Lecture 3C: The Continuous-Time Overlapping-Generations … · 2011. 12. 28. · Lecture 3C: The...
Transcript of Lecture 3C: The Continuous-Time Overlapping-Generations … · 2011. 12. 28. · Lecture 3C: The...
Motivation & ModelAnalysis
Conclusions & Further Research
Lecture 3C: The Continuous-TimeOverlapping-Generations Model:
Beyond the Basic Model
Ben J. Heijdra
Department of Economics, Econometrics & FinanceUniversity of Groningen
13 January 2012
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 1 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Outline
1 Motivation & ModelMotivation & OverviewModel: householdsModel: firms
2 AnalysisInformational assumptionsSeparating equilibriumPooling equilibrium
3 Conclusions & Further Research
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 2 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Motivation & Overview (1)
In the absence of a bequest motive, annuities are veryattractive insurance instruments in the presence of longevityrisk (Yaari, 1965; Davidoff, Brown, Diamond, 2005)
The macroeconomic literature has mostly restricted attentionto the case of perfect annuities, one in which annuity rates areactuarially fair.
In reality, “money’s worth” calculations reveal that annuityrates are typically not actuarially fair. Possible reasons:
Administrative costs incurred by the annuity companiesnecessitates a loading factor [ignored].Imperfect competition in the annuity sector [ignored].Adverse selection effects: people who believe themselves to behealthier than average are more likely to buy annuities[stressed in this paper].
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 4 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Motivation & Overview (2)
Based on: Heijdra, B.J. & Reijnders, L.S.M. (2009),Economic growth and longevity risk with adverse selection.CESifo Working Paper, Nr. 2898, December 2009 (Rev.November 2010).
Objective of the paper: to study the growth and welfareimplications of adverse selection effects in the annuity market.
Key model features:
Endogenous growth model of a closed economy.Overlapping generations of heterogeneous finitely-lived agents.Privately known mortality process.Competitive annuity market, offering linear annuity contracts(cf. Pauly, 1974; Abel, 1986).
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 5 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Motivation & Overview (3)
Main findings of the paper:
If health status were observable, there would be a separatingequilibrium (SE). Each health type would get actuarially fairperfect insurance.With unobservable health, there is a pooling equilibrium (PE).The healthy have an incentive to misrepresent their healthstatus. The healthy get a better than actuarially fair rate,whilst the unhealthy get a less than actuarially fair rate.In the PE the unhealthy encounter a self-imposed borrowingconstraint if they live long enough.For a plausible calibration, the SE is slightly better in welfareterms than the PE.
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 6 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Individual agents (1)
Finitely-lived agents of different health types
Lifetime utility function of type-j agent:
Λj (v, v) =
∫ v+Dj
v
cj(v, τ )1−1/σ − 1
1− 1/σe−ρ(τ−v)−Mj(τ−v)dτ,
(1)
Dj is the maximum attainable agecj (v, τ ) is consumption of vintage-v agent at time τρ is the pure rate of time preferenceσ is the intertemporal substitution elasticityMj (τ − v) ≡
∫ τ−v
0µj(s)ds is the cumulative mortality rate
µj (s) is the instantaneous mortality rate at s (0 ≤ s ≤ Dj ,µ′
j (s) > 0, µ′′
j (s) > 0, and lims→Djµj (s) = +∞).
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 8 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Individual agents (2)
Household budget identity:
˙aj (v, τ ) = [r + pj (τ − v)] · aj (v, τ ) + w (τ)− cj (v, τ ) (2)
aj (τ ) is financial assetsr is the interest rate; r + pj (s) is the annuity rate of interestw (τ) is the wage rate
Labour supply is exogenous and equal to unity, so w (τ) alsostands for the household’s wage income
At time v, the agent chooses time paths for cj (v, τ ) andaj (v, τ ) (τ ≥ t) in order to maximize (1) subject to (2),taking into account the initial and terminal conditions onassets, aj (v, v) = aj
(
v, v + Dj
)
= 0, and the TVC.
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 9 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Individual agents (3)
Solutions:
˙cj(v, t)
cj(v, t)= σ
[
r + pj (t− v)− µj (t− v)− ρ]
(3)
cj(v, v) =
∫ v+Dj
v w (τ) e−r(τ−v)−Pj(τ−v)dτ∫ v+Dj
v e−(1−σ)[r(τ−v)+Pj(τ−v)]−σ[ρ(τ−v)+Mj(τ−v)]dτ(4)
aj (v, t) e−r(t−v)−Pj(t−v) =
∫ t
vw (τ) e−r(τ−v)−Pj(τ−v)dτ
−cj(v, v)
∫ t
ve−(1−σ)[r(τ−v)+Pj(τ−v)]−σ[ρ(τ−v)+Mj(τ−v)]dτ (5)
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 10 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Demography
Non-zero population growth
Population proportion of type-j people, πj, is constantDemographic steady state:
βj ·
∫ Dj
0e−ns−Mj(s)ds = 1 (6)
βj is the crude birth rate of type j cohortsn is the growth rate of the population (for a given value of nand a given mortality process Mj (s), (6) defines the coherentsolution for βj)
Relative cohort size of type j agents of age t− v evolvesaccording to:
lj (v, t) ≡Lj (v, t)
L (t)=
βjπje−n(t−v)−Mj(t−v) for 0 ≤ t− v ≤ Dj
0 for t− v > Dj
(7)
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 11 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Figure 1(a): Demographics, surviving fraction
e−Mj(u)
20 30 40 50 60 70 80 900
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
biological age (u+18)
HealthyUnhealthy
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 12 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Figure 1(b): Demographics, instantaneous mortality rate
µj(u)
20 30 40 50 60 70 80 900
0.01
0.02
0.03
0.04
0.05
0.06
biological age (u+18)
HealthyUnhealthy
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 13 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Aggregate household sector (1)
Focus on steady-state growth path:
w (t) = w (v) · eg(t−v) (8)
Per capita consumption:
cj (t) ≡
∫ t
t−Dj
lj (v, t) cj (v, t) dv
Can be written as:
cj (t)
w (t)= βjπj
[
cj(v, v)
w (v)
∫ Sj
0e−(n+g)s−(σ+1)Mj(s)+σ(r−ρ)s+σPj(s)ds
+
∫ Dj
Sj
e−ns−Mj(s)ds
]
(9)
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 14 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Aggregate household sector (2)
Per capita asset holdings of type j agents,aj (t) ≡
∫ tt−Dj
lj (v, t) aj (v, t) dv, evolves over time accordingto:
aj (t) = (r − n) aj (t) + πjw (t)− cj (t)
+
∫ t
t−Dj
[
pj (t− v)− µj (t− v)]
lj (v, t) aj (v, t) dv (10)
Per capita total assets, a (t) ≡∑
j aj (t), satisfy thedifferential equation:
a (t) = (r − n) a (t) +w (t)− c (t) (11)
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 15 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Motivation & OverviewModel: householdsModel: firms
Firms (1)
Firm-level relationships:
Yi (t) = Ω (t)Ki (t)ε Li (t)
1−ε (13)
w (t) = (1− ε)Ω (t) ki (t)ε (14)
r (t) + δ = εΩ (t) ki (t)ε−1 (15)
External effect as in Heijdra & Mierau:
Ω (t) = Ω0k (t)1−ε (16)
Aggregate relationships:
Y (t) = Ω0K (t) (17)
w (t)L (t) = (1− ε)Y (t) (18)
r (t) = r = εΩ0 − δ (19)
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 17 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Assumption about the annuity market
Assumption (A1) The annuity market is perfectly competitive. Alarge number of firms offer annuity contracts toindividuals. Firms entry and exit is unrestricted.
Assumption (A2) Annuity firms do not use up any real resources.
Assumption (A3) The annuitant’s health status is privateinformation and cannot be observed by the annuitycompanies. Annuity firms know all the features of themortality process of each health group.
Assumption (A4) The annuitant’s age is public information andcan thus be observed by the annuity companies.
Assumption (A5) Annuitants can buy multiple annuities fordifferent amounts and from different annuity firms.Individual annuity firms cannot observe an annuitant’sholdings with their competitors.
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 19 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Rothschild-Stiglitz-Wilson versus Pauly-Abel (1)
Most papers on adverse selection use Rotschild-Stiglitz (1976QJE ) or Wilson (1977 JET ) equilibrium concept (R-S-Whereafter).
non-linear pricing (price and quantity controlled)requires annuity companies (a) to perfectly monitor annuitypurchases of all annuitants and (b) to prevent customers frombuying more than one annuity contract.relevant for some insurance markets but not for annuities.
Reasons for rejection of R-S-W:
Monitoring is very difficult. Receipts of annuity payments fromother annuity firms are not observable.Death of the annuitant ends the liability of the annuity firm:withholding payments and investigating compliance notpossible (as they are with fire- or car insurance)Menus with co-payments (as with medical insurance) areinfeasible.
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 20 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Rothschild-Stiglitz-Wilson versus Pauly-Abel (2)
In favour of Pauly (1974 QJE ) – Abel (1986 Ectrica) linearpricing:
Quantity of annuity purchases cannot be controlled, only theprice of annuities.Alternative interpetation: Nash equilibrium. Firms deviatingfrom the zero-profit price incur losses.
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 21 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Separating equilibrium (1)
If health type is observable to annuity firms:
pj (u) = µj (u)
Full longevity risk insurance
Analytical model in Table 1
Visualized in Figures 2(a)–(d)
Quantified in Table 2
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 23 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Table 1: Balanced growth in the separating equilibrium
(a) Microeconomic relationships:
cj (v, v)
w (v)=
∫ Dj
0 e−(r−g)s−Mj(s)ds∫ Dj
0 e−(1−σ)rs−σρs−Mj(s)ds(T1.1)
(b) Macroeconomic relationships:
c (t)
w (t)=
∑
j
βjπjcj (v, v)
w (v)
∫ Dj
0e−(n+g)s−Mj(s)+σ(r−ρ)sds (T1.2)
g ≡k (t)
k (t)= r − n+
[
1−c (t)
w (t)
]
·w (t)
k (t)(T1.3)
w (t)
k (t)= (1− ε)Ω0 (T1.4)
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 24 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Table 2: Growth and retirement: quantitative effects(a) SE (b) PE (c) SE (d) PENo retirementb With retirementc
cH (v, v)
w (v)0.7668 0.7396 0.8267 0.8147
cU (v, v)
w (v)0.7858 0.7775 0.8492 0.8476
SH (years) DH DH DH DH
SU (years) DU 61.63 DU 63.93
c (t)
w (t)1.0714 1.0753 0.8966 0.8979
g (%year) 2.00 1.89 2.00 1.96
w (t)
k (t)0.2800 0.2800 0.3346 0.3346
ΛH (v0, v0) 27.4619 26.4639 11.9477 11.8281
ΛU (v0, v0) 21.8047 20.6189 10.3314 9.7873
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 25 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Figure 2(a): Separating equilibrium, scaled individual
consumption
cj(v,v+u)w(v)
20 30 40 50 60 70 80 900
1
2
3
4
5
6
7
8
9
biological age (u+18)
HealthyUnhealthy
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 26 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Figure 2(b): Separating equilibrium, scaled cohort assets
aj(v,v+u)w(v)
20 30 40 50 60 70 80 900
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
biological age (u+18)
HealthyUnhealthy
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 27 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Separating equilibrium (2)
Proposition 1. Consider the separating equilibrium (SE) inwhich annuity firms can observe the health type of annuitants.Provided σr − g > ρ, agents of all health types are net saversthroughout life, i.e. aj(v, v) = aj(v, v + Dj) = 0 andaj(v, v + u) > 0 for 0 < u < Dj .
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 28 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Pooling equilibrium (1)
If health type is unobservable to annuity firms, the competitivepooling rate is:
p (u) =
µH (u) · aH (v, v + u) + µU (u) · aU (v, v + u)
aH (v, v + u) + aU (v, v + u)for 0 < u ≤ DU
µH (u) for DU < u < DH
(25)
For 0 < u ≤ DU , µH (u) ≤ p (u) ≤ µU (u). No actuarialfairness.
Analytical model in Table 3
Visualized in Figures 3(a)–(d)
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 30 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Pooling equilibrium (2)
Proposition 2. Consider the case in which annuity firms areunable to observe the health type of annuitants. Assume thata pooling equilibrium (PE) exists and that σr − g > ρ. Then:
(i) healthy agents are net savers throughout life, i.e.aH(v, v) = aH(v, v + DH) = 0 and aH(v, v + u) > 0 for0 < u < DH ;
(ii) unhealthy agents are net savers until age SU < DU afterwhich they adopt a self-imposed borrowing constraint, i.e.aU (v, v) = 0, aU (v, v + u) > 0 for 0 < u < SU , andaU (v, v + u) = 0 and cU (v, v + u) = w (v + u) forSU ≤ u ≤ DU .
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 31 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Table 3: Balanced growth in the pooling equilibrium
(a) Microeconomic relationships:
cH (v, v)
w (v)=
∫ DH0 e−(r−g)s−P(s)ds
∫ DH0 e−ρ∗s−(1−σ)P (s)−σMH (s)ds
(T3.1)
cU (v, v)
w (v)=
∫ SU0 e−(r−g)s−P (s)ds
∫ SU0 e−ρ∗s−(1−σ)P (s)−σMU (s)ds
(T3.2)
cU (v, v)
w (v)= e
−(σ(r−ρ)−g)SU+σ[MU (SU )−P (SU )] (T3.3)
aH (v, v + u)
w (v)= βHπHe
(r−n)u−MH (u)+P(u)[ ∫
u
0e−(r−g)s−P(s)
ds
−cH (v, v)
w (v)
∫
u
0e−ρ∗s−(1−σ)P(s)−σMH (s)
ds
]
, (0 ≤ u ≤ DH ) (T3.4a)
aU (v, v + u)
w (v)= βUπU e
(r−n)u−MU (u)+P(u)[ ∫
u
0e−(r−g)s−P (s)
ds
−cU (v, v)
w (v)
∫
u
0e−ρ∗s−(1−σ)P (s)−σMU (s)
ds
]
, (0 ≤ u < SU ) (T3.4b)
aU (v, v + u)
w (v)= 0, (SU ≤ u ≤ DU ) (T3.4c)
p (u) =µH (u)aH (v, v + u) + µU (u)aU (v, v + u)
aH (v, v + u) + aU (v, v + u)(T3.5)
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 32 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Table 3: Balanced growth in the pooling equilibrium
(b) Macroeconomic relationships:
c (t)
w (t)= βHπH
cH (v, v)
w (v)
∫
DH
0e(r−n−g−ρ∗)s−(1+σ)MH (s)+σP (s)
ds
+βUπU
[
cU (v, v)
w (v)
∫
SU
0e(r−n−g−ρ∗)s−(1+σ)MU (s)+σP (s)
ds
+
∫
DU
SU
e−ns−MU (s)
ds
]
(T3.6)
g ≡k (t)
k (t)= r − n +
[
1 −c (t)
w (t)
]
w (t)
k (t)(T3.7)
w (t)
k (t)= (1 − ε)Ω0 (T3.8)
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 33 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Figure 3(a): Pooling equilibrium, scaled individual
consumption
cj(v,v+u)w(v)
20 30 40 50 60 70 80 900
1
2
3
4
5
6
7
8
9
biological age (u+18)
HealthyUnhealthyWage income
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 34 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Figure 3(b): Pooling equilibrium, scaled cohort assets
aj(v,v+u)w(v)
20 30 40 50 60 70 80 900
0.02
0.04
0.06
0.08
0.1
0.12
biological age (u+18)
HealthyUnhealthy
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 35 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Figure 3(c): Pooling equilibrium, relative pooling rate (U)
p(u)− µU (u)
20 30 40 50 60 70 80 90−2
−1.8
−1.6
−1.4
−1.2
−1
−0.8
−0.6
−0.4
−0.2
0
biological age (u+18)
annu
al p
erce
ntag
e po
ints
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 36 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Figure 3(d): Pooling equilibrium, relative pooling rate (H)
p(u)− µH(u)
20 30 40 50 60 70 80 900
0.1
0.2
0.3
0.4
0.5
0.6
0.7
biological age (u+18)
annu
al p
erce
ntag
e po
ints
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 37 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Welfare analysis
Clearly ΛSEj (v0, v0) > ΛPE
j (v0, v0) for j ∈ H,U
Lost growth years (were you born too early?):
LGYij =
1
gi·ΛSEj (v0, v0)− Λi
j (v0, v0)∫ Dj
0 e−ρs−Mj(s)ds. (28)
From Table 2:
Difference between SE and PE is smallFirst version of the paper: difference between PE and NAE islarge. Depends on recycling of accidental bequests. Tragedy ofannuitization.Adverse selection problems are small!
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 38 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Informational assumptionsSeparating equilibriumPooling equilibrium
Robustness check
Two conclusions emerge from the analysis conducted thus far:
Conclusion (C1): Unhealthy individuals adopt a self-imposedborrowing constraint fairly early on during old age.Conclusion (C2): The welfare effects of the annuity marketimperfection due to asymmetric information are rather modest.
Robustness checks with respect to:
Mandatory retirement under a PAYG systemEndogenous versus exogenous growth model
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 39 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Main findings
SE is slightly better in welfare terms than PE.
But an H-type individual has an incentive to misrepresenthealth status. SE is not credible.
In PE, H-types benefit for part of life from presence ofU -types.
In PE, U -types run out of assets if they live long enough.
Conclusions are robust. They hold under mandatory retirementand also in an exogenous growth model.
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 40 / 41
Motivation & ModelAnalysis
Conclusions & Further Research
Extensions
Endogenous labour supply, optimal retirement, and growth.
Transitional dynamics.
Health shocks
Differential productivity.
Long-term annuities and lock-in effects
NAKE Dynamic Macroeconomic Theory Lecture 3C (January 13, 2012) 41 / 41