목차
Title page
Contents
Abstract 3
1. Introduction 4
2. Institutional Setting 8
3. Data 10
3.1. Main data and sample 10
3.2. Survey data on health outcomes 12
4. Empirical Strategy 13
4.1. Difference-in-Differences Method 13
4.2. Instrumental Variables Strategy 16
5. Results 17
5.1. Effects on Pension Income and Retirement 18
5.2. Effects on Mortality 19
5.3. Placebo and Robustness 20
5.4. Heterogeneous Effects on Mortality 21
5.5. Effect of Pension Income on Mortality 24
5.6. Mechanisms: Effects of Pension Income on Health Outcomes 25
6. Discussion 28
6.1. Comparisons with existing literature 28
6.2. Gender Differences 30
6.3. Policy implications 32
7. Conclusion 33
References 34
Online Appendix 44
A. Appendix Tables and Figures 45
B. Additional Details on Institution 78
C. Data Appendix 81
D. Details on Robustness 81
E. Calculation of the Monetary Gain in Life Expectancy 82
Table 1. Impact of subsidy eligibility (DID estimates) 39
Table 2. Heterogeneous effects (DID estimates) 40
Table 3. Impact of pension income on mortality (IV estimates) 41
Table 4. Impact of pension income on health outcomes (IV estimates) 42
Table 5. Impact of pension income on other outcomes (IV estimates) 43
Figure 1. Event study coefficients in the baseline sample, first stage 37
Figure 2. Event study coefficients in the baseline sample 38
Table A.1. Summary statistics (RTWF) 58
Table A.2. Summary statistics by gender by treatment status (RTWF) 59
Table A.3. Sample selection 60
Table A.4. Impact on mortality before 60 and after 75 (DID estimates) 61
Table A.5. Definition of health, financial constraints and optimism variables in SHARE-RV data 62
Table A.6. Summary statistics (SHARE-RV) 63
Table A.7. Summary statistics by gender (SHARE-RV sample) 64
Table A.8. Event study estimates in baseline sample 65
Table A.9. Impact of subsidy eligibility on other measures of mortality (DID estimates) 66
Table A.10. Placebo checks (DID estimates) 67
Table A.11. Robustness checks (DID estimates) 68
Table A.12. Heterogeneity effects by marital status and gender (DID estimates) 69
Table A.13. Heterogeneity effects by subsidy size and gender (DID estimates) 70
Table A.14. P-value on significance in difference of point estimates for heterogeneous effects (Table 2 and Table A.12) 71
Table A.15. Summary statistics for the compliers 72
Table A.16. Impact of pension income on other diseases (IV estimates) 73
Table A.17. Heterogeneity by share of pension income over total household income (IV estimates) 74
Table A.18. Heterogeneity by any household member owning a house (IV estimates) 75
Table A.19. Summary statistics for people with aep 〉 0.75 and more than 35 years of contribution by gender 76
Table A.20. Impacts of Eligibility on sickness leaves before age 50 (VSKT data) 77
Figure A.1. Relationship between subsidy size and aep 46
Figure A.2. Distribution of aep by contribution years above / below 35, full sample and by gender 47
Figure A.3. Distribution of contribution years by aep below and above 0.75, full sample and by gender 48
Figure A.4. Policy schedule of subsidy size by contribution years and by treatment status 49
Figure A.5. First stage: mean probability of being a recipient and amount of pension subsidy by contribution years 50
Figure A.6. Scatter plot of pension benefits over contribution years by treatment status 51
Figure A.7. Scatter plot of mortality outcomes over contribution years by treatment status 52
Figure A.8. Scatter plot of main outcomes over contribution years by treatment status for women and men 53
Figure A.9. Effect of eligibility on the probability of dying after retirement 54
Figure A.10. Placebo checks: event study coefficients in the 1922-1931 sample 55
Figure A.11. Placebo checks: event study coefficients in the aep ∈ (0.8-1.25) placebo sample 56
Figure A.12. Robustness checks: event study coefficients by contribution semester in the baseline sample 57