The SET is widely applied to unravel the employer–employee relationship, especially in the employee turnover and retention literature (Coyle-Shapiro and Conway, 2005; Gopalan et al., 2020). According to this theory, a person, who benefits from someone, feels obligated to repay that person through positive behaviours and devotion. Furthermore, this theory postulates that employees deliver their optimum performance upon achieving support and perceiving value from their employers (Eisenberger et al., 2001). Hence, the theory is used to investigate the employees’ behaviour, enabling organisations to enforce certain HRM practices and igniting unique social exchange relationships.

From the SET perspective, employee retention can be induced by training and development offered by employers, which facilitate mutual benefits and create reciprocated obligations. This phenomenon occurs because individuals and organisations are involved in exchange relationships (Raihan, 2012). Employees perceive responsibility to repay their employers upon providing a conducive working environment. This repayment may derive in the form of increased loyalty, commitment, and stay for a long time (Settoon et al., 1996). Simultaneously, their job satisfaction and proper behavioural responses will increase, owing to the perception of fulfilling emotional needs (Iqbal and Hasnah, 2016; Latorre et al., 2016; Iqbal et al., 2017; Rubel et al., 2021) and improving employee retention (Rubel et al., 2021). Therefore, the current study posits that job satisfaction and work environment followed by training and development are positively related to employee retention.

A specific type of leadership is required to tackle the distinct needs and current challenges of a particular environment. Hence, according to the SLT, a single leadership style is insufficient for every situation (Hersey and Blanchard, 1969). In this case, effective leadership, that is, transformational leadership, emerges as a promising idea and is applicable across diverse fields. This concept facilitates the concept of adapting to varying circumstances and work environments (Hersey and Blanchard, 1969). In the context of employee retention, an environment must be structured where they feel empowered, valued, and connected to their employers (Ohunakin et al., 2019; Frye et al., 2020); thus, this leadership style is consistent with these requirements (Kim and Park, 2020). Consequently, this leadership style enhances the employees’ capability and reshapes the organisational image in the marketplace (Mwita and Tefurukwa, 2018; Singh et al., 2020). Drawing on the SLT, the current research claims the moderating role of transformational leadership on the relationship of training and development, work environment, and job satisfaction with employee retention.

Job satisfaction concerns employees’ evaluations of their jobs based on perceptions by comparing their actual job outcomes with desired ones (Schleicher et al., 2011). The concept is defined as a positive state where employees share their feelings about their job (Locke, 1976), ranging from moderate- to low-level satisfaction (Locke, 1976; Quigley et al., 2007). Furthermore, the idea is considered a causal factor that promotes intentions to stay with the organisation because it is a pleasant psychological state. In this case, individuals perceive content with their work (Fletcher et al., 2018). Therefore, job satisfaction fosters the social exchange between employers and employees, where satisfied employees exemplify positive experiences. This experience is supported by the social exchanges between the employee and the organisation, reinforcing their intentions to stay with the organisation (Koster et al., 2011).

Employees from varying organisations, industries, and geographical locations exhibit different levels of job satisfaction (AbuAlRub et al., 2009). However, job satisfaction positively affects employees’ intentions to stay irrespective of industries and region (AbuAlRub et al., 2009). For instance, a multi-level study found that job satisfaction is positively related to the employees’ intentions to stay in the united kingdom (UK) (Fletcher et al., 2018). Moreover, meta-analytic evidence has demonstrated that those who are satisfied at work are more likely to retain their employment (Fishbein and Ajzen, 2011); therefore, the hypothesis H3 is proposed on the basis of SET as follows:

Transformational leadership is considered highly effective in driving employee retention (Kossivi et al., 2016; Amankwaa et al., 2019), in which the leaders initiate, execute, and interconnect change with the desired outcomes (Bass and Riggio, 2006). This form of leadership influences employees by altering their perceptions, views, ambitions, and moral standards (Bass and Avolio, 1996). Transformational leaders also offer an ideal approach to employees and exhibit the attributes of significant faith, effective communication, uprightness, trust, and innovative thinking (Tian et al., 2020). These leaders promote intellectual stimulation, idealised influence and inspirational motivation, and individualised consideration. In addition, leaders can facilitate employees by understanding their issues and creating a psychologically safe environment through individualised consideration (Iqbal et al., 2021b).

Transformational leaders are viewed as role models and counsellors who encourage them to participate in organisational activities. Supervisor support, training, information accessibility, and counselling lead to higher employee retention (Ooi et al., 2021) and higher employability (Matsuo, 2021). Furthermore, healthy communication positively influences the work environment (Denton, 2011) via sharing of innovative ideas and intellectual stimulation, an indicator of open communication. This approach culminates in improved work that enhances talent management (Perlow and Kelly, 2014). Moreover, open communication and improving resource management in employee allocation ease employees’ adaptation to new work environments (Castrogiovanni et al., 2011).

Organisational support in the form of supervisor support, rewards, and favourable working conditions are vital to stimulating organisation-related outcomes, that is, reduced withdrawal behaviours and commitment (Rhoades and Eisenberger, 2002; Gillet et al., 2022). Supervisor support is critical to replenish employee physical and psychological resources to increase their retention probability (Kalliath and Kalliath, 2014). Furthermore, transformational leaders affect employees’ behaviour resulting in high employee retention (Sow et al., 2016). Leaders can develop a reward system within their organisation which is highly effective in retaining employees (Adekanbi, 2016). They must also ensure sufficient resources to employees following the organisational goals. Moreover, shared vision is directly related to the employees’ engagement (Boyatzis et al., 2017) and continuous improvement (Fardazar et al., 2015; Iqbal and Piwowar-Sulej, 2022). This leadership style presents a shared vision and elaborates the underlying reasons, enhancing their participation and involvement in decision-making. Previous studies found that transformational leadership negatively impacts employees’ turnover intention (Maaitah, 2018) but positively affects their knowledge base (Fletcher et al., 2018). For instance, a positive impact of various factors on employee retention, that is, idealised influence, inspirational motivation, individualised consideration, and intellectual stimulation (Jiang et al., 2017). Therefore, the following hypotheses, H4, H5, and H6, are proposed:

With reference to the above discussion, the hypothesized model is drawn as shown in the below Figure 1.

In China, organisations face fierce pressure to retain employees due to the shortage of skilled and talented employees (Fu et al., 2020). Statistically, 44% of the top management in organisations operating in China view this issue as a critical barrier to employee retention (Mashiah, 2021). China’s manufacturing makes up two-thirds of SMEs (Zhu et al., 2012; Iqbal et al., 2021c; Xuecheng et al., 2022), where most operate in the northern region. Therefore, this study focuses on the SME employees in this region. Moreover, this study requires a minimum of 185 responses based on the sample-item ratio (Hatcher and O’Rourke, 2013). Online survey forms were structured to collect data, comprising six sections that measure various factors. The factors include training and development, work environment, job satisfaction, employee retention, transformational leadership, and respondents’ demographics.

We adopted a convenience sampling approach for data collection, considering China’s time and financial constraints and current COVID-19 restrictions. The present study is cross-sectional in design, where data are collected at a specific time from the employees. The online survey link was disseminated via 500 email addresses with the assistance of the human resource department. Accordingly, 287 responses were received, which is sufficient, indicating a response rate of 57.40%. Furthermore, a gentle reminder was included during data collection to increase the number of responses. In this study, we also marked it mandatory to check against each item in the online survey form, ensuring no missing values in the dataset.

We adopted measurement scales of four continuous variables in the current study. Previous studies have reported reduced quality and high cognitive ability required to collect data using a high Likert scale (Cummins and Gullone, 2000; Iqbal et al., 2020). Therefore, a 5-point Likert scale was employed, ranging from 1 = strongly disagree to 5 = strongly agree. Delery and Doty (1996) have defined training and development as the degree to which organizations offer training to employees to foster their skills. We adopted the 4-item scale from Delery and Doty’s (1996) study to measure training and development. For example, one item is “We receive formal development training which increases our promotion chances within the organisation.” Similarly, Bibi et al. (2018) used an identical scale in the context of Pakistan and found it highly reliable (α = 0.918). In this study, the Cronbach’s alpha value of this scale is 0.719.

The work environment is defined as the degree to which employees perceive the availability of a safe and conducive workplace (Edgar and Geare, 2005). Subsequently, 4-measurement items were adopted from Bibi et al.’s (2018) study to assess the work environment. For instance, one item is “We always feel safe working here in this environment.” In the current study, the Cronbach’s alpha value of this 4-item scale is 0.928. Meanwhile, job satisfaction is defined as the pleasurable emotional state emerging from the job appraisal as facilitating the accomplishment of one’s job values (Locke, 1976; Zhang M. M. et al., 2016). In this case, we adopted a 3-item scale from Cammann et al.’s (1979) study to measure job satisfaction. An example of this item is “In general, I like working here.” This scale was utilised by Zhang L. et al. (2016), who found it highly reliable (alpha = 0.870). Accordingly, the Cronbach’s alpha value in this study for the 3-item scale is 0.921.

Transformational leadership is defined as those who have idealised influence, intellectual stimulation, inspirational motivation, and individualised consideration. We adopted 20 items of the Multifactor Leadership Questionnaire (MLQ) to analyse transformational leadership. This analysis was related to the four items, namely intellectual stimulation, idealised influence, inspirational motivation, and individualised consideration. A sample of the item is “my leaders give me tasks with enthusiasm.” A previous study (Ohunakin et al., 2019) indicates high reliability where Cronbach’s alpha values of its four dimensions were between 0.88 and 0.92. In this study, the Cronbach’s alpha values are in the range of 0.875–0.918.

Employee retention is defined as the effort by an organisation to keep desirable employees to fulfil business objectives (Frank et al., 2004; Govaerts et al., 2011). We adopted six items (Govaerts et al., 2011) to measure employee retention, for example, “I love working for this company.” The present study’s measurement scale was highly reliable, that is, Cronbach’s alpha = 0.794, aligning with Khalid and Nawab’s (2018) findings.

In this study, the research framework is complex due to its prediction-oriented feature and the presence of independent variables, dependent variables, and moderators. Therefore, partial least squares-structural equation modelling (PLS-SEM) was selected following Hair et al. (2020). This approach is considered a proper analytical strategy compared to covariance-based structural equation modelling (CB-SEM) (Ringle et al., 2020). The technique revolves around the assessment of the measurement model and structural model, though it requires prior evaluation of the former measurement.

Before analysis, data screening must be conducted, which concerns missing values, outliers, data normality, and common method bias. We ensured the absence of missing values in the current dataset by marking mandatory against each measurement item in the online survey form. Moreover, univariate outliers and multivariate outliers were investigated through Z-score and the Mahalanobis distance test. Accordingly, three responses were removed in the univariate outlier because of the Z-score values greater than 3.29 (Tabachnick and Fidell, 2007). Meanwhile, the Mahalanobis distance test revealed that the P-value of 15 cases is less than 0.001, a clear indicator of multivariate outliers; therefore, the 15 responses were excluded from the dataset.

The normality was assessed based on the skewness and kurtosis values, and in this study, the skewness values of all continuous variables extend from −1.307 to 1.531, which are within ± 3. However, the kurtosis values fall out of the ± 3 range (DeCarlo, 1997), and thus, the data are not univariate normal. The Mardia’s coefficient skewness 0.005 and kurtosis values (β=172.761,ρ < 0.005) confirmed the multivariate normality. Nevertheless, the PLS-SEM does not require data normality; hence, non-normal distribution is not an issue. Next, we applied Harman’s single factor test and the correlation matrix procedure to examine the common method bias. Harman’s single factor test revealed that a single factor only counts for 39.41% < 0.50% of the total variance (Podsakoff et al., 2012), confirming no issues with method bias. The correlation matrix exposed that no single correlation between continuous variables is greater than 0.90 (Bagozzi et al., 1991); thus, the current study is free from common method bias. We have also examined the model fit of hypothesized model based on two-index strategy recommended by Hu and Bentler (1999). In the present study, hypothesized model was found fit based on comparative fit index (CFI) (0.96 > 0.95) and standardised root mean square residual (SRMR) (0.049 < 0.09) (Hu and Bentler, 1999), as compared to alternative models.

The present study is dominated by male participants (n = 176, 61.32%), where most participants (n = 109, 37.98%) fall between the ages of 25 and 35, followed by 85 between 36 and 45. The most significant number of participants (n = 144, 50.17%) in current research possessed a bachelor’s degree and nine participants (3.14%) with PhDs. Finally, more than 50% of the participants (n = 153) acquired 5–10 years of working experience, while nine participants with more than 20 years of professional experience.

In this study, the mean values of training and development (M = 4.145), work environment (M = 4.334), job satisfaction (M = 4.322), and employees’ retention (M = 4.167) were found significant. Notably, job satisfaction and employee retention values were higher than those reported by a previous study among employees of multi-national enterprises (MNEs) in China (Zhang M. M. et al., 2016). Regarding four dimensions of transformational leadership, idealised influence (M = 4.132) has the highest mean value. This result is followed by inspirational motivation (M = 4.121), individualised consideration (M = 3.973), and intellectual stimulation (M = 3.670). Moreover, the current descriptive analysis revealed that participants seek more idealised influence than other dimensions of transformational leadership in Chinese SMEs. In other words, a leader’s ability to exhibit high morality, ethics, and personality enables the realisation of high performance within organisations.

The measurement model analysis examines the construct reliability and validity, in which the former is assessed with reference to its indicator and internal reliability. This study revealed that all indicator loadings are more significant than 0.50 and less than 0.944, which are deemed acceptable. An item has sufficient indicator reliability provided its factor loading is more significant than 0.50 (Chin, 1998). Moreover, we examined the construct reliability based on Cronbach’s alpha and composite reliability values. A construct has acceptable reliability, provided its Cronbach’s alpha or composite reliability value is greater than 0.70 (Sarstedt et al., 2019).

Table 1 shows that the Cronbach’s alpha values of various items are greater than 0.70. These items comprise training and development (0.719), work environment (0.928), job satisfaction (0.921), employees’ retention (0.794), and individualised consideration (0.889). Other items include idealised influence (0.918), inspirational motivation (0.875), intellectual stimulation (0.882), and transformational leadership (0.885). Similarly, the composite reliability values of these variables are greater than 0.70 (see Table 1). Hence, it is evident that all the continuous variables exhibit acceptable construct reliability.

Construct validity is formulated on the convergent and discriminant validity, where a construct has sufficient acceptable convergent validity provided its factor loadings are greater than 0.70. Furthermore, its average variance extracted (AVE) must be higher than 0.50 (Hair et al., 2020). The items of all continuous variables must possess factor loadings greater than 0.70 (see Table 1). The AVE values of multiple items revealed values higher than 0.50, that is, training and development (0.541), work environment (0.823), and job satisfaction (0.864). Others include employee retention (0.576), individualised consideration (0.706), idealised influence (0.712), inspirational motivation (0.669), intellectual stimulation (0.692), and transformational leadership (0.677) (see Table 1).

Based on the above results, all continuous variables possess acceptable convergent validity. We employed Fornell–Larcker Criterion to examine the discriminant validity of the variables. This method confirms the discriminant validity of a variable provided that the square root of its AVE is greater than its inter-constructs correlation values (Henseler et al., 2009). Table 2 indicates that the square root of AVE of all variables is greater than their inter-constructs correlation values; hence, these variables exhibit acceptable discriminant validity.

The structural model analysis revealed that training and development significantly influence employee retention (β=0.824,ρ < 0.05) (Table 3). In essence, one unit change in training and development brings 82.40% variations in employee retention, and thus hypothesis H1 is supported. Notably, the findings indicated the significant positive impact of the work environment on employee retention (β=0.274,ρ < 0.05), supporting hypothesis H2. Meanwhile, job satisfaction significantly influences employee retention (β=0.824,ρ < 0.05) (see Table 3). In other words, there is a 20% change in employee retention among SME employees in China for one unit change in job satisfaction, supporting hypothesis H3.

The effect of its interaction terms with training and development, work environment, and job satisfaction were estimated. This approach was conducted to examine the moderating effect of transformational leadership. In this case, the interaction term of transformational leadership with various dimensions does not significantly influence employee retention in SMEs in China. The dimensions include training and development (β = −0.081,ρ=0.233 > 0.05), work environment (β=0.021,ρ=0.856 > 0.05), and job satisfaction (β=0.078,ρ=0.101 > 0.05) (see Table 3). Therefore, moderation hypotheses H4, H5, and H6 are rejected.

The present study significantly contributed to the extension of the theory. First, the literature is enriched by offering empirical support on the integrated understanding of the three dimensions and employee retention. Furthermore, extant literature provided contradicting findings on the relationship between the three dimensions. However, the role of transformational leadership as the conditional variable in their relationship is still missing. Second, this study augments the SET by examining the effect of training and development, work environment, and job satisfaction on employee retention. Third, the SLT is extended by providing evidence on the non-significant moderating role of transformational leadership, specifically the correlation of the three dimensions on employee retention.

The current research offered several recommendations for practitioners and policymakers. Extant literature claims that employee retention is a significant challenge for SME employees in China. In this case, the current work emphasises the three dimensions critical to enhancing employee retention within China’s SMEs. The present findings found that training and development is the most crucial element which increases employee retention within organisations. Chinese SMEs may, therefore, plan their training and development programme more effectively based on their employee’s needs.

China’s SMEs should design training and development programmes that parallel their employees’ career growth. However, previous findings focused on offering training programmes related to company-specific skills rather than general skills. Accordingly, general skills enable employees to effectuate professional opportunities outside the parent organisations. Therefore, these SMEs must evaluate their programmes and perform changes concerning their company-specific skills. Equally important, the top management should employ specific strategies to foster a conducive workplace to improve the work environment and cope with employees’ burnout. Such strategies may include special counselling sessions for stress-induced employees and enforcing flexible work schedules.

The management should centre on developing the work environment, ensuring satisfied employees, establishing open communication, and fostering ideas while offering peer support. Moreover, practitioners should work on the proper fund allocation to develop a positive work environment. The current study reported a significant positive effect of job satisfaction on employee retention. Hence, the human resource and departmental managers are advised to make incremental changes, spurring employee satisfaction, that is, providing a fair salary, rewards, and incentives to their employees. Nevertheless, the moderating effect of transformational leadership does not appear significant in retaining employees for SME management in China.

Based on the above, it is proven that transformational leaders are highly effective in designing an empowered and meaningful work environment. For instance, this form of leadership offers individualised consideration and idealised influence, stimulates intellectually, and motivates inspiration. Thus, practitioners and managers should evaluate the development of leadership. Specifically, in China’s SMEs, policymakers, and managers must critically assess the leadership development programmes in their organisations.

The current study possesses several limitations despite the significant theoretical and practical contributions. First, this study was conducted in the northern region of China, hence, increasing the generalisation in the context of China. Moreover, the sole focus on China SMEs may not provide comprehensive information on other emerging and developed economies. Therefore, future studies can enrich the quantitative findings by assessing other developing countries, such as Pakistan and India. The survey form was distributed through the human resource management department, decreasing the chance of randomly distributing the survey form to employees. This situation will hinder the generalisation issue further, especially in a broader population.

Second, this study subjectively measured all continuous variables, and such measurement might influence the presence of common method bias. Practitioners and academicians face extreme difficulty in collecting objective data from organisations. Thus, future endeavours could overcome these limitations by adopting improved research design and employing a qualitative approach to unravel the causal relationship. Third, a direct connection was found between the three dimensions with employee retention. The upcoming research must then investigate the potential mechanisms of these relations. Finally, the current study provided multiple shreds of evidence about the non-significant moderating role of transformational leadership; thus, further examinations must be made on the conditional part of other leadership styles, such as sustainable and servant leadership.