Na zaproszenie Pani Profesor Elżbiety Gołaty, prorektora Uniwersytetu Ekonomicznego w Poznaniu Paweł Strzelecki przedstawił zebranym specjalistom zajmującym się statystyką i demografią wyniki symulacji efektów różnych scenariuszy wzrostu dzietności za pomocą modelu makroekonomicznego nakładających się pokoleń (OLG). Szczególnie ożywioną dyskusję wzbudziła możliwość kwantyfikacji dobrobytowych skutków polityki prorodzinnej oraz efekty makroekonomiczne w dłuższym terminie. Dyskusja dotyczyła także możliwości pomiaru stanu zdrowia ludności za pomocą dostępnych danych. Bardzo dziękujemy za możliwość podzielania się wynikami badań oraz bardzo ciekawe uwagi.
1. Simulations of macroeconomic effects caused by changes in
fertility and health status of the ageing Polish population
Paweł Strzelecki,
Krzysztof Makarski, Joanna Tyrowicz, Magda Malec i Oliwia Komada
05/03/2019
Uniwersytet Ekonomiczny w Poznaniu
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 1
2. Outline
Motivation and research questions
Fertility scenarios
Mortality scenarios
OLG model – short introduction
Demographic scenarios in OLG
Results of the selected OLG simulations
Summary
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 2
4. Motivation
• Policy call for costly natalist policies and instruments
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 3
5. Motivation
• Policy call for costly natalist policies and instruments
• substantial decline in population due to lowering fertility and longevity in most of
advanced and middle income economies
• declining population and multiple long-term implications
=⇒ social security, pension system and health care expenditures
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 3
6. Motivation
• Policy call for costly natalist policies and instruments – are they worth it?
• substantial decline in population due to lowering fertility and longevity in most of
advanced and middle income economies
• declining population and multiple long-term implications
=⇒ social security, pension system and health care expenditures
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 3
7. Motivation
• Policy call for costly natalist policies and instruments – are they worth it?
• substantial decline in population due to lowering fertility and longevity in most of
advanced and middle income economies
• declining population and multiple long-term implications
=⇒ social security, pension system and health care expenditures
• mixed empirical literature on previous policy interventions
=⇒ negligible effects, "too soon to tell", methodological issues
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 3
8. Motivation
• Policy call for costly natalist policies and instruments – are they worth it?
• substantial decline in population due to lowering fertility and longevity in most of
advanced and middle income economies
• declining population and multiple long-term implications
=⇒ social security, pension system and health care expenditures
• mixed empirical literature on previous policy interventions – can we explain why?
=⇒ negligible effects, "too soon to tell", methodological issues
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 3
9. Literature review – fertility
• empirical evaluation with negative effects
Olivetti and Petrongolo (2017), Baizan et al. (2016), Rossin-Slater (2018)
• empirical evaluation with positive effects
Drago et al. (2011), Milligan (2005), Brewer et al. (2012), Frejka and Zakharov (2013), Garganta et al. (2017),
Lalive and Zweimueller (2009), Rindfuss et al. (2010), Havnes and Mogstad (2011), Bauernschuster et al. (2015),
Del Boca et al. (2009)
• OLG framework with more explicit fertility
Fehr et al. (2017), Georges and Seekin (2016), Mamota (2016), Hock and Weil (2012)
• fertility may be endogenous
Liao (2011), Ludwig et al. (2012), Hock and Weil (2012)
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 4
10. Our questions – fertility
1. What are welfare effects of fertility changes?
⇒ costs are immediate and private, gains are delayed and public
2. What are macroeconomic effects of fertility changes?
⇒ assuming we know how to achieve given rise in fertility, how much to spend
3. Does it matter what kind of policy we do?
⇒ intensive (families with kids have more) vs extensive (more families has kids)
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 5
11. Literature review – disability/health
• mixed empirical observations regarding relation between increasing e0 and
HALE and different theories that explains main possible outcomes
Olshansky et al.(1991), Verbrugge (1984)., Fries (1989), Manton (1982)
• however no doubts that HALE is increasing in working age population
EHLEIS (2018), Salomon et al. (2012)
• in most publications improving health has clearly positive influence on growth
see Garibaldi (2010) for extended literature review
• different approaches to health changes in macromodels
French, Song(2012), DeNardi(2018)
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 6
12. Our questions – disability/health
1. Does improving health in production age have only positive impact on economy?
⇒ potential selection to disability
2. Improving health and retirement age?
⇒ to what extend potential from health improvement is hampered by
retirement age
3. What are macroeconomic effects of longer healthy life?
⇒ assuming we know how to achieve given rise in healthy life, how much to
spend
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 7
13. Value added of simulations in OLG model
Value added of simulations in OLG model
• time mismatch: immediate costs and delayed benefits
• beneficiary mismatch: private costs and public gains
• general equilibrium effects: people adjust to expected fertility and better health
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 8
15. TFR in population projections for Poland
Polish CSO - TFR scenarios Children ever born in the age 30 by cohorts
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 9
16. Fertility rates by birth order
ASFR first chld ASFR second chld ASFR 3+
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 10
18. Men - perspective of the e0 and HALE increase for Poland
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 11
19. Women - perspective of the e0 and HALE increase for Poland
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 12
20. HALE and length of labour market career
Men Women
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 13
21. Disability in Poland – changes in time
Men Women
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 14
22. Trends in disability in Poland - APC analysis
The results of the age − period − cohort decomposition of the changes in disability.
Men – Labour Force Survey data:
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 15
23. Trends in disability in Poland - APC analysis
The results of the age − period − cohort decomposition of the changes in disability.
Women – Labour Force Survey data:
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 16
24. Relative wage level of employed persons before transition to disability
Men Women
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 17
26. Macro model with family structure
Our approach
We develop large OLG model with family structure and types of agents in a
household. Things we really care for:
• family structure – households with κ = 0, 1, 2, 3+ children
• heterogeneity – two types of agents within a household
• exogenous fertility
• extensive and intensive margin adjustments
• calibrating the model closely to the data
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 18
27. Macro model with family structure
Our approach
We develop large OLG model with family structure and types of agents in a
household. Things we really care for:
• family structure – households with κ = 0, 1, 2, 3+ children
• heterogeneity – two types of agents within a household
• exogenous fertility
• extensive and intensive margin adjustments
• calibrating the model closely to the data
Main novelty
We can analyze various paths of fertility ⇒ sensitivity of macro to demographics
Model Calibration Demography
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 18
28. Labor augmenting technological progress
• Projection from European Commission AWG Aging Report, constant TFP
growth at 1.4%
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 19
29. Fertility scenarios
What is baseline?
• status quo demographic projection
• unchanged fertility 1.44 (data averaged for 2006-2014)
• data on household structure
What is fertility change scenario?
• 1.44 −→some higher level
• How many combinations of household structure can generate a given fertility
increase path?
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 20
30. Fertility scenarios
What is baseline?
• status quo demographic projection
• unchanged fertility 1.44 (data averaged for 2006-2014)
• data on household structure
What is fertility change scenario?
• 1.44 −→some higher level
• How many combinations of household structure can generate a given fertility
increase path? Countless.
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 20
31. Fertility scenarios
What is baseline?
• status quo demographic projection
• unchanged fertility 1.44 (data averaged for 2006-2014)
• data on household structure
What is fertility change scenario?
• 1.44 −→some higher level
• How many combinations of household structure can generate a given fertility
increase path? Countless.
• Does it matter?
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 20
32. Fertility scenarios
What is baseline?
• status quo demographic projection
• unchanged fertility 1.44 (data averaged for 2006-2014)
• data on household structure
What is fertility change scenario?
• 1.44 −→some higher level
• How many combinations of household structure can generate a given fertility
increase path? Countless.
• Does it matter? Yes.
Generating scenarios
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 20
33. Health/disability scenarios
What is baseline?
• status quo demographic projection
• constant productivity by age and preference for leisure (φ) to match labour
force participation
• calibrated to match parameters in real economy
Proposed health change scenarios?
• change in labour force participation preference for leisure (φ)
• change in the productivity endowments by age (ωi )
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 21
35. Measuring macroeconomic effects
• net surplus of the government budget
difference between government spending in baseline (GB
t ) and reform (GR
t )
discounted for the moment of fertility change and expressed in terms of GDP
per capita
• it accounts for GE effects (i.e. public gains from fertility increase)
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 22
36. Measuring fiscal effects
• fiscally beneficial
• (−) labor market (women)
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 23
37. Measuring fiscal effects
• fiscally beneficial
• (−) labor market (women)
(−) higher government expenditures
• (+) labor market (men)
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 23
38. Measuring fiscal effects
• fiscally beneficial
• (−) labor market (women)
(−) higher government expenditures
• (+) labor market (men)
(+) higher consumption tax base
(+) higher labor tax base
• effects are small
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 23
39. Measuring fiscal effects
How much can we spend every year, assuming increase of fertility?
target fertility 1.50 target fertility1.85
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 24
40. Measuring fiscal effects
Distribution of the fiscal effects over time
target fertility 1.50 target fertility1.85
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 25
41. Measuring welfare effects
Do I prefer to live in the world with increased fertility?
(discounted expected utility in the form of consumption equivalent)
• raised fertility is not increasing
individually measured utility
• negative effect:
wages ↓ > pension benefits ↑
• really small effects: individual
welfare is not particularly responsive
to the population dynamics
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 26
42. Measuring welfare effects
Does intensive and extensive margin matter?
target fertility 1.50 target fertility1.85
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 27
44. Insights from our study – fertility
1. Fiscal: net surplus in government budget, but small
• small, but universal fiscal gains
• for central path 0.2% GDP
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 28
45. Insights from our study – fertility
1. Fiscal: net surplus in government budget, but small
• small, but universal fiscal gains
• for central path 0.2% GDP
• not in the long run (!)
2. Welfare: negative welfare effect
• fertility↑ −→ welfare↓
• intensive and extensive margin matters (change in sign), but extensive margin can
be unrealistic (no trend in data)
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 28
46. Insights from our study – fertility
1. Fiscal: net surplus in government budget, but small
• small, but universal fiscal gains
• for central path 0.2% GDP
• not in the long run (!)
2. Welfare: negative welfare effect
• fertility↑ −→ welfare↓
• intensive and extensive margin matters (change in sign), but extensive margin can
be unrealistic (no trend in data)
3. Methodology: mixed empirical results make sense
• intensive vs extensive fertility margin tilts the sign of the outcomes
• even with unidirectional labor supply effects
• effects are small (false rejection of null hypothesis in empirical research)
• labor supply of men dominates that of women
• potential perverse incentives should drag even further towards zero
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 28
47. Insights from our study – fertility
1. Fiscal: net surplus in government budget, but small
• small, but universal fiscal gains
• for central path 0.2% GDP
• not in the long run (!)
2. Welfare: negative welfare effect
• fertility↑ −→ welfare↓
• intensive and extensive margin matters (change in sign), but extensive margin can
be unrealistic (no trend in data)
3. Methodology: mixed empirical results make sense
• intensive vs extensive fertility margin tilts the sign of the outcomes
• even with unidirectional labor supply effects
• effects are small (false rejection of null hypothesis in empirical research)
• labor supply of men dominates that of women
• potential perverse incentives should drag even further towards zero
• it is not likely that technological progress is a “replacement” for fertility
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 28
48. Expected results – disability/health
1. Fiscal: net surplus in government budget
2. Welfare: probably ↑
3. Methodology: searching for nontrivial assumptions
• better health can mean more less productive persons on the market
• retirement age limits gains from better health
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 29
49. Thank you for your attention!
w: grape.org.pl
t: grape_org
f: grape.org
e: pawel.strzelecki@sgh.waw.pl
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie 30
50. Producers – very standard
• Perfectly competitive representative firm
• Standard Cobb-Douglas production function
Yt = Kα
t (zt Lt )1−α
,
• Profit maximization implies
wt = (1 − α)Kα
t zt (zt Lt )−α
rt = αKα−1
(zt Lt )1−α
− d
where d is the capital depreciation rate and zt is technological progress
Go back
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
51. Consumers
• live up to j = 1, 2, ..., J years (J = 100)
• face time and age specific mortality
• labor supply l endogenous until retirement age ¯J = 65
• until adult j < 21 they live in the household of birth
• reaching adulthood j = 21 they form their own household
• and observe the realization of the fertility
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
52. Households
• consist of men and women (the latter denoted by *)
• differ by the number of children κ = 0, 1, 2, 3+
• collective decision making within households
• optimize lifetime utility derived from leisure and consumption
J
j=21
βj−21
πj,t+j−21[uj (˜cκ,j,t+j−21, lκ,j,t+j−21)
+ u∗
j ˜cκ,j,t+j−21, l∗
κ,j,t+j−21 ]
• with individual consumption as follows
˜cκ,j,t =
1
(2 + ϑκ)
cκ,j,t = Ξκcκ,j,t
ϑ child consumption scaling factor,
consumption scaling factor, Ξκ scale effect
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
53. Households II
• during child rearing “female” labor supply is reduced following ϕκ
• households maximize utility:
men in age j < ¯J : uj (˜cκ,j,t , lκ,j,t ) = log ˜cκ,j,t + φ log(1 − lκ,j,t )
women in age j < 41 : u
∗
j (˜cκ,j,t , l
∗
κ,j,t ) = log ˜cκ,j,t + φ log(1 − l
∗
κ,j,t − ϕ(κ))
women in age 41 ≤ j < ¯J : u
∗
j (˜cκ,j,t , l
∗
κ,j,t ) = log ˜cκ,j,t + φ log(1 − l
∗
κ,j,t )
men in age j ≥ ¯J : uj (˜cκ,j,t , lκ,j,t ) = log ˜cκ,j,t
women in age j ≥ ¯J : u
∗
j (˜cκ,j,t , l
∗
κ,j,t ) = log ˜cκ,j,t
• subjected to:
(1 + τc )cκ,j,t + ˜sκ,j+1,t+1 = (1 − τ − τl )wj,t lκ,j,t + (1 − τ − τl )wj,t l
∗
κ,j,t
+ (1 + rt (1 − τk )) ˜sκ,j,t
+(1 − τl )bκ,j,t + (1 − τl )b
∗
κ,j,t
+beqκ,j,t + Υt (1)
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
54. Government
• collects taxes
Tt = τl (1 − τ)wt Lt + τl Bt + τc Ct + τk rt St + Υt
where Lt , Ct , St , Bt denote labor, consumption, savings and benefits
• finances spending on public goods and service Gt = gt Yt ,
• and services debt ∆Dt = (1 + rt )Dt−1 − Dt
Tt = Gt + ∆Dt
• PAYG defined contribution pension system with mandatory τ
bκ,¯J,t =
¯Jt −1
s=1
Πs
ι=1(1 + rI
t−j+ι−1) τwt−j+s−1lκ,s,t−j+s−1
J
s=¯J
πs,t
• pensions indexed annually with the rate of payroll growth
Back to main
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
56. Calibration to replicate 2014 Polish economy
• Discounting rate (δ) matches interest rate of 6.5%
• Depreciation rate (d) matches investment rate of 21%
• Contribution rate (τ) matches benefits to GDP ratio of 7%
• Labor income tax (τl ) matches revenues to GDP ratio of 4.5%
• Consumption tax (τc ) matches revenues to GDP ratio of 11%
• Capital tax (τk ) de iure = de facto
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
57. Calibration to replicate 2014 Polish economy
• Discounting rate (δ) matches interest rate of 6.5%
• Depreciation rate (d) matches investment rate of 21%
• Contribution rate (τ) matches benefits to GDP ratio of 7%
• Labor income tax (τl ) matches revenues to GDP ratio of 4.5%
• Consumption tax (τc ) matches revenues to GDP ratio of 11%
• Capital tax (τk ) de iure = de facto
• Technological progress according to EC AWG projections, growth at 1.4%
Note: averages for 2000-2010 (investment rate) and 2005-2014
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
58. Preferences
• Preference for leisure (φ) matches participation rate of 56.8%
• Female child rearing time (ϕκ) according to Time Use Survey 2013, approx.
0.231 for κ = 1, 0.236 for κ = 2 and 0.257 for κ = 3
• Consumption scaling factor ( ) and child consumption scaling factor (ϑκ)
matches OECD equivalence scale
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
59. Demographic assumptions
• no mortality until children are raised (j < 41)
• historical data on fertility and mortality 1964-2014
• AWG projections until 2080, and stable afterwards
• completed fertility data from household structure for 2006-2014
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
62. Data match – calibrating to path
Data Model
Completed fertility 1.38-1.52 1.44
Share of cohorts at j < 21 0.23 0.23
Share of cohorts at 20 < j < 41 0.31 0.30
Share of cohorts at j ≥ ¯J 0.18 0.19
Life expectancy at j = 1 73.47 73.83
Life expectancy at j = ¯J 15.41 15.42
Shares of childless women 0.36 0.35
s1 : s2 : s3+ 0.16 : 0.28 : 0.2 0.16 : 0.29 : 0.2
Note: Completed fertility measured as realized fertility for women aged 45 years, data averaged over 2006-
2014. Shares of age groups based on population structure data, averaged over 2006-2014. Data from
Eurostat.
Back to model
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
63. Generating fertility scenarios
1. Pick target (completed) fertility rate
target = reach that rate in 35 years,
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie
64. Generating fertility scenarios
1. Pick target (completed) fertility rate
target = reach that rate in 35 years, gradually
We test target fertilities of 1.5, 1.85 and 2.1
2. For intensive scenarios (families have more kids)
2.1 Keep childless ratio constant
2.2 Randomly generate N=100 paths yielding given fertility, from compositions of 1,
2 and 3+ kids per household
2.3 Number of paths can be any
3. For extensive scenarios (more families has kids)
3.1 Keep ratio of 1, 2 and 3+ families constant
3.2 Randomly simulate N=100 paths yielding given fertility, from varying the share of
households without kids and adjusting remaining shares (with a fixed ratio)
Back to main
FAME|GRAPE, Szkoła Główna Handlowa w Warszawie