Utility Maximising Design of Means Tested Retirement Benefits

1
Utility Maximising Design of Means Tested
Retirement Benefits 
James Sefton Justin van de Ven

• National Institute of Economic and Social
  Research
• jvandeven_at_niesr.ac.uk
• July 2007

2
Outline

• Policy Background
• Costs and Benefits of Means Testing
• Framework of Analysis
• analytical question
• analytical method
• policy comparisons
• Selected Results
• Summary and Directions for Further Research

3
Policy Background

• Increased in the UK following introduction of the
  Minimum Income Guarantee in 1997
• Withdrawal rates were lowered with introduction
  of the Pension Credit in 2003
• Pensions Commission recommendations
• will see means testing scaled back over a period
  of time

Reforms to make the state system less
means-tested and closer to universal than it
would be if current indexation arrangements were
continued indefinitely. Pensions Commission
(2005)
4
Criticisms of Means Testing

• Economic distortions
• Means testing of discourages private provision
  for retirement
• Take up
• Means tested benefits commonly suffer from low
  rates of take up in the UK
• Politically backward
• Means testing is adapted for application at the
  family or household level, which runs counter to
  welfare provision to the individual

5
So why Bother with Means Testing?

• Means testing limits the costs of benefits
  provision by targeting benefits receipt
• A satisfactory policy evaluation must take into
  account the distortions associated with the
  alternative methods of benefits finance

Every citizen, rich or poor, receives the same
tax-free Basic Income but the Surcharge on the
first slice of other income is the equivalent of
a withdrawal of part of the Basic Income, not
pound for pound but, say, one pound for every two
pounds earned. The Surcharge may weaken but will
not eliminate the incentive to earn more income
it can however, enormously reduce the cost of the
Basic Income. Can we learn a third way from
the Agathotopians?, Meade (1993, p. 94).
6
Framework of Analysis

• The analysis is based on the following thought
  experiment
• Suppose that each birth cohort is asked, at an
  age before they have learned anything about their
  individual specific circumstances, to select
  their preferred policy specification from a menu
  that is limited to revenue neutral alternatives.
  Given the assumptions embodied by our analytical
  model, what combination of policy parameters is a
  cohort likely to choose?

7
Framework of Analysis

• Compare various pension policy regimes in terms
  of the expected lifetime utility of a
  representative household.
• behind a veil of ignorance Rawls (1971)
• In the spirit of optimal tax literature that
  focuses upon the social insurance-efficiency
  trade-off
• Mirrlees (1974), Varian (1980), Low and Maldoom
  (2004)
• The motive for redistribution here is not a
  desire for equity per se, but rather a desire for
  social insurance Varian (1980, p. 51)

8
Relevant Results

• Means-testing of pension benefits is a non-linear
  capital tax in retirement.
• If preferences for different goods (i.e. leisure
  and consumption) are separable Atkinson and
  Stiglitz (1976) demonstrate that capital taxes
  are suboptimal.
• However Browning Meghir (1991), Heckman (1974)
  argue that consumption and leisure are not
  separable.
• BM reject separability at all standard
  confidence levels.
• Work and Consumption track over the lifetime
• (The current trend for downsizing the work/life
  balance!)
• Does this generate a significant role for
  means-testing?

9
Model Description

• Numerical simulations based on an articulated
  structural Dynamic Programming model
• households choose consumption and labour supply
  to maximise expected lifetime utility subject to
  the prevailing policy environment, period
  specific liquidity constraints, and uncertainty
  regarding labour incomes and the time of death
• lifetime divided into annual increments from age
  20 to 110
• detailed description of the UK tax and benefits
  system

10
Model Description

• Objective function
• The model is specified to ensure that control
  variables only have a direct effect on the
  immediately succeeding state variables through
  two pathways
• Intertemporal budget constraint (the entire
  lifetime)
• Potential labour income (the working lifetime)

11
Policies Considered for Analysis

• Three policy parameters are subject to variation
• the value of means tested retirement benefits
• the withdrawal rates on means tested benefits
• tax rates during the working lifetime
• Sensitivity analysis is conducted with regard to
  the tax adjustments considered during the working
  lifetime
• proportional tax poll tax

12
Diagrammatically
From Age 65
Below Age 65
BSP mtb
proportional tax
poll-tax / universal benefit
13
Selected Results

• Policy parameters that describe local maxima for
  expected lifetime utility under our analytical
  framework
• prop tax value of benefit 3.9 x value at 2003
• withdrawal rate 70
• poll tax value of benefit 1.8 x value at 2003
• withdrawal rate 70
• A significant role for means testing of
  retirement benefits
• Sensitivity analysis no variation of BSP

For single from 24.65 p.w. to 96.14 (BSP
77.45 p.w.) For couple from 32 p.w. to
124.80 (BSP 123.80 p.w.)
14
The Proportional Tax Scenario
15
Effects for 100 Withdrawal Rate
16
Effects when Means Tested Benefit 3.9 x 2003
Value
17
Summary and Directions for Further Research

• Given the assumptions embodied by our analytical
  model, the current analysis suggests that there
  is a substantial role for means testing in the
  provision of retirement benefits
• these results are in the absence of an exogenous
  preference for equity
• The need for further sensitivity analysis
• preference parameters
• policy environment