Transcript Use Arial Narrow in grey

Incentivising Retirement Saving –
A Waste of Money?
AUG 2009
Context
Social security and retirement funding review by
government
 Re-evaluation of equity of tax incentives
 ‘Gap’ solution needed for those for whom tax relief means
little
 Competing national budget priorities

Context
5.0
Formally employed excludes
agricultural workers
4.5
People in Millions
4.0
Formally Employed = 9.3m
Existing Provision = 5.8m
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
< R12k
R12 - R75k
R75k - R150k
R150k - R1m
Annual Income
Formally Employed
Existing Provision
Courteousy Old Mutual
> R1m
Retirement Saving: Typical Policy Levers
1.
2.
3.
4.
5.
Increasing income
Improved returns (credits, contribution matching & tax
incentives)
Financial education (informing rationality)
Financial regulation (essential for long-term time horizons)
Mechanical inducement to counter inertia, procrastination
& myopia (auto-enrolment, compulsion, limits on liquidity
& preservation)
‘Improved Returns’ via the Tax System
1.
2.
3.
4.
5.
Tax allowance (fixed percentage from taxable income)
Tax deduction (deducted in proportion to income level)
Tax credit (fixed amount deducted from tax liability)
Tax exemption (part or source is exempted from tax)
Preferential tax rate (preferring certain income or sources)
Distributional Effects of Relief Mechanisms
Antolin and Ponton (2007)
Relief Mechanisms for Retirement Saving


TEE / ETE / EET
General international practice
1.
2.
3.
4.
Most countries provide tax relief for mandated systems
In unfunded systems, contributions normally exempt, while
benefits are fully taxed
For funded systems, same as (2) and the interest earned is
not normally taxed (EET)
For retirement income: consumption-like tax is applied
(exempts at accumulation stage, but taxes when drawn
down at decummulation)
Holtzmann and Hinz (2005)
Design in Retirement Systems
Yoo & de Serres (2004)
Estimated Tax Cost for SA
•
Deduction
Fund income at 18% (no CGT)
Lump-sum formula
Total (at least)
•
R31.5 bn = 1.9% of GDP vs 1.7% in Ireland & UK
•
•
•
R 27.0 bn
R 4.5 bn
?
R 31.5 bn
Tax Statistics: NT, SARS (2008)
EU Social Protection Committee (2008)
Literature: Effect of Subsidies on Saving
•
•
•
•
Ambiguous in theory. Difficult to estimate response – people
might simply spend more in the present rather than save more.
(Neuberger and McCarthy:2004)
Literature divided on empirical results, but on balance favours
positive response, especially with lower income cohorts.
Besley & Meghir (1998) call it at best a marketing opportunity
for governments and Antolin et el (2004) estimate range of new
saving in contributions of between 25% to 40%.
For funded retirement saving, offset against household saving
quite high, greater redistributive component, less offset
(Disney,2005).
Effectiveness of Incentives re New Saving
Blundell, Emmerson & Wakefield (2006)
Funded Retirement Sav on Household Saving
Funded Retirement Sav on Household Saving
A Few Further Pointers from the Literature
•
Consumer as rational optimiser:
•
50% of working adults say they have ‘no idea’ of their likely
retirement income; those with 15yrs to retirement, 50% say
they do not know if they have sufficient to retire on
(Mayhew:2003)
•
Optimisation requires complex calcs, past experience
limited (only retire once) and info from others limited.
People are myopic and use ‘rules of thumb’ (Thaler: 1994)
•
People tend to choose a lifetime savings pattern separately
from distribution. Rise in pension or housing wealth offsets
other saving, especially when close substitutes (Hawksworth:
2006) - e.g. provident funds vs annuitisation
Conclusions on Incentives ?
Political Economy
1.
2.
3.
4.
5.
Low-income groups like large, highly distributive systems
Middle income groups favour earnings related systems
High-income earners prefer no system at all – have access to
private systems
Acknowledgement at least of some form of altruism leads to
solidarity between these groups
The tax system plays a vital role in re-distribution and cannot be
viewed in isolation
Conde-Ruiz and Profeta (2002)
Net Replacement Rates (Mandatory Schemes)
Percentage Change between Gross and Net Replacement, Mandatory Schemes
Calculated using Whitehouse (2007), World Bank
Regressive Bias in Incentives
Regressive Nature of Tax Incentives (individual earnings, multiple of average)
Calculated using Whitehouse (2007), World Bank
Introduction to CGE Modelling






Based on previous modelling done by Go, Kearney, Korman,
Robinson and Thierfelder (2008).
Utilise a Computable General Equilibrium (CGE) model.
Extended the modelling to include the informal sector as a
recipient of the wage subsidy.
Modelled a 5,10 and 15 percent wage subsidy to medium and low
skilled workers in the formal and informal sector.
Modelling is performed against different assumptions of labour
market flexibility; medium and low market flexibility is
investigated.
Assume that the informal labour category is unemployed. The
demand for more workers will therefore lead to an increase in
employment, while real wages will remain constant.
Background to the CGE Model Used










Follows modelling tradition set by Dervis and De Melo (1982).
Solved using GAMS software program.
Set of equations solved simultaneously.
Imposes a structure of behaviour based on microeconomic theory.
Neoclassical assumptions:
 Optimising behaviour
 Non-linear first order conditions
 Maximise profits and utility
Demand equals supply in goods and factor market.
Imposes a relationship between prices and taxes.
Most important data source is a Social Accounting Matrix (SAM).
For this analysis a 2005 SAM is used as compiled by Quantec.
Labour is highly disaggregated in this SAM for the purpose of this
analysis (9 labour categories).
Schematic View of the Model
Factor
Costs
Activities
Domestic Private Savings
Factor
Markets
Wages
& Rents
Intermediate
Input Cost
Gov. Savings
Taxes
Households
Government
Sav / Inv
Transfers
Private
Consumption
Sales
Commodity
Markets
Exports
Imports
Rest of the
World
Government
Consumption
Investment
Demand
Foreign Transfers
Foreign Savings
LLöfgren, et al (2001)
21
Results - Summary
Employment (% change)
GDP (% change)
Initial wage subsidy shock (R billion)
Wage subsidy cost (R billion)
Net cost (R billion)
Cost per job created (R million)





Wage Subsidy to Formal Semi- and
Unskilled Labour
5%
10%
15%
2.09
4.34
6.80
0.86
1.75
2.67
12.18
24.37
36.55
12.65
26.33
41.16
-6.77
-14.33
-22.79
48585.23
48570.02
48543.99
Wage Subsidy to Formal Semi- and
Unskilled Labour as well as Informal
5%
10%
15%
3.01
6.28
9.87
0.97
1.97
3.01
13.75
27.51
41.26
14.30
29.79
46.64
-7.72
-16.35
-26.04
38087.33
37993.16
37887.16
Medium elasticity case shown.
Size of wage subsidy in 10% case is approximately 1,5 percent of GDP or
R24,3 billion in 2005 terms.
As a result of the additional economic activity generated the net cost is
R14,3 billion.
When the subsidy is extended to the informal sector the cost increases
by R3,1 billion, and the net cost to R16,3 billion.
The cost per job created falls to R37 993 as more jobs per rand spend is
created when the subsidy is extended to the informal sector.
Results – Employment and GDP
Results - Employment
High-Skilled Formal Employed
Skilled Formal Employed
Semi- and Unskilled Formal Employed
Informal Employed
Self Employed
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Wage Subsidy to Formal Semi- and
Unskilled Labour
5%
10%
15%
0.447
0.908
1.384
0.247
0.497
0.753
-0.035
-0.071
-0.108
4.330
9.060
14.250
2.688
5.589
8.732
3.115
6.505
10.211
4.330
9.060
14.250
2.088
4.341
6.782
3.418
7.139
11.208
1.168
2.379
3.636
0.804
1.630
2.481
0.770
1.561
2.377
0.166
0.339
0.522
0.072
0.146
0.224
-0.029
-0.058
-0.090
Wage Subsidy to Formal Semi- and
Unskilled Labour as well as Informal
Labour
5%
10%
15%
0.529
1.075
1.640
0.306
0.618
0.935
-0.043
-0.088
-0.133
4.504
9.428
14.836
2.815
5.853
9.146
3.214
6.712
10.537
4.504
9.428
14.836
2.193
4.558
7.121
3.493
7.295
11.453
4.504
9.428
14.836
3.265
6.792
10.619
3.981
8.318
13.066
0.221
0.453
0.697
0.121
0.247
0.379
-0.042
-0.085
-0.131
Agriculture sees largest percentage gains in employment, however
overall share in total employment relative low.
 Largest employment gains is in services sector.

Results – Equivalent Variation
Gives an approximation of the welfare impact of the wage subsidy.
 Wage subsidy extended to the informal sector increases welfare in
general; but also benefits the poor more relative to the high income
groups.

Results – Low Elasticity Case
Employment (% change)
GDP (% change)
Initial wage subsidy shock (R billion)
Wage subsidy cost (R billion)
Net cost (R billion)
Cost per job created (R million)
Wage Subsidy to Formal Semi- and
Unskilled Labour
5%
10%
15%
1.10
2.25
3.48
0.45
0.91
1.37
12.18
24.37
36.55
12.43
25.38
38.90
-8.23
-16.90
-26.07
90832.26
90220.66
89556.92
Wage Subsidy to Formal Semi- and
Unskilled Labour as well as Informal
5%
10%
15%
1.56
3.21
4.97
0.51
1.02
1.54
13.75
27.51
41.26
14.04
28.68
43.99
-9.34
-19.20
-29.63
72266.70
71613.93
70915.63
Low elasticity case shown.
 Employment gains are much lower; 3.21% of 10% wage subsidy when
extended to informal sector compared to 6.28% for medium elasticity
case.
 Cost per job created is therefore significantly higher.
 How flexible is South Africa’s labour market?

Conclusions






The cost per job created of a wage subsidy is high, but can be lowered if
extended to the informal sector.
The cost per job created is high because all the workers receive the
wage subsidy and not only the new entrants.
It is lower when extended to the informal sector because the average
wage in the informal sector is lower.
The labour market flexibility assumed has a significant impact on the
results. Under a low market flexibility assumption, the employment
gains under a wage subsidy are much lower. How flexible is SA’s labour
market? Literature indicates that SA’s labour market flexibility is
relatively low.
The wage subsidy should therefore, ideally, be accompanied by policies
to improve labour market flexibility in South Africa for it to be effective.
The paper did not consider alternative designs for the wage subsidy.
Tax Wedge as % of GDP SA
%
GDP
Taxes on individuals
Skills Development Levy
UIF
Wage subsidy (gross)
Total
7.9
0.3
0.1
1.9
10.2
Individual’s Tax Disaggregation
Practical Policy




A social security tax of 5% applied to earnings of R120 000 and
more is estimated to yield about R16.5 bn, leaving a considerable
shortfall
This would need to be sacrificed elsewhere or be deficit financed
Hardly the time to raise taxes on employers or individuals
We need an alternative that does not have the potential deadweight loss for the economy, yet is effective ‘at the margin’
An Alternative: State Sponsored Co-contribution
Papke (1995) cited by Neuberger and McCarthy (2004)
An Alternative: State Sponsored Co-contribution
Finding
Author
Existence positively correlated with
participation, level has little effect
Poterba, Venti & Wise (1992)
Corroborate above finding
Papke, Petersen & Poterba (1993)
Participation & contributions insensitive to
match rate over time
Kusko, Poterba & Wilcox (1993)
Existence of match, not quantum that matters
Scott (1994)
Concur with above finding
Papke (1995), Even & Macpherson (1996),
Basset et el (1998), Kusko et el (1998)
Increases in match rate raise saving
Clark and Schieber (1998), VanDerhei and
Copeland (2001), Choi, Laibson et el (2002)
Papke (1995) cited by Neuberger and McCarthy (2004)
An Example: Direct Subsidy for Low Income
Voluntary coverage rate by deciles of income
German Riester Pensions pay a basic subsidy
of €154 per adult, and an additional child
subsidy of €185.
EU Social Protection Committee: 2008
Another Example: New Zealand
In 2005 NZ announced auto-enrolled KiwiSaver, with eight weeks for opt
out. Tax incentives were added on 1 July 2007 after 20 yrs of neutrality:
 New member sign-on incentive of $1000 tax free;
 Matching contribution of up to $20 per week ($1043 per year);
 Subsidy for the purchase of first home of up to $5000;
 Fee subsidy of $40 per year;
 Wage subsidy for employers (offset to compulsory employer
contributions) of up to $20 per week;
 Employer contributions set to rise from 1% of gross pay to 4% by
2011;
 Investment income receives favourable tax treatment.
 2008 Nationwide survey showed auto-enrollees had made a relatively
small contribution by Dec 2007 (33%, 32% opt-out)
 Incentives made the difference – ‘new saving’ range between 9-19 cents
per dollar (Gibson and Le: 2008)

Conclusions

On tax incentives:




Adage: “an old (and well-understood) tax is a good tax” probably implies to
existing retirement incentives too;
30 yr review of literature for Mirrlees Review: incentives do matter, but
results differ based on education and demographics (Meghir and Phillips:
2008)
People use ‘capping’ or ‘limits’ as guides to sufficiency optimisers (Thaler:
1994) and could interpret tax limits as guides to income replacement in
retirement;
Combine existing EET with direct incentives for low income individuals and
employers to overcome initial inertia (sign-up incentive) with cocontribution of ratio which can be gradually phased in (test elasticity);
Achieve ‘solidarity’ through tax mechanism (social security tax) rather
than in the fund, which can increase labour force resistance, distort
actuarial neutrality and ‘line of sight’ for member;
 Use mixture of compulsory annuitisation and commutation to reduce
substitutability of retirement saving for discretionary saving
 Don’t forget the other policy levers (income growth, mechanical
inducement, regulation, education)

Conclusion
“…although common sense has been described
as ‘that most blunt of intellectual
instruments’, it remains the most useful
tool in deciding the issue.”
Leach AJA, 2009, WJ Fourie Beleggings CC v Commissioner for SARS,
Supreme Court of Appeal.
THANK YOU