Chapter 4. Managing across levels of government: The challenge of pension reform in China

Christine Wong
East Asian Institute
National University of Singapore
Randong Yuan
Centre for Contemporary Chinese Studies
University of Melbourne

Research for this chapter was partially funded by a grant from the Ford Foundation. We thank participants of the OECD Workshop on Ageing and Long-Term Challenges across Levels of Government for helpful comments.

The People’s Republic of China (hereafter ‘China’) is facing a “population ageing tsunami”,1 with the share of the population aged over 65 expected to double between 2010 and 2030. This has already led to a surge in public pension expenditures, with a near-quadrupling of outlays in the Urban Employee Scheme (UES), the largest scheme by expenditure, from CNY 1.05 trillion in 2010 to CNY 3.8 trillion in 2017, an annual growth of nearly 24% (National Bureau of Statistics, 2018[1]). This alarming demographic trend coincides with a broadening of public pension coverage in China’s recent effort to improve public services, reinforcing the urgent need for reform to find a way to render pension programmes sustainable in the long term.

This chapter examines the issues and challenges facing the pension system in China. The focus will be on the public pension system, particularly the Urban Employee Scheme – often called the Basic Old Age Insurance Scheme – that provides coverage for 400 million urban employees and pensioners. While the pension system in the wider sense would include employer-based annuity funds and commercial annuity insurance programmes to provide supplementary sources of retirement income, these non-public elements have received little policy support and have seen little development in China (Fang and Feng, 2018[2]).

This chapter is organised as follows: the next section provides a brief description of the current public pension system and recent progress in broadening coverage. This is followed by an overview of the Urban Employee Scheme and its operations. The fourth section examines the problems of fragmentation and managing across levels of government. The final section describes current reforms and the way forward.

China’s public pension system consists of three schemes: the Urban Employee Scheme (UES); the Public Employees Scheme (PES); and the Basic Residents Scheme (BRS). All three schemes have two pillars – a defined benefit pillar funded by pooling contributions from employers and fiscal subsidies, and a defined contribution pillar funded by employee contributions. The first two schemes are compulsory, while the third is voluntary. A summary of their main features and current status are presented in Table 4.1.

The Urban Employee Scheme was launched in 1997 to provide coverage for urban employees (State Council, 1997[3]). As China urbanised rapidly over the next two decades, this scheme has grown with the size of the urban labour force, and its scope was expanded to take in self-employed urban residents. In 2017, the UES had a total enrolment of 402.9 million, comprising 292.7 million active employees and 110.2 million retirees (National Bureau of Statistics, 2018[1]).

The Public Employees Scheme was launched in 2015 as part of public service reform and provides coverage to the estimated 40 million employees working in government and public institutions (State Council, 2015[4]).2 Before the PES was launched, these pensions were funded on a pay-as-you-go basis from the current operating budgets of government agencies and public institutions – a carryover from the planned economy that had applied to state-owned enterprises (SOEs) as well. Without a formal scheme, public employees did not pay contributions and enjoyed generous retirement benefits that were 80-100% of final salary, with adjustments pegged to current civil service pay. From the inception of the UES, public employee pensions diverged increasingly from urban employee pensions, growing to nearly double the level of UES pensions. These costly pensions were widely resented and perceived to be a barrier to reforms to devolve public services to market provision.

The PES is a transitional scheme toward integrating public employees into the main Urban Employee Scheme. Under the PES, public employees and their agencies pay contributions under the same rules that apply to the UES (see Table 4.1). At the inception of the scheme, civil service and public employee salaries were augmented to ensure the same level of take-home pay. Under the PES, retirement benefits will converge to those under the UES. To preserve some differentiation, public pension benefits will also be supplemented with an annuity financed by an annual 8% payroll contribution from the employers and 4% of contributing wage from the employees. A ten-year transition period guarantees that those retiring in this period will not see their pension benefits reduced due to the reform.3 For now, the financing and management of the PES remain separate from the UES.

The Basic Residents Scheme was created in 2014 through the merger of the New Rural Pension Scheme and the Urban Residents Scheme (State Council, 2014[6]). The scheme is designed to provide coverage to urban and rural residents without formal employment. In practice, it also acts as the default scheme for a large number of migrant workers not covered under the UES – an issue discussed later in this chapter. Thanks to its residual status, it is the largest of the three schemes, with over 500 million participants.

As of 2008, only around 40% of the labour force was covered by government-sponsored pension arrangements – the UES, a rural pension system existing in some counties (covering 28.3% of the labour force),4 and pension plans of government agencies (covering 7.2% of the labour force) and public institutions (covering 5.2% of the labour force) (Impavido, Hu and Li, 2009[7]). In the decade since, pension coverage was greatly expanded under the aegis of the Harmonious Society programme when the government began in the mid-2000s to re-orient budget spending toward public services (Wong, 2010[8]).

The New Rural Residents programme was introduced in 2009, offering eligibility to all rural residents regardless of employment status (State Council, 2009[9]). This was followed in 2011 by the Urban Residents Pension programme. With these programmes, the government extended public pensions to the non-employed and the labour force outside the formal sector, filling a huge gap in coverage. Although the schemes are voluntary, they are popular because they are heavily subsidised, and the threshold for joining is a contribution of as little as CNY 100 per annum to qualify for a basic pension. By 2010, over 100 million were enrolled, and participation jumped to 332 million in 2011 when the programmes were in operation in all provinces (National Bureau of Statistics, 2018[1]).

In 2014, the two residents’ programmes were merged into the Basic Residents Scheme, and enrolment now exceeds 500 million. The scheme has a basic pension component funded by the central government for the central and western provinces (which together comprise two-thirds of China’s population), and jointly with local governments for the eastern provinces. In some provinces, this basic pension is topped up by local government contributions. The scheme also has a defined-contribution component funded by individual contributions.

The Urban Employee Scheme has also seen significant progress in coverage in recent years. During 2000-17, the number of workers contributing to the scheme grew from 104.4 million to 292.7 million, when the coverage rate rose from 45% of the urban workforce to 69% (National Bureau of Statistics, 2018[1]; Zuo, 2014[10]) (see Figure 4.1). Figure 4.2 shows the growth in participation under the UES and the BRS. When the 40 million in PES are added, a total of 955 million people were covered under public pensions in 2017, nearly 90% of the population over age 16.5

The recent expansion of pension coverage has made China stand out among emerging economies. Indeed, the Chinese government was given the International Social Security Association’s Award for Outstanding Achievement in Social Security, which the association gives every three years, in recognition of China’s “unprecedented progress in the development of its social security system in the past decade” (ISSA, 2016[11]). However, much work remains to achieve full social welfare coverage in China. As shown in Table 4.1, at present the benefit levels differ significantly across the three schemes, with average monthly pensions of nearly CNY 5 000 in the PES, almost CNY 3 000 in the UES, and only CNY 127 in the BRS. The reality is that more than half of the covered population are on the BRS, which will provide a pension that covers only a fraction of living costs in retirement, at a level that is far below the official poverty line.6

The basic framework for the UES was set in the 1997 document, “Decision of the State Council on the Establishment of a Unified Basic Old-Age Insurance System for Employees of Enterprises”. The scheme moved the financing of pensions out of state-owned enterprises into pension pools at the county or city level. The document called for a three-part pension system. The first part is a defined benefit plan called the basic pension funded by enterprise contributions. The second part is a defined-contribution component funded jointly by employer and employee contributions into individual employee accounts. The third part would comprise voluntary, private pensions such as annuities or individual retirement savings plans. In practice, there has been little take-up of the third part of the policy, and private pensions remain underdeveloped in China.

With subsequent tweaks to the basic design (State Council, 1998[12]; 2000[13]; 2005[5]) and the 2010 Social Insurance Law, the UES calls for enterprises to contribute up to 20% of payroll into the social pension pool. With an accrual rate of 1% per annum, it is designed to offer a basic pension of 35% of the city’s average wage after 35 years of work. This basic pension would be supplemented with benefits from the individual account, to which employees contribute 8% of salary, and is designed to yield a target of 24.2% of the city’s average wage.7 Together, the target retirement benefit is 59.2% under the UES.

Although nominally a national scheme directed by the State Council and placed under the administration of the Ministry of Human Resources and Social Security (formerly the Ministry of Labour and then the Ministry of Labour and Social Security), at the operational level the UES is managed as local pension pools under city and county governments. This decentralised management feature is a legacy of the economic and political circumstances of the 1990s, as the scheme’s inception coincided with the prolonged fiscal decline during the first two decades of transition when market reforms undermined the traditional support of state-owned enterprises provided by central planning mechanisms (Wong, 1993[14]). With the budget falling from one-third of the gross domestic product (GDP) in the late 1970s to only 10% of GDP during 1995-96, the central government was unable to commit fiscal support to public pensions. At the same time, the rollout of a public pension programme was urgently needed as a precondition to the reform of SOEs, whose deepening losses were further draining fiscal resources.

At inception, then, the UES left in place the pooling and management of pensions at the city and county level that were characteristic of pilot reforms at the time in Guangdong, Shanghai, etc.8 Because of the differences in economic and demographic structures across localities, the contribution rates required to meet pension expenditures varied substantially across pools. In Guangdong province they reportedly ranged in the mid-1990s from 12% in Zhuhai (a newly emerging hub of labour-intensive exports) to 26% in Guangzhou, the provincial capital. Within the province of Hubei, contribution rates ranged from 19% to 32% across 70 counties and cities (West, 1999[15]). In light of these differences, the State Council launching document had expressly allowed provinces to set contribution rates in accordance with local needs, albeit in consultation with the Ministry of Finance and the Ministry of Labour (State Council, 1997[3]).

Today the UES remains a patchwork of thousands of local pension pools, some with as few as 30 000 participants. The differences in the economic and demographic conditions they face have only grown through the past two decades, as urbanisation accelerated, and young migrants moved in increasing numbers to growth centres where the jobs are located. This is reflected in the data presented in Figure 4.3, which shows the disparities in the ratio of workers to pensioners in the UES across provinces. At the top, Guangdong, whose export industries are magnets drawing young workers from other provinces, has more than eight workers to support each retiree. At the bottom are seven provinces that have fewer than two workers per retiree. The worst-off are Liaoning, Jilin and Heilongjiang, the three north-eastern provinces in China’s “rust belt”, which are burdened with declining heavy industries and aged workforces. Between 2014 and 2017, Heilongjiang reportedly added 640 000 retirees and only 165 000 workers to the province’s UES (Guo, 2018[16]).

To gauge the size of disparities across the UES, ideally, data at the city and county levels should be examined, where the pension pools are located. Unfortunately, such data are not available in the public domain. Instead, data on provincial-level data are primarily used. However, aside from the four provincial-level municipalities of Beijing, Shanghai, Tianjin and Chongqing, an average province has around 100 cities and counties, with the largest – Sichuan, having 204 – provincial-level data are highly aggregated and mask many inter-local differences (National Bureau of Statistics, 2018[17]). Whenever possible, information from local pension pools collected during fieldwork is used to supplement them.

Under the decentralised UES, pension benefits differ substantially across regions. In 2015, the average monthly pensions ranged from CNY 1 816 in Chongqing to CNY 3 644 in Shanghai, although their replacement ratios (derived as average benefits divided by the provincial average wage) were less dissimilar: 36% in Chongqing and 40% in Shanghai. As can be seen in Figure 4.4, however, there are substantial variations in the replacement ratio across provinces, ranging from 36% in Chongqing to a high of 63% in Shanxi. While the trend line suggests a positive relationship between pension benefits and the replacement ratio (with higher wage areas providing pensions at a higher replacement ratio), there are significant deviations, with the rich, coastal provinces of Beijing, Tianjin, Guangdong, Zhejiang and Jiangsu offering some of the lowest replacement ratios. Death benefits, too, can differ significantly across localities, ranging from around CNY 4 000 in the Gansu province in the west to more than CNY 50 000 in the coastal Shandong province.9

Even with the substantial variations in benefit levels that could help to alleviate pressures, many pension pools are unable to cover their costs. In 2010, it was reported that 14 provinces were in deficit and required subsidisation, including the rich coastal city of Shanghai (Fang and Feng, 2018[2]). In 2017, data from provincial budgetary accounts showed that among the 13 provinces reporting, 8 had expenditures exceeding receipts in their UES accounts before subsidies. In Jilin province, expenditures were nearly double what they took in from contributions (Table 4.2). At the sub-provincial level, the gaps may be larger still. In one fieldwork county, annual receipts from contributions are able to cover only one-third of pension outlays.10

Several studies have noted fragmentation as a salient feature of China’s pension system and identified many problems that stem from it (World Bank, 1997[19]; West, 1999[15]; Zhao and Xu, 2002[20]; Cai and Cheng, 2014[21]; Zhang and Li, 2018[22]; OECD, 2019[23]; OECD, 2019[24]). Fragmentation of the pension system, especially in the presence of significant income disparities, places barriers on inter-regional and social mobility in the labour market (Zhang and Li, 2018[22]). It hinders the system from capturing the gains of increased risk sharing or taking advantage of the economies of scale that could be achieved in administration. Administrative costs are high, estimated by the Ministry of Finance (2016[25]) to be almost 4% of collections. Supervision and regulation are complex and challenging, and the lack of accurate information about local pools hinders long-term, strategic planning for social protection (Lou, 2019[26]).

In many respects, the deep problems of the UES are intergovernmental in nature, of managing across levels of government. The UES is a social insurance programme that is mainly managed by the cities and counties, with some provincial-level involvement but ultimately guaranteed by central government. This disarticulation between central policy making and fiscal responsibility from local policy implementation and fiduciary management has created serious agency and moral hazard problems.

One manifestation is the trend of growing budget subsidies to the UES, which is shown in Figure 4.5. Since 2010, subsidies have more than quadrupled, from CNY 195.4 billion to CNY 800.4 billion in 2017. Even since 2014, it has more than doubled in size and now covers 21% of national outlays in the programme. Mr Jiwei Lou, the just-retired Chairman of the National Social Security Fund and former Minister of Finance, pointed to this rapid growth in subsidisation as indicative of “too many defects in the system” and attributed them to the moral hazard arising from decentralised and fragmented management (Lou, 2019[26]).

Non-compliance is a pervasive feature of this disarticulated system, and it comes in many forms. The first and most glaring is the fragmentation itself – the persistence of pension management at the county and city level, long after it ceased to be the government’s model of choice.

From the start of pension reform, policymakers were concerned that disparities in contribution and benefit levels would place unequal burdens on enterprises across pension pools, creating an un-level playing field that could perpetuate and exacerbate long-term regional inequalities. To tamp down these disparities, the government began to promote a two-pronged strategy: consolidation of pools at the provincial level as a first step toward eventual national integration, and harmonising contribution rates across pools.

As early as 1991, official documents called for the UES pension pools to be consolidated to the provincial level and eventually to the national level (State Council, 1991[27]). These calls have been reiterated periodically, but have found little traction (State Council, 1995[28]; State Council, 1997[3]; Ministry of Labour and Social Security and Ministry of Finance, 2007[29]; National People’s Congress, 2010[30]). The explicit redistributive intent of consolidation makes the effort unpopular with rich pension pools. When threatened with consolidation and losing their reserves, local governments reduced collection efforts, down-sized surpluses and even created deficits. Zhao and Xu (2002[20]) attributed two waves of early retirements to local government resistance to consolidation efforts – the first in the early 1990s after the State Council 1991 document called for moving pension pooling to the provincial level, and the second in the late 1990s, following the closure of industry pension programmes and merging them with local pension pools.11 Similar tactics can be seen in recent times. Over the past decade, under the growing threat of consolidation since the promulgation of the 2010 Social Insurance Law, local governments have adopted a variety of ways to reduce surpluses and accumulate future liabilities. For example, it was learned in fieldwork that some of the cash-rich pension pools in Guangdong province are busy enrolling local residents in the UES regardless of their employment status and eligibility, thereby shifting pension liabilities from the budget-funded BRS to soak up some of the surpluses in the UES. Even in deficit counties, officials are promoting “buy-ins” to the UES to collect an upfront payment in exchange for future liabilities (see Box 4.1).

Whether it was due to such resistance or the lack of fiscal capacity or political will, to date the effort to promote consolidating pension pooling has been largely stymied. As of 2017, only six provinces have unified the collection of premium contributions, benefit payments, and management of reserves. Of the six, four are provincial-level municipalities (Beijing, Tianjin, Shanghai, and Chongqing), leaving only two provinces that are managing pension pools. Another 13 provinces have set up provincial risk funds that hold reserves by collecting a contribution from the local pools, to be used to bail out shortfalls when necessary (Zhang and Li, 2018[22]). In 12 provinces, cities and counties remain entirely on their own in managing their pension programmes.

The effort to harmonise contribution rates to eliminate unfair inter-regional competition, which intensified from 2006 onwards, has met with a similar fate, albeit in a more subtle way. The 2010 Social Insurance Law stipulates contribution rates of “not more than 20%” from employers, 8% from employees, and committed central and local government subsidies to meet shortfalls. Official data seem to show pension pools mostly complying with rates that are at or near the targeted levels, except for the Guangdong and Zhejiang provinces (Table 4.3). Data available on cities, though, show that within provinces there are some differences in reported rates. In Fujian, for example, Xiamen, which has a special status as a semi-provincial level city, has a reported rate of 12%, compared to 18% in Fuzhou, the provincial capital of Fujian. In all cases, though, the provincial governments appear to play a role in setting contribution rates for the lower level pools (State Council, 1995[28]; 1997[3]).

These official contribution rates, however, are undermined by the existence of widespread and serious problems of underpayment that are found in more granular data and from fieldwork investigations. Nyland, Smyth and Zhu (2006[32]) reported that 71% of a sample of 2 600 firms audited in Shanghai in 2001 were found to have paid less than their mandated social security contributions.12 Recent studies have similar findings. Zheng (2016[33]), for example, found that nationwide, 70% of firms had paid less than the prescribed levels of contributions in 2015.13

Under-reporting wages is a common tactic for getting around mandated contribution rates. One version of this is deliberately choosing the wrong basis for calculation. For example, in 2015 employers in Beijing were found to have paid only 64% of their mandated contributions because they used imputed wages based on the previous year’s “local average wage”, rather than actual payroll, in making pension contributions (China International Economic and Exchange Center, 2017[34]). In fact, using imputed wages based on the “local average wage” is widely adopted even though the Social Insurance Law stipulates that “the employer shall pay the UES insurance premium in proportion to the total wages of the employees of the unit” (National People’s Congress, 2010[30]). In Hunan, provincial authorities go further, by allowing local pension pools to use the 2016 local average wage as the pension contribution basis for three years in a row in 2017, 2018 and 2019, thus increasing the amount of underpayment each year.14

The practice of using imputed rather than actual wages and applying a band of 60–300% of the local average wage as the contributing basis was introduced in the 1990s. It was applied to the employees’ contributions to their individual accounts – as distinct from employer contributions, and the 60% minimum was meant to prevent firms from under-reporting and using artificially low wage figures (West, 1999[15]). Over time, however, imputed wages came to be widely adopted in calculating employers’ contributions as well. Deviating from actual wages allowed for further distortions in the calculation of pension contributions, by grouping employees into a few tiers. For example, a 2018 document from the Beijing Social Security Bureau specified the contributing base wages to be applied for four categories of employees, those paying at 300% of the local average wage, and at 70%, 60% and 40%.15 A 2018 survey conducted by 51Shebao, a social insurance information service provider, found that only 27% of companies have made full payment of their social insurance contributions, while 31% paid at 60% of the local average wage (Cheng, Yu and Han, 2018[35]).

Local governments condone these underpayments. In one small county visited by the authors in a central province, officials in the social security bureau admitted that even though it violates the Social Insurance Law’s mandates, most private companies in the county are allowed to pay social insurance premiums at 60% of the local average wage, if at all. According to one official, “We do want to collect the right amount of pension contributions, but we can’t kill these companies only for the sake of doing that.” In that county, most, but not all, state-owned enterprises pay pension contributions according to the real level of salaries. One of them, a local branch of China Tobacco, the highly lucrative state monopoly for tobacco products, pays pension contributions at the upper end, at 300% of the local average wage.16

In China’s decentralised system, where local governments compete fiercely for economic growth and investment, firms with mobile capital have significant bargaining power. Indeed, the officials in the visited central provincial county lamented a lost opportunity in 2017, when a large heavy machinery manufacturer was considering building a plant in the local industrial zone. The company demanded a concessionary pension contribution rate of 14%, which the county government supported. However, the provincial government turned down the special request, and the firm went elsewhere.

In this environment, it comes as no surprise that a large employer like Foxconn, which employs more than 350 000 workers to make iPhones in Zhengzhou, can operate with impunity in its labour practices. China Labour Watch, an international watchdog non-governmental organisation (NGO), reported that it found the workforce in the Foxconn facilities in Zhengzhou to comprise about 50% of temporary workers in August 2019, in gross violation of the 10% upper limit imposed by the Chinese Labour Law. This came after it had been condemned in 2018 for using temporary workers to fill 55% of the workforce, including high school students and student interns, and vowed to do better under pressure from Apple (Gurman, 2019[36]). Many of these temporary workers may not be covered under social insurance.

As for the fiscal backstop for the UES, the 2010 Social Insurance Law stipulates that subsidies would be provided in cases of a shortfall in the social insurance accounts, shared between the central and local governments. However, it is silent on the basis for the allocation and apportionment of responsibility between the central and local governments. In practice, the annual allocation is ad hoc for nearly all transfers in the central government budget.17 Their distribution across provinces appears to be similarly ad hoc and negotiated. For example, to cover the large deficit and to meet social security payments obligations in 2016, the province of Heilongjiang had to negotiate for a special loan from the central government on top of the subsidy it received.18

In provinces that have set up provincial risk pools, the coverage of pension shortfalls may be treated more systematically, but there is little public information about the arrangements. In the other provinces, case-by-case negotiations appear to permeate the system at all levels. A Hunan provincial government document on pension policy explains that the province shares the responsibility for pension subsidies with prefectural and county-level governments “based on factors of the local economy and society such as medium-to-long-term socio-economic development planning, economic development status, demographic structure, employment situation, fiscal capacity, number of participants in the pension insurance schemes and dependency ratio” (People’s Government of Hunan, 2019[37]).

Lacking accurate information on the local pension pools and given weak enforcement capacity, the system often falls back on a form of contracting.19 In the poor central China county visited, officials explained that the county is given an annual quota for pension contributions, which they must fulfil before higher levels will provide assistance, and then only after much pleading and sometimes theatrics.20 This echoes a report from the Chongqing Municipality, where a deputy director in the social security bureau explained that they treat each county or district differently depending on how the targets for pension revenues and expenditures are met each year (Xinhua Net, 2019[38]).

Over the past decade and especially since 2013 under Xi Jinping, China has rolled out a co-ordinated programme of reform and institution building aimed at creating a more centralised and rules-based system of governance and public service delivery (Wong, 2016[39]; Wong, 2018[40]; Gruenberg and Drinhausen, 2019[41]). To combat the longstanding weaknesses in policy implementation, especially in monitoring and enforcement at the local levels, reforms have over the past six years focused on strengthening top-down vertical controls, boosting the legal framework and monitoring capacity, with inspections being the hallmark of monitoring efforts in recent years.21 Examples include reforms in public financial management processes that have greatly improved the information content and reporting of budgets at all levels of government, assigned responsibilities to various agencies and levels, built performance evaluations into the budgeting process, and subjected all public spending to regular, external audits. In environmental protection, another key policy area prioritised by the top leadership, laws, regulations and action plans have proliferated to support climate goals. Reforms have also assigned more resources and elevated the status of environmental agencies, and created inspection mechanisms to strengthen monitoring and enforcement capacity to ensure implementation at the lower levels (Wong, 2019[42]).

Social security reform is part of this agenda, with a co-ordinated and highly ambitious programme aimed at addressing critical issues in financing, incentives, and administration of the UES. The efforts began with the introduction of stricter rules for accounting and transparency in 2010, when the State Council called for the creation of a “Social Insurance Fund Budget” (SIFB), and laid out the principles, scope and methods of compilation (State Council, 2010[43]). The SIFB would include the accounts of the public pension schemes (the UES and, at the time, the New Rural Pension Scheme), the medical insurance schemes, and the unemployment insurance, disability insurance, and maternity insurance programmes.22 The Social Insurance Law, passed in October 2010, stipulated that financial accounts for each of the social insurance programmes must be reported in the SIFB using a uniform system of national accounting standards (National People’s Congress, 2010[30]).

When the revised Budget Law was promulgated in 2015, the SIFB came to be included as one of four budgets that governments at all levels are required to compile and publish on-line within 30 days of their approval by the People’s Congress (Wong, 2016[39]; Wong, 2018[40]).23 As the National Audit Office has expanded authority to conduct regular, rather than ad hoc, audits of the budget, its oversight responsibility extends to the social insurance funds as well (Wong, 2019[44]).

More dramatic reforms came only recently. For the UES, one of the most important changes is an overhaul of the system of collecting social insurance levies. In March 2018, the government announced that starting from 1 January 2019, tax bureaus would be responsible for the collection of all social insurance premiums nationwide, replacing the previous arrangement where they were collected either by the social security bureaus or local tax bureaus (CPC Central Committee, 2018[45]). Significantly, an announcement soon followed that implementation is put off temporarily to allow for preparations, and the State Council called for a phased, step-by-step, province-by-province transition (State Council, 2019[46]).

This is a dramatic change. Although many provinces had turned over the collection task in recent years to the local tax bureaus, this change is significant because it comes on the heels of the consolidation of tax administration under the central government in June 2018 (CPC Central Committee and State Council, 2018[47]).24 When implemented, it will potentially remove local governments from the assessment and collection of pension contributions and transform the social security contributions into a tax obligation – closing a loophole and plugging a significant gap in compliance.

To improve incentives, and betting that improved collection will enable the government to reduce contribution rates while maintaining solvency of the system, in April 2019, the State Council issued a document allowing provincial governments to reduce the pension contribution rate to 16% for employers (while keeping employee contributions unchanged at 8%) (State Council, 2019[46]). This follows a 1% reduction that had been allowed in 2016 for nine months as a stimulus measure in response to a slowing economy (Ministry of Human Resources and Social Security and Ministry of Finance, 2016[48]), and which had been extended for another year in 2018 (Ministry of Human Resources and Social Security and Ministry of Finance, 2018[49]).

Other reforms outlined in the 2019 State Council document include a call for the consolidation of the UES at the provincial level to be achieved by 2020. It also called for the creation of a Central Adjustment Fund (CAF) as a risk-sharing mechanism managed at the central level to boost the central government’s capacity to provide funding to bail out ailing local pension pools.

The CAF began operation in July 2018. It is designed to draw from each province a set percentage of UES pension revenues, with the pooled funds returned to the provinces by a formula taking into account the number of retirees in each province (State Council, 2018[50]). At the start, the fund drew 3% of “base revenues” from each province, with formulas set as follows:

  • Contribution from province i = 3% * (90% of the average wage in province i * number of UES participants in province i)25

  • Disbursement to province i = the number of retired persons in province i * national disbursement per retiree; with

  • National disbursement per retiree = total in CAF / total number of retirees in UES

In April 2019, the first accounting of the CAF was published, showing a total CAF pool of CNY 484.5 billion collected and disbursed. Guangdong, the richest provincial pool, contributed CNY 74.2 billion and drew CNY 26.8 billion from the CAF. At the low end, Liaoning contributed CNY 13.1 billion and received CNY 34.7 billion (Table 4.4). Even at this initial rate of 3% contribution, the CAF is already comparable in size to the central government’s current subsidies to the UES, which totalled CNY 528.5 billion in 2019 (Xinhua Net, 2019[51]). With the contribution rate raised to 3.5% in 2019 and further in the future, the CAF is designed to alleviate fiscal pressures. Equally important is that the CAF is seen as the first step toward the eventual integration of the pension system at the national level.

These are early days, and much work remains to be done, including the verification of the parameters underlying the CAF calculations, but this appears to be an important foundational first step towards building a nationally integrated system of social insurance for urban employees.

In summary, after a long lag, the pace of reform in the pension system accelerated sharply in 2018. As many of the changes were only introduced in 2018 and 2019, and are therefore just underway or are still under preparation, it is much too soon to assess their impact, especially given the long history of reforms that have been thwarted by local resistance.

What is different this time is the co-ordinated nature of the reform measures. When completed, the accounting and public financial management reforms will improve the uniformity and veracity of reporting on social insurance accounts. Including the Social Insurance Budget in the annual budget report to the National People’s Congress and sub-national congresses has elevated the issues of social insurance on the policy agenda. Moving the collection of social insurance contributions from the social insurance departments to the State Tax Administration will remove local government meddling and improve compliance with contribution requirements, and the creation of the CAF has enlarged the risk pooling. When implemented, these reforms will bring far-reaching changes and, in the aggregate, go a long way toward strengthening the governance and sustainability of the programme.

To strengthen the financial sustainability and fairness of the pension system further, reforms must improve the incentives for attracting and retaining the younger migrant workers and service sector workers – important concerns that are beyond the scope of this chapter.


[31] 51-HR (2017), Human Resources Management Commonly Used Standard Data Query, (accessed on  March 2017).

[60] Bureau of Human Resources and Social Security of Beijing (2018), Notice on Unifying the Salary Bases and Contributions of Various Social Insurances in 2018,

[21] Cai, Y. and Y. Cheng (2014), “Pension Reform in China: Challenges And Opportunities”, Journal of Economic Surveys, Vol. 28/4, pp. 636–651,

[35] Cheng, S., H. Yu and W. Han (2018), “In Depth: China’s Social Security Balancing Act”, Caixin,

[34] China International Economic and Exchange Center (2017), Accelerate the implementation of national pooling of the basic employees pension scheme to further improve the integrated account for two pillar multi-level pension system.

[10] Clements, B., F. Eich and S. Gupta (eds.) (2014), Reforming Pensions to Ensure Equitable and Adequate Retirement Incomes in China, International Monetary Fund, Washington, DC.

[45] CPC Central Committee (2018), Plan for Deepening the Institutional Reform for the Party and the State,

[47] CPC Central Committee and State Council (2018), Reform Plan for the Collection and Management System for State Tax and Local Tax,

[57] Department of Human Resources and Social Security of Gansu (2016), The Full Coverage of the Pension Insurance System in the Province since the 12th Party Congress in Gansu Province,

[58] Department of Human Resources and Social Security of Shandong (2019), What Will Happen When an Insured Person Pass Away: The Personal Account Can Be Inherited, and the Government Provides Subsidies and Relief Funds,

[59] Department of Human Resources and Social Security of Shandong (2017), Notice on the Announcement of the Salary Base for Calculating Contributions and Benefits for the Basic Employees Pension Insurance for 2017’. 2017,

[2] Fang, H. and J. Feng (2018), “The Chinese Pension System”, NBER Working Papers, No. 25088, National Bureau of Economic Research, Cambridge,

[14] Galenson, W. (ed.) (1993), Public Finance and Economic Decentralization, University of Michigan Press.

[40] Garnaut, R., L. Song and C. Fang (eds.) (2018), An Update on Fiscal Reform, ANU Press,’s-40-years-reform-and-development-1978–2018.

[41] Gruenberg, N. and K. Drinhausen (2019), The Party Leads on Everything: China’s Changing Governance in Xi Jinping’s New Era, Mercator Institute for China Studies.

[16] Guo, J. (2018), “The central pension adjustment system was implemented on July 1, and provinces with more retirees took more money”, Sina,

[36] Gurman, M. (2019), “Apple, Foxconn Broke a Chinese Labor Law to Build Latest iPhones”, Bloomberg,

[56] Hussain, A. (1994), “Social Security in Present-Day China and Its Reform”, American Economic Review, Vol. 84/2, pp. 276–280.

[7] Impavido, G., Y. Hu and X. Li (2009), “Governance and Fund Management in the Chinese Pension System”, IMF Working Papers, International Monetary Fund,

[11] ISSA (2016), Government of China Receives International Social Security Award,

[26] Lou, J. (2019), “Understanding the Budgetary Issues Based on the National Conditions”, Public Finance Research, Vol. 5, pp. 3–6.

[54] Ministry of Education (2018), Statistics Bulletin for Education Development in China in 2017,

[52] Ministry of Finance (2019), 2019 Central Transfer Fund Expenditure (Allocation) Table,

[53] Ministry of Finance (2019), 2019 Central Transfer Fund Revenue (Turn Over) Table,

[25] Ministry of Finance (2016), Statistics of Final Accounts of Revenues and Expenditures of Social Insurance Funds in 2015, Ministry of Finance, Beijing.

[49] Ministry of Human Resources and Social Security and Ministry of Finance (2018), Notice on the Extension for the Temporary Reduction of Social Insurance Contribution Rates,

[48] Ministry of Human Resources and Social Security and Ministry of Finance (2016), Notice on the Temporary Reduction of Social Insurance Contribution Rates,

[29] Ministry of Labour and Social Security and Ministry of Finance (2007), Notice on Promoting Provincial Coordination of Basic Pension Insurance for Enterprise Workers,

[1] National Bureau of Statistics (2018), China Labour Statistical Yearbook 2018, China Statistics Press, Beijing.

[17] National Bureau of Statistics (2018), China Statistical Yearbook 2018, China Statistics Press, Beijing.

[18] National Bureau of Statistics (2016), China Labour Statistical Yearbook 2016, China Statistics Press, Beijing.

[30] National People’s Congress (2010), Social Insurance Law of the People’s Republic of China.

[32] Nyland, C., R. Smyth and C. Zhu (2006), “What Determines the Extent to Which Employers Will Comply with Their Social Security Obligations? Evidence from Chinese Firm-Level Data”, Social Policy and Administration, Vol. 40/2, pp. 196–214,

[24] OECD (2019), Country Profiles: China, Pensions at a Glance 2019, OECD, Paris,

[23] OECD (2019), Pension Markets in Focus, OECD, Paris,

[55] People’s Government of Chongyi County (2016), National Poverty Standard by Year,

[37] People’s Government of Hunan (2019), Notice of the People’s Government of Hunan on Improving the Provincial-Level Coordination System for Basic Old-Age Insurance for Enterprise Employees,

[46] State Council (2019), Circular of the General Office of the State Council on Issuing a Comprehensive Plan for Reducing Social Insurance Contribution Rates,

[50] State Council (2018), Notice of the State Council on the Establishment of the Central Adjustment System of the Basic Pension Insurance Fund for Enterprise Employees,

[4] State Council (2015), Decision of the State Council on the Reform of the Old-Age Insurance System for Civil Servants and Public Institutions Employees.

[6] State Council (2014), Opinion of the State Council on the Establishment of a Unified Basic Old-Age Insurance System for Urban and Rural Residents.

[43] State Council (2010), Opinions on the Trial Implementation of Social Insurance Fund Budget,

[9] State Council (2009), Guidance of the State Council on the Pilot Reform of the New Rural Old-Age Social Insurance,

[5] State Council (2005), Decision of the State Council on Improving the Basic Old-Age Insurance System for Employees of Enterprises,

[13] State Council (2000), Notification of the State Council on Issuing the Pilot Programme for the Improvement of the Urban Social Security System,

[12] State Council (1998), Notice of the State Council on Issues Related to the Implementation of Provincial Coordination of Basic Pension Insurance for Enterprises Employees and the Transfer of Industry-Based Pension Insurance under Local Administration,

[3] State Council (1997), Decision of the State Council on the Establishment of a Unified Basic Old-Age Insurance System for Employees of Enterprises,

[28] State Council (1995), Notification on Deepening the Reform of the Old-Age Insurance System for Employees in Enterprises.

[27] State Council (1991), Decision on the Reform of the Old-Age Insurance System for the Employees of Enterprises.

[15] West, L. (1999), “Pension Reform in China: Preparing for the Future”, Journal of Development Studies, Vol. 35/3, pp. 153–83,

[42] Wong, C. (2019), More Sticks and Fewer Carrots? Looking Ahead to Implementation of Environmental Policy Under the “New Normal”.

[44] Wong, C. (2019), Public sector evaluation in the People’s Republic of China.

[39] Wong, C. (2016), “Budget reform in China: Progress and prospects in the Xi Jinping era”, OECD Journal on Budgeting, Vol. 15/3,

[8] Wong, C. (2010), “Paying for the Harmonious Society”, China Economic Quarterly, Vol. 14/2, pp. 20–25,

[61] Wong, C. (1992), “Fiscal Reform and Local Industrialization: The Problematic Sequencing of Reform in Post-Mao China”, Modern China, Vol. 18/2, pp. 197–227,

[19] World Bank (1997), Old Age Security: Pension Reform in China, World Bank, Washington, DC.

[51] Xinhua Net (2019), “660 billion state-owned capital “blood transfusion” social security fund”, Xinhua Net,

[38] Xinhua Net (2019), “Several Provinces Speed up Provincial Unification of Pension Insurances”, Xinhua Net,

[22] Zhang, L. and M. Li (2018), “Acquired but Unvested Welfare Rights: Migration and Entitlement Barriers in Reform-Era China”, The China Quarterly, Vol. 235/September, pp. 669–92,

[20] Zhao, Y. and J. Xu (2002), “China’s Urban Pension System: Reforms and Problems”, The Cato Journal, Vol. 21/January.

[33] Zheng, B. (2016), China Pension Report 2016, Economic Management Press, Beijing.


← 1. This is a term coined by Cai and Cheng (2014[21]).

← 2. There are no public accounts of the PES, so the actual enrolment in the PES is unknown. Estimates run from 37 million to nearly 50 million. Our estimate is based on current public employment of around 41 million, but among whom many of the staff at public institutions had been moved to the UES over the past decade, and they are offset by the retirees in the system.

← 3. During the transition period, benefits will revert to those under the old rules if benefits calculated under the new rules are lower.

← 4. In the 1990s, there were in some localities rural pension schemes funded by contributions from rural households and subsidised by rural collectives, set up under an initiative by the Ministry of Civil Affairs, but they were largely unsuccessful.

← 5. The potential target population consists of the population aged 15 and above, excluding full-time students in high schools, colleges and universities. In 2017, the population aged 15 and above in China was 1 156.6 million (National Bureau of Statistics, 2018[17]), while the number of full-time students in high schools, colleges and universities was 77.5 million (Ministry of Education, 2018[54]), resulting in a target population of 1 079.1 million.

← 6. The national poverty line in 2017 was CNY 3 335 for annual disposable income per capita (People’s Government of Chongyi County, 2016[55]). The average pension for the BRS was less than half of the poverty line.

← 7. From 2006 onwards, the individual accounts are funded only by employee contributions.

← 8. For the history of the Chinese pension system and reasons for its fragmentation, see (Hussain, 1994[56]; West, 1999[15]; Cai and Cheng, 2014[21]).

← 9. According to a document issued by the Department of Human Resources and Social Security of Gansu, the amount of death benefits for the UES was increased to CNY 3 676 in 2014, which was the latest adjustment to the death benefits in Gansu (Department of Human Resources and Social Security of Gansu, 2016[57]). According to the website of the Department of Human Resources and Social Security of Shandong, the amount of death benefits for the UES was set to equal to 10 months of the average monthly wage in the province in the previous year since 1993 (Department of Human Resources and Social Security of Shandong, 2019[58]). Therefore, in 2017, the death benefits were CNY 3 676 in Gansu and CNY 52 970 in Shandong, as the 2016 average monthly wage in Shandong was CNY 5 297 (Department of Human Resources and Social Security of Shandong, 2017[59]).

← 10. Fieldwork interviews in a central province in July 2019.

← 11. At the time the UES was created, 11 industries were allowed to set up their own social pension pools. These were shut down by order of the State Council in 1998.

← 12. The audit had covered only 2 600 firms out of 100 000.

← 13. Cited in Fang and Feng (2018[2]).

← 14. Fieldwork interviews in 2019.

← 15. For more information, see Bureau of Human Resources and Social Security of Beijing (2018[60]).

← 16. Fieldwork interviews in a central province in July 2019.

← 17. The exceptions are tax rebates, which are specified by law; and the funding for the equalisation transfer, which comes from a portion of the corporate income tax.

← 18. Interview with an expert on Chinese pension reform in Beijing in July 2017.

← 19. This situation is highly reminiscent of the revenue-sharing system before the Tax Sharing System reform was implemented in 1994, when the central government had little information about local tax bases and tax effort, and resorted to fiscal contracting to ensure it could extract at least some revenue from local governments (Wong, 1992[61]).

← 20. In one episode recounted in a poor county visited, the director of the Human Resources and Social Security Bureau said that he had to write special reports to the Provincial Department of Human Resources and Social Security to plead for the transfers needed, and forced the whole team in his bureau to work until midnight of the day before pension payments were issued, to collect enough to narrowly avoid a delay in payments (Fieldwork interviews in a central province in July 2019).

← 21. For other aspects of institution building in the Xi Jinping administration, see (Wong, 2018[40]).

← 22. There are currently eight programmes in the Social Insurance Fund Budget: the UES, PES, BRS, the Urban Employees Basic Medical Insurance, the Basic Residents’ Medical Insurance, and the Disability, Maternity and Unemployment insurance programmes.

← 23. The “four budgets” includes the general budget, the “government fund budget” comprising earmarked “funds” including revenues from land conveyance, the “state capital operating budget” of profit remittances from SOEs, and the SIFB.

← 24. The central and local tax administrations were merged in June 2018 under the State Tax Administration.

← 25. The document calls for using the total number of workers (all of whom should be enrolled) rather than the actual UES participants. As a compromise, the average of the two is used for the transition period.

Metadata, Legal and Rights

This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area. Extracts from publications may be subject to additional disclaimers, which are set out in the complete version of the publication, available at the link provided.

© OECD 2020

The use of this work, whether digital or print, is governed by the Terms and Conditions to be found at