1. Introduction Model Scenarios Calibration Results
The shadow of longevity – does social security reform
reduce gains from increasing the retirement age?
with Karolina Goraus, Krzysztof Makarski and Joanna Tyrowicz
Marcin Bielecki
Faculty of Economics, University of Warsaw
First World Congress of Comparative Economics
Rome, 25-27 June 2015
1 / 30

2. Introduction Model Scenarios Calibration Results
Motivation
Major issues in pension economics:
increasing old-age dependency ratio
majority of pension systems fail to assure actuarial fairness
in most countries people tend to retire as early as legally allowed
2 / 30

3. Introduction Model Scenarios Calibration Results
Motivation
Major issues in pension economics:
increasing old-age dependency ratio
majority of pension systems fail to assure actuarial fairness
in most countries people tend to retire as early as legally allowed
Typical reform proposals
switch to DC systems and strengthen the link
between contributions and benefits
raise the social security contribution rate
cut government expenditure
increase minimum eligibility retirement age
2 / 30

4. Introduction Model Scenarios Calibration Results
Effective retirement age in OECD
3 / 30

5. Introduction Model Scenarios Calibration Results
Participation rates in OECD
4 / 30

6. Introduction Model Scenarios Calibration Results
Outline
1 Introduction
2 Model
3 Scenarios
4 Calibration
5 Results
5 / 30

7. Introduction Model Scenarios Calibration Results
Literature review
Two streams of literature:
1 Answering the question about optimal retirement age
Gruber and Wise (2007), Galasso (2008), Heijdra and Romp (2009)
2 Comparing different pensions system reforms: increasing retirement age
vs cut in benefits/privatization of the system/...
Auerbach et al. (1989), Hviding and Marette (1998), Fehr (2000),
Boersch-Supan and Ludwig (2010), Vogel et al. (2012)
Fehr (2000)
Macroeconomic effects of retirement age increase may depend on the existing
relation between contributions and benefits
Remaining gaps in the literature
how the macroeconomic effects differ between various pension systems?
what happens to the welfare of each affected generation and why?
6 / 30

8. Introduction Model Scenarios Calibration Results
Goals and expectations
Goal
Analyse macroeconomic and welfare implications of retirement age
increase under DB (defined benefit), NDC (notionally defined
contribution) and FDC (funded defined contribution) systems
Expectations
under DB: leisure ↓, taxes ↓, welfare?
under DC: leisure ↓, pensions ↑, welfare?
difference between FDC & NDC: pricing of capital?
Why a full model? Labor supply adjustments & GE effects
7 / 30

9. Introduction Model Scenarios Calibration Results
Model structure: consumers I
”born” at age 20 (j = 1) and live up to 100 years (J = 80)
subject to time and cohort dependent survival probability π
choose labor supply l endogenously until exogenous
retirement age ¯J (forced to retire)
8 / 30

10. Introduction Model Scenarios Calibration Results
Model structure: consumers I
”born” at age 20 (j = 1) and live up to 100 years (J = 80)
subject to time and cohort dependent survival probability π
choose labor supply l endogenously until exogenous
retirement age ¯J (forced to retire)
optimize remaining lifetime utility derived from leisure 1 − l
and consumption c
Uj,t =
J−j
s=0
δs πj+s,t+s
πj,t
u(cj+s,t+s, lj+s,t+s)
with
u(c, l) = log(cφ
(1 − l)1−φ
)
8 / 30

11. Introduction Model Scenarios Calibration Results
Model structure: consumers II
receive market clearing wage for labor
receive market clearing interest rate on private savings
receive pension income
receive unintentional bequests
pay taxes
9 / 30

12. Introduction Model Scenarios Calibration Results
Model structure: consumers II
receive market clearing wage for labor
receive market clearing interest rate on private savings
receive pension income
receive unintentional bequests
pay taxes
Subject to the budget constraint
(1 + τc
t )cj,t + sj,t = (1 − τl
t )(1 − τι
)wj,tlj,t ← labor income
+ (1 + (1 − τk
t )rt)sj−1,t−1 ← capital income
+ (1 − τl
t )pι
j,t ← pension income
+ bj,t ← bequests
− Υt ← lump-sum tax
9 / 30

13. Introduction Model Scenarios Calibration Results
Model structure: producers
perfectly competitive representative firm
standard Cobb-Douglas production function
Yt = Kα
t (ztLt)1−α
profit maximization implies
wt = zt(1 − α)kα
t
rt = αkα−1
t − d
with k ≡ K
zL
10 / 30

14. Introduction Model Scenarios Calibration Results
Model structure: government
collects taxes on earnings, interest and consumption (sum up to T)
spends GDP fixed share of GDP on government consumption G
collects social security contributions and pays out pensions
of DB and NDC system
subsidyt = τι
¯J−1
j=1
wj,tlj,t −
J
j= ¯J
pj,tNj,t
services debt D and maintains debt/GDP ratio fixed
lump-sum taxes Υ adjust to satisfy the govt budget constraint
Gt + subsidyt + (1 + rt)Dt−1 = Tt + Dt + Υt
J
j=1
Nj,t
11 / 30

15. Introduction Model Scenarios Calibration Results
Pension systems
Defined Benefit: constructed by imposing a mandatory exogenous
contribution rate τ and an exogenous replacement rate ρ
pDB
¯J,t
= ρw ¯J−1,t−1l ¯J−1,t−1
indexed by 25% of total payroll growth
Defined Contribution: constructed by imposing a mandatory
exogenous contribution rate τ and actuarially fair individual
accounts
pDC
¯J,t
=
accumulated sum of contributions ¯J,t
expected remaining lifetime ¯J,t
Notional: contributions before retirement and pensions are indexed
by 25% of total payroll growth
Funded: contributions before retirement and pensions are indexed
by market interest rate
12 / 30

16. Introduction Model Scenarios Calibration Results
What we do
What happens within each experiment?
1 Run the no policy change scenario ⇒ baseline
2 Run the policy change scenario ⇒ reform
3 For each cohort compare utility, compensate the losers from the
winners
4 If net effect positive ⇒ reform efficient
Welfare analysis – like Nishiyama & Smetters (2007)
Macroeconomic analysis
13 / 30

17. Introduction Model Scenarios Calibration Results
Reforms
Three experiments:
1 DB with flat retirement age → DB with increasing retirement age
2 NDC with flat retirement age → NDC with increasing retirement age
3 FDC with flat retirement age → FDC with increasing retirement age
Increasing retirement age
14 / 30

18. Introduction Model Scenarios Calibration Results
Robustness: age-productivity profile
Heterogeneity between cohorts due to age-specific productivity: wj,t = ωjwt
Deaton (1997) decomposition on Polish LFS data
15 / 30

19. Introduction Model Scenarios Calibration Results
Calibration to replicate 1999 economy of Poland
Preference for leisure (φ) chosen to match participation rate of 56.8%
Impatience (δ) chosen to match interest rate of 7.4%
Replacement rate (ρ) chosen to match benefits/GDP ratio of 5%
Contributions rate (τ) chosen to match SIF deficit/GDP ratio of 0.8%
Labor income tax (τl
) set to match PIT/GDP ratio
Consumption tax (τc
) set to match VAT/GDP ratio
Capital income tax (τk
) set de iure = de facto
16 / 30

20. Introduction Model Scenarios Calibration Results
Calibrated parameters
Age-productivity profile
ω – D97 ω = 1
α capital share 0.31 0.31
τl
labor income tax 0.11 0.11
τc
consumption tax 0.11 0.11
τk
capital income tax 0.19 0.19
φ leisure-consumption preference 0.578 0.526
δ discounting rate 0.998 0.979
d depreciation rate 0.045 0.045
τ social security contribution 0.060 0.060
ρ replacement rate 0.138 0.227
resulting
(dk)/y investment rate 21 21
r interest rate 7.4 7.4
17 / 30

21. Introduction Model Scenarios Calibration Results
Exogenous processes in the model I
Demographic projection until 2060, after that 80 years, and after
that “new steady state”
“Births” of 20-year olds from the projection, constant afterwards
Mortality rates from the projection, constant afterwards
18 / 30

22. Introduction Model Scenarios Calibration Results
Exogenous processes in the model II
Productivity growth
Labor augmenting productivity parameter z
Projection from AWG, after that “new steady state”, 1.7%
19 / 30

23. Introduction Model Scenarios Calibration Results
Is the reform efficient?
Yes!
Net consumption equivalent ω – D97 ω = 1
DB 9.88% 3.70%
Transition to NDC 11.31% 4.41%
Transition to FDC 11.81% 4.70%
20 / 30

24. Introduction Model Scenarios Calibration Results
Everybody gains
21 / 30

25. Introduction Model Scenarios Calibration Results
Why they gain? Benefits under DC systems ...
22 / 30

26. Introduction Model Scenarios Calibration Results
... and taxes under DB system
SIF deficit
Lump-sum taxes
23 / 30

27. Introduction Model Scenarios Calibration Results
Is there any behavioral response? Of course!
24 / 30

28. Introduction Model Scenarios Calibration Results
Labor supply in the final steady state
Labor supply Labor supply with MERA increase
(no reform) j < 60 j ≥ 60 Total
Average Average Aggregate Average Aggregate
(base=100%) (base=100%)
DB 63.2% 59.6% 94.4% 71.8% 113.7%
NDC 62.0% 58.8% 94.8% 72.3% 114.7%
FDC 61.7% 59.0% 95.5% 72.2% 115.4%
25 / 30

29. Introduction Model Scenarios Calibration Results
Aggregate labor supply (in millions of individuals)
26 / 30

30. Introduction Model Scenarios Calibration Results
Capital per effective unit of labor decreases ...
27 / 30

31. Introduction Model Scenarios Calibration Results
... but mostly due to decrease in “precautionary savings”
28 / 30

32. Introduction Model Scenarios Calibration Results
Conclusions
extending the retirement age is universally welfare enhancing
some downward adjustment in individual labor supply,
but the aggregate labor supply increases
effects on capital are “overstated”
29 / 30

33. Introduction Model Scenarios Calibration Results
Thank you for your attention!
Questions or suggestions?
30 / 30