The food manufacturing industry in South Africa employs around 288 000 people (representing 21% of the manufacturing industry’s workforce). In addition, the industry has contributed around R80.9 billion (approximately US$4.5 billion), which is 21% of the manufacturing gross domestic product (GDP) and roughly 2.7% of South Africa’s total GDP (Republic of South Africa 2018). According to Mercer Selected Intelligence (Mercer 2017), the worldwide average rate of voluntary labour turnover is 4.7%. According to South African statistics data from 2019, the average employee turnover rate in most industries was between 5% and 10% (Adcorp Holdings 2021). Although a lack of skills in the food sector is a significant problem in South Africa, employees may experience the effects of organisational systems, depersonalisation, physical environment, work role and organisational climate, which Liu et al. (2019) mention are factors that relate to the work environment and are considered necessary for an employee’s intention to leave the organisation. A study conducted by Schlechter, Syce and Bussin (2016) in various occupational work environments in South Africa showed that more than 66% of office-based employees might be experiencing excessive stress, anxiety and depression, with the overall national average labour turnover rate of 57.3% in 2020 and the average labour turnover in the manufacturing industry ranging from 30.6% in 2017 to 44.2% in 2020 (Marshall & Stephenson 2020). However, labour turnover is usually affected by a variety of factors such as stress, burnout and OS (Liu et al. 2019). Emotionally exhausted food industry workers who work in poor conditions and lack support as well as sufficient energy to carry out their activities may wish to leave their employer organisation as a way of facing this loss of energy (Lopez-Cabarco 2021). While Liu et al. (2019) conducted their study in the nursing work environment, similar relationships have not been extensively explored in the food services industry. Given the unique stressors and high turnover rates in food services, it is crucial to demonstrate how OS may function differently in this setting. Therefore, this study aims to examine the effect of OS on the relationship between EE and TI in the South African food industry sector.

Turnover intention refers to an employee’s deliberate consideration or willingness to leave their current organisation (Hom et al. 2012, 2017). Turnover intention is characterised as the cognitive process of contemplating resignation and actively seeking alternative employment opportunities (Cicek, Turkmenoglu & Ozbilgin 2021). Because TI denotes an employee’s desire to leave the employer organisation for better working prospects, the immediate phase before leaving is the intention to depart, which is commonly seen as the intention to leave the employer within the next 6 months (Mathebula & Mitonga-Monga 2022; Rubenstein et al. 2019). Rajput and Kumari (2023) contend that an individual who intends to leave the employer portrays actual leaving behaviour. Individual employees may be willing to quit their employer organisation for several reasons, which include retirement, unfair labour practices, inadequate salary and working conditions, lack of competent supervision and support, lack of career growth and promotional chances and unfairness (Markham & Gobind 2022). Based on previous studies by Fried et al. (2008), role stress was found to have a moderate association with the intention to leave.

Furthermore, lack of good working conditions, inadequate pay, lack of support and job dissatisfaction are key factors that contribute to employee TI; in other words, these factors are inversely related to TI, as they often drive employees to consider leaving their organisation (Cicek et al. 2021). Employee turnover can be quite costly for employer organisations, as it affects the strategic objectives and competitiveness, resulting in additional costs of hiring and training new employees (Yang et al. 2022; Zhang et al. 2023). Employee turnover may be devastating for the employer if they lose competent employees, not only because they cannot replace them but also because they must train and advise newly hired workers to match the job and contribute to the organisation’s goals (Hou et al. 2021; Sun & Wang 2017). Employer organisations can reduce intention to leave by developing a high-performing culture (Markham & Gobind 2022), creating an ethical climate (Mitonga-Monga 2020) and providing a positive organisational justice characterised by fair human resource practices (Sharma & Tiwari 2023). This study examined TI with EE.

Emotional exhaustion is described as a significant element of career burnout that often occurs when an individual’s emotional resources are depleted because of experiencing stressful circumstances (Sun & Ma 2022). Maslach and Goldberg (1998:64) define EE as ‘feelings of being emotionally overextended and depleted of one’s emotional resources’. Shah et al. (2022) posit that working conditions, especially psychological aspects, are detrimental because they can either stimulate or impede an employee’s well-being and lead to withdrawal, disruptive behaviour and fuel intention to leave, which may affect organisational goal achievement. As an individual’s emotional resources, EE may occur for multiple reasons (Maslach & Jackson 1981). Firstly, several researchers have argued that EE is a critical element of burnout (Knudsen et al. 2009; Maslach, Jackson & Leiter 1996; Schaufeli & Taris 2005). Secondly, there is evidence that EE precedes the development of other components of burnout (Toppinen-Tanner, Kalimo & Mutanen 2002). Emotional exhaustion also has adverse health implications (Taris 2006) and has been linked to mediocre performance (Cropanzano, Rupp & Byrne 2003).

Furthermore, EE may spread to other organisational members (Halbesleben & Buckley 2004). In addition, EE can lead to several negative work-related physical and psychological outcomes such as high turnover, absenteeism, poor sleep, demotivation and role conflict (Park, Rhee & Lee 2021). Employer organisations must initiate a quality relationship and support good working conditions and a positive organisational atmosphere, which may help employees regulate their feelings and emotional competence when responding to their job demands (Ali et al. 2022). Implying regulating their feelings and emotional competence, employees will experience resource balances, leading to a low level of EE, low feedback avoidance and a prominent level of performance (Rajput & Kumari 2023). In contrast, if employees experience an important level of EE and then attempt to avoid feedback and portray unproductive behaviours (Korunka et al. 2008), all lead to job burnout from excessive job demands caused by negative employee behaviours such as absenteeism and TI (Shah et al. 2022). Employees who are aware of personal dysfunction owing to perceived conditions or events in the workplace are likely to experience an elevated level of occupational stress, which, in turn, would lead to a higher TI (Hills 2019).

Perceived organisational support is a construct of interest for both human resource practitioners and academics (Kurtessis et al. 2017). Organisational support is understood as an employee’s overarching belief regarding the degree to which their organisation values their contributions and prioritises their well-being (Eisenberger et al. 1986). Employees’ perceptions about the degree to which the employer values their well-being are entrenched, like the exchange relationship between the employee and employer (Zhai, Wang & Weadon 2020). Similarly, Mathebula and Mitonga-Monga (2022) postulate that employees are in an exchange relationship with the organisation for expected rewards for which they strive. The OST states that when the employer provides emotional or social support to their employees so that their needs are fulfilled, positive outcomes will emerge for both the employer and the employee (Eisenberger et al. 1986). Based on the exchange relationship mechanisms, employer organisations should support employees who are willing to trade their efforts to help them (Blau 1960; Mitonga-Monga 2020). This theory emphasises that employees who evaluate their employer’s policies, procedures and actions develop a broader view of the steps adopted by the employer are favourable for employees. For example, if the employer’s desired goals are met, employees would expect tolerance from their employer regarding sickness and mistakes in the future, better working conditions, fair rewards and opportunities for advancement Stefanidis & Strogilos (2020). Such support from the employer organisation would lead to positive work-related outcomes such as increased organisational commitment (Arasanmi & Krishna 2019), job satisfaction (Nijhawan et al. 2022) and reduced TI and absenteeism (Park et al. 2021). Employees who believe their employer organisation is supportive will more likely be willing to achieve their personal and organisational goals, effectively balancing the psychological resources. Alternatively, unsupportive employers foster a win-lose approach (Akhtar et al. 2018). Unsupportive employers may create a work environment in which employees’ individual goals are incompatible with other employees or even with the employer’s goals (Giao et al. 2020). This incompatibility may lead some employees towards goal attainment and others less likely to achieve their goals (Gurbuz, Turunc & Celik 2013). It is likely that an employer organisation that is supportive or non-supportive can influence whether there will be positive outcomes to reduce TI and EE (Caesens et al. 2017; Chatzittofis et al. 2021; Marchand & Vandenberghe 2016).

The COR theory is one of the most unified theories to explain the link between EE and TI (Saleh et al. 2023; Wright & Cropanzano 1998). According to this theory, individuals work actively to maintain, protect and build things that they consider to be centrally valued (Hbfoll 2011). The theory suggests that because individuals have the motivation to protect their resources when resources continue to decrease, those individuals will withdraw to protect their resources (Zhang et al. 2013). They suffer stress when adverse conditions develop (Knudsen, Ducharme & Roman 2006; Park et al. 2021). Drawing on this theory, we propose that TI is more likely to arise when employees feel that their emotional resources are emotionally overextended or depleted owing to a lack of fairness, good working conditions, adequate pay and opportunities for advancement (Arshadi & Shahbazi 2013; Nguyen et al. 2022). For example, a previous study by Cao and Qu (2014) found that EE related positively to TI. They imply that employees who experience high emotional exhaustion, weakness in functioning and adverse health outcomes because of stressful work conditions are likely to perform poorly and demonstrate negative behaviour such as absenteeism and turnover. A study by Hills (2019) revealed that employees who experience a prominent level of EE are likely to avoid or reduce resource losses by quitting their employer organisation. Wen et al. (2020) examined the effect of occupational stress on TI and found that an important level of occupational stress related to a high TI. A study by Knudsen et al. (2009) reported that EE related positively to TI. Overall, from a COR theory perspective, it can be explained that perceptions of work overload, lack of support, elevated levels of job demands and poor working conditions exacerbate the degree of employees’ EE directly. Moreover, the theory indicates that EE can lead the employer organisation to notice negative work behaviour on behalf of its members, which may, in turn, result in the intention to quit the job. Thus, the researchers hypothesised that EE relates positively to TI (see H₁).

Perceived organisational support is an individual’s belief that their employer organisation values their contribution and cares for their well-being, which can be seen as an essential job resource that can help to achieve job demand (Demerouti et al. 2001). Employees who have a positive belief in support from their employer for their career growth and who care about their career needs are likely to see the support that they receive as an essential resource to help them cope with stress while focussing more on work goals (Shabbir, Naz & Trivedi 2021). A positive feeling of perceived employer support relates to reduced EE and TI (Kurtessis et al. 2017). Also, employees with a high feeling of POS have a sense of work-related achievement (Sya & Mangundjaya 2020). A study by Yang (2018) emphasised the moderating effect of POS in the association between burnout and job satisfaction. Studies conducted on the relationship between POS and EE have established that with high POS, EE will decrease among employees (Ali et al. 2022), and they will contribute towards achieving the employer organisation’s goals (Palmer et al. 2017). For example, employees with high POS have been found to have a prominent level of affective commitment (Kurtessis et al. 2017). Srivastava and Agrawal (2020) claim that POS is a positive feeling that employees share, and its foundation is based on employees’ emotions. A study by Eisenberger et al. (2001) found that a joyous feeling of POS leads to a low level of EE and a high social exchange relationship, which in turn results in a positive work outcome for both employees and the employer organisation. There is also evidence that elevated levels of POS are likely to generate a feeling of employer support, increasing emotional resources and decreasing the intention to leave. It is, therefore, hypothesised that an important level of POS moderates the relationship between EE and TI (see H₂).

The following hypotheses were developed for the research:

The respondents completed a biographical and demographic survey, using the following standardised instruments, namely the Emotional Exhaustion Scale (EES) (Maslach & Jackson 1986), the Turnover Intention Scale (TIS) (Halbesleben & Wheeler 2008) and the Perceived Organisational Support Scale (POSS) (Eisenberger et al. 1986). The measurement qualities of these instruments are described in the next paragraph.

The EES was used to measure an employee’s emotions of being overwhelmed and exhausted by their work. The EES consists of 5 items, which are rated on a seven-point Likert scale (0 = Never, Everyday = 7). Maslach and Jackson (1986) reported a Cronbach’s alpha coefficient of 0.81 for internal consistency for the EES, while the present study realised a Cronbach’s alpha coefficient of 0.81 for the EES.

The TIS is a five-point measurement of participants’ assessment of whether they want to stay or leave the company. It is rated on a six-point Likert scale (0 = never, very satisfactory or very unlikely; 5 = always, unsatisfactory or very likely). Examples of the measurement included: ‘I intend to leave the company in the next 12 months’. Halbesleben and Wheeler (2008) provided a Cronbach’s alpha of between 0.88 and 0.91 for the TIS scale. This study resulted in a Cronbach’s alpha coefficient of 0.66 for the TIS.

The POSS was used to measure the degree to which employees perceived support from their organisation. The six POSS items were used in a five-point Likert scale ranging from ‘completely agree (5)’ to ‘completely disagree (1)’. Example items included: ‘The organisation values my contribution to its well-being’; ‘The organisation and my supervisor care about my well-being’; and ‘The organisation and supervisor care about my general satisfaction at work’. In terms of the POSS, Rhoades and Eisenberger (2002) reported a Cronbach’s alpha coefficient of 0.85 for the scores from the overall POSS. This study obtained a Cronbach’s alpha coefficient of 0.84 for the POSS.

The study’s collected data were analysed using the Statistical Packages for Social Sciences (SPSS, version 29.0). Firstly, confirmatory factor analysis (CFA) determined the factorial structure. The CFA was calculated using Varimax, with a Kaiser–Meyer–Olkin (KMO) measure of sampling precision greater than 0.50, factor loadings of 0.6 or higher and an eigenvalue of 1. The researchers chose to examine only components with a loading of 0.50 or higher, as indicated by Hair, Howard and Nitzl (2020). Secondly, structural modelling of POS, EE and TI was performed and the following goodness of fit indices (GFIs) were considered for all the constructs: the Chi-square (CMIN/df < 0.05); the GFI; the adjusted goodness of Fit Index (AGFI); the normed Fit Index (NFI); the comparative Fit Index (CFI); the Tucker–Lewis Index (TLI > 0.95) and the root mean square errors of approximation (RMSEA < 0.05). Thirdly, the descriptive statistics were calculated to determine the mean and standard deviations, and correlation analysis was calculated to establish the relationship between the variables. In the fourth stage, hierarchical moderator regression analysis was computed to verify if POS moderates the relationship between EE and TI. Before performing the moderator regressions, a collinearity diagnostic was performed to check whether the zero-order correlations were below the threshold of 0.80, that the variance inflation factors did not exceed 10, and that the tolerance values were close to 1 (Hair et al. 2020). The researchers conducted a simple propensity test to examine the interactions by varying the value of the moderator at different standard deviations from the mean (Hair et al. 2020). The critical value was set at a 95% confidence interval (p < 0.05) to reduce the likelihood of Type I errors. In this article, the practical significance of the R² values was assessed by calculating effect sizes (f²).

An application for full ethical approval was made to the University of Johannesburg’s Department of Industrial Psychology and People Management (IPPM) Research Ethics Committee and ethics consent was received on 03 June 2021. The ethics approval number is IPPM-2021-495(M).

Participants agreed to participate digitally in the survey after being informed in writing about the aim of the study and its voluntary nature. The researcher guaranteed the participants that their responses would remain confidential and anonymous. Participants received a Google Forms link from Alphabet, Inc., based in Mountain View, California, United States of America. They were instructed to complete the survey during their working hours.

To assess the factorial structure and reliability of the study’s variables, principal component analysis (PCA) with varimax rotation was employed. As shown in Table 1, the KMO index scores were 2.847 for EE, 2.305 for TI and 3.331 for POS. According to Table 1, the eigenvalues for all constructs in the study were > 1, and all factor loadings were above 50. Table 1 shows that all indices of the goodness of fit statistics were within the threshold: CMIN/df < 0.05; GFI, AGFI, NFI, CFI, TLI > 0.95; and RMSEA < 0.05 for POS, EE and TI.

Table 2 illustrates that, except for POS, the composite reliability of EE and TI was above 0.70. Table 2 indicates that in addition to TI, the average extracted variance for EE and POS was above 0.50. Also, Table 2 shows that the Cronbach’s alpha coefficients were between 0.66 and 0.84.

The descriptive statistics for the variables included in the study are presented in Table 3. The mean values ranged between (M = 2.50 to M = 3.15). Notably, POS exhibited the highest mean value (M = 3.15, s.d. = 0.87), followed by TI (M = 3.00, s.d. = 0.92), while EE displayed the lowest mean value (M = 2.50, s.d. = 0.86).

Analysis of the interrelationships between POS, EE and TI revealed significant correlations. The results show that POS related negatively to EE (r = –0.50; large effect size, p ≤ 0.05, and TI (r = –0.30; medium effect size, p ≤ 0.05). The results reveal that EE related positively with TI (r = –0.46; medium effect size; p ≤ 0.05). These findings provide empirical support for the hypothesised relationships between these variables and the theoretical frameworks underpinning this study.

As shown in Table 4, EE did not act as a predictor of TI for the main effects (F [3; 381] = 31.12; p ≥ 0.05), (B = –0.31; SEB = 0.28; 95% CI = [–0.86; 0.24]; p = 0.27). This means that EE was not associated with a decrease in the TI percentage. In addition, POS acted as a predictor of TI, as (F [3; 381] = 31.12; p ≥ 0.05), (B = –0.93; SEB = 0.23; 95% CI = [–1.38; –0.49]; p = 0.00). This suggests that POS was associated with a decrease in the TI percentage. The above results were found after analysing an interaction by using the moderator values at the mean and standard deviations above and below the mean (Hair et al. 2020). Table 4 illustrates that POS acts as a moderator in the relationship between EE and TI, as (F [3; 381] = 31.12; p ≤ 0.05), B = –0.22; SEB = 0.09; 95% CI = [–0.05; –0.38]; p ≤ 0.05). The data were depicted earlier in Figure 1.

The study’s findings have clearly established the role that OS can play as a significant variable in lessening EE and eventually reducing TI among food sector employees in South Africa. This is a clear call for human resource managers to make effective use of the given study findings by designing specific interventions that will improve POS such as introducing recognition programmes, developing a supportive work environment and promoting initiatives for work–life balance. Moreover, organisations can work at reducing the root causes of EE by establishing better workload management and role clarity and by providing sufficient resources for employees. If there is a focus on employee well-being and organisational culture, this will go a long way towards improving the retention of staff and the organisation’s overall performance.

This study is not without its limitations. The use of a convenience sample places limitations on generalisation to the broader population. Future studies may target more representative sampling to enhance external validity. Besides, cross-sectional design precludes causal inference. Longitudinal research designs will provide more insight into the temporal dynamics of the effects of EE, OS and TI. In addition, further examination of influences by other variables like job demands, job resources and individual differences may add more value to TI in the food sector.