Introduction
Since the late 1950s, the household registration (hukou 户口) system in China has been an institution that regulates urban-to-rural migration and confines migrant workers’ rights to access public benefits available for local residents. Hukou is a major institution that has perpetuated social inequality and stratification in China.Footnote 1 Although several reforms of the hukou system at the local level have taken place since the 1980s, these reforms did not effectively ameliorate the fundamentally unequal system, and none has been at the scale of the nationwide hukou reform in 2014.Footnote 2 This study aims to describe the patterns of subnational variations in hukou reform policy design through collecting and coding policy announcements on the 2014 hukou reform from 231 cities, and we further examine economic, political and sociocultural drivers of variations in hukou reform approaches.
This study contributes to prior literature in four ways. First, we systematically coded migration policies in 231 cities to depict patterns stemming from the 2014 hukou reform, which has been described as the greatest reform in scale in the past two decades. The reform is crucial to the new social and spatial landscape in China because it affects hundreds of millions of migrants’ rights and socio-economic opportunities. Second, our investigation into drivers of migration policies in China expands the understanding of migration policy determinants to a non-democratic context, whereas prior literature has predominantly focused on the Western and democratic contexts. We exploited rich multidisciplinary data combining standard city statistics, the official migrant survey of more than 100,000 migrants every year, night-time lighting data from satellites, hand-collected local politicians’ bibliographies, and applied machine learning to mass media to derive data on the sociocultural environment that migrants encounter. These efforts demystify how migration policies take shape in a certain direction. Third, prior studies on drivers of policymaking in China have not covered the area of domestic migration policies. This study can further contribute to understanding policymaking mechanisms in China, a decentralized authoritarian regime. Fourth, identifying intervention patterns of a major institution to Chinese socio-economic life – hukou policy – offers a foundation for subsequent policy-evaluation studies to utilize a quasi-experimental design in assessing the causal consequences of hukou reform.
The hukou system and hukou reforms in China
All Chinese individuals are assigned an agricultural or non-agricultural hukou in a county/city, which is often referred to as rural or urban hukou. The hukou system was established in 1958 for population and resource control. Hukou status divided people's rights and privileges in domains of consumption, access to education, employment, housing, social insurance and social safety nets, with far more generous benefits attached to urban hukou.Footnote 3 Originally, one's hukou was inherited from one's mother, but since 1998 individuals have been allowed to choose between either parent's hukou if they were born after 1998. Hukou status can only be altered through limited mechanisms.Footnote 4 Even after the economic reform in 1978, when rural-to-urban migration was gradually permitted to meet the labour demand, the majority of peasant workers were still not entitled to urban hukou or benefits.Footnote 5 Because the hukou system confines migrants’ economic and social positions, the selection and integration policy features resemble those of international migration.Footnote 6 Hukou is effectively an “internal passport” affecting people moving within or across provincial boundaries in analogy to moving across international boundaries.Footnote 7 As such, this article is based on theoretical and empirical literature on both international and internal migration to inform our inquiry into patterns and determinants of migration policies.
Hukou reforms had taken place since the mid-1980s in various shapes to boost the labour supply, while they made little progress towards promoting equal access to welfare.Footnote 8 Urban hukou have remained difficult to attain.Footnote 9 In 2014, China issued the State Council “Opinion on hukou reform” (or “Advice on further promoting the reform of the hukou system”) in response to the national goals of supporting urbanization and building a prosperous society (xiaokangshehui 小康社会) by 2020, as set out by the 18th National Congress of the Chinese Communist Party (CCP). The 2014 hukou reform (hereafter, the reform) aimed to end the distinction between agricultural and non-agricultural registration and recognize the residential status of migrants through replacing the old temporary residential permit (zanzhuzheng 暂住证) with the new residential permit (juzhuzheng 居住证).Footnote 10 It is deemed a “major step towards the structural integration of the urban–rural dichotomy.”Footnote 11
The reform adopted a tiered approach to accommodate the situations of different localities, in which “(1) hukou restrictions will be completely lifted in towns and small cities with below 0.5 million population, (2) hukou restrictions will be gradually lifted in cities with 0.5 to 1 million population, (3) hukou restrictions will be reasonably lifted in large cities with 1–3 million population, (4) hukou requirements will be reasonably set for large cities with 3–5 million population in the districts, and (5) the population of megacities with over 5 million population will be strictly controlled.”Footnote 12
An emerging body of research on the reform has described it as unprecedented in magnitude.Footnote 13 Following the national reform agenda, provincial and municipal governments have released policy plans to implement the reform. The responsibility for formulating hukou policies has devolved to local governments.Footnote 14 Local policymakers can set up “hidden” rules or barriers to protect local residents and to maintain political support from pivotal groups.Footnote 15 The reform encompasses policies aimed at selecting and integrating migrants into a destination. Policy documents released from prefecture-level cities (hereafter, city-level) contain the most detailed policy plans, so we focus on depicting patterns of the reform at the city level. There are many case studies on the reform,Footnote 16 and quantitative studies mostly could not precisely measure local variations of the reform.Footnote 17 Zhang et al. measured and quantified variations in selection policies across more than 100 cities, yet they exclude integration policies.Footnote 18 Our study is the first to cover both selection and integration policy domains across 231 cities that have published hukou policy documents.
Determinants of migration policies
Prior research on determinants of international migration policies has mostly been grounded in European, Anglo-Saxon and democratic contexts.Footnote 19 The major debates of these studies concern whether economic interests (e.g. labour market competition among migrants and locals), ideological/cultural factors (e.g. ethnocentrism, racism, nationalism and cultural prejudice) or political institutions (e.g. political orientation of government, institutional veto points and the degree of electoral competition) is a stronger determinant of migration policymaking. Some studies have found support for ideological and cultural factors as the major drivers,Footnote 20 whereas other studies have found support for politicalFootnote 21 and economic factors playing a major role. With regard to determinants of internal migration policies, studies in this area are scarce because limited subnational variations exist in internal migration policies, and the only exceptions can be found in the US context. In the United States, harsh immigration enforcement policies and migrant labour regulation policies were mostly driven by political and ideological factors, and economic and demographic factors show inconsistent predictive power.Footnote 22
In summary, prior research was mostly grounded in democratic contexts, and evidence from non-democratic contexts is rare. In democratic contexts, legislators represent public opinions, and legislators channel the public's economic and cultural concerns in policy debates and legislative voting. In China's autocratic, one-party regime, these mechanisms are not applicable. Under an authoritarian circumstance, policy formulation is less of a joint outcome achieved through negotiations and compromises by competing parties but more of a state-dominated process. Autocratic migration policies are often designed to meet the state's economic needs and plans. Our investigation into the determinants of hukou policy in China can extend the understanding of drivers of migration policies beyond a democratic context and deepen extant theories on forces influencing migration policymaking.
Policymaking in China
China has been described as a decentralized authoritarian state, with the central government setting the grand policy direction and local governments – especially at the sub-provincial level – assuming the duty of adapting the grand direction to local circumstance.Footnote 23 The contradiction of having subnational policy variations under an authoritarian regime has drawn academic interest in examining the forces driving subnational policymaking in China. Economic concerns, such as the level of economic developmentFootnote 24 and fiscal resources,Footnote 25 have been found to play an important role in explaining subnational policy variation. Other forces driving local policymaking include ensuring social stability,Footnote 26 the governance styleFootnote 27 and local government officials’ concerns over their political careers.Footnote 28 Although studies examining the determinants of subnational policymaking in China have covered a diverse range of domains that include social policy,Footnote 29 agrarian policy,Footnote 30 economic policyFootnote 31 and environmental policy,Footnote 32 migration policy is an untapped area in studies concerning policymaking in China. Our investigation into the determinants of internal migration policies can enrich the literature on policymaking in China.
Literature Review
Factors influencing migration policies and hypotheses
Our theoretical framework on determinants of migration policymaking in China is in Figure 1. We posited that approaches to migration policies in China are driven by economic, political and sociocultural factors,Footnote 33 and we elaborate each factor below.
Figure 1. Conceptual Model for Migration Policy Determinants in China
Economic factors
Economic considerations are related to migration policies through several channels. First, migration policies can be part of an economic development plan. As far as economic expansion and growth are concerned, localities may adopt relatively lenient selection and integration policy approaches to attract and incorporate migrants into the local labour market.Footnote 34 Second, stringent migration policies can be established to curb the excess inflows of migrants in economically prosperous regions. Empirical evidence grounded in large cities in China supports the second channel. Specifically, case studies examining hukou policies in large cities (Shanghai, Shenzhen, Chongqing and Chengdu) have shown that these municipalities adopted more stringent requirements (e.g. a higher education or capital level, having a work contract) for migrants to acquire local hukou as a way to advance human and financial capital.Footnote 35 Meanwhile, the surge of migrants also created a financial burden on welfare and public service systems, and many localities chose to exclude migrants from entitlements that local residents receive, resulting in rights deprivation and social exclusion of migrants.Footnote 36
• H1.1: Cities intending to maintain economic growth are more likely to adopt lenient requirements in selecting migrants to receive local hukou to fulfill the local labour demand.
• H1.2: Cities with higher levels of economic development are more likely to adopt stricter requirements in selecting eligible migrants to acquire local hukou and are more reluctant to provide migrants with the welfare and services available to locals.
• H1.3: Cities with high levels of economic hardship (e.g. high unemployment) are more likely to adopt stringent migration policies,Footnote 37 whereas cities not in economic distress are more likely to enact pro-migration policies.Footnote 38
Labour skill composition and demand is another factor that shapes the selection of a migration policy approach. In the United States, localities heavily represented by industries relying on migrant workforce (i.e. agriculture, mining and construction) are less likely to pursue stringent immigration enforcement policies.Footnote 39 In authoritarian regimes, migration policies are tools designed to redistribute low-skilled workers to meet the domestic demand for labour.Footnote 40 Local governments in China can flexibly adjust migration policies to favour rural migrant labour according to evolving local needs. Government officials permitting the massive inflow of low-skilled migrants while simultaneously rejecting their rights and naturalization is possible because officials have limited incentive to assimilate migrant labour systematically into the locality.Footnote 41
• H1.4: A high demand for low-skilled workers is related to lenient selection policies and stringent integration policies.
Political factors
Different from a democratic context where right-wing parties generally support conservative migration policies and left-wing parties often adopt liberal migration policies,Footnote 42 in China politicians’ ideological stances converge more than they diverge.Footnote 43 Politicians in China are not subject to electoral checks, yet their policy decisions can be driven by incentives for career advancement and the top-down control of superior governments. Politicians have to perform according to promotional expectations (e.g. attaining policy objectives or meeting evaluation metrics) to advance their careers to larger cities or the national level.Footnote 44 When the top-down control from the upper-level government is stronger (e.g. setting specific targets), local leaderships face stronger pressure to comply and perform.Footnote 45
Additionally, government leaders’ characteristics (e.g. background and demographics) also drive policy decisions.Footnote 46 Specifically, prior literature has suggested that politicians’ policy selections and incentives to perform are associated with their tenure in their positions, affinity to the locality, and prior education and career trajectories (e.g. loyalty to the CCP, educational background and administrative experience). First, politicians often set a higher performance target at the beginning of their tenureFootnote 47 or make extra efforts to perform or please their constituencies at the end of their first term to ensure reappointment or to win re-election.Footnote 48 However, during their second or last term, politicians channel less funding to their locality.Footnote 49 Second, politicians having stronger private relationships in a locality or who work in their home region are more generous in investing resources there, a phenomenon termed “home bias.”Footnote 50 Third, politicians who have demonstrated stronger loyalty to the CCPFootnote 51 or who embarked upon their civil service careers earlierFootnote 52 have a higher prospect of receiving promotion, which leads to their higher motivation to achieve policy objectives. Fourth, politicians’ disciplinary training background and prior work experiences within the government influence their competency in administration and policy preferences.Footnote 53
• H2.1: City governments facing superior governments with stronger top-down control forces are more like to adopt policies consistent with the central government's policy direction.
• H2.2: City Communist Party secretaries (CPSs), the main leaders of municipal affairs under the CCP, will adopt more policy reform measures during the earlier or later years of their first terms than those in midst of their first term or in their second term would.
• H2.3: CPSs who work in their home regions will adopt more generous integration measures if their favouritism extends beyond local residents to migrant populations.
• H2.4: CPSs who have longer histories as CCP members and who have educational background and work experience in public administration are more likely to select policies consistent with the central government's policy direction.
Sociocultural factors
In democratic societies, a distaste for multiculturalism predicts lower support for integration policy measures, and some studies suggest that social attitudes are even stronger determinants of migration policy approaches than economic factors are.Footnote 54 In China, the dichotomy of “locals” and “outsiders” is deeply ingrained in the country's cultural context. Discrimination can be found between groups from different origins.Footnote 55 Regionalism in the labour market can limit migrant workers’ opportunities,Footnote 56 while fluency in the local dialect can enhance wage outcomes.Footnote 57 Different from the authoritarian context, however, policymaking in China is not a direct transfer of public opinion. Public opinion and sociocultural contexts indirectly affect migration policies to the extent that these factors shape the views of policymakers.
• H3: The public's higher receptivity to migrants and fewer migrant-related conflicts are associated with adopting more lenient selection and integration policies.
Methods
In the first stage, we described patterns of hukou reform through collecting policy documents available at the city level, coding policy documents, and describing patterns of reform using descriptive statistics and cluster analysis. In the second stage, we examined determinants of hukou reform policy approaches through collecting determinant indicators and employing regression models.
Data: construction of data set on city-level hukou reform policies
Out of 337 prefecture-level cities,Footnote 58 231 cities have retrievable hukou-reform policy documents,Footnote 59 and we constructed a policy data set that captures variations of reform policies. Prior studies that systematically compiled and coded international migration policies have offered rigorous guidelines for collecting migration policy data in the Chinese context.Footnote 60 To ensure the constructed policy data are valid, reliable and replicable, we maximized the transparency of decisions and guidelines in the policy coding through following the DEMIG policy-coding guideline.
Policy-coding process
The purpose of policy coding is to transform hukou policy reform content into an analysable format. Each policy article represents one policy change. The number of articles in each policy document ranges between 4 in Huzhou and 57 in Kunming, with a mean of 29.9. The content of each policy article was first coded into four variables: policy area (what?), policy tool (how?), migrant category (who?), and migrant origin (from where?). Then we assessed each policy article from the perspective of restrictiveness and magnitude.Footnote 61
Policy area refers to the policy domain of an article, including selection, conversion and integration. Policy tool indicates the specific policy measure, ranging from a points system under selection to the issuance of residential permits under conversion (see Appendix Table 2). Migrant category is defined by the migrant type that an article targets, such as college/university graduate, migrants’ spouse or real estate investor. Migrant origin specifies the place from which the migrants move (e.g. rural locals for within-province rural migrants, non-rural outsiders for cross-province urban migrants). Restrictiveness of a policy change assesses whether policy changes became less restrictive (value = −1) or more restrictive (value = 1). The magnitude of change captures the degree of change and provides the weight for restrictiveness of each policy article. The magnitude of change is measured by the extent of coverage and radicalness, with a range between 1 (fine-tuning change) and 4 (major change).Footnote 62 Detailed coding criteria for restrictiveness and magnitude are presented in Appendix 3.
Aggregation of policy scores
Finally, we constructed the policy scores for each article through multiplying the restrictiveness by the magnitude of change. The policy score ranges from −4 to 4, with smaller values indicating more lenient changes and larger values indicating more stringent changes.Footnote 63 We constructed each city's policy scores in the policy domains of selection, conversion and integration. The list of indicators is shown in Figure 2 (definitions are detailed in Appendix Table 2). Selection policies include policy tools that specifically aim at attracting high-skilled migrants (e.g. talent scheme and investment scheme), policies that do not target high-skilled migrants (available for non-high-skilled migrants) and policies that allow family unification. Conversion policies refer to policies that allow migrants to convert their status to local hukou and include tools that issue local residential permits to migrants and convert agricultural hukou to non-agricultural hukou (nongzhuanfei 农转非). Integration policies focus on policy measures that allow migrants to receive benefits similar to local residents and protect their rights associated with rural identity. The benefits include access to education and public services, and rural rights entail the protection of rural property and an adaptation period for rural migrants to adjust to urban fertility regulation.
Figure 2. Conceptual Framework of Policy Indicators
Measures and empirical strategies
Patterns of hukou reform
To describe policy variations across cities, we first summarized policy scores by city size using descriptive statistics and then we performed hierarchical cluster analysis. An agglomerative clustering procedure was chosen due to its data-driven, bottom-up nature, as opposed to a divisive (top-down) procedure.Footnote 64 As one of the most commonly applied standards, Euclidean distance was adopted as the measurement of dissimilarities between clusters and the Ward method as the principle of linkage to minimize the within-cluster variance.
The objective of cluster analysis is to summarize policy approaches adopted across cities in a parsimonious manner. We specifically conducted three sets of analysis to synthesize policy approaches: (1) across all policy areas, (2) selective policy approaches, and (3) integrative policy approaches (including both conversion and integration policies). We standardized the policy score for each indicator for cluster analysis. The final choice of cluster number was decided by three statistical indices, the Silhouette score, the Calinski–Harabasz score, and the Davies–Bouldin score, as well as the conceptual meaning of each cluster.
Determinants of hukou reform policies
Economic indicators include gross domestic product (GDP) growth, the level of economic development (as measured by night-time lighting data collected by satellites to avoid multicollinearity with GDP growth rates), the proportion of low-skilled workforce and unemployment rates. The GDP growth rates covering 2010–2013 were calculated based on GDP data in China City Statistical Yearbooks. The density of night-time lighting detected by satellites covering 2010–2013 was derived from the US Air Force's Defense Meteorological Satellite Program/Operational Linescan System at the National Oceanic and Atmospheric Administration. We took average values of the above economic indicators over 2010–2013 to address the fluctuation in annual data. The proportion of low-skilled workforce was derived from the 2008 Economic Census. The unemployment rate was collected from the China Statistical Yearbook for Regional Economy 2013.
Political factors cover (1) the extent of top-down control, as measured by whether the provincial government set specific policy targets (e.g. allowing a specific number/proportion of migrants to obtain local hukou) and specified the division of labour among bureaus in policy implementation in provincial policy announcements; and (2) municipal CPS characteristics, including the CPS's tenure (0–20, 21–40, 41–60 and 60+ months given the five-year term length) at the time the city released the reform document, whether the CPS was born in the province where they work, the CPS's duration as a CCP member, whether the CPS's undergraduate degree was in social sciences (i.e. business management and economics) and whether the CPS had ever served as a secretary in a government bureau. To collect CPSs’ characteristics, we relied on a website (www.hotelaah.com/) with all the historical records of leading government officials and then we used Baidu to search for CPSs’ biographies to code their characteristics.
To capture the sociocultural context faced by migrants, we constructed two measures: the intensity of labour disputes covered in provincial media and the social integration levels experienced by migrants. First, we collected media reports on migrant workers and measured the frequency of local employer–migrant employee disputes within text data. The provincial daily newspaper series was selected as our data source because it is the primary source of a newspaper in each province and the only newspaper available across all provinces in China.Footnote 65 This enabled us to derive comparable data on the media environment.Footnote 66 We used machine-learning techniques to code 3,214 news articles between 2010 and 2014, labelling articles as one if they mentioned infringement of labour rights (e.g. overwork, wage arrears and safety violations) and zero otherwise. Eventually, 3,214 binary labels were disaggregated by province and year, resulting in a continuous variable falling between zero and one, with a higher value representing greater labour dispute intensity. Appendix 4 provides a detailed description of the variable construction process. Second, we derived the cities’ social integration levels for migrants by pooling data from the official China Migrant Dynamic Survey (CMDS; 2011–2013) conducted annually by the National Health Commission. Each wave includes 128,000 to 198,795 internal migrants out of 326 to 344 cities, providing sufficient sample size to depict migrants’ experiences. The CMDS collected responses to five items (e.g. I would love to fit in and become one of them; I think local residents are willing to accept me as one of them). Each question had four response categories (1 – completely disagree to 4 – completely agree). We calculated the standardized mean scores for responses to the five items. The internal consistency level was optimal (Cronbach's α = 0.77).
We controlled for demographic and geographical indicators as well as CPSs’ characteristics. We collected demographic indicators from the 2010 census, including total population, percentage of migrants from other provinces, percentage of ethnic minorities (non-Han population), percentage of elderly (65 or older) and the number of college-degree holders. As for geographical indicators, we controlled the geographical region (east, central and west) for each city. We also controlled for gender, education levels, age of CPSs at the time of the hukou reform, the proportion of welfare expenditures (in social security and employment),Footnote 67 the reliance on land conveyance incomeFootnote 68 and hukou policy approaches from an economically neighbouring city.Footnote 69 Descriptive statistics of all variables are in Appendix Table 5.
To study the determinants of hukou reform policies, we used cluster memberships as the outcome to run multinomial models. The above-mentioned variables were included as independent and control variables to predict hukou reform approaches.
Results
Patterns of hukou reform policies
In Table 1, we present policy scores by city size.Footnote 70 Based on the reform policy, small cities are defined as cities with fewer than 500,000 residents within an urban area (chengqu 城区) in a city, medium cities as 500,000–1 million, big cities as 1–5 million, and megacities as more than 5 million. The summary statistics show a clear distinction on policies adopted in megacities and non-megacities, with megacities adopting less lenient selection policies. However, the distinction among non-megacities (i.e. small, medium and big cities) is not salient. Regarding selection policies, megacities adopt more lenient talent schemes, less lenient investment and family reunification schemes and more stringent point systems than smaller cities. With regard to conversion policies, we find that most cities provided opportunities for migrants to receive a resident permit or had unified urban–rural systems, with small cities being the most lenient in conversion measures. Regarding integration policies, the total integration policy scores were less lenient in megacities than in smaller cities. Specifically, we found that megacities granted less lenient access to education, employment services, social insurance and social assistance than smaller cities did. Regarding rural rights protection, we found that megacities are more reluctant to grant migrants a fertility-control adaptation period.
Table 1. Policy Scores by City Size
Next, we conducted cluster analysis to classify hukou-reform policy measures. We selected three cluster solutions for overall, selection and integrative policy approaches. The list of cities in each cluster is presented in Appendix Table 7. Table 2 shows the clustering results for each policy approach and their average standardized policy scores.Footnote 71 Regarding overall policies, most cities (n = 108; 46%) adopted more selective policies with some integrative measures, followed by 94 cities (41%) that adopted the most welcoming policies, offering lenient selective and integration policies. The remaining 29 cities (13%) comprised the smallest group, which provided lenient selection measures yet the least lenient integrative policies. Compared to the average likelihood of being in each cluster, megacities are more likely to adopt the least integrative policies and the least likely to adopt welcoming policies. Small cities, on the other hand, are more likely to adopt welcoming policies and less likely to adopt selective policies.
Table 2. Standardized Policy Scores by Policy Clusters
With regard to selection policies, the majority of cities (n = 183; 79%) adopted the least lenient (relatively stringent) policy measures towards all types of migrants, followed by 36 cities (16%) that adopted welcoming policies towards non-high-skilled migrants. These 36 cities were the least likely to adopt the points system. Instead, they offered lenient policies towards investors and broader migrant groups. The least represented group is the 12 cities (5%) that provided easier access to local hukou for high-skilled and high-income migrants through talent and investment schemes. They are also more likely to adopt a stringent point system. Overall, we found that megacities are overrepresented in the cluster that welcomes high-skilled workers and underrepresented in cities that welcome non-high-skilled workers.
Regarding integration policy approaches, the majority of cities (n = 173; 75%) adopted integrative social welfare policies, granting migrants’ children more lenient access to education and extending access to social assistance programs to migrant families. Another 30 cities (13%) adopted the least integrative policies. The remaining 28 cities (12%) adopted integrative labour market policies by providing migrants with less generous access to social assistance and education yet providing access to welfare measures that encourage labour force participation. These pro-labour market measures included granting access to employment services, employer-provided insurance and housing. We found that megacities were overrepresented in the least integrative cluster and underrepresented in the integrative social welfare cluster.
The third panel of Table 2 summarizes the distribution of 231 cities across their selection and integration policy approaches. The majority (63.2%) of cities adopted selective and integrated social welfare measures. The next represented group (9.5%) adopted policies welcoming non-high-skilled migrants while also providing integrative social welfare measures. The third commonly seen combination occurred in 8.2 per cent of cities that adopted selective policies but did not provide integrative policies, followed by 7.8 per cent of cities that adopted selective policies and integrative labour market policies. The rest of the policy combinations were represented by no more than 3 per cent of cities.
Determinants of hukou reform policies
Table 3 present results from multinomial regression models using the most represented cluster as the reference group and examining determinants of hukou reform policies. With respect to economic determinants, we found that more lenient selection policies towards high-skilled workers were related to higher levels of economic growth and economic development prior to hukou reform (Model 2.2), whereas lenient selection policies towards non-high-skilled workers were related to lower economic development levels prior to the reform (Model 2.1). Cities facing higher unemployment rates were more likely to adopt the least integrative policies (Model 3.2).Footnote 72 With regard to political factors, cities facing stronger top-down control, as measured by clear division of implementation duties, were more likely to adopt friendly policies towards migrants. CPSs in their second term of tenure (61+ months), who faced a lower prospect of promotion, were less likely to adopt progressive policy approaches that deviated from the norm or reference groups (integrative social welfare policies; Models 1.2 and 3.2). CPSs in the earlier phase of their tenure were more likely to adopt welcoming policies (Model 1.1). CPSs who were from the provinces where they served and those who had a social science background were less likely to adopt selective policies in general (Model 1.1). CPSs’ membership history and experiences as a secretary in the government were not associated with policy approach selection. In terms of sociocultural factors, we found that cities in provinces with more labour disputes were less likely to adopt integrative social welfare policies and were more likely to adopt integrative labour market or least integrative policies (Models 3.1–3.2). Migrants’ perceived integration levels were not associated with policy adoption. We summarize our findings in relation to our hypotheses in Appendix Table 9. Appendix 10 reports several sensitivity analyses to account for additional factors and to test the robustness of our findings on determinants of hukou reform policies, such as controlling for pre-existing integration levels, using the number of dibao 低保 (welfare) recipients as an alternative indicator for economic hardship, and controlling for integration policies in selection analyses and controlling for selection policies in integration analyses.
Table 3. Multinomial Regression Results on Determinants of Hukou Reform Policies
Notes: Each two columns are based on a multinomial regression model. Each model controls for population, % of migrants, ethnic minorities and elderly, the number of college-degree holders, region, proportion of welfare expenditure, reliance on land conveyance income, and CPS's gender, education, age, and hukou reform policies from neighboring regions. Coefficients are presented, and standard errors clustered by province are in brackets.
*p < 0.05; **p < 0.01; ***p < 0.001.
Discussion and Conclusion
This study investigates the patterns of the largest nationwide hukou policy reform in the past two decades across 231 cities in China and the economic, political and sociocultural forces that drive the policy directions. We found that the majority of cities (63.2%) adopted less lenient selective policies in screening migrants while offering integrative social welfare policies. This pattern implies that rights to internal migration would remain to be selectively granted in China after the reform, yet rights to welfare and services in China may be fairer than they were in the past because of the reform.
The 2014 national hukou reform policy states that localities should relax selection policies strategically and selectively to meet labour demands while offering increased social rights to migrants. The local reform patterns revealed through our inquiry suggest that the majority of local governments implemented national policy by closely following the guidelines from the central government so political leaders could gain performance recognition.Footnote 73 Fewer cities leveraged opportunities arisen from vertical decentralization to adopt national policy guidelines flexibly. Furthermore, while the 2014 national hukou reform policy directed cities to adopt more lenient hukou policies as city size decreases, we did not find city-level hukou reform policies reflecting such gradation. We only found that megacities adopted selection policies that were substantially more stringent, and the policy approach distinctions between big, medium and small cities were negligible. Our empirical finding contradicts the assumed “common knowledge” of hukou reform policy patterns whereby smaller cities adopt policies that are more lenient, and the findings revealed a more nuanced depiction of local reform patterns. Researchers should be cautious in using city size as a basis to evaluate the effects of hukou reform.Footnote 74
In examining determinants of local hukou policies in China, our findings suggest that economic, political and sociocultural factors influenced both selection and integration policies. Most studies on determinants of migration policies were based in Western or democratic contexts. To our knowledge, our study is the first to investigate determinants of migration policies covering hundreds of cities in an authoritarian context. Findings in the Chinese context are distinct from Western literature in how political factors exert influences on migration policymaking. In China, the extent of top-down political control and the characteristics of political leaders (e.g. tenure, educational background and area of origin) shape migration policies. Specifically, because politicians’ demonstrated performance in meeting policy objectives or evaluation metrics drives their career advancement in China,Footnote 75 we found that CPSs who have lower prospects for promotion, such as CPSs in their second term,Footnote 76 were less motivated to adopt distinct policy approaches that were different from the majority. Furthermore, we found support for hometown favouritism.
Another contribution of this study is its examination of the role of sociocultural factors in migration policymaking in a Chinese context. Little is known about the extent to which local officials in China consider public sentiments when designing policies. First, we found cities located in provinces with higher levels of labour disputes were more likely to adopt less lenient integration policies. More frequent labour disputes imply a higher propensity for local officials to regard accommodating migrants as problematic. This conflictual status quo may contribute to city officials’ preference for less integrative policies, suggesting that local officials do not adopt integration measures to resolve pre-existing conflicts or to appease migrants.Footnote 77 Second, we found that migrants’ higher perceived social integration level in a locality were not associated with the selection of policy approaches. This finding in contrast with the first shows that local governments’ policy selection is related to how thorny migrant issues are or government media's framing of migrant issues but not related to the integration levels of migrants or public sentiment towards migrants, a population group of marginal political significance. Taken together, our findings on sociocultural factors corroborate that in China policymaking takes into account existing social conflicts but does not respond to migrants’ integration situation. This nuanced finding contributes to the growing inquiry into the extent to which public opinion influences policymaking in China.Footnote 78
This study has several limitations. First, coding hukou reform policy documents can only capture policy plans rather than actual policy actions and implementation process (e.g. how governments allocate quotas to enterprises when executing the point systems). Second, despite our efforts to exhaust all known information sources and to contact municipal offices directly to obtain city-level hukou reform documents, we only retrieved 231 documents from 337 cities. Third, our measures of economic, political and sociocultural factors are confined to available data sources, and some indicators (e.g. unemployment rates) from official statistics are subject to accuracy concerns.Footnote 79 Fourth, the State Council also released related policies after the 2014 hukou reform, such as “Several opinions on further advancing the new type urbanization”Footnote 80 and “The circular on promoting the urban settlement of 100 million non-local hukou residents.”Footnote 81 We focused on identifying the sources of variations in the initial policy set-up, while investigating these subsequent policies requires complex identification strategies addressing dynamic endogeneity in a series of policy decisions and goes beyond the scope of the present study. Fifth, there are data and space constraints for us to evaluate the efficacy of the reform in this article. While consistent with the existing literature,Footnote 82 we did not find that the 2014 reform changed the gap in numbers between urban residents and the urban hukou population in our exploratory analysis,Footnote 83 and we were unable to thoroughly assess the impact of the reform in this article. Future researchers can consider employing other innovative approaches and data sources to overcome these limitations and extend knowledge on the effects of hukou reform and the forces that drive policy landscapes in China.
Supplementary material
To view supplementary material for this article, please visit http://doi.org/10.1017/S0305741023001674
Acknowledgements
We acknowledge the funding support from the Contemporary China Research Cluster Postdoctoral Fellow Scheme from the Faculty of Social Sciences, University of Hong Kong (HKU), and the Seed Funding for Basic Research from HKU. Julia Shu-Huah Wang also appreciates the support from the Yushan Young Scholar Program from Taiwan Ministry of Education (NTU-112V1018-2). Jing You appreciates the support from the Beijing Social Science Foundation (17YJC028).
Competing interests
None.
Julia Shu-Huah WANG is an associate professor at the Department of Social Work, National Taiwan University, and an honorary associate professor at the Department of Social Work and Social Administration, University of Hong Kong. Her research focuses on social welfare policies, immigration policies and the well-being of families. She is currently investigating the design and impacts of social safety nets in East Asia and beyond as well as the patterns and effects of hukou reforms in mainland China.
Yiwen ZHU is a computer science master's student with professional experience in data science and data analytics, and a strong drive and capability for autonomous learning. She is keen to explore new opportunities in the area of cloud infrastructure and deployments.
Chenhong PENG is an assistant professor at the Department of Social Work and Social Administration, University of Hong Kong. Her research interests include social policy (social protection, housing and old-age income protection) and poverty alleviation. Her research has been published in The China Quarterly, Journal of Social Policy, Social Science and Medicine and Health Policy and Planning.
Jing YOU is a professor at the School of Agricultural Economics and Rural Development, Renmin University of China. She specializes in development economics and applied microeconomics. Her research interests include household livelihood in terms of poverty and inequality analyses, evaluation of social policies on education, health and labour, and recently the origins of individual beliefs, preferences and behaviors.
