Date post: | 15-Jan-2016 |
Category: |
Documents |
Upload: | bruce-hildreth |
View: | 216 times |
Download: | 0 times |
The dark side of wage indexed The dark side of wage indexed pensionspensions
Evert Carlsson
&
Kalle Erlandzon
FUR 2006
RationaleRationale
• This paper investigates some welfare effects of forced saving through a mandatory pension scheme.
• The framework for the analysis is a life-cycle model of a borrowing constrained individual’s consumption and portfolio choice in the presence of:– Uncertain labour income– Realistically calibrated tax and pension
systems.
The Swedish mandatory The Swedish mandatory pension schemepension scheme
• Launched in 1999, the reformed Swedish pension system has attracted a lot of interest around the world.
• 16 % of all wages and benefits are paid as contributions to an individual notional account.
• The return on these accounts are set to equal aggregate labour income growth.
• The system is actually a PAYGO system, but it is set up as a defined contribution system.
• Therefore, the system is called a Notional Defined Contribution system – NDC.
Past researchPast research
• Modigliani & Brumberg(54), Friedman (AER57): No correlation between predictable income changes and consumption. However, empirical data shows a positive correlation.
• Deaton (EM91), Carroll(QJE97), Gourinchas & Parker(EM02): Life-cycle models with borrowing constraints, stochastic labour income and consumption choice. These models reconcile PIH with data.
• Cocco et al. (ReFinStud05): Extends Carrol by including portfolio choice.
• Campbell et al. (2001): Extends Cocco et al. to include an additional state variable, retirement wealth.
ContributionContribution
Our model extends Campbell et al. by including a realistically calibrated tax and pension system. We can therefore:
• attribute a value to mandatory pension savings,
• analyse the welfare effects of pension returns being linked to aggregate labour income growth.Results:– Young constrained individuals attribute little
value to their pension savings.– The welfare loss associated with uncertain
pension returns stems primarily from the dependency between labour income growth and pension returns rather than the volatility of pension returns.
Model – Individual Model – Individual preferencespreferences
C – consumption – discount factor – relative risk aversionpj – conditional probability to survive
b – bequest parameterD – bequest amount
Model – Labour incomeModel – Labour income
likt – log labour real income
f(t,Zit) – deterministic function of characteristics
it ~N(0,) idiosyncratic temporary shock
uit ~N(0,) permanent shock
t ~N(0,) group aggregate component
it ~N(0,) idiosyncratic component
Model – Mandatory savings Model – Mandatory savings & retirement benefits - & retirement benefits - NDCNDC
NDC – Notionally defined contribution accountRl – Return of NDC pensionsl – Expected national income growth ~N(0,) national aggregate component
L – Gross labour incomePO– Annualised payout from NDC
Model – Mandatory savings Model – Mandatory savings & retirement benefits – & retirement benefits – Defined benefitDefined benefit
Defined benefit payout
0
100
200
300
400
500
600
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Gross income
Payout
Model – TaxesModel – Taxes
Model - AssetsModel - Assets
Rs - After tax real simple risky returnRf - After tax real simple risk-free returns - Expected risk premium- Innovation in excess returns, ~N(0,)
- Correlation between excess returns and the aggregate component of permanent innovations to labour income
Model – Private savings & Model – Private savings & consumptionconsumption
X – Cash on HandTwo control variables: – consumption share– risky weightF – initial wealth
Model – OptimizationModel – Optimization
Optimization problem now has4 state variables , v, X and NDC2 choice variables and 4 stochastic variables u and
V – Value function – Vector of state variables
Risky and consumption weight Risky and consumption weight at age 64at age 64holding NDC constantholding NDC constant
v
X
v
X
Calibration - EstimationCalibration - Estimation
• Income is defined as all taxable social benefits and wages for an individual before taxes.
• The components of the labour income process were estimated using LINDA panel data 1992-2002.
• LINDA is a register based data bas covering 3.35 % of the Swedish population.
• The data was divided into six non-intersecting groups defined by sex and educational status.
CalibrationCalibration
Our estimates of transitory shocks are lower than in C&S and CGM:- LINDA is register based data while C&S and CGM used survey
data, which is less precise (measurement errors will increase estimated variance).
Calibration – deterministic Calibration – deterministic labour income profile – elabour income profile – ef(t, f(t,
Z)Z)
0
100
200
300
400
500
600
700
800
900
22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64
Univ Men
Comp Wom
Calibration – Key Calibration – Key parametersparameters
Aggregate labour income growth
l 1.8%
Risk-free rate Rf-1 1.5%
Expected risky premium s 3%
Asset volatility 20%
Coefficient of relative risk aversion
5
Discount factor 0.98
Bequest parameter b 1
Volatility in agg. labour income growth
2%
Averages from the Averages from the simulated distributionssimulated distributions
Frequency distributions – Frequency distributions – log of Xlog of X
log(X)
Age
Frequency distributions – Frequency distributions – risky weightrisky weight
Age
Frequency distributions – log Frequency distributions – log of riskyof risky
Age
Log[(X-C)]
Frequency distributions – Frequency distributions – consumption shareconsumption share
Age
Valuing the NDC Valuing the NDC accountaccount
• Forced saving when wage profile is increasing and consumption is preferred
• NDC account cannot serve the precautionary motive
Risk in the NDC accountRisk in the NDC account
• The model has five assets: risk-free, risky, defined benefit, future wages and the NDC asset.
• In the reference case NDC return is volatile and correlated with future wages and defined benefits as well as with the risky asset.
• Volatility (2%) in the NDC return is low, but NDC is the largest asset at retirement.
Risk in the NDC accountRisk in the NDC account
• Two alternative regimes:– NDC return is risk-free– NDC return is independent (but volatile)
• Gain in consumption and bequest equivalent units(a 1% increase represents the increase in utility that would be produced by a 1% increase in cons. and beq. for the rest of the life)
Risk in the NDC accountRisk in the NDC account
• Group 3 is men with a university degree.
• Group 4 is women with compulsory school only.
Risk in the NDC accountRisk in the NDC account
• We note that 2/3 of the gains originates from the elimination of the dependency. This holds irrespective of age.
• Without a risky asset in the model, the gains would have been generally underestimated.– After retirement, the
gains from independency would have been zero and very small in the risk-free case.
Risk in the NDC accountRisk in the NDC account
Risky weight
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Age
Risk-free NDC
Reference
Robustness & sensitivityRobustness & sensitivity
• Conservative parameterisation wrt. the valuation of the NDC account.– High NDC return, low equity premium,
high discount factor and low correlation.
• Two tests:– Lowering the risk aversion from 5 to 2.– Increasing the equity premium from 3%
to 4%.
ConclusionConclusion
• This paper contributes to the understanding of the risk characteristics and welfare effects of NDC pension systems.
• First, individuals attribute little value to their pension savings in early life.
• Second, it is the dependency rather than the
volatility of the NDC returns which is of most importance.
Total inkomst per gruppTotal inkomst per grupp
Total inkomst över 10 Total inkomst över 10 basbeloppbasbelopp
Individual wealth – cross-Individual wealth – cross-section 2002section 2002
Medi an 75% per cent i l e
1000 SEK
- 50
0
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
800
850
900
950
1000
Age
20 30 40 50 60 70 80 90 100
Robustness & sensitivity – Robustness & sensitivity – Reference caseReference case
Robustness & sensitivity – Robustness & sensitivity – Valuing NDCValuing NDCHigher risk premium, makes the NDC asset less attractive.Lower risk aversion:• Higher acceptance of risk makes the NDC less
desirable and the short-sales constraint often effective.• Higher elasticity of intertemporal substitution makes
the agent willing to substitute consumption over time (variation across time is less disadvantageous).
Robustness & sensitivity – Robustness & sensitivity – Valuing NDCValuing NDC• In early life, these two effects cancel out.• In mid life, as consumption reaches the lifetime average,
the effect of a higher intertemporal substitution diminishes.
• Finally, in late life the ratio of private to NDC wealth increases giving a higher preference for risk-free assets.
Robustness & sensitivity – Robustness & sensitivity – Risk CharacteristicsRisk CharacteristicsRisk-free or independent NDC return
Higher premium is associated with more risky assets and hence larger gains from a risk-free or independent NDC.Making the NDC risk-free or independent for an agent with lower risk aversion has less effect.
The proportional gain that stems from independency is still 2/3.
Contour plots of risky and Contour plots of risky and consumption weight at age 64 consumption weight at age 64 holding NDC constantholding NDC constant
X
v v
Frequency distributions – Frequency distributions – marginal utilitymarginal utility