The Productivist Construction of Selective Welfare Pragmatism in China

Transcription

1 SOCIAL POLICY &ADMINISTRATION ISSN DOI: /spol VOL. 51, NO. 6, November 2017, PP The Productivist Construction of Selective Welfare Pragmatism in China Ka Ho Mok a, Stefan Kühner a * and Genghua Huang b a Department of Sociology and Social Policy, Lingnan University, Tuen Mun, Hong Kong b Department of Asian and Policy Studies, The Education University of Hong Kong, Ting Kok, Hong Kong Abstract This article discusses whether Mainland China under the Hu Jintao and Wen Jiabao leadership ( ) has developed a new welfare settlement, the emphasis of which is to shift away from a productivist focus on education and healthcare investment towards a more protective approach, characterized by investing into social protection and establishing a minimum living guarantee for both the urban and rural poor. In so doing, this article reviews the conceptual debate on Chinese social policy development and explores whether there is any evidence to substantiate a gradual decrease of fragmentation in social provision among the Chinese provinces. With regard to the former question, the article finds that the various social policy initiatives have yet to amount to a qualitative shift in the core foundation of the human capital-focused welfare production logic in China. With regard to the latter question, we argue that considerable fragmentation of social provision at the Chinese provincial level continues to hamper attempts to define a coherent Chinese social model. Indeed, we find considerable diversity in terms of the co-operative state-local interactions within China leading to varying trajectories of social decentralization. Unlike much of the current research in comparative social policy analysis, which continues to treat Mainland China as a single case, this article provides a strong account of a productivist construction of selective welfare pragmatism, which reproduces social policy gaps for different groups of the Chinese population, and suggests that determining multiple welfare types within China might be the most fruitful path for future academic inquiry. Keywords Chinese welfare transition; Chinese social model; Chinese social policy fragmentation; Welfare productivism; Welfare modelling business Author stefankuehner@ln.edu.hk 2017 John Wiley & Sons Ltd

2 Introduction While the discussion of real and ideal types of welfare models continues to influence the field of comparative social policy analysis (e.g. see Emmenegger et al. 2015; Powell and Barrientos 2015), the focus of this debate has expanded considerably since the publication of Esping-Andersen s (1990) ground-breaking book, The Three Worlds of Welfare Capitalism. There has been a considerable internationalization of research agendas, with a particular focus on East Asia (e.g. see Hwang 2011; Ahn and Lee 2012; Lin and Yi 2013; Hudson et al. 2014; Yu et al. 2015; Abrahamson 2016). Although included in this academic discourse with a slight delay, several recent studies have discussed the welfare transition in Mainland China (henceforth: China) under the lens of the welfare modelling debate in order to explore what kind of social model is emerging in the world s most populous country (e.g. see Li 2012; Gao et al. 2013; Ringen and Ngok 2013). 1 Here, it is typically argued that the marketization of social welfare services and the privatization of social policy provision in China during the 1990s resulted in a host of negative consequences (Mok and Ku 2010; Leung and Xu 2015), which were addressed under the Hu Jintao and Wen Jiabao leadership ( , henceforth: Hu-Wen leadership) by implementing various social policy programmes in an attempt to facilitate social harmony and enhance its political legitimacy (Ngok 2013; Ngok and Chan 2016). In light of this emerging literature, this article discusses whether China under the Hu-Wen leadership has developed a new welfare settlement, the emphasis of which is to shift away from the productivist focus on education and healthcare investment towards a more protective approach, characterized instead by investing into social protection and establishing a minimum living guarantee (MLG) for both the urban and rural poor. In so doing, this article reviews the conceptual debate on Chinese social policy development, and explores whether there is any evidence to substantiate a gradual decrease of fragmentation in social provision. With regard to the former question, the article finds that the various social policy initiatives have yet to amount to a qualitative shift in the core foundation of the human capital-focused welfare production logic in China. With regard to the latter question, we argue that considerable fragmentation of social provision at the Chinese provincial level continues to hamper attempts to define a coherent Chinese social model. Indeed, by combining the latest empirical data with theoretical accounts of the politics of the Chinese welfare transition, we find diversity in terms of the co-operative state-local interactions within China, leading to varying trajectories of social decentralization. Unlike much of the current research in comparative social policy analysis, which continues to treat China as a single case, this article provides a strong account of a productivist construction of selective welfare pragmatism reproducing social policy gaps for different groups of the Chinese population, and suggests that determining multiple welfare types within China might be the most fruitful path for future academic inquiry. The article is structured as follows. In the next section, we briefly outline the key aspects of the Chinese welfare transition under the Hu-Wen leadership. This is followed by a discussion on how these changes have been 2017 John Wiley & Sons Ltd 877

3 conceptualized in the scholarship on Chinese social policy development. We then offer an empirical analysis of the latest available local statistics to map the fragmentation of social policy provision among the 31 Chinese provinces over the course of , and review recent accounts to consider the politics behind the varying social policy dynamics among the Chinese provinces. Lastly, we conclude by summarizing our key findings and by placing them into context of contemporary pressures that have led to a different framing of key debates within Chinese political circles and its growing social policy academy. The Chinese Social Model in Transition The marketization of social welfare services and the privatization of social policy provision in China during the 1990s have resulted in negative consequences, including intensifying social inequality and worsening regional disparities (Ngok and Chan 2016; Mok and Qian 2016). Since the initiation of the Chinese economic reforms in the late 1970s, the rigid control through the hukou system was relaxed in response to rapid urbanization and industrialization. As a consequence of the mobility of the rural labour force to work and live in cities, a new form of labour force between the rural and urban residents emerged, namely migrant workers (also: peasant workers or nongmingong, in Chinese) (Ngok 2012). Local governments and enterprises initially adopted an attitude of economic acceptance and social rejection (Chan 2001), leading to a deliberate exploitation of these migrant workers, who were treated as a source of cheap labour without any entitlement to labour insurance. As a consequence, they had to confront the cruel reality of having to accept the dirtiest, toughest and most dangerous jobs with minimal wages (Ngok 2013). In order to rectify this situation, urban social insurance schemes have been gradually extended to migrant workers as part of a much wider government policy agenda to facilitate Chinese social development and to achieve greater social cohesion. Indeed, the Hu-Wen leadership initiated a host of social policy initiatives in order to prevent social instability and political crisis from negatively affecting China s developmental trajectory (see table 1). In realizing that the singular quest for economic growth had created social problems, but also adversely affected the political legitimacy of the Chinese state, the Hu-Wen leadership put forward the scientific outlook of development as a guiding principle in the Third Session of the Sixth Central Committee of the Chines Communist Party. In 2007, the people s livelihood program was introduced, stressing that more attention should be placed on the construction of social harmony and the importance of improving people s livelihood. As social policy or people s livelihood policy became one of the core responsibilities of the Chinese government (Xiong 2014), social security expenditure soared from 3,116,000 Yuan in 2004 to 15,969,000 million Yuan in 2014; albeit at different absolute levels, the expenditure for healthcare and social security investment expanded at a roughly similar pace, while education investment surged particularly after 2010 (see figure 1). The total social expenditure in China reached a total 7.63 per cent of gross domestic product (GDP) in John Wiley & Sons Ltd

4 Table 1 Major policy initiatives in China since 2003 Policy areas New social policy initiatives since 2003 Healthcare Social security Employment Education Housing Personal social services 2003, New rural co-operative medical system for farmers 2006, Healthcare insurance for migrant workers 2007, Contributory health insurance for urban residents 2006, Budgetary fund for the most disadvantaged rural residents ( Five Guarantees ) 2007, Minimum living standard scheme for rural residents 2009, Old age insurance scheme for rural residents 2011, Old age insurance for urban residents 2006, Employment services for migrant workers 2008, Implementation of the Labour Contract Law and Employment Promotion Law 2009, Active employment policy particularly for college graduates and migrant workers 2006, Free compulsory education for all rural and urban children 2006, Education service for migrant children 2007, Welfare housing for low income urban families 2006, Community-based services for the elderly, comprehensive family services, professional social work services, and voluntary services Source: Adapted from Ngok and Huang (2014). Figure 1 Social expenditure in China, Source: NBS (various years). Compiled by the authors according to the China Statistical Yearbook (NBS various years), of which 2.48 per cent was allocated to social security, 1.58 per cent to healthcare and 3.57 per cent to education investment (note the slower growth of social 2017 John Wiley & Sons Ltd 879

5 expenditure figures measured as share of GDP due to the strong average real economic growth of around 7 per cent in China throughout the same time period, see Feenstra et al. 2015). 2 China has also experienced a remarkable expansion of basic medical insurance through enhancing the coverage of different types of workers such as employees in mixed-ownership enterprises, non-public economic organizations, and migrant workers. For instance, the Chinese government promoted a New Agricultural Co-operative Medical Scheme, which was implemented to ease the separation of social welfare provision, covering those in rural and urban areas, which together reached around hundred million Chinese by 2013 (figure 2). In terms of education, the so-called twoexemptions and one subsidy policy was launched in 2003 to improve basic education services for rural children, mainly by reducing the financial costs of education for rural families (Han 2015). Additional government funding was also allocated to schools to rectify the situation, whereby schools attempted to generate revenues by charging for textbooks and other miscellaneous fees. Free nine-year compulsory education for rural school children was implemented in 2006; this policy also provides subsidies to boarding students from poor families. China has experienced a human capital supply shock in the labour market as the number of university students increased from 3.4 million in 1998 to 20.2 million in 2008 (Knight et al. 2017). Not least, a total of million Chinese million of whom live in rural areas are now covered by the MLG Scheme (dibao) (Leisering et al. 2017). Figure 2 The number of people on basic medical insurance Source: NBS (various years), MOH (various years). Compiled by the authors. Note: Unit = one hundred million John Wiley & Sons Ltd

6 The Productivist Construction of Welfare in China In view of these ambitious social policy initiatives, some scholars have argued that China has undergone a historic welfare transition signalling a new era of improved people s livelihoods (Wang 2007;Li2012). Yet again, others have posited that China s attempts to integrate economic and social development goals are indicative of the creation of a new Chinese social model (Leung and Xu 2015), which has fast garnered international attention by lifting hundreds of millions of Chinese out of poverty (Lakner and Milanovic 2016). Welfare productivism in East Asia has typically been characterized as focusing on a growth-oriented development strategy, in which investment into human capital predominates over cash transfers in order to secure labour market fluidity and to increase skilled labour supply (Mok et al., Introduction to this issue, pp ). The prototype policy mechanisms employed by productivist welfare types across East Asia have been extensive investments into education and healthcare paired with flexible labour regulations and strictly means-tested social assistance (e.g. see Gough 2004). Based on this conceptualization of productivist welfare policy, Choi (2012: 275) sees strong signs that several East Asian cases, including China, are moving out of their productivist nature. Equally, Han and Kim (2014: 87) compared the social welfare expansions in China and South Korea and argued that the former has gone beyond the initial socialist collective phase, which emphasises the work unit and collective agriculture for the distribution of welfare, to the social protection expansion phase with the key feature of replacing the selective welfare of liberalism to the institutional welfare of universalism. More broadly still, Lin and Wong (2013) reviewed social policy transformations in Japan, South Korea, China, Taiwan and Singapore, and observed a general shift away from a productivist policy focus towards a more redistributive and inclusive orientation, based on investment into cash transfers and other income protection policies. Thereby, the aforementioned authors stressed that the problem pressures created by demographic and labour market changes and/or political factors such as democratization and the influence of new social movements, prompted East Asian governments to reconfigure their social policy settlements, particularly after the 1997 Asian financial crisis (Kim and Shi 2013). Yet, not all researchers agree with this assessment. For instance, Mok and Hudson (2014) did not find a fundamental change in the Greater China region despite China s various social policy initiatives outlined above. Similarly, Mok and Lau (2014) argued that the Chinese ideological development first, redistribution later paradigm still remained as popular as ever, despite heightened welfare expectations among different groups of the Chinese population. Taking a more historical perspective, Xiong (2014: 36) argues, the long-standing absence of the state in social protection [in China] reflects the government s primary desire to fully control the social system and augment its economic power through its centralised administrative system. For this reason, Xiong (2014: 36) suggests that the MLG Scheme (Dibao) remains deeply rooted in the long-standing Chinese tradition of social relief rather than developing a fully-fledged social right enjoyed by all Chinese citizens John Wiley & Sons Ltd 881

7 Building on their previous work on rich OECD/EU democracies, Hudson and Kühner (2011) using secondary data provided by Botero et al. (2003) and the Asian Development Bank s (ADB s) Key Indicators series classified China as protective at the turn of the 21st century due to relative high social security and employment protection legislation for workers in state-owned enterprises. However, a more recent update of this analysis using data from the ADB s landmark Social Protection Index (ADB 2013) and the World Bank s Doing Business report (2014) is more in line with the East Asian productivism thesis, and suggests that China has not experienced a qualitative shift in regards to its income protection and training investment (Kühner 2015a). Instead, China is merely classified as weak productive-protective due to high education investment and high rigidity of employment protection (particularly compared to very low rigidity scores in productivist proto-types such as Singapore and Malaysia). By contrast, both China s total social protection spending (as a share of GDP) and its overall social protection index score (computed by dividing the total social protection spending by the total number of intended beneficiaries) remained considerably short of the respective scores for South Korea, which was highlighted [by the ADB] as an attainable [protective] benchmark for at least middle-income countries in Asia and the Pacific (ADB 2013: xiii xv). In what is probably the most detailed appraisal of the welfare transition in Greater China and East Asia to date, Yang (2017) also considers the protective intent of Chinese old-age income protection, public housing and passive labour market policy and again finds quite substantial differences between China and the other post-productivist cases highlighted in the literature, namely Japan, Taiwan, but especially South Korea. Based largely on qualitative historical policy data, Yang s (2017) contribution suggests that China fails to pass the qualitative break-point of membership for each of these protective fuzzy sets resulting in China being classified as weak productive (based on comparatively high scores in healthcare and family policy investment, but interestingly not education due to relatively low accessibility of tertiary education, compared to five other East Asian economies). There is no doubt that the Hu-Wen leadership has taken great strides in terms of its implementation of social policy programmes, but given these results and the fact that China is still lagging behind in terms of its human development and income inequality compared to other high-income East Asian economies (Kühner 2015a), it appears difficult to substantiate claims made in the literature of a tangible shift towards a new welfare settlement in China. China joins other East Asian cases by still focusing a large share of its total social expenditure on education investment around 47 per cent according to the latest figures published in the China Statistical Yearbook (see figure 1). This is a conservative approach to measure productive intent as it does not count the positive effect of healthcare on the commodification of labour (Rudra 2007). Suffice it to say that there is a good deal of conceptual disagreement and ambiguity about how to conceptualize productive and protective welfare policies in the literature (Nolan 2013), but overall the evidence suggests that the Chinese social model has thus far remained predominantly productivist in nature John Wiley & Sons Ltd

8 Regional Fragmentation of the Chinese Social Model The previous assessment of China s welfare transition under the Hu-Wen leadership is further corroborated when considering the structural imbalance of welfare provision between rural and urban China. For instance, He and Wu (2016) have argued that the path towards universal health insurance in China continues to be severely hampered by the existence of multiple social health insurance schemes and their limited portability across provincial lines. Focusing on the new rural co-operative medical system for farmers, Li et al. (2015) found that this programme has only achieved limited positive effects in regards to household financial risk from medical expenses and rural-urban income inequity. In 2007, 4.9 hospital beds were available for every 1,000 urban residents, but only two hospital beds were available for every 1,000 rural residents. In 2014, the number of hospital beds per 1,000 population increased to 7.8 and 3.5 in the urban and rural areas, respectively, indicating that the gaps in provision of healthcare services between the rural and urban areas have so far persisted (figure 3). Wu (2013) has come to a very similar conclusion i.e. stating unequal institutional arrangements, increasing benefit level differentials between rural and urban areas and the persistence of gaps in coverage for self-employed and unemployed for pension arrangements among different segments of the Chinese labour force. Employees in Chinese government departments and official social organizations continue to benefit from more comprehensive social insurance coverage and enjoy relatively higher benefit levels compared with the urban employees serving in the non-public urban sector. In 2001, the average monthly pensions for retirees from government departments and official social organizations was 9,766 Yuan per year, almost 1.5 times the average monthly payments for retirees from other enterprises (6,674 Yuan per year); while pension benefits for both groups increased rapidly after 2003, the gap between the two has only decreased marginally (figure 3). Figure 3 Healthcare services and pension provision in China, Source: NBS (various years). Compiled by the authors John Wiley & Sons Ltd 883

9 While the above issues have been discussed at length in the literature, a major additional policy challenge has been identified in regards to the diverse welfare provision among the different Chinese provinces. This challenge has been recognized by the Chinese government, which issued a White Paper in 2004 entitled China s Social Security Situations and Policies underlining that the Chinese government s approach to social welfare development has been built upon stark imbalances in regional development (State Council Information Office of China 2004). In the preface of this document, it is stated that: China is the largest developing country in the world, with a huge population, weak economic base, and unbalanced development between different regions. The Chinese government has established and modified social security systems that are compatible with the levels of economic development under its conditions. 3 There have been numerous quantitative accounts of fiscal decentralization in China and its impacts on economic growth, inequality and other social and economic outcomes (e.g. see Wang and Ma 2014; Liu et al. 2016). In this article, our interest rests on the extent to which regional variations in China s socio-economic development have resulted in considerably variation in social provisions for the citizens living in different Chinese provinces. To explore this question empirically, we compiled a data set, which includes the expenditures for the main social welfare programmes (i.e. education, healthcare and social security) across the 31 Chinese provinces to enable us to compare the difference in social welfare investment and to answer whether and how regional provision gaps have been affected during the Hu-Wen leadership. Given the Table 2 Differences in social expenditures among Chinese provinces Year Maximum Minimum Maximum/ Minimum Mean Coefficient of variation ,497 (Beijing) 281 (Guizhou) % ,813 (Beijing) 333 (Guizhou) % ,097 (Beijing) 423 (Anhui) % ,479 (Beijing) 520 (Guizhou) % ,348 (Beijing) 735 (Guangxi) 4.6 1,283 51% ,787 (Beijing) 953 (Guangxi) 4.0 1,628 46% ,121 (Beijing) 1,210 (Shangdong) 3.4 1,991 41% ,607 (Qinghai) 1,507 (Henan) 4.4 2,533 46% ,874 (Qinghai) 2,033 (Henan) 3.9 3,334 45% ,657 (Qinghai) 2,463 (Hebei) 3.5 3,898 39% ,729 (Xizang) 2,566 (Hebei) 3.4 4,133 36% ,925 (Xizang) 2,747 (Hebei) 4.0 4,568 39% Source: NBS (various years). Compiled by the authors. Notes: Per capita, Yuan. n = John Wiley & Sons Ltd

10 oft-discussed limitations of expenditure data to measure the degree of welfare policy change (Kühner 2015b), we combine this new data set with newly available secondary data on the generosity of the urban MLG Scheme (Wang and Bai 2015) to gain a better understanding of changes in the protective intent of welfare policy across the different regions of China. Table 2 shows that the maximum and minimum total social expenditure across Chinese provinces in 2003 were 1,497 Yuan per capita in Beijing and 281 Yuan per capita in Guizhou, respectively. The ratio of these two extreme values (i.e. maximum/minimum) reached 5.3, which was among the highest scores in our available data for the time period (second only to the maximum/minimum gap in 2004). A slightly more sophisticated measure of regional differences in social expenditure in China is the coefficient of variation, which describes the amount of variability across all Chinese provinces relative to the mean. From 2003 to 2005, the coefficient of variation of the total social expenditure per capita among 31 provinces was greater than 50 per cent compared to below 40 per cent throughout The data confirms that China has experienced not only a general increase of social expenditure the mean across all provinces increased from 564 Yuan per capita in 2003 to 4,568 Yuan per capita in 2014 but also that it has experienced an upward convergence of overall expenditure levels. The findings are a little different when we consider regional developments in the MLG Scheme (table 2). Wang and Bai s (2015) data show that the maximum and minimum benefit levels measured as a share of the average wage were 17.0 per cent in Hebei province and 8.6 per cent in Tibet in 2003, respectively. The ratio of extreme values (i.e. maximum/minimum) reached 2.0, which again was the highest score for the available data and reduced to 1.6 in 2013 (the latest point of available data). However, the coefficient of variation does not suggest that the generosity gap was diminished across 31 regions as the findings for 2003, 2008 and 2013 remain broadly similar. These findings are corroborated when we consider the development of regional disparities of benefit levels as a share of the minimum wage. It is worth mentioning that there was an overall decrease in the mean benefit level replacement rates among Chinese provinces probably owing to increased wages in China (Athukorala and Wei 2017). The average benefit level across all Chinese provinces was 14.1 per cent in 2003 and reduced to 9.4 per cent in 2013; the drop of benefit levels as a share of the minimum wage was even more pronounced from 41.1 to 30.2 per cent over the same time period. The generosity of income protection schemes is typically not only based on discussions of benefit levels, but also by the number of designated beneficiaries the programme reaches (see e.g. ILO 2017). Such data is generally more difficult to retrieve, but a look at the number of urban minimum living standard scheme for rural residents (MLSS) recipients measured as a share of the non-rural population reveals a similar decrease in the number beneficiaries across all Chinese regions, while the regional gaps ranged between 13.4 percentage points in 2003 (14.1 in Qinghai 0.7 in Zhejiang) and 11.4 percentage points in 2008 (11.9 in Gansu 0.5 in Zhejiang) (see also Wang and Bai 2015) John Wiley & Sons Ltd 885

11 The data presented in table 2 and table 3 indicates that China experienced a gradual closing of the overall social expenditure gap across the different Chinese regions, but not the regional generosity gap of the urban MLSS programme under the Hu-Wen leadership. Hierarchical cluster analysis can be undertaken on the above data to assess whether there have been any significant changes in the composition of disaggregated social policy expenditures, i.e. education, health, social security as well as in urban MLG recipients and standards across the Chinese provinces. In 2003, the 31 Chinese provinces are classified into seven clusters, with Tibet behaving differently to all other provinces in terms of its social spending profile (table 4). Clusters A 1 (Henan, Guangxi, Guizhou, Jiangxi, Sichuan, Hunan) and B 1 (Heilongjiang, Hubei, Hebei, Hainan, Anhui) were differentiated by overall low welfare investments. Cluster C 1 (Zhejiang, Guangdong, Jiangsu, Fujian, Shangdong) comprises the Eastern coastal provinces and is different from Clusters A 1 and B 1 mainly due to much higher investment into education services. Cluster D 1 (Chongqing, Shaanxi, Shanxi, Yunnan, Inner Mongolia, Gansu) comprising mainly of Central-Northern Chinese provinces has similar overall levels of social expenditure than Cluster C 1, yet the data suggests a bias towards social security rather than education investment. In other words, one can identify significantly different social spending patterns at the provincial level in China at the outset of the Hu-Wen leadership. We also identify a Chinese city Cluster H 1 (Beijing, Tianjin, Shanghai), which combines relatively high investment into education and healthcare (generally more than twice as much investment than all other provinces) with high social security spending. However, the highest level of social security spending was evidenced in Cluster F 1 (Liaoning, Qinghai) despite comparatively low levels of healthcare investment. Table 3 Differences in MLG standard among Chinese provinces (n = 31) Year Maximum Minimum Maximum/ Minimum Mean Coefficient of variation As a share of average wage (%) (Hebei) 8.6 (Tibet) % (Tianjin) 7.0 (Xinjiang % and Tibet) (various) 6.8 (Ningxia) % As a share of minimum wage (%) (Beijing) 28.3 (Xinjiang) % (Beijing 24.4 (Guizhou) % and Tianjin) (Beijing) 22.1 (Ningxia) % Source: Wang and Bai (2015) John Wiley & Sons Ltd

12 Table 4 Chinese welfare policy clusters in 2003 and A 1 : Henan, Guangxi, Guizhou, Jiangxi, Sichuan, Hunan B 1 : Heilongjiang, Hubei, Hebei, Hainan, Anhui C 1 : Zhejiang, Guangdong, Jiangsu, Fujian, Shangdong D 1 : Chongqing, Shaanxi, Shanxi, Yunnan, Inner Mongolia, Gansu E 1 : Jilin, Ningxia, Xinjiang F 1 : Liaoning, Qinghai G 1 : Tibet H 1 : Beijing, Tianjin, Shanghai A 2 : Henan, Guangxi, Anhui, Hubei B 2 : Heilongjiang, Jiangxi, Hunan C 2 : Inner Mongolia, Liaoning D 2 : Jiangsu, Zhejiang, Hebei, Shangdong E 2 : Fujian, Guangdong F 2 : Chongqing, Shaanxi, Hainan G 2 : Ningxia, Xinjiang, Gansu H 2 : Jilin, Yunnan, Shanxi, Guizhou, Sichuan J 2 : Tibet, Qinghai K 2 : Beijing, Tianjin, Shanghai Notes: Hierarchical cluster analysis using Ward linkage (n = 31). There are five indicators in the two cluster analyses for 2003 and 2013, respectively: education expenditure per capita, health expenditure per capita, social security expenditure per capita, urban MLG recipients as a share of non-agricultural population, and urban MLG standard as a share of average wage. Housing expenditure per capita is only available after 2010 including this indicator into the cluster analysis for 2013 does not change our substantive findings. Repeating the same exercise for the most recent data in 2013 suggests a formation of a new separate Cluster H 2 comprising of different cases from across the previous Clusters A 1,D 1, and E 1 in 2003 (table 4). The Eastern coastal provinces grouped into two separate Clusters D 2 and E 2 in All Clusters A 2 -K 2 share in common a significant increase in education and healthcare investment (roughly a seven- to eightfold increase between 2003 and 2013), which outpaced still significant increases in social security (roughly fivefold). Beijing, Tianjin and Shanghai continue to form a separate Cluster K 2 characterized by relatively high expenditures on education, healthcare and social security. This cluster is differentiated from Cluster J 2 (Tibet, Qinghai) only by a relatively lower share of urban MLG recipients of the non-agricultural population; in regards to social security per capita (and also housing expenditure per capita), the Chinese city-states had been surpassed by the cases in Cluster J 2 by Despite catching-up in terms of the overall levels of welfare policy spending, the increase in the number of clusters among the Chinese provinces in 2013, from eight total clusters to ten, suggests an increased divergence of disaggregated welfare regime characteristics throughout the Hu-Wen leadership. This finding lends further credence to the characterization of the Chinese welfare transition not only as primarily productivist, but also as characterized by a continued and even growing degree of regional fragmentation. Appendix table A1 summarizes the findings of these relatively simple cluster analyses in more detail John Wiley & Sons Ltd 887

13 The Construction of Selective Welfare Pragmatism in China So far, our review of the Chinese welfare transition and our own data analysis have suggested that despite implementation of various social protection measures under the Hu-Wen leadership, the Chinese social model has largely remained biased towards human capital production through social investment in education and healthcare services. We have also shown that the fragmentation of policy outputs has persisted despite concerted efforts to address the imbalances in Chinese regional development since publication of the 2004 White Paper. Indeed, we have not only shown that gaps in healthcare services and pension provision in China between urban-rural and occupational divides remain, but also identified a further differentiation of empirical clusters of welfare provision among Chinese provinces despite a notable convergence of overall welfare policy spending levels. In other words, rather than converging towards a coherent Chinese social model, the productive and protective intent to borrow the wording from Hudson and Kühner (2011) of local governments appears to have diverged to some extent in a period of significant overall social policy expansion. In an attempt to explain these developments the literature on Chinese social policy development has begun to provide useful theoretical accounts of the strategies adopted by local Chinese governments to integrate the contradicting demands of simultaneously promoting economic growth and social harmony. It is these accounts of the politics of persisting regional fragmentation in China that we now turn to in the final substantive section of this article. Qian and Mok (2016) analyzed longitudinal government expenditures in education, health, social security and assistance programmes for three city governments in the Eastern coastal areas in China, and highlighted the unintended consequences of what they called a dual decentralization process. More precisely, Qian and Mok (2016) posit a crowding out effect of social programme delivery at the local level in China due to the adverse consequences of a performance evaluation system related to social welfare delivery, and argued that the fragmentary authority and information structure may cause policy ineffectiveness. Thus, they present an argument that is in line with the idea of paternalistic welfare pragmatism in China for two reasons: first, government social expenditures are based on the fiscal capacity of city-level governments; second, variations of welfare programs are associated with the dichotomy between the urban formal and informal sector. Several other studies have equally highlighted the role of local government leadership in social policy implementation thereby further strengthening the notion of pluralist welfare production among the Chinese provinces (e.g. see Shih 2012; Leung and Xu 2015; Ngok and Chan 2016). In view of these diverse regional social protection experiences, Shih (2017: 1) proposed three different but interrelated modes of territorial politics in China that have emerged when local governments implement social protection programmes in response to the call of the central administration in Beijing. These modes are: John Wiley & Sons Ltd

14 (1) social dumping, in which regions compete to attract inward investments or subsidies from above at the expense of distributional objectives; (2) regional protectionism, in which well-off localities tend to erect administrative barriers to prevent welfare migration from other regions; and (3) intricate control coordination of local implementation that renders social policy development even more regionalized, with only the least common denominator (i.e. minimum benefit levels set in major social programmes) applicable across the country while supplementary benefits remain variegated across the region. Shih (2017) posits that Chinese citizens have experienced differential welfare treatments because of a social decentralization in which social provision relies substantially on local fiscal and implementation capacity of the local government as well as its political will. With fiscal decentralization being adopted in public administration, the social policy provisions that Chinese citizens enjoy primarily depend on local governments to achieve policy objectives because the progress of social policy implementation [ ] is closely related to local officials incentive structures, individual efforts and institutional recognition (Xiong 2014: 51). However, in realizing the limitations of such a dualistic approach to analyze the central local relationship, an increasing number of studies has identified the importance of co-operative interactions in Chinese policy implementation (Chien 2007; Zheng 2007). In this perspective, China s central government has adopted a soft centralization approach, with the concentration of power at the upper administration level (i.e. mostly province and prefecture), thereby leaving low levels of governments (e.g. township government) to become hollow shells (Mertha 2005; Smith 2010). Conceiving the Chinese state at the interventionist end of a continuum running from laissez-faire to central planning, the directionality of innovation and social policy reforms are essentially top down: the central government sets the policy directions at the commanding highest, and through the rigid control of the Chinese government over the official approval in which subnational actors compete with each other for promotion within the hierarchical party system by meeting the settled criteria (e.g. see Cai and Treisman 2006; Landry 2008). Indeed, the Chinese state has vacillated between centralization and decentralization in governing to address the huge geographical size and complexity/diversity of local interests. Therefore, the centralized decentralization that occurs in implementing policies of strategic importance for securing state legitimacy and asserting control over economic and social activities should not be surprising (Hsueh 2011; Mok and Wu 2013; Lardy 2014). Meanwhile, economic and social reforms have also dispersed power within the state and introduced market forces to manage economic affairs and to govern social sectors such as housing, health, education, social services and social welfare (Wong and Flynn 2001; Mok and Ku 2010; Zhao and Zhou 2017). A competition under hierarchy or experimentation-based policy cycle has evolved in China with local experiments under the patronage of senior leaders at the central level. This 2017 John Wiley & Sons Ltd 889

15 system typically shows how steering at a distance operates in Chinese social administration (Heilmann and Perry 2011; Gobel 2011) and is consistent with the notion of social decentralization proposed by Shih (2017). Public policy research on China ought to appreciate territorial politics, governance, and the different ranks of government affecting policy implementation (Cartier 2016). The emerging literature on Chinese social policy development equally accounts for the differentiated experience of social policy implementation by pointing to the specific financial and administrative arrangements under the central local government structures and their specific interactions in Chinese context. As the result of centralized decentralization, the central government in Beijing assumes the role of policy coordinator and supervisor, and defines broad national policy frameworks and principles. Thereafter, policy implementation relies substantially upon political will and state capacity at the local level. The central government is thereby in agreement with local governments and multiple actors to diversify funding sources and mobilize different non-state actors/sectors to facilitate the implementation of public and social policies decided by the central government. Indeed, maximizing multiple levels of governance to manage the multi-faceted characteristics of social administration has become a prominent trend in China (Chou and Ravinet 2015; Mok and Han 2017). Taken together, the different political economy accounts enable a better understanding why selective welfare pragmatism is observed across the different regions in China. Conclusion The call for the development of a harmonious society and the strong intention of the state to make welfare protection a policy tool for social stability and regime legitimation have fundamentally changed the social policy landscape in China. Indeed, the social policy expansion that has taken place under the Hu-Wen leadership at first glance suggests a fundamental paradigm shift of the Chinese social model departing from a predominantly productivist human-capital approach towards a more protective welfare settlement, with a stronger orientation towards social rights of all Chinese citizens including the rural population and previously excluded migrant workers. However, we have argued in this article that China is currently experiencing a transitional period in which the formation of a new post-productivist welfare regime has not yet fully materialized. Overall, the evidence presented in this article suggests that the Chinese production of welfare has thus far remained productivist in nature. In addition, given China s vast geographical size and the diverse regional development patterns highlighted, we have argued that defining only one Chinese social model should not restrict the conceptualization of China s social development. We have argued that considerable fragmentation of social provision at the Chinese provincial level persists, and we find considerable diversity in terms of the co-operative state-local interactions within China leading to varying trajectories of social decentralization. Determining multiple John Wiley & Sons Ltd

16 typologies within China might be a more fruitful path for future academic inquiry: local governments in China have addressed the tension between achieving economic growth and social harmony, but they have done so in various ways owing to different central-local relationships as well as different interactions between their own capacity and the market in policy implementation. China is confronted with huge regional disparities, a persistent urban-rural fragmentation in its social security system, and complex social inequalities across her 31 provinces. The present study indicated how the Chinese government has tactically walked with different legs that correspond to the diverse social, economic, and political needs of these different Chinese localities and regions. By cautiously engaging in a productivist construction of solidarity, the Chinese government has adopted a paternalistic approach in governing its multiple welfare models across different regions. Adopting social policy as a tool to manage old and new social risks and contingencies, this article has shown how Chinese adaptive state entrepreneurship functions through partnership with local governments in managing competing and contradicting economic and welfare demands against the context of compressed development. Taken together, the instrumental and pragmatic approach taken by local governments in addressing diverse development needs and the ensuing differential treatment experienced by citizens across the Chinese provinces provide a strong characterization of Chinese welfare state development as a productivist construction of selective welfare pragmatism. However, in view of increasing uncertainty after the global financial crisis of 2008, Chinese leaders have begun to call for a critical reflection of the strategies adopted under the Hu-Wen leadership, particularly in anticipation of a further slowdown in economic growth (Su and Tao 2017; Bulman et al. 2017; Paus 2017). Recent developments threatening established global value chains and generating new international movements of people, such as the Brexit from the EU, growing American protectionism under the Trump administration, climate change, the Syrian refugee crisis and global terrorism, have already rendered Chinese leaders considerably more cautious to embrace further social policy expansion (Li 2016). A more complete understanding of the multiple typologies of welfare and thence the future social policy development in China should be obtained through the lens of political economy and historical institutionalism, which together acknowledge path dependence as a temporal process that produces self-reinforcing effects on social policy governance and delivery (e.g. see Shih 2006; Xiong 2009; Zhang 2013). This, in turn, will facilitate endeavours to overcome the dependent variable problem in Greater Chinese and East Asian social policy research more generally, and prepare the ground for further theoretical innovation on the basis of more systematic historical comparative research based on multiple Chinese cases. This article was not able to deliver such an ambitious research objective, but through mapping out the most recent discussions and available data on the contemporary Chinese welfare transition the hope is that is makes a contribution by setting the research compass in this future direction John Wiley & Sons Ltd 891

17 Appendix Table A1 Chinese welfare policy clusters in 2003 and 2013 Cluster membership Education expenditure (per capita, Yuan) Health expenditure (per capita) Social security expenditure (per capita) Urban MLG recipients (% nonagricultural population) Urban MLG standard (% of average wage) 2003 A: Henan, Guangxi, Guizhou, Jiangxi, Sichuan, Hunan B: Heilongjiang, Hubei, Hebei, Hainan, Anhui C: Zhejiang, Guangdong, Jiangsu, Fujian, Shangdong D: Chongqing, Shaanxi, Shanxi, Yunnan, Inner Mongolia, Gansu E: Jilin, Ningxia, Xinjiang F: Liaoning, Qinghai G: Tibet H: Beijing, Tianjin, Shanghai 2013 A: Henan, Guangxi, 1, Anhui, Hubei B: Heilongjiang, 1, , Jiangxi, Hunan C: Inner Mongolia, 1, , Liaoning E: Jiangsu, Zhejiang, 1, Hebei, Shangdong E: Fujian, Guangdong 1, G: Chongqing, 1, , Shaanxi, Hainan H: Ningxia, 1, , Xinjiang, Gansu J: Jilin, Yunnan, Shanxi, Guizhou, Sichuan 1, , (Continued) John Wiley & Sons Ltd

18 Table A1 (Continued) Cluster membership Education expenditure (per capita, Yuan) Health expenditure (per capita) Social security expenditure (per capita) Urban MLG recipients (% nonagricultural population) Urban MLG standard (% of average wage) K: Tibet, Qinghai 2, , , L: Beijing, Tianjin, 3, , , Shanghai Notes: Unweighted cluster averages. Hierarchical cluster analysis using Ward linkage method, n = 31. Acknowledgements We would like to thank the regional issue editors and two anonymous reviewers for their constructive comments that enabled us to improve this article. All errors and omissions remain our own. Notes * Stefan Kühner, corresponding author. 1. The same can be said about Greater China, comprised of Hong Kong SAR, Taiwan ROC, and Macau, but this will not be the focus of this article. For a good summary of key debates see, for example, Mok and Lau (2014). 2. There is a slight difference in the social expenditure levels reported in the Chinese Statistical Yearbook compared to other major international data collection initiatives (see, e.g. ADB 2013). 3. Available for download at: content_18024.htm (accessed 4 July 2017). References Abrahamson, P. (2016), Returning to the Welfare Modelling Business in East Asia with an Eye to Care, Paper presented at the 12th ALIN Expert Forum, Social Welfare Laws in Asian Countries in Relation with Economic Development, 28 April, Hotel President, Seoul, Korea. Asian Development Bank (ADB) (2013), The Social Protection Index, Mandaluyong City: ADB. Ahn, S. H. and Lee, S. Y. S. (2012), Explaining Korean welfare state development with new empirical data and methods, Asian Social Work and Policy Review, vol. 6: Athukorala, P. C. and Wei, Z. (2017), Economic transition and labour market dynamics in China: An interpretative survey of the turning point debate, Journal of Economic Surveys, (accessed 7 July 2017). Botero, J., Djankov, S., La Porta, R., Lopez-de-Silanes, F. and Shliefer, A. (2003), The Regulation of Labor, National Bureau of Economic Research Working Paper 9756, Cambridge, MA: NBER John Wiley & Sons Ltd 893