Page 118 - IJPS-11-5

International Journal of
Population Studies Endowment insurance and family consumption in China

Table 8. Regression results of household consumption for Table 9. Number of participants in different endowment
different ages insurance
Independent variable Type I Type II Types of pension insurance Sample Sample size
households households number of for this type of
(under (60 years old and participants insurance only
60 years old) above) Pension received from the 1051 556
Dependent variable: Household institution where he or she used to
consumption work after retirement
Participation in 0.099*** (9.104) Basic endowment insurance 3634 2610
endowment insurance Supplementary enterprise 493 195
Pensions (log) −0.008** (−2.356) endowment insurance
Constant 9.808*** (195.646) 9.931*** (58.019) Commercial endowment insurance 498 173
Control variables Join Join Old rural endowment insurance 1236 1045
Household fixed effects Yes Yes New rural social endowment 8070 7543
Year fixed effects Yes Yes insurance
Fixed effects F test 2.820 (p=0.000) 2.870 (p=0.000) Urban and rural residents' 935 594
endowment insurance
Model significance test 103.400 (p=0.000) 17.070 (p=0.000)
Within R² 0.112 0.116 Initially, we selected ages 55 and 65 as alternative
Sample size 22440 4484 breakpoints. Table 13 presents the results, revealing that
Notes: *p<0.1; **p<0.05; ***p<0.01. age effects on pension receipt are generally not statistically
significant at these breakpoints, except for a few marginal
consistency, the analysis focuses on households with male cases. Similarly, the impact of pension receipt on
heads and employs a fuzzy breakpoint regression approach consumption remains mostly insignificant across different
to account for variations in pension receipt relative to the age ranges. These findings indicate that the treatment effect
retirement age. is specific to age 60, thereby confirming the robustness of
Figures 1 and 2 reveal distinct jump points near the the original breakpoint regression results.
retirement age, indicating a relationship between pension Subsequently, we excluded 302 samples near the
receipt and household consumption. The fuzzy breakpoint original breakpoint of age 60 to address potential
regression results presented in Table 11, utilizing the manipulation concerns. Table 14 shows that even after this
IV-2SLS method, isolates the effect of pension receipt by exclusion, the effects of being over 60 on pension receipt
using age 60 as an instrumental variable. remain statistically significant at the 1% level. In addition,
The results demonstrate a significant negative effect of the impact of pension receipt on consumption continues to
pension receipt on household consumption, suggesting be significant across various age ranges, suggesting that the
that receiving a pension leads to a reduction in spending. causal relationship persists despite the removal of samples
Table 12 demonstrates a significant negative effect of near the breakpoints.
pension receipt on household consumption, suggesting Finally, we performed a placebo test by omitting all
that receiving a pension leads to a reduction in spending. control variables to determine if the observed effects were
Furthermore, the analysis in Figure 3 confirms that the driven by covariates. The results in Table 15 demonstrate
conditional densities of covariates remain continuous at the
breakpoints, showing no significant jumps. This reinforces that the effects of age over 60 on pension receipt and the
the validity of the fuzzy breakpoint regression model and subsequent impact on household consumption remain
supports the conclusion that receiving a pension may have statistically significant at the 1% level. This confirms that
an adverse effect on household consumption. the consumption jump at the breakpoint is not attributable
to the inclusion of control variables, thereby supporting a
3.5.3. Placebo test causal relationship between pension receipt and household
The placebo test was conducted to evaluate the robustness consumption.
of the breakpoint regression results by introducing Overall, the placebo tests validate the robustness
alternative breakpoints at ages 55 and 65. This involved of the breakpoint regression findings, indicating that
assessing the impact of these new breakpoints on pension the significant increase in consumption at age 60 is not
receipt and household consumption. influenced by alternative breakpoints, sample exclusion, or

Volume 11 Issue 5 (2025) 112 https://doi.org/10.36922/ijps.4857