:
LIBRARIES)
^
©
I.I.T.
LIOKAK&E©
-
UCVVCi
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium
Member
Libraries
http://www.archive.org/details/insulationofpensOOdiam
Oewe^
working paper
department
of economics
INSULATION OF PENSIONS FROM POLITICAL RISK
Peter Diamond
94-20
June 1994
massachusetts
institute of
technology
50 memorial drive
Cambridge, mass. 02139
INSULATION OF PENSIONS FROM POLITICAL RISK
Peter Diamond
94-20
June 1994
MASSACHUSETTS INSTITUTE
OF TECHNOLOGY
OCT 2 1 1994
LIBRARIES
Insulation of Pensions
Peter
from
Political
Risk
Diamond
June 1994
Abstract
There are many sources of
political risk to public provision
analyzes legislation to alter the retirement income system.
that
some changes
risks.
in the
is in
This paper
This approach naturally recognizes
system are good responses to social
Thus the discussion
of pensions.
risk,
terms of the legislative actions.
while others generate such
Particular attention
is
paid to
the roles of automatic pension adjustment and pension professionals in providing insulation.
Briefly touched
upon
system being created.
is
the tendency of legislation to redistribute as a function of the type of
A.. Introduction
There are many sources of
of excessive benefits
political risk to the public provision
people retiring when the system
to
combined with promises for future
of pensions. One
the granting
not mature, particularly
is
A
can not be met.
retirees that
is
second
is
the
when
making of
promises to future retirees that can not be met, perhaps because assets that could have financed
such benefits have been used for other purposes.
to the short-run condition
against political risks
I
mean
that
we have
third is excessive responsiveness
of the government budget.
benefits to the long-run condition of the
The Chilean
A
A
fourth
of benefits
excessive responsiveness of
is
government budget. 2
privatized mandatory pension system appears to provide
(Diamond and Valdes-Prieto (1993), Diamond
(1994)).
good
insulation
By good insulation,
not seen a number of actions that might have occurred.
Under
the
new
system, there has been no large infusion of government funds to preserve the expectations of
Since the interest rates earned on the privatized funds have been large and there has
workers.
been no massive fraud,
structure.
budget.
this
may
There has been no adjustment of pension benefits with the
In 1985
when
frozen, but nothing
the
government budget was
was changed on
being paid were very small; in part
didn't
reflect these circumstances as well
the
it
new system.
was,
in trouble, the
In part this
I think, politically
state
COLA
may
as the social security
of the government
on the old system was
reflect the fact that benefits
insulated by the fact that the
money
flow through the government. 3 There was no use made of the mandatory savings plan
-2-
to
do redistribution, even though there were
presumption
Also
I
see
Perhaps,
is
that the
no reason
need
to identify a source
to anticipate a
if interest rates
major
from the
crisis as
crisis;
However, with a single data point so
social security
of financing affected
particular
My
the political outcome.
Chile goes through the aging transition.
go down, the mandatory savings
adjustment that should happen without a
crisis.
improve pensions for coal miners.
calls to
rate will
have to change, but
that is an
not without a political struggle, but without a
far, it is
economic and
hard to separate the role of privatization of
political climate in Chile, and,
of course, the
particular political structure at the time of the Pinochet government.
This paper explores the meaning of excessive in the opening paragraph and a description
of changes
in retirement
income systems
in
terms of
risks.
Chilean approach that also offer insulation against some of these
It
considers alternatives to the
risks,
permitting a consideration
of possible tradeoffs of other properties of retirement income systems with the degree of
insulation.
The approach taken
is
the analysis of legislation to alter the retirement income system.
This approach naturally recognizes that some changes in the system are good responses to social
risks,
while others generate such
institutional structure
risks.
Thus the discussion
on the likelihood of
is in
terms of the effect of
alternative legislative actions.
Particular attention
is
paid to the roles of automatic pension adjustment and pension professionals in providing
insulation.
relationship
The focus on
legislation after the creation
between the approach taken
pattern of outcomes.
This consideration
to initiation
is
briefly
-3-
of the system omits consideration of the
of the retirement income system and the
touched upon
in
terms of the tendency of
a function of the type of system being created.
legislation to redistribute as
To frame
the economy.
models
It
the analysis, the paper begins with discussion of the allocation of other risks in
briefly
mentions the income distribution system.
of
to illustrate patterns
constitutional
and
risk bearing
institutional effects
modeled
are
all
risk creation.
on the patterns of
B..
We
and
risk bearing in the asset
is
the social allocation that
all
Individuals are
market
is
Pareto efficient.
generically not Pareto efficient.
comes with these
choices.
These have
If markets are complete, then the allocation of
4
However,
if
markets are not complete, then
Incomplete markets imply that allocations can
56
In considering efficiency, a critical element
and for
legislation.
Other risks
different properties in different market settings.
be improved upon.
followed by discussion of
making) choices across different risk-expected return combinations. More
important for our purposes
is
is
familiar with issues of risk bearing in the asset market.
as facing (and
the allocation
This
There are then three simple
is
whether the allocation process happens once
(with complete markets) or whether there are repeated times of decision making (with
incomplete markets). Repeated decision making by multiple economic agents in the presence of
a changing ability to choose
complete markets).
Of
among
alternatives is the source
of inefficiencies
(in contrast with
course, banning future trades and forcing people to stay with whatever
incomplete market plans they
made
earlier is not a cure for inefficiency.
-4-
In considering political risk, there are
social risk that
must be adapted
One
two elements.
is
the presence of significant
a once-and-for-all allocation will not be efficient.
to;
the combination of limited complexity in legislation and a lack of
Second
commitment devices
government. Taken together, these restrictions are akin to the incomplete market problem
markets.
When
the lack of
commitment devices by government has been approached
of a time-consistency vocabulary,
government and
7
its citizens.
It
has generally been modeled as a
this
seems
to
me
at least as
is
for
in asset
in
terms
game between
a
important to recognize the risk associated
with political competition as the risk associated with lack of commitment by a single government
player.
Moreover,
in the
game
time-inconsistency problem.
other issues.
8
context, reputation effects can play a major role in mitigating this
In any event
These resemble
it
has been
much
political business cycle issues
9,
studied, so
although
I
I
will concentrate
will
view them
on
in the
social insurance context.
The mention of time inconsistency
individual choice.
10
Once
individuals
wants to do welfare analysis.
I
in
government policy reminds us of the same issue
show time
inconsistency,
think that the
same
it is
not generally clear
in
how one
issue arises in the context of repeated
democratic decision making that shows time inconsistency because of shifting political outcomes
and preferences. But
this is
an issue that
C.
When
creating
a
new
I
will not explore.
Redistribution
retirement income system,
intertwined with other design issues.
While
it
is
-5-
distribution
issues
are inevitably
analytically convenient to ignore distribution
issues, they
his
AEA
should not go unmentioned. This concern has been well stated by Robert Solow
in
Presidential Address:
"Most of us are conscious of a conflict
we
because, while
think
achieve distributional goals,
minds and consciences
usually a mistake to fiddle the price system to
is
it
that arises in our
we
realize that the public
and the
political process are
make the direct transfers we would
If we oppose all distorting transfers, we end up opposing transfers
prefer.
Some of us seem to welcome the excuse, but most of us feel
altogether.
uncomfortable. I don't think there is any very good way to resolve that conflict
perversely
more
in practice."
willing to do that than to
(1980, page
1.)
In the realm of redistribution one can ask about the relationship between political insulation of
the mandatory savings part of the pension scheme, and political insulation of the redistribution
coming from the minimum pension.
In Chile, any annuity purchased with mandatory savings
must be indexed for
minimum
inflation, but the
are periodic adjustments.
A
providing to the poor.
pension
is
not automatically indexed.
This does not appear to be a good
more
way
So there
system aimed
to design a
at
integrated pension system might have had a different balance of
insulation of the different parts.
P..
It
is
natural to
generations model.
Nevertheless,
I
12
A
examine
simple example of risk sharing
social provision
1 '
of retirement insurance
in
an overlapping
This permits consideration of some issues not present in
will consider
two period models,
examples, reflecting aggregate income
risk,
for their simplicity.
pure political
-6-
risk,
I
want
finite
models.
to consider three
and inaccuracy of expectations.
Consider a two-period model with two types of agents.
period one, before the government passes any legislation.
two, which
two
random
is
legislation, if any.
There
market plays a central role
is
no production and no storage
in the allocation
B
Type
time of period-one legislation, but
at the
function
u'(c',, c' 2 ),
The
set
receives
before the time of period-
in this
economy. Thus the
way
in
which
in
credit
which each
risks
of second
maximize the expected value of the
to
y, in
receives income y 2 in period
of resources, representing both a way
Each agent seeks
income
known
type obtains consumption in the period without income and the
period income are shared.
A
Type
utility
i=a,b.
of Pareto optimal allocations has both types of agents having the same realized
(ex post) intertemporal marginal rates of substitution.
(We do
individuals of a given type have different consumption levels.)
not consider allocations where
Thus the Pareto optimal
allocations satisfy
1
u',(cV
.
<=',
+
c
c' 2
+
c
b
,
b
2
Debreu theory,
= y2
it
b
b
2
b
)/u
2
(c „ c
b
2 ),
.
is to
at
hand
is
how
to achieve
assume a complete
conditional on the level of
that
c
= y„
The question
income
= u b ,(c\,
c' 2 )/u» 2 (c'„ c' 2 )
in period two.
income
in period
We know that
would be very expensive
set
the
to try to
one of these
of markets.
two
is
allocations.
That
is,
One way, from Arrow-
consumption
a different good for each possible level of
economy does not have
this
complete
have them; and that the attempt to
-7-
in period two,
set
of markets;
list all
states
of
nature would
fail,
since something
would happen
Consider next the allocation that occurs
assume that marginal
2.
lump sum (ex
E{u',(cY
c",
+
c
c'2
+
c
I
b
,
b
2
in
anticipated.
only trading in a safe (real) bond.
ante) redistributions
c' 2 )}/E{u' 2 (c'„ c"2 )}
The
set
between the two types,
= E{u b ,(c b „
c
b
b
2
b
)}/E{u 2(c „ c
satisfy the equations
b
2 )},
= y„
= y2
.
We know these
with complete markets (in
Now,
let
B
2,
i.e.,
everyone knows the true distribution of incomes
(1)).
us add to this system a pay-as-you-go
on type
A
A
consumers.
We
(PAYG)
consumers used
In the second period, the poll tax,
given to type
in taking expectations,
allocations (in (2)) are less efficient than those described above
period, there is a poll tax T,
consumers.
We
of equilibrium allocations,
have assumed that everyone has the same subjective probabilities
in period two.
benefit,
if there is
a position to go bankrupt.
E, and that these expectations are correct;
first
was not
goes to infinity sufficiently quickly with low incomes that individuals
utility
never allow themselves to be
for different
that
T2 on
,
type
social insurance system.
In the
to finance a benefit B, for type
B
consumers
is
B
used to finance a
have simplified the story by having no distortions
associated with the social insurance system.
-8-
Let us consider restrictions on allowable legislation in period one and the possibility of
new
Limited
legislation in period two.
with complexity, or equal treatment
ability to deal
constitutional restrictions, or the determination
of legislated outcomes might require
or benefits or their relationship be the same for each generation.
Assuming
are of equal size reduces these possibilities to a single restriction,
social
that the generations
T,=T 2 =B,=B 2
insurance accomplishes nothing in risk sharing; the allocation
that taxes
is
.
In this case,
simply one of the
allocations in (2), with the given level of redistribution.
One method of introducing
That
is,
we
risk sharing is to
introduce proportional taxes,
t,
the break even condition requires B]=t,y,.
One
period two.
the
PAYG
and
t
2
assume
rather than the poll taxes.
The question
possibility is to legislate the
that taxes are proportional to income.
same
is
what
tax rate,
is
legislated in period
government does not choose
becomes more complicated. Let
t,^, with benefits determined by
the
to legislate such an indexed system, then the story
government choose an income tax
break even in period one, and legislate both values for both periods.
government
will
need
wants to break even.
do
one for
break even condition.
If the
will
In the first period,
in period
to legislate a
change
rate
and benefit
Then,
in benefits or taxes (or both)
in period two, the
almost certainly,
if
it
Provided that people can accurately predict exactly what the government
two and provided
that the
government, ex post, will do the correct thing from
an ex ante point of view, the government can produce an allocation satisfying the
conditions in
level that
(1).
-9-
full efficiency
However, we have introduced two complications
in period
two and what
-
what
will the
will people predict in period one.
A
E..
simple example of risk creation
Let us continue with the same model as above, except that
period income
insurance.
is
determinate rather than random. There
There
also
is
government actually do
no efficiency cost
described here, provided that
first
Such reconsideration introduces
to
now no
that
second
risk-sharing reason for social
having a social insurance system of the kind
period legislation
political risk
is
we now assume
not reconsidered in the second period.
is
from the perspective of individuals.
Moreover,
perceptions of the possible different outcomes affects the functioning of the capital market, even
though there
is
no taxation of
market income.
capital
the determination of political outcomes.
14
And
13
In order to proceed,
we need
a model of
for the purposes of the design of social insurance,
a model of the relationship between political outcomes and the social insurance system structure.
To make
this issue precise,
(Different population sizes
government
in
legislates
we now assume
would be a reasonable
t,^ and B,=B 2
that
y2
not random, but
is
not equal
to y,
alternative to consider.) In the first period the
This works fine
.
is
in the first period, B,=t,y,,
but results
a system with a surplus or deficit in period two, depending on the relative sizes of y and y2
Thus second period
,
legislation is
now
uncertainties that impact
on the
legislation yields a tax rate
and a benefit
break even condition,
B 2 (L) =
t
2
needed.
Let
legislative process.
(L)y 2
.
L be
.
a random variable reflecting the
Thus we assume
level that are functions
that
of L, and that
second period
satisfy the
PAYG
In this setup, the credit market in period one depends on
-10-
expectations about the legislative process.
following
first
8
,
c' 2
=
+
s
=
s(1+r)
(l-t,)y„ c
b
,
+ B 2 (L),
=
c
c' 2 )}
s
in
balance in period two
b
2
is
= E{yi\(c\,
b
b
A
if there is
random
period one.
It is
=
L.
The assumption
that the social insurance
it
may
still
system will not
not a necessary condition for there to be legislation in period two,
It
does seem more likely that there will
how
the dependence of legislated taxes and benefits, depicted as
variable L, varies with the structure of social insurance put in place in
two pieces
a probability of
-
and a relationship between period-two legislation and period-one
advance,
2 )},
(l-t 2 (L))y 2 - s(l+-
natural to divide the question into
legislation is in terms
b
imbalance.
central question is
functions of the
b
c 2 )}/E{u 2 (c „ c
or uncertainty about period-two legislation in period one.
be legislation
and
+ B„
where expectations are taken over
be
s
r.
E{u',(c'„ c'2 )}/E{u' 2 (c'„
c
equality of marginal expected utilities yields the
order condition for the determination of the level of lending and borrowing,
the rate of return,
3.
The
legislation.
new
legislation
For example,
if the
of poll taxes, T, and benefits so that budget balance can be legislated
happen
that these levels are
changed
in the
second period. Having
first
in
and
second period benefits and taxes the same might minimize the probability of legislation to change
their levels.
In addition,
we would
expect second period taxes and benefits to be a (stochastic)
function of both what happened in period one and what
That
is,
the legislated tax, t^L),
is
was voted
in
period one for period two.
plausibly a function oft, and also of what
-11-
t2
would be
if
there
were no
The
legislation.
would
legislation; this
This setup lends
inherited legislated
naturally affect
Below we
previously legislated levels.
There
as
outcomes with bargaining models of
will turn to the further issue
of the relationship of
of the program.
a second dimension to this issue
is
legislated outcomes.
consideration of legislated levels of parameters as functions of
itself to
legislation to the structure
outcomes are a default should there be no
if
one views the
first
period legislative process
no more worthy of respect from the analyst than the second period process.
legislation is
prevents
its
viewed as unfair
in the
second period, would a design of the legislative process that
undoing be a good thing?
F..
In both sections above,
(possibly)
know
If first period
random incomes and
Perceptions
we assumed
(possibly)
these distributions; no one does.
that
random
consumers knew the distributions of both
legislated redistributions.
Thus we need
Consumers
to consider the implications
don't
of possibly
inaccurate forecasts. For example, consider the model of pure political risk (no aggregate income
risk)
where the
social insurance system legislated in the first period is definitely not viable in the
second period. Then, consumers planning their transactions
the political outcomes.
4.
E'{u',(cY
c\
+
s
=
Thus
the equilibrium will satisfy
c' 2 )}/E'{u' 2 (c'„ c'2 )}
y,
-
T„
c
b
,
=
in the capital
s
= E b {u b ,(c b „
+ B„
-12-
b
b
b
b
b
c 2 )}/E {u 2 (c „ c 2 )},
market must forecast
c\ = s(l+r) +
B2
,
c
b
2
= y2
-
s(l+r)
-
T2
.
This allocation differs from the one above in that
have added individual tags
I
operator, reflecting the possibility of different beliefs
by different consumers (although
assumed the same for everyone of the same type) over the
second period.
While
first
G..
It
some
of taxes and benefits
in the
Some
are needed.
Hypothetical Examples
might be useful to consider some circumstances where each of the above models has
relevance.
While
it
would be
hypothetical ones. Consider an
are
levels
still
period benefits and taxes have been legislated, the second period
random variables over which expectations
levels are
to the expectation
growing rapidly and returns
worked and saved
elderly are
at
economy
that is
real
examples,
make do with
I will
undergoing rapid economic growth so that wages
to capital are high, but the current elderly are not well off, having
times of low wages and low rates of return. Possibly a large fraction of the
below the poverty
for elderly to pull
some
best to have
threshold.
many of them
Politically,
out of poverty.
it is
difficult to raise the
But increasing benefits
security system is easy since payroll tax revenues are high.
viewed as an example of the model
aggregate incomes in a
way
in section
that tends to
minimum
in the
PAYG
social
Passing such legislation can be
2 since the level of benefits
move
pension
is
sensitive to
intertemporal marginal rates of substitution
toward each other.
As an example of political
risk,
consider a government that has tax revenues that are more
-13-
than proportionally sensitive to the business cycle (or are particularly sensitive to other variables).
Assume that the government responds to
retirement pensions.
fiscal difficulties
Then, viewing the
state
efficiency of risk bearing.
is
all
expenditures, including
of the budget as a random variable, the variation
of government provided retirement pensions with the
of the variation in other incomes
by adjusting
state
of the government budget
in excess
an example of government created risk that worsens the
This example differs from simply assuming that two political parties
repeatedly change in controlling the government, with one party expanding retirement benefits
and the second party reducing them since
additional
methods of
it
identifies a source
insulation, rather than just a
In both of these examples,
I
of variation
need for insulation.
have identified the pattern of consumption or income per
capita as the basis for judging risk reducing and risk increasing changes.
economic theory of
risk sharing
and
that suggests
is
the basis for the proposal by Robert
This reflects the
Merton
that social
security benefits be indexed to consumption per capita (Merton, 1983).
In sum,
we would
like to identify
good reasons
for varying benefits and taxes and
bad
reasons for varying them, and then to choose a structure for provision of retirement income that
encourages the former and discourages the
latter.
encouragement and discouragement are probably the
in politics.
-14-
It
is
probably worth remembering that
right vocabularies
-
there are no guarantees
H.. Legislated
We
have identified a pair of needs
circumstances and a need not to modify
it
-
Outcomes
the need to modify social security in
in others.
Moreover, there
that affect the likelihood of
of such mechanisms on changes
in
change
in the system.
both circumstances.
We want to
And we
to ask
about
ask about the effect
probably want to distinguish
between circumstances of increases and circumstances of decreases
All of the simple
a need for accurate
Thus we want
perceptions about the likelihood of both types of circumstances.
mechanisms
is
some
in both taxes
models above incorporate some legislated outcomes.
and
At
benefits.
this stage
of
our knowledge of the determination of legislated outcomes, this process must remain something
of a black box, with the argument relying on intuitions of the effect on
derivations from a
model of
the legislative process.
One
legislation.
source
is
There are three different types of sources of effects on
elements of the constitution that affect
process works in the consideration of social security; second
benefit and cost determination process
institutions
on
legislative behavior.
culture that limits
on
to do,
the legislative
the role of the social security
even when they could
of other factors and
(at least temporarily)
are distinguishing between changing the rules covering legislation
is,
on the one hand from
institutions that affect political context.
reflect
how
This can be considered to be the development of a political
do other things. That
Of course
is
likely legislation; third is the role
what governments choose
we
legislation, rather than
these issues are not unique to social security.
on the recent newspaper
articles
For example,
on the opening up of the Japanese
-15-
rice
it is
interesting to
market
to imports
and
part,
its
on incomes of
effect
from the
rise
rice farmers.
of countervailing
lack of an independent claim for
Japanese
rice).
political
protection system
was
politically vulnerable, in
power (exporters concerned about GATT) and
income protection (other than the
As economists,we may
interesting to consider
The
historic cultural continuity
not like the income protection to begin with, but
what would have made
it
more
secure.
the
Do
tariffs
of
it is still
have more (or
less)
political viability than quotas?
Constitutional Rules
I
will refer to the rules
These rules
affect
governing the functioning of the legislature as a constitution.
how a government that
the constitution might be
continues to live by the constitution might behave,
amended within
the constitutional process and
how
how
(and whether) the
constitution might be replaced by an extra-constitutional process.
There are many actions that governments might
government.
Any
take.
Any
asset
might be taken by
expected receipt might be blocked by government. Conversely an asset might
be given by government, an expectation that might otherwise be unrealized might be
government.
These are some of the second period legislated
begin by considering
Some
common
constitutional restrictions
acts that
fulfilled
might be taken.
by
Let us
on asset seizure generally.
constitutions protect private property and private contractual arrangements by a
restriction that
government must compensate
definition of a taking
is
if
it
"takes" private property.
Of
course the
unclear so the implications on behavior of such a rule depend on ongoing
-16-
political
and judicial processes.
government contract
Similarly, a
manipulated.
same damage-payment
In addition the definition
to
of "private property" could be
purchase an item
rules as a private contract. Thus,
one approach
is
normally subject to the
to constitutional protection
for social security is to give the social security system's promises the
same
which
contracts or private property rather than the status of legislated behavior,
future legislated changes.
the promised level.
benefits. This
to
status
One would hope
side.
13
the legislative process
would make more
a
more
little
There are two sides
may
difficult the reduction
that this difficulty in reduction
may make predictability
be on the low
that
Such
to
difficult since legislated
such a change.
have occurred or a change
in the desired
outcome
subject to
of benefits below
act to limit
promised
promises would tend
Generating such sluggishness in
However, there
(in
someone's eyes) while enhancing the
is
an asymmetry here between benefit
increases and benefit decreases.
Thus, desired outcomes
a time of temporary political
power and might be hard
An
is
reduce the ability to respond to either the particular risky outcomes
protection from pure political risk.
at
would
status as private
(in
someone's eyes) can be legislated
reverse.
alternative approach is to put the rules covering social security directly into the
constitution,
making changes more
difficult since
amending a
constitution
is
generally a process
requiring greater majorities or additional steps than with ordinary legislation.
Social Security Design
The
full privatization
and
full
funding of a defined-contribution plan, as in the Chilean
model, seems to provide the most insulation. This suggests that
this insulation
off against other elements, such as the ability to provide insurance and cost.
-17-
should be traded
However,
this
ignores the issue that insulation without limit
is
not uniformly good.
It
seems useful
to
distinguish different degrees of resistance to change along a path connecting a traditional
unfunded defined benefit system with the Chilean model.
The minimal design element
for public provision of retirement
no specific identification of a revenue source.
plan, with
income would be a
Such a system would be very
vulnerable to the state of the government budget. Traditionally, a payroll tax
for the
promised
benefits.
of the budget, although
Earmarking
it
earmarked
to
pay
This earmarking offers some insulation from fluctuations in the
rest
is
increases sensitivity to fluctuations in the earmarked source of revenue.
will generally require periodic legislation to align taxes
automatic adjustment.
benefit
Automatic adjustment rules
in a totally
and benefits unless there
is
unfunded system may provide
considerable insulation to benefit increases in that any benefit increase requires a simultaneous
tax increase, although an immature system will have
low tax
such changes. Such an automatic-adjustment mechanism
visible than in a
nonautomatic system.
rates
may make
and offer
it
resistance to
the future rise in taxes
The presence of earmarking
the revenue will be used for other purposes;
little
more
affects the likelihood that
also affects the likelihood that other sources of
revenue will be used for this expenditure and that the level of earmarked funds will be increased.
Thus the absence of a fund may increase
insulation in a system
where benefits and taxes are
adjusted to each other automatically.
It
is
common
for traditional systems to partially fund benefits so that any excess of
earmarked revenue over benefits goes
insulation
from vagaries
in the
into a fund.
The presence of a fund does provide some
government budget, but
-18-
it
probably increases the likelihood of
on the basis of fund
benefit increases (possibly unsustainable)
for other purposes, a
more
difficult trick if there is
Funding can be done with public
in the
trust
debt.
That
availability
and also of fund use
no fund.
this is primarily
a political act can be seen
following quotation from Bodie and Merton (1993). In commenting on the social security
fund buildup
in the
U.
S.,
they write:
"In a private plan, funding is used to insure against default by the plan sponsor.
Here the promise to pay benefits has the same level of full faith and credit of the
government as the bonds used to fund the plan. Yet there seems to be a belief
that this change may help to ensure that, when they reach retirement, workers will
indeed receive benefits approximating those promised under the current benefit
formula (i.e., the one in effect in 1992)."
There
is
a role for such a promise
in the effect
on the legislative process.
probably different degrees of likelihood in the following actions.
just implicit social security debt,
system,
it
might repudiate
security system,
it
all
might seize
it
government
all assets.
the pooling of groups of assets or
legislative process.
be taxed, but not
or wealth.
at
That
is,
might repudiate
debt,
The
it
by the
social security
might seize private assets held by the social
idea behind the reduction in likelihood
there tends to be a
-
government might repudiate
explicit public debt held
income claims so
a discriminatory rate
A
There are
that they tend to
norm of
comes from
be treated the same by the
uniformity, assets or income flows can
a rate different from other (similar) sources of income
Insofar as groups are politically accepted as equal, they tend to get treated equally.
This does not underestimate the ability of the political system to identify (or create) differences
in
order to have targeted benefits.
By
placing retirement income in the
same equal treatment
category with other forms of income, particularly privately provided retirement income or income
-19-
from
assets generally, there
in that
it is
to cite
it
to
may be more
political protection.
Again, there
easier to invoke equal treatment to guard against losses than for the general public
guard against increased benefits. This
similar to the possibility that the idea of a
is
comprehensive tax base affects the extent to which tax loopholes get
The next
step
beyond funding with government debt
probably increases the insulation from political
assets
may seem more
From
in that the
may be asymmetry
"real" to voters,
this perspective,
risk,
is
although
more of an obligation
legislated.
funding with private
it is
a
little
to continue
unclear
assets.
why
promised
private
benefits.
promised benefits are automatically adjusted to the
availability
of funds. That
is,
no
This can be
done with earmarking without funding. But earmarking without automatic adjustment
is
not the
as earmarking with automatic adjustment; just as funding a defined benefit system
different
16
a defined contribution system differs from a defined benefit system
further legislative actions are required in order to carry out the promised benefits.
same
-
This
is
from funding a defined contribution system.
In addition to the issue of automatic adjustment, there is an issue of the visibility of the
automatic adjustment.
In defined contribution systems, benefits are automatically adjusted to
measures of the performance of the economy. This adjustment
at the
time of setting benefits, rather than changing their
by the "market."
I
think that a legislature
would have
It
seems
to
me
not very visible since
It is
even
difficulty trying to
between annuities and accumulations on a year by year
be out of sight of the public.
level.
is
basis.
less visible if
done
it is
set
change the relationship
This adjustment seems to need to
that visibility in the public arena is a
-20-
it is
two-edged
Visibility prevents
sword.
light
of day then
of increased
Thus
But higher
result
visibility will also
it is
who pays
a
little
for a
harder to steal in the
new
benefit
is
a form
tend to limit good professional reactions.
useful to try to build a respected professional expertise,
political pressures, that
happen.
political action:
Forcing identification of
at night.
visibility.
may be
it
it is
some kinds of bad
somewhat
insulated from
can try to guide the system into the kinds of changes that ought to
Thus the Chilean system seems
to provide insulation
of individual accounts and by having decreased
by having added
visibility
visibility as a
from the absence of a
stated
benefit formula.
It is
there
is
natural to think of an annuity market as adjusting benefits to available resources, but
no need for the resources
to various
to
be physically present. That
the benefits could be adjusted
accounting measures that have no backing in physical assets. This
Boskin, Kotlikoff and Shoven (1988) propose.
additional (real) benefits (on retirement) based
rates are set
By doing
on
that years taxes paid.
of these calculations from
even
less
professionalism, but
visible
it
it is
political forces.
than
is
a single
essentially
much
However
the accrual
cycle affecting current
easier to preserve the professional character
Having many small decisions should make
decision.
can also be insulation for
what
forecasted to be in balance over
these adjustments on an annual
contributors and not current beneficiaries,
is
In their system, each year people accrue
by actuaries based on forecasts so that the system
the projection period.
process
is,
Lack of
visibility
is
insulation
this
for
theft.
Contrast this approach with one that adjusts annual benefits to annual revenues by
proportionally changing
all
benefits to be paid that year.
-21-
While benefit
recipients might be
passive in the face of such announcements,
it
and so pressure on the system from
politicians
is
more
likely that there will
politicians to protect benefits.
be pressure on
There will also be
pressure to enlarge entitlements on the ground that actual benefits will be cut in the future.
Taking away from people what they do not realize they have
is
much
easier.
Similarly, the
public seems to think largely in nominal terms, so that reductions in cost-of-living adjustments
are politically easier than adjustments in benefit levels,
Of
economists.
course, this
viability, since the
is
however curious
the sort of political evaluation that
this distinction is to
may
not have long run
degree of understanding of the nominal-real distinction and
its
political
acceptance by the public could change.
More
generally, the degree of
money
history, particularly exposure to inflation.
illusion (inflation illusion) in the public
Thus,
in
systems, protecting retirement incomes from inflation
incomes.
The obvious
indexed system
may be no
Political
is
institutional response is to
somewhat vulnerable
some inadequately indexed
is
have an indexed system.
of high
social security
an important part of protecting retirement
to sufficiently high inflation rates
direct cure for this, with avoidance
depends on
inflation the
However, even an
because of
lags.
There
obvious solution.
Context
In addition to constitutional changes and different choice of parameters in the insurance
institution, there are other
ways
to affect the distribution
of legislated outcomes.
One
is
by
adding an institution that has some of the powers of the legislative body (but subject to override
-22-
by the
banks.
legislature), for
Such an
example an
institution is then
institution with
some of the workings of independent
charged with setting some parameters of social security and
staffed with independent professionals.
1718
In addition to having direct control of
some parameters, an
that is required to report annually and/or to
official actuarial office
comment on any proposed
legislation gives higher
(and presumably resistance) to changes that
the Office of the Actuary in Social Security
75 years.
While
prominence
it is
to these
must forecast the
Another approach
in the
to a benefit
unsustainable.
In the
US
financial balance of the system for
little
more
political
long run issues.
comment on proposed
system
may be
easy to ignore these pronouncements, this does give a
With privatized management of a separate
to
institution with professional
Having an
independence can affect the dialogue about social security.
political visibility
central
trust fund, there is again a set
of professionals
legislated changes in the private fund managers.
to affecting political
outcomes
is
by changing the description of the
eyes of the public. For example, the vocabulary of individual accounts (as opposed
formula) has suggestions to the public that
to social security as a tax-transfer
vulnerable than referring to
it
mechanism
may
affect legislated outcomes.
in ordinary discourse
may make
Referring
the system
more
one where benefits have been earned by
as an insurance system,
paying earlier premiums. Such a conversational change results in a primarily one-sided protection
-
harder to cut benefits, but not harder to raise them.
in unsustainable benefit levels, then this
If the greater risk is increases that result
does not offer
-23-
much
protection.
An
balance.
-
where vocabulary matters
additional area
One example of a
whether
this includes or
subtler
form of pressure
deficit.
the public definition of the budget balance
One
such a vocabulary choice.
role for earmarking
An
To
get
some
insulation
from the
social security
This can be done in a variety of ways with
One can simply
differing degrees of insulation.
government budget
government budget, one can move the
and expenditures outside the budget.
of the budget
the vocabulary of
excludes the social security system.
short run fluctuations in the state of the
receipts
is
is
not include social security in official definitions
and a
trust
fund
is to
add
to the sustainability
independent institution managing the fund
may
of
also add to this
sustainability.
The
political viability
If people
creation.
politically
more
Working
of
this paper.
19
do pay for the assets that go into
in
an exchange economy has kept
Capital accumulation
concern with insulation.
A
is
my
that period
Perceived fairness has some
analysis focused on risk, the central charge
a major issue in the behavior of governments.
comes from a
desire to see
One can
more
Some of
capital accumulation, not
consider the extent to which one wants
accumulation than occurs in equilibrium
how
may be
from
similar analysis might be done of the capital accumulation question.
Consider a three period model.
consider
its
clearly perceived unfairness.
the academic support for funding
capital
their accounts, these accounts
insulated than if they are given such accounts.
So too does
political effects.
of an arrangement can also depend on the circumstances of
in
each of the
first
two
more or
less
One
can
periods.
the level of capital accumulation that occurs depends on institution design in both
and
earlier periods.
The government has many
-24-
tools to affect capital accumulation,
of which funding retirement pensions
Thus one needs
is
just one.
to consider the relative merits
of funding different
institution designs that affect capital accumulation.
idea that national defense
is
activities, as
For example one might
well as other
try to inculcate the
shaky unless future defense expenditures are funded. The politics
of such an institution would probably work differently from one focused on retirement income.
Conclusion
I..
Right now, the two most visible risks to retirement income (other than through risks to
the entire
economy and
budget and the
formula.
risk
society
20
)
of excessive redistribution to early generations, masked
The Chilean
government
are the risks of oversensitivity to the state of the
structure does well with both of these
incomes outside the government budget and by seeming
in
an unsustainable
problems by moving retirement
to require the identification
source of financing for benefits for any group of individuals.
It is
of an explicit
probably too soon to rule out
the ability of the political system to tap these resources for government revenue needs or for
redistributions.
market rates
is
Requiring investment
government
the obvious route for the former.
to "needier" accounts
against such
in
(which
moves probably
may have
Redistributing part of the return on portfolios
an age component)
lies in the identification
them) as private property, entitled
to the
the public conception of the retirement
debt, particularly debt paying less than
is
one route for the
as other assets.
If this
income system plays an important
-25-
Protection
of individual accounts (and the returns on
same protection
actual institutional structure.
later.
is
correct, then
role along with the
These considerations suggest a value
visibility to professional evaluation
in
it.
in the creation
of an
institution giving status
and
of the condition of the retirement system and possible changes
Since some systems are easier to communicate to the public, such systems are more easily
protected by such an institution.
Ongoing defined
benefit systems are probably in greater need for protection than are defined
contribution systems.
may
help provide
It
Automatic adjustment rules (or adjustment by independent professionals)
more
protection for such systems.
would be good
to
have a more factual basis on the behavior of legislatures toward
retirement systems in different settings in order to identify the set of risks
be
less speculative
more
carefully and to
about the legislative process in thinking about this important topic.
-26-
A ppendix. A
I
Funded System
started with a pure
exchange economy to highlight the issue of
creating behavior of legislatures in a setting
happening
economy. The use of a
in the
where there was
fully
To make
of savers
and
legislative action
want
to consider a
bring out
some
many
possibilities
impacts on the likely pattern of
two-period economy with two groups
not interact with each other in the private market.
considered the present generation and future generations.
generation ignores
risk
inaction.
this point formally, I
who do
it
and
else (besides redistribution)
funded system does not change the
of legislative actions to create risks or share them, although
outcomes of
little
risk sharing
Lumping
all
The two groups can be
of the future into a single
issues of the carryover of institution design into the future, but does
issues in a simple setting.
Consider a two period model with two types of agents.
period one, before the government passes any legislation.
one, but this income
is
random
of period-two legislation,
if
at the
any.
Type
Type B
There
is
its
known
b
in period
before the time
a stochastic linear technology for converting
R
b
savings decisions based on
-27-
first
be the quantities of period-two income
received for each unit of period-one income saved by the two types.
makes
receives income y" in
receives income y
time of period-one legislation, but
period income into second period income. Let R" and
trade with each other, so each
A
its
The two groups can not
(stochastic) options.
Thus
the credit market plays
no
separately by each group
the
way
in
which
risks
is
role in the allocation
the
way
to
consumption
maximize the expected value of the
subject to the budget constraint
we assume
that each type obtains
that the
c' 2 =(y'-c',)R',
i=a,b.
government can not
Using
realized rate of return.
transfers, but
does not
constraint of each group
the
second period and
is
this
utility
function u^c
1
,,
c' 2 ),
i=a,b
In keeping with the interpretation of the
alter the levels
by setting up a rule transferring between the two groups
random
in the
of second-period income are determined.
Each agent seeks
model,
of risks between the groups. Direct investment
of investment, but can
in period
two
risk share
as a function of the
formulation has individual savers recognize the presence of
alter the rate
of return on marginal investments. Thus the budget
altered to read c'^Cy'-c'JR
1
+ B'(R
a
,R
b
),
i=a,b; with the constraint
on
government B*(Ra ,R b )+B b (R',R b )=0.
b
Purely risk-sharing legislation would have E{B'(R",R )}=0. Redistribution from the future
to the present generation
benefit level
would be
would have
this expectation positive.
risk sharing or risk creating
Legislation with a zero expected
depending on whether
intertemporal marginal rates of substitution towards each other or
pure political risk as discussed above).
-28-
it
tended to
move
away from each other
the
(as with
References
Aghion, Philippe, and Patrick Bolton, 1990, Government domestic debt and the
a political-economic model of the strategic role of debt, in R. Dornbusch and
risk
of
default:
M. Draghi
,eds.,
Public Debt Management: Theory and History. Cambridge: Cambridge University Press.
1992, Picoeconomics: The Strategic Interaction of Successive Motivational
Ainslie, George,
States within the Person. Cambridge:
Alesina, Alberto, 1988,
Bodie, Zvi, Alan
J.
Contribution Plans:
(eds.),
Pensions
Cambridge University
Macroeconomics and
Politics,
Press.
NBER
Macroeconomics Annual
.
Marcus, and Robert C. Merton, 1988, Defined Benefit versus Defined
are the Real Trade-offs?, Chapter 5 in Z. Bodie, J. Shoven and D. Wise
What
in the
U.
S.
Economy. Chicago: University of Chicago
Press.
Bodie, Zvi, and Robert C. Merton, 1993, Pension Benefit Guarantees in the United States:
A
Functional Analysis, in R. Scmitt (ed.), The Future of Pensions in the United States University
.
of Pennsylvania Press.
Kotlikoff, and John B. Shoven, 1988, Personal Security
J.
Fundamental Social Security Reform, in Susan M. Wachter (ed.), Social
Security and Private Pensions Lexington, Massachusetts: Lexington Books.
Boskin, Michael
Accounts:
Laurence
J.,
A Proposal
for
.
Cukierman, Alex, 1992, Central Bank Strategy. Credibility, and Independence: Theory and
Evidence. Cambridge: MIT Press.
Diamond, Peter A., 1994.
Analisis Economico.
Privatization of Social Security: Lessons
from Chile
,
in Revista
de
and James A. Mirrlees, Optimal Taxation of Identical Consumers when
Markets are Incomplete, in P. Dasgupta, D. Gale, O. Hart, and E. Maskin, eds., Essays in Honor
Diamond, Peter
A.,
of Frank Hahn (Cambridge:
MIT
Press, 1992).
Diamond, Peter A., and Valdes-Prieto, Salvador, 1993, "Social Security Reform," in The Chilean
Economy: Policy Lessons and Challenges B. Bosworth, R. Dornbusch and R. Laban (eds),
.
Washington: Brookings
Gale, Douglas, 1990,
Institution.
The
Efficient Design of Public Debt, in R.
Dornbusch and M. Draghi,
eds.,
Public Debt Management: Theory and History. Cambridge: Cambridge University Press.
Geanakoplos, John and Heraklis Polemarchakis, 1986, Existence, Regukarity, and Constrained
Suboptimality of Competitive Allocations when the Asset Market is Incomplete, in W. Heller,
R. Starr, and D. Starrett (eds), Essays in
Cambridge University
Honor of Kenneth Arrow.
Press.
-29-
3,
65-95, Cambridge:
Geanakoplos, John, M. Magill, M. Quinzii, and
J.
Dreze, 1990, Generic Inefficiency of Stock
Market Equilibria, Journal of Mathematical Economics.
19, 113-151.
and Edward C. Prescott, 1977, Rules Rather Than Discretion:
Inconsistency of Optimal Plans, Journal of Political Economy. 85, 3, 473-492.
Kydland, Finn
E.,
Loewenstein, George, and Jon Elster, 1992, Choice Over Time.
New
The
York: The Russell Sage
Foundation.
Malinvaud, Edmond, 1972, The Allocation of Individual Risks
Economic Theory. 4,
Merton, Robert
and
J.
Shoven,
C,
2,
1983,
in
Large Markets, Journal of
312-328.
On Consumption-Indexed Public
eds., Financial
Pension Plans, Chapter 10
in Z.
Bodie
Aspects of the U.S. Pension System. Chicago: University of
Chicago Press.
Merton, Robert C, and Zvi Bodie, 1992,
Management.
On
the
Management of Financial Guarantees,
Financial
21, 4, 87-109.
Nordhaus, William, 1989, Alternative Approaches to the Political Business Cycle, Brookings
Papers on Economic Activity.
2.
Thomas, 1984, Self Command
Thought, American Economic Review. 74
Schelling,
in Practice, in Policy
and
in
a Theory of Rational
(2), 1-11.
Social Security Technical Panel, 1991, Report to the 1991 Advisory Council on Social Security.
Washington, DC: 1991 Advisory Council on Social Security.
Solow, Robert M., 1980,
On
Theories of Unemployment, American Economic Review. 70,
1, 1-
11.
Stokey, Nancy, 1989, Reputation and
Time Consistency, American Economic Review. 79
(2),
134-139.
Strotz,
Robert H., 1955, Myopia and Inconsistency
Economic
Studies. 23, 165-180.
-30-
in
Dynamic
Utility
Maximization, Review of
Endnotes
Paper prepared for the conference Mandatory Pensions: Funding, Privatization and
Policy, Santiago, Chile, January 26-27, 1994
1
Macroeconomic
This paper will not consider the risks to the rest of government budgeting from the
2.
financial condition
The
3
of the retirement system.
fact that the funds
do not flow through the government does not guarantee that they
Since the earnings on pension assets are particularly easy to
are not particularly vulnerable.
tax,
because
in the short
run
it is
very hard to do tax evasion, they might have been subjected
to special taxes.
4.
Geanakoplos and Polemarchakis (1986), Geanakoplos, Magill, Quinzii, and Dreze (1990).
5.
There
is
a similar, although less studied, tradeoff in the labor market.
individuals are choosing different risk-expected return tradeoffs as well.
careers
is
far riskier than pursuing others.
In choosing careers,
Pursuing some
Again, incomplete markets imply a lack of Pareto
efficiency in general.
The
6.
literature
on insurance normally assumes
that there is
no
social risk,
and then explores
the role of insurance markets in spreading individual risks (Malinvaud, 1972).
social risks, then the allocation
may have
7.
Kydland and Prescott (1977).
8.
Stokey (1989).
9.
Alesina (1988), Nordhaus (1989).
Once
there are
different properties.
10. Ainslie (1992), Strotz (1955), Schelling (1984),
and the collection edited by Loewenstein
and Elster (1992)).
1 1
There are large aggregate
risks associated with the provision
of retirement income.
For
US, the annual percentage reduction in age-adjusted central death rates for
males was -0.19% for 1954-1968, but 1.56% for 1968-1988 (Social Security Technical Panel,
example,
in the
1991).
For a discussion of the use of public debt to share risks in an OLG model with
incomplete markets, see Gale (1990). For a discussion of optimal taxation in a Ramsey
12.
economy with incomplete markets and
social risk, see
-31-
Diamond and Mirrlees
(1993).
were real investments, with random aggregate returns, then full risk sharing would
depend on two sources of risky income. Similarly, if type A's received random income in
period two, with a different random structure and an identifiably different character (from the
perspective of political identification) we would again have two sources of income risk. For a
setup of this model with real investments, see the appendix.
13. If there
14.
I
will treat the political process as
random
relative to social security.
That
is, I
assume
in social security do not play a major role in the determination of
For a model that determines voting for political parties as a function of
expenditure, and debt policies the parties will follow, see Aghion and Bolton (1990).
proposed changes
that
political
the tax,
outcomes.
For example, before the U.S. system was indexed, the benefit formula and the calculation
of lifetime average wages were both in nominal terms, making for periodic increases in
15.
benefits in response to inflation.
16.
The use of
may
private assets does affect the nature of political process since the rate of return
well be different
on the funds probably
17.
This was not a satisfactory system.
-
even
if
the aggregate return to the
affect legislated
outcomes on
economy
is
unchanged, and returns
benefits.
For a discussion of independent central banks, see Cukierman (1992).
of Boskin, Kotlikoff and Shoven (1988), actuaries set the
rate of conversion of current taxes paid into claims on annuitized retirement income.
18.
For example,
19.
A
20.
Civil
in the proposal
model with
war
is
capital is briefly presented in the appendix.
tough on publicly provided retirement income.
-32-
)
1
7 8
J L
Date Due
m 24
W»-
Lib-26-67