The Impact of Higher Education Expansion on Subjective Well-Being during the COVID-19 Pandemic: Evidence from Chinese Social Survey

Abstract

The rage of the COVID-19 pandemic, coupled with the downward trend seen in the economy, has further aggravated the downturn of the job market and diminished people’s sense of wellbeing in recent years. To mitigate the detrimental effects of the pandemic on college students’ employment, the Chinese government has further expanded the enrollment of postgraduate and undergraduate students. This study leverages data from the 2019 and 2021 waves of the Chinese Social Survey and constructs a difference-in-differences (DID) model to examine the effect of the higher education expansion (HEE) policy, initiated in 1999, on individuals’ subjective well-being during the COVID-19 pandemic. The results show that HEE policy could significantly improve individuals’ subjective well-being (SWB) during the pandemic, and that social class mobility emerges as a vital mechanism through which HEE policy impacts individuals’ SWB. Furthermore, there is a local-ladder effect due to reference dependence, with socio-metric status having a greater impact on SWB than socioeconomic status. This study reveals that the level of an individual’s happiness depends largely on whom they are compared with. This implies that the key focus of the HEE policy is to stimulate individuals’ potential and motivation for upward social mobility, ultimately enhancing their overall sense of well-being.

1. Introduction

The outbreak of COVID-19 had a detrimental effect on individuals’ physical and mental health, as well as their overall well-being. This was not only due to the restrictions on daily life caused by the pandemic, but also its negative effects on individuals’ psychology, social distance, job opportunities, and financial burdens [1]. From the enterprise perspective, the persistence of the pandemic, combined with economic pressures, has plunged numerous small and medium-sized enterprises to the brink of survival, further exacerbating the already sluggish job market in recent years. In 2020, 8.74 million college graduates faced the worst job prospects in history. Against this backdrop, the higher education expansion (HEE) policy, aiming to moderately increase the enrolment of postgraduate and doctoral students, has been regarded not only as a short-term measure to rapidly stabilize social sentiment and secure social stability, but also as a long-term plan to enhance national quality and promote people’s well-being. The premier Li Keqiang, in his 2020 government work report, clearly proposed the task of expanding enrolment in vocational colleges and universities by two million for two consecutive years. Naturally, one question is asked: does the HEE policy help to mitigate the negative effects of the COVID-19 epidemic and improve individuals’ subjective well-being (SWB)?

SWB is typically defined as an individual’s overall evaluation of their quality of life. It is often used interchangeably with terms such as life satisfaction, quality of life, and happiness [2]. Prospect theory posits that happiness is derived from a comparison between their expectations and perceived reality. That is, happiness comes from relative success in actualizing expectations, while unhappiness arises from relative failure [3].

HEE, as a major reform of China’s higher education system, has profoundly impacted both the labor market structure and even the entirety of Chinese society. The existing literature on the higher education expansion policy primarily focuses on its impact on the employment status of college graduates and the return rates seen in college education [4]. However, higher education is not simply an important means of improving employment and increasing income [5], but it also plays a critical role in mate selection, interpersonal interaction, and social stratification [6]. Therefore, a comprehensive evaluation of the gains and losses of the HEE policy should go beyond its impact on employment or the objective income of college graduates, and should consider its impact on individuals’ SWB.

As China transitions from a survival-oriented society to a development-oriented one, the criteria, expectations, and target references of people’s satisfaction have undergone profound changes. The sense of relative deprivation caused by educational injustice, wealth disparity, and the stratification of society has become a prevalent social psychology. Class stratification is not inherently bad, but “class solidification” is. Narrower pathways of upward mobility and “despair from the bottom” are more appalling social issues than poverty. Education, as a potential criterion for social stratification and an important driving mechanism for upward social mobility, can help individuals obtain well-paying jobs, better social resources, and improved social status, and thereby allow them to reach the ultimate goal of the pursuit of happiness. Therefore, higher education policy has traditionally been viewed by sociologists as an effective tool for promoting social mobility and increasing people’s well-being [7]. Based on the prospect theory and the idea of the “sorting machine” in education proposed by American educator Joe Spring, this paper hypothesizes social mobility as a key intermediate variable affecting national happiness during the pandemic; it thus utilizes a difference-in-differences method to identify the causal relationships between the shock of the epidemic, higher education expansion, and individual SWB.

The contributions of this paper are threefold. First, unlike the focus of the previous literature on the employment or income of college graduates, this paper evaluates the pros and cons of the HEE policy during the pandemic from the perspective of SWB. Second, most extant studies focus on the average happiness-enhancing effect of educational attainment. In this article, along with the average effect, we see how the extent of heterogeneity in the pro-happiness effect of a college degree changes within HEE, thus depicting a more comprehensive picture relative to that of previous studies. Third, China has witnessed unprecedented economic success in recent decades, but people’s perceived quality of life has not grown concomitantly. On the contrary, previous studies have underscored a decline in people’s SWB in the Reform Era [8,9]. The findings of this study shed practical light on public policies during the pandemic and can serve as a basis for governmental departments to mitigate the negative impact of the pandemic through the policy tool of HEE. This may also be enlightening for other developing nations where both rapid economic growth and a declining SWB are observed.

This paper is arranged as follows. Section 2 reviews the relevant literature and proposes the research hypotheses. Section 3 describes the data and the empirical identification. Section 4 presents the results and conducts a series of robustness tests. Section 5 further analyzes the potential mechanism and the heterogeneity based on different types of regions. The final section concludes the paper.

2. Literature Review and Hypothesis Development

Over the past few decades, there has been abundant research on the effect of college education on happiness. Most researchers believe that college education is positively associated with happiness, and individuals who have received a college education usually have a higher level of happiness than those who have not [10,11,12,13,14]. Although the positive effect of higher education on SWB has been widely confirmed, we know surprisingly little about how such an effect may vary in an era of higher education expansion (HEE), especially under the impact of the COVID-19 pandemic.

Many scholars have agreed that the pandemic has had various adverse effects on the psychology and SWB of individuals [15]. Wettstein et al. [16] also found that, three to four months after the outbreak in Germany, there was a significant increase in the number of depressed patients, particularly among the elderly, those who were less healthy overall, and those with a greater prevalence of anxiety and pneumonia among adults. The pandemic has hindered interaction among individuals and has continued to cause harm to self-esteem [17], cause mental stress [18], enhance negative emotions [19], and even produce depression [20]. Therefore, overall SWB decreased during the COVID-19 pandemic.

The COVID-19 pandemic has not only posed a serious threat to human life and health, but also presented unprecedented difficulties that hinder economic and social progress. The rising costs of controlling the pandemic, the diminishing effective demand from society, and the restrictions on personnel activity have not only impeded regular recruitment and internship activities, but also reduced the available job opportunities for recent college graduates. In 2020, the number of college graduates in China reached 8.74 million, and approximately 600,000 graduates returned from studying abroad, making it the most challenging employment season in history.

As one of the policy initiatives of the State Council’s joint prevention and control mechanism, the Ministry of Education has vigorously expanded the enrolment scale of higher education since 2020, with the enrolment scale of master’s degree students increasing by approximately 189,000 students annually and the scale of college graduates in general universities increasing by approximately 322,000 students annually. The goal of this expansion policy is, on the one hand, to satisfy graduates’ pursuit of higher education, society’s demand for high-level talent, and further optimize the structure of the labor market, and, on the other hand, to buffer the employment pressure caused by the pandemic and create conditions that allow for the full employment of graduates.

Currently, there is still a wide difference of opinion among scholars concerning the association between HEE and individuals’ SWB. On the one hand, the decrease in the admission threshold provides the possibility of pursuing higher education to those who would otherwise have limited access, thus likely increasing their well-being [6]. It has been argued that higher education can bring both material and non-material benefits to SWB. For example, it helps individuals accumulate human capital [21], obtain higher wages [22], achieve better regional mobility [23], experience an augmented sense of self-control [24], increase their options in the marriage market [25], widen their social network [26], enhance their socioeconomic status and reputation [27], and extend their life expectancy [28]. On the other hand, with the implementation of the HEE policy, the proportion of highly educated people could drastically increase within a short time frame, thus causing the value of the college diploma as an indicator of competence to decrease [29], which may reduce the well-being of those who could pursue further education without the expansion. Thus, this paper proposes the following hypothesis.

H1: 

HEE can enhance individuals’ SWB in the context of the COVID-19 pandemic.

Since the 20th century, the expansion of higher education has become a common phenomenon worldwide, leading to discussions about the impact of this expansion on social class mobility. Does mass higher education truly help promote social class mobility? Previous research has not reached a consensus on the degree of educational access and its fluctuating trends in relation to HEE, social class mobility, and social opportunity structure; it contains divergent views and advocates for models such as the “open school”, “solidified school”, and “stable school”. According to the acquisition model proposed by Blau and Duncan [30], education is both a catalyst for intergenerational upwards mobility and a means of status reproduction. Other theories, such as maximally maintained inequality theory (MMI) [31], effectively maintained inequality theory (EMI) [21], the Lipset–Zetterberg hypothesis [32], industrialization theory [33], new institutionalism theory [34], signal screening theory [35], cultural reproduction theory [36], and rational choice theory [37], have also been proposed in this regard.

Scholars have endeavored to uncover the development logic of HEE and social class mobility from various theoretical perspectives, resulting in the formation of various models regarding education and social class mobility. One of these is the performance mobility model. The concept of “Open school” emphasizes the importance of human capital accumulation and holds the view that the influence of educational factors is more prominent than that of pre-existing factors, such as family background [38]. According to the principle of performance, the market develops a free competition system in which the most capable individuals rise and the less capable fall; consequently, education serves as a ladder that facilitates the vertical flow of society, thus increasing social mobility.

Another model is the family status inheritance model. The solidification school emphasizes the role of cultural reproduction in the intergenerational transmission of family status. In a socially unequal structure, the unequal distribution of social resources by the parent generation can be transformed into a competitive advantage for their children’s generation, thus enabling the inheritance of the family’s social status. The greater that the level of inequality in the distribution of social resources in the parent generation is, the greater that the inequality of social status is for their offspring. Through this process, education has become a tool for class inequality and social reproduction, leading to a solidification trend in the structure of social stratification.

The third model is the state-sheltered mobility model, in which the state intervenes in the process of social class mobility through policy design, thus providing certain classes with more mobility opportunities. Both MMI theory and EMI theory assert that increasing the total number of educational opportunities does not lead to a significant reduction in the inequality of educational opportunities as it is intended to do; unless the dominant class’s demand for quantity and quality in education is met, the inequality of education will remain in effect or even worsen.

Although many previous studies have considered education as a control variable in the SWB model and have thus reported the correlation between education and happiness, few people have discussed and elucidated this, and comprehensive investigations examining this relationship are scarce. Dolan et al. [26] and Dockery [39] have both advocated for further research on this issue. Powdthavee et al. [40] have further suggested that when assessing the association between education and happiness, it is essential to identify the key intermediate variables that influence happiness, as otherwise the effects of these variables on happiness will be “absorbed”, leading to the bias of omitted variables and the misinterpretation of the relationship between education and happiness.

Sociologist Havighurst highlights that, as society progresses, education becomes the main pathway for individuals to move up the social ladder. Subsequent research has confirmed this, with Mizobuchi [41] demonstrating the positive effect of social class identity and social class mobility on SWB across countries, and DiTella et al. [42] providing evidence of a similar effect in a particular country or region. In light of this, this paper proposes that social class mobility should be viewed as a key intermediate variable in the relationship between HEE and individual well-being and, therefore, presents the following research hypothesis H2.

H2: 

HEE can improve individuals’ SWB through promoting social class mobility.

According to prospect theory, proposed by Nobel Laureate Daniel Kahneman and psychologist Amos Tversky [3], an individual’s SWB arises through comparison, which is a consequence of the contrast between an individual’s psychological expectations and their perceived reality (see Figure 1). SWB is determined by the disparity between the reference point and the actual accomplishment. When the gap between the two is positive (that is, relative gains), the individual’s SWB increases; conversely, when the difference between the two is negative (that is, relative loss), the individual’s SWB decreases. In other words, happiness is a result of relative success, while unhappiness is an outcome of relative deprivation. Diener et al. [43] argued that the primary determinant of individual well-being is the subjective difference rather than the actual difference. Prospect theory challenges the basic assumption of the ‘rational individual’ in traditional economics and suggests that human rationality is limited. When making decisions, individuals will have different risk attitudes and behaviors according to their reference points. Prospect theory is characterized by the following three fundamental principles: loss aversion, reference dependence, and diminishing sensitivity.

Loss aversion theory suggests that people are more concerned with avoiding losses than gaining an equivalent amount of income. Therefore, they will work harder to avoid losses than to obtain gains of a similar size. Losses that are more important than gains can be expressed as unfulfilled personal goals. Thaler [44] further explains that individuals tend to value the items they possess more than what they would be willing to pay for the same item. This is known as the endowment effect, where people would require more compensation if they were to give up the item they already own. Loss aversion means that when a reference point is established for a desired outcome, if that result is not achieved, the individual will experience a greater sense of dissatisfaction than the satisfaction of surpassing the goal. That is, people have different sensitivities to gain and loss, with the pain of loss being far greater than the pleasure of gain. For example, the pleasure of picking up USD 100 for nothing can hardly offset the pain of losing USD 100. Since people are more likely to be risk-averse in regard to gains, one can assume that a decrease in social status has a greater effect on SWB than an increase in social status. This can be attributed to individuals’ preference for the status quo. If individuals prefer to maintain their current status, their happiness function will be much steeper when it comes to losses rather than to gains. This is one kind of loss aversion (see Figure 1). McBride [45] conducted research using the GSS and concluded that individuals who had achieved greater social status than their parents were considerably more content than those who had not. Dolan and Lordan [46] utilized data from the UK Cohort Study and established that while upwards mobility did not have a statistically significant influence on life satisfaction, downward mobility was extremely detrimental to both mental health and life satisfaction. Therefore, this paper proposes the following hypothesis.

H3: 

The effect of social class mobility on SWB is asymmetric due to loss aversion, with downwards class mobility having a more negative effect on SWB than the positive effect incurred by upwards class mobility.

The theory of reference dependence highlights the significance of reference point utility rather than absolute value utility. Individuals make decisions based on whether an event is expected to incur a loss or provide a gain, which is evaluated by the value of a benchmark reference point. In other words, people’s assessment of gains or losses is frequently determined by the reference point; for example, when asked to choose between “other people earn USD 50,000 a year and they earn USD 70,000 a year”, and “other people earn USD 120,000 a year and they earn USD 100,000 a year”, most people opt for the former.

Bourdieu proposed a renowned theory of capital transformation, suggesting that people can acquire wealth (economic status), power (political status), and prestige (social status) through education, thus leading to upwards social class mobility. The pursuit of social status is a powerful driver of many social behaviors. However, does higher social status lead to a greater sense of well-being? Diener et al. [43] argue that socioeconomic status (SES) is a poor indicator of SWB. In fact, those who prioritize material wealth tend to experience lower levels of well-being. SES emphasizes an individual’s economic position in society, which is often measured with objective indicators such as economic income, education, and occupation. However, previous research on the correlation between SES and SWB has focused almost exclusively on social status as determined by income and wealth, and has not examined whether other forms of social status exert an impact on SWB.

Based on SES, Anderson et al. [47] proposed an alternative form of social status, known as socio-metric status (SMS). SMS is a measure of the respect and admiration that individuals receive in face-to-face groups, such as those comprising neighbors, colleagues, or classmates, over a long-term period. Compared to SES, an individual’s level of SMS can more accurately reflect their sense of control, acceptance, and influence. Anderson et al. [48] used the method of the MacArthur ladder to examine the effect and relative magnitude of SMS and SES on individuals’ SWB, and found that SMS had a stronger predictive effect on SWB than SES.

The “local-ladder effect” suggests that, as the level of SMS rises or falls, the level of SWB rises or falls accordingly, and these effects are driven by power and social identity, both of which are key determinants of SWB [49]. SMS is defined in a ‘face-to-face group’ and is related to the psychological and social processes that shape one’s social position. In contrast, SES is often described as the economic position of a person in society as a whole. According to Festinger [50], comparing themselves with those in their vicinity is more likely to influence individuals’ well-being than comparing themselves with those at a distance. Similarly, Bertrand Russell [51] asserted that beggars are not jealous of millionaires, but they are certainly jealous of other, more successful beggars. Based on the above analysis, hypothesis H4 is proposed.

H4: 

Reference dependence leads to a local-ladder effect, where the SMS perceived in face-to-face groups is much more important than an individual’s SES in determining their SWB.

3. Empirical Design

3.1. Data and Sample

The data used in this paper were sourced from the Chinese Social Survey (CSS), a continuous sampling survey project that covers the entire nation. This survey utilizes the multi-stage mixed probability sampling (PPS) household survey method, with Chinese citizens aged 18–70 years old as the survey subjects. The questionnaire covers topics such as work status, living conditions, life satisfaction, and social evaluation. This paper utilizes the 2019 and 2021 waves of the CSS, considering the impact of COVID-19 in 2020 and the current data release of CSS. After dropping observations with missing values, a total of 15,301 valid samples were obtained, including 6679 rural samples and 8622 urban samples.

3.2. Models

Since the implementation of the higher education expansion policy in 1999, the rate of higher education enrollment in China has experienced exponential growth. As of 2022, there are 3012 higher education institutions across the country, with a total enrolment of 44.3 million students, making China the largest provider of higher education in the world. This unique context for China’s university enrolment expansion, coupled with the impact of the COVID-19 pandemic, provides a valuable quasi-natural experimental setting for the research topic of this paper. Therefore, this paper constructs the following two-way fixed effects model and employs the DID approach to explore the effects and mechanisms of HEE on individuals’ SWB. The baseline model is structured as follows:

$$SW{B}_{it}={\beta }_0+{\beta }_{1}COVI{D}_{it}\ast HE{E}_{it}+{\sum }_j{\beta }_{j}Contro{l}_{it}+{\gamma }_t+{\mu }_i+{\epsilon }_{it}$$

(1)

where $SW{B}_{it}$ is the dependent variable for individual i in year t. $COVI{D}_{it}$ is a binary variable taking the value of 1 if individual i is affected by the pandemic (that is, t > 2020), and 0 otherwise. $HE{E}_{it}$ is a dummy variable indicating whether an individual is affected by the HEE policy. $Contro{l}_{it}$ represents all control variables. The time fixed effect (${\gamma }_t$) and province fixed effect (${\mu }_i)$ are included as well. It is necessary to further explain the treatment group and the control group. Given the outbreak of the COVID-19 pandemic at the end of 2019, the level of SWB reported by individuals after 2019 is subject to the pandemic. Based on the data from the 2019 and 2021 waves of the CSS, we define $COVI{D}_{it}$ as 1 if the observations occurred in 2021 and 0 if otherwise. Considering that the HEE policy was implemented in 1999, in accordance with the Law of the People’s Republic of China on Compulsory Education, individuals are typically 18 years old if they have completed 9 years of compulsory education, 3 years of high school education, and have taken the college entrance examination. Drawing on Hao and Zhang [52], $HE{E}_{it}$ is defined as the difference between whether a sample has received higher education (including undergraduate education, college/vocational education, and graduate education) and whether their birth year is greater than or equal to 1981. In other words, this paper assumes that those born in 1981 or later and who have received higher education are affected by the HEE policy, and therefore, defines $HE{E}_{it}$ as 1 and 0 if otherwise. This paper focuses on the coefficient ${\beta }_1$ of $COVI{D}_{it}\ast HE{E}_{it}$. If ${\beta }_1$ is significantly positive, it indicates that the HEE can improve an individual’s SWB within the context of pandemic shock.

3.3. Variables

We used self-reported life satisfaction to measure SWB. Specifically, the CSS asked the respondents, “How satisfied are you with your life in general?” The responses were recorded using a ten-point numeric scale between 1 (totally dissatisfied) and 10 (totally satisfied). The Cronbach’s alpha of this measure was 0.863, which indicates the good reliability of this happiness measure. This study focuses on the effects of both SMS and SES on SWB, as a comprehensive and accurate measure of an individual’s social status is a prerequisite for identifying the mechanism of social status acquisition. SMS is the self-identification of an individual’s social status in “face-to-face groups” and it is influenced by reference groups and value judgements. Therefore, this paper adopts class self-identification as a subjective reflection of SMS and also uses income to measure individuals’ SES. Mobility is categorized into upwards mobility and downward mobility with reference to intergenerational succession.

This paper controls for a variety of individual characteristics, such as age and its squared term, gender, marriage, education, political affiliation, and household registration in order to reduce omitted variable problems [53]. Furthermore, a collinearity diagnosis for the model was also conducted, finding VIF values for the independent variables that were lower than 3, which implied that collinearity was not likely to be a big issue.

Table 1. Variable definitions.

Variable	Definition
SWB	A questionnaire was administered using the question, “Overall, how satisfied are you with your life?” as a measure of subjective well-being, with values ranging from 1 to 10 signifying increasing levels of life satisfaction.
COVID	Dummy variable for COVID-19′s shock, which are equal to 1 if the observations occurred in 2021 and 0 if otherwise.
HEE	Dummy variable for HEE, which equals 1 if the observations were influenced by HEE and 0 if otherwise.
Mobility	For social class mobility, the current social status is compared to the social status that respondents self-assessed five years prior, with a result of 1 (Upwards mobility) if it is greater, −1 (Downwards mobility) if it is less, and 0 (no change) if it remains the same.
SES	The respondent’s SES was classified into three categories based on the P25 and P75 of the reported income, with high being set to 3, medium to 2, and low to 1.
SMS	Each respondent’s SMS was measured according to the responses given in the questionnaire, “Which social class do you consider yourself to be currently in?”, with the values ranging between upper = 3, middle = 2, and lower = 1.
Gender	Male = 1, and female = 0.
Age	Year of research minus the reported year of birth.
Age squared	The square of age.
Marriage	Dummy variable for Marriage, where married = 1, and unmarried (including divorced, widowed) = 0.
Education	Years of education, where illiterate = 0, primary or private school = 6, middle school = 9, high school (including general high school and other equivalent secondary vocational and technical schools) = 12, specialist = 15, bachelor’s degree = 16, and master’s degree and above = 19.
Party	Political affiliation, where Communist Party member = 1, Communist Youth League member = 2, Democratic Party member = 3, and General = 4.
Register	Dummy variable for household registration, where urban = 1, and rural = 0.
Han ethnicity	Dummy variable for ethnicity, where Han Chinese = 1, and non-Han Chinese = 0.

Notes: This table provides the definitions of variables used in the analysis.

provides definitions of the variables mentioned above.

3.4. Descriptive Statistics

Table 2. Summary statistics.

Variable	Observation	Mean	SD	Min	Max
SWB	15,301	7.167	2.231	1.000	10.000
COVID × HEE	20,391	0.074	0.261	0.000	1.000
Mobility	20,419	0.134	0.661	−1.000	1.000
SES	20,419	2.135	0.741	1.000	3.000
SMS	20,130	1.562	0.626	1.000	3.000
Gender	20,419	0.435	0.496	0.000	1.000
Age	20,419	46.602	14.371	18.000	70.000
Age squared	20,419	2378.269	1292.549	324.000	4900.000
Marriage	20,410	0.797	0.402	0.000	1.000
Education	20,391	9.393	4.410	0.000	19.000
Party	20,418	3.536	1.008	1.000	4.000
Register	20,301	0.329	0.470	0.000	1.000
Ethnicity	20,325	0.919	0.272	0.000	1.000

Notes: This table presents the summary statistics of the variables used in the analysis. All variables are defined in detail in Table 1.

presents the descriptive statistics. The sample average of SWB is 7.167, indicating a relatively high level of well-being. Consistent with the expectations of industrialization theory, China’s social class structure is in an open state of upwards mobility, as the average value of mobility is greater than 0. There is no clear evidence of an increasingly solidified mobility pattern in China’s social classes. The average value of SMS (1.562) is in the lower-middle range and is significantly lower than that of SES, indicating that, although the middle social stratum in China’s social structure continues to expand, the self-positioning of the stratum is relatively low.

4. Empirical Results

4.1. Baseline Regression

Table 3. Baseline regression.

	(1)	(2)
	SWB	SWB
COVID × HEE	0.417 ***	0.202 ***
	(0.047)	(0.052)
Gender		−0.060 **
		(0.029)
Age		−0.087 ***
		(0.008)
Age squared		0.001 ***
		(0.000)
Marriage		0.331 ***
		(0.045)
Education		0.027 ***
		(0.005)
Party		−0.139 ***
		(0.014)
Register		0.134 ***
		(0.032)
Ethnicity		−0.086
		(0.057)
Year FEs	Yes	Yes
Region FEs	Yes	Yes
Wald chi2 (p value)	79.81 (0.000)	518.20 (0.000)
Log pseudolikelihood	−30,571.851	−30,067.41
Pseudo R2	0.001	0.008
Observations	15,278	15,135

Notes: This table presents the baseline regression results of the impact of COVID-19 interaction with higher education expansion (HEE) on subjective well-being (SWB). Variable definitions are explained in detail in Table 1. The reference categories for the yearly dummies and regional dummies are 2019 and the mid-western region, respectively. Robust standard errors clustered at the individual level are provided in parentheses, and cut-off points are included but not reported. ***, ** denote statistical significance at the 1% and 5% levels, respectively.

reports the average treatment effects of higher education expansion on individuals’ SWB under the COVID-19 shock. Column (1) does not include control variables while Column (2) includes all the control variables. Each column of regression controls for both time and province fixed effects, and the reported standard errors are clustered at the province level. The results displayed in Table 3 show that the coefficient of COVID × HEE is significantly positive regardless of whether the control variables are included or not. This indicates that the expansion of higher education can enhance individuals’ SWB under the context of pandemic shock, and can be considered as an effective policy tool for reducing the negative effects of the pandemic, thus verifying research hypothesis H1.

In addition, the effects of the control variables on SWB are in line with the literature. For instance, women are generally reported to have higher levels of well-being than men [54]. There is a U-shaped relationship between age and SWB, with younger and older people being the happiest [55]. Married individuals tend to experience higher levels of well-being, as marriage is believed to bring about a more lasting intimate relationship with one’s partners [56]. Furthermore, higher levels of education are linked to greater happiness levels [27], and urban residents are typically happier than rural residents [57].

4.2. Parallel Trends Tests

Passing the test of parallel trends is a prerequisite for the validity of a DID model. Following Beck et al. [58], this paper conducts an event study to decompose the impact of higher education expansion in specific years under the epidemic shock. Specifically, we first add 18 to the birth year of the respondents who have received higher education, and then subtract it from 1999, that is, the year when the college enrollment expansion policy was implemented. Then, we construct several binary variables that indicate each value of the above variable, such as HEE (−2) and HEE (−1), and multiply these binary variables with the variable of COVID. Finally, we include these new interaction terms in the baseline regression model.

Table 4. Test for parallel trends.

	(1)
	SWB
COVID × HEE (−2)	0.186
	(0.312)
COVID × HEE (−1)	0.111
	(0.397)
COVID × HEE	0.490 **
	(0.250)
COVID × HEE (+1)	0.444 **
	(0.245)
COVID × HEE (+2)	0.176 ***
	(0.054)
Gender	−0.058 **
	(−1.997)
Age	−0.091 ***
	(−11.523)
Age squared	0.001 ***
	(12.292)
Marriage	0.330 ***
	(7.295)
Education	0.024 ***
	(4.834)
Party	−0.136 ***
	(−9.487)
Register	0.128 ***
	(3.962)
Ethnicity	−0.084
	(−1.482)
Year FEs	Yes
Region FEs	Yes
Wald chi2 (p value)	538.80 (0.000)
Log pseudolikelihood	−30,060.782
Pseudo R2	0.0082
Observations	15,135

Notes: This table presents the results of the parallel trends test. All variable definitions are provided in Table 1. The reference categories for the yearly dummies and regional dummies are 2019 and the mid-western region, respectively. Robust standard errors clustered at the individual level are provided in parentheses, and cut-off points are included but not reported. ***, ** denote statistical significance at the 1% and 5% levels, respectively.

shows the results of the test of parallel trends. The coefficients of the two interaction terms, that is, COVID × HEE (−2) and COVID × HEE (−1), are statistically non-significant. This indicates that the SWB between the experimental group (influenced by HEE and COVID) and the control group (not influenced by HEE and COVID) before the expansion policy had no significant differences, satisfying the assumption of parallel trends. Meanwhile, the coefficients of the other interaction terms, that is, COVID × HEE, COVID × HEE (+1), and COVID × HEE (+2), are significantly positive at the 1% or 5% level, and a decreasing effect can also be observed among them, indicating that the average effect of the HEE on SWB decreases over time as the expansion in college enrollment continues.

4.3. Placebo Tests

In this section, we conduct three placebo tests to address the concerns regarding endogeneity problems caused by potential omitted variables and confounding factors. We assume that there were fictitious expansions in higher education occurring in 1996, 1997, and 1998, and substitute the dependent variable of interest in the baseline model. The corresponding results of the fictitious policy implementation periods are shown in

Table 5. Placebo tests.

	(1)	(2)	(3)
	SWB	SWB	SWB
COVID × HEE1998	0.002
	(0.003)
COVID × HEE1997		0.002
		(0.003)
COVID × HEE1996			0.003
			(0.003)
Gender	−0.066 **	−0.066 **	−0.066 **
	(0.029)	(0.029)	(0.029)
Age	−0.088 ***	−0.088 ***	−0.088 ***
	(0.008)	(0.008)	(0.008)
Age squared	0.001 ***	0.001 ***	0.001 ***
	(0.000)	(0.000)	(0.000)
Marriage	0.329 ***	0.330 ***	0.330 ***
	(0.045)	(0.045)	(0.045)
Education	0.028 ***	0.028 ***	0.028 ***
	(0.005)	(0.005)	(0.005)
Party	−0.144 ***	−0.144 ***	−0.144 ***
	(0.014)	(0.014)	(0.014)
Register	0.131 ***	0.131 ***	0.131 ***
	(0.032)	(0.032)	(0.032)
Ethnicity	−0.083	−0.083	−0.083
	(0.057)	(0.057)	(0.057)
Year FEs	YES	YES	YES
Region FEs	YES	YES	YES
Wald chi2 (p value)	546.51 (0.000)	547.28 (0.000)	548.03 (0.000)
Log pseudolikelihood	−30,050.368	−30,050.333	−30,050.298
Pseudo R2	0.009	0.009	0.009
Observations	15,135	15,135	15,135

Notes: This table presents the results of the placebo test with fictitious event time. All variable definitions are provided in Table 1. The reference categories for the yearly dummies and regional dummies are 2019 and the mid-western region, respectively. Robust standard errors clustered at the individual level are provided in parentheses, and cut-off points are included but not reported. ***, ** denote statistical significance at the 1% and 5% levels, respectively.

, where the coefficients of COVID × HEE1998, COVID × HEE1997, and COVID × HEE1996 are all insignificant. This indicates that the finding from the baseline model is robust.

4.4. Other Robustness Tests

We further test the sensitivity of our baseline regression results using alternative explained variables, the propensity score matching (PSM) method, and subsample regression. First, we use another SWB measure, which was elicited by asking respondents how much they agree with the statement “I am a happy person in general”. The answers range from “strongly disagree (1)” to “strongly agree (6)”, with higher values signifying a higher level of SWB. Column (1) in

Table 6. Other robustness tests.

	Substituting the Dependent Variable	Estimation of PSM–DID	Excluding Province with Ethnic Minorities
	(1)	(2)	(3)
	Well-being	SWB	SWB
COVID × HEE	0.527 **	0.273 ***	0.176 ***
	(0.210)	(0.069)	(0.067)
Gender	−0.095 ***	−0.085 **	−0.046
	(0.036)	(0.040)	(0.038)
Age	−0.087 ***	−0.082 ***	−0.088 ***
	(0.027)	(0.011)	(0.010)
Age squared	0.001 ***	0.001 ***	0.001 ***
	(0.000)	(0.000)	(0.000)
Marriage	0.393 ***	0.543 ***	0.298 ***
	(0.061)	(0.062)	(0.059)
Education	0.024 ***	−0.021 ***	0.028 ***
	(0.006)	(0.007)	(0.006)
Party	−0.171 ***	−0.118 ***	−0.139 ***
	(0.017)	(0.020)	(0.019)
Register	0.074 *	0.013	0.127 ***
	(0.041)	(0.044)	(0.042)
Ethnicity	−0.075	−0.022	−0.126 **
	(0.072)	(0.078)	(0.064)
Year FEs	YES	YES	YES
Region FEs	YES	YES	YES
Wald chi2 (p value)	346.17 (0.000)	219.87 (0.000)	319.12 (0.000)
Log pseudolikelihood	−20,879.132	−10,395.991	−17,734.729
Pseudo R2	0.0075	0.0111	0.0081
Observations	10,425	10,046	8847

Notes: This table presents the results of the robustness test over the alternative explained variables, PSM–DID matching, and subsample regression. All variable definitions are provided in Table 1. The reference categories for the yearly dummies and regional dummies are 2019 and the mid-western region, respectively. Robust standard errors clustered at the individual level are provided in parentheses, and cut-off points are included but not reported. ***, **, * denote statistical significance at the 1%, 5% and 10% levels, respectively.

shows the result of this new dependent variable. The coefficient of the interaction term of interest is still significantly positive at the 1% level, implying that the benchmark result is reliable.

Second, we employ the PSM method to counterbalance the potential selection bias. When employing the DID model to assess the policy effect, the optimal situation should be one in which the individual characteristics of the treatment group and those of the control group are identical prior to policy implementation. However, due to individual heterogeneity, the sample used in this paper may not have satisfied this requirement, thus potentially leading to selection bias. To address this issue, this paper further applies PSM to reassess and examine the matching control group, and Column (2) in Table 6 shows the test result. The coefficient of COVID × HEE remains significantly positive.

Third, considering the particularity of the temperament and living habits of ethnic minorities, this paper excludes observations from provinces with a high density of ethnic minorities, including Xinjiang, Tibet, Ningxia, Qinghai, Guizhou, and Yunnan, and re-estimates the baseline model. The result using this subsample is presented in Column (3) of Table 6, and again, the coefficient of COVID × HEE is significantly positive.

5. Further Analysis

5.1. Potential Mechanism

Next, we examine the potential mechanism through which the HEE policy enhances individuals’ SWB in response to the pandemic shock. Before, we argued that the expansion of higher education can promote social class mobility and hence enhance SWB. Thus, we use the Baron and Kenny [59] method to test the mediating role of social class mobility. First, we replace the dependent variable in the baseline model with mobility, and Column (1) shows that COVID × HEE is significantly associated with higher social class mobility during the COVID-19 pandemic. Then, we include mobility in the Equation (1) and drop the interaction of interest. The coefficient of mobility in Column (2) is also significantly positive, indicating that social class mobility can enhance individuals’ SWB. In Column (3), we further introduce both COVID × HEE and mobility in the regression model, which shows that their coefficients are still significantly positive, with the coefficient of COVID × HEE decreasing from 0.202 to 0.170 and the R-squared increasing from 0.008 to 0.012. The overall results in

Table 7. Channel tests.

	(1)	(2)	(3)
	Mobility	SWB	SWB
COVID × HEE	0.231 ***		0.170 ***
	(0.061)		(0.052)
Mobility		0.345 ***	0.344 ***
		(0.022)	(0.022)
Gender	−0.150 ***	−0.048 *	−0.047
	(0.028)	(0.029)	(0.029)
Age	−0.076 ***	−0.080 ***	−0.078 ***
	(0.007)	(0.008)	(0.008)
Age squared	0.001 ***	0.001 ***	0.001 ***
	(0.000)	(0.000)	(0.000)
Marriage	0.077 *	0.314 ***	0.321 ***
	(0.042)	(0.045)	(0.045)
Education	−0.006	0.030 ***	0.028 ***
	(0.004)	(0.005)	(0.005)
Party	−0.033 **	−0.139 ***	−0.138 ***
	(0.015)	(0.014)	(0.014)
Register	−0.067 **	0.139 ***	0.139 ***
	(0.032)	(0.032)	(0.032)
Ethnicity	−0.094 *	−0.071	−0.071
	(0.052)	(0.057)	(0.057)
Year FEs	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes
Wald chi2 (p value)	286.04 (0.000)	755.31 (0.000)	773.86 (0.000)
Log pseudolikelihood	−19,716.083	−29,945.325	−29,942.081
Pseudo R2	0.0073	0.0120	0.0121
Observations	20,173	15,135	15,135

Notes: This table presents the results of the potential mechanism test. Dependent variables in columns (1) to (3) are mobility, SWB, and SWB, respectively. All variable definitions are provided in Table 1. The reference categories for the yearly dummies and regional dummies are 2019 and the mid-western region, respectively. Robust standard errors clustered at the individual level are provided in parentheses, and cut-off points are included but not reported. ***, **, * denote statistical significance at the 1%, 5% and 10% levels, respectively.

suggest that social class mobility acts as a significant mediator in the relationship between HEE and SWB under a pandemic shock, which validates Hypothesis H2.

5.2. Test for Different Directions of Social Mobility

To test hypothesis H3, we include downward social mobility (Down_mob) and upwards social mobility (Up_mob) separately and aggregate them into the baseline model;

Table 8. Test for different directions of social mobility.

	(1)	(2)	(3)
	SWB	SWB	SWB
COVID × HEE	0.174 ***	0.191 ***	0.171 ***
	(0.052)	(0.052)	(0.052)
Up_mob	0.405 ***		0.322 ***
	(0.031)		(0.032)
Down_mob		−0.489 ***	−0.375 ***
		(0.040)	(0.042)
Gender	−0.053 *	−0.051 *	−0.047
	(0.029)	(0.029)	(0.029)
Age	−0.080 ***	−0.083 ***	−0.078 ***
	(0.008)	(0.008)	(0.008)
Age squared	0.001 ***	0.001 ***	0.001 ***
	(0.000)	(0.000)	(0.000)
Marriage	0.322 ***	0.328 ***	0.322 ***
	(0.045)	(0.045)	(0.045)
Education	0.028 ***	0.027 ***	0.028 ***
	(0.005)	(0.005)	(0.005)
Party	−0.139 ***	−0.138 ***	−0.138 ***
	(0.014)	(0.014)	(0.014)
Register	0.146 ***	0.127 ***	0.138 ***
	(0.032)	(0.032)	(0.032)
Ethnicity	−0.073	−0.080	−0.071
	(0.057)	(0.057)	(0.057)
Year FEs	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes
Wald chi2 (p value)	688.75 (0.000)	675.63 (0.000)	774.64 (0.000)
Log pseudolikelihood	−29,983.37	−29,990.336	−29,941.682
Pseudo R2	0.0107	0.0105	0.0121

Notes: This table presents the results of the H3 test. All variable definitions are provided in Table 1. The reference categories for the yearly dummies and regional dummies are 2019 and the mid-western region, respectively. Robust standard errors clustered at the individual level are provided in parentheses, and cut-off points are included but not reported. ***, * denote statistical significance at the 1% and 10% levels, respectively.

shows the results. Comparing the regression results across all three columns, we can see that the negative impact of downward social mobility on individuals’ SWB during the pandemic is significantly higher than the positive impact of upwards social mobility. This demonstrates that the impact of social mobility on SWB is asymmetric when taking intergenerational succession as a reference. The empirical results follow the theoretical expectation of hypothesis H3, which suggests that, due to loss aversion, individuals feel differently when experiencing upwards or downward social mobility, and the losses from downward social mobility are much greater than the gains from upwards social mobility. This may also be attributed to a preference for the status quo, as some individuals may prefer to maintain their current social status and living conditions. Faced with pandemic shock, social competition, and economic downturn, an increasing number of young people are choosing to ‘lie flat’.

5.3. Test for Different Types of Social Status

As with the above section, we separately and aggregately include SMS and SES into the baseline model. The regression results in

Table 9. Test for different types of social status.

	(1)	(2)	(3)
	SWB	SWB	SWB
COVID × HEE	0.176 ***	0.168 ***	0.150 ***
	(0.052)	(0.056)	(0.056)
SES	0.173 ***		0.126 ***
	(0.021)		(0.022)
SMS		0.879 ***	0.869 ***
		(0.024)	(0.024)
Gender	−0.102 ***	−0.016	−0.047
	(0.030)	(0.030)	(0.030)
Age	−0.097 ***	−0.081 ***	−0.088 ***
	(0.008)	(0.008)	(0.008)
Age squared	0.001 ***	0.001 ***	0.001 ***
	(0.000)	(0.000)	(0.000)
Marriage	0.310 ***	0.243 ***	0.229 ***
	(0.045)	(0.046)	(0.046)
Education	0.022 ***	0.015 ***	0.012 **
	(0.005)	(0.005)	(0.005)
Party	−0.137 ***	−0.109 ***	−0.108 ***
	(0.014)	(0.015)	(0.015)
Register	0.100 ***	0.151 ***	0.125 ***
	(0.032)	(0.033)	(0.033)
Ethnicity	−0.099 *	−0.004	−0.016
	(0.057)	(0.057)	(0.057)
Year FEs	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes
Wald chi2 (p value)	774.64 (0.000)	1775.21 (0.000)	1790.17 (0.000)
Log pseudolikelihood	−29,941.682	−30,033.898	−28,945.674
Pseudo R2	0.0091	0.0305	0.0311
Observations	15,135	14,934	14,934

Notes: This table presents the results of the H4 test. All variable definitions are provided in Table 1. The reference categories for the yearly dummies and regional dummies are 2019 and the mid-western region, respectively. Robust standard errors clustered at the individual level are provided in parentheses, and cut-off points are included but not reported. ***, **, * denote statistical significance at the 1%, 5% and 10% levels, respectively.

show that both SMS and SES have a significant positive effect on individuals’ SWB, but SMS has a much higher effect than SES. It is believed that those with a higher SMS tend to have more friends, are more involved in social activities, have more control over group decisions and have more sway over the opinions of others. Additionally, SMS, as a symbol of mutual respect and admiration among peers, can have a profound effect on an individual’s SWB and social acceptance. In fact, economic conditions have a marginal diminishing effect on SWB, while the respect, trust, and approval of those around us are invaluable and do not diminish marginally by contrast. The theory of reference dependence highlights the value of ‘reference points’, as individuals are more likely to assess their well-being in comparison to those around them than those at a distance, resulting in those with greater SMS attaining higher levels of well-being. This finding demonstrates that there is a local ladder effect, which supports hypothesis H4; this also provides an explanation for the notion that “money cannot buy happiness”.

5.4. Heterogeneity Analysis

In contrast to the average effect, the degree of heterogeneity is related to whether people consistently benefit from the HEE policy. Thus, we separate the sample based on individuals’ household registration, for which the corresponding regression results are provided in

Table 10. Heterogeneity analysis based on different types of region.

	(1)	(2)	(3)	(4)
	Mobility	Mobility	SWB	SWB
	Rural	Urban	Rural	Urban
COVID × HEE	0.146 *	0.269 ***	0.096	0.312 ***
	(0.089)	(0.088)	(0.067)	(0.087)
Gender	−0.205 ***	−0.055	−0.079 **	−0.024
	(0.035)	(0.049)	(0.036)	(0.051)
Age	−0.057 ***	−0.111 ***	−0.069 ***	−0.130 ***
	(0.009)	(0.013)	(0.010)	(0.013)
Age squared	0.001 ***	0.001 ***	0.001 ***	0.002 ***
	(0.000)	(0.000)	(0.000)	(0.000)
Marriage	−0.009	0.234 ***	0.212 ***	0.579 ***
	(0.053)	(0.069)	(0.057)	(0.077)
Education	−0.005	−0.000	0.021 ***	0.050 ***
	(0.005)	(0.008)	(0.006)	(0.009)
Party	−0.070 ***	0.013	−0.156 ***	−0.119 ***
	(0.022)	(0.022)	(0.020)	(0.022)
Ethnicity	−0.077	−0.140	−0.084	−0.051
	(0.060)	(0.103)	(0.066)	(0.104)
Year FEs	Yes	Yes	Yes	Yes
Wald chi2 (p value)	160.85 (0.000)	148.11 (0.000)	237.79 (0.000)	237.34 (0.000)
Log pseudolikelihood	−13,365.87	−6320.1581	−20,554.747	−9365.1769
Pseudo R2	0.0060	0.0122	0.0049	0.0132
Observations	13,509	6664	10,238	4897

Notes: This table presents the results of heterogeneity tests based on different types of region. Dependent variables in columns (1) to (3) are mobility, SWB, and SWB, respectively. All variable definitions are provided in Table 1. The reference categories for the yearly dummies and regional dummies are 2019 and the mid-western region, respectively. Robust standard errors clustered at the individual level are provided in parentheses, and cut-off points are included but not reported. ***, **, * denote statistical significance at the 1%, 5% and 10% levels, respectively.

. Columns (1) and (2) show that the positive effect of HEE on social class mobility is significantly higher for the urban sample than for the rural sample within the context of pandemic shock. Furthermore, the regression results in Columns (3) and (4) demonstrate that HEE has a significant positive effect on the SWB of urban residents, while it does not have a significant influence on the SWB of rural residents. This means that the increased opportunity to enroll in higher education was not evenly distributed to benefit both urban and rural residents. Instead, individuals from urban areas benefit more and thus experience a greater increase in social mobility and SWB. Despite the surge in higher education opportunities during the pandemic, family background still plays a major role in determining whether individuals have access to higher education.

Compared with urban individuals of the same age, the chances of rural children receiving higher education have become increasingly smaller. Therefore, expanding enrollment in higher education does not significantly improve the SWB of rural individuals. It has been widely suggested by researchers [60] that factors such as family background, ethnic and racial origin, and gender determine the distribution of higher education opportunities, which can result in individuals of lower social classes, ethnic and racial minorities, and women facing disadvantages in the competition for higher education opportunities, thus reinforcing social stratification. The household registration system of China divides Chinese society into two distinct regions, urban and rural, with a significant disparity in the availability and quality of educational resources between the two. Although the implementation of higher education expansion results in a substantial increase in the number of opportunities to enroll in higher education, the educational inequality that exists between the urban and rural locations remains a cause for concern.

6. Concluding Remarks

The emergence of the knowledge economy and the increase in high-skilled employment opportunities have driven the implementation of policies to expand higher education on a global scale. It is widely believed that education and human capital accumulation are crucial for economic growth, poverty reduction, and human development in developing countries. In the face of the unprecedented challenge posed by the COVID-19 pandemic, which has become a global crisis, the employment situation for college students has become much more severe, thus making the implementation of HEE policies a special historical mission. Given the inevitable trend seen in the expansion of higher education, it is necessary to investigate the relationship between HEE and individual SWB in the context of the proliferation of college enrollment.

Based on the prospect theory and the idea of the “sorting machine” in education, proposed by American educator Joe Spring, this paper uses the DID approach to evaluate the effects of the HEE policy during the pandemic. This study confirms that during the COVID-19 pandemic, the HEE policy was able to significantly increase individual SWB, through which social class mobility plays a significant mediating role. It can be concluded that the HEE can elevate national happiness by facilitating social class mobility. Due to loss aversion, the effect of social class mobility on SWB is asymmetric, with downward class mobility having a negative effect on SWB that is greater than the positive effect of upwards class mobility. We also find that there is a local ladder effect due to reference dependence, and that SMS among face-to-face groups has a greater impact on SWB than SES does.

According to Easterlin’s happiness paradox, economic growth does not necessarily lead to an increase in people’s SWB; thus, to improve people’s SWB, it is essential to not only satisfy their material needs, but also to enhance their nonmaterial living conditions. One of the most important aspects of this process is social mobility, which determines whether people have fair opportunities to progress upwards and achieve a certain status. As the saying goes, “running water does not rot”. An ideal social structure should be flexible, dynamic, and invigorating, allowing individuals from different family backgrounds to have an equal opportunity to compete in the marketplace.

The findings of this paper have multiple policy implications. First, the government should strive to create a fair educational environment, provide a reasonable path for social mobility, stimulate individuals’ potential and motivation to move up the social ladder, and thereby enhance their overall well-being. Secondly, while expanding higher education, efforts should be made to shift our focus from providing a wide coverage of higher education to providing high-quality higher education, in order to ensure that the increasing number of college students can be accommodated by the existing higher education facilities and faculty capacity; a balance between quantity and quality should be strived for. Lastly, happiness comes from comparison, and what matters for happiness is not what individuals have but what their reference group does not have. The government should make an effort to reshape a universally recognized public value orientation, vigorously cultivate a proactive, rational, peaceful, open, and inclusive social mentality that can help people reduce grievances and complaints, and improve people’s awareness of how to pursue happiness.

However, several limitations of this study must be mentioned. First, HEE can function as a ladder for upward social mobility for members of society, but it can also solidify social mobility through social reproduction. Therefore, HEE can impact an individual’s educational decisions, making it challenging to clarify the pure effect of education on social class mobility and individual SWB. Second, the educational decisions made by individuals with varying unobserved and missing variables, such as their personal aptitudes and family background, and the model’s strong endogeneity, pose a challenge to examining the impact mechanisms of HEE on SWB. Therefore, further research will be necessary in order to establish the causal effects via examining the experimental design and effective instrumental variables. Third, there is a lack of empirical research on reference points, particularly on the mechanisms involved in individual decision-making processes based on different reference points under conditions of uncertainty. An obvious extension to the general analysis would be to specifically examine whether our findings remain unchanged when using various approaches to define reference groups. This should be explored in greater depth for further investigation, which is beyond the scope of our study.