Defined Contributions from Workers, Guaranteed Benefits for Bankers:
The World Bank’s Approach to Social Security Reform
By Dean Baker and Debayani Kar [1]
July 16, 2002
Executive Summary
In the last decade the World Bank
has actively promoted the partial or complete replacement of public Social
Security systems with systems of individual accounts. While proponents of such
accounts had originally hoped that they would boost growth by increasing
national saving, the evidence to date has convinced even most advocates of
individual accounts that the net effect on national saving will be minimal.
However, the increase in the
government deficit, due to the loss of Social Security tax revenues during a
transition period, can lead to serious financial problems. In the case of
Argentina, the current budget crisis can be attributed largely to the decision
to privatize its Social Security system. The lost tax revenue, plus the interest
resulting from the additional incurred expenditure, exceeded its central
government budget deficit in 2001. In other words, if Argentina had not
privatized it Social Security system in 1994, and done everything else exactly
the same, it would have run a budget surplus in 2001.
This paper compares the
administrative costs associated with individual accounts, measured as a share of
contributions to the system, with the costs of operating an efficient public
Social Security system like the one in the United States.
Among the findings:
1) According to data from the
World Bank, the administrative cost of running privatized systems of individual
accounts is between ten and fifty times as much as the administrative cost of
running the public Social Security system in the United States. These additional
fees are direct transfers from workers’ retirement income to the financial
sector.
2) According to data from the
World Bank, the cost of the running the public agency that supervises the
operation of a system of individual accounts (the equivalent of a Securities and
Exchange Commission for these accounts), is between 62 percent and 400 percent
of the administrative cost of running the entire Social Security system in the
United States. In most countries, the cost of running this oversight body is far
greater than the cost of actually running the whole Social Security system in
the United States.
3) The cost of converting funds
accumulated in individual accounts into an annuity that provides a lifetime
stream of earnings is between 11 and 22 times the cost of operating the Social
Security system in the United States. These fees are direct transfers from
workers’ retirement income to the financial sector.
4) Proponents of individual
accounts have failed to consider the opportunity cost to workers, in the form of
the time needed to oversee their accounts. If the time required to manage these
accounts is equal to half an hour per year, the opportunity costs would be
between 55 percent and 280 percent of the administrative costs of the Social
Security system in the United States.
Introduction
Over
the last decade, more than a dozen countries in Latin America and Central and
Eastern Europe have partially or completely replaced public defined benefit
pension systems with defined contribution systems managed by private financial
institutions. Many other nations are considering partial or complete
privatization of their pension systems. The World Bank has been a major catalyst
for this shift, providing loans and technical support. The track record of these
reforms to date is not promising. The new programs have incurred substantial
administrative costs, which come directly out of workers’ retirement income,
and accrue largely to the financial industry. Even in an optimistic scenario,
these privatized systems will cost several hundred percent more to administer
than a typical public system.
The
shift from a public system to a system of private accounts also implies a loss
of tax revenue, as money that had been paid into the public system is diverted
into private accounts. The size of this loss will depend on how large a portion
of current tax revenue is diverted, but it can be substantial. For example, in
the case of Chile the revenue loss was about 8 percent of GDP in the years
immediately following the reform (Acuna and Iglesias 2001). In developing
nations, with limited ability to collect taxes and poor credit ratings, a
revenue loss of this magnitude can have serious consequences. It can force major
cuts in essential public services and/or contribute to the sort of financial
crisis recently experienced by Argentina. This is a serious risk that appears to
have been underestimated by the World Bank in its advocacy of privatization.
Accounting Illusions
Before
assessing the administrative costs of private accounts, it is worth briefly
dispelling an illusion about the potential for higher returns in private
accounts relative to a traditional pay-as-you-go defined benefit system.
Typically, stocks have offered a higher rate of return than the government bonds
that are usually held as assets by government run programs. For example, in the
United States, annual returns on stocks have averaged 4.0 percentage points
higher than returns on government bonds. This difference is often seen as a way
to increase the rate of return to individual workers—by investing their money
in stocks through individual accounts, as opposed to government bonds held by a
central fund.
While this switch may lead to
higher returns, it is important to recognize that the additional returns are
simply transfers from elsewhere, not new wealth for the nation as a whole.
Unless the switch to private accounts significantly increased savings, a
possibility that few economists regard as likely, then the nation as a whole is
no wealthier as a result of the switch to private accounts. This means that if
the returns to retirees are higher, then someone else must be getting less.
There are two obvious sources
for the additional returns from private accounts. The first is the increased
interest payments that the government is likely to incur, as a result of the
fact that money that had been designated to purchase government bonds is instead
being used to buy stocks. In order to attract alternative purchasers on
government bonds, it will be necessary for the government to pay a higher rate
of interest on all of its bonds. Since the switch to private accounts raises the
amount of interest that the government pays on its debt, it can be viewed as a
transfer from general government revenue to the private accounts.
The
second source of the additional returns on the individual accounts is a lower
return on stocks in the period after the creation of private accounts. As more
money flows into stocks, the returns to future purchasers of stocks will be
lower. This will be due to the fact that the inflow of new funds can be expected
to push up the ratio of stock prices to earnings, and therefore reduce the ratio
of the dividend payout to the share price.[2]
Since the switch to private accounts will lower the future returns to other
holders of stock, it can be viewed as a tax increase on the holding of stock.
If
it is considered desirable to raise the returns to retirees on the money they
pay into a national Social Security system, then this can be accomplished
without incurring the administrative costs associated with administering private
accounts. Instead of incurring additional interest costs on its outstanding
debt, this money could just be paid directly from general revenue into the
Social Security system. From the standpoint of the national budget, there is no
difference if the government incurs higher costs due to higher interest rates on
its debt or due to additional payments to the Social Security system.
Similarly,
it is possible to raise the tax rate on income from stocks as an alternative to
depressing returns through the creation of private accounts. Holders of stock
will end up no worse in this scenario, since their return—net of taxes—will
be the same regardless of whether yields fall due to a rise in share prices, or
whether after-tax returns are lowered due to higher tax rates.[3]
By
tapping these sources of revenue to increase the rate of return to the Social
Security system, the drain on the government and holders of stock would be
exactly the same as if it had switched to a system of private accounts.[4] The only difference would
be the savings in administrative costs.
The Costs of Private Accounts
The
experience of developing nations in administering national systems of private
accounts, combined with evidence on the costs of defined contribution pension
plans in the United States, clearly shows that even the best-managed systems of
private accounts will cost several times as much to administer as a well-run
public system.[5]
These costs take four forms:
1)
the direct administrative fees charged for managing and maintaining the
accounts,
2)
the cost of purchasing annuities after retirement, for workers who want
to be assured of a lifetime income flow,
3)
the costs of maintaining an oversight agency to ensure the proper
management of these accounts, and
4)
the time required by workers to oversee their own accounts.
There is a significant literature
devoted to the first two types of costs, which are the largest costs associated
with private accounts. The latter two costs have been less widely researched,
but each one, individually, could be as large as the cost of operating an
efficient public sector system.
Direct Administrative Fees
The direct
administrative costs of private accounts vary substantially across nations. The
table below shows the range of costs (measured as a percentage of annual
contributions) for a representative group of nations. It also shows the
administrative costs of the Social Security system in the United States, which
is taken as a model of an efficiently run public system. While some nations have
been more successful than others in restraining the costs of private accounts,
even Bolivia, the lowest cost country, still pays fees that (measured as a share
of contributions) are more than ten times the expense of running the entire
United States system.
Table 1—Administrative Costs as a Percent of
Annual Contributions
Net Fees/Contributions[6]
Argentina
23.0%
Bolivia
4.8%
Columbia
14.1%
Chile
15.6%
El Salvador
19.0%
Mexico
22.1%
Uruguay
14.3%
United States
0.5%
Source:
James, Smalhout, and Vittas 1999, table1; and Social Security Trustees Report
2000, table IV.A.1.
It is
important to recognize that the fees for the private systems shown in the table
understate the actual size of the transfers from workers to the financial sector
each year, because they exclude brokerage fees and commissions which are charged
to the firms that manage the accounts (e.g. spreads on stock trades). These fees
are deducted directly from the returns on the accounts. While such fees are
likely to be small relative to the fees charged for managing the accounts, they
are likely to still be substantial relative to the cost of administering a
well-run public system, especially for actively traded funds. For example, in
the case of Argentina, if these trading fees were equal to just 0.05 percent of
the value of the assets in the accounts, it would raise the administrative costs
by an amount equal to approximately 1.0 percent of annual contributions. This is
more than twice the entire cost of administering the United States system. A
full measure of the drain on workers’ savings, and the economic resources used
to maintain a system of private accounts, should include these trading fees.
Annuities
The second well-documented
cost associated with systems of individual accounts is the cost of issuing
annuities for workers at the point that they start collecting benefits. There
has been considerable research on the cost of annuities in the United States,
which generally finds that workers must accept a benefit that is 15-20 percent
below an actuarially fair payment (Mitchell et al 1997). A significant portion
of this payment, usually estimated at approximately 10 percentage points, is
attributable to adverse selection. This is a result of the fact that in a system
in which annuities are voluntary, the subset of people who opt to buy an annuity
are likely to be longer-lived than the average for the population as a whole. To
compensate for the longer average lifespan of the people who buy annuities,
insurance companies pay lower annual benefits than if annuity buyers were
typical of the population as a whole.
But the loss that workers incur
due to adverse selection—while it does mean that the system is functioning
more poorly in meeting the goal of providing a secure retirement income—is a
transfer between groups of workers. By contrast, the fees charged by insurance
companies to issue annuities are direct transfers from workers to the financial
industry. These fees correspond to the use of real resources (labor and capital)
to research and administer the annuity system. In a traditional defined benefit
system, in which the old-age benefits take the form of an annuity (adjusted for
inflation, in most cases), these costs are unnecessary. Therefore the 5-10
percent of savings, which are assessed as a fee when an annuity is issued and
are attributable to actual costs incurred by insurers, can be added to the
expense of operating systems of private accounts.
Costs of Supervisory Agencies
The third cost of maintaining a
system of private accounts is the cost of operating the supervisory agency that
oversees the accounts. This is the agency that must protect workers against
fraud or bad management practices, and prevent workers from holding assets that
are excessively risky. The cost of running these agencies has proven to be
relatively small compared to the direct administrative expenses of individual
accounts, but it is not small relative to the cost of a well-run public sector
system. The table below shows the cost of operating the supervisory agencies,
relative to annual contributions. The total administrative cost of the United
States Social Security system is included at the bottom to provide a basis of
comparison. While these costs are far lower than the direct administrative fees
charged to workers, they are still quite large relative to the entire cost of
administering the Social Security system in the United States. The
administrative costs of operating the oversight system in Chile, the most
efficient of the group, measured relative to contributions, is 62 percent of the
entire cost of administering the Social Security system in the United States.
Bolivia, the country with the most inefficient system, pays four times as much
to operate its oversight system, as the United States pays to administer its
whole Social Security system.
Table
2—Expense of Supervisory Agencies for Overseeing Individual Accounts Compared
With Total Administrative Costs in the United States
Annual Costs/Contributions[7]
Argentina
0.36%
Bolivia
1.80%
Chile
0.28%
Mexico
0.95%
Peru
1.23%
U.S. (total administrative cost)
0.45%
Source:
Demarco and Rofman 1998, table3; and Social Security Trustees Report 2000, table
IV.A.1
The Opportunity Cost of Workers' Time
A significant cost associated with
individual accounts, which has been largely neglected in the literature on the
issue, is the opportunity cost of the time required by workers to manage their
accounts. Since this issue has been almost completely neglected by the World
Bank and other proponents of individual accounts, there is little basis for
estimating this cost. However, even if the time involved in managing these
accounts is very limited, it would still be substantial compared to the
administrative costs of a well-run defined benefit system. For example, if
a worker spent half an hour a year overseeing his or her account on
average, this would be equal to 0.25 percent of the annual contributions in a
large scale privatized system and more than 1.0 percent of the annual
contributions in a partial privatization plan—similar to ones often suggested
for the United States.[8]
It is surprising that the
opportunity costs of workers’ time have been largely ignored in discussing the
relative merits of individual accounts and defined benefit systems. Defined
benefit systems generally make no demands on workers’ time between the period
when they first register and the point at which they arrange to start receiving
benefits. Therefore the time that workers spend managing individual
accounts—and therefore not caring for their children, going to school, or
engaging in some type of leisure activity —should be viewed as an additional
cost of individual accounts. Ignoring this cost would be comparable to
advocating a new type of car without considering the time that drivers had to
spend maintaining it.
One piece of evidence suggesting
that the time involved in overseeing these accounts may be substantial is the
recent decision of the Chilean government to start offering “benefits
awareness” courses in school, so that workers will be better prepared to
manage their accounts. A recent World Bank study recommended that the other
countries in Latin America with individual account systems follow Chile's
example (Devesa-Carpio and Vidal Melia 2001, p 26). This means that students
will forego time spent learning math, language, or other skills in order to
learn about the country's public pension system. While learning these financial
skills may offer students other benefits, it is not obvious that it is the best
use of their time.
The Total Costs of Individual Accounts
The discussion above
indicates that systems of individual accounts impose a substantially greater
economic drain on society than a well-run defined benefit system. In other
words, they divert resources, in the form of labor and capital, which could be
used productively elsewhere. Most of these resources, such as management and
annuity fees, are income to the financial sector. Table 3 summarizes the costs
of operating a defined contribution system measured as a share of contributions.
This table shows that even when using the low cost estimates, it is far more
expensive to run a defined contribution pension system than a defined benefit
plan. The low cost estimate is approximately 2300 percent of the cost of
operating the defined benefit system in the United States. The high cost
estimate implies that the operational expenses of a defined contribution system
is approximately 70 times that of the United States. These expenses imply a
serious drain of resources. If a mature system has contributions equal to 6
percent of GDP, the low cost estimate implies that the amount of waste, compared
with the cost of running a defined benefit system, is equal to approximately 0.6
percent of GDP annually. This would be equivalent to approximately $65 billion a
year in the United States at present. The high cost estimate would imply a level
of waste equal to 2.1 percent of GDP, which would be the equivalent of $230
billion a year in the United States.
Table 3—Operating
Costs of Defined Contribution Pension Systems as a Share of Contributions
Administrative
Oversight Annuity
Opportunity Cost
Total
Fees [9]
Agency [10]
Fees[11]
of Time[12]
Low Estimate
4.8%
0.28%
5.0%
0.25%
10.33%
High Estimate
23.0%
1.80%
10.0%
1.0%
35.80%
Financing the Transition
The revenue lost due to the
diversion of payroll taxes to individual accounts has long been recognized as a
serious problem of switching to a defined contribution system. In the period
immediately after the transition, the government’s obligations to current
retirees is largely unchanged, even though it no longer has as much payroll tax
revenue to pay these benefits. This shortfall can be quite large. For example,
in Chile it exceeded 8.0 percent of GDP in the early eighties. Some proponents
of individual accounts have argued that this shortfall should not be viewed as
increasing the government’s debt, since it is just replacing implicit debt in
the form of pension obligations with explicit debt owed by the government.[13]
Regardless of the theoretical
merits of this argument, the recent experience of Argentina demonstrates that
neither the financial markets, nor the I.M.F. accept it in practice.[14]
According to the I.M.F., the transition costs of Argentina’s Social Security
privatization increased its budget deficit by an amount equal to approximately
1.0 percent of GDP (I.M.F. 1998). This shortfall, and the subsequent interest
payments, fully accounted for Argentina’s central government budget deficits
in the three years prior to its financial collapse. However, the financial
markets evidently focused on Argentina’s current deficit and ignored the
reduction in its long-term pension liabilities. The I.M.F. adopted a similar
stance in setting a zero deficit budget target as a condition of new lending.[15]
In retrospect, it seems clear that
the decision to privatize Argentina’s Social Security system was a disaster,
given the nation’s precarious financial position. However, it should have been
possible to recognize this problem before pushing Argentina down this path. Tax
evasion by the wealthy in Argentina is not a new phenomenon, and it was not
reasonable to believe that it would be qualitatively reduced at the time the
privatization plan was put into effect in 1994. This meant that Argentina’s
government would be foregoing a large source of revenue with no obvious
alternative replacement in sight.
Since most nations in the
developing world do have chronic budget problems and pay extraordinarily high
interest rates on their debt, the push to privatize Social Security and
therefore forego substantial amounts of revenue seems very reckless. The waste
of resources implied by privatization, outlined in the first section above,
should provide further caution against going this route. In Argentina, the
financial crisis that resulted in part from the deficit created by privatization
has led to a general economic collapse and the collapse of the system of private
accounts, which were seized by the government to pay its debt obligations.[16] Requiring workers to
place their Social Security contributions in the private financial system may
provide large benefits to bankers, but it is not evident how it helps anyone
else.
References
Acuna, R. and A. Iglesias, 2001. "Chile's Pension
Reform After 20 Years." Washington, DC: World Bank.
Demarco, G. and R. Rofman, 1998. "Supervising
Mandatory Pension Funds: Issues and Challenges." Washington, D.C.: World
Bank.
Devesa-Carpio, José, and Carlos Vidal-Meliá, 2001.
"The Reformed Pension Systems in Latin America." Pension Reform
Primer. Washington, DC: World Bank.
Garcia-Mujica, Jorge, 1996. “Fiscal Impact of Switching
from a Pay as You Go to a Capitalization System,” Washington, D.C.: World
Bank.
International Monetary Fund, 1998. "Argentina: Recent
Economic Developments." IMF Staff Country Report No. 98/38. Washington, DC: International Monetary Fund.
James, E., J. Smalhout, and D. Vittas, 1999.
"Administrative Costs and the Organization of Individual Account Systems: A
Comparative Perspective." Washington, DC: World Bank.
Mitchell, O., J. Poterba, and M. Warshawsky, 1997.
"New Evidence on the Money's Worth of Individual Annuities." Working
Paper # 6002, Cambridge, MA: National Bureau of Economic Research.
United States Department of Labor, 1998. "Study of
401(k) Plan Fees and Expenses." Washington, D.C.: Pension and Welfare
Benefits Administration of the United States Department of Labor.
United States Social Security Administration, 2000. 2000
Annual Report of the Board of Trustees of the Federal Old-Age and Survivors
Insurance and Disability Insurance Trust Funds.
Washington, DC: United States Social Security Administration.
[1] Dean Baker is co-Director of the Center for Economic and Policy Research. Debayani Kar is a research associate at the Center for Economic and Policy Research.
[2] Firms typically pay out a fixed percentage of their profits as dividends or use this money to buy back shares. If the share price rises, then this payout is smaller as a percentage of the share price.
[3] It is worth noting that there will be a one-time capital gain to current holders of stock at the time when privatization is put in place, as the price to earnings ratio rises to a permanently higher level. For many shareholders this windfall will be at least partially offset by a capital loss due to the decline in bond prices associated with higher interest rates on government bonds.
[4] On accounts, see James, Smalhout, and Vittas 1999; on annuities, see Mitchell, Poterba, and Warshawsky 1997.
[5] See United States Department of Labor 1998.
[6] These fees exclude expenses associated with disability or survivor insurance, except in the case of the United States, for which the fees include the cost of administering the survivors’ insurance portion of the program.
[7] These fees exclude expenses associated with disability or survivor insurance, except in the case of the United States, for which the fees include the cost of administering the survivors’ insurance portion of the program.
[8] If a typical worker spends 2000 hours a year on the job and pays a total of 10 percent of their wages into a system of accounts, then their annual contribution is equal to 200 hours of wages. If they have to spend half an hour managing their accounts, then this is 0.25 percent of the time they spent working to earn the money contributed to their accounts. If the contribution rate is 2.0 percent of wages, then this would be equivalent to 40 hours for a full-time full year worker and the half hour spent on management is 1.25 percent of their annual contribution.
[9] The low estimate is for Bolivia and the high estimate is for Argentina. Both are taken from James, Smalhout, and Vittas 1999, p 38.
[10] The low cost estimate is for Chile; the high cost estimate is for Bolivia. The estimates are taken from Demarco and Rofman 1998, table 3.
[11] These estimates are taken from research on the costs of annuities by Mitchell, Poterba, and Warshawsky 1997. They refer only to actual expenses, not additional costs associated with adverse selection.
[12] The low estimate assumes that workers spend an average of 0.5 hours annually on their accounts and that the annual contribution is equal to 200 hours of work (10 percent of 2000). The high end assumes that workers spend an average of 2 hours per year making decisions on their accounts.
[13] See Garcia-Mujica 1996.
[14] Nor, implicitly does the World Bank, since it generally defers to the I.M.F. in assessing a country’s eligibility for non-project lending.
[15] If the I.M.F. was considering the impact of Social Security privatization in setting a zero deficit target, then it implies that the Fund would have set a target of a surplus equal to 1.0 percent of GDP if Argentina had not partially privatized its Social Security system.
[16] An account of this seizure can be found on the British Broadcasting Corporation's website [http://news.bbc.co.uk/hi/english/business/newsid_1696000/1696010.stm].