-type scale varying from 1 (strongly disagree) to 5 (strongly agree) was used to score responses. No items of the Psychological Meaningfulness Scale were reversescored.

Utrecht Work Engagement Scale
The Utrecht Work Engagement Scale (UWES-9) (Schaufeli, Bakker, & Salanova, 2006) was used to measure employee engagement. This self-report measure, which consists of nine items, assesses the three dimensions of employee engagement, namely vigour (three items, e.g. 'at my work, I feel bursting with energy'), dedication (three items, e.g. 'I am enthusiastic about my job') and absorption (three items, e.g. 'I am immersed in my work'). A cross-cultural analysis across 10 countries (i.e. Australia, Belgium, Canada, Finland, France, Germany, the Netherlands, Norway, South Africa and Spain) showed support for the internal consistency reliability of the UWES-9 (α = 0.85-0.92) and its subscales (vigour: α = 0.60-0.88; dedication: α = 0.75-0.90; absorption: α = 0.66-0.86) . The three subscales of the UWES-9 were combined in the study. The composite measurement indicator measured the degree to which participants were engaged in their work. A seven-point Likert-type scale varying from 0 (never) to 6 (always or everyday) was used to score responses. No items of the UWES-9 were reverse-scored.

Turnover Intention Scale
Turnover was measured with the Turnover Intention Scale (Moore, 2000). This self-report measure consists of four items. An example item is 'I will probably look for a job at a different company in the next year'. A validation study among a random sample of 214 employees in the US demonstrated support for the internal consistency reliability of the Turnover Intention Scale (α = 0.92) (Moore, 2000). The Turnover Intention Scale measured the probability that participants would quit their job in the near future. A five-point Likert-type scale varying from 1 (very unlikely) to 5 (very likely) was used to score responses. Two items of the Turnover Intention Scale were reverse-scored.

Research procedure
Once the study was evaluated and approved by the Ethics Committee of the university where the research was undertaken, the researchers approached the human resource managers of financial service-providing firms with regional offices across South Africa. The human resource managers received an email that contained a letter of approval from abovementioned Ethics Committee and an institutional permission form. Human resource managers who agreed to participate in the study were asked to send a signed copy of the institutional permission form to the principal researcher. This gave the researchers written permission to invite employees to complete the web-based survey.
The web-based survey was sent electronically to employees of participating regional offices. Participation was anonymous and voluntary. Informed consent was obtained from individual participants. Participants were given 3 weeks to complete the survey. Two reminders (i.e. weekly) were sent to employees to encourage their participation.

Statistical analysis
A number of statistical procedures were used to analyse the data and to evaluate the proposed employee engagement and turnover intention structural model, depicted in Figure 1. These techniques included item analysis to determine the reliability of the latent variable scales, confirmatory factor analysis to validate the measurement model and covariance-based Structural Equation Modelling (SEM) to evaluate the structural model. In addition, partial least squares (PLS) SEM, a prediction-oriented SEM technique, was performed as an alternative method to evaluate the structural model.

Testing the measurement model Item analysis
An item analysis was done on each latent variable scale that was included in the web-based survey. Table 2 shows that the Cronbach's alpha for each latent variable scale were satisfactory (≥ 0.70) (Nunnally, 1978). Moreover, the average inter-item correlation of the Psychological Meaningfulness Scale, UWES-9 and Turnover Intention Scale was > 0.50. Even though the average inter-item correlation for the abbreviated Proactive Personality Scale and Job Crafting Scale was 0.45 and 0.26, respectively, this was still considered satisfactory (≥ 0.15) (Clark & Watson, 1995).

Evaluating measurement model fit
Confirmatory factor analysis was performed to evaluate the measurement model fit. Notably, the researchers did not test all the model parameters in a single measurement model. Three measurement models were constructed. These measurement models were the employee engagement measurement model, the job crafting measurement model, Note: It was assumed that the structural error terms were fixed to zero. ξ 1 , proactive personality; η 1 , turnover intention; η 2 , employee engagement; η 3 , meaningful work; η 4 , job crafting.
FIGURE 1: Proposed employee engagement and turnover intention structural model. and the meaningful work, proactive personality and turnover intention measurement model.
Results showed that the empirical data reproduced the employee engagement measurement model reasonably well: Sattora Bentler chi-square value = 85.84 (p < 0.00), root mean square error of approximation (RMSEA) value = 0.08, goodness-of-fit index (GFI) value = 1.00 and adjusted goodness-of-fit index (AGFI) value = 0.99. The factor loadings of the indicator variables, which were included in the employee engagement measurement model, were statistically significant (p < 0.05) and the t-values exceeded the critical cut-off values (−1.96 < t-value < +1.96). The factor loadings ranged from 0.71 to 0.97, and t-values ranged from 18.68 to 96.16.
Comparable results were reported for the job crafting measurement model. The job crafting measurement model was also reproduced reasonably well by the empirical data: Sattora Bentler chi-square value = 480.41 (p < 0.00), RMSEA value = 0.07, GFI value = 0.97 and AGFI value = 0.97. The factor loadings of the indicator variables, which were included in the job crafting measurement model, were statistically significant (p < 0.05) and the t-values exceeded the critical cut-off values (-1.96 < t-value < +1.96). The factor loadings ranged from 0.58 to 0.93, and t-values ranged from 8.61 to 38.13.
In line with the above results, the meaningful work, proactive personality and turnover intention measurement model also reproduced by the empirical data reasonably well: Sattora Bentler chi-square value = 277.41 (p < 0.00), RMSEA value = 0.07, GFI value = 0.92 and AGFI value = 0.89. The factor loadings of the indicator variables, which were included in proactive personality, meaningful work and turnover intention measurement model, were statistically significant (p < 0.05) and the t-values exceeded the critical cut-off values (-1.96 < t-value < +1.96). The factor loadings ranged from 0.53 to 0.92, and t-values ranged from 13.03 to 93.33.
Empirical data thus corroborate the three measurement models reasonably well. Each measurement model showed reasonable fit (0.05 < RMSEA < 0.08) and met the criteria for the GFI value (> 0.90). The factor loadings of the indicator variables, which were included in each of the measurement models, were statistically significant (p < 0.05) and the t-values exceeded the critical cut-off values (-1.96 < t-value < +1.96).

Testing the structural model Fitting the employee engagement and turnover intention structural model
It is important to mention that the subscales of the Job Crafting Scale  and the UWES-9  were operationalised as manifest variables of job crafting and employee engagement, respectively, during covariance-based SEM. This enabled the researchers to fit the employee engagement and turnover intention structural model despite restrictions in terms of sample size (n = 391).
The proposed employee engagement and turnover intention structural model, depicted in Figure 1, did not fit the empirical data exceptionally well: Sattora Bentler chi-square value = 549.22 (p < 0.00), RMSEA value = 0.06, GFI value = 0.89 and AGFI value = 0.86. Although the RMSEA value (0.06) showed reasonable fit, this was not verified by the GFI value (0.89) and the AGFI value (0.86). These values indicated mediocre fit. To ensure the accuracy and consistency of the results obtained from subsequent statistical analyses concerning the strength and significance of the hypothesised relationships that exist among the latent variables, covariance-based SEM was supplemented with PLS SEM. Table 3 shows that five p-values in the proposed employee engagement and turnover intention structural model were statistically significant. The strength of the influence of ξ j and/or η j on η i was substantial and significant between proactive personality and job crafting (p = 0.73; accept Hypothesis 1); job crafting and meaningful work (p = 0.34; accept Hypothesis 3); meaningful work and employee engagement (p = 0.67; accept Hypothesis 4); job crafting and employee engagement (p = 0.29; accept Hypothesis 6); and employee engagement and turnover intention (p = -0.50; accept Hypothesis 8) at the 95% confidence level.

Testing the hypothesised relationships
Three p-values were not statistically significant. The strength of the influence of ξ j and/or η j on η i was not substantial and significant between proactive personality and turnover intention (p = 0.12; reject Hypothesis 2); meaningful work and turnover intention (p = -0.01; reject Hypothesis 5); and employee engagement and job crafting (p = 0.05; reject Hypothesis 7) at the 95% confidence level.

Partial least squares path analysis
Before commencing with PLS SEM, the proposed employee engagement and turnover intention structural model was modified. As shown in Figure 2, the revised structural model excluded the hypothesised positive influence of employee engagement on job crafting. Hypothesis 7 could, therefore, not be tested during PLS SEM because of the model modification.

Reliability analysis
A reliability analysis was done on all the latent variable scales that were included in the web-based survey. Table 4 shows  Table 4 shows that the R 2 values of employee engagement (0.57) and turnover intention (0.21) were satisfactory. By implication, the inner-model measurement fit was satisfactory. The latent variables in the revised employee engagement and turnover intention structural model explained 57% of variance in employee engagement and 21% of variance in turnover intention. Figure 3 depicts the path coefficients between the latent variables of interest. Six path coefficients in the revised employee engagement and turnover intention structural model were statistically significant. The strength of the influence of ξ j and/or η j on η i was substantial and significant between proactive personality and job crafting (accept Hypothesis 1); job crafting and meaningful work (accept Hypothesis 3); meaningful work and employee engagement (accept Hypothesis 4); meaningful work and turnover intention (accept Hypothesis 5); job crafting and employee engagement (accept Hypothesis 6); and employee engagement and turnover intention (accept Hypothesis 8) at the 95% confidence level.

Testing the structural model
One path coefficient was not statistically significant. The strength of the influence of ξ j and/or η j on η i was not substantial and significant between proactive personality and turnover intention (reject Hypothesis 2) at the 95% confidence level.

Discussion
The researchers drew on the central tenets of the JD-R theory to examine the role of job crafting, proactive personality (i.e. a personal resource) and meaningful work (i.e. a job resource) in predicting employee engagement and turnover intention among employees in service-providing firms that operate in the financial services industry. Results showed that employees exhibit average levels of employee engagement and turnover intention. It is, therefore, necessary to develop and implement human resource practices and interventions that concentrate on addressing employee disengagement and turnover intention.
Research has directed attention to a number of antecedents of variance in employee engagement and turnover intention. However, in response to recent developments in job crafting and proactive personality literature, only meaningful work, job crafting and proactive personality were addressed within the scope of the study. The researchers hypothesised that employees who have a proactive personality would be more inclined to exhibit crafting behaviour. This, in turn, may contribute to the meaningfulness of work-related activities, foster employee engagement and discourage the development of turnover intention.
Results generally supported the hypothesised relationships.
Covariance-based SEM showed that hypotheses 1, 3, 4, 6 and 8 were statistically significant, and hypotheses 2, 5 and 7 were not statistically significant. Results of PLS SEM were mostly consistent with the results of covariance-based SEM. Note: It was assumed that the structural error terms were fixed to zero. ξ 1 , proactive personality; η 1 , turnover intention; η 2 , employee engagement; η 3 , meaningful work; η 4 , job crafting.   Partial least squares SEM showed that hypotheses 1, 3, 4, 5, 6 and 8 were statistically significant, and Hypothesis 2 was not statistically significant. As mentioned, Hypothesis 7 could not be tested during PLS SEM because of the model modification. The only difference in the results of covariancebased SEM and PLS SEM related to Hypothesis 5. In summary, results revealed that proactive personality was positively related to employee engagement and negatively related to turnover intention through job crafting and meaningful work.
Unexpectedly, the hypothesised positive influence of proactive personality on turnover intention was not statistically significant. Covariance-based SEM (p = 0.12) and PLS SEM (PLS path coefficient = 0.11) demonstrated that proactive personality was not a predictor of turnover intention. By drawing on the results obtained for Hypothesis 1 (p = 0.73; PLS path coefficient = 0.59), the researchers argued that employees who have a proactive personality rather exhibit crafting behaviour when their need for congruence is not met by their current work environment, as opposed to seeking alternative employment opportunities (i.e. turnover intention).

Practical implications
The practical implications of the study are threefold. Firstly, covariance-based SEM and PLS SEM demonstrated that job crafting is a predictor of meaningful work (p = 0.34; PLS path coefficient = 0.33), as well as employee engagement (p = 0.29; PLS path coefficient = 0.27). It is, therefore, necessary to encourage crafting behaviour among employees in the financial services industry in order to contribute to the meaningfulness of work-related activities and to foster employee engagement. The antecedents of job crafting can broadly be divided into individual differences (e.g. psychological states and personality traits) and organisational factors (e.g. organisational climate and supervisory support) (Petrou, 2013). The researchers recommend that at an individual level, human resource practices and interventions should be tailored to recruit and select employees who are more inclined to exhibit crafting behaviour. At an organisational level, supervisors should cultivate an organisational climate that encourages job crafting. In addition, supervisors should model crafting behaviours to their subordinates.
Secondly, covariance-based SEM (p = 0.73) and PLS SEM (PLS path coefficient = 0.59) demonstrated that proactive personality is a predictor of job crafting. It is, therefore, necessary to encourage proactivity among employees in the financial services industry in order to foster job crafting. Theories of proactivity suggest that the antecedents of proactivity can broadly be separated into individual differences (e.g. knowledge, skills and abilities and personality traits), motivational forces (e.g. extrinsic motivation and role-based motivation) and organisational factors (e.g. organisational climate and supervisory support) (Strauss & Parker, 2014). At an individual level, human resource practices and interventions should be tailored to recruit and select employees who are more inclined to be 'self-starting and change-orientated to enhance personal and organisational effectiveness' (Unsworth & Parker, 2003, p. 178). In addition, human resources practitioners should implement developmental interventions that equip employees with the knowledge, skills and abilities required for proactivity. At a motivational level, it is important to recognise and reward employees who exhibit proactivity.
Recognising and rewarding employees who exhibit proactivity, will reinforce its value. Lastly, at an organisational level, supervisors should cultivate an organisational climate that encourages proactivity (McCormick et al., 2018).
Thirdly, covariance-based SEM (p = 0.67) and PLS SEM (PLS path coefficient = 0.62) demonstrated that meaningful work is a predictor of employee engagement. In addition, PLS SEM (PLS path coefficient = −0.14) demonstrated that meaningful work is a predictor of turnover intention. It is, therefore, necessary to improve the meaningfulness of work in organisations in the financial services industry in order to foster employee engagement and to discourage the development of turnover intention. As mentioned, job design (e.g. task significance and comprehensibility), person-job fit (e.g. values and mission), work relationships (e.g. colleagues and supervisors) and work beliefs (e.g. calling orientation) contribute to the meaningfulness of work-related activities (Pratt & Ashforth, 2003). With this in mind, the researchers recommend that human resources practitioners draw on the Job Characteristics theory (Hackman & Oldham, 1976) in order to improve the meaningfulness of employees' workrelated activities through job enrichment. In addition, it is recommended that human resource practices and interventions be tailored to consider person-job fit and work beliefs when recruiting and selecting employees.

Limitations of the study and recommendations for research
Although the study offers valuable insight into the relationships between job crafting, proactive personality, meaningful work, employee engagement and turnover intention, it is not without limitations. The most noteworthy limitation of the study is its cross-sectional survey design. This design prevents the examination of causal relationships between latent variables and additional (including longitudinal) research is needed to understand the interplay and relational dynamics between the latent variables.
Another limitation relates to the nature of self-reported measures. Owing to the use of self-report measures, it is plausible that common-method variance negatively affected the accuracy of associations between the latent variables of interest. Reasonable steps were, however, taken to assure participants of the complete protection of their identity in order to reduce evaluation apprehension and, as a consequence, common-method variance (Podsakoff, MacKenzie, Lee, & Podsakoff, 2003). Nonetheless, multiple-source studies are needed to fully understand how job crafting, proactive personality and meaningful work influence employee engagement and turnover intention.
The mean age of respondents was 26.53 years old, and 34% of participants had been employed for less than a year at their current service-providing firm. Research has revealed that demographic characteristics, such as age and organisational tenure, influence employee engagement (Berg, Wrzesniewski, & Dutton, 2010) and turnover intention (Peltokorpi, Allen, & Froese, 2015). Although the sample size of the study was satisfactory, it is recommended that the study should be replicated among a sample with broader representation in terms of age and organisational tenure. Similarly, women comprised 61% of the sample. While research concerning the association between gender and proactivity has not been conclusive (Berdicchia, Nicolli, & Masino, 2016), it is recommended that future studies explore the salience of gender.
The four subscales of the Job Crafting Scale  were combined to obtain a composite measurement of crafting behaviour among participants. Research is needed to understand how each type of job crafting (i.e. increasing social or structural job resources and challenging job demands, and decreasing hindering job demands) is related to proactive personality, meaningful work and employee engagement. In addition, job crafting was operationalised as an individual-level phenomenon. It would, however, be valuable to investigate how the opportunity to perform crafting behaviours, as well as the effects of job crafting, may be influenced by other individuals such as colleagues and supervisors (Bakker, Rodríguez-Muñoz, & Sanz Vergel, 2016;Tims, Bakker, Derks, & Van Rhenen, 2013). Research which operationalises job crafting as a team-level phenomenon is needed.
The demarcation of the study within the financial services industry limits the findings and conclusions to the context of organisations within this industry. Research is needed to determine whether results obtained in the study hold true in other occupational contexts in order to generalise results to other work environments.

Concluding remarks
The study illustrates that the central tenets of JD-R theory can be applied to employees in the financial services industry to address challenges regarding employee engagement and turnover intention. Results demonstrated that job crafting, proactive personality and meaningful work significantly predict variance in employee engagement, which in turn significantly predicts turnover intention along with meaningful work. Employees with a proactive personality were more inclined to exhibit crafting behaviour to align job demands and job resources with their needs and abilities. By making their work more meaningful, these employees were more engaged in their work and, in turn, less inclined to develop turnover intention.