Paths and Forks or Chutes and
Ladders?: Negative Feedbacks and the
Dynamics of Pension Regime Change
Kent Weaver
Georgetown
University
and the
Brookings
Institution
Policy Regimes, defined:
•
A system of organizing, financing and
delivering policy that:
– Has a distinctive distribution of costs
and benefits
– Is relatively stable over time
Examples:
•
•
•
•
The German pension system
The U.S. health care system
Cheap energy policies
“Socialism market economy” in China
The Questions:
1. How frequent is a major change in
policy regimes?
2. How much do past choices constrain
the range of future regime options?
Do policymakers have significant
discretion in shifting policy regimes?
Is the best metaphor for policy regime
transitions…
Paths and Forks?
Old
103
Path Choice at Some Times and Not at
Others?
Cul-de-sac or “No Exit”?
Boomerang
or policy
reversal?
..or Driving off a Cliff?
Cul-de-sac:
Regime Type
at t1
Regime Type 1
Regime Type 2
Regime Type 3
Regime Type 4
Regime Type 5
Regime Type 6
Regime Type 7
Regime Type
at t2
Unconstrained choice:
Regime Type
at t1
Regime Type 1
Regime Type 2
Regime Type 3
Regime Type 4
Regime Type 5
Regime Type 6
Regime Type 7
Regime Type
at t2
Paths and Forks:
Regime Type
at t1
Regime Type 1
Regime Type 2
Regime Type 3
Regime Type 4
Regime Type 5
Regime Type 6
Regime Type 7
Regime Type
at t2
Chutes and Ladders:
Regime Type
at t1
Regime Type 1
Regime Type 2
Regime Type 3
Regime Type 4
Regime Type 5
Regime Type 6
Regime Type 7
Regime Type
at t2
Mixed Patterns Across Regimes:
Regime Type
at t1
Regime Type 1
Regime Type 2
Regime Type 3
Regime Type 4
Regime Type 5
Regime Type 6
Regime Type 7
Regime Type
at t2
More choice at t1:
Regime Type
at t1
Regime Type 1
Regime Type 2
Regime Type 3
Regime Type 4
Regime Type 5
Regime Type 6
Regime Type 7
Regime Type
at t2
Regime Type
at t3
More choice at t2:
Regime Type
at t1
Regime Type 1
Regime Type 2
Regime Type 3
Regime Type 4
Regime Type 5
Regime Type 6
Regime Type 7
Regime Type
at t2
Regime Type
at t3
Boomerang:
Regime Type
at t1
Regime Type 1
Regime Type 2
Regime Type 3
Regime Type 4
Regime Type 5
Regime Type 6
Regime Type 7
Regime Type
at t2
Regime Type
at t3
3. What factors determine whether
policy regime change occurs?
The Conventional (Piersonian) Wisdom:
•
Once policies are in place they are
generally reinforced by:
–
–
–
•
Adaptive expectations and sunk costs (lock-in
effects)
Political support coalitions that grow up around
the policies
Multiple veto points in political systems
So policy regime change mostly results
from exogenous shocks
A Framework for Explaining Policy
Regime Change:
Much policy regime change has endogenous
roots, and depends on:
• Balance between positive and negative
feedbacks from existing policy
• Incremental reform options available to
policymakers—and whether they have
been exhausted
• Regime transition opportunities available
to policymakers
A Quick Overview of Pension
Policy in Wealthy Countries
OECD
countries
vary
substantially
in what they
spend on
pensions
Pension systems vary substantially across
countries in their impact on poverty
Most western
countries face a
severe decline in
the ratio of
workers to
retirees….
….that gets
worse the
further out
projections
are made
And many countries have experienced falls in
labor force participation for older men….
Male Labor Force Participation Rates
age 60 to 64 c. 1980 and 1999 (approx.)
100
90
1980
86.4
1999
80
70
68.8
65.9
57.4
60
50
46.6
55.5
52.2
60.4
54.8
45.7
44.2
39.1
40
30.3
30
31.7
29.1
16.4
20
10
0
Canada
France
Germany
Italy
New
Zealand
Sweden
United
Kingdom
United
States
Source: Social Security Administration, An Aging World, 2001. The initial year is 1981 for Canada and New
Zealand, 1982 for France. The final year is 1996 for France.
…while levels for older women are lower and
show more uneven trends
Female Labor Force Participation Rates
age 60 to 64 c. 1980 and 1999 (approx.)
100
90
1980
1999
80
70
60
46.5
41.4
50
40
30
38.8
34
32.5
28.3
26
20
27.8
24.7
22.3
15.2
13 12.7
8 8.1
10
11.7
0
Canada
France
Germany
Italy
New
Zealand
Sweden
United
Kingdom
United
States
Source: Social Security Administration, An Aging World, 2001. The initial year is 1981 for Canada and New
Zealand, 1982 for France. The final year is 1996 for France.
A variety of incremental policy responses have
been tried to address pension funding issues:
• Refinancing
– Increase payroll tax base and rates
– Add dedicated revenue sources or increase
general revenue subsidies
• Retrenchment
– Change indexation formulas
– Punish early retirement
– Increase retirement ages
But what about more fundamental
reforms―shifts in pension regimes?
The Questions on Pension Regime
Change:
• How frequent is a major change in
pension policy regimes?
• How much do past choices constrain
the range of future pension regime
options? Do policymakers have
significant discretion in shifting
pension policy paths?
• What are the forces that determine
whether pension regime changes
occur?
The Convention Wisdom on
Pension Regime Change
Categorizing Pension Regimes:
• Welfare states can be divided into three
categories
– Universal/citizenship regimes
(Scandinavia)
– Social insurance “Bismarckian” regimes
(continental Europe)
– Residual regimes (U.K., Canada, United
States, Australia)
Esping-Andersen, The Three Worlds of Welfare Capitalism
The Frequency of Pension
Regime Restructuring:
• Welfare states have survived
recent economic and demographic
pressures relatively intact
• Pension reform has been largely
incremental rather than
fundamental “regime change”
Myles and Pierson, The New Politics of the Welfare State
Explaining Patterns of Pension
Restructuring:
• “Positive policy feedbacks” limit the
pension reform options of
policymakers:
– Constrain choice sets
– Create constituencies who resist any
change that would make them worse off
• Age and maturity of pension regime
matter (e.g., “double payment
problem”)
A Revised Approach
Categorizing Pension Regimes:
• Esping-Andersen’s tripartite categories
are overly broad and misleading, e.g.:
– Residual category is overly broad mixture of
• means-tested
• “Bismarckian Lite”
• mixed regimes
with distinctive challenges and regime
transition opportunities
– New “Notional Defined Contribution” (NDC)
pension has different challenges and
transition opportunities from
continental/Bismarckian regimes
Recategorizing Pension Regimes:
– Universal/citizenship regimes (New
Zealand)
– Social insurance “Bismarckian” regimes
(continental Europe)
– “Bismarckian Lite” regimes (U.S.,Canada)
– NDC regimes (Sweden, Italy)
– Residual regimes (formerly Australia)
– Mixed regimes (U.K., Netherlands,
Switzerland, Denmark)
– Privatized regimes (none among rich
countries)
Pension Regime Transitions
1950
1974
1985
1995
NDC
Bismarckian
Bismarckian Lite
Universal
Mixed
Residual
Asterisk indicates that a country has added a small mandatory or
quasi-mandatory defined contribution individual account tier.
2009
Sweden *
Italy *
Germany *
Austria
France
U.S.
Canada
N.Z.*
Ireland
Denmark
Switz.
U.K.
Neth.
Australia
Available regime transition options
depend on your starting point…
Residual (Means-Tested) Pension Regime
Transitions
1950
1974
1985
1995
2009
NDC
Bismarckian
Bismarckian Lite
Canada
Universal
Denmark
Mixed
Australia
Residual
Universal Pension Regime Transitions:
Early exits and multiple destinations
1950
1974
1985
1995
2009
Sweden *
NDC
Bismarckian
Bismarckian Lite
Universal
Mixed
Residual
Asterisk indicates that a country has added a small mandatory or
quasi-mandatory defined contribution individual account tier.
Canada
N.Z.*
Ireland
Denmark
Switz.
U.K.
Bismarckian Pension Regime Transitions:
Very late exits and only 1 destination (plus*)
1950
1974
1985
1995
NDC
Bismarckian
Bismarckian Lite
Universal
Mixed
Residual
Asterisk indicates that a country has added a small mandatory or
quasi-mandatory defined contribution individual account tier.
2009
Sweden *
Italy *
Germany *
Austria
France
“Bismarckian Lite” Pension Regime
Transitions: High durability
1950
1974
1985
1995
2009
NDC
Bismarckian
Bismarckian Lite
Universal
Mixed
Residual
U.S.
Canada
Mixed Pension Regime Transitions:
Multiple Precursors, No Exits, and **
1950
1974
1985
1995
NDC
2009
Sweden *
Italy *
Germany *
Bismarckian
Bismarckian Lite
Universal
Mixed
Residual
Asterisk indicates that a country has added a small mandatory or
quasi-mandatory defined contribution individual account tier.
N.Z.*
Denmark
Switz.
U.K.
Neth.
Australia
The Frequency of Pension Regime
Restructuring:
• Pension regime change is fairly frequent
– 9 of 14 countries in sample have at least one
– Only four (Sweden, Canada, Denmark and New
Zealand) have more than one
– Regime reversals (“Boomerangs”) are very rare
• Bismarckian systems are beginning to
“thaw” with shifts to NDC regimes and
addition of small DC tiers
• The U.S. is an outlier in having virtually no
incremental or regime changes since 1983
Pension Regimes differ
substantially in their durability:
• “Bismarckian Lite” and mixed
regimes are highly durable (cul-desac) in post WW II period
• Universal and residual regimes
virtually disappeared after World War
II, with multiple destinations (paths
and forks)
• Bismarckian regimes were very
durable until mid-1990s when shift to
NDC began (precursor to “Chutes and
Ladders?”)
Pension Regime Restructuring—
Timing:
• Different types of regime transitions are
concentrated in different periods:
– Shifts to Bismarckian regimes stop pre1973
– Shifts to mixed regimes concentrated
post-1973
– Shifts from Bismarckian to NDC regimes
post-1994 (after incremental reform
options exhausted)
– Individual account account “add-ons” to
dominant DB regimes grow beginning in
1990s
Argument: Prospects for pension
regime change depend on:
• The balance of positive and negative
feedbacks
• The availability and efficacy of
incremental reforms to address those
challenges
• The availability and political
acceptability of regime transition
options
which vary across specific pension
regimes
Challenge and Change in
Bismarckian Pension
Regimes
Challenges for Bismarckian social
insurance systems are severe:
• Severe sustainability issues with aging
• Need to address problems of low labor
market participation in 55-64 age group
Male Labor Force Participation Rates
age 60 to 64 c. 1980 and 1999 (approx.)
100
1980
90
1999
80
70
65.9
60.4
60
54.8
50
40
44.2
39.1
30.3
30
20
55.5
29.1
32.3
31.7
18.6
16.4
10
0
France
Germany
Italy
US
Sweden
Belgium
Incremental reform options for Bismarckian
social insurance systems are limited:
• Payroll taxes perceived to hurt competitiveness
• Less visible incremental benefit cuts mostly
exhausted, leaving only painful options (e.g.
retirement ageSocial
increase)
Security Contributions as % of GDP in 2000
18
16.4
16.1
14.8
16
15
14.1
11.9
14
12.4
12
%
10
8
6.9
6.1
5.1
6
4
2
S
U
K
U
en
ai
n
Sw
ed
et
h
N
Sp
s
er
la
nd
ly
Ita
y
m
an
G
er
ce
Fr
an
da
an
a
C
Be
lg
i
um
0
Transition Opportunities for Bismarckian
regimes are highly constrained:
• Shift to mixed regime regimes unlikely
due to double payment problem (only
small DC “add-on” possible)
• Can’t shift to universal, residual, or
Bismarckian Lite regimes because of
adequacy concerns
• NDC regime is only remaining regime
transition option (“single chute”)– and
it is a recent innovation
Sweden in the 1990s—Policy
Feedbacks in a Bismarckian System
• Universal pension
• Earnings-related pension on top
• Generous income-tested pension
removes almost all seniors from
poverty
Sweden-Strong Negative Feedback Effects
• Very serious fiscal challenge in both short run
and long run as population ages
• Very high payroll taxes and overall tax burden
spark competition concerns
Exhausted Incremental reform options
• Strong resistance to payroll tax increases
• Strong union resistance to visible benefit cuts
Limited Regime Transition Opportunities:
• Shift to a Mixed System very difficult given high
current commitments and payroll tax
• Shift to NDC system compatible with existing
earnings-related system
Sweden Today—An NDC System with an
Individual Account Add-On:
• NDC tier financed by 16%
contribution rate-- risk of poor
economic performance and
increased longevity shifted from
state to workers
• Individual account tier financed by
2.5% contribution rate
Challenge and Change in
“Bismarckian Lite” Pension
Regimes
Challenges for “Bismarckian Lite”
social insurance systems include:
• Developing adequate mechanisms to
deal with senior poverty
• Adapting to changes in
supplementary occupational and
personal pension sectors
• Addressing long-term pension
funding problems in the absence of
an immediate funding crisis
“Bismarckian Lite” pension regimes contain room
for refinancing without restructuring
Social Security Contributions as % of GDP in 2000
18
16.1
16.4
16
14.8
15
14.1
11.9
14
12.4
12
%
10
8
6.9
6.1
5.1
6
4
2
S
U
K
U
en
ai
n
Sw
ed
et
h
N
Sp
s
er
la
nd
ly
Ita
y
m
an
G
er
ce
Fr
an
da
an
a
C
Be
lg
i
um
0
“Bismarckian Lite” pension regimes have
multiple transition opportunities:
• Can shift to Bismarckian regime only
before demographic crisis hits
• Can shift to mixed regime as “addon” with higher contributions
• Can shift to NDC regime
But also have less need to shift
because of availability of
incremental reform options
United States:
Negative Feedbacks
• Moderate fiscal challenge in short run and
relatively modest in longer run
• Adequacy: High senior poverty rates
especially very old single women
Incremental reform options
• Benefit cuts (e.g., retirement age
increases and increased income-testing
at upper end)
• payroll tax increases off the table because
of Republican opposition
United States:
Regime Transition Opportunities
• Shift to mixed system inhibited by
financing constraints unless new
revenues added
• NDC system possible but inhibited
by internal cross-subsidies unless
new revenues added
United States Today—
“Bismarckian Lite” Stability :
Social Security in the U.S.:
• Incremental reform in 1977 and 1983
including benefit cuts and increases
in standard retirement age
• Virtually no policy change since then
• Efforts by Bush II to get opt-out
individual account reform on the
agenda failed
Concluding Thoughts
1. How much pension regime change?:
• How much regime change you see
depends on how you categorize
regimes
• Tri-partite conceptualization of pension
regimes is inadequate
• Amount of pension regime change over
last fifty years has been substantial in
OECD countries
2.
Amount and direction of pension
policy regime change depends in large
part on:
• Balance between positive and negative
policy feedbacks
– Negative policy feedbacks can be strongly
transition-encouraging (e.g., affordability of
Bismarckian regimes)
• Incremental reform options available to
policymakers—and whether they have been
exhausted
• Regime transition opportunities available to
policymakers
– Only available after invented (e.g., NDC)
– Adoption depends in part on political resources
3. Prospects for specific pension regime
transitions vary over time
(Most common timing of pension regime transitions is shown next time to the
corresponding arrow. Pension regimes with a low probability of regime exit in
the current “late” period of public pension development are shown with a
shaded background)
Residual
regime
early
No
public
pensions
early
middle
or late
middle
Universal
regime
middle
Mixed
regime
middle
or late
middle
middle
early
“Bismarckian
Lite”regime
middle
Bismarckian
regime
late
NDC
regime
4.
Negative feedback effects are a
necessary but not sufficient
explanation of pension regime
transitions, e.g., why:
• Some Bismarckian systems (e.g.,
Sweden, Italy) shift to NDC while
others (e.g., France) do not
• New Zealand and Ireland remain
outliers as residual systems
• Timing differs in regime transitions
5.
A broader set of explanations of
pension policy regime change
includes:
Economicdemographic
variables (e.g., aging,
competitiveness,
fiscal crisis
Agenda-setting
Policy feedbacks
(e.g., policy
regimes, program
maturity, “micro”
rules)
Political
Environments and
strategies
(e.g.,macroinstitutions,
coalitions,
corporatism
The Problem stream
The Policy stream
The Politics stream
Policy Adoption
Policy Implementation
Policy survival
But examining negative feedbacks is a good place to start!
6.
A framework focused on negative feedbacks
and reform options has wide applicability—
e.g., where Marx and Engels went wrong:
Balance of positive and negative feedback effects
• Anticipated near-universal immiseration did not occur
Incremental Reform Options
• Many incremental patches available (e.g., welfare state
programs, better regulation of business cycles)
• Patches became more politically feasible as voting
franchise was expanded
Regime Transition Options
• High costs of transition
• Major opposition from entrenched interests
• Socialist and communist alternatives proved to have
substantial negative impacts in practice
• Redistributive welfare state (with great variations) proved
“easier regime transition option
The End– Finally!