Determinants of Accountants’ Turnover Intention in Independent CPA Firms: Empirical Evidence from Vietnam

Introduction

Accountants must have great professional competence, high ethical standards, and well-developed professional skills in a world economy that is becoming more unstable and competitive. The nature of audit work, characterized by intensive audit cycles, multiple concurrent deadlines, and strict compliance with professional ethics such as integrity and objectivity, exposes accountants to heightened risks of work overload and occupational burnout. These psychological and job-related pressures constitute important antecedents of turnover intention (TI) among accountants. The departure of accountants not only results in significant recruitment and training costs but may also undermine service quality, team morale, and the professional reputation of CPA firms. Therefore, identifying the factors that shape turnover intention and clarifying their underlying mechanisms is essential for modern human resource management within the auditing profession.

The psychological mechanisms through which work overload (WO) influences accountants' decisions to leave an organization remain understudied, especially in the Vietnamese auditing context, although many investigations have looked at the relationship between job-related factors and TI. In addition, limited empirical evidence exists regarding the moderating role of personal resources-such as individual motivation (MO)-in buffering or mitigating these adverse effects.

The current study intends to close these gaps by: (1) examining the direct effects of WO, burnout (BU), work-life balance (WLB), and other control factors like job attitude (JA) and organizational justice (OJ) on TI; (2) examining the mediating roles of burnout and WLB in the relationship between WO and TI; and (3) evaluating the moderating effect of motivation (MO) on the relationship between WLB and TI.

This study advances the theoretical knowledge of turnover behavior and offers useful empirical data to assist CPA companies in creating successful human resource plans by utilizing an integrated research methodology based on the Job Demands-Resources (JD-R) theory. These insights help minimize turnover risks, strengthen employee commitment, and sustain a stable, high-quality professional workforce.

WO has a positive and statistically significant direct effect on TI and indirectly increases TI through burnout (BU), according to findings from a study of 225 accountants in Vietnam. This confirms burnout as a crucial mediating mechanism. WLB negatively influences TI, underscoring its protective role. JA demonstrates a dual effect-enhancing WLB while simultaneously increasing BU, resulting in an overall ambiguous net impact on TI. Notably, MO acts as a positive moderator that strengthens the protective influence of WLB on TI. These findings provide additional empirical support for the JD-R framework and suggest that CPA firms should concurrently reduce work pressure, manage burnout effectively, and foster mindfulness and motivation to improve retention and mitigate turnover risk among accountants.

By elucidating the mechanism WO → BU → TI and emphasizing the moderating function of motivation in the link between WLB and TI, this study further strengthens and expands the JD-R theoretical framework. Additionally, the study sheds light on the dual nature of job attitude, serving as a resource that enhances balance while also potentially contributing to overload and exhaustion, thereby offering empirical evidence for the “engagement-exhaustion paradox.” Finally, the study provides important empirical support for the regulatory effect of psychological factors, particularly motivation, in shaping the link between WLB and TI.

Beyond the introduction, conclusion, and implications, the article presents a comprehensive literature review, research model, empirical findings, and discussion.

Literature Review

In the accounting industry, employee turnover has long been a major concern, especially for CPA companies, both Big4 and non-Big4. High TI increases recruitment and training costs, reduces audit quality, and adversely affects firms’ professional reputation (Kyire et al., 2023; Poornachitra & Maheswari, 2023; Ma et al., 2024). Therefore, understanding the determinants of accountants’ TI is essential for developing effective human resource strategies.

Previous research has often used the JD–R model (Demerouti et al., 2001) and Role Stress Theory (Fogarty et al., 2000). According to these frameworks, employees are more likely to feel stress, burnout, and eventually TI when job expectations are not met by sufficient resources.  Furthermore, it has been demonstrated that both individual motivation and organizational fairness are crucial in reducing the detrimental consequences of job demands (Nieoczym & Rybak, 2023; Alves et al., 2024; Prada et al., 2024).

Theoretical Foundations

Job Demands-Resources Model (JD-R)

A thorough theoretical framework for elucidating how work conditions impact employee well-being and behavioral outcomes is provided by the JD-R model, which was first put forth by Demerouti et al. (2001) and subsequently updated by Bakker and Demerouti (2017). Job characteristics are divided into two main groups under the JD-R model:

- Job demands include things like severe workloads, strict deadlines, and peak-season audit pressures that call for consistent mental or physical effort. Prolonged exposure to high job demands can result in elevated stress levels and burnout.

- Job Resources: These include factors that facilitate task completion, alleviate job demands, and enhance motivation, for example, supervisory support, flexible work arrangements, and opportunities for career development.

According to the JD-R model, burnout and increased TI result from a health-impairment process that arises when job demands surpass available resources. Conversely, strong job resources promote motivation, engagement, and resilience, thereby reducing the adverse effects of demanding work conditions (Demerouti et al., 2001; Bakker & Demerouti, 2017).

Social Exchange Theory (SET)

Homans (1958) proposed Social Exchange Theory (SET), which was further conceived by Blau (1964), arguing that workplace relationships are based on reciprocal trades between employees and their businesses.  Workers are more likely to react with loyalty and organizational dedication when they perceive chances for progress, fairness, and support. Conversely, perceptions of unfairness may weaken employees’ commitment and heighten their TI (Cropanzano & Mitchell, 2005).

Overview of Determinants and Hypotheses

Work Overload / Work Demand

WO is one of the most extensively examined constructs in the accounting and auditing literature and is considered a defining characteristic of the profession. Accountants in CPA companies frequently put in 65 to 80 hours a week during busy times, which is more than the amount of time they think audit quality starts to decline. Extended working hours, tight deadlines, and insufficient staffing heighten the risk of reduced audit quality (Persellin et al., 2019).

Alves et al. (2024), using a sample of 301 accountants in both Big4 and non-Big4 in Portugal and employing structural equation modeling (SEM), found that WO negatively affects WLB, which in turn increases TI. Their study also documented clear distinctions between Big4 and non-Big4 environments: Big4 firms tend to exhibit more competitive work settings and heavier workloads, yet concurrently offer greater opportunities for learning and career advancement.

In Indonesia, Pradana and Salehudin (2015) surveyed 141 early-career accountants working in independent CPA firms and applied SEM for data analysis. According to their findings, WO increases TI both directly and indirectly through increased stress, a worsened WLB, and lower job satisfaction. This reflects the vulnerability of young accountants, particularly newcomers, to high job demands and limited organizational support.

In Vietnam, Pham et al. (2022) found that WO and working conditions are primary turnover drivers. Their study emphasizes that professional certification pressure and organizational culture significantly influence the engagement and retention of Vietnamese accountants.

H1: Work overload positively affects accountants' TI.

Work-Life Balance

The ability to preserve harmony between professional obligations and personal demands, such as family, health, and social activities, is known as WLB. When WLB is achieved, employees can allocate their time and energy effectively, thereby reducing stress and enhancing job satisfaction.

In the accountancy profession, the peak-season workload, characterized by extended working hours and high time pressure, often compromises WLB. Previous research has demonstrated that WLB acts as a mediating element in the relationship between WO and TI in addition to having a direct and negative impact on TI (Pradana & Salehudin, 2015; Alves et al., 2024). Furthermore, Nazaripour and Babazadeh Hashinon (2025) stress that this effect is shaped by individual motivation; in particular, greater motivation can mitigate the detrimental effects of WO on WLB, allowing accountants to sustain balance and long-term commitment to their organizations.

H2: Work-life balance negatively affects accountants'TI.

H3: Work-life balance mediates the relationship between work overload and TI.

Burnout

When people are subjected to ongoing work pressures that surpass their capacity for adaptation, they experience burnout, a state of physical, emotional, and mental tiredness (Maslach et al., 2001; Efremov, 2023). In the auditing profession - where high responsibility, strict deadlines, and substantial workloads are inherent characteristics - burnout is considered a particularly serious issue, directly influencing job performance and long-term career commitment.

According to Role Stress Theory, burnout typically stems from role conflict, role ambiguity, and role overload. When these stressors persist over time, accountants tend to experience chronic strain and diminished motivation, which ultimately elevates their TI (Fogarty et al., 2000; Nguyen & Hoang, 2022; Trung et al., 2022; Ncube et al., 2023).

Burnout is a direct result of role stress and one of the best indicators of TI, according to Fogarty et al. (2000), a study that is considered a landmark in the accounting and auditing field.  Building on this framework, Smith, Emerson, and Everly (2017) presented additional data demonstrating that the two main mediating mechanisms connecting role stressors to detrimental work outcomes, such as decreased performance, lower job satisfaction, and increased intention to leave the profession, are stress arousal and burnout. Their findings highlight that burnout - particularly emotional exhaustion - is the critical pathway through which job pressure translates into turnover behavior.

More recently, Weissmantel et al. (2024) offered international evidence from Brazil, confirming that burnout exerts a strong direct influence on TI while simultaneously serving as the principal explanatory mechanism connecting work demand to TI.

H4: Burnout positively affects accountants' TI.

H5: Burnout mediates the relationship between WO and TI.

Job Satisfaction & Organizational Commitment

Two important factors that are regularly associated with lower TI in the accounting profession are job satisfaction and organizational commitment. Accountants are more likely to be satisfied and show a greater long-term commitment to their companies when they believe that their work is relevant, that their contributions are valued, and that they have prospects for career progression (Tam et al., 2023; Alves et al., 2024; Cissé et al., 2024; Kariri et al., 2024).

The critical functions of these elements have been highlighted by recent studies.  According to Alves et al. (2024), organizational commitment and job satisfaction both directly lower TI and somewhat moderate the link between WO and TI. Weissmantel et al. (2024) further found that strong organizational commitment buffers the negative impact of work demand on TI, particularly within independent CPA firms. In the Vietnamese context, Pham et al. (2022) confirmed that job satisfaction and organizational commitment function as protective factors that significantly decrease the likelihood of accountants leaving the profession, especially among younger employees working in Big4 firms.

H6: Job satisfaction negatively affects accountants' TI.

H7: Organizational commitment negatively affects accountants' TI.

Organizational Justice

OJ reflects employees’ perceptions of fairness in various aspects such as compensation, promotion opportunities, decision-making procedures, and work assignment practices. When accountants perceive fairness within their organizations, they are more likely to develop trust, strengthen their work motivation, and exhibit lower TI.

Recent studies continue to emphasize the importance of OJ in sustaining and developing human resources in the accountancy profession. Zhao et al. (2024) reported that both distributive justice and procedural justice exert strong negative effects on TI, indicating that when employees perceive transparency and fairness, they are less inclined to leave the organization (Machate et al., 2022). Hajanirina (2025) further demonstrated that OJ indirectly reduces TI through affective commitment, particularly among millennial accountants - a group highly sensitive to workplace transparency and fairness.

In a similar vein, Putri and Christi (2024) found that lower TI is linked to higher levels of perceived OJ in a study of accountants in Bali (Brekeit et al., 2022). They also confirmed that the effect of job stress on TI is amplified when accountants experience high levels of WIPL, underscoring the importance of both WLB and OJ in keeping accountants in CPA firms.

H8: OJ negatively affects accountants' TI.

Motivation - Moderator

In a similar vein, Putri and Christi (2024) found that lower TI is linked to higher levels of perceived OJ in a study of accountants in Bali (Wal et al., 2023). They also confirmed that the effect of job stress on TI is amplified when accountants experience high levels of WIPL, underscoring the importance of both WLB and OJ in keeping accountants in CPA firms.

Nazaripour and Babazadeh Hashinon (2025), surveying 239 accountants in Iran, demonstrated that motivation serves as a significant moderator in both the WLB-TI relationship and the relationship between WO and WLB (Danchin et al., 2024). Their findings reveal that when accountants exhibit high motivation, they are better able to maintain balance and job satisfaction even under heavy workloads, thereby reducing their TI. Conversely, low motivation magnifies the negative consequences of poor WLB, increasing susceptibility to burnout and intention to leave the profession.

Complementing these results, Alves et al. (2024) showed that motivation not only moderates the relationship between WO and WLB, but its effects vary according to firm type. In Big4 CPA firms - where job demands and performance expectations are more intense - intrinsic motivation plays a crucial role in sustaining WLB and reducing TI (Balaji et al., 2022). Their findings indicate that highly motivated accountants in Big4 firms are better able to buffer the negative impact of WO on both WLB and TI, whereas in non-Big4 firms, motivation primarily exerts its influence through organizational commitment.

Materials and Methods

Research Model

This study suggests the following research models based on an analysis of earlier representative studies on the factors influencing accountants' propensity to leave their jobs:

The first model examines the direct effects of the key factors on turnover intention.

As a result, TI (turnover intention) is the dependent variable; WO (work overload), OJ (organizational justice), JS (job satisfaction), and OC (organizational commitment) are the independent factors; and WLB (work-life balance) and BU (burnout) are the mediating variables.

The models examine the effects of WO on TI through the mediating roles of WLB and BU, while also assessing the direct effects of WLB and BU on TI.

The indirect and direct effects of WO through WLB and BU are determined as follows: DE = c

The total effect of WO: TE = c + α1b1 + δ1b2

The model investigates how MO moderates the association between TI and WLB:

Accordingly, MO serves as the moderating variable; φ represents the interaction coefficient, and if φ  ≠ 0, it indicates that motivation changes the strength of the effect of WLB on TI.

The marginal effect of WLB at different levels of MO is determined as follows:

The measurement scales for the variables included in the research model are presented in Table 1.

Table 1. Measurement Scales and Observed Variables

Source: Compiled by the author

Because the variables employed in the model are latent constructs, the study conducts Cronbach’s Alpha reliability testing to assess the measurement quality. The results indicate that all scales achieve a Cronbach’s Alpha greater than 0.6, and all observed items have a corrected item–total correlation above 0.3. These results confirm that the measurement scales in this study exhibit satisfactory reliability, and no observed variable was removed, as both criteria recommended by Nunnally and Bernstein (1994) are satisfied: a Cronbach’s Alpha greater than 0.6 and a corrected item–total correlation above 0.3; when Cronbach’s Alpha exceeds 0.8, the scale is considered highly reliable

After assessing the reliability of the measurement scales, we proceeded to conduct Exploratory Factor Analysis (EFA) to evaluate the underlying structure of the scales used in the research model, determine the degree to which the observed variables converge into their respective factors, and eliminate any items that do not load appropriately. A summary of the EFA results is presented in Table 2.

Table 2. Summary of EFA Results

Source: Author’s computation

The EFA results reported in Table 2 indicate that the KMO values for all factor groups exceed 0.5, satisfying the requirement for sampling adequacy. Bartlett’s Test of Sphericity yields a significance level of 0.000 for all constructs, suggesting that the correlations among items are sufficiently strong for factor analysis. All factor loadings are greater than 0.5 (most above 0.7), indicating that the observed variables adequately represent their latent constructs. The extracted variance for every factor is greater than 70%, exceeding the traditional threshold of 50%, and the eigenvalues are greater than 1. Thus, all measurement scales exhibit satisfactory convergent validity, and no item needs to be removed based on the EFA criteria. 

However, the EFA results in Table 2 show that the observed variables of Organizational Commitment (OC) and Job Satisfaction (JS) load onto the same factor rather than separating into two distinct factors as initially expected. This suggests that, within the context of this study, employees may not clearly distinguish between their commitment to the organization and their satisfaction with it; instead, they perceive these as interconnected dimensions of positive attitudes toward their job and their organization. Therefore, in this study, the author combines OC and JS into a single construct, JA. After merging JS and OC into JA, we re-ran the Cronbach’s Alpha test for the JA construct. The results show a Cronbach’s Alpha of 0.940, with all corrected item–total correlations exceeding 0.7, confirming excellent internal consistency.

Because the independent variables in the model are latent constructs measured through multiple items, we employ Principal Component Analysis (PCA) to construct composite variables for TI, WO, WLB, BU, OJ, JA, and MO. To assess the suitability of the data for PCA, both Bartlett’s Test of Sphericity and the Kaiser–Meyer–Olkin (KMO) measure were conducted. As reported in Table 3, all variables (TI, WO, WLB, BU, OJ, JA, and MO) exhibit KMO values greater than 0.5 and within the acceptable range of 0 to 1. Moreover, Bartlett’s tests for all constructs yield p-values lower than the alpha significance level of 0.01. Following Hair et al. (2006) and Tabachnick et al. (2007), these results confirm that PCA is appropriate for measuring TI, WO, WLB, BU, OJ, JA, and MO in this study. Accordingly, PCA is employed to compute the composite scores for these constructs.

Table 3. Scale Validity and Reliability Test Results

Source: Author’s computation

Data Source

All variables in the research model are latent constructs measured by multiple items; therefore, the parameters for exploratory factor analysis (EFA) are used in this study to estimate the sample size. According to Hair et al. (2006), the required sample size (n) can be estimated as: n=kj=1mPj  where Pj is the total number of observed items and m is the number of measurement scales.

Given the constraints in survey resources and the characteristics of our target population - employees working in the accountancy and auditing profession - we adopt k = 5 for this study.

The research model consists of 8 measurement scales with a total of 36 observed variables, resulting in a minimum required sample size of: n = 5 x 36 = 180. Employees of CPA companies participated in the poll, and 225 valid responses, more than the required number, were gathered. As a result, the sample size is thought to be completely sufficient for the analysis.

Among the 225 respondents, 78 were male (34.7%), and 147 were female (65.3%). Regarding job positions, 67 were Assistants (29.8%), 34 Interns (15.1%), 45 Managers (20.0%), 58 Seniors (25.8%), and 21 at the Manager/Director/Partner level (9.3%). In terms of firm type, 78 respondents (34.7%) worked in Big4 firms, and 147 respondents (65.3%) worked in Non-Big4 firms.

Results and Discussion

Estimation Results

The descriptive statistics of the variables used in the empirical model are presented in Table 4. Among all measurement items, only two burnout indicators - BU2 and BU3 have average scores below 3. All remaining measurement items exhibit mean values above 3.

Table 4. Descriptive Statistics of Variables Used in the Model

Source: Author’s computation

Baseline Regression Results

The estimation results for Models (1), (2), and (3) are presented in Table 5. Model (1) shows the effects of all independent variables and mediators (WLB and BU) on the dependent variable (TI). The coefficients of BU and WLB in this model represent their direct effects (DE) on TI. Model (2) examines the relationship between the independent variable (WO) and the first mediator (WLB), while Model (3) captures the relationship between the independent variable (WO) and the second mediator (BU).

Table 5. Estimation Results for Models (1), (2), and (3)

Source: Author’s computation

Notes: Values in parentheses are standard errors. Significance levels: ∗∗∗p<0.01, ∗∗p<0.05, *p<0.10.

As shown in Table 5, all models are statistically significant (F-test, p < 0.01). In particular, Model (1) yields an R² of 0.5162, indicating that the model is well-fitted and that the independent variables jointly explain 51.62% of the variation in TI.

In Model (1), WO exerts a positive and highly significant effect on TI (coefficient = 0.363; p < 0.01), indicating that greater work overload substantially increases accountants’ turnover intention. This finding is consistent with prior studies (e.g., Pham et al., 2022), which emphasize that the high-pressure working environment in the accountancy profession significantly heightens turnover risks (Çınaroğlu et al., 2023; Yang et al., 2023; Delcea et al., 2024). Conversely, WLB has a negative effect on TI (coefficient = –0.1181; p > 0.10), suggesting that maintaining a balance between work and personal life may help reduce TI, although the effect lacks statistical significance.

Notably, BU has a strong and statistically significant impact on TI (coefficient = 0.4751; p < 0.01). This indicates that when accountants experience emotional exhaustion and psychological fatigue, their likelihood of wanting to leave their job increases markedly. This result reinforces earlier findings (Fogarty et al., 2000; Smith et al., 2017; Weissmantel et al., 2024), which identify burnout as a core mediating mechanism linking job demands to TI among accounting professionals. In addition, OJ shows a positive and statistically significant effect on TI (coefficient = 0.2554; p < 0.01), whereas JA (Job Attitude) has a negative but statistically insignificant effect (coefficient = –0.1728; p > 0.10), suggesting that the level of organizational attachment is not sufficiently strong to reduce turnover intention in this sample. Overall, Model (1) provides evidence that WO and burnout are the two strongest determinants of TI among accountants working in CPA firms.

In Model (2), WO has a negative and statistically significant effect on WBL (β = –0.227; p < 0.01), indicating that higher work demands make it more difficult for accountants to maintain a balance between work and personal life. This relationship reinforces the argument that misalignment between professional demands and personal time is a key contributor to job-related stress and burnout. JA exhibits a positive and statistically significant impact on WBL (coefficient = 0.6722; p < 0.01), suggesting that accountants with higher levels of engagement and organizational attachment are better able to regulate their time and energy to sustain WLB. OJ shows a small positive but statistically insignificant effect, implying that perceptions of fairness may also contribute to better WLB, although the evidence in this sample is not strong enough to confirm this effect.

Model (3) shows that WO has a positive and highly significant effect on BU (coefficient = 0.3654; p < 0.01). This finding suggests that burnout is directly caused by work overload. Although the evidence is not strong enough to support this effect, OJ shows a slight negative but statistically insignificant effect that suggests beliefs of justice may assist reduce fatigue to some extent. Notably, and contrary to expectations, JA has a positive and statistically significant effect on BU, indicating that employees with higher job attachment are more prone to experiencing burnout. A possible explanation is that highly committed accountants may invest substantial time and energy into their work, potentially resulting in imbalance and emotional exhaustion. Excessive dedication beyond personal limits can lead to burnout.

Mediation Analysis Results

Table 6 summarizes the direct effects (DE), indirect effects (IE), and total effects (TE) of WO, WBL, and BU on TI.

Table 6. Summary of Direct, Indirect, and Total Effects on Turnover Intention