COVID-19 and social inequality in China: the local–migrant divide and the limits of social protections in a pandemic

Abstract

The existing literature suggests that external shocks, such as pandemics, stimulate people’s demand for social protections and prompt them to favor short-term social consumption over long-term investments. However, this argument may not apply fully in a society with an urban–rural divide in addition to an unequal welfare system. Through a telephone survey conducted in July 2020, this study sought to investigate public opinions on the social policy response to the coronavirus disease pandemic in China. Quantitative evidence showed large economic hardship among the respondents, who expressed a strong expectation for labor market interventions instead of social assistance. This study reveals that the preexisting inequalities in people’s access to welfare benefits have led local residents and migrants to develop differential preferences for social policies. This attitudinal heterogeneity is illustrative of the inequalities in the Chinese welfare system as well as of the labor market dynamics that have resulted from massive internal migration and the informalization of the workforce. The division between locals and migrants in China’s urban welfare system has shaped a demarcation of welfare preferences between the two groups through peculiar interpretive feedback effects.

COVID-19, social protection, attitudes, China, inequality

The economic devastation triggered by the coronavirus disease (COVID-19) pandemic has exacerbated existing social inequalities in most countries. The disease’s sudden and significant disruption of economic activities has led to massive unemployment, widespread cuts in wages, and sharply rising household hardship. As expected, the people hardest hit in the pandemic have often been the groups that were already vulnerable before the crisis, thus reflecting significant preexisting inequalities. Social policies are an indispensable part of a coordinated policy response to the crisis and have been necessary to keep individuals and households afloat. Most governments have introduced short-term social protection arrangements and fiscal packages. Oft-used policy tools include wage subsidies, paid leaves, job retention programs, and new cash transfer schemes targeting those who are not covered by existing social welfare programs (Béland et al., 2021). At the same time, existing programs, such as unemployment benefits and minimum wage, have continued to play a role and to be further strengthened in many countries (Gentilini et al., 2020; Soon et al., 2021).

This study examines the situation in China, where the containment of COVID-19 was extraordinarily swift. While the rest of the world was still wrestling with the soaring pandemic, China had largely managed to tame it. The whole-of-the-nation approach, strengthened by the high mobilizational and coercive capacity of the state, created a distinctive style of crisis governance that was conducive to pandemic control (He et al., 2020; Mei, 2020). Notwithstanding its role as the global outperformer in its COVID-19 response, however, very little is known regarding the implications of this pandemic for social inequalities in China. Although the “torture” that the pandemic had imposed in China was considerably short vis-à-vis other countries, anecdotal and scholarly data have revealed a disproportionate impact on low-skilled workers, rural-to-urban migrants, and fresh graduates from university (Che et al., 2020; Wang et al., 2021). Given the soaring income inequalities prior to the crisis (Xie & Zhou, 2014), one can reasonably posit that inequalities may further increase during and possibly after the pandemic.

The existing literature suggests that external shocks, such as pandemics and economic crises, stimulate people’s demand for social protections and prompt them to favor short-term social consumption—cash transfers, in-kind assistance, fee waivers, and the like—in preference to long-term investments such as education and training (Han & Kwon, 2020; Häusermann et al., 2014; Rehm, 2009; Thewissen & Rueda, 2017). However, our study found that the argument does not fully apply in a society with an urban–rural divide in conjunction with an unequal welfare system. Social security serves to cushion the adverse impact caused by labor market shocks and individual-level risks. However, as has been seen elsewhere, these policies themselves are often structured along residential, occupational, ethnic, age, and gender lines that may restrict necessary access to social protections for vulnerable populations during large-scale crises such as the current pandemic. The Chinese welfare system has long been known as a fragmented, transitional one that is built primarily on occupational status and local citizenship (Ringen & Ngok, 2017). The urban–rural dichotomy and the divide between formal and informal sectors have created an unequal welfare system, despite governmental efforts in recent years to reduce such disparities (Jiang et al., 2018; Shi, 2012). This study found that the country’s preexisting inequalities in access to welfare benefits have led local residents and migrants to exhibit differential preferences for social policies. That attitudinal heterogeneity is illustrative of the inequalities in the Chinese welfare system as well as of the labor market dynamics resulting from massive internal migration and the informalization of the nation’s workforce.

COVID-19 and the social policy response in China

The COVID-19 outbreak caught most countries unprepared. Soon after China’s nationwide concerted response had yielded positive outcomes, the Chinese government started to pay serious attention to social protections. Two directives were issued by the central authorities, on 21 February and 6 March 2020, respectively, announcing a social policy package. Table 1 summarizes the details. Notably, stabilizing employment was given the utmost priority, with the key instrument being to substantially relieve the burden on enterprises of making social insurance contributions. In February 2020 alone, a total of 123.8 billion yuan in social insurance contributions was waived (Xinhua News Agency, 2020a). Small and medium enterprises that refrained from enacting major employee layoffs were rewarded with a maximum full rebate of the premiums they had paid to the unemployment insurance scheme in 2019. As of late April 2020, 1.3 million enterprises had been rewarded with a total of 18.6 billion yuan in rebates (Lu et al., 2020). In addition, active labor market policies were pursued to create more jobs through public works. The central authorities urged local governments to accelerate the resumption of government-funded projects in infrastructure, energy, power, and IT networks. Released in May 2020, China’s fiscal stimulus package highlighted promoting employment through cutting taxes and social insurance contributions. Notably, the stimulus package did not include initiatives to expand social welfare programs (Wong & Qian, 2020).

Table 1.

Social policy package of the Chinese government from early- to mid-2020.

Policy category	Content
Employment promotion: corporate support to stabilize employment	(1) Waiving half social insurance contributions of large enterprises for up to 3 months and up to 5 months for small and medium enterprises (2) Deferment of social insurance contributions for hard-hit enterprises for up to 6 months (3) Conditional refund of unemployment insurance benefits
Employment promotion: active labor market policies	(1) Central government urged localities to accelerate the resumption of public construction projects (2) New public works were supported by special local government bonds (3) Local government subsidy to enterprises offering special vocational training programs (4) Promoting online employment service (5) Loans to enterprises to support employment
Wage protection	Central government ordered employers to pay full wage for the first salary payment cycle
Unemployment subsidies	Offered to laid-off employees with less than one year contribution history
Consumption vouchers	Local governments with necessary fiscal capacity offered consumption voucher to residents
Existing social protection benefits	Central government ordered the timely payment and discretionary increase of existing cash transfer programs (especially dibao) to poor households and other vulnerable individuals

Source: State Council, Notice on strengthening work related to employment and poverty alleviation in response to COVID-19, 21 February 2020, retrieved from http://www.gov.cn/zhengce/zhengceku/2020-02/24/content_5482770.htm; Notice on social assistance for poor households during COVID-19 response, 6 March 2020, retrieved from http://www.gov.cn/gongbao/content/2020/content_5496767.htm (Gentilini et al., 2020; Lu et al., 2020; Soon et al., 2021).

China made much less use of direct subsidies, as far as central government policies were concerned, than other East Asian economies did (Soon et al., 2021). Instead, local governments with necessary fiscal capacity were encouraged to launch discretionary initiatives, such as top-up cash allowances, extension of the unemployment benefit period, and consumption vouchers. For example, in Xiamen City, local residents who were ineligible for claiming unemployment benefits were offered a monthly allowance of up to 1000 yuan.¹ A total of 155 prefectural cities had issued consumption vouchers to local residents by June 2020 (Cao et al., 2020). Taken as a whole, the social policy response package of the Chinese government was undertaken by (a) introducing emergency intervention programs, (b) enhancing existing programs, and (c) relaxing eligibility criteria for individuals. Whereas policy-makers tried to employ both types of conventional social protection measures—short-term social consumption and longer-term investments, especially via cash transfers and active labor market policies—the latter was given prominence. That emphasis partly reflects the long-held developmental logic of social policy in China (Soon et al., 2021).

The social welfare system and inequality in urban China

The urban–rural divide and social inequality

As is widely documented in the literature, various forms of social inequalities in contemporary China are deeply rooted in the country’s urban–rural dichotomy and its market transition. The current dichotomy began in the 1950s with the institution of the hukou (household registration) system, which was established to cope with rising demographic pressures in the course of rapid industrialization and assigned citizens into either an agricultural (rural) status or a non-agricultural (urban) status that spatially segregated both populations (Wu & Treiman, 2007). The divisive hukou system was associated with a sharp disparity of rights and privileges. Essentially “locked” to the countryside by their rural hukou status, the vast rural population was not entitled to the same social services and welfare benefits that their urban peers were (Chan & Zhang, 1999).

The market transition that commenced in the late 1970s unleashed China’s immense economic potential. More than 100 million rural-to-urban migrant workers, or approximately 11% of the total population, became the chief labor force for China’s vast labor-intensive industries (Zhang & Wu, 2017). Despite certain relaxations of mobility and employment restrictions, the urban–rural segregation persisted in many ways. Migrant workers’ chances of obtaining an urban hukou were very slim, regardless of their long urban residency or employment history. Moreover, access to many highly paid occupations is tied to an urban hukou that excludes rural residents, irrespective of their individual merits. That labor market segregation has been exacerbated by some exclusionary policies instituted by local governments that have essentially relegated rural migrants to physically demanding, low-skilled, and potentially hazardous jobs (Knight et al., 1999).

China’s urban–rural divide has led to significant income inequalities. A study using national data determined that rural migrants earned barely two-thirds of what their urban counterparts did (P. Li & Li, 2007), whereas another study found that migrants earned less than half of urban workers’ income (Zhang et al., 2016). Using multiple data sources, widely cited research by Xie and Zhou (2014) established that a substantial part of China’s large income inequality was due to regional disparities and the urban–rural gap. The urban–rural dichotomy is also reflected in welfare entitlements. Rural health insurance, old-age pensions, and poverty alleviation programs were largely non-existent until the 2000s. Once established, unfortunately, the rural benefits were significantly less than those of the urban programs. Furthermore, welfare entitlements in the cities were largely determined by hukou status, thus excluding most non-local people. Given the large population of rural migrants and their vulnerability, their marginalization in the urban welfare system has been a key subject of scholarly and policy research.

The urban social welfare system and unequal accessibility

The current social welfare system in urban China consists of contributory social insurance programs and tax-funded social assistance programs. The contributory insurance programs mainly include retirement insurance (old-age pensions), health insurance, unemployment insurance, and work injury insurance. Most urban social insurance programs are occupation-based, jointly contributed by employers and employees at a fixed rate. The social assistance category includes primarily the Urban Minimum Livelihood Guarantee Scheme (dibao), China’s last-resort social safety net, and a few other cash transfer schemes for disadvantaged households, all of which require a means test. The social insurance programs in the cities were initially intended to cover local employees, but in the past decade, various motives have prompted some local governments to include migrants in their urban systems (Shi, 2012; Yang, 2021).

One change has resulted from local governments struggling to solve the increasing labor shortage in the coastal regions, making the governments eager to secure economic growth by offering welfare entitlements to skilled migrant laborers, who have since come to possess bigger bargaining power in the labor market (Meng, 2020). Some local governments have introduced progressive policies that allow the conversion of rural non-local hukou into urban local hukou, the attainment of which grants migrants legitimate access to local welfare benefits. A second change has arisen because, in 2010, a new labor contract law stipulated that employers must enforce migrant workers’ participation in the urban social insurance programs, providing that an official employment contract had been signed. Third, local governments have been motivated to include migrants in their social insurance programs for a practical but often unspoken reason—because migrant workers are predominantly young and healthy, the bulk of their monthly premium contributions is retained in the urban risk pool of the host city, whereas their use of benefits is relatively small. In other words, migrants are net contributors to the urban social insurance system and essentially cross-subsidize the locals, and that cross-subsidization effect is in turn particularly valuable for rapidly aging coastal regions such as Guangdong (He, 2021).

However, such inclusionary policies appear to have yielded a limited outcome, because the participation rate among migrants remains much lower than expected. Several reasons are at play. First, many employers circumvent the labor regulations by offering migrant employees informal contracts that do not require contributing to the social insurance. In addition, many migrant workers appear to accept or even favor non-participation, because then they are able to save the often considerable sum of their premium contribution (Jiang et al., 2018). Second, the portability of social insurance in China remains limited. Migrant workers tend to perceive that the value of joining the urban system is low because most of the benefits—especially old-age pensions—are not portable back to their hometowns.

In brief, the urban social insurance system has loosened its barriers to the migrant population, and in many localities, migrants are actually welcome to join. Thus, the urban–rural divide seems to have eased in regard to people’s access to contributory welfare programs. In reality, however, voluntary opting-out has limited the coverage of those programs for migrants and particularly for those in the informal sector. Furthermore, migrants’ access to non-contributory social assistance programs remains highly restricted. Heavily built on the notion of local citizenship, those tax-funded programs still predominantly exclude migrants, thereby giving rise to another form of inequality.

The COVID-19 shock and its impact on migrants

The COVID-19 pandemic hit China heavily in early 2020. Official statistics reported a 6.8% decline in gross domestic product in the first quarter of the year, whereas the urban unemployment rate increased from 5.3% in January to 6.2% in February. Six percent of employees nationwide were forced to leave the labor market. At least 50 million rural migrant workers quit the urban labor market, for both voluntary and involuntary reasons (Zeng, 2020). Indeed, as a result of issues related to how the unemployment rate is calculated in China, some experts believe that the official figures were substantially underestimated (Barrett, 2020; Wang et al., 2021), with Che et al. (2020) estimating the job losses of migrants to be in the range of 40–70 million.

Migrants have been among the most vulnerable people in this pandemic. Because the outbreak in China occurred during the Spring Festival, when a majority of migrants had returned home, the stringent containment measures adopted in cities since late February 2020 greatly limited those migrants’ mobility (Wang et al., 2021). Their inability to return to their urban workplace, compounded by the high perceived health risk from traveling, caused a significant proportion of them to lose their jobs (An & Sun, 2021). Che et al.’s (2020) survey found that more than 90% of rural migrant workers in their sample were stranded after they returned home and had failed to find a job as of late February 2020. Another nationwide survey found that the unemployment rate of migrants was considerably higher than the national average, thus leaving them more vulnerable (Cai et al., 2021). Moreover, the sudden drop of foreign orders resulted in tremendous pressure for China’s export-led economy, which hires a large number of migrants. Many labor-intensive industries bore the brunt of the pandemic’s economic hardship, and layoffs became unavoidable. In January and February 2020 alone, 247,000 enterprises were shut down.²

Unemployment insurance was expected to provide emergency relief, but nationwide, the program covered only 17% of the migrant population.³ Its limited protections have once again become evident in this COVID-19 response (Lu et al., 2020). The tiny number of migrant participants reflects the limited relief that the unemployment insurance program has offered to the huge population of unemployed migrants. Even among the general labor force, a mere 8% of the individuals laid off by November 2020 were protected by unemployment insurance, and 86% of the laid-off workers did not receive any form of social assistance (Cai et al., 2021). Although the central government indicated flexibility in offering one-off cash transfers to laid-off migrants who were not enrolled in the unemployment insurance program, in reality, the provision was limited to very few cities (Che et al., 2020). Furthermore, even when the benefits were provided, only unemployed migrants working in the formal sector were eligible. Sadly, the vast informal sector has been under-covered by the nation’s social protection measures. A telling piece of evidence is that from February to April 2020, only 67,000 laid-off migrant workers had received the one-off living allowance (Xinhua News Agency, 2020b), in stark contrast to the huge size of the migrant population and the large-scale unemployment experienced during the pandemic (Che et al., 2020; Wang et al., 2021). Apparently, the local–migrant divide has constituted a major source of social inequality during the pandemic.

Theories and hypotheses

Conventional wisdom posits that exposure to major socioeconomic and health risks stimulates people’s demand for social protections and for increases in welfare spending (Han & Kwon, 2020; Rehm, 2009). Unemployment, catastrophic diseases, and chronic disability are typically among such risks. Furthermore, social policies themselves encompass many distinct welfare domains, meaning that individuals’ attitudes and preferences are multifaceted too. The problem of “who supports what” has been extensively examined in the recent social policy literature (He, Qian, et al., 2021; Q. Li & He, 2019). However, the major drawback of the existing literature is that “risks” are often narrowly conceived at the individual level, such as employment insecurity or sickness, and we know much less about people’s attitudinal reactions to grand-scale crises—such as the current COVID-19 pandemic—that challenge virtually all walks of life. When systemic crises occur, governmental policies can hardly be confined to the economic sphere or to social protections alone, but the “whole-of-the-nation” response often requires a full range of policy interventions. The larger menu of policies employed in a real-world crisis setting offers researchers a much more favorable opportunity to elicit individuals’ relative policy preferences.

Beramendi et al. (2015) drew a useful bi-dimensional distinction between social investment and social consumption. Social investment refers to social policies that are intended to boost the economy’s overall productivity, whereas social consumption focuses on compensating for current and immediate losses and tackling people’s urgent needs. Inspired by this conceptual framework, we contend that real-world policies in fact may not all neatly fit into the bi-dimensional framework. For example, some policy instruments, such as vouchers observed in the COVID-19 fight, essentially serve the purposes of both social investment and social consumption (Cao et al., 2020). Furthermore, in contrast to the situation in mature welfare states in which legal citizens are granted largely equal welfare entitlements, accessibility to welfare in China remains predominantly unequal, particularly between local residents and migrants. Hence, in order to elucidate the “who supports what” question against the backdrop of COVID-19 in China, one must carefully operationalize “who”—locals and migrants—and “what”—the range of policy instruments employed during the pandemic.

The existing literature maintains that major exogenous shocks or threats induce individuals to favor immediate, short-term social protections at the expense of gradual, longer-term social investments in such efforts as education and job-training programs (Häusermann et al., 2014; Thewissen & Rueda, 2017). The rationale behind this argument is that stronger shocks are likely to force individuals to focus heavily on immediate compensation in the form of cash transfers. This effect is particularly pronounced among those who suffer from or are more vulnerable to employment insecurity (Anderson & Pontusson, 2007). COVID-19 has created tremendous labor market turbulence and shaken the employment foundations of many industries, and we expect the preference for social consumption to be held by both locals and migrants who are more vulnerable in the labor market. We thus posited a general hypothesis:  

H₁: Ceteris paribus, individuals who occupy a weaker position in the labor market are more likely to prefer cash transfers over other policy options.

A sizable literature has elucidated that individuals’ welfare attitudes are shape by institutional setup (He, Ratigan, et al., 2021; Larsen, 2008) and labor market structures (He, 2021; Q. Li & He, 2019). Therefore, we contend that the divide between locals and migrants in China’s urban labor market and social welfare system can alter the aforementioned attitudinal pattern to some extent for two reasons. First, migrants are aware of their marginalized position in the local welfare system (particularly in government-funded social assistance) and thus tend to hold very low expectation for direct monetary benefits that are considered to be the privilege of locals. Second, the vast majority of Chinese migrants are economically motivated mobile laborers whose primary purpose for residing in cities is for employment and earning wages (Watson, 2009). Compared with locals who enjoy access to social assistance benefits and hence expect direct monetary relief, we posit that migrants tend to value employment promotion policies that can provide them with jobs and therefore with wages. As a result, we hypothesized:  

H₂: Ceteris paribus, migrants are less likely than locals to prefer cash transfers over employment promotion policies.

The first two hypotheses sought to understand the social policy preferences of individuals faced with the pandemic. We were also interested in investigating to what extent those policies have truly met their needs for the social protections that are supposed to mitigate the hardship caused by COVID-19. We speculated that individuals enrolled in the local social security system, both locals and migrants, should be more likely than the unenrolled to appreciate a local social policy response and hence offer relatively high evaluation of these policies. We therefore hypothesized:  

H₃: Ceteris paribus, individuals enrolled in the local social security system, both locals and migrants, are more likely than their unenrolled counterparts to offer high evaluation of local social policy responses.

By the same token, one can anticipate that individuals who have borne strong COVID-19 shocks and thus have been especially vulnerable in the labor market should be more likely than others to offer favorable evaluation of local government policies. However, scholarly as well as anecdotal evidence gleaned from the above-cited materials appears to suggest a rather large gap between soaring welfare demands and limited protections amid such a sweeping crisis. As a result, we posit that individuals who have suffered from greater COVID-19 shocks and who have been especially vulnerable in the labor market are less satisfied than others with local government policies. We consequently put forth the following hypotheses:  

H₄: Ceteris paribus, individuals who have suffered from greater COVID-19 shocks, both locals and migrants, are more likely than others to offer lower evaluation of the local social policy responses.

 

H₅: Ceteris paribus, individuals who are more vulnerable in the labor market, both locals and migrants, are more likely to offer lower evaluation of the local social policy responses.

Methodology

Study setting

This study was set in Guangdong Province. Located in southern coastal China, this province has been the principal economic powerhouse in the country’s economic growth since the 1980s. Its export-led industry has been supported by a highly vibrant labor-intensive sector, as well as thriving high-tech enterprises. We chose Guangdong for investigation for two reasons. First, it has been an extremely popular destination of rural-to-urban migration in China. Despite a decline in recent years, in 2018 there were still 45.4 million rural migrant workers residing in Guangdong and employed primarily in manufacturing, construction, and the service industries.⁴ Second, Guangdong’s export-led economy was badly hit by the COVID-19 pandemic, with a significant drop in orders. Nearly 30,000 enterprises in Guangdong were shut down in the first two months of the outbreak. Although the local government never officially released unemployment figures, anecdotal materials revealed that a rather formidable situation prevailed this prosperous province until May 2020.⁵

Local governments in Guangdong responded with several rounds of social policy interventions. As seen elsewhere, social insurance contributions were reduced for enterprises to alleviate their financial stress and encourage job retention. Small and medium enterprises were further offered with premium rebates. Its fiscal strength enabled the Guangdong government to introduce considerably more generous emergency relief. Poor unemployed individuals who were not covered by unemployment insurance were granted a one-off living allowance of 5000 yuan.⁶ Local authorities also increased the payment rate of cash transfer programs for vulnerable individuals and households. Various regular and one-off cash transfer payments amounted 2 billion yuan in Guangdong between early February and late March 2020.⁷

Design and sampling

Given the overarching concerns for social distancing at the time, contactless telephone surveys were naturally the most appropriate method of data collection amid the pandemic. The data collection work for this study was outsourced to a professional CATI (computer-assisted telephone interview) center in Guangdong that has rich experience in social surveys. The survey was undertaken from 1 to 10 July 2020, by a group of well-trained enumerators. Ethical clearance was obtained from the first author’s university.

We employed a two-stage sampling strategy to draw respondents. The primary sampling frame included 6 out of 21 prefectural cities of Guangdong: Guangzhou, Shenzhen, Zhuhai, Dongguan, Foshan, and Zhongshan. We did this stage through purposive selection because we wished to focus primarily on those major industrial hubs in the well-known Pearl River Delta region. In the second stage, we performed random-digit dialing in a preexisting database of local mobile numbers. Only adults were invited for interview. Those who had a mobile number registered in Guangdong but did not live there on a regular basis were excluded. Stringent quality control protocols were applied to ensure the quality of the data. A total of 25,814 successful dials were performed, and the final sample consisted of 2,160 individuals, thus yielding a response rate of 8.37%. We dropped 14 observations that had a substantive number of missing values. The sample included a small number of current university students (N = 189) who were not employed. We decided to drop those observations from the sample, too, because the primary focus of this study was on the working population.

When the survey was undertaken in early July, the Chinese economy was on the mend, but the recovery was far from a full return to normality. For instance, unemployment rate as released by the government remained high at 5.7% by 30 June (National Bureau of Statistics, 2020), but for the reasons explained earlier, the real situation of unemployment should be still rather grave. Anecdotal materials suggest slow recovery of business among small enterprises in Guangdong and considerable unemployment of migrant workers even by September 2020 despite their return to cities.⁸ Another piece of evidence was that the local authorities in Guangdong decided in July 2020 to extend the temporary reduction of social insurance contributions for enterprises, in order to offer more financial relief.⁹ Therefore, the data collected by this study in July 2020 still reflected public opinions amidst serious adverse impact on livelihood imposed by the pandemic.

Dependent variables

We listed five oft-used categories of policy interventions and invited respondents to indicate the most urgent one for the local government to undertake: (a) transferring cash ad hoc to middle- and low-income households (cash transfer), (b) increasing the benefits of existing social assistance programs (social assistance), (c) improving employment services and helping create more jobs (employment promotion), (d) providing financial and policy support to small- and medium-size enterprises in order to secure employment (enterprise support), and (e) distributing consumption vouchers to all residents (voucher). The question items were phrased in lay language so that even individuals with limited literacy were able to reasonably understand. Encompassing social assistance policies, active labor market interventions, and market-stimulating instruments in options, this single-choice item gauged an individual’s policy preference in the midst of a pandemic that had caused both labor market turbulence and household hardship. We subsequently invited respondents to rate the performance of the municipal government in its social policy response following the outbreak, on a scale of 0–10, with higher values representing more favorable evaluation (variable: policy evaluation). We focused on municipal governments because they were the principal authorities responsible for financing and delivering these policy programs

Explanatory variables

The key explanatory variable was hukou status (variable: local hukou) and whether the respondent was enrolled in the local social security system that granted him/her access to essential social protections (variable: local enrollment). We speculated that individuals who had suffered more from COVID-19—in general terms—might exhibit salient preferences for certain policies and levels of recognition of local government policies. We hence invited respondents to rate the impact of COVID-19 on their personal and professional lives on a −10 to +10 numerical scale, with lower values indicating more negative impacts (variable: COVID-19 shock). We then recoded the variable to 1 if the score was lower than −5, representing severe negative impacts, and to 0 if the score was −5 or above.

We created three binary variables to represent the participants’ employment status: (a) unemployed, (b) self-employed (including business owners), and (c) professional (such as teachers, doctors, engineers, and managerial professionals); non-professional employees were taken as the reference group in our statistical analysis. Self-employed and unemployed individuals were understood to occupy a weaker position than employed participants in the labor market. Vulnerability in the labor market was gauged not only by objective proxies, but also by subjective perceptions in relative terms. Respondents were asked “To what extent do you rate your comparative advantage against locals/migrants in the labor market?”. Options included (a) no advantage at all, (b) small disadvantage, (c) equal, (d) small advantage, (e) big advantage, and (f) no competition at all. We then combined category (a) and category (b) and created a binary variable “lack of competitiveness” to represent each participant’s labor market vulnerability. Migrants/locals perceiving equal or higher competitiveness in the labor market compared with that of locals/migrants, or perceiving no competition at all, were taken as the reference group in our statistical analysis.

Several important factors were included as control variables, and most of them were measured in binary scales: (a) gender (variable: women = 1; men = 0), (b) age (variable: age < 45 = 1; 45 or above = 0), (c) marital status (variable: married = 1; single, widowed, or divorced = 0), (d) college education (1 = had a bachelor’s degree or above, 0 = no degree), (e) working or ever worked¹⁰ in a secondary industry (variable: secondary industry; 1 = working/worked in manufacturing or construction industries, 0 = not working/worked in those industries), (f) with child care responsibility (variable: child care; responsible for looking after children = 1, not responsible for caring for a child = 0), and (g) with elderly care responsibility (variable: elderly care; responsible for looking after elders = 1, not responsible for elder care = 0). Dummy variables were created for each city in the primary sampling frame and were controlled for in multivariate analyses. Table 2 reports the descriptive statistics of key variables. The survey instrument included several question items that were primarily for descriptive purposes and were not used in the multivariate analyses. Those items are reported in the next section.

Table 2.

Descriptive statistics.

Variable	Obs.	Mean	S.D.	Min.	Max.
Employment promotion	1957	0.453	0.498	0	1
Enterprise support	1957	0.216	0.412	0	1
Cash transfer	1957	0.174	0.379	0	1
Social assistance	1957	0.049	0.215	0	1
Voucher	1957	0.086	0.280	0	1
Policy evaluation	1957	8.344	2.168	0	10
Local hukou	1957	0.483	0.500	0	1
Local enrollment	1957	0.714	0.452	0	1
Lack of competitiveness	1957	0.386	0.487	0	1
Unemployed	1957	0.175	0.380	0	1
Self-employed	1957	0.063	0.243	0	1
Professionals	1957	0.291	0.454	0	1
2nd industry	1957	0.392	0.488	0	1
COVID-19 shock	1957	0.295	0.456	0	1
Women	1957	0.428	0.495	0	1
Age <45 years	1957	0.855	0.352	0	1
College education	1957	0.567	0.496	0	1
Married	1957	0.676	0.468	0	1
Child care	1957	0.581	0.494	0	1
Elderly care	1957	0.725	0.447	0	1

Results

Descriptive results

Table 3 reports the descriptive results of key variables, disaggregated by local and migrant statuses. Compared with locals, migrants were significantly more likely to (a) be employed in a secondary industry, primarily in labor-intensive manufacturing and construction enterprises; (b) have attained a lower educational level; and (c) perceive themselves as having an inferior status in the labor market relative to that of locals. It should be noted that both groups contained a very similar percentage of professionals, suggesting the presence of white-collar migrants in our sample. We also observed that the migrants were not necessarily more likely than the locals were to be unemployed or out of the labor force. A plausible explanation could be that when the survey was being conducted in July 2020, a gradual resumption of work had gained progress and unemployment had been mitigated to a certain extent in the migrant population.

Table 3.

Descriptive results of key variables, sorted by local and migrant groups.

	Overall	Migrants (M)	Locals (L)	M − L
Observations	1957	1011	946	—
Impact of COVID-19
Spending cuts (%)	42.8	46.4	38.9	7.5**
Severely affected by COVID-19 (%)	29.5	31.6	27.3	4.3*
Job loss (%)	5.4	6.2	4.4	1.8*
Relying on savings after job loss (%)	44.7	48.0	41.2	6.8
Relying on family members after job loss (%)	35.4	32.8	38.2	−5.4
Relying on social assistance after job loss (%)	2.0	0.6	3.6	−3.0*
Social insurance enrollment
Old-age pension (%)	90.0	86.7	93.6	−6.9**
Health insurance (%)	97.7	96.9	98.7	−1.8**
Unemployment insurance (%)	83.2	78.9	87.8	−8.9**
Local enrollment (%)	71.4	64.1	79.2	−15.1**
Labor market characteristics
Lack of competitiveness (%)	38.6	42.9	34.0	8.9**
College education (%)	56.7	42.8	71.6	−28.7**
Self-employed (%)	6.3	6.9	5.6	1.3
Unemployed (%)	17.5	17.5	17.4	0.1
Professionals (%)	29.1	28.3	29.9	−1.6
2nd industry (%)	39.2	47.9	30.0	17.9**

Note: **p < 0.01, *p < 0.05 (one-tailed T-test).

In the full sample, 42.8% of the respondents had experienced spending cuts due to the pandemic, while 5.4% had been laid off after February 2020. Nevertheless, we believe that this unemployment figure did not accurately portray the actual situation, because many migrants who lost their jobs had already left Guangdong and therefore their part of the migrant population was not represented in our sample. Savings were the respondents’ principal source of income support after they had lost their job, and only 2% of the laid-off respondents relied on social assistance benefits. Approximately 30% of the respondents characterized the negative impact of COVID-19 on their personal and professional lives as “severe”. The migrants were clearly more vulnerable than the locals were, because they had a substantially higher likelihood of experiencing spending cuts but a lower chance of receiving social assistance benefits. The migrants also tended to be more severely hit by the pandemic itself.

While three major social insurance programs covered a large majority of the respondents in the sample, that coverage was comparatively low in the migrant group vis-à-vis the local group. As expected, the inclusionary reforms summarized above have extended the coverage of social insurance programs to the migrant population, but still, fewer than 80% of the migrants in the sample were insured by the unemployment insurance program. More than one-third of the migrants in our sample either did not participate in the social security system at all or were enrolled elsewhere, presumably in their hometowns. Given the limited portability of social security benefits in China, they had a very slim chance of receiving benefits of any sort in Guangdong.

A remarkable finding emerging from Table 2 was that nearly half of the respondents (45.3%) expressed a clear preference for employment promotion policies, whereas just 4.9% favored strengthening the existing social assistance programs. The second most preferred policy response was to secure employment through supporting small- and medium-size enterprises (21.6%), followed by ad hoc cash transfers (17.4%), and finally by consumption vouchers (8.6%). The shallow protections provided by the existing social assistance programs, such as dibao, are known among ordinary people who may well recognize their limits. Instead, ad hoc cash transfers that offer direct monetary relief may seem more attainable. The respondents’ strong overall preference for labor market interventions mirrors the domination of the economic individualist ideology held by most Chinese, which assumes that individuals should finance their welfare income through their own hard work (Cheng & Ngok, 2020). The respondents generally offered high evaluation of the local governmental policies, with the mean score being 8.34 out of 10 (standard deviation = 2.17).

Multivariate results

We ran multinomial logistic regression models to analyze the determinants of people’s social policy preferences, and the statistical results are reported in Table 4. Local hukou holders demonstrated a very significant preference for cash transfers as opposed to employment promotion programs. Migrants, however, understanding their marginalized position in the local welfare system, tended to show very low expectation for cash transfers and instead strongly favored job creation programs. Individuals who perceived that they lacked competitiveness (i.e., were vulnerable) in the labor market showed a significant preference for cash transfers, but self-employment status and unemployment status were not associated with such a preference at a statistically significant level. In brief, H₂ was well endorsed and H₁ was partially endorsed.

Table 4.

Multinomial logistic regression on policy preferences.

	Enterprise support		Cash transfer		Social assistance		Voucher
	Coef.	S.E.	Coef.	S.E.	Coef.	S.E.	Coef.	S.E.
Local hukou	0.058	0.136	0.424**	0.145	0.330	0.246	0.306	0.194
Local enrollment	−0.023	0.185	0.185	0.209	−0.251	0.314	−0.318	0.250
Lack of competitiveness	0.053	0.130	0.271*	0.134	0.123	0.229	0.211	0.179
COVID-19 shock	−0.030	0.135	0.123	0.140	0.158	0.237	−0.135	0.196
Self-employed	1.308**	0.244	0.151	0.321	−0.428	0.627	0.157	0.384
Professionals	0.291*	0.144	−0.018	0.161	−0.125	0.270	−0.344	0.210
Unemployed	−0.308	0.238	0.159	0.240	−0.765	0.401	−1.208**	0.345
2nd industry	−0.132	0.129	0.027	0.137	0.055	0.231	−0.414*	0.189
Women	−0.170	0.128	−0.097	0.135	0.267	0.227	−0.019	0.177
Age <45 years	0.229	0.192	0.113	0.194	0.132	0.341	−0.066	0.252
College education	0.449**	0.141	−0.265	0.145	−0.417	0.250	0.161	0.197
Married	0.118	0.205	−0.151	0.213	−0.032	0.334	0.272	0.282
Child care	0.154	0.194	0.170	0.201	−0.484	0.322	0.025	0.258
Elderly care	−0.173	0.149	0.098	0.163	0.028	0.261	−0.070	0.212
Constant	−1.565**	0.325	−1.700**	0.348	−2.141**	0.559	−1.473**	0.423
City dummy	Yes		Yes		Yes		Yes
Log likelihood	−2521.749
LR |${\chi ^2}$|(df)	159.11 (76)
Observations	1913

Note: “Employment promotion” as the base outcome; **p < 0.01, *p < 0.05. Those who support none of these policies are excluded (N = 44). Omitted variables as reference categories include (a) migrants, (b) not enrolled in local social security system, (c) perceiving equal or higher competitiveness or no competition with locals/migrants in labor market, (d) not severely affected by COVID-19, (e) employed in non-professional occupations, (f) working in service industry, (g) men, (h) aged 45 years or above, (i) without college education, (j) not in marriage, (k) having no children to look after, and (l) having no elders to look after.

Self-employed workers and professionals were associated with a greater expectation for policies that supported (small- and medium-sized) enterprises and job creation programs, thus reflecting those respondents’ self-interest motives. Either being unemployed or working in labor-intensive industries significantly reduced the appeal of consumption vouchers compared with employment-promotion initiatives. Whereas hukou, employment status, occupational status, and educational attainment all held explanatory powers to varying extents, most of the demographic characteristics, such as gender, age, and marital status, did not predict the dependent variables at a statistically significant level.

We used ordinary least squares (OLS) models to analyze the factors explaining people’s evaluation of the local social policy response. We also regressed the migrant and local groups separately to gain deeper insights into the formation of their respective attitudinal structures. Table 5 presents the regression results. Local respondents who were enrolled in the local social security system were significantly more likely than others to offer high evaluation of the current governmental policies. Among the migrants, their local enrollment status did not generate a powerful effect in the dependent variable, but the statistical association was actually marginally significant (p < 0.1). Therefore, H₃ was largely endorsed. As expected, a heavy COVID-19 shock on the part of individuals was associated with a very strong negative, rather than positive, evaluation of the local social policy response. This result offers a piece of strong, albeit indirect, evidence of the limited social protections provided during the pandemic. H₄ was therefore accepted.

Table 5.

OLS regression on evaluation of local government policies.

	Overall		Migrants		Locals
	Coef.	S.E.	Coef.	S.E.	Coef.	S.E.
Local hukou	−0.198	0.108	—		—
Local enrollment	0.377*	0.148	0.313	0.178	0.757**	0.287
Lack of competitiveness	−0.323**	0.102	−0.211	0.145	−0.463**	0.142
COVID-19 shock	−0.533**	0.107	−0.379*	0.150	−0.696**	0.151
Self-employed	−0.438*	0.208	−0.344	0.289	−0.608*	0.301
Professionals	−0.056	0.118	−0.101	0.173	−0.010	0.160
Unemployed	−0.369*	0.179	−0.573*	0.233	0.188	0.316
2nd industry	−0.137	0.103	0.053	0.145	−0.376*	0.147
Women	0.066	0.101	0.007	0.151	0.094	0.135
Age <45 years	0.280	0.145	0.458*	0.221	0.131	0.191
College education	−0.052	0.110	−0.160	0.156	0.102	0.157
Married	0.204	0.157	0.665**	0.233	−0.320	0.215
Child care	−0.110	0.148	−0.436	0.223	0.152	0.197
Elderly care	−0.248*	0.120	−0.321	0.172	−0.142	0.168
Constant	8.429**	0.249	8.236**	0.341	8.079**	0.414
City dummy	Yes		Yes		Yes
Observations	1957		1011		946
R2	0.053		0.052		0.082

Note: **p < 0.01, *p < 0.05. Omitted variables as reference categories include (a) migrants, (b) not enrolled in local social security system, (c) perceiving equal or higher competitiveness or no competition with locals/migrants in labor market, (d) not severely affected by COVID-19, (e) employed in non-professional occupations, (f) working in service industry, (g) men, (h) aged 45 years or above, (i) without college education, (j) not in marriage, (k) having no children to look after, and (l) having no elders to look after.

Attitudinal heterogeneity across the two groups was observed when we tested H₅. We initially expected that the individuals in weaker labor market positions would be more appreciative of social policy interventions, regardless of their hukou status. However, the statistical results portrayed a slightly different picture. In the migrant sub-sample, being unemployed powerfully predicted a lower evaluation of local government policies, partly reflecting a strong dissatisfaction of migrants with the employment assistance that was provided during the pandemic. Interestingly, the same independent variable registered a positive, albeit insignificant, relationship with the evaluation of local government policies in the sub-sample of locals. Moreover, self-perceived inferiority among the local respondents was strongly associated with a lower evaluation of local government policies. In short, H₅ was largely endorsed. Additionally, we found that married migrants offered a higher evaluation of local government policies than singletons did. That difference can be viewed as evidence of the protective function that marriage provided to vulnerable groups during the pandemic. Migrants, who were far away from their hometown and lacked social capital in Guangdong, were particularly vulnerable during the pandemic. Serving as a traditional safety net, families and spouses provided the migrants with important financial and emotional support to overcome those difficulties.

Discussion and conclusion

This study used a telephone survey to investigate public opinions about China’s social policy responses to the COVID-19 pandemic. The survey was conducted in Guangdong Province, China’s principal economic engine, which houses a large migrant population. Against the backdrop of the nation’s urban–rural dichotomy and unequal access to social welfare, we paid particular attention to the attitudinal dynamics of both local residents and migrants, who together represent a key source of social inequalities in contemporary China (Xie & Zhou, 2014). We were also interested in understanding individuals’ preferences for various social policies and their level of satisfaction with those policies.

A striking finding was that more than 40% of the respondents had experienced spending cuts, reflecting the heavy economic hardship shouldered by households. When laid off from work, an extremely small percentage of individuals had received social assistance benefits, and even fewer were migrants, who clearly suffered from heavier hardship. Respondents who had suffered more severely from COVID-19 and employment insecurity offered significantly lower evaluation of the local social policy response. Apparently, social assistance played a very residual role, and making the situation even worse was the fact that more than one-third of the migrant respondents were not enrolled into the local system. The fragmentation of the Chinese social security system and the limited portability of the benefit package have essentially excluded the vast number of vulnerable migrants from receiving necessary social protections. Preexisting inequalities have persisted and become entrenched, and little evidence indicates that social policy interventions have helped mitigate those inequalities when a crisis massively increased people’s vulnerability. Paradoxically, people’s evaluation of the social policy response to the pandemic was on a fairly high level, suggesting that the government’s social policy response was largely expected by citizens.

This study defied the long-held argument that major external shocks elevate people’s desire for immediate social consumption, at the expense of long-term social investment. Instead, our findings argue that social policy preferences are shaped by dominant welfare ideologies and ultimately by institutional arrangements. When the participants were invited to indicate their preferences for types of social policy response, the majority (nearly 70%) favored labor market interventions rather than social assistance. Their preference for strengthening the existing social assistance programs was unexpectedly low, whereas additional cash transfers received moderate support. Taken as a whole, these results mirror the residual nature of the Chinese welfare system, which seems to have framed prevalent welfare attitudes characterized by economic individualism in favor of employment. In contrast, people held remarkably low expectation for monetary benefits.¹¹ This attitudinal pattern was particularly salient among the migrant population. We argue that attitudinal feedback effects are at play.

Social policies produce interpretive effects by framing people’s cognitive processes of what they consider to be “normal” and “appropriate” with respect to welfare arrangements (Jordan, 2013; Pierson, 1993). Those effects, through policy narratives and education, engender generalized frames of reference in members of society about their deservingness and entitlements. The recent literature discovered that the fragmented residual social policies in China have indeed followed a norm-shaping function through which central values undergirding the social policies have become internalized into people’s cognitive systems (He, Ratigan, et al., 2021). This current study offers additional evidence that the division between locals and migrants in China’s urban welfare system has similarly penetrated attitudes to reinforce the demarcation within the local citizenship. The migrants’ policy preferences and satisfaction were not only determined by purely utilitarian-based reasons but also shaped by migrants’ cognitive interpretation of welfare entitlements. In other words, institutional inequalities within the Chinese social welfare system have gradually penetrated to self-justify the differential distribution of entitlements between locals and migrants.

Zooming out from plain statistics, this study draws two important conclusions. First, whereas many recent studies have optimistically assumed that the urban–rural divide is diminishing in the Chinese social welfare system, we argue that the divide has been eased in contributory social insurance programs only, while the tax-funded social assistance system remains very exclusionary. This contrast was not highlighted until COVID-19 struck China and caused mounting needs for social assistance. Our study, together with other recent ones, has revealed the limited provision of social assistance in China during the pandemic. Moreover, this study has elucidated the fact that the preexisting inequalities in the Chinese welfare system have created social policy preferences that differ between local residents and migrants. We argue that this attitudinal heterogeneity not only has resulted from utilitarian self-interest motives alone but also has been profoundly shaped by interpretive effects that internalize demarcated welfare institutions into people’s welfare expectation. The “who supports what” question in regard to social policy preferences in a residual and divided welfare system is configured by people’s expectation of what kind of social policy benefits they are able to access.

Acknowledgements

The authors thank Daniel Béland for useful comments.

Funding

This project was funded by the Department of Asian and Policy Studies, The Education University of Hong Kong.

Ethical approval

Ethical clearance was obtained from the Human Research Ethics Committee of The Education University of Hong Kong.

Conflict of interest

None

Footnotes

1

“Xiamen expanding the coverage of unemployment insurance”, China News, 15 July 2020, retrieved from http://www.fj.chinanews.com/news/fj_zxyc/2020/2020-07-15/471103.html (accessed on 2 April 2021).

2

“247,000 enterprises shutdown, foreign ordered dropped by 80%: enterprise forced to lay-off?” Sina, 2 April 2020, retrieved from http://jiaju.sina.cn/news/20200402/6651300020463403579.shtml (accessed on April 9 2021).

3

Ministry of Human Resources and Social Security, 2017 Human Resources and Social Security Statistical Reports, retrieved from http://www.mohrss.gov.cn/SYrlzyhshbzb/zwgk/szrs/tjgb/201805/t20180521_294287.html (accessed on March 22 2021).

4

National Bureau of Statistics, 2018 Migrant Worker Monitor, retrieved from http://www.stats.gov.cn/tjsj/zxfb/201904/t20190429_1662268.html (accessed on January 2 2020).

5

“Coronavirus: China’s economic woes could be worse than thought as legions of migrant workers return home,” South China Morning Post, April 29 2020, retrieved from https://www.scmp.com/economy/china-economy/article/3081953/coronavirus-chinas-economic-woes-could-be-worse-thought (accessed on January 29 2021); “Coronavirus has hit China’s migrant workers harder than SARS and the financial crisis, but worse yet to come,” South China Morning Post, 25 May 2020, retrieved from https://www.scmp.com/economy/china-economy/article/3085904/coronavirus-has-hit-chinas-migrant-workers-harder-sars-and (accessed on January 29 2021).

6

Guangdong Provincial Government, Notice on policies regarding further securing and promoting employment, 20 February 2020, retrieved from http://www.gd.gov.cn/zwgk/wjk/qbwj/yf/content/post_2903650.html (accessed on 2 April 2021).

7

State Council, “Guangdong hands out 2 billion yuan to vulnerable people affected by Covid-19”, March 24 2020, retrieved from http://www.gov.cn/xinwen/2020-03/24/content_5494857.htm (accessed on August 27 2021).

8

“China’s two-speed economic recovery leaves migrant workers and small businesses lagging behind”, South China Morning Post, 26 September 2020, retrieved from https://www.scmp.com/economy/china-economy/article/3102861/chinas-two-speed-economic-recovery-leaves-migrant-workers-and (accessed on August 28 2021).

9

Guangdong Provincial Government, “Reduction of social insurance contribution extended”, 10 July 2020, retrieved from http://www.gd.gov.cn/gdywdt/zwzt/yqfk/gdzxd/content/post_3041672.html (accessed on August 26 2021).

10

If unemployed or out of the labor force at the time of the interview.

11

Cross-country evidence suggests that national governments elsewhere clearly resorted to social consumption more than social investment strategies in their social policy response to the COVID-19. According to the estimation of Gentilini et al. (2020), 87 countries all over the world had introduced 193 cash transfer programs until mid-April 2020. In comparison, a total of 78 active labor market policy programs were noted.