For income or for recognition: a comparative study of happiness among employees in state-owned and private enterprises

Abstract

In China, state-owned enterprises (SOE) and private enterprises represent the two main sectors absorbing labor. While previous studies have extensively examined differences between these two types of enterprises in terms of market performance and human resource management, little attention has been paid to differences in employee well-being. Using 3,178 observations from the 2016 wave of the China Labor-force Dynamics Survey (CLDS), including 555 SOE employees and 2,623 private enterprise employees, this study examines the relationship between enterprise ownership and employee well-being. The results indicate that SOE employees report significantly higher levels of well-being than their counterparts in private enterprises. Heterogeneity analyses further show that this advantage is more pronounced among older workers, as well as in the eastern and central regions of China, and within the secondary and tertiary industries. Mediation analysis suggests that the well-being gap is mainly explained by lower work burnout and stronger perceived work meaningfulness among SOE employees. These findings highlight a significant well-being disparity between employees in SOE and private enterprises and suggest that improving rest rights, job stability, and perceived work meaningfulness may help enhance well-being among employees.

Introduction

Since the reform and opening-up, China has achieved tremendous success in market-oriented reforms and economic development, with both state-owned enterprises (SOE) and private enterprises playing significant roles. The primary difference between them lies in the ownership and control of enterprise assets and capital. SOE are owned or controlled by the state and are considered to have attributes of the public sector. The government widely supports China’s SOE, and their heads have bianzhi and administrative rank. Therefore, it can be interpreted as a quasi-public sector. Private enterprises refer to those where the means of production are privately owned, traditionally regarded as embodying capitalist ownership¹. In other words, to some extent, the difference between SOE and private enterprises in China involves a gap between the public and private sectors.

Currently, scholars’ comparative studies of Chinese SOE and private enterprises have primarily focused on the organizational level, emphasizing the differences and reasons behind the performance of these two types of business entities in market activities^2–6. However, with the improvement in economic levels and societal development, an increasing number of studies are shifting focus to individuals within these enterprises, particularly in regards to their mental needs, such as happiness^7,8. Nevertheless, systematic comparative research on the differences in well-being between employees in state-owned and private enterprises, as well as the underlying causes of these differences, remains relatively scarce. This theoretical gap not only limits our understanding of the micro-level psychological characteristics of the labor force within China’s distinctive ownership structure, but also makes it difficult to explain the increasingly pronounced divergence in employment preferences observed in the labor market. In the context of China’s ongoing structural transformation of the labor market and employees’ rising expectations regarding job quality and psychological well-being, it is both urgent and necessary to uncover the differences in well-being between employees in enterprises with different ownership types and to explore their deeper causes. Therefore, this study aims to fill this research gap by comprehensively analyzing various aspects of work across different types of enterprises in order to identify the key factors influencing employee well-being.

the distinct characteristics of SOE and private enterprises contribute to the complexity observed in the happiness of their employees. In traditional Chinese society, it was commonly believed that working for a SOE meant having an “iron rice bowl”, which means that not only will you not be fired, but you will also have higher income and fringe benefits. This results in a higher level of happiness among employees. However, SOE employees’ job security has been affected by the wave of SOE reforms since the 1990s, leading to layoffs. Meanwhile, people’s favorability towards private enterprises have changed. On the one hand, with the development of the market economy, the income and benefits of private enterprise employees have also increased, narrowing the income gap between them and SOE employees⁹. For example, emerging internet companies in China have been labeled with high income and high welfare benefits. Higher work income is often associated with greater happiness¹⁰. However, on the other hand, some private enterprises in China have adopted a work pattern known as “996”¹¹. In such high-intensity work, the “work 996, get sick ICU” verse has become common, highlighting the physical and mental health risks and reduced happiness levels associated with excessive overtime in private enterprises.

Pursuing happiness is an important goal in most societies¹². When choosing a job, the expectation of happiness has become an increasingly significant factor. In the perception of Chinese people, compared to private enterprises, SOE seem to be a happier choice. But does this mean that employees in SOE actually have a higher level of happiness compared to those in private enterprises? Existing research has clearly indicated significant differences between SOE and private enterprises in human resource management and many other working conditions^13,14.

Therefore, exploring whether these differences lead to variations in employee happiness is both necessary and valuable. Additionally, this study recognizes the diversity of factors affecting employee happiness differences. By introducing the Two-Factor Theory, it attempts to explain the reasons for happiness differences between employees in SOE and private enterprises from the perspectives of motivation factors and hygiene factors. Specifically, this study uses data from the 2016 wave of the China Labor-force Dynamics Survey (CLDS) and employs a range of empirical methods, including Ordinary Least Square (OLS), Propensity Score Matching (PSM), Double Machine Learning (DML), mediation analysis, and Ordered Logit models, to examine the differences in well-being between employees in state-owned and private enterprises and to explore the underlying mechanisms.

The marginal contributions of this study can be summarized in three aspects. First, in terms of research perspective and theoretical extension, this study introduces Two-Factor Theory into the explanation of well-being differences associated with enterprise ownership, thereby enriching the interdisciplinary literature at the intersection of organizational behavior and labor economics. Second, in terms of methodology and empirical depth, this study not only employs traditional OLS and PSM models but also introduces the more advanced DML approach, providing more robust and precise empirical evidence for understanding the relationship between enterprise type and employee well-being, and thereby enhancing the scientific rigor and reliability of the findings. Third, in terms of practical implications and policy relevance, this study reveals heterogeneity in the well-being gap between SOE and private enterprise employees across age groups, regions, and industries. These fine-grained empirical findings not only validate the applicability of existing theories in the Chinese context but also offer targeted empirical evidence to support government efforts in optimizing employment guidance policies and helping private enterprises improve human resource management practices more effectively.

Literature review

Researchers have yet to reach a concrete conclusion regarding the comparison of employee happiness in SOE versus private enterprises. Different arguments have been presented based on samples from various countries and industries.

One view suggests that private enterprise employees tend to have a higher sense of happiness than those in SOE. For instance, an empirical study based on Polish society found that enterprise ownership type predicted stress related to a lack of organizational support and work overload, which were significantly higher for managers in SOE¹⁵. Similarly, a comparison of the Australian nursing profession found that nurses in the private sector had higher levels of happiness, satisfaction with their relationships, and teamwork¹⁶. Comparable results were observed when comparing South African public sector nurses with private sector nurses¹⁷. Researchers in Finland found that private sector physicians had higher life satisfaction and fewer psychological problems than their public sector counterparts¹⁸. On the other hand, some studies have come to the opposite conclusion that employees in SOE are happier than those in private enterprises. For example, in Pakistan, female employees in the private sector were more likely to be sexually harassed and had higher levels of depression and anxiety than those in the public sector¹⁹. Likewise, employees of state-owned banks in India were found to be happier than those of private banks²⁰. However, it is also argued that there is no significant difference in happiness between employees of SOE and private enterprises. A comparison of nurses in the UK public sector with those in the private sector found no significant differences in their experiences of stress²¹. In Malaysia, physicians in the public and private sectors had similar overall stress levels, although they differed in terms of stressors²². Other researchers, using propensity value matching, found no significant difference in happiness between Korean public sector employees and private sector employees²³.

These varying findings demonstrate that researchers have not yet reached a consensus on the happiness of employees in SOE compared with those in private enterprises. As for the mediation mechanisms involved, researchers have mainly discussed three variables: job burnout, job security, and income.

Job burnout is a well-established construct that encompasses two main dimensions: emotional exhaustion and depersonalization²⁴. It can leave individuals feeling physically, emotionally, and psychologically depleted, as well as disconnected from their colleagues and clients²⁵. Empirical studies, including those based on Chinese society, have identified a negative association between job burnout and happiness among employees²⁶. For example, A study of primary school teachers in Portugal’s public and private sectors found that public sector teachers experience higher levels of job burnout²⁷, while studies based on Indonesia suggested that nurses in the private sector were more likely to suffer from job burnout²⁸. Overall, researchers generally agree that job burnout is negatively correlated with happiness, but more research is needed to better understand the differences in job burnout levels between employees in SOE and private enterprises.

Job security is crucial to employees’ happiness as demonstrated in the literature on unemployment²⁹. In China, employees of SOE generally have higher job security than those of private enterprises, as SOE positions are commonly referred to as “iron rice bowls”³⁰. This provides a significant positive impact on employees’ happiness³¹ by eliminating negative emotions such as anxiety and worry caused by job insecurity. However, this security comes at a cost to upward mobility, as seniority ranking is prevalent in SOE, limiting opportunities for promotion among younger employees. This trade-off between job security and upward mobility reduces the happiness effect of job security, and may lead employees to seek greater possibilities in private enterprises. In fact, during the 1990s reform wave, many public sector employees willingly gave up their “iron rice bowls” to pursue entrepreneurship or invest in private enterprises³². Therefore, while the security of SOE jobs enhances employee happiness, it may also reduce employee happiness due to its associated costs.

Job income is another important factor affecting happiness. Many empirical studies have supported a positive association between income and happiness^33–35, indicating that higher incomes lead to higher levels of happiness. However, the effect of relative income on happiness cannot be overlooked. The well-known “Easterlin paradox” suggests that although higher personal income is associated with higher happiness, when a country’s average GDP rises, the average happiness of its citizens does not increase with it³⁶. This phenomenon can be explained by the fact that it is relative income rather than absolute income that affects happiness. In China, except for state monopolies such as electricity and tobacco, most industries coexist with both SOE and private enterprises. In these industries, employees of SOE and private enterprises compare each other’s income, which significantly affects their happiness. In the early stages of China’s market economy, private enterprises were treated less favorably than SOE. However, with the development of the market economy, the income gap between private enterprises and SOE has gradually narrowed³⁷. This makes it difficult to determine the happiness comparison between SOE employees and private enterprise employees, as it may reduce the happiness of employees in SOE and enhance the happiness of employees in private enterprises.

Theoretical analysis and hypotheses

Among the various factors may influence the happiness of employees in SOE and private enterprises, job security and job income play a positive role, while job burnout may have a negative impact. These factors, as intrinsic or extrinsic aspects of work, are related to the core of Herzberg’s Two-Factor Theory—motivation factors and hygiene factors. As intrinsic elements of work, motivation factors include promotion, the work itself, potential for growth, responsibility, recognition, and achievement; whereas hygiene factors, as extrinsic elements, encompass interpersonal relations, wages, company policies and administration, relationships with supervisors, and working conditions^38,39.

In previous studies, the Two-Factor Theory has often been used to explain job satisfaction. For workers, work is an important part of life, and job satisfaction is closely linked to their happiness. Existing research also extensively focuses on the enhancing⁴⁰ and protective⁴¹ effects of job satisfaction on happiness. Therefore, this study attempts to expand the application boundaries of the Two-Factor Theory and use it to analyze the overall well-being of employees, that is, examining the differences in happiness among employees of SOE and private enterprises and their underlying mechanisms from the perspectives of motivation factors and hygiene factors. Specifically, the satisfaction of motivation factors could potentially stimulate positive psychological experiences and thereby increase happiness, the hygiene factors can reduce dissatisfaction with work, and their absence can lead to unhappiness. Furthermore, it is important to note that workers vary in age, region, and industry, which not only gives them different work expectations but also changes the path of how the macroeconomic environment affects their well-being through motivation and hygiene. Therefore, the study also examines the potential differences in happiness among workers of different ages, regions, and industries in SOE or private enterprises.

Differences in happiness between employees of SOE and private enterprises

In China, it is possible that SOE employees have higher levels of happiness than their counterparts in private enterprises. This is because SOE may excel in certain motivation factors and hygiene factors, thereby enhancing their employees’ happiness. In terms of motivation factors, SOE may have certain advantages. Compared to private enterprises that primarily pursue economic benefits, Chinese SOE tend to have a stronger public attribute. Working in SOE aligns with the Chinese societal emphasis on devotion to public service⁴², allowing SOE employees to derive greater sense of achievement and happiness from their work. Regarding hygiene factors, the advantages of SOE could help reduce job dissatisfaction and promote happiness. For instance, benefits like lifelong employment, affordable housing, free medical care, and pensions are less common in private enterprises⁴³. Additionally, private enterprises tend to emphasize efficiency and continuously raise job demands⁴⁴. This not only deprives employees of favorable working conditions and engenders unhappiness, but also, greater stress itself is associated with lower levels of well-being (Valletta, 1999)⁴⁵. Hence, the following hypothesis is proposed:

Hypothesis 1

Employees in SOE are likely to have higher levels of happiness compared to those in private enterprises.

Differences in individual expectations and happiness

Employees of different ages, regions and industries may have varying expectations and needs for their jobs. Work tends to be central to a worker’s life, and job risks could potentially lead to anxiety and loss of hope⁴⁶. At the same time, the external market environment in which they operate also influence the relative differences between state-owned and private enterprises in their ability to provide hygiene factors and motivation factors.

Firstly, compared to younger workers, older workers are disadvantaged in the labor market due to age-related limitations, resulting in fewer job opportunities⁴⁷. They also tend to exhibit a stronger risk-averse orientation. As a result, stable employment relationships and well-developed pension and medical security (i.e., hygiene factors) become core criteria in their evaluation of job desirability⁴⁸ In contrast, private enterprises are often constrained by intense market competition and limited resources, which tends to shorten their organizational life cycle. According to the Report on Financial Services for Small and Micro Enterprises released by the People’s Bank of China, the average lifespan of small and medium-sized enterprises is only about three years. Moreover, during economic downturns, private firms are more likely to transfer risks through layoffs or wage reductions⁴⁹. Therefore, in the central region, the gap between SOE and private enterprises in providing hygiene and motivation factors appears to be the largest, which may be accompanied by the most pronounced difference in happiness between the two groups.

Second, from the perspective of regional economic characteristics, there are significant differences in the levels of marketization and industrial structures across eastern, central, and western China, which profoundly influence the well-being gap between employees in state-owned and private enterprises. In the eastern region, although the private sector is highly developed and offers higher potential earnings, it is also characterized by intense market competition and heavier work intensity. In this context, the advantages of SOE in providing hygiene factors, such as guaranteeing statutory rest rights and promoting work–life balance, become particularly prominent. In the western region, economic development remains highly dependent on state investment and policy support due to geographical and historical conditions. As key pillars of the local economy, SOE not only offer compensation and welfare packages far superior to those of local private firms but also confer high social status on their employees. The central region, meanwhile, has faced the dual pressures of industrial restructuring and the outflow of talent to the eastern region in recent years. Compared with the strong market dynamism of the east and the policy advantages of the west, the private sector in the central region is relatively weaker and lacks high-quality employment opportunities. Consequently, SOE in the central region represent not only stability but also scarce and highly desirable employment opportunities. As a result, the gap between SOE and private enterprises in providing both hygiene and motivation factors is the largest in the central region, leading to the most pronounced difference in employee well-being.

Third, from the perspective of industrial characteristics, private firms in the primary and secondary sectors are often concentrated in traditional labor-intensive industries, which tend to involve harsher physical working environments and stricter management systems. In contrast, SOE in these sectors typically control key resources and are often capital-intensive or technology-intensive, providing better working conditions and more standardized labor protection. By comparison, private firms in the tertiary sector (including services and high-tech industries) are more diverse and can offer younger workers a wider range of career development opportunities and higher income prospects. As a result, the gap in working conditions between SOE and private enterprises is relatively smaller in the tertiary sector. Therefore, we propose the following hypotheses:

Hypothesis

2a: The effect of happiness in SOE is higher for older employees compared to younger employees.

Hypothesis

2b: In both the eastern and western regions, employees in SOE.

Hypothesis

2c: The happiness gap between employees of SOE and private enterprises is greater in the primary and secondary industries than the tertiary industry.

Mechanisms of happiness differences between employees of SOE and private enterprises

Based on Two-Factor Theory, this study attempts to explores the mechanisms underlying the well-being differences between employees in state-owned and private enterprises through two pathways: hygiene factors and motivation factors. Although Herzberg’s Two-Factor Theory encompasses multiple dimensions, this study uses three indicators as proxies, work burnout, income, and perceived work meaningfulness, to operationalize hygiene and motivation factors. Admittedly, this study relies on specific measurement indicators that have clear limitations, as they cannot exhaust all extensions of the Two-Factor Theory. Variables such as interpersonal relationships or specific corporate administrative policies, for instance, were not incorporated into the model. However, in empirical research, the adequacy of measurement indicators does not equate to exhausting all theoretical dimensions; rather, it depends on whether the selected indicators can adequately capture the core variation within a particular research context. The central focus of this study is the divergence in well-being associated with institutional differences between state-owned and private enterprises. The selection of these three indicators is not random but is deeply rooted in the specific context of the Chinese labor market, which may reflect the most fundamental structural differences between the two types of enterprise systems.

Contemporary social debates in China regarding career choices between state-owned and private enterprises often center on three aspects: First, the prevalent “996 overtime culture” in private enterprises corresponds to the absence of hygiene factors regarding working conditions. Second, the structural differences in explicit compensation and implicit benefits between the two types of enterprises correspond to the hygiene factor of income security. Third, the attribute of SOE as serving the public interest corresponds to the satisfaction of motivating factors such as achievement and recognition. These three aspects constitute key points of divergence between the two types of enterprises at the level of employees’ psychological perception. Therefore, the introduction of the three variables, job burnout, income, and work meaningfulness, sufficiently covers the core heterogeneity between state-owned and private enterprises with respect to working environment, material security, and intrinsic motivation. Within the specific context of this study, these variables provide adequate theoretical explanatory power and enable a meaningful assessment of the theoretical mechanisms proposed in this paper.

First, working conditions constitute a core hygiene factor, and work burnout serves as a direct indicator of unfavorable working conditions and the associated physical and psychological strain. In terms of working conditions, SOE perform better than private enterprises in reducing job burnout. In general, social cognition, the ideal job is often characterized as “more money, less work, and closer to home”. Among these factors, “less work” means that the workload should not be too heavy, as it can lead to physical and mental exhaustion. Compared to private enterprises, which are highly competitive and demand more energy from employees, SOE have lower pressure on competition and performance⁵⁰ and require less overtime work and can better protect employees’ right to rest⁴⁴, resulting in less burnout among their employees. Many studies have documented a significant negative correlation between job burnout and employee happiness⁵¹. Therefore, lower levels of work burnout, as a reflection of more adequate hygiene factors, reduce the distress and dissatisfaction experienced by employees in SOE.

Second, income, as another key hygiene factor, exhibits structural differences between the two types of enterprises, which may be closely related to the degree to which employees’ basic material needs are met. China’s labor security system is not sound, and many private enterprises have problems with long working hours and low salaries⁵⁰. High income only exists in some private enterprises and is not universal. In contrast, SOE provide not only normal salaries but also many fringe benefits, such as medical care, children’s education, and housing⁵², indirectly increasing the income of their employees. This helps reduce dissatisfaction with work and is positively associated with their higher levels of happiness.

Finally, beyond hygiene factors, the realization of intrinsic goals (motivation factors) is also be an important determinant of well-being⁵³. According to Two-Factor Theory, a sense of achievement, social recognition, and the intrinsic meaning of work are often considered core sources of positive work attitudes and deep psychological satisfaction. In this study, perceived work meaningfulness is used as a proxy variable representing these core motivation factors. This indicator may also play a mediating role in the differences in happiness between employees of SOE and private enterprises. Specifically, the public attribute of SOE, along with the strong collectivist values of SOE employees⁵⁴, not only provide them with a stronger sense of achievement and value; it also helps them to gain more social recognition, which in turn is positively associated with happiness⁵⁵. Therefore, the following hypotheses are proposed:

Hypothesis

3a: Employees in SOE report higher levels of happiness than those in private enterprises, and this difference is associated with lower levels of job burnout.

Hypothesis

3b: Employees in SOE report higher levels of happiness than those in private enterprises, and this difference is associated with higher income levels.

Hypothesis

3c: Employees in SOE.

Materials and methods

Data

The data used in this study were collected from the 2016 CLDS, organized and conducted by Sun Yat-sen University. The CLDS is an extensive interdisciplinary survey that covers various aspects of demographic, economic, social, psychological, and health-related factors. It targets the working-age population and aims to capture the current state and changes in education, employment, occupational protection, health, and happiness of the labor force, providing high-quality basic data for empirical research and policy formulation. The survey employs a multi-stage, multi-level, probability sampling method. It is proportional to the labor force size, fitting China’s rapidly evolving environment, while accommodating the characteristics of a cross-sectional survey. The CLDS2016 sample covers 29 provinces in China (excluding Hong Kong, Macao, Taiwan, Tibet, and Hainan), making it nationally representative. For the purpose of this study, the research sample was limited to respondents who work in either a SOE or a private enterprise. After excluding samples with missing or invalid responses to key variables, the effective sample size for analysis totaled 3178, comprising 555 SOE employees and 2623 private enterprise employees. The empirical analysis of this study was based on Stata and Python.

It is worth noting that, as of 2026, the most recent publicly available data from the CLDS is from the 2018 wave. However, the 2018 survey exhibits inconsistencies or missing values in several key variables—such as job satisfaction and job burnout, when compared to the 2016 version. To ensure consistency in variable construction and accuracy in empirical analysis, this study primarily utilizes the 2016 CLDS data. The 2018 data are incorporated in robustness checks to enhance the reliability of the findings through cross-validation. Although the data may appear somewhat outdated, this study focuses on the underlying mechanism between ownership structure and employee well-being, a relationship whose theoretical logic remains relatively stable over time. Therefore, the 2016 data still offer considerable explanatory power and academic value.

Measures

Dependent variable

The dependent variable in this study is employee well-being, measured using a single survey item from the CLDS 2016 questionnaire: “Overall, do you consider your life to be happy?” Respondents rated their answers on a five-point scale ranging from 1 (very unhappy) to 5 (very happy). This item has been widely used in both domestic and international research on well-being and demonstrates good face validity⁵⁶. Although this indicator is based on a single-item subjective evaluation scale, prior studies have shown that single-item measures of well-being are highly correlated with multi-item scales (such as the Satisfaction with Life Scale, SWLS) and are widely adopted in large-scale social surveys due to their simplicity and efficiency⁵⁷. As well-being is inherently a subjective evaluative construct, this measurement approach has strong comparability within the broader literature on happiness.

Independent variable

The independent variable in this study was the type of enterprise, which was obtained from the question “Which of the following is your workplace?“. This study was limited to employees of SOE and private enterprises. A value of 1 was assigned to individuals who responded that their workplace was a SOE and a value of 0 was assigned to those who responded that their workplace was a private enterprise.

Control variables

Following the approach of Yamashita et al. (2016)⁵⁸, and considering both the relevance of variables and data availability, control variables from four dimensions were selected. The first category comprises demographic control variables, including gender, age, years of education, and marital status. These variables are fundamental demographic factors that influence employee well-being. The second category consists of social relationship characteristics, including neighborhood relations and life choice latitude. These two variables reflect employees’ social support networks and subjective autonomy, which are closely associated with sources of well-being outside the workplace. The third category includes health-related characteristics, namely self-rated health status and exercise frequency. Health is a significant predictor of well-being, and exercise frequency serves as a proxy indicator for health behaviors. The final category comprises regional characteristics, including household registration type and a geographic variable. A dummy variable for province is employed as a control variable, with samples from enterprises located in the eastern coastal provinces assigned a value of 1 and those from other regions assigned a value of 0, so as to control for the macro-level influence of regional economic development on employee well-being. The measurement methods for the control variables are presented in Table 1.

Table 1.

Variable Definitions and Descriptive Statistics (N = 3,178).

Variable	Variable description	Variable definition	Mean	Std.Dev.	State-owned enterprise (SOE)	Private enterprise (PE)
Dependent Variable
Happiness	Respondents’ level of happiness	Assign a value from 1–5, the higher the score, the higher the happiness	3.819	0.898	3.98	3.78
Independent Variable
Worktype	Employment type	State-owned enterprise = 1, Private enterprise = 0	0.175	0.380	-	-
Control Variables
Gender	Respondent’s gender	Male = 1, female = 0	0.55	0.498	0.68	0.52
Age	Respondent’s age	Actual age of respondents	39.257	11.513	42.08	38.66
Education	Educational attainment	Respondents’ years of education	10.722	3.581	12.08	10.43
Marital status	Marital status	Married = 1, Unmarried = 0	0.795	0.404	0.84	0.78
Social support	Frequency of mutual help among neighbors	Assign a value from 1–5, the higher the score, the more the social support	2.619	1.266	2.76	2.59
Life choice	Perceived freedom of life choice	Assign a value from 1–10, the higher the score, the more the life choice	6.544	2.112	6.46	6.56
Health	Self-rated health status	Very healthy = 1, healthy = 2, average = 3, rather unhealthy = 4, very unhealthy = 5	3.855	0.832	3.85	3.85
Exercise	Whether the respondent exercises regularly	Yes = 1, no = 0	0.38	0.485	0.50	0.35
Registered residence	Household registration type	Urban = 1, Rural = 0	0.598	0.490	0.25	0.67
Province	Region indicator	Eastern province = 1, other province = 0	0.62	0.485	0.46	0.65
Mediating Variables
Job burnout	Job burnout	Assign a value from 1–5, the higher the score, the higher the job burnout	2.832	1.344	2.65	2.87
Sense of value	Sense of work meaningfulness	Assign a value from 1–5, the higher the score, the higher the sense of value	2.845	1.354	2.97	2.82
Income	Income group	Income group from 1–7	2.984	1.136	3.27	2.92

Mediating variables

According to Herzberg’s Two-Factor Theory, factors influencing employees’ work attitudes can be categorized into two types: hygiene factors and motivating factors (Herzberg et al., 1959). Hygiene factors primarily pertain to external environmental conditions such as compensation, working conditions, and interpersonal relationships. Their absence leads to employee dissatisfaction, yet their improvement does not necessarily generate satisfaction. Motivating factors, on the other hand, involve intrinsic elements such as a sense of achievement, recognition, and the meaningfulness of the work itself. Their presence genuinely stimulates employees’ motivation and well-being.

Based on this theoretical framework and with reference to existing studies, three key indicators were selected to capture the mechanisms of the two factors. First, income is one of the core hygiene factors, directly reflecting the economic rewards employees obtain from their work and serving as an important measure of extrinsic motivation. The CLDS2016 survey inquired about respondents’ total annual personal income. In this study, income is classified into seven hierarchical levels, assigned values from 1 to 7 in ascending order, with higher scores indicating higher income levels of the respondents. Second, job burnout, as an extension of hygiene factors, reflects the extent to which objective working conditions, such as work intensity, working hours, and work-related stress, consume employees’ physical and mental resources. It serves as an important proxy variable for the quality of the working environment. Referring to the study by Xu et al. (2017)⁵⁹, the survey item “My work makes me feel physically and mentally exhausted” is defined as job burnout. Responses are assigned values from 1 to 5 in ascending order, with higher scores indicating a greater degree of work-related job burnout. Third, work meaningfulness is a core manifestation of motivating factors, reflecting employees’ perception of the meaningfulness of their work, their recognition of social contribution, and their intrinsic sense of achievement. It constitutes one of the most critical dimensions of motivating factors. Referring to the study by Yang et al. (2021)⁶⁰, the survey item “I feel that I have accomplished a lot of valuable work” is defined as work meaningfulness. Responses are assigned values from 1 to 5 in ascending order, with higher scores indicating a greater sense of work meaningfulness.

The three indicators outlined above correspond to the core mechanisms of the Two-Factor Theory from two dimensions: external conditions (income and job burnout) and intrinsic meaning (work meaningfulness). It should be noted that all three indicators are measured using single-item scales, which may entail certain measurement limitations. Nevertheless, existing research suggests that well-designed single-item measures can achieve acceptable validity and reliability for conceptually focused constructs (Matthews et al., 2022)⁶¹. Therefore, within the context of this study, these three indicators are employed as an initial test of the mechanisms underlying the Two-Factor Theory.

Common method bias test

It is important to note that since both the dependent variable and the mediating variables in this study are derived from respondents’ self-reported data, there is a potential risk of Common Method Bias. To examine this issue, we conducted Harman’s single-factor test by including the key self-reported variables in an unrotated principal component factor analysis. Table 2 reports the results of the factor analysis. The results show that the eigenvalue of the first factor is 1.255, explaining 31.39% of the total variance, which is below the commonly used 40% threshold. This indicates that no single factor accounts for the majority of the variance in the sample data. Therefore, the influence of common method bias in this study is considered to be limited.

Table 2.

Results of Harman’s Single-Factor Test.

Factor	Eigenvalue	Variance explained(%)	Cumulative(%)
Factor1	1.255	31.39	31.39
Factor2	1.157	28.92	60.31
Factor3	0.895	22.37	82.67
Factor4	0.693	17.33	100.00

Model selection

Ordinary least square model

To test the relationship of enterprise ownership type on employees’ happiness, we first employed theOLS model for estimation. The model was set up as follows

1

Among them, indicates respondent i’s happiness. indicates whether respondent i is an employee of a SOE. indicates other control variables. represents the constant term. indicates the random error term. is the coefficient to be estimated in this study, which reflects the magnitude and direction of the association of SOE job on employees’ happiness.

Double machine learning model

Given the significant correlation between individual happiness and various covariates (such as personal health status, family economic condition, etc.), failure to effectively control these covariates might have impaired the accurate identification of differences in happiness levels between employees of SOE and private enterprises. Recognizing the limitations of traditional econometric models in handling high-dimensional data with limited sample sizes, this study adopted the DML approach proposed by Cherno Zhukov et al.⁶². A random seed of 20260214 was set. Five-fold cross-validation was employed for sample partitioning to reduce the risk of overfitting and improve estimation efficiency. For the fitting of conditional expectation functions, two machine learning algorithms, random forest and support vector machine, were respectively adopted, and the two algorithms were combined to yield a total of four estimation specifications: (1) both the outcome variable and the treatment variable using random forest; (2) the outcome variable using random forest and the treatment variable using support vector machine; (3) the outcome variable using support vector machine and the treatment variable using random forest; (4) both the outcome variable and the treatment variable using support vector machine. Standard errors were constructed using the robust option in the ddml command to address heteroskedasticity. This method employs machine learning models repeatedly in auxiliary equations to eliminate bias introduced by regularization terms. A notable advantage of this approach is that it not only overcomes the ‘curse of dimensionality’ problem faced by traditional linear regression due to an excess of control variables but also significantly enhances estimation accuracy due to its non-parametric machine learning models, which can handle non-linear relationships between variables. Leveraging these benefits, this study further employed the DML model to deeply assess the differences in happiness levels between employees of SOE and private enterprises. The specific models were set up as follows:

2

3

In this model, signifies the individual’s level of happiness. indicates whether the ith respondent is employed by a SOE. represents a series of control variables. and denote the conditional expectation functions of the outcome variable given the control variables and the treatment variable given the control variables, respectively. U and V are the error terms in the model. Finally, denotes the treatment association under scrutiny.

Propensity score matching model

The research question addressed in this paper was whether there is a difference in happiness levels between employees of SOE and those of private enterprises. However, individuals’ decisions to work for either a SOE or a private enterprise were not random and may have been influenced by selection bias. Therefore, to address potential endogeneity issues, we employed Rosenbaum’s PSM method proposed in 1983⁶³. This method, based on the counterfactual intervention framework model, was effective in mitigating the impact of endogeneity.

The PSM method compressed the sample information through a logit or probit model to generate a propensity score for acceptance of treatment. The treatment group in this study comprised employees of SOE, while the control group consisted of employees of private enterprises. These groups were matched based on their propensity scores, and the average treatment association was then calculated using three methods: nearest neighbor matching, kernel matching, and radius matching. By employing the PSM method, we were able to more accurately estimate the association of working for a SOE versus a private enterprise on happiness levels, while also accounting for the potential selection bias in the sample.

Mediation effect model

To test whether the difference in happiness between SOE employees and private enterprise employees was generated by the difference in job burnout, income and sense of work meaningfulness between the two, this study drew on the mediating effect test proposed by Baron and Kenny⁶⁴ to construct Eqs. (4) and (5) on the basis of Eq. (1):

4

5

Among them, represents the mediating variable. indicates whether the ith respondent is employed by a SOE. signifies the individual’s level of happiness. , capture the total and direct associations between SOE status and happiness. refers to the vector of coefficients for the association of control variables on the mediating variable. indicates the vector of coefficients for the association of other control variables on happiness. describes the coefficient of influence of the mediating variable on happiness. , articulate the constant terms. designates other control variables. Finally, represents the random error term.

The overall study flow of this study is shown in Fig. 1. Figure 1 shows the various steps of the study and their interrelationships, helping to visually present the structure of this study and provides a clear operational path for subsequent analysis.

Fig. 1.

Study Flow Chart.

Results

Descriptive statistics

Table 1 reports the descriptive statistics of the main variables. The sample size is 3,178. The mean level of happiness in the full sample is 3.819 (SD = 0.898), indicating a moderately high level of well-being among respondents. In the subsample comparison, employees in SOE report a higher level of happiness (mean = 3.98) than those in private enterprises (mean = 3.78). In terms of sample composition, males account for 55% of the respondents. The average age is about 39.3 years, the average years of education is 10.7, and 79.5% of the respondents are married. Regarding health and lifestyle, the average self-rated health is 3.855, while only 38% of respondents report exercising regularly. For perceptions of the social environment, the mean frequency of neighborhood mutual assistance is 2.619, and the mean level of perceived freedom of life choice is 6.544. In addition, 59.8% of the respondents hold urban household registration, 62% come from eastern provinces, and the mean value of the income group variable is 2.984.

Benchmark regression

Table 3 presents the results of the benchmark regression. Column (1) reported the regression of the dependent variable on the core independent variable, while Column (2) included demographic characteristics variables. Column (3) added social relationship characteristic variables, and Column (4) further controlled for health and region characteristic variables. The interpretation of the results was based on Column (4). The results indicate that the happiness of SOE employees was 0.164 (p < 0.001) units higher compared to private enterprise employees, supporting hypothesis 1.

Table 3.

Benchmark Regression Results.

	(1)	(2)	(3)	(4)
Worktype	0.200***	0.180***	0.174***	0.164***
	(0.041)	(0.044)	(0.042)	(0.042)
Gender		-0.035	-0.048	-0.062*
		(0.032)	(0.031)	(0.031)
Age		-0.004**	-0.005**	-0.002
		(0.002)	(0.002)	(0.002)
Education		0.025***	0.018***	0.016**
		(0.006)	(0.005)	(0.005)
Marital status		0.310***	0.329***	0.323***
		(0.046)	(0.043)	(0.043)
Registered residence		0.004	-0.021	-0.026
		(0.039)	(0.038)	(0.037)
Province		0.088**	0.062	0.064*
		(0.033)	(0.032)	(0.032)
Social support			0.080***	0.072***
			(0.011)	(0.011)
Lifechoice			0.108***	0.095***
			(0.008)	(0.008)
Health				0.177***
				(0.020)
Exercise				0.046
				(0.033)
_cons	3.784***	3.407***	2.601***	1.952***
	(0.018)	(0.117)	(0.125)	(0.146)
N	3178	3178	3178	3178
R 2	0.007	0.033	0.115	0.141
adj. R2	0.007	0.031	0.113	0.138

Note: *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Robustness tests

Double machine learning

To mitigate as much as possible the potential issues of functional form misspecification and high-dimensional control variables that traditional linear regression may face, this section further employs the double machine learning method as a robustness check. This method flexibly fits conditional expectation functions using machine learning algorithms, effectively captures nonlinear relationships among variables, and yields consistent estimates under high-dimensional covariate specifications.

Table 4 reports the DML estimation results for the four algorithm combinations. The estimated values for the four combinations are 0.126, 0.123, 0.124, and 0.160, respectively, all significant at the 1% level, and are highly close to the value of 0.154 obtained from the OLS baseline regression. Notably, regardless of whether random forest or support vector machine is used to fit the conditional expectation functions, and irrespective of how the algorithms for the outcome variable and the key independent variable are combined, the estimation results remain stable, indicating that the core finding of this study does not depend on the specific choice of machine learning algorithm. Compared with OLS, the DML method, under more stringent control conditions (allowing for nonlinear relationships between covariates and both the key independent variable and the outcome variable), still yields significantly positive estimates. These further rules out the possibility of bias due to functional form misspecification. In summary, the DML estimation results not only are consistent with the baseline regression but also reinforce the robustness of the core conclusion that the well-being of employees in SOE is significantly higher than that of employees in private enterprises. This finding is consistent with Hypothesis 1.

Table 4.

Double Machine Learning Estimation Results.

Variables	Estimation Method for Dependent Variable: Random Forest		Estimation Method for Dependent Variable: Support Vector Machine
	Estimation Method for Independent Variable: Random Forest	Estimation Method for Independent Variable: Support Vector Machine	Estimation Method for Independent Variable: Random Forest	Estimation Method for Independent Variable: Support Vector Machine
Worktype	0.126***	0.123***	0.124***	0.160***
	-0.044	-0.041	-0.04	-0.038
Controls	Yes	Yes	Yes	Yes
N	3178	3178	3178	3178

Notes: Column (1) presents the results where both the outcome variable and the treatment variable are fitted using random forest; Column (2) presents the results where the outcome variable is fitted using random forest and the treatment variable using support vector machine; Column (3) presents the results where the outcome variable is fitted using support vector machine and the treatment variable using random forest; Column (4) presents the results where both the outcome variable and the treatment variable are fitted using support vector machine. All estimations are based on five-fold cross-validation. Robust standard errors are reported in parentheses. *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Robustness checks using alternative models

In addition, this study adopts several strategies to further enhance the robustness of the results. First, winsorization is applied. To examine the influence of extreme values on the estimation results, all continuous variables are winsorized at the 1% level. After winsorization, the coefficient of worktype is 0.164, which is identical to the main regression result (0.164). This finding suggests that the data cleaning procedure has effectively addressed the issue of extreme values, and the core findings are not driven by a small number of outliers.

Second, to further test the robustness of the core conclusion and avoid the possibility that the research findings depend solely on the specific characteristics of the 2016 survey sample, we conducted a robustness check by replacing the sample using data from the 2018 wave of the same survey project. It should be noted that, although the 2018 questionnaire differs from the 2016 version in the measurement of certain control variables, making it inappropriate to directly pool the two waves of data for longitudinal comparison, the core variables of interest in this study, enterprise type, employee well-being, income, work meaningfulness, and job burnout, are measured in an entirely consistent manner across the two years and are therefore comparable. Accordingly, independently replicating the main analysis using the 2018 sample can effectively test whether our core finding remains stable across different survey years. Table 5 reports the regression results based on the 2018 sample. The results show that the well-being of employees in SOE is significantly higher than that of employees in private enterprises (0.106, p < 0.01), indicating that the core finding of this study still holds in the 2018 sample, thereby reducing the concern that the finding is driven solely by the specific characteristics of the 2016 sample and, to a certain extent, supporting the robustness of the research findings.

Table 5.

CLDS2018 Inspection.

	Happiness
Worktype	0.106**
	(0.049)
Gender	-0.057*
	(0.033)
Age	-0.008***
	(0.002)
Education	0.019
	(0.015)
Marriage	0.278***
	(0.047)
Social support	0.052***
	(0.018)
Life choice	0.117***
	(0.010)
Health	-0.196***
	(0.022)
Exercise	0.079**
	(0.034)
Registered residence	-0.009
	(0.038)
Province	0.076**
	(0.034)
_cons	3.273***
	(0.136)
N	2221
R2	0.170

Note: *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Third, the model specification is changed. To ensure the robustness of the benchmark regression results, this study employs an Ordered Logit model (Ordered Logistic Regression) to re-examine the relationship between enterprise type and employee well-being. Since the dependent variable, well-being, is an ordered discrete variable ranging from 1 to 5, the Ordered Logit model is better suited to capture its ordinal nature and thus serves as an effective complement to the OLS estimation. As shown in Table 6, the estimation results from the Ordered Logit model are highly consistent with those of the OLS benchmark regression, and the core findings remain unchanged. This further supports the robustness of the positive relationship between enterprise type and employee well-being identified in this study.

Table 6.

Robustness Test of Ordered Logit Model.

	(1)	(2)	(3)	(4)	Marginal Effect
Worktype	0.404***	0.370***	0.387***	0.366***
	(0.087)	(0.095)	(0.095)	(0.097)
Gender		-0.084	-0.127	-0.158*
		(0.068)	(0.069)	(0.069)
Age		-0.008*	-0.010**	-0.004
		(0.004)	(0.004)	(0.004)
Education		0.051***	0.039**	0.033**
		(0.012)	(0.012)	(0.012)
Marital status		0.617***	0.697***	0.702***
		(0.094)	(0.093)	(0.094)
Registered residence		0.010	-0.054	-0.058
		(0.081)	(0.082)	(0.083)
Province		0.153*	0.108	0.107
		(0.068)	(0.069)	(0.070)
Social support			0.170***	0.153***
			(0.026)	(0.026)
Lifechoice			0.252***	0.226***
			(0.020)	(0.020)
Health				0.419***
				(0.047)
Exercise				0.130
				(0.073)
1._predict					-0.004***
					(0.001)
2._predict					-0.011***
					(0.003)
3._predict					-0.068***
					(0.018)
4._predict					0.020***
					(0.006)
5._predict					0.063***
/
cut1	-4.183***	-3.455***	-1.759***	-0.221
	(0.151)	(0.282)	(0.312)	(0.361)
cut2	-2.812***	-2.078***	-0.352	1.193***
	(0.080)	(0.248)	(0.279)	(0.334)
cut3	-0.506***	0.257	2.096***	3.683***
	(0.039)	(0.240)	(0.277)	(0.335)
cut4	1.190***	1.982***	3.939***	5.570***
	(0.044)	(0.243)	(0.285)	(0.347)
N	3178	3178	3178	3178	3178

Note: *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

PSM test

The benchmark regression results indicate that employees in SOE report significantly higher levels of happiness. However, it cannot be ignored that systematic differences in individual characteristics may exist between employees in state-owned and private enterprises, which may lead to potential selection bias. To address concerns regarding endogeneity arising from observable factors, this study employs PSM as a robustness check.

First, Table 7 presents the balance test results for the covariates before and after matching. After matching, the standardized biases for all covariates are reduced to below 10%, and all corresponding t-tests become statistically non-significant (p > 0.1). In addition, the pseudo-R² decreases from 0.149 to 0.002, and the likelihood-ratio (LR) test changes from statistically significant to non-significant (p = 0.969). These results suggest that the matching procedure substantially improves the balance of observable characteristics between the SOE group and the private enterprise group.

Table 7.

Balance Test Results.

Variables	Unmatched	Mean			Reduct Bias%	T- value	P- value
	Matched	Treated l	Control
Gender	U	0.67928	0.5223	32.5		6.80	0.000
	M	0.67928	0.68649	-1.5	95.4	-0.26	0.797
Age	U	42.077	38.66	31.0		6.39	0.000
	M	42.077	41.063	0.1	99.6	0.02	0.982
Education	U	12.081	10.435	48.0		9.99	0.000
	M	12.081	11.995	2.5	94.7	0.45	0.654
Marital status	U	0.84324	0.7846	15.1		3.11	0.002
	M	0.84324	0.85225	-2.3	84.6	-0.42	0.676
Registered residence	U	0.24685	0.67213	-94.3		-19.65	0.000
	M	0.24685	0.24685	0.4	99.6	0.07	0.944
Province	U	0.45946	0.65383	-39.9		-8.67	0.000
	M	0.45946	0.46126	-0.4	99.1	-0.06	0.952
Socialsupport	U	2.7622	2.5883	13.5		2.94	0.003
	M	2.7622	2.6577	8.1	39.9	1.29	0.196
Lifechoice	U	6.4613	6.5616	-4.8		-1.02	0.309
	M	6.4613	6.5622	-4.8	-0.6	-0.78	0.434
Exercise	U	0.5009	0.35417	30.0		6.51	0.000
	M	0.5009	0.50811	-1.5	95.1	-0.24	0.810

Subsequently, to examine the robustness of the matching results, we employ three matching methods to estimate the average difference in outcomes for the treated group. Specifically, the ATT is estimated using 1:1 nearest-neighbor matching, kernel matching, and radius matching (caliper = 0.01). For all matching procedures, standard errors are computed using the bootstrap method with 1,000 replications, and a random seed of 10,101 is set to ensure replicability. The results are reported in Table 8. The estimated ATT values are 0.201, 0.148, and 0.143, respectively, all of which are statistically significant at the 1% level. These findings are consistent with the direction and statistical significance of the baseline OLS results.

Table 8.

Propensity Score Matching Estimates of the ATT.

Matching method	Treatment group mean	Control group mean	ATT	Bootstrap standard error	z-value	p-value
1:1 Nearest Neighbor	3.456	3.255	0.201	0.0715	2.81	0.005
Kernel (Epanechnikov)	3.456	3.308	0.148	0.0421	3.51	0.000
Radius(caliper = 0.01)	3.456	3.313	0.143	0.0434	3.29	0.001

Note: N = 3,178. The mean for the treatment group is based on observed outcomes. The control group mean represents the estimated counterfactual outcome derived from matched observations. The reported ATT reflects the estimated difference in outcomes between treated and matched control units after adjusting for observable characteristics. All standard errors are obtained using 1,000 bootstrap replications.

Overall, this evidence suggests that, after accounting for potential selection bias based on observable characteristics to the extent possible, the positive association between employment in SOE and employees’ happiness remains robust. This provides additional support for the validity of the baseline findings related to Hypothesis 1.

Heterogeneity analysis

This study aimed to explore whether the differences in happiness between employees of SOE and private enterprises were moderated by various factors. Therefore, the study examined the differences in happiness between employees of SOE and private enterprises by age, region and industry.

Firstly, employee age is closely associated with work attitudes, job performance, and workplace behaviors⁶⁵. In conducting the heterogeneity analysis by age, this study adopts the age of 35 as the threshold to distinguish between younger and middle-aged workers. This choice is based on two main considerations. First, it carries practical significance in the social context. In China’s labor market, the age of 35 is widely regarded as a critical turning point in one’s career trajectory. Particularly in the private sector, workers above 35 often face higher risks of layoffs and greater workplace pressure. Thus, the 35-year threshold not only reflects a socially recognized divide but also captures institutional and employment-related disparities across age groups. Second, from the perspective of career life cycle theory, the age of 35 typically marks a transition from the phase of rapid growth driven by passion to a stage of more stable development. During this period, individuals experience notable shifts in career expectations, risk tolerance, and perceptions of work meaningfulness. As a result, employees’ preferences for job-related goals such as “stability,” “growth,” and “achievement” may differ significantly before and after the age of 35, offering a theoretical basis for examining age-specific differences in happiness between employees in state-owned and private enterprises^66,67.

Second, China’s regional economic development imbalance constitutes an important macro-level context influencing employee well-being. The eastern region, characterized by a high degree of marketization, a developed private economy, and diverse employment opportunities, may see the compensation and career development prospects of private sector employees partially offsetting the institutional advantages of SOE. In the central region, which is in a period of accelerated industrialization, SOE still occupy a significant position, while the compensation, benefits, and social security levels of private enterprises are relatively lagging; thus, the role of SOE may be more prominent. In contrast, the situation in the western region is more complex. On the one hand, the western region has a relatively lagging level of economic development and limited private sector activity, where SOE often assume important functions in providing stable employment and comprehensive social security. On the other hand, SOE in the western region are mostly concentrated in traditional industries such as resource extraction and heavy industry, and due to limited local fiscal capacity, the compensation advantages of SOE may be less pronounced than those in the central region. Therefore, the direction of the association between SOE and employee well-being in the western region remains uncertain and requires empirical assessment. Based on this context, and following the official regional classification criteria of the National Bureau of Statistics of China, the sample is divided into three sub-groups, eastern, central, and western regions, to examine whether there is regional heterogeneity in the association between SOE and employee well-being.

Third, significant differences exist across industries in terms of capital intensity, market competition, and ownership structure, which may moderate the relationship between enterprise type and employee well-being. In the primary industry, the proportion of SOE is relatively low and the sample size is small, which may lead to unstable estimation results. In the secondary industry, SOEs are typically larger in scale, have more standardized working conditions, and provide more comprehensive labor protection, whereas private enterprises often face more intense market competition and greater cost-control pressures, resulting in higher work intensity and instability for employees. Therefore, the well-being advantage of SOE may be more pronounced in the secondary industry. In the tertiary industry, the private economy is active, and emerging service sectors provide a substantial number of high-paying and highly flexible employment opportunities, which partially offset the institutional advantages of SOE. Consequently, the well-being advantage of SOE is relatively attenuated in the tertiary industry. Based on this, following the industrial classification standards of the national economy, this study divides the sample into three sub-groups, primary, secondary, and tertiary industries, for heterogeneity analysis.

Table 9 reports the results of the heterogeneity analysis. The age heterogeneity analysis shows that the coefficient of worktype is not significant for the group of young employees aged 35 and below (β = 0.097), but is significantly positive for the group of middle-aged and older employees above 35 (β = 0.187, p < 0.001), indicating that the well-being advantage of SOE is mainly concentrated among middle-aged and older workers. The regional heterogeneity analysis reveals that the well-being advantage of SOE is most pronounced in the central region (β = 0.246, p < 0.01), followed by the eastern region (β = 0.140, p < 0.05), whereas the coefficient for the western region, though positive, is not statistically significant (β = 0.157). The industry heterogeneity analysis indicates that the coefficient of worktype is not significant in the primary industry (β = 0.607), possibly due to the relatively small sample size, but is significantly positive in both the secondary industry (β = 0.276, p < 0.001) and the tertiary industry (β = 0.129, p < 0.05), with the coefficient in the secondary industry being larger than that in the tertiary industry. Overall, the association of enterprise type on employee well-being exhibits clear heterogeneity across age, region, and industry. The well-being advantage of SOE is more pronounced among middle-aged and older employees, employees in the central region, and those working in the secondary industry.

Table 9.

Summary of Heterogeneity Analysis.

Variables	By age		By province			By industry
	Age < = 35	Age > 35	Eastern	Central	Western	Primary industry	Secondary industry	Tertiary industry
Worktype	0.097	0.187***	0.140*	0.246**	0.157	0.607	0.276***	0.129*
	(0.070)	(0.053)	(0.061)	(0.081)	(0.084)	(0.306)	(0.063)	(0.055)
Controls	YES	YES	YES	YES	YES	YES	YES	YES
_cons	2.055***	2.031***	2.179***	1.817***	1.609***	3.643***	3.756***	3.815***
	(0.257)	(0.226)	(0.182)	(0.336)	(0.346)	(0.189)	(0.024)	(0.026)
N	1328	1850	1970	665	543	36	1524	1618
R2	0.173	0.134	0.142	0.191	0.152	0.071	0.012	0.003

Note: *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Mechanism analysis

Based on the Two-Factor Theory, this study examines the potential mediating roles of job burnout, sense of work meaningfulness, and income in the relationship between worktype and employee well-being. Table 10 reports the corresponding regression results.

Table 10.

Results of Mediation Analysis.

Variables	Job Burnout	Work Meaningfulness	Income	Happiness	Happiness	Happiness
Worktype	-0.232***	0.151*	0.324***	0.149***	0.150***	0.155***
	(0.060)	(0.063)	(0.053)	(0.041)	(0.042)	(0.042)
Job burnout				-0.078***
				(0.012)
Work meaningfulness					0.038**
					(0.012)
Income						0.021
						(0.015)
Controls	YES	YES	YES	YES	YES	YES
_cons	4.384***	2.346***	1.860***	2.368***	1.904***	1.971***
	(0.142)	(0.152)	(0.116)	(0.155)	(0.146)	(0.145)
N	3178	3178	3178	3178	3178	3178
R2	0.053	0.011	0.044	0.152	0.143	0.140

Note: *, **, *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively.

Columns (1)– (3) present the effects of worktype on the three proposed mediators. The results show that worktype is significantly associated with lower levels of job burnout (Column 1), a higher sense of work meaningfulness (Column 2), and higher income (Column 3). Columns (4)– (6) report the associations between the mediators and employee well-being. In Column (4), job burnout is negatively and significantly associated with well-being. In Column (5), sense of work meaningfulness is positively and significantly associated with well-being. In contrast, in Column (6), the coefficient on income is not statistically significant, suggesting that income is not associated with well-being in this specification.

Taken together, these results are consistent with the presence of partial mediation through job burnout and sense of work meaningfulness. Specifically, worktype is associated with lower job burnout and a higher sense of work meaningfulness, both of which are also associated with higher levels of well-being. By contrast, although worktype is associated with higher income, income does not exhibit a significant association with well-being, suggesting that it is unlikely to is unlikely to represent a major channel in this relationship.

Overall, the findings suggest that worktype is associated with employee well-being may operate through non-monetary channels, including work environment and perceived meaning in work, rather than through material income. This pattern is broadly consistent with the two-factor theory, which emphasizes the roles of motivators and certain non-monetary conditions over purely financial rewards.

In summary, the empirical results are consistent with Hypotheses 3a and 3c, while Hypothesis 3b is not supported.

Discussion

Employees are an enterprise’s intangible assets and the most important competitive advantage⁶⁸. Paying attention to and enhancing employee happiness contributes to the healthy development of both the enterprise and society. This study explores the differences in happiness between employees of SOE and private enterprises from the perspectives of motivation factors and hygiene factors, and summarizes the related factors causing these differences.

Firstly, the study’ results indicate that employees of SOE in China are happier than those of private enterprises. Previous studies have mainly focused on differences in market performance between SOE and private enterprises, neglecting the relationship between enterprise ownership type and employee happiness. The study shows that Chinese individuals are happier when working in SOE than in private enterprises. Specifically, compared to employees of private enterprises, employees of SOE perform better in terms of both motivation factors and hygiene factors. This not only directly enhances their happiness but also suppresses their job dissatisfaction. The combined association of these two types of factors results in higher levels of happiness among employees of SOE⁶⁹. However, the results differ from those found in Europe and the United States^16,18. One possible explanation for this disparity is that Chinese SOE possess public sector attributes and receive extensive government support.

Secondly, the findings show that both job burnout and sense of work meaningfulness significantly influence the happiness differential and serve as mediating mechanisms underlying the gap in happiness between employees in state-owned and private enterprises. A brief qualifying note is necessary here: while the empirical results provide evidence supporting the above mediating associations, the deeper institutional explanations should be understood as contextual interpretations grounded in the national macro environment rather than mechanisms directly tested within the current survey data. Building on this inferential interpretation, SOE, compared with private firms, possess a public sector nature. On the one hand, as an explanatory comment, the public attributes that embodies the state’s political will makes it inferentially more inclined to strictly abide by labor law provisions in its institutional arrangements and to resist the excessively demanding overtime culture of “996”. It must be made clear that the discussion here concerning labor-law compliance and the “996” culture is entirely part of an extended contextual interpretation intended to provide a plausible real-world background for the empirical results; it is not an empirical finding directly measured by this study. Within this interpretive context, privatization is inferred to potentially intensify employees’ work stress under certain circumstances⁴⁴. On the other hand, the public attribute is also inferred to enhance employees’ sense of value and achievement derived from serving the collective. Overall, the absence of favorable working conditions in private enterprises, as manifested in higher levels of job burnout, along with the exemplary performance of SOE in terms of work meaningfulness, leads to differences in the happiness of their employees.

Notably, although SOE employees earn higher incomes, this is not the reason for their higher levels of happiness. One possible explanation for this is that employees of SOE and private enterprises have different expectations and levels of emphasis on income. The reform of the salary system in Chinese SOE, which stresses adjusting excessively high incomes and highlighting fairness, indicates that choosing to work in an SOE to some extent reflects an employee’s lower expectation for high income, with more focus on other factors associated with SOE. The study by Wang and Xie also indirectly verifies this point. They found that, those who gave up their iron rice bowl in the public sector and chose to jump into the sea of jobs in the private sector did not experience higher happiness with higher incomes³¹. For employees of private enterprises, as Goerzes et al. have found, they place greater emphasis on financial incentives and promotion opportunities⁷⁰. Therefore, a high income may be more effective in reducing job dissatisfaction and enhancing happiness for employees of private enterprises, while its positive relationship on employees of SOE is relatively limited.

Finally, the study finds that due to different expectations of happiness, the happiness gap between employees of SOE and private enterprises varies among groups of different ages, regions, and industries. In terms of age, working in SOE has a stronger positive association on the happiness of older employees compared to younger ones. This is because employees of different age groups have varying needs and interests in the workplace. Older employees tend to prioritize job stability, retirement benefits, and healthcare coverage. First, job security and stability are crucial for them⁷¹, as unemployment can pose a significant financial shock to the family. As the primary breadwinners, they need to ensure economic stability. In addition, due to relatively lower work efficiency, they are more likely to face the risk of dismissal in efficiency-driven private enterprises⁷², which subjects them to greater work pressure and uncertainty. In contrast, SOE have established regulations that provide a sense of job security⁷³, creating a safe and stable work environment. Second, SOE typically offer more comprehensive retirement and healthcare benefits, which not only alleviate older employees’ concerns about the future but also increase their satisfaction with their current jobs, thereby significantly enhancing their happiness. In other words, high stability and comprehensive welfare provisions precisely meet the needs of older employees, significantly boosting their overall well-being.

However, for younger employees who are at the early stage of their careers, greater emphasis is placed on career development opportunities, skill enhancement, and work flexibility. The positive relationship of working in SOE on their happiness is not as significant. This may be because the work environment and culture in SOE tend to be more rigid and conservative⁷⁴, which may not fully align with the younger generation’s aspirations for innovation and change⁷⁵. Moreover, SOE often have relatively fixed compensation and reward systems that lack strong incentives tied to individual performance and contributions. As a result, younger employees may feel that their efforts are not adequately recognized or rewarded. Therefore, although SOE provide job stability, this alone is insufficient to significantly enhance the happiness of younger employees, who place greater value on personal growth and career advancement.

In China, employees in SOE in the eastern and western regions report higher levels of happiness compared to their counterparts in private enterprises, with this disparity being even more pronounced in the central region. This can be attributed to the differing regional characteristics and economic advantages, which shape workers’ expectations across these areas. The eastern region, benefiting from geographic advantages and early reform and opening-up policies, has a higher degree of market openness that has attracted substantial foreign investment and private enterprises^76,77, fostering economic diversification and innovation. The relatively developed private economy in the east enables workers to select jobs that better meet their expectations, such as incentives related to promotion opportunities or benefits related to income. The western region has benefited from favorable national policies, including the “Western Development” strategy, the Belt and Road Initiative, and fiscal transfer payments. These policies have created a conducive environment for the development of the private sector, leading to a significant acceleration in its growth compared to previous periods, with promising momentum and prospects^78,79. Although the absolute level of private sector development in the west remains relatively limited, its rapid growth and positive outlook have attracted workers, who consequently hold more optimistic attitudes toward job stability and career prospects, thereby enhancing their happiness. The fulfillment of private-sector employees’ expectations helps narrow the happiness gap between SOE and private enterprise workers.

In contrast, the central region is often affected by the “siphoning effect” from the eastern region, resulting in a loss of human resources⁸⁰. Moreover, as an important grain production base in China, the central region shoulders the critical responsibility of ensuring national food security and the effective supply of agricultural products. This functional positioning leads to a larger share of labor being absorbed by agriculture and other traditional industries. At the same time, government policies and resource allocation tend to prioritize SOE that ensure stable grain production and supply. These factors collectively contribute to the relatively lagging development of the private sector in the central region compared to the east, and a slower growth rate than in the west. There is a widespread perception that jobs in SOE offer greater stability and security³⁰, which influences employees’ occupational choices in the central region. Under these circumstances, job stability and security become primary expectations for workers. Securing employment in relatively well-developed SOE implies stability, safety, and better working conditions, which in turn contribute to higher levels of employee happiness. Therefore, employees in the central region tend to prefer jobs in SOE, which partially explains why their happiness levels are significantly higher than those of private enterprise employees. These regional differences reflect the combined influence of economic policies, market openness, resource distribution, and socio-cultural perceptions.

In terms of industry, for workers in the primary and secondary industries, working in SOE brings greater happiness. On one hand, the labor market in these industries is becoming saturated. The National Bureau of Statistics of China shows that compared to 2012, in 2022, the employment share in the primary and secondary industries decreased by 10.6 and 1.3% points, respectively. With limited job absorption in these industries, SOE have a higher degree of involvement, meaning more job positions. On the other hand, people in these industries generally have limited education and lower income, making job security their biggest expectation⁷⁵. Therefore, working in SOE is their relatively best option, thereby enhancing their happiness. As for workers in the tertiary industry, firstly, the adjustment of China’s economic structure has made the tertiary industry the main driver of economic growth and employment absorption. Research also finds that private enterprises in the tertiary industry are more developmental and offer more job opportunities than any other industry^76,77. More job choices enable workers in the tertiary industry to choose jobs that better meet their expectations, making the traditional stability advantage of SOE no longer the dominant factor in their happiness.

Conclusion

Since the reform and opening up, China has achieved great success in economic development. However, as the economic level has risen, the happiness of the Chinese people has declined. In the 1990 World Values Survey, 28% of Chinese survey respondents reported being happy. By 2000, this percentage had dropped to 12% ⁷⁸. Given that the private economy has been instrumental in China’s economic growth, we may wonder if employees of private enterprises are any less happy than employees of SOE. To address this question, this study used data from the CLDS2016 survey and employed OLS and PSM models to examine the difference in happiness between employees of SOE and those of private enterprises. The empirical results show that employees of SOE have higher happiness than those of private enterprises. Specifically, the happiness of SOE employees is 0.163 units higher compared to that of private enterprise employees, and the results are significant at the 1% level. Further heterogeneity analysis reveals that the significant differences in well-being between employees in state-owned and private enterprises are primarily manifested across different age cohorts, geographic regions, and industry sectors.

Based on the above findings, we propose the following policy recommendations to improve the happiness of private enterprise employees. First, considering that employees in SOE are generally less tired than those in private enterprises, it is suggested that job burnout may play a role in the differing levels of happiness between these groups. Therefore, private enterprises should protect employees’ right to rest and avoid the damage of excessive overtime work on their happiness. Rest is one of the basic rights of employees, and they should be allowed to rest and take vacation to ensure that they can relax, regain their strength, and maintain good physical and mental health. The labor security department of the government should also pay more attention to employees’ right to rest, strengthen the supervision of excessive overtime work in enterprises, and fully guarantee employees’ right to rest. Second, given that the sense of work meaningfulness plays an important role in the difference between the happiness of employees in SOE and those in private enterprises, private enterprises should focus on enhancing the sense of work meaningfulness of their employees. In this regard, private enterprises should give humanistic care to their employees, establish an emotional bond between the enterprise and the employees, and enhance the employees’ sense of belonging to the enterprise. This can also help to improve their work efficiency. Third, given the heterogeneity of the happiness association of employees in SOE in terms of age due to job security, efforts should be made to strengthen the supervision of private enterprises and innovate re-employment mechanisms. On the one hand, employees of enterprises should actively promote the establishment of labor union organizations and make them actively participate in the labor employment process to ensure that enterprises sign fair labor contracts with their employees and avoid arbitrary dismissal. On the other hand, the government’s labor security department should improve the unemployment protection system, provide skill training and employment information to workers in unemployment status, and promote their smooth re-employment.

This study examines the relationship between enterprise type and employee happiness within a rigorous methodological framework; however, certain limitations remain. These limitations also point to directions for future research.

First, the study uses CLDS2016 data. While we prioritized the operational consistency of core variables when measurement validity and data recency could not be simultaneously ensured, the data do not fully capture the impacts of recent structural adjustments in China’s economy and labor market on employee happiness. Future research could use CLDS or other survey data to recalibrate and re-examine these findings over time. In particular, studies should focus on major institutional changes since 2016, such as the mixed-ownership reform of SOE and the development of new forms of employment, to assess whether these have altered the structural relationship between enterprise type and happiness. Second, regarding the reliability issue of single-item measurement. The dependent variable and mediating variables in this study were measured using single-item scales. Although we employed Harman’s single-factor test in the analysis, which indicated no serious common method bias, and applied various robustness checks to control for observable and unobservable confounding factors, the limitations inherent in single-item measurement and the potential risks associated with same-source data cannot be overlooked. Future research could employ multi-item scales to measure the core constructs, and incorporate multi-source data or longitudinal designs to more accurately estimate causal relationships between variables and further reduce the influence of common method bias. Such efforts would help to further validate and deepen the theoretical findings of this study. Third, based on Herzberg’s Two-Factor Theory, this study examined three mediating pathways: job burnout, work meaningfulness, and income, finding that the first two were significant while income was not. This finding carries certain theoretical implications. The analysis based on the 2016 survey data suggests that in the Chinese labor market during this period, the work environment-related dimension of hygiene factors (job burnout) may play a more prominent role than purely material rewards (income) in accounting for the relationship between ownership structure and happiness, while the motivating factor (work meaningfulness) also appears to be noteworthy. This pattern offers some insight into potential differences in the mechanisms underlying employee happiness between state-owned and private enterprises. That said, given recent changes in the labor market (such as the rise of the platform economy and the spread of flexible employment), it is possible that the specific effect sizes have evolved, and further research would be needed to explore this question.

In summary, this study provides theoretical evidence for understanding the relationship between enterprise ownership and employee happiness. The limitations outlined here define the boundaries of this research and indicate directions for future investigation. We hope that subsequent studies, using richer data, can further refine and deepen the exploration of this important topic.

Author contributions

Chao Xu and Mingyue Li were responsible for the overall conceptualization and design of the study. They contributed significantly to the drafting and revision of the manuscript. Rongling Long and Jiangyin Wang led the quantitative analysis, including data collection, statistical modeling, and interpretation of the results. All authors contributed to the final manuscript revision and approved the submitted version.

Funding

This research was funded by the “Research on China’s Pathways to Cooperating in East Asian Public Health Security under the Vision of a Community with a Shared Future for Mankind” project, supported by Shandong University of Political Science and Law (2020Q12B). The funding body had no role in the design of the study, data collection, analysis, interpretation, or manuscript writing.

Data availability

Data from the China Labor-force Dynamics Survey (CLDS) 2016/2018. The datasets used during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.