8/11/2019 Diamond Barr 2010
1/24
Reforming Pensions: Lessons from Economic Theory and Some Policy Directions [withComment]Author(s): NICHOLAS BARR, PETER DIAMOND and Eduardo EngelSource: Economa, Vol. 11, No. 1 (Fall 2010), pp. 1-23Published by: Brookings Institution PressStable URL: http://www.jstor.org/stable/25800053.
Accessed: 17/04/2014 17:57
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.http://www.jstor.org/page/info/about/policies/terms.jsp
.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
.
Brookings Institution Pressis collaborating with JSTOR to digitize, preserve and extend access toEconoma.
http://www.jstor.org
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=brookingshttp://www.jstor.org/stable/25800053?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/25800053?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=brookings8/11/2019 Diamond Barr 2010
2/24
NICHOLAS
BARR
PETER DIAMOND
Reforming
ensions:
Lessons
from
Economic
Theory
nd
Some
Policy
Directions
Pension
systems
have
wide-ranging
and
important
effects.
They
influ
ence
the
living
standards of
older
people
and hence thewelfare of both
older
people
and
their children.
They
can
also
affect
national economic
performance through potential
effects
on
the labor
supply
and
saving.
The
design
of
pensions
therefore
matters.
In
discussing
the
topic,
this
paper
draws
on
two
of
our
earlier works.1
It
starts
by
setting
out
some
central lessons
from
economic
theory.
The
second
section
derives
some
policy
implications
of
particular relevance toLatin America. The policy chapters inour two earlier
books
focus
particularly
on
two
countries in which
we
have worked:
China,
which is
important
because
of
its
size,
and
Chile,
which
is
important
because
its
1981
reforms
were
widely adopted
in Latin
America
and elsewhere and
which has become
an
exemplar
in
the
pension
debate.
Box
1
defines
some
of
the relevant
terms.
The final section concludes.
KeyLessonsfrom conomicTheory
Economic
theory
offers
a
series
of conclusions
that should
frame
policy
design.
Some
of
them,
though
apparently
obvious,
are
frequently forgotten;
others
can
be counterintuitive.
We
focus
on
five
sets
of lessons:
pension
sys
tems
have
multiple
objectives;
different
pension
systems
share
risks
differ
ently,
both
across
people
and
over
time;
there
is
no
single
best
pension
system;
pensions
should
be
analyzed
in
a
second-best
context,
that
is,
taking
account
of market
imperfections
and other
distortions;
and amove to
funding
may
or
may
not
be the
right
policy.
Ban
is with the London School
of
Economics;
Diamond is
with
the
Massachusetts
Institute
of
Technology.
1. Barr
and
Diamond
(2008,
2010).
1
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
3/24
2
ECONOMIA,
all 010
BOX
1.
Terminology
Defined-benefit
DB)
pensions
are
systems
n
hich
the
pension
benefit
s
determined
s
a
function
f
the
worker's
historyf coveredearnings.Theformulamay be based on theworker'sfinalwage and lengthfservice ronwages overa
longer eriod
(for
xample,
a
full
areer).
A DB
system
may
be
fully
r
partially
unded,
r
it
may
be unfunded.
n
pure
DB
arrangement,
nsofar
s
the
degree
of
funding
s
maintained,
the
sponsor's
ontributions
re
adjusted
tomeet
anticipated
obligations;
hus,
he
risk f
varying
ates
f
return
o
pension
assets
falls
n
the
sponsor.
Defined-contribution
DC)
pensions
are
systems
n
hich the
benefit
s
etermined
by
the
value
of assets
accumulated
toward
person's
pension.
Benefits
may
be taken
s a
lump
um,
s
a
series
f
withdrawals,
or
through
n
annuity.
he
expected
discountedvalue
of
benefits
s hus
equal
to
the value of
assets
(in
technical
erms,
he
benefits
re
determined
actuarially).1
pure
DC
plan adjusts
obligations
o
match available
funds,
o
that the
individual
ears
the
portfolio
isk.
Fully
unded
pensions
pay
all benefits rom
ccumulated
funds.
ee also
partially
unded
pensions.
A
noncontributory
ension
isbased
on
age
and
years
of
residence.
Notional efined-contributionNDC) ensionsarefinanced n a pay-as-you-go rpartially undedbasis,with a person's
pension
bearing
quasi-actuarial
relationship
o his
or
her
lifetime
pension
contributions.
Partially
unded
pensions
pay
benefits oth from
ccumulated
ssets
and
from
urrent
ontributions.
Pay-as-you-go
PA
YG)
ensions
are
paid
out
of current
evenue
(usually
y
the
state,
from
ax
revenue)
rather
han
out
of
accumulated funds.
Partially
unded
pensions
are
often
referred
o
as
PAY6.
A
provident
und
ays
a
defined-contribution
pension
based
on
the
performance
f
a
single,
entral und
rather han
on
the
performance
f individual
ccounts.
1. The
discountedalue
epends
n
therelevantnterest
ate
nd
load actors.
Pension
Systems
ave
Multiple
bjectives
For
the individual
or
family,
the
major objectives
of
pensions
are
consump
tion
smoothing
(redistribution
from
one's
young
self
to
one's older
self)
and
insurance.
Governments
have
additional
objectives, including
poverty
relief
and redistribution.
Any
analysis
of
pension
reform needs
to
take
account
of
the full
range
of
objectives
alongside
other
policy goals,
such
as
economic
efficiency
and
output
growth.
Differentension
Systems
hare isks
Differently
Separate
from
their
redistributive
effects,
different
pension
systems
share
risks
differently.
Pension
systems
are
subject
to
multiple
sources
of
risk,
and
they
have different
underlying
philosophies
of
who
should bear
those
risks. This
section looks
at
four
types
of
pension,
with
a
focus
on
how
they
distribute
risk.
We startwith two polar extremes of fully funded systems: pure funded
defined-contribution
(DC)
plans
and
pure
funded
mandatory
defined-benefit
(DB)
plans.2
In
the
case
of
the
former?also known
as
funded
individual
accounts?individual
workers
set
aside
a
given
fraction
of their
earnings
to
2.
See the
glossary
in
box
1
for
a
brief
definition of
these
terms.
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
4/24
Nicholas
Barr
nd
Peter
Diamond
3
buy
private
financial
assets,
which
are
accumulated
until retirement.
At
that
point,
the retirees
can
either
purchase
an
annuity
or
take
a
series of
with
drawals from the accumulation. Fluctuations in the cumulative return on
assets
during working
life
affect the individual
account
holder
by
affecting
the
amount
available
to
finance retirement.
If
theworker
buys
an
annuity,
he
or
she
will have faced the risk
in the
pricing
of
annuities,
reflecting
both
mor
tality projections
and
asset returns
from this
point
forward. Once the
annuity
is
purchased,
however,
further fluctuations
in
asset returns
and the
develop
ment
of
mortality compared
to
the
projections
used
in
pricing
the
annuity
are
borne
by
the insurance
company,
unless
annuity
benefits
are
indexed
for
asset
returns
(a
variable
annuity)
or
for
mortality
realizations. As the insurance
company
adapts
to
the realizations of
returns
and
mortality,
it
may
in
turn
adjust
the
price
of its
policies,
the
compensation
it
pays
its
workers,
and the
returns
to
its
shareholders
(if
it is
not
a
mutual
company).
If the retiree does
not
buy
an
annuity,
he
or
she faces
mortality
and
return
risks.
That
is,
the risks
of different
outcomes
up
to
retirement
fall
on
the individual
retiree,
since
benefits
adjust
to
what
is in
a
worker's individual
account at
the time of retire
ment.
Annuitization shifts
risks
after retirement
to
insurers,
but the retiree
still
faces
the risk of the
pricing
of
annuities
at
the
start
of retirement.
In
the
case
of
pure
funded
mandatory
defined-benefit
(DB)
plans,
individ
ual workers receive retirement
benefits based
on
a
formula
relating
benefits
to
their
earnings history.
Thus,
both
asset
return
and cohort
mortality
risks
are
managed
through
adjustments
in contributions and
are
therefore borne
by
the
workers
during
their
working
lives. That
is,
in
order
to
sustain
full
funding
for the
provision
of
future
benefits
as
set
out
in
the benefit
formula,
contribu
tions need to be adjusted as returns vary and mortality projections change.
Since
a
single
fund is available
for
all benefits
and
changes
in
contribution
rates
are
normally
uniform
across
workers,
there
is
a
collective dimension
in
that the need
at
any
time
depends
on
the full
array
of workers
who
share
in
the need
for
aggregate
contributions.
A
plan
does
not
have
to
be
fully
funded
at
all times?fluctuations
in
the
degree
of
funding
are
a
device
for
shifting
risks
intertemporally
and,
therefore,
across
workers born
in different
years.
The
fund
may
purchase
annuities,
shifting
risks
to
insurance
companies.
A
plan may have a sponsor who absorbs the risk, such as a corporate employer
or
the
government relying
on
taxes
in addition
to
contributions.
The former
shifts risks
to current
and
future
workers
at
the
firm,
shareholders,
and
possi
bly
customers,
if
the
company
responds
to
a
pension
deficit
by
raising
its
prices.
If the
government
is the
sponsor,
the
risks
are
shifted
to
current
and
future
taxpayers.
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
5/24
4
ECONOMIA,
all 010
Thus,
one
element
in the distinction between
DB
and DC
plans
is the
focus
on
adjusting
contributions
or
adjusting
benefits.
In
practice,
corpora
tions and
governments
commonly adjust
both,
so
that
plans
are
not
typically
pure,
although
the
labeling
can
affect
the
legislative
outcome.
A
second ele
ment
is
the
collective
aspect
of
having
a
single
fund
rather
than
individual
accounts.
This
opens up
opportunities
for
risk
sharing
within
the DB
con
text,
although
governments
could redistribute
across
individual DC
accounts
and
may
provide
insurance
on
the
rate
of
return,
financed
from
outside the
pension
system.
A
central fund has
lower
administrative
costs
and avoids
poor decisionmaking by workers, although it is at risk of poor investment
decisions
by
the
fund
managers.
However,
a
central
fund
forgoes
the
oppor
tunity
to
have
different
degrees
of
asset
return
risk forworkers with different
levels of risk aversion.
Corporate pensions
should
remain close
to
fully
funded
since
corpora
tions
can
fail,
leaving
the
workers with less than
planned
or
leaving
it
to
the
government
to
insure the
workers.
While
occasionally
governments
go
bank
rupt
as
well,
this
risk is
not
significant
in
countries
with
sound economies
and effective governments. Governments can therefore have systems thatare
less than
fully
funded,
but still well
designed.
The
extreme
version has
no
funding?something
that
is
not
standard
practice
since
most
systems
main
tain
a
buffer fund for short-run
fluctuations,
and
government
plans
sometimes
provide
significant
funding.
A third
option
is
a
pay-as-you-go (PAYG)
defined-benefit
system
financed
by
social
security
contributions.
In
this
arrangement,
there
are no
assets,
so
the rate-of-return risk is
of
no
significance.
Instead,
the risk
to
the
plan's income comes from fluctuations in the earnings of covered workers,
given
the contribution
rate.
Since there is
no
opportunity
for
surpluses
and
deficits,
the risks
are
shared
among
current
workers
through
changes
in
contributions.
If
the fund
can run
surpluses
(partial funding)
or
deficits
(borrowing
against
future contributions
or
using previous
surpluses),
the
risks
can
be
shared with
future
workers
or
eased
by
the
presence
of
accu
mulated
past
contributions.
Any
accumulation,
whether
positive
or
negative,
involves
risk in
rates
of
return.
There is also the risk
of the
evolution of
mor
tality
rates:
if
people
live
longer,
the
cost
of
paying
a
given
level
of
benefits
will
increase.
A
final
option
involves
systems
financed
at
least in
part
by
general
tax
rev
enues.
For
example,
in
a
noncontributory
pension
(referred
to
as a
citizen's
pension),
the
risks
are
shared
by
all
taxpayers
and
are
thus
spread
across
gen
erations
(since
future
taxes
as
well
as
current taxes
can
change
as
debt
varies).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
6/24
Nicholas
Barr
nd PeterDiamond
5
In
practice,
plans
are
not
pure,
and countries
frequently
adjust
both
con
tributions and
benefits,
thus
sharing
risks between workers
and
pensioners.
A
central
question
for
policymakers
is how risks should be
shared,
a
question
with both
efficiency
and
equity implications.
As with
redistribution,
different
answers are
possible,
but it is
a
major
error
to
ignore
the
question.
There
s
No
Single
est ension
System
Policymakers
face
a
series of
constraints
in
pursuing
the
multiple objectives
mentioned
above,
including
fiscal
capacity
(stronger
fiscal
capacity
makes
it easier for the system to find additional revenues for a pension system);
institutional
capacity
(stronger
institutional
capacity
makes
feasible
a
wider
range
of
design options);
the
empirical
value of behavioral
parameters
(such
as
the
responsiveness
of labor
supply
to
the
design
of
the
pension
system
and
the
effect
of
pensions
on
private saving);
and the
shape
of the
pretransfer
income distribution
(a
heavier lower tail increases the need
for
poverty
relief).
The
reason
why
there
is
no
single
best
system
is
simple: policymakers
at
different times and in different
places
attach different relative
weights
to the
various
objectives;
and
the
pattern
of
constraints,
including
political
and his
torical
constraints,
will
differ
across
countries.
If
objectives
and constraints
differ,
the
optimum
will
generally
differ,
as
well.
Pensions
hould
e
Analyzed
n
Second-Bestontext
The
assessment
of
the
optimal pension
system
should
be made
in
a
second
best context.3 Simple theory assumes that individuals make optimal choices
and that labor
markets,
savings
institutions,
and insurance markets exist and
function
ideally.
Those
assumptions
do
not
apply
to
pension
systems
because
workers
and
pensioners
face information
problems,
behavioral
problems,
missing
markets
(for
example,
the lack
of
private
instruments
to
provide
indexation),
and broader
factors such
as
the
inescapable
existence
of distor
tionary
taxation.
Framing
the
argument
in
second-best
terms starts
from
the
multiple objec
tives of pension systems. Thus, policy has tooptimize (notminimize ormax
imize)
across
a
range
of
objectives,
which
cannot
all
be achieved
fully
at
the
same
time.
Policy
has
to
seek
the
best balance between
consumption
smooth
ing,
poverty
relief,
and
insurance,
and
this
balance
will
depend
in
each
society
3.
For
a
fuller
discussion,
see
Barr
and
Diamond
(2008,
section
4.2 and
box
9.6).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
7/24
6
ECONOMIA,
all 010
on
the
weights given
to
those
and other
objectives
and
to
the different
con
straints that societies face.
AMove
o
Funding
ay
r
May
ot ethe
ight
olicy
In
assessing
whether,
and
to
what
extent,
a
move
to
funded
pensions
might
increase
welfare,
policymakers
need
to
ask
two sets
of
questions.4
First,
is
a
move
toward
funding
welfare
improving?
For
example,
does it increase
out
put,
either
by increasing savings
in
a
country
that is short of
saving,
or
by
strengthening capital
markets,
thereby
improving
the
efficiency
with
which
savings
are channeled into
productive
investment? Does it have desirable
intergenerational
redistributive effects?
Second,
even
if
a
move
toward fund
ing
is
in
principle
welfare
improving,
is such
a move
feasible? Does the
coun
try
in
question
have the
economic
conditions and
institutional
capacity
necessary
to
implement
schemes that
are
safe and
administratively
cheap?
The
answers
to
these
questions
will
vary
over
time and
across
countries.
First,
an
increase in
saving
may
not
be the
right objective, particularly
in
a
country
where the
saving
rate
is
already
high
(for
example,
China).
Second,
a move to
funding
may
or
may
not
increase
saving.
For
example,
fund
ing
through
the
issue of
new
government
bonds
will not
do
so.
Similarly,
increased
mandatory pension saving
may
be
largely
offset
by
declines
in
voluntary private saving
or
by
increases
in
government
borrowing,
perhaps
to
help
finance the transition
costs
of
a move
to
a
regime
with
more
funding.
Third,
formal
capital
markets
may
or
may
not
allocate
funds
to
investment
more
effectively
than
informal
capital
markets. Gains in the
effectiveness
of
capital
markets will
depend
on
effective
administration
and
on
political sup
port
for
improved regulation.
Thus,
funding
may
increase national
saving
or
expand
explicit public
debt,
or
both;
and it
may
improve
the
operation
of
cap
ital
markets. Either is
possible;
neither
is
inevitable.
The economic
case
for
funding
has
to
be
analyzed
in
each
country.
In
addition
to
the
potential
effects
on
output,
a
move
to
funding
has inter
generational
effects.
If
funding
is
to
raise
output
growth
in the
future,
it
must
increase
saving today.
But for
saving
to
increase,
there
must
be
a
decline in
consumption, by government or by today's workers or by today's retirees.
Thus,
a
move
to
funding
generally imposes
a
burden
on
today's
generations
to
the
benefit of
future
generations,
an
outcome
that
may
or
may
not
be
good
policy.
More
generally,
introducing
a new
PAYG
system
allows the
early
4. For
a
fuller
discussion,
see
Barr and
Diamond
(2008,
section
6.3).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
8/24
NicholasBarr nd Peter
Diamond
7
cohorts
to
receive
larger
pensions
than
if
the
new
system
were
fully
funded.
Consequently,
any
choice between
PAYG,
partial funding,
and full
funding
is also and
necessarily
a choice about the
intergenerational
distribution of
income and of risks.
Even if
funding
does
increase
output,
the
change
cannot
be
presented
as an
unambiguous improvement.
The
answers
to
these
questions
shed
light
on
whether
a move
to
funding
might
be
optimal.
A
separate
issue is whether it is feasible. Funded
pensions
make
significant political
demands
on
government
and
require
significant
institutional
capacity
in both the
public
and
private
sectors.
The
litany
of
funded schemes that have
failed
completely
or
that have
not
lived
up
to
the
promises
thatwere made for them attests to the
importance
of these
opera
tional
issues,
as
well
as
to
the
dangers
of
basing policy
on
untested and
per
haps
excessively
optimistic predictions.
Policy
mplications
In discussing policy direction in the context of Latin America, it is helpful
to start
with
consumption
smoothing.
This discussion establishes
the
con
text
for
examining
poverty
relief
generally
and the
2008
reforms
in
Chile
in
particular.
Consumptionmoothing
It
is
a
mistake
to
conflate
the
separate
concepts
of individual
accounts
and funding. Individual accounts can be fully funded (such as the Chilean
DC
scheme),
partially
funded,
or
PAYG
(for
example,
a
notional defined
contribution
scheme
with
no reserves
beyond
a
buffer fund
to
cover
a
few
months of
payments).
The discussion
below looks
first
at
notional
defined
contribution
(NDC)
pensions
and then
at
reform
options
for funded indi
vidual
accounts.
ndc
pensions.
A recent
innovation
internationally,
pure
NDC
systems
mimic funded
individual
accounts,
but
on a
largely
or
wholly
pay-as-you-go
basis. In the simplest such scheme, each worker pays a contribution of a fixed
percentage
of
his
or
her
earnings,
which
is
credited
to
a
notional
individual
account,
while
theworkers'
contributions
this
year
largely
or
wholly
pay
this
year's pensions.
The
government
keeps
a
record
of
individual
contributions,
each
year
attributing
a
notional
interest
rate
to
each
worker's
accumulation.
When the
worker
retires,
his
or
her
notional
accumulation
is
converted
into
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
9/24
8
ECONOMIA,
all
010
an
annuity.
In
a
pure
NDC
system,
benefits
are
strictly
related
to
the size
of
a
person's
notional accumulation.5
The
system
can
also
incorporate
redistribu
tion,
such as minimum benefits or
pension
credits for
caring
activities,
and
the scheme
can
incorporate
partial
funding.
NDC schemes
have
a
range
of
potential
advantages.
The
system
is sim
ple
from
the
point
of
view of the
worker,
and
the fact that
it is administered
centrally keeps
administrative
costs
low.
Risk is
kept
low,
since
an
unfunded
NDC
pension
system
is
not
affected
by capital
market
volatility,
themain
source
of
risk
to
funded individual
accounts. An NDC
system
does
not
require
the
institutional
capacity
to
manage
funded schemes.
In
addition,
saving
may
be the
wrong
policy,
or
people
may
not want to
save.
Finally,
NDC
can
be
partially
funded
and
can
be
the basis for
a
future
move
to
full
funding,
so
the
approach
may
provide
a
starting point
if financial
market
tur
bulence continues.
As discussed
earlier,
solid
analytics
should
underpin
the choice
between
NDC and
funded
accounts.
The choice should
be
economic,
not
primarily
political.
funded individual accounts. If policymakers want a system with
funded
individual
accounts,
there
are
different
ways
to
implement
the
policy.
In the best-known
arrangement,
workers have
significant
individual
choice
over
their
pension provider,
as
in
Chile
and
Sweden.
Choice
is
generally
assumed
to
be
welfare-enhancing. Simple
economics
argue
that
policy
should
allow
people
to
choose
their
own
pension
provider
in
a
competitive
market.
Such
choice,
it is
argued,
benefits
the individual
in
the
same
way
as
choice
and
competition
for
clothes,
cars,
restaurants,
and
MP3
players.
That view
needs to be tempered by two sets of factors, however: impediments to good
choice
by
individuals;
and the fact that choice has
costs.
The model of thewell-informed
consumer
does
not
hold
in
many
areas
of
social
policy,
as
demonstrated
by
the economics
of information.
In
the
con
text
of
pensions,
there
is
ample
evidence
of
poor
information.
Many
do
not
understand
basic
concepts
in
finance:
Orszag
and
Stiglitz
quote
the
chairman
of theU.S.
Securities
and
Exchange
Commission
as
stating
that
over
50
per
cent
of Americans do
not
know the
difference
between
a
stock and
a
bond.6
5.
The
system
is
quasi-actuarial
in
that
the
interest
rate
attributed
to
a
worker's notional
accumulation is
not the market
return
to
financial
assets,
but
an
interest
rate
determined
by
the relevant
pensions legislation?for example,
the
growth
in the total
wage
bill of covered
workers
each
year.
6.
OrszagandStiglitz(2001,p.
37).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
10/24
Nicholas
Barr
nd
Peter
Diamond 9
Most
people
with
an
individual
account
do
not
understand
the
need
to
shift
from
equities
to
bonds
as
they
age,
and
virtually
nobody
realizes the
signifi
cance
of administrative
charges
for
pensions.
As
noted,
choice has
costs,
and
in the
case
of
individual
accounts
those
costs
are
significant:
over a
full
career,
an
annual
management
charge
of 1
percent
of the
individual's
accu
mulation reduces the accumulation
(and
hence
the
pension) by
20
percent.7
Recent lessons
from behavioral economics
also
yield
powerful
lessons,
explaining
such
phenomena
as
procrastination (people
delay saving,
do
not
save,
or
do
not
save
enough),
inertia
(people
stay
where
they
are),
and
immo
bilization
(whereby
conflicts and confusion lead
people
to
behave
passively,
like
a
deer in the
headlights).8
These
bodies of
theory
suggest
that the
benefits
of
wide
choice
are
likely
to
be
very
limited,
and
for
most
people
likely
to
be
outweighed by
the
costs
of
choice.9
These considerations
suggest
a
series of
guidelines
for the
design
of individual
accounts:
?Use automatic
enrollment;
?Keep
choices
simple
(for
most
people,
highly
constrained choice
is
a
deliberate and welfare-enhancing feature of good pension design,
although
one
of
the
options
could be
to
allow
individual
choice);
?Design
a
good
default
option
for
people
who make
no
choice;
and
?Decouple
fund
administration
from fund
management,
with
account
administration
centralized
and
fund
management
organized
on a
whole
sale,
competitive
basis.
The U.S.
Thrift
Savings
Plan for federal civil
servants
complies
with
these
criteria.10
The
plan
offers
participants
a
very
limited
choice
of
portfolios.
Ini
tially therewere three: a stock market index fund, a fund holding bonds
issued
by
private
firms,
and
a
fund
holding
government
bonds.
In 2007 work
ers
could
choose from
six
funds,
including
a
life-cycle option
(that
is,
an
option
inwhich
a
person's
portfolio
shifts
automatically
from
mainly
equi
ties
to
mainly
bonds
as
he
or
she
ages).
A
government
agency
keeps
central
ized
records of
individual
portfolios.
Fund
management
is
on a
wholesale
basis.
Investment
in
private
sector
assets
is handled
by
private
financial
firms,
which
bid
for
the
opportunity
and
which
manage
the
same
portfolios
in
the
voluntary private
market,
providing
insulation from
political
interference.
7.
See
Barr
and
Diamond
(2008,
Box
9.4).
8.
For
a
fuller
discussion,
see
Barr
and
Diamond
(2008,
Box
9.6).
9.
For
a
fuller
discussion,
see
Barr and
Diamond
(2008,
section
9.3).
10. For
more
information,
see
the
U.S.
Thrift
Savings
Plan
website
(www.tsp.gov).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
11/24
10
ECONOMIA,
all 010
This
plan
thus
(a)
simplifies
choice
for
workers,
respecting
information
constraints,
(b)
includes
automatic
enrollment,
(c)
has
a
default
option,
and
(d)
keeps
administrative costs
astonishingly
low,
at as little as 6 basis
points
annually,
or
60
cents
per
$1,000
of
account
balance.
By
the end of
2009,
the
program
had
grown
to
include 3.2
million active
participants
and 4.3
million
total
accounts
and
held
assets
of
$244
billion.
The
United
Kingdom
is intro
ducing
a
similar
arrangement.11
Poverty
elief
Pure DC individual accounts provide consumption smoothing, but theydo not
provide
adequate
poverty
relief for
workers with limited
lifetime contribu
tions,
typically
as a
result of
low
earnings, fragmented
careers,
or
a
work his
tory
substantially
in the informal
sector.
In this
regard,
the
post-1981
system
in
Chile,
which is based
heavily
on
individual
accounts,
offers three
primary
lessons:
first,
mandatory
funded individual
accounts
can
be
part
of
a
good
reform,
but such
a
reform
is
not
easy
and
depends
on
complementary
reforms;
second,
private supply
and
competition
alone
are
not
sufficient
to
keep
down
transactions costs or
charges;
and,
finally,
unless
accompanied by
a robust
sys
tem
of
poverty
relief,
individual
accounts
do
not
constitute
a
pension
system,
but
only
part
of
a
pension
system.12
To
explain
the
shape
of
Chile's 2008
reforms,
it is
necessary
to
consider
the
changes
in the economic and social environment
within which social
pol
icy
operates.
The
contributory
principle
assumed thatworkers had
long,
sta
ble
employment,
so
that
coverage
would
grow.
History
has
not
sustained this
supposition.
To
explain why,
consider
the
way
the world has
changed
over
the
past
sixty
years.
Social
policy
in
1950
was based on a series of
assump
tions: theworld
was
made
up
of
independent
nation
states;
employment
was
generally
full time
and
long
term;
international
mobility
was
limited;
the
sta
ble nuclear
family
with
a
male breadwinner and
a
female
caregiver
was
the
norm;
and
skills,
once
acquired,
were
lifelong.
While these
assumptions
were
not
entirely
true
even
then,
they
held well
enough
to
be
a
realistic basis for
social
policy.
The
world
today
is
very
different.
There
is
increasing
international
com
petition.
The
nature
of work is
changing,
with
more
fluid labor markets.
11.
See
the website of the
U.K. National
Employment Savings
Trust
(www.nestpensions.
org.uk).
12. Barr and Diamond
(2008, p. 238).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
12/24
NicholasBarr nd Peter
Diamond
11
International
mobility
is also
increasing,
and
it
is
likely
to
continue
to
do
so.
The
nature
of the
family
is
changing,
with
more
fluid
family
structures
and
rising
labor-market
participation
by
women.
Finally,
the half-life of skills
has declined.
Given this
new
environment,
the
key
drivers of
change
when
considering
pension design
are more
diverse
patterns
of
work,
which
create
problems
in
terms
of
coverage
for
contributory
benefits tied
to
employment,
and
increas
ingly
fluid
family
structures,
which make itdifficult
to
base women's
benefits
on
their husbands'
contributions.
Those circumstances
create
a
case
for
a
noncontributory
basic
pension,
that
is,
a
pension
financed from
general
taxation and awarded
at
a
flat
rate to
any
one
who
meets
an
age
and residence
test.
The
Chilean Presidential
Advisory
Council
on
Pension Reform
put
the
point succinctly:
The
prevailing image
at
the time of the
pension
reform
[in
Chile in
1981],
of
a
workforce
composed
mainly
ofmale
heads of
household,
with
permanent
jobs,
contributing
contin
uously throughout
their active
lives,
has become less and
less
representative
of the real situation of
the
country
and will become
even
less
so
in
the future.
This
means
that the system designed at thatpoint in time is also gradually los
ing
its
ability
to
respond
to
the
needs of the
population
as a
whole. 13 For
these
reasons,
Chile
introduced
a
noncontributory pension
(the
Pension
Bdsica Sol
idaria)
in
2008,
initially
for
pensioners
in
the
poorest
40
percent
of the
popu
lation and
rising
to
60
percent
when
fully phased
in.
The
case
for
a
non-contributory pension
is that it
strengthens
poverty
relief
in
terms
of
coverage,
adequacy,
and
gender
balance;
improves
incentives rel
ative
to
income-tested
poverty
relief;
provides good targeting
(in
that
age
is
a
useful indicator of poverty); and can assist international labormobility. The
obvious
question
is how
to
pay
for such
a
benefit. Three instruments
can
match
expenditures
with
budgetary
constraints: the size
of the
pension,
the
age
at
which it is first
paid,
and the
option
of
an
affluence
test,
which
keeps
benefits
from the best-off
segment
of the
population.
The Chilean mechanism for
con
trolling
beneficiaries
was
mentioned above.
In
Canada,
95
percent
of older
people
get
the
full
flat-rate
benefit,
and
only
2
percent
are
entirely
screened
out.
In
the
Netherlands,
the benefit is awarded
only
on
the
basis of
age
and
res
idence, with no testof income or assets. Box 2 summarizes the international
experience
with this
type
of
pension.
13.
Presidential
Advisory
Council
on
Pension Reform
(2006).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
13/24
BOX
2.
NoncontributoryPensions
A number
f
countries
ave
implemented
noncontributory
ension
system.2 igh-income
ountries
ith
this
arrangement
nclude
ustralia,
anada,
the
Netherlands,
nd
New
Zealand.
For
middle-income
countries,
he
noncontributory
ension
introduced
n
hile in 008
is
discussed
in he text.
outh
Africa lso
has
a
noncontributory
pension,
namely,
he
State
Old
Age
Grant. he
case
is
interesting
n hat
it
reaches
not
only
urban
pensioners,
ut
also the
rural
elderly.
he
pension,
which is
paid
to
men
at
age
65
and
women at
60,
is
financed
rom
general
revenue
ith
no
contribution
onditions. he
benefit
s
round
half f
average
household
income
nd is
hus
high
relative
o
the
very
ow
incomes f
most
nonwhites
n
outh
Africa,
ut low
relative
o the incomes
f
the better ff.
Originally
ntroduced
s
poverty
elief
or
hites
during
he
1930s,
the
plan
was
gradually
xpanded
tocover ll race
groups.
Research
suggests
that it s
highly
ffective
oth in
erms
f social
policy
nd
in he
way
the
plan
is
implemented:
The South
African
ocial
pension
is
n
example
of
a
transfer
plan
where
eligibility
s
determined
by
age.
In
spite
of the
simplicity
f the
targeting
indicator,
he
pension
is
ffective
n
reaching
he
poorest
households
nd thosewith children_The
South
African
authorities
ave
overcome
the
difficultiesf
making
cash transfers
o
even
remoterural
reas
and
of
checking ligibility
among
even
illiterate
pensioners. 3
nmost
urban
reas,
people
receive
he
pension
through
ank
accounts
or
post
offices.
In
rural
reas,
government
as
outsourced
delivery
o
the
private
ector,
rganized
at
the
provincial
evel.At its
est,
the
system
s ffective
nd innovative.
n
ome
areas,
vehicles fitted
ith cash
dispensers
go
to
designated
places
at
preordained
imes.Pensioners
nter
their dentificationumber
or
ingerprint),
nd
their
pension
is
paid
out.
Notionally,
there is
government
fficial
n
hand to
provide
help,
but this
facility
s
patchy.
Anumber f low-income ountries lso have
noncontributory
ensions
(sometimes
alled social
pensions), including
Bolivia,
Botswana, Namibia,
nd
Nepal.
Total
spending
s
typically
mall
in
hese
cases
(below
1
percent
f
GDP
in
Botswana, Namibia,
nd
Nepal),
and
the benefit
s lso
generally
mall.4
Pensions
of
this
ort
have the
great
potential
dvantage
of
extending
overage
to
people
with
limited ontributions
records,
specially
women
andworkers
in he
informal
ector. n
assessing
their
desirability
nd
feasibility
n
particular
country,
olicymakers
eed
to
consider
range
f factors:
?How well
could the
pension
be
targeted?
he
cost
effectiveness
f
a
noncontributory
niversal
pension
depends
on
the
accuracy
f
age
as a
targeting
evice. In
principle,
he
more
poor
people
a
country
as,
the
greater
he
importance
f
poverty
elief nd
the
better-targeted noncontributory
ension
will be.
The
extent
to
which
age
alone is
good
indicator,
however,
ill
vary
rom
country
o
country, epending,
for
example,
on
the
extent
to
which old
people
live
lone
or as
part
ofan extendedfamily.
?Is administrative
capacity
ufficient?ven
simple
pension
has
administrative
equirements.
he
government
must
be able
to
establish
people's
ages
and
to
guard
againstmultiple
claims
by
one
person
and
claims
by
relatives
n
behalfof
a
pensioner
who
has
died.
?Is
the
cost
of
delivery
ow
enough
relative
o
the
size
of
pension
being
considered?
Finally,
here
a
government
as
the
necessary
mplementation
apacity,
olicymakers
ave
a
range
f
options
for
containing
osts.
First,
he level f the
pension
can
be
kept
low.
For
example,
it s
only
10
percent
f
GDP
per
capita
in
Botswana and
Nepal.
Second,
the
age
at
which the
pension
is
irst
paid
can
be
set
high.
In
Nepal,
only
1.1
percent
f the
population
is lder
than the
qualifying
ge.
Finally,
f
dministrative
capacity
ermits,
further
option
is
o
pay
a
smaller
pension
at
first
for
xample,
to
pensioners
ged sixty-five
o
seventy-five)
nd
a
larger
ne
thereafter
to
those
seventy
five nd over).
2.
For urther
iscussion
f
noncontributory
ensions,
ee
Willmore
2007).
3. Case ndDeaton
1998,
.
1359).
4.
Willmore
2007,
able
).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
14/24
Nicholas
Barr
nd
Peter
Diamond
13
Conclusion
In considering the implications of economic theory for thedesign of pension
systems,
a
number of
conclusions stand
out.
First,
the
analysis
should
con
sider the
pension
system
as
a
whole. Pension
design
affects the labor
market,
economic
growth,
the distribution
of
risk,
and the distribution of
income,
including
effects
by gender
and
generation. Analysis
should consider the
entire
pension
system.
There is
no
efficiency gain
from
designing
one
part
of
the
system
(such
as
individual
accounts)
without distortions
if
distortions
are
then
placed
elsewhere
to
accomplish
other
objectives.
Hence,
there
can
be
no
gain
from an actuarial second-tier
pension given
the need for a
poverty-relief
element
in
the first tier.What is relevant for
analysis
is the combined effect
of the
system
as a
whole.
Consequently,
analysts
must
simultaneously
con
sider the
parts
of the
pension
system
that
provide
poverty
relief and those
whose
primary
focus is the
pursuit
of other
objectives.
Second,
the economic crisis has
provided
some
strategic
lessons.
Perhaps
the
key
lesson
for
pensions
is the
importance
in
any
reform
of
explicitly
ask
ing
how risk should be shared.
With
pure
funded individual
accounts,
all
of the risk falls
on
the
worker,
and modifications
frequently
leave the
worker
bearing
a
large
fraction of
the risk.
Many
(including
us)
regard
this concentration of risk
as
undesirable. One
way
of
sharing
risk
more
widely
is
to
buttress individual
accounts
with
a
tax-financed noncontribu
tory
pension.
Third,
there
are
many
choices,
which widen
as
a
country's
economic and
administrative
capacity
grows.
To
illustrate,
we
briefly
describe
pension
sys
tem
options
for
a
middle-income
developing country
and for
an
advanced
country.
A
middle-income
developing
country
has
two
main
options
for its
first-tier
pension:
a
noncontributory
tax-financed
pension
(see
box
1),
with
or
without
an
affluence
test
(examples
include
Australia,
the
Netherlands,
New
Zealand,
South
Africa,
and
Chile);
or a
simple
contributory
PAYG
pension,
such
as a
flat-rate
pension
based
on
number of
years
of contributions
(such
as
the basic
state
pension
in the
United
Kingdom).
For the second tier, the options include a publicly organized earnings
related defined-benefit
pension,
or
possibly
a
notional
defined-contribution
pension;
or a
provident
fund defined-contribution
pension
(Malaysia,
Singa
pore),
inwhich the
design
of
any
tax
concessions
should
consider
the
extent
to
which
tax
benefits
are
regressive.
If
the
first
tier
includes
a
contributory
pension,
the second-tier
mechanisms
can
be
separate
from it
or
integrated.
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
15/24
14
ECONOMIA,
all
010
The third
tier
comprises
voluntary
defined-contribution
pensions
at
the
level of the
firm
or
the
individual;
any
tax
benefits
should be
designed
to
avoid
excessive
regressivity.
In
an
advanced
country,
the
options
for the first-tier
pension
are
either
a
contributory pension
aimed
at
poverty
relief,
with
a
large
array
of
possible
designs
(many
countries
use
this
type
of
system,
including
theUnited
King
dom);
or
a
noncontributory
tax-financed
pension,
with
an
affluence
test
(for
example,
Australia
and South
Africa)
or
without
(for
example,
the Nether
lands
and New
Zealand).
There
are
several
options
for the second
tier:
a
publicly organized
defined
benefit
pension,
which
may
be
integrated
with the
first-tier
contributory
pension
(the
United
States)
or
operated
separately
(France,
Germany,
and
Sweden);
a
notional defined-contribution
pension
system
(Sweden);
an
administratively cheap savings plan
with
access
to
annuities
(for
example,
the
Thrift
Savings
Plan
in
theUnited
States);
mandatory
occupational
funded
defined-benefit
pensions
(this
is the de facto
system
in
the
Netherlands);
or
funded defined-contribution
pensions
(Chile,
Sweden),
possibly
including
an
anti-poverty element (Mexico).14
As in the
case
of the
developing
country,
the
third tier
comprises
volun
tary
defined-contribution
pensions
at
the level of the
firm
or
the
individual;
any
tax
benefits should
be
designed
to
avoid
excessive
regressivity.
Finally,
for
an
effective
pension
system,
two
things
matters
above
all:
effective
government
and
output
growth.
These factors
are
simple
to
identify
but difficult
to
achieve.
An effective
government
will
manage
a
PAYG
sys
tem
responsibly,
and
itwill
create
themacroeconomic and
regulatory
stability
within which funded schemes can flourish. In contrast, ineffective govern
ments
are
prone
to
make
irresponsible
PAYG
promises
and
to
pursue
policies
leading
to
macroeconomic
instability.
Robust
output
growth,
by
relaxing
resource
constraints,
makes it easier
to
realize the
plans
of
both workers and
pensioners.
14.
The
system
in
the Netherlands is
now
moving
more
toward
a
defined-contribution
arrangement.
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
16/24
Comment
Eduardo
Engel:
This
paper
provides
a
masterful
primer
on
the economics of
pensions.1
Barr and
Diamond also
cover
the
shortcomings
of
arguments
for
privatizing pensions
in Latin
America,
as well as
problems
thatwere not
envisioned
at
the time of
reform.
I
organized
a
lecture
on
pensions
in
the
undergraduate
course
I
teach
at
Yale
on
economic
policy
in
Latin America
based
on
this
paper,
and itworked
very
well.
In
these
comments,
I
provide
additional evidence
illustrating
the
paper's
main
points
and discuss
issues
on
which I differ.
Among
the
latter,
I
argue
that
political
economy
considera
tions
may
temper
some
of the
paper's policy
recommendations.
Promise
A
pension
system
based
on
individual
accounts
as
part
of
a
fully
funded
defined-contribution
system
was
introduced
in
Chile
in
1981.
Argentina,
Bolivia, Colombia,
Costa
Rica,
theDominican
Republic,
El
Salvador,
Mexico,
Peru,
and
Uruguay
introduced similar
systems
in
the decades that
followed,
sometimes
as one
ofmany options and
in
other
cases as
themain
or
only
option
available
for
new
entrants to
the
labor force.2
Four main
arguments
for the Chilean
pension
reform
were
given
in
the
early
1980s.
First,
providing
well-defined
ownership rights
reduces the risk
of
opportunistic
behavior aimed
at
obtaining
a
larger
share
of
accumulated
funds and
current taxes
destined
to
finance
pensions.
Both advocates
and
detractors
of
theChilean
reform
provided ample
evidence
that the old
system
was
plagued
by
blatantly
unfair
and
corrupt
practices.3
Second,
competition
1.
I
thank Eduardo
Bitran,
Ronald
Fischer,
Eduardo
Lora,
and Claudio Raddatz
for
com
ments
and
suggestions.
2. Arenas and
Mesa-Lago
(2006).
3.
Arellano
(1985);
Pinera
(1991).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
17/24
16
ECONOMIA,
all 010
among
pension
providers
would lead
to
better
service and
higher
returns
on
savings,
translating
into
higher
pensions.
A
third
argument,
closely
related to
the first ne, was that the new system would foster sounder
public
finances
by
reducing
the
burden
on
the
government
budget
imposed
by
the
partially
unfunded
systems
in
place
at
the
time.
Jose
Pinera,
labor
minister at
the
time
of
the
Chilean
reform,
went
so
far
as
to
claim
that
the
fiscal
cost
of
the
reform
would
be
zero.4
Finally,
many
arguments
were
presented
linking
a
system
based
on
individual accounts to
the
development
and
deepening
of
financial
markets.
Experience
As is
so
often
the
case
with
economic
reforms,
the
benefits
are
oversold
and
a
number of
problems
are
not
anticipated.
The
transition
from
a
system
with
defined
benefits
to
one
with
defined
contributions
based
on
individual
accounts
evidently
has
a
cost
for
public
finances.
Taxes must
finance
pen
sions
for
those
belonging
to
the old
system
for
a
period lastingmany decades,
with
no
recourse
to
social
security
contributions
from
workers
entering
the
new
system.
Only
a
comparison
of
steady
states
that
ignores
the
transition
costs
involves
no
fiscal
cost.5
The
analysis by
Barr
and
Diamond
also
tempers
the
enthusiasm for
pen
sion
systems
based
on
individual
accounts
as
a
means
to
develop
financial
markets.
I
do
not
have
space
to
address
the
debate
on
whether
pension
reform
fosters
the
development
and
deepening
of
financial
markets.6
The
potential
is
there, especially
for
countries thatprivatized publicly held utili
ties
concomitantly
with
reforming pensions,
since
the
long-term
financing
needs of
privatized
utilities
match
the
pension
fund
managers'
demand
for
long-term
assets.
Nonetheless,
the
challenges
faced when
addressing
this
question
empirically
based
on
aggregate
evidence
are
large,
since
other
4.
20
respuestas
acerca
de
la
prevision
dio
ministro
Pinera,
El
Mercurio,
15
November
1980,
p.
Al.
5.
The
present
discounted
value
of
government
outlays
will
also
be
higher
with
reform
than
without
reform,
since
the
transition
from
the
partly
unfunded
defined-benefits
system
to
the
fully
funded
defined-contribution
system
with
individual accounts
is
the
converse
of
the
well
known
result
according
to
which
introducing
social
security
constitutes
a
Pareto
improvement
(Samuelson,
1958).
6.
See
Raddatz and
Schmukler
(2005)
for
a
good
summary
of the
main
issues and
references.
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
18/24
Nicholas
Barr
nd
Peter Diamond 17
reforms coincided
with
pension
reform
in
many
countries,
making
identifi
cation
close
to
impossible.7
A more promising route to study the impact of pension reformon financial
markets is
to
use
microeconomic data
to test
specific
channels for
potentially
beneficial
effects. Raddatz and Schmukler
provide
an
interesting example
along
these
lines.8
They
look
at
investment
strategies by
Chilean
pension
fund
managers
and
find that
theymainly
hold bank
deposits,
government
paper,
and short-term
assets;
that
they
tend
to
hold similar
portfolios
and fol
low
momentum
trading strategies;
and that
they
do
not
actively
trade the
assets
they
hold. All of
this
suggests
that
even
if
pension
funds
helped
create
markets for
long-term
instruments, as
they
most
likely
did,
they
did not con
tribute
to
making
these markets
particularly liquid.
Raddatz and Schmukler
conclude
that
the
observed investment
patterns
are
not
consistent with the
initial
expectations
that
pension
funds
would be
a
dynamic
force
stimulating
the overall
development
of
capital
markets,
especially
that
of
secondary
trad
ing
markets. 9
Competition
The
advantages expected
from
competition
did
not
materialize. As noted
by
Barr and
Diamond,
the economics
of information and behavioral economics
both
help
explain why.
For
example,
95
percent
of
account
holders of Chile's
private pension providers
do
not
know the administrative
charges they
pay,
despite
the
fact
that
they
receive
quarterly
statements
containing
these
charges
and that this is themain determinant of
differences in
returns
across
pension
funds.10
It
is
therefore
not
surprising
that
competition
among
providers
has
been
weak,
leading
to
average
annual
returns
on
capital
for these
companies
that
are
many
times those obtained
by
assets
facing
similar
risk
profiles
else
where
in
the
economy.
For
example,
the
average
annual
return
of
pension
administrators
in
1998-2003
was
53
percent,
which is four times
larger
than
normal returns.11
Despite
such
high
returns,
no new
firm
entered
the
industry
7. For
example,
orande
(1998) argues
that
hile's
high
growth eriodbeginning
n
1985
is
explained
by
increased
savings
that
resulted from
pension
reform.
8.
Raddatz and
Schmukler
(2005).
9.
Raddatz and
Schmukler
(2005,
p.
7).
10. Berstein
and Ruiz
(2004).
11.
Valdes-Prieto and
Marinovic
(2005).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
19/24
18
ECONOMIA,
all
010
between
1998 and
2010,
when the
system
was
reformed
to
introduce
more
competition
(see
below).
Interestingly,well-designed competition for thefield can be a good surro
gate
when
competition
in the
field
does
not
work
properly.12
Private
pension
markets
provide
an
excellent illustration
of this
powerful
idea.
In
February
2009,
theChilean
government
auctioned
the
individual
accounts
of all
work
ers
entering
the
labor market
over
a
two-year
period,
with the
accounts
being
awarded
to
the
pension provider
bidding
the lowest
fees.
If
an
incumbent
won
the auction
with
a
bid lower than
the fees
it
charged
existing
customers,
it
had
to
pass
on
the
reduction
in
charges
to
all
customers. Workers
assigned
to
the
winner are free tomove to another
provider
if
they
so desire.
The
winner of the auction offered
a
commission
16
percent
below
the low
est
commission
in themarket.
Furthermore,
the
winner
was
new
to
the indus
try,
the first
entrant in
more
than
a
decade.
Having
a
Demsetz
auction lowers
entry
costs
significantly,
since
new
firms
can
count
on a
sufficiently large
number
of clients
(approximately
7
percent
of
themarket
in theChilean
case)
to
operate
at
an
efficient
scale without
having
to
invest
in
marketing. Despite
the
fact that incumbent
pension providers
lobbied
strongly
against
the
auc
tion
and succeeded
in
watering
down the
extent to
which it reduced
entry
costs,
the
outcome
is
promising
and
may
lead
to
significant
fee
reductions in
years
to
come.
Administrative
osts
Arellano
was
one
of the first
to
point
out
the
high
administrative
costs
of the
Chilean
pension
system.13
In
general,
the
administrative
costs
of
private
investment
funds
in Latin America
are
high.
As of December
2002,
fees
as
a
percentage
of total contributions
were are
follows:
Argentina,
36.2
percent;
Chile,
15.0
percent;
Colombia,
14.0
percent;
El
Salvador,
12.5
percent;
Peru,
22.1
percent;
and
Uruguay,
13.5
percent.14
Bolivia is the
one
exception,
with
administrative
fees
of
only
4.8
percent
of total contributions.
Why
are
admin
istrative
costs
so
much lower in Bolivia?
Bolivia learned
from the
Chilean
experience and used aDemsetz auction to reduce administrative costs by two
thirds.
Instead
of
allowing
free
entry,
Bolivia
opted
for
two
pension providers
12.
Chadwick
(1859);
Demsetz
(1968).
13. Arellano
(1985).
See
also Diamond and Valdes-Prieto
(1994).
14.
Barr and
Diamond
(2008,
table
9.2,
p.
166).
This content downloaded from 192.188.53.215 on Thu, 17 Apr 2014 17:57:51 PMAll use subject toJSTOR Terms and Conditions
http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp8/11/2019 Diamond Barr 2010
20/24
Nicholas
Barr
nd
Peter
Diamond
19
selected via
competitive bidding
based
on
the
lowest
average
monthly
admin
istrative
charges.15
Pensions