Presentation is loading. Please wait.

Presentation is loading. Please wait.

Public Pension Reforms and Private Savings

Similar presentations


Presentation on theme: "Public Pension Reforms and Private Savings"— Presentation transcript:

1 Public Pension Reforms and Private Savings
Minki Hong Korea Labor Institute August, 2009

2 Contents Previous Studies Korea National Pension Scheme
Estimation Method: CIC / DID Results

3 Previous Empirical Studies
(1) Estimate pension wealth(PW) (2) Regress private wealth on PW Calculating PW is very complicated: Earning profiles, retirement age, Results are sensitive to estimated PW Ideally require perceived wealth, smaller than PW by law.  Using PW overestimates the effects.

4 Desirable Solutions 1. Use information on perceived pension wealth by individuals  Bernheim & Levin (1989) (+ pension reforms & panel data would be more desirable) 2. Compare outcomes between the treatment group and the control group → This paper’s approach

5 Korea National Pension Scheme
Funded, defined benefits program Before 1999, covered workers workplace with 5+ workers In 1999, extended to all residents (1) the self-employed (2) non-regular workers (3) workers in small business

6 Korea National Pension Scheme
The self-employed account for 34% of total employment. Among the self-employed (aged 26-59), about 74% participated in the National Pension program as of 2002. Many self-employed run a very small business ( very volatile).

7 Korean National Pension : Contribution
Contribution rate workers : 4.5% (U.S. : 6.2%) self-employed : 9% (U.S. : 12.4%) (U.S. : 2.9% Medicare..)

8 Korean National Pension : Benefits

9 Structure of KNPS by cohort
Age As of 2002 Eligible Ages Contribution Years Replacement Rates 25-29 65 36-40 0.54 30-34 64-65 30-35 0.45 35-39 63-64 24-29 0.36 40-44 62-63 18-23 0.26 45-49 60-61 11-16 0.16 50-54 60 6-10 0.1

10 Structure (cont.) Change in saving by policy reform will be greater the closer the individual is to retirement age (Gale, 1998) Stronger effect of the pension on older individuals. For older individuals, the replacement rate is low due to the short period of contribution by KNPS  Weaker effect on older individuals The actual effect is level across ages.

11 Data Dependent variable: saving rates
Korea Labor and Income Panel Study HHs whose head is the self-employed aged t=0 (before reforms) : 1998 t=1 (after reforms) : each. Dependent variable: saving rates No information after 2004 on participation in KNPS. No information in 1998 on assets.

12 Summary Statistics (’98/’02)
Mean S.D. 10th perc 25th 50th 75th 90th N Control t=0 0.155 0.24 0.020 0.232 0.490 131 t=1 0.164 0.19 0.096 0.248 0.409 Treat 0.198 0.22 0.151 0.331 0.498 439 0.197 0.30 0.041 0.143 0.278 0.468

13 Estimation Method Changes-in-Changes by Athey and Imbens (2006) CIC
DID Effect of Policy Heterogeneous Same Adoption of Endogenous Exogenous Rescaling Not matter Matter

14 Changes-in-Changes : Observable
Comparing two produces the whole distribution of effects Fs are estimated nonparametrically Control (g=0) Treatment (g=1) t=0 F(0,0) F(1,0) t=1 F(0,1) F(1,1)

15 CIC with covariates 1. Regress D: four group-time dummy
X: size, own a house, net income, age, gender, marital status… 2. Construct residuals with the group/time effect be added back in Then apply CIC

16 Estimation Results (CIC)
Saving Rates 1998 and 2000 1998 and 2001 1998 and 2002 1998 and 2003 mean 0.0097 (0.0280) (0.0343) (0.0290) (0.0491) 10th perc 0.0114 0.007 (0.0134) (0.0159) (0.0174) (0.0117) 25th perc 0.0055 0.0026 (0.0120) (0.0227) (0.0192) (0.0254) 50th perc. 0.0202 (0.0413) (0.0335) (0.0484) (0.0483_) 75th perc. 0.0235 -0.046 (0.0428) (0.0547) (0.0504) (0.0704) 90th perc. 0.0428 0.0451 0.0267 (0.0379) (0.0742) (0.0902) (0.0964)

17 Estimation Results (DID)
Saving Rates 1998 and 2000 1998 and 2001 1998 and 2002 1998 and 2003 mean 0.0094 (0.0276) (0.0316) (0.0256) (0.0364) 10th perc 0.0478 0.0002 0.0247 (0.0235) (0.0296) (0.0232) (0.0326) 25th perc 0.0441 0.0125 0.0172 (0.0236) (0.0257) (0.0212) (0.0343) 50th perc. 0.0426 -0.022 (0.0333 (0.0328) (0.0438) 75th perc. (0.03 (0.038) (0.0321) (0.0395) 90th perc. (0.0420) (0.0325) (0.0428)

18 Results Mean effects are close to zero. General reason :
(1) precautionary and bequest motives of savings (2) uncertainty on pension benefits HH below 50th perc. (have zero saving rates) were not affected ← liquidity constraints

19 Results (cont.) HH around 50 to 75 perc.:
SR drop by 5-6% points, corresponds 25% offset HH 90th + perc are not affected ← (1) replacement rate is low (2) pension wealth accounts for small portion of their wealth.

20 Results from DID Mean effects are close to CIC
DID estimator tends to overestimate the negative effect for HH with high SR.


Download ppt "Public Pension Reforms and Private Savings"

Similar presentations


Ads by Google