Whether Social Security Funds Should be Invested Collectively or Through a System of Individual Accounts

September 6, 2001

Thank you for the opportunity to testify
before the Commission to Strengthen Social Security on the critical
issue of whether Social Security funds should be invested collectively
or through a system of individual accounts. This question is of course
related to the larger question of the extent to which Social Security
should retain its character as a program paying a “defined benefit”
that is protected from risks of market fluctuations and inflation, or
whether individuals should be exposed to increased risk from financial
market fluctuations.

Almost everyone accepts the need for a
multi-tiered retirement income system, including a minimum floor, an
income-related defined benefit, some form of tax-advantaged and/or
mandatory retirement savings, and voluntary savings for retirement.
And most of the advanced industrial countries have adopted a
multi-tiered approach. The questions we must all weigh are: what mix
of these tiers is appropriate? What is affordable? How much leeway do
we have for change given past policy choices? And if we do decide on
policy change, over what time frame should it be imposed?

I will focus my testimony on four issues.
What are the relative advantages of collective and individual
investment? How can we minimize the political risks associated with
collective investment? What mechanisms could mitigate problems
associated with individual accounts? Are the serious problems
associated with a partial opt-out from Social Security into individual
accounts solvable?

I believe that there is a stronger case
for some form of collective investment of trust fund surpluses than
for a move to individual accounts as a way of addressing Social
Security’s financing problems. But I also believe that the
Commission must move beyond the ideological debate for or against
privatization to consider the details of particular proposals. There
are better and worse ways of organizing and implementing both
collective investment and individual accounts. Whatever you decide to
recommend, the details count.

1. Collective investment of
Social Security surpluses has important advantages over
individual account plans.

Collective investment of Social Security
funds in a broader range of instruments than government-guaranteed
securities has several important advantages over the status quo. The
most important is that it would allow a greater return on the trust
funds over the long term through the higher returns (with higher
risks) associated with equity investment.

Just as important as its advantages over
the status quo, collective investment of Social Security trust funds
has three important advantages over a system of individual accounts,
with lower risks and costs. First, by pooling investments and keeping
transaction, marketing, and reporting costs to a minimum, collective
investments can lower the costs of investing funds dramatically and
produce higher net returns than individual retirement savings
accounts. How much administrative costs reduce the ultimate return on
individual accounts depends heavily on the particulars of how that
system is structured as well as the mix of assets that they invest in,
as will be discussed further below. But an individual account system
is inherently costly to administer, especially for small employers and
firms with large labor turnover. A recent study by Estelle James,
James Smalhout and Dimitri Vittas estimates that administrative and
marketing costs in decentralized individual account systems where
pension funds are “retailed” to individual consumers are likely to
lower eventual pensions by fifteen to thirty percent. Those where
there are more constraints on individual choice (e.g., a limited
number of funds are offered) or where administrative functions are
centralized (as in Sweden) lower pension accumulations by ten percent
or less. Because many of the costs associated with maintaining
individual accounts are fixed costs—notably record-keeping and
communication with shareholders those charges are likely to hit small
accounts held by persons working at low wages particularly hard unless
fees are regulated in a way that protects small accounts (for example,
capping fees at a total percentage of annual account balances while
barring front-loaded fees, as in the new U.K. stakeholder

A second advantage that collective trust
fund investment has over individual accounts is that it lowers
information costs for consumers, as the costs of evaluating
alternative investments are spread over huge groups. There are also
distributive issues here as well: low-wage workers are likely to bear
the highest information costs in seeking information about investment
options, because they are less likely to be targeted in the marketing
efforts of fund managers who believe that they can make higher profits
concentrating their marketing efforts on those with higher incomes and
higher fund balances. Moreover, those with lower incomes will face a
lower return on any information-gathering efforts because their fund
balances are lower.

A third important advantage that allowing
Social Security trust fund equity investments has over individual
accounts is that it is doing so would not undermine or erode the
defined benefit structure of Social Security, which provides a
predictable retirement income that spreads the risks of fluctuating
asset values and annuity prices across the population and over
generations. The leading work on this topic has been done by my
Brookings colleague Gary Burtless, who has estimated the initial
replacement rates for successive cohorts of hypothetical workers over
most of the past century who invested a constant six percent of their
salaries in a broad stock market index and then converted the fund
value to a level-rate annuity upon retirement at age 62. Burtless
found that the initial replacement rates for workers ranged between 20
and 110 percent, with an average rate of 53 percent. This difference
of more than 5 to 1 in replacement rates is a fatal flaw for a program
designed to ensure a basic income level. These variations can also be
seen in very recent history: individuals under an individual account
system who retire today rather than eighteen months ago would be hit
by the double whammy of a falling stock market and higher annuity
prices resulting from lower interest rates.

Workers in an individual account system
would also be exposed to substantially varying degrees of inflation
risk after retirement. Participation in an inflation protected-defined
benefit plan eliminates the risk that whole cohorts of individuals who
have played by the rules and done the right thing by saving may be
left with an inadequate retirement income because of market conditions
over which they have no control.

2. The risks of political
interference with collective investment can be minimized through
proper insulation mechanisms.

Critics of broadening the investment
options for the Social Security trust funds fear that such funds would
inevitably be subject to political interference, and that they would
be so big that they would disrupt private capital markets. However,
both of these concerns can be addressed in designing a set of
safeguards for Social Security investment funds drawing on both
domestic and foreign experience. I will focus here on what I believe
are the most appropriate models for the United States relating to
three issues: insulation mechanisms, fund size, and corporate

&nbsp Many state
pension plans in the United States, as well as the Federal Thrift
Savings Plan and the Canada Pension Plan, have managed to achieve
excellent financial returns, while keeping costs low and avoiding
political interference in investment decisions. Although the
governance structure of these plans varies substantially, their
experience suggests several “best design practices” that are likely to
minimize political interference, notably:

  1. Give the investment funds explicit
    organizational mandates to maximize return on contributors’
    investment consistent with a prudent approach to risk rather than
    including social considerations in investment.


  2. Have independent boards of trustees for
    the funds, serving long terms. Expertise in financial services
    should be an explicit requirement for appointment to the boards.
    Appointment of politicians on a partisan or regional basis should
    be avoided.


  3. Have the investment fund trustees

    contract out portfolio management to professional fund managers on
    a competitive basis. Contracting out allows the Canada Pension Plan
    Investment Board to manage more than $C8 billion (with growth to
    $C130 billion planned by 2011) with a staff that currently totals
    about 15 persons.


  4. Invest funds primarily in broad,
    indexed investments. This need not preclude more active investment
    policies entirely, however. The Canada Pension Plan Investment
    Board, for example, has recently begun to implement a policy to
    actively invest up to half of its Canadian equity assets. It has
    also begun working in partnership with merchant banks and other
    pension funds to take advantage of venture capital opportunities
    while spreading risks.

The four mechanisms outlined above are not
particularly difficult to design and maintain. With tens of millions
of current and future Social Security beneficiaries looking on to make
sure that Congress does not meddle with “their” retirement futures, it
is almost certain that Congress would maintain a hands-off

Fund Size: Once concerns
about political interference have been addressed, worries about the
size of public investment funds can be addressed in two ways. One is
simply to limit the size of any single Social Security fund, creating
new funds that are separately (and also privately) managed once a
public fund reaches a certain size. Multiple funds are already used in
Sweden, which has six separate funds to manage accumulated surpluses
in Sweden’s pay-as-you-go public pension system, and a seventh
to manage the funds of workers who do not designate a fund choice in
the new individual accounts tier of the pension system. The government
has put explicit limits on how much individual funds, and all the
funds collectively, can own of a single firm and of the total

But how big a fund is too big? One simple
standard would be to limit the size of any one Social Security
investment fund to roughly the size of the largest private investment
funds—a position currently held by Fidelity with 3.3 percent of
domestic equities, followed by Barclay’s Global Investors with
2.1 percent, and State Street Global Advisors with 1.6 percent. Once a
Social Security investment fund reached this size, it would not
receive any new investment funds from Social Security surpluses, and a
new investment fund or funds, again privately-managed, would be set up
to receive new funds.

A second and somewhat more convoluted
mechanism for limiting Social Security investment fund size would
involve tracking employee contributions into tax-favored 401(k) plans.
Social Security trust fund surpluses would be distributed among fund
managers in proportion to 401(k) contributions—or at least to
those fund managers who agreed to provide a hefty discount to the
Social Security system in recognition of the vastly lower costs of
administering one large account than tens of thousands of individual
401(k) accounts. The logistical difficulties and financial costs of
setting up and administering such a system are not inconsiderable, but
they are minute in comparison to the costs of setting up roughly one
hundred million individual accounts, many of which would receive very
small and irregular contributions from low-earners. The size-limited
Social Security investment funds and “401(k) mirror” options outlined
above could also be combined, each receiving half of Social Security

Corporate Governance:
Another concern that has been raised about collective investment of
Social Security trust fund surpluses is that they will be the object
of repeated initiatives aimed at affecting corporate governance and
the investment practices of corporations. For example, liberal members
of Congress might try to require fund managers to support shareholder
resolutions forbidding a company from investing in Myanmar or in
tobacco stocks, while social conservatives might try to get
publicly-owned biotechnology companies to refrain from engaging in
stem cell research. The record of some state public employee
retirement funds in the 1980s has increased concern in the corporate
sector about such risks. However, a recent study of state retirement
funds by Alicia Munnell and Annika Sunden suggests both that there has
been a move away from such practices in recent years by state
retirement systems and that such funds have earned returns that
compare well to those of private retirement funds. To prevent such
initiatives from occurring, the simplest solution is probably to
establish in legislation that shares held by Social Security
investment funds will not be voted by fund managers.

Is there a guarantee that the steps
suggested above will completely eliminate all risk of political
interference in how those funds are invested? Of course not. But
neither can proponents of individual accounts guarantee that all
interference (for example, domestic investment requirements) would be
avoided with private accounts that provide compulsory or
tax-advantaged retirement savings. However, the steps suggested above
should keep the risks of political interference very
low—certainly low enough that the overall return on such funds
is likely to be higher than for any plausible system of individual
accounts, once the administrative and marketing costs of the latter
are taken into account.

3. Individual accounts pose a
more complex set of design issues than collective investment of
Social Security trust funds, but once again there are better and
worse options.

While collective investment funds present
important and complicated issues of program design, designing an
effective individual account system is far more complex. I assume that
later speakers will focus on the many difficult issues related to the
benefit structure in an individual account system. These issues

  1. the extent to which individuals are
    required to convert their account balances into some form of income
    stream at retirement, and the conditions under which that
    conversion takes place;

  2. the problems that a system of
    individual accounts creates for spouse’s benefits, and for
    survivors and disability insurance; and

  3. (3) the higher annuity prices charged
    to women given their longer life expectancy. Equally important is
    the issue of how to finance a transition from the current largely
    pay-as-you-go system to a system that includes fully-funded
    individual accounts, the so-called “double payment problem.”

All of these issues deserve the
Commission’s full attention.

Even leaving these aside, individual
account plans pose a formidable list of design choices. In addition to
the question of whether to have individual accounts as an opt-out from
the defined benefit plan or as a mandatory component, which will be
the focus of the next section of my testimony, there are important
issues of administering accounts and investment practices. I will
focus briefly on each in turn, again seeking to draw out potential
“best practices.”

Account Administration: Administration of individual account systems differs on
several dimensions. One is whether they are administered by employers,
by government, or by some combination of the two (as in the U.K.). A
second is whether government regulates entry and/or fees charged by
pension providers. These governing choices have important implications
both for the number of pension options available (in the aggregate and
to subsets of the population) and for the costs of pension

Many options are possible. Sweden’s
new individual account tier, for example, features a highly
centralized system of account administration. Funds flow into the
Treasury and out through a specialized state agency that moves
resources into and out of investment funds, with the individual fund
managers knowing only the amount of the funds to be moved rather than
the identity of their owners. The advantages and disadvantages of this
approach are clear. It facilitates maximum fund choice (regarding
entry and switching of funds as well as distribution of assets among
multiple funds) at minimum cost. When the program debuted in the fall
of 2000, Swedes could choose from approximately 450 funds. The Swedish
system also minimizes the additional paperwork burden for employers,
who can follow existing procedures for submitting payroll taxes and do
not need to get involved in administering fund choices and payments to
multiple funds by their employees. Thus it almost certainly weakens
opposition from employers (and especially small employers) to
participation in such a system. Central administration of funds also
makes it easier to negotiate reductions in management fees by fund
providers, but it contributes to a very long lag time in crediting of
individual pension accounts.

Other countries have made very different
choices. Bolivia, for example, created a duopoly of pension providers
when it created its privatized pension system, minimizing choice but
also lowering administrative costs. Australia, with a decentralized
individual account system run by individual employers or on an
industry basis (fund management functions are generally contracted
out), has a high cost system where the number of options available to
employees differs greatly—but is usually quite limited.

Which set of options might be most
appropriate for the United States if there is a move toward individual
accounts in this country? Given the large size of the U.S. economy,
limiting entry to a small number of competitors in an individual
account system does not seem either politically sustainable or
desirable. The five fund options offered by the Thrifty Savings Plan,
for example, would be enormous if expanded to cover all persons
currently paying Social Security taxes. On the other hand, the
evidence presented by James et al and others suggests that the cost
savings from centralized administration can be substantial. Overall,
the best option is probably a variant of the Swedish approach, in
which record-keeping and administrative functions are
centralized—thus lowering the cost of fund administration—and a
large number of fund options are permitted. This method would also
lower costs to (and probably political opposition from) employers,
although it does have the disadvantage of delaying the movement of
funds into individual accounts.

A key issue in
the design of individual account systems is whether those accounts
should be required to limit risk by holding several different assets.
Issues relating to diversification requirements are complex, and have
been politically contentious in several countries. Should small
business people be allowed to invest most or all of their individual
retirement savings accounts in their businesses, for example, which
may boost their long-term income if the business succeeds but leave
them with nothing if the business fails? What about investing in the
house that they occupy, which might allow them to pay off a mortgage
faster and thus enter retirement with a lower drain on their income?
Or concentrating most of their individual retirement savings in the
stocks or bonds of their employers, if they work for a relatively
large firm? Or putting all of their savings into a low-yielding bank
account, which is insured against loss of principal by government but
provides very limited opportunities for growth? All of these issues
have surfaced in other countries. Note that Congress has successively
broadened the conditions under which individuals are allowed to borrow
against tax-advantaged retirement savings in the United States.
Clearly such pressures would also be felt in an individual account
system that was part of Social Security, as individuals asked why they
could borrow against one form of retirement savings but not

There are no simple answers to these
questions, but there is a principle that can help: the more that
income from an individual investment account is expected to supply a
“basic” level of income to a future retiree rather than supplemental
income above a basic minimum or replacement rate, the stronger the
case for investment diversification requirements and for prohibitions
against borrowing. Certainly any type of account that is expected to
supplant or offset current levels of Social Security income fits
within the category of accounts where diversification requirements and
borrowing prohibitions are essential.

4. Partial opt-outs from Social
Security into individual accounts are not a compromise between
the status quo and privatization but rather the worst of both

The mandate that President Bush gave to
this commission was to develop a proposal that would permitworkers to
shift some of their payroll taxes to individual retirement investment
accounts but not require anyone to do so. Permitting rather than
requiring sounds great. What could be more American? No one would be
forced to do anything, but everyone would enjoy increased choice. But
experience from abroad suggests that Social Security opt-outs pose
some very serious problems.

Several advanced industrialized countries
have adopted or, at least considered, mandatory savings programs for
all workers. However, only the United Kingdom and (to a very limited
degree) Japan use an opt-out approach for privatized pensions. In the
U.K., an opt-out system emerged not as a planned outcome but as a
by-product of the fact that earnings-related pensions were not adopted
until the 1970s, after a private system of occupational pensions was
already highly developed.

The British experience with opt-out public
pensions offers a number of cautionary lessons about the perils of
this approach. Social Security opt-outs have all of the disadvantages
associated with mandatory saving through individual
accounts—notably high administrative costs and increased risk
across individuals and cohorts. But opt-outs also pose an additional
set of problems not found in mandatory individual accounts. One
problem with an opt-out reflects the fact that the present system
offers higher returns on contributions of low-wage workers to help
provide them with a decent retirement income. If an opt-out were
available, higher-income workers would be more likely to opt-out,
seriously undermining the current Social Security system’s financing.
Conversely, opting out of Social Security wouldn’t make sense for many
low-wage workers. But they’re likely to be the least sophisticated
investors, so they might opt out when they would be better off staying
in the current system.

A second major opt-out problem is the
differing returns offered by contributions to an individual retirement
investment account during a worker’s life. The earlier in one’s
career these contributions are made, the likelier they are to generate
higher pension value. Conversely, contributions to Social Security are
indexed for wage growth. Contributions of equal real value will
provide relatively equal returns regardless of when they are made. As
a result, many opted-out workers will find it advantageous to opt back
into a state-defined benefit plan at some point. However, British
experience suggests that it is unclear where that point is, given
uncertainties about future returns on investments and prices for
annuities. Given the complex and confusing choices faced by British
workers, it is no wonder that when the U.K.’s Financial Services
Authority recently prepared a decision tree to help individuals make
pension choices, almost all paths led to the same end point: consider
getting professional financial advice.

In the U.K., incentives to opt back into
the state pension have been addressed through age-related rebates for
National Insurance contributions: older workers get higher rebates as
an incentive to continue to opt out of state pensions. These
age-related rebates make the British system complicated and expensive
to administer. Age-related rebates make even less sense in the U.S.
system, where there is a closer linkage between contributions and
benefits. The absence of general revenue financing in Social Security
means that more generous Social Security contribution rebates for
older workers would undermine the financing of Social Security as a
whole. An alternative solution would be to require young workers to
make a one-time, irrevocable choice to opt-out or opt-in from Social
Security. But this option is almost certainly not appropriate given
unforeseen changes in earning potential, and it is even less likely to
be politically sustainable.

Problems concerning who should opt out and
when to opt back in raise a third critical problem with opt-outs: To
whom could workers turn for impartial advice on whether opt-outs were
an appropriate choice for them? Pension fund providers and many
financial advisers have a vested interest in selling their products.
And the Social Security Administration would likely resist such a role
under intense pressure from the administration, Wall Street and the
pension industry not to weaken the message that privatization is a
good thing.

Unfortunately, the outcome in the United
States could mirror the British experience: workers may respond to
high pressure sales practices by pension providers who “mis-sell”
pension products. In the U.K., mis-selling in the late 1980s is
estimated to have cost more than 15 billion dollars. Were this to
happen in the U.S., litigation would surely follow. A U.S. pension
mis-selling scandal could be the biggest boon to trial lawyers since
the Ford Pinto.

In short, opt-out plans for Social
Security impose too much additional complexity in an already very
complex pension system. Potential implementation problems could
undermine the legitimacy of both Social Security and the private
pension industry. While universal mandatory savings plans have merit,
they should be considered as a supplement to rather than an opt-out
from Social Security.


In closing, I would like to leave you with
five thoughts, most of which reflect my political scientist’s
orientation to issues of pension program design, implementation and
political viability.

First, investment of Social Security funds
in a broader range of financial instruments, whether done collectively
or through individual accounts, is not a panacea that will solve all
of Social Security’s long range funding problems, which flow
from the demographic bulge of the Baby Boom’s retirement and
more fundamentally from longer life expectancies. To use a nutritional
metaphor, collective or individual investments in equities are not a
free lunch that will cause this financial problem to disappear, but
more like a healthy snack that can help to make it more

Second, any proposal that the Commission
recommends should have as a central objective strengthening the long
term financial viability of the current defined benefit system. Social
Security is by far the most popular federal program, and the most
successful in reducing poverty. Defined benefit replacement rates are
very low in comparison to most other advanced industrial countries.
Any plan that is perceived by the public as weakening the ability of
Social Security to pay currently promised defined benefits—and
almost all opt-out plans fit in this category—will be political
non-starters, moving the debate on Social Security reform backward
rather than forward. Any plan that does not reduce the projected
long-term deficit is no plan at all. The Bush administration’s
Social Security proposals suffer from a widespread perception that it
is a mechanism to send billions of taxpayers’ money to Wall
Street for the administration of individual accounts. Unless this
perception is credibly addressed, policy stalemate and heightened
political division are the likely outcomes. Strengthening Social
Security is likely to require going beyond the Commission’s
mandate to design a viable opt-out system: as talented as this group
is, being given a mandate to make two plus two equal five
doesn’t mean that it can be done.

Third, getting program design right is as
important as the choice between individual accounts and collective
investment. For collective investment, it is imperative that a strong
set of protections be put in place to prevent political interference
in investment decisions. As noted above, I believe that both foreign
experience and experience with state retirement systems and the
Federal Employee Retirement System suggest that these risks are
manageable. Individual accounts pose a far more complex set of design
issues, including how to prevent the erosion of accounts by
administrative expenses, minimizing risks posed by market fluctuations
in asset values and annuity prices, whether and how to require
conversion of account balances into income streams, and whether or not
to allow inheritability of fund balances if a worker dies before
retirement or shortly thereafter. Again, there are better and worse
solutions to these issues. I believe that any move to individual
accounts should focus on lowering administrative and marketing costs
and should be integrated as much as possible with the current program
in terms of conversion of fund balances into retirement income

Fourth, even if the commission does
recommend a system of individual accounts, some element of collective
investment is still desirable, and probably necessary. It will still
be desirable to increase the returns on the surpluses in the Social
Security trust funds, and it will probably be necessary to have
collective funds to deal with individuals who do not make a choice
among funds and to deal with transitional periods between when funds
are collected and when they can be attributed to individual

Finally, although it is important to
preserve the current, relatively modest level of Social Security
defined benefits, that benefit is clearly insufficient to provide a
retirement income that most Americans will find adequate. The federal
government should do more to encourage individual savings for
retirement by its citizens. This can be done in a number of ways. A
broader public education campaign modeled after the “Choose to Save”
campaign by the American Savings Education Council and Employee
Benefit Research Institute could be useful in increasing public
awareness of the importance of retirement savings. More direct steps
to encourage retirement savings among all income groups should also be
undertaken. The 2001 tax bill makes important strides in this
direction for upper-income workers, but more needs to be done to help
low-income workers who cannot save enough on their own by providing
tax subsidies or direct subsidies to these workers. Strengthening
Social Security should be the keystone of a broader effort to improve
the retirement income system in the United States.