The rural pension system in China

In rural China, due to the prevalence of the traditional idea that people should raise children in preparation for their old age, caring for the elderly people was a family matter prior to the middle of the 20^th century [14]. However, this traditional notion has changed dramatically since then owing to the country’s significant social and economic transformation, thereby the Chinese government began to play a major role in old age protection in rural China [15]. By the end of the 1990s, the Chinese authorities had initiated several pilot projects aimed at creating an old age insurance program in rural areas (also called the Old Rural Pension Scheme, ORPS). It has been argued that due to lack of trust in non-public pension plans, only about 54 million rural residents had participated in the ORPS by 2004, resulting in a failure of the pilot program [15]. Along with the new administration in the Chinese central government in 2002, the role of social protection, social welfare, and improvement in the well-being of socially vulnerable rural residents once more came to the fore [16]. Over the next few years, the Chinese government initiated several pilot programs once again based on the experience of ORPS. In 2009, a new rural pension plan called the New Rural Pension Scheme was set up for all rural residents.

The NRPS consists of a non-contributory basic social pension (SP) component and a funded defined contribution (FDC) component [14]. The basic pension is available to people who have reached the age of sixty. Those who had already reached this age when the NRPS was initially implemented were eligible to receive the basic pension even if they had never contributed to the scheme, provided that their adult children “voluntarily” enrolled in and contributed to the FDC component. The FDC component is an aggregate of the personal contributions and the subsidies provided by central and local governments. Those younger than sixty become eligible for the basic pension and an individual pension amount after having contributed to the FDC for at least fifteen years and having reached age sixty. There are a variety of alternative contribution levels, ranging from ¥100 to ¥500 per year [17]. If one contributes more to the FDC each year, she/he can get more pension benefits when reaching age sixty. In 2014, the central government decided to merge the NRPS and the Urban Residents Pension Plan and build a unified basic pension insurance plan for both urban and rural residents. In addition, on the basis of residents’ previous contribution levels, several other contribution choices (¥600, ¥700, ¥800, ¥900, ¥1,000, ¥1,500 and ¥2,000 per year) were made available [18]. Table 1 presents a brief overview of China’s rural pension reforms.

Table 1

Overview of the rural pension systems in China.

Year	Pension scheme	Pension amount (¥/month)	Number of participants (million RMB)	Nature of participation	Source
1999	Old Rural Pension Scheme (ORPS)	2–20	80	Voluntary	Liu and Sun (2016) [15]
2009	New Rural Pension Scheme (NRPS)	55–593	90	Voluntary	Deng and Xue (2010) [19]
2014	A combination of NRPS and Urban Residents Pension Plan	84–2800	498	Voluntary	Huang (2015) [20]

Notes: ¥6.8 (RMB) = $1 (USD).

Gansu Province and the New Rural Pension Scheme

Gansu Province is located in the northwest of China and composed of fourteen prefectures. The land area of Gansu Province is 42.6×10⁴ square kilometres, covering approximately 4.7 per cent of the total area of China. The total population of Gansu was 26.1 million in 2018, with 55.3 per cent living in rural areas [21]. In 2009, the population of the elderly people aged over sixty was 3.18 million in Gansu, accounting for 12.4 per cent of the province’s total population [22]. Gansu Province is one of China’s least developed regions, with relatively poor agricultural production conditions and a low level of urbanisation. The per capita disposable income of rural residents in Gansu was about ¥8,804 in 2018, much lower than the national average (¥14,617) [21].

The New Rural Pension Scheme provides pensions for millions of rural residents and represents a major step forward for the rural pension reform in China. Fig 1 shows the changes in the rural population and insured rural residents in Gansu Province and the whole country from 2009 to 2018. Since the implementation of NRPS in 2009, there has been a rapid increase in coverage over the past seven years. In Gansu Province, about 13.2 million (96 per cent) of rural residents were covered by the end of 2018 (Fig 1). This includes 3.2 million residents over the age of sixty–almost the entire rural retirement-age population in Gansu Province [21].

Fig 1

Trends in rural population and coverage of NRPS in China (a) and Gansu Province (b). Data source: China Statistical Yearbook.

Actual replacement rate

This study follows Wang & Béland (2014) to build the actuarial model. The model is based on several assumptions: (1) the New Rural Pension Scheme policy framework will remain stable until 2058; (2) both central government and Gansu provincial government subsidies are collectively referred to as government subsidies; (3) according to the People’s Republic of China Social Insurance Law, the central government will adjust the basic pension and funded defined contribution subsidies based on the growth of residents’ income and inflation level; (4) because multiple payment were added in 2014 when the NRPS and the Urban Residents Pension Plan were merged, 2014 was set as the baseline year; (5) at the beginning of each year, the insured residents routinely pay the FDC component and the payment level they have selected remains stable for all years thereafter; (6) for the elderly people, all their adult children have participated in the NRPS and they can thus receive the basic pension amount.

Actual replacement rate (ARR) actuarial model

Considering that the elderly people only have access to the SP component, the ARR values of the elderly, middle aged and young people are calculated separately. The ARR of the basic pension of the elderly people is ARR_elderly, expressed as:

The pension income for the middle aged and young people after they retire consists of the basic SP pension and their personal FDC pension. Therefore, their ARR includes two parts, i.e., ARR₁ for the basic pension and ARR₂ for the personal pension. The ARR₁ can be expressed as:

According to the third assumption above, the adjustment of the basic pension coincides with the growth of rural residents’ per capita net income, and thus the two parameters g₁ and g₂ are considered to be equal (replaced by g). Therefore, the Eq (2) can be further expressed as:

The ARR for the middle aged and young people is denoted by ARRyoung+middle, expressed as:

Parameters of the ARR actuarial model

According to The Guidance for Carrying Out the NRPS and the People’s Republic of China Social Insurance Law, this study utilizes the following parameters in the ARR actuarial model (Table 2):

Interest rate of personal account (e). The interest rate for personal accounts is equal to China’s annual bank interest rate on deposits [21]. This study uses an average interest rate from 2008 to 2019, and e ≈ 3 percent.

Inflation rate (h). According to the announcement issued by the People’s Bank of China, the inflation rate is expected to be maintained at about 2 percent over following years [21]. Thus, the inflation rate is set at h = 2 percent.

Growth rate of per capita net income of rural residents (g). According to the statistics of the net income of rural residents in Gansu Province from 2008 to 2019 [21], this model uses the growth rate at g ≈ 6 percent.

Mean number of years for which rural residents can enjoy the pension benefits (m). According to The Guidance for Carrying Out the NRPS, the average number of years is calculate and equal to 11.58 years (i.e., 139 divided by 12) [26]. Thus, the model parameter m = 12.

Government subsidies (T). If the insured residents pay ¥100-¥400 for FDC each year, the central government will subsidise them with ¥30 per year; if they pay ¥500 or more, they will obtain ¥60 from the government per year [13]. Thus, T is equal to ¥30 (resident payment ¥100-¥400) or ¥60 (resident payment ≥ ¥500).

Basic pension (P₀). The central and provincial governments provide ¥88 and ¥15 per month, respectively, for each insured resident [13]. Thus, the basic pension amount, P₀, is equal to ¥1,236 per year.

Per capita net income (Y₀). According to the China Statistical Yearbook, the per capita net income of rural residents is calculated at Y₀ = ¥5,107.76 (EBGDY, 2019).

Table 2

Parameters used in the ARR actuarial model.

Parameters	h (%)	e (%)	g (%)	m (years)	P0 (¥)	Y0 (¥)	ARRelderly (%)	T (¥/year)
								Payment of 100–400	Payment of 500 or more
Value	2	3	6	12	1236	5107.76	24.2	30	60

Notes: h, inflation rate; e, interest rate of personal account; g, growth rate of per capita net income; m, the mean time period to receive pension benefits; P₀, the basic pension amount in 2014; Y₀, the per capita net income of rural residents in 2014; T, annual subsidy per person.

¥6.8 (RMB) = $1 (USD).

Extended linear expenditure system model

To calculate the MLSRR, we first predict rural residents’ basic living expenses in the studied time period. We employ an extended linear expenditure system (ELES) model to measure the basic living expenses and then calculate the MLSRR by dividing the predicted basic living expenses by the per capita net income.

The linear expenditure system (LES) is developed to examine the demand relationships of various commodities, and the extended linear expenditure system (ELES) has two modifications based on the LES model. First, the total consumption expenditure is substituted by income to obtain exogenous variables, through which the consumption-saving decisions are also included in the model. Second, the marginal budget is replaced with marginal propensity to consume. The ELES model divides people’s needs for various commodities into two parts: basic demand and demand beyond the basic demand. This study considers the protection purpose of NPRS in the poor regions, and thus the demand for commodities by rural residents refers to the basic demand. In addition, the ELES model adds three assumptions:

In a fixed period of time, residents’ demand for various commodities depends on their income and commodity prices;

The basic demand is not associated to the resident income;

The marginal budget is not related to the consumption.

According to Gansu’s Development Yearbook, the rural residents’ living expenditure can be categorized into eight types: food, clothing, residence, household facilities, transport and communication, cultural and recreation education services, health care and medical services, as well as miscellaneous goods and services. According to the ELES, the total consumption of goods consists of two parts: the basic demand for daily necessities and the extra demand [27]. The basic demand for daily necessities is independent on residents’ income. Because almost all of the pension benefits are used to meet the elderly person’s basic needs in China’s poor regions [18], the modified living standards replacement rate (MLSRR) is adopted in this research, which removes the extra demands for commodities.

The basic expression of the ELES model can be expressed as:

Since the price (p_i) and the actual demand (q_i) of the ith good consist of household-level data that are not available, the following equation is derived:

We can substitute Eqs (7) into (6) and obtain:

Table 3

Parameter estimation of the ELES model.

Year	Parameters	Food	Clothing	Residence	Household facilities, articles and services	Transport and communications	Education, cultural, recreation and services	Health care and medical services	Miscellaneous goods and services
2007	ai’	551.053*	32.467**	37.036	-14.67	19.024	1.39	6.489	-0.848
	bi’	0.169*	0.034*	0.111*	0.045**	0.072*	0.089**	0.062**	0.013**
	R2	0.996	0.997	0.994	0.968	0.994	0.975	0.976	0.982
2008	ai’	674.590*	51.14**	39.279	-2.857	57.178***	-2.315	13.969	2.02
	bi’	0.168*	0.031*	0.128**	0.036**	0.065*	0.082**	0.055*	0.011**
	R2	0.998	0.995	0.968	0.968	0.993	0.985	0.99	0.969
2009	ai’	710.914*	48.728**	87.626	11.42	51.431***	6.118	24.364***	-3.079
	bi’	0.145*	0.036*	0.188**	0.044**	0.063**	0.071*	0.052*	0.015**
	R2	0.997	0.99	0.965	0.930	0.988	0.991	0.997	0.988
2010	ai’	837.453*	74.013**	-16.25	34.690***	32.484	7.451	20.984	-0.154
	bi’	0.140*	0.032*	0.165*	0.033*	0.066*	0.067*	0.053*	0.013*
	R2	0.989	0.985	0.959	0.982	0.98	0.996	0.971	0.982
2011	ai’	949.97*	104.697**	288.834**	74.226**	72.657	-15.035	147.358***	-2.551
	bi’	0.153*	0.036**	0.079*	0.032**	0.075*	0.079*	0.049**	0.020*
	R2	0.994	0.976	0.987	0.988	0.981	0.987	0.967	0.992
2012	ai’	949.763*	90.267*	248.309***	79.843***	117.542	-13.117	136.217**	0.356
	bi’	0.155*	0.047*	0.097**	0.038*	0.071**	0.075*	0.058**	0.022**
	R2	0.996	0.998	0.929	0.984	0.944	0.994	0.988	0.986
2013	ai’	1272.285*	218.382*	519.219***	184.557*	218.015***	86.639	361.014**	60.845**
	bi’	0.103*	0.026**	0.054***	0.023**	0.074**	0.055**	0.03***	0.012**
	R2	0.994	0.986	0.814	0.985	0.959	0.978	0.86	0.973
2014	ai’	1330.351*	249.725*	561.770**	195.390**	315.585**	145.689	277.706**	68.699***
	bi’	0.109**	0.023*	0.054**	0.021***	0.052**	0.045**	0.040**	0.009***
	R2	0.969	0.999	0.817	0.909	0.976	0.934	0.935	0.808
2015	ai’	1507.663*	298.211**	658.824**	297.940**	463.603**	537.605**	235.055***	73.523**
	bi’	0.106**	0.024**	0.081**	0.021***	0.050**	0.046**	0.063**	0.006**
	R2	0.985	0.978	0.944	0.919	0.923	0.969	0.953	0.969
2016	ai’	1328.069*	296.870**	847.919**	282.554**	541.695**	725.582*	222.408***	49.258***
	bi’	0.057**	0.025**	0.064**	0.024**	0.055**	0.032**	0.080*	0.010*
	R2	0.976	0.978	0.962	0.956	0.927	0.959	0.977	0.959
2017	ai’	1693.24**	324.852*	706.294	311.555**	433.49**	775.637**	297.624**	64.376***
	bi’	0.092**	0.023*	0.063	0.021**	0.072**	0.027***	0.073*	0.009**
	R2	0.964	0.992	0.282	0.971	0.984	0.805	0.992	0.961
2018	ai’	1918.35*	440.103**	1416.661**	383.792**	751.025**	1046.761**	748.089**	65.3***
	bi’	0.088**	0.013***	0.035***	0.015	0.037**	0.018	0.044**	0.011**
	R2	0.983	0.876	0.818	0.768	0.961	0.670	0.959	0.951

Note: ***p < 0.05,

**p < 0.01,

* p < 0.001. ai’ and bi’ are the fitted values of ai and bi.

Data and analytical procedures

The data related to rural residents’ net income and their expenditure were obtained from the Gansu Development Yearbook [13, 28]. The data on rural population and NRPS coverage were drawn from the China Statistical Yearbook [21]. In addition, the data about rural residents’ income and consumption in Gansu Province were from the Editorial Board of Gansu Development Yearbook from 2007 to 2019 (http://www.gstj.gov.cn). The data about macroeconomic indicators were from the People’s Bank of China (http://www.pbc.gov.cn/).

All the statistical analyses were conducted using Stata Version 14.0. First, the actuarial model of the NRPS was constructed using the actuarial principle, and subsequently the actual replacement rate (ARR) was measured and predicted. Second, the extended linear expenditure system (ELES) model and the OLS method were used to measure rural residents’ expenditure on the basic consumption of items falling into the eight categories between 2008 and 2019. Then, the linear fitting was used to calculate the rural residents’ basic expenditure and per capita disposable income between 2020 and 2058. Finally, the modified living standards replacement rate (MESRR) was calculated.

Actual replacement rate

As a benchmark, the actual replacement rate of the elderly people’s basic pension, ARR_elderly, is found to be 24.2 percent. Table 4 presents the results of the ARR actuarial model for middle aged and young people. The actual replacement rate increases as the individuals’ FDC payment level increases. Similarly, at each payment level, the ARR_young+middle increases as the length of payment increases (i.e., the younger the resident starts to enroll in the program). For example, at the contribution level of ¥100, the ARR increases from 24.4 for a resident starting enrolment at 59 years old to 29.2 starting at 16 years old. Especially, at the high payment levels (e.g., ¥1,500 and ¥2,000), the ARR increases dramatically as the insured age decreases.

Table 4

The predicted actual replacement rate (ARR, %) of the insured residents aged below 60 (i.e., the middle aged and young people).

Age insured	Year eligible to receiving pension benefits	Contribution levels (¥)
		100	200	300	400	500	600	700	800	900	1000	1500	2000
16	2058	29.2	33.1	36.9	40.8	45.8	49.7	53.5	57.4	61.3	65.1	84.4	103.7
19	2055	29.0	32.8	36.5	40.2	45.1	48.8	52.5	56.2	59.9	63.7	82.3	100.9
24	2050	28.7	32.0	35.6	39.1	43.6	47.1	50.6	54.0	57.5	61.0	78.3	95.6
29	2045	28.3	31.5	34.7	37.8	42.0	45.1	48.3	51.4	54.7	57.8	73.7	89.5
34	2040	27.9	30.7	33.5	36.4	40.1	42.9	45.7	48.5	51.4	54.2	68.4	82.5
39	2035	27.4	29.8	32.2	34.5	37.8	40.1	42.7	45.2	47.6	50.0	62.2	74.4
41	2033	27.1	29.4	31.7	33.9	36.9	39.1	41.2	43.7	45.9	48.2	59.5	70.8
44	2030	26.8	28.8	30.7	32.7	35.3	37.3	39.3	41.2	43.2	45.2	55.1	65.0
49	2025	26.1	27.6	29.0	30.1	32.3	33.8	35.3	36.7	38.2	39.7	46.9	54.2
54	2020	25.3	26.2	27.0	27.9	29.0	29.8	30.4	31.3	32.4	33.2	37.5	41.7
59	2015	24.4	24.6	24.7	24.9	25.1	25.2	25.4	25.5	25.7	25.8	26.6	27.3

Notes: the baseline year is 2014.

Modified living standards replacement rate

Table 5 presents the estimated basic consumption and modified living standards replacement rate (MLSRR) of the elderly. The MLSRR of the retired residents in 2017 and 2018 is 92.0 percent and 88.2 percent, respectively. Because that the per capita net income is not available for 2019 and after, the MLSRR is forecast from 2019 to 2058 (Table 6). The model forecast indicates that the MLSRR will experience a hump-shaped trend from 2019 to 2058. Specifically, from 2019 to 2023, the MLSRR shows a slight increasing trend and gets the peak at 102.1 percent in 2023. Subsequently, the MLSRR experiences a downward trend and stands at 40.6 percent in 2058.

Table 5

The results of basic consumption expenditure and modified living standards replacement rate (MLSRR) for the elderly from 2007–2016.

Year	∑αi	1-∑βi	piri (¥)	Y (¥)	MLSRR (%)
2007	631.9	0.41	1560.3	2328.9	67.0
2008	833.0	0.42	1964.6	2723.8	72.1
2009	937.5	0.39	2428.8	2980.1	81.5
2010	990.7	0.43	2298.5	3424.7	67.1
2011	1620.2	0.48	3396.6	3909.4	86.9
2012	1609.2	0.44	3682.3	4506.7	81.7
2013	2920.9	0.62	4688.5	5107.8	91.8
2014	3144.9	0.65	4860.8	5736.0	84.7
2015	4072.4	0.60	6753.6	6936.0	97.4
2016	4294.4	0.65	6576.3	7457.0	88.2
2017	4607.1	0.62	7430.8	8076.0	92.0
2018	5470.1	0.76	7159.8	8804.0	81.3

Notes: ∑α_i: sum of α_i’ estimates of the eight commodities; ∑β_i: sum of β_i’ estimates of the eight commodities; p_ir_i: total basic expenditure of the eight commodity categories; Y: per capita net income.

Table 6

The prediction results of the modified living standards replacement rate (MLSRR) for the elderly from 2017 to 2058.

Year	MLSRR(%)	Year	MLSRR(%)	Year	MLSRR(%)
2019	98.6	2033	90.6	2047	60.8
2020	100.2	2034	88.6	2048	58.8
2021	101.2	2035	86.6	2049	56.8
2022	101.9	2036	84.5	2050	54.8
2023	102.1	2037	82.4	2051	52.9
2024	102.0	2038	80.2	2052	51.0
2025	101.6	2039	78.0	2053	49.2
2026	100.9	2040	75.8	2054	47.4
2027	99.9	2041	73.7	2055	45.6
2028	98.7	2042	71.5	2056	43.9
2029	97.4	2043	69.3	2057	42.2
2030	95.9	2044	67.1	2058	40.6
2031	94.2	2045	65.0
2032	92.4	2046	62.9

Comparison of the ARR and MLSRR

The actual replacement rate (ARR) for the elderly people is 24.2 percent in 2014, which is far lower than the modified living standards replacement rate (MLSRR) (84.7 percent) (Table 5). For the middle aged people, no matter which payment level they choose, the NRPS will not meet their basic living needs when they reach the retirement age (Fig 2). For example, even if a person aged 45 chooses the highest payment level (¥2,000 per year) and contributes to the FDC for fifteen years, the predicted ARR (63.0 percent) in 2029 would still be much lower than the predicted MLSRR (97.4 percent).

Fig 2

Comparison of actual replacement rate (ARR) and modified living standards replacement rate (MLSRR) of the insured residents aged below 60 (the middle aged and young people) when they join the New Rural Pension Scheme.

The ARR100, ARR200,…, and ARR2000 indicate the predicted ARR when selecting the payment levels of ¥100, ¥200, …, and ¥2,000 per year.

Regarding the young people, the results show their basic needs after retirement can be achieved at certain payment levels. The lowest level of FDC payments is ¥400 per year. At this payment level, the ARR in 2058 will be 40.8 percent, slightly higher than 40.6 percent of the MLSRR (Fig 2). This means that a 16 years old person would need to continuously contribute to the FDC for 44 years. If a young person chooses the highest payment level (i.e., ¥2,000 per year), she/he needs to contribute to the FDC for at least 24 years in order to meet the basic needs after retirement (Fig 2).

Increasing the basic pension and government subsidies

China is currently spending only about 0.11 percent of its GDP on the SP component [21]. This is substantially lower than the average of 0.42 percent in some Latin American countries which have a similar level of economic development to China [14]. In addition, the Chinese government’s subsidies to the FDC component are also very limited and they only provide 8 percent to match the rural residents’ contribution payment [21]. By contrast, in most OECD countries with a similar FDC scheme, the matching proportion is around 20 percent on average [14]. Chinese people have the traditional culture and preference of saving money in banks, which is much stronger than that in the Latin America and Caribbean countries [32]. If the Chinese government matchs a substantially larger portion of the voluntary FDC contributions, it is likely that the rural residents would invest more of their savings in the NRPS and this also helps shift to higher voluntary contribution levels. In this context, the Chinese government–including central, provincial and local governments–should increase the basic pension and its matching proportions across the country, especially in the poor western regions, to address the concerns on pension adequacy.

Shifting from voluntary FDC to modest mandatory pillar

China’s NRPS has been very successful in promoting rural coverage over the past years. This is mainly attributed to the “family-binding” policy, whereby it is only after their adult children’s “voluntarily” participation in the FDC component that the elderly who are already in retirement age (60 years or older) can receive the basic pension benefits. However, the “voluntary” participation may decline when new retirement-aged rural residents become eligible for the NRPS pension benefits.

Moreover, most rural residents currently opt for the lowest level (¥100) of annual contribution to the FDC component, which results in very low pension benefits that they can receive after retirement. Even if the interest earned on their contributions to the FDC exceeds the inflation rate, the pension benefits generated through the FDC accounts cannot meet the pension needs of the rural population [14]. Considering a possible decrease in future enrolments and low levels of pension benefits, it is suggested that the current voluntary FDC element should shift to a modest mandatory FDC pillar with at least ¥400 per year, that is, the lowest contribution level necessary to meet basic needs after contributing to the FDC for 44 years. This reform option could arguably facilitate the retention of the currently high rural pension coverage and an increase in pension benefits for the rural residents.

Exploring and enhancing other support for retired residents

The comparative evaluation shows that the pension benefits of current NRPS cannot meet the basic needs of the elderly and middle aged people living in rural areas. However, the living expenses of the rural elderly people are not just met with the state-organized pensions. The diversified funding sources for the elderly people in rural China include individual savings, family financial support (also called Confucian welfare familism), intergenerational monetary transfer, formal or informal employment after retirement, and access to farmland. Therefore, these alternative ways facilitate an increase in the social security of those rural residents.

First, the traditional role of family support in caring for the elderly should not be ignored. In recent years, China’s rural population has declined substantially due to the migration of young people looking for better job opportunities. This outmigration has resulted in a decrease in family bonds and disintegrated the traditional family structure, to some extent, endangering the traditional role of family support in senior care. Therefore, the Chinese government should adopt certain strategies (e.g., creating more employment opportunities in rural areas) to support the young population, which helps strengthen the link between the elderly and their adult children.

Second, increasing the return rate of the rural pension fund can be another way to improve the pension adequacy of the NRPS. Currently, China’s social pension fund calculates earnings based on a fixed interest rate for deposits, i.e., an annual rate of 3 percent, which is lower than of the average annual rate of 8 percent in other countries [33]. The Chinese government can, therefore, expand the scope for financial investment of the pension fund. A higher return rate is possible if investing in local government bonds, equity investments, trust loan investments and interbank deposit certificates [34]. Third, commercial pension insurance companies can be encouraged to enter the rural market in western China to diversify the rural pension insurance plans. This will make the commercial pension programs more widely available in rural areas and achieve their complementary but equally important role in supporting the rural residents.