Does pension affect labour supply? New evidence from large-scale rural pension policy reform in China

ABSTRACT

Using national longitudinal survey data, we study how the Chinese New Rural Pension Scheme (NRPS) reform affects the rural labour supply structure of individuals aged 20–70. Applying multiple regression approaches, we find NRPS does not significantly affect Chinese rural residents’ weekly working hours, labour force participation, or migration choices; only those aged 50–59 change their labour supply pattern after joining the pension system. Parental pensions motivate adult children to migrate out only when both parents receive pensions. Various factors are explored to understand our findings; the low subsidy that Chinese rural pensioners currently receive is the major factor. These results indicate that from the perspective of improving the welfare of the rural elderly and encouraging the more free flow of young rural labour force, it is necessary to increase the compensation level of NRPS.

KEYWORDS:

• Chinese pension reform
• rural labour
• migration choice
• labour participation
• multiple econometric methods

JEL CLASSIFICATION:

• I38
• J00
• J43
• O53

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

¹ We only use the 2012, 2014, and 2016 surveys for several reasons. The test surveys in 2008 and 2009 have not been completely published, and data for 2018 are still being updated, so we are unable to use them. Although 2010 survey data has been completely published, it is also not available for our use. In reality, it will take some time for the pension policy reform to affect the labour supply. The NRPS was initially piloted in late 2009, while the 2010 survey was conducted in March 2010. Therefore, 2010 data are unlikely to reflect the true impact of pension reform on labour supply due to the very limited time between those two points. Considering all this, we use the data from the 2012, 2014, and 2016 surveys for our analysis. The data can be accessed at http://www.isss.pku.edu.cn/cfps/.

² The Chinese NRPS and Chinese Endowment Insurance of Urban Residents (EIUR) have been combined into the Chinese Insurance System for Urban and Rural Residents since 2014. However, due to the short implementation time of the new policy and no significant adjustment of the content, the names of pension schemes before the reform were used in the 2016 questionnaire. Interviewees were asked what specific kind of pension scheme they participated in. Therefore, we can directly determine what kind of pension scheme rural respondents participated in (Question I301). Additionally, we can also determine what kind of pension rural respondents received, combined with the question ‘Have you gone through the procedures for receiving pensions’ (Question I1011).

³ A total of 220 interviewees aged 50–59 who reported that they had received pension benefits, accounting for 7.55% of the pension beneficiaries in the sample. To obtain better estimation results using the RD method, we choose age 50 as the lower cut-off. To be consistent, we choose age 70 as the upper cut-off.

⁴ It is a common practice for self-rated health to be used as a proxy for health status; however, some studies suggest that self-rated health as an indicator of an individual’s state of health may create an endogeneity problem (e.g. Zhang, Cheng, and Liu 2016). Since CFPS surveys provide no better alternative indicators, our study employs the common practice and uses self-rated health as a proxy indicator for health status.

⁵ The NRPS was initially piloted in a few selected districts in late 2009 and was adopted by other areas in the following years. When the NRPS was launched, those aged 60 or above became entitled to receive pension benefits each month without making any contribution. For instance, if an area launched the NRPS in 2010, the local residents who reached age 60 in 2010 were eligible to obtain subsidies despite never having contributed. However, to receive benefits, residents who had not reached age 60 in 2010 were required to pay the premium each year before they reached the cut-off age. For example, a person who was 58 years old in 2010 was required to contribute for at least two years to be subsidized after age 60. This means that in the 2012, 20,014, and 2016 CFPS surveys, some of the pension recipients may not have ever contributed to the NRPS, while others may have contributed for a few years. An endogeneity problem may arise for those who had made contributions. However, since such individuals account for an extremely small proportion of the total recipients, the endogeneity caused by their existence can reasonably be ignored.

⁶ According to the NRPS regulations, participants must contribute to the NRPS for at least 15 years before they reach the eligible age of 60 to receive pension benefits. Therefore, 45 may be a breakpoint age for rural residents to choose to participate. However, a graphical analysis on the distribution of residents’ participation rate over age shows no significant jump in the rate around age 45. According to the graphical results, we do not apply the RD method when we examine the effects of participating in the NRPS on the labour supply of people under age 60.

⁷ The Relevance tests results in Appendix Table A2 shows that the IV we selected is considered to meet the principle of relevance.

⁸ The optimal bandwidth corresponding to the local linear regressions of labour force participation status and working hours are 2.027 and 5.554, respectively. Hence, when using the2SLS method for regression, we choose sample observations within the ±2 year and ±6 year intervals with age 60 as the origin.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China [No. 71873063].