It's been a little over 30 years since Milton Friedman published Capitalism and Freedom. In this book, which was cited
in Friedman's Nobel prize award, he argues passionately and cogently
in support of free markets. He stresses that government interventions
in markets generally restrict individuals' freedom of choice and
impair the efficiency of the economy.
Friedman, and later his critics, pointed out that efficiency is
not the whole story. Although free markets generally lead to efficient
outcomes, that alone is not sufficient to guide policymaking. Policymakers
must be guided by other social objectives--namely, fairness, or
equity. Friedman's critics argued that efficient outcomes are not
necessarily fair because some individuals can receive too little
of the economy's goods or services.
Policymakers, then, can be viewed as facing a trade-off between
efficiency and fairness. According to this common view, increasing
government intervention in markets leads to more fairness at the
cost of less efficiency. But, because government interventions in
the economy have been allowed to expand, the public must view the
cost of this lost efficiency as low.
However, a lot has happened since Capitalism and Freedom was published. New knowledge and evidence have accumulated on the
relationship between economic performance and government involvement
in the economy. These new developments suggest that the costs of
government interventions typically are higher than formerly thought.
Such interventions not only result in one-time losses in economic
efficiency, as is commonly thought, but they typically also reduce
growth over time.
If the costs of achieving fairness are higher than formerly thought,
government interventions to achieve that goal ought to be scaled
back. The public must realize that the trade-off between efficiency
and fairness is not so favorable when viewed in a dynamic, or growth,
To make my arguments concrete, I examine just three of the many
ways the government intervenes in markets: trade protection, redistributive
taxes and transfers, and social security. (Of course, I could have
also included government interventions in health care, agricultural
price support policies, industrial policies, and so on.) In each
case, I explain why the government has intervened to make things
fairer and then show how the new tools and research reveal much
higher economic costs to these interventions than has been commonly
thought. The case of social security is slightly different. Some
government intervention in annuity markets can be defended on efficiency
grounds. But actual government operation of the social security
system can be defended only on fairness grounds. So the argument
here is that once the efficiency concerns are addressed, the costs
of government operation of the system are much higher than commonly
I begin with the trade-off between efficiency and fairness as
it is usually argued. A key reason the standard analysis fails to
capture all of a policy's costs is that it usually considers costs
at a point in time rather than costs over time. The difference between
the two can be dramatic.
Restoring the balance between efficiency and fairness is simple
economically, but difficult politically. Economically, the solution
is to scale back government interventions made in the name of fairness
by better targeting benefits to the poor and needy. However, politically,
that is hard to do. Government programs spread benefits widely to
low-, middle-, and high-income people in order to buy support for
the program. But there is no clear rationale for the government
to intervene in the name of fairness and distribute benefits to
middle- and high-income people. People in these income classes must
be convinced that such interventions are not in their collective
best interests. Until they are, politics suggests that the government's
role in the economy will continue to grow.
The Cost of Government Interventions
in Theory ...
The Public Demands Fairness ...
By almost any measure, the government's role in the economy has
been growing. For example, the ratio of total government expenditures
to gross domestic product has risen steadily in the last 30 years,
from less than 28 percent in 1963 to more than 34 percent in 1993.
And even these figures vastly understate the government's role in
the economy because they exclude mandates, regulations, tax subsidies,
and other types of interventions. The government's role has expanded
even though economists since the time of Adam Smith have agreed
that, except in special cases, private unfettered markets are the
most efficient way of delivering goods. So why the expansion?
The short answer is that the public wants to be fair. Admittedly,
some interventions can be attributed to special cases. Examples
include interventions in decreasing cost industries, such as price
controls in the cable television industry; interventions because
of externalities, such as pollution abatement programs; and interventions
in certain private information, insurance environments, such as
the mandate of automobile insurance for all drivers. But these special
cases cannot explain the extent of the expansion. Rather, the increase
in government interventions reflects a dissatisfaction with the
ability of markets to provide fair outcomes. Departures from free
trade are one example. Concessions under the North American Free
Trade Agreement (NAFTA) to individual industries such as sugar beet
growers, textile producers, and truckers were justified because
it would be unfair to cause workers in these industries to lose
their jobs or receive less income. Recent changes in the income
tax structure are another example. The Clinton administration's
1993 tax bill was justified on the grounds that it was taking income
from those who had profited (unfairly) from the 1980s tax cuts and
giving the proceeds to the less fortunate. And finally, the greatest
expansion in federal government expenditures over the last number
of years was in entitlements, which exist primarily to give more
income or services to the poor or unfortunate; that is, they owe
their existence to the public's support of the goal of fairness.
... But What's Fairness?
Although policies seek to achieve fairness, there's a surprising
lack of agreement among both economists and noneconomists on what
fairness really means and on how important it is relative to other
social goals. Yet most defenders of government interventions in
the name of fairness do tend to hold some similar views. They typically
agree that fairness does not require everyone to get the same size
slice of the economic pie, but they argue that market outcomes are
often unfair because some people get too little. Defenders also
agree that fairness is not all that counts. They argue that efficiency,
or the size of the pie, counts too. However, between two systems
producing the same size economic pie, they agree that the one with
slices of more equal size is better.
Costs Are Small Initially ...
The defenders have a strong case. Under unfettered markets, individuals
would be rewarded according to their contributions to the economic
pie. The rewards for some, however, would be too small to afford
them a minimal standard of living. Changes in the reward system
under unfettered markets could also be very harsh. Since industries
decline as they become obsolete or noncompetitive, workers in those
industries would lose their jobs. Since the public appears to demand
fairer outcomes, government intervention is seen as a desirable
way to bolster the standard of living of low-income individuals
and to protect the jobs of individuals in threatened industries.
But the defenders of this type of intervention recognize that
interventions incur costs. As economist Arthur Okun (1975) put it
many years ago, there is a trade-off between equity and efficiency.
Efficiency is generally best served under unfettered markets. In
a market system, prices signal how to allocate resources, and rewards
provide individuals with the proper incentives. Government interference
in markets alters the price and reward structure and causes inefficiencies.
The trade-off is easily seen in an extreme case. Suppose initially
that a market economy generates a wide distribution of income. Suppose
next that the government intervenes by taxing all individuals earning
above the average (or mean) income and transferring the proceeds
to those whose income is below the average--in essence making everyone's
income the same. Individuals would then have little incentive to
work because they would get the same income whether they worked
or not. In this case, total income, or the size of the economic
pie, would shrink considerably.
In actual practice, though, the loss of efficiency from interventions
is typically seen as modest. That is a major reason for their proliferation.
Okun likened government redistribution schemes to leaky buckets
that carry water from the haves to the have-nots. Although some
water is lost in transit, the task is still worthwhile because the
unfortunate end up with more than they had before (1975, pp. 91-106).
Empirical studies generally confirm that the costs of such interventions
are small. For instance, a recent study done by the Congressional
Budget Office (CBO) finds that removing the barriers to trade with
Mexico would lead to little gain in output in the United States
(CBO 1993, p. 23). It follows that the existence of these barriers
costs little in terms of lost efficiency. Similarly, another study
finds that the tax distortion costs of the U.S. redistributive income
tax system are no larger than the compliance and collection costs--a
modest 5-10 percent of revenues collected (Slemrod 1992, p. 46).
... But Huge Over Time
Although I have no quarrel with the argument that some government
interventions are necessary to provide adequate income or services
to the poor and unfortunate, I do quarrel with the common assessment
of their costs. If resources are being taken from one group and
given to another based on the income of each, the redistributions
necessarily distort. That is, these interventions necessarily alter
the reward structure and thereby alter incentives.
The common view of the costs of these distortions to incentives,
such as that in the two studies above, is based on an essentially
static, or point-in-time, analysis. However, interventions based
on fairness not only lead to static distortions, but they also can
reduce growth--an effect that can only be measured over time. Such
interventions typically reduce the rewards to innovation and investment
in human and physical capital. The costs of underinvestment in developing
new methods, new skills, and new equipment can become staggering.
It is possible that the recipients of the government's redistribution
schemes eventually would be better off without them: a small slice
of a big pie could eventually exceed an equal slice of a small pie.
That is essentially what happened under Eastern European socialism,
leading to the fall of Communism. Although this brand of socialism
was intended to promote fairness, the economic pie in this part
of the world became relatively so small that the middle class there
became worse off than the poor in capitalist countries.
Restraining government interventions made in the name of fairness
could lead to more growth by encouraging more innovation and investment.
More growth is desirable because it can provide larger slices of
the economic pie for everyone in the society.
The ability to raise both efficiency at a point in time and growth
over time by restraining government is not just a theoretical possibility.
It's a real option. The next three sections illustrate how government
interventions get started, how much damage they actually do, and
how they could be scaled back.
... And in Practice
On many different occasions and in many different ways, the U.S.
government has erected barriers to the free exchange of goods and
services across its national borders. The government, for example,
has placed explicit import quotas on textiles and voluntary import
quotas on Japanese automobiles, collected tariffs on liquor and
taxes on high-priced autos (the incidence of which falls predominantly
on foreign imports), and enacted anti-dumping laws to limit Japanese
computer chips. The primary reason these barriers exist is to maintain
the jobs and incomes of those in the protected industries. For instance,
Ross Perot [with Pat Choate (1993, p. 29)] argues against NAFTA
because it "will accelerate the loss of manufacturing jobs in the
United States .... Eventually, companies that choose to stay in
the U.S. will need to reduce employee wages and benefits." Economists
point out that Perot's argument is not balanced. He cites some of
the costs but ignores all of the benefits to the U.S. economy from
free trade with Mexico. Standard economic analysis suggests that
although moving to freer trade may indeed cause some temporary job
displacement, it leads to greater economic efficiency.
Standard economic analysis also suggests that only a minimal connection
exists over time between a country's level of employment and the
openness of its markets. A country can have full employment if it
is an island economy that doesn't engage in any foreign trade or
if it is an open economy that is completely free of barriers to
foreign trade. The difference in employment between the economies
would be not in the total number of jobs but in the types of work
the jobholders do. This analysis suggests that moving to freer trade
can have some temporary costs because workers will be relocated
from industries that have become less competitive in a global marketplace
to industries that are experiencing a greater demand for their goods
Against this temporary cost, standard economic analysis suggests
the need to consider the benefits of a more efficient economy. The
efficiency gain is thought to come from comparative advantage (Gould,
Ruffin, and Woodbridge 1993, pp. 1-2). The idea of comparative advantage
is that each country is relatively better than others at producing
some goods or services and so should specialize in producing what
it does best and then trading. For example, Canada is better off
specializing in growing wheat, exporting it, and using the proceeds
to purchase oranges rather than growing oranges at home.
Under standard analysis, the policy issue is this: Do the benefits
of freer trade, in terms of more efficiency, outweigh the costs
of temporary job displacement? Although economists using the standard
analysis typically favor moving to freer trade, empirically it tends
to be a close call. The costs of lost jobs and income in some vulnerable
industries are readily apparent, while the comparative advantage
gains are usually found to be fairly small.
Standard analysis, however, understates the costs of protectionism.
It fails to consider how barriers to trade impede growth. Thus the
standard approach will find only modest advantages to opening U.S.
markets. Standard analysis considers comparative advantage, but
sees only a once-and-for-all efficiency gain. After that is realized,
standard analysis suggests there is no reason why economic growth
should be affected. That is because standard analysis assumes that
the rate of technological advance is unaffected by barriers to trade.
New theory and new observations are not consistent with that assumption.
They indicate that freer trade promotes technological advances and
In recent years, macroeconomists have become increasingly interested
in why some countries' economies grow faster than others. Their
studies conclude that differences in economic growth across countries
cannot be explained simply by differences in inputs of labor and
capital (Lucas 1988 and Schmitz 1993). But economists are not in
agreement on what other factors explain differences in economic
growth. Some cite varying degrees of resistance to adopting new
technologies. Such resistance can come from capitalists or workers
who have stakes in maintaining old technologies. (See, for example,
Holmes and Schmitz 1994.) Others cite differences in human capital,
that is, in the education, training, and experience of workers that
affect their skills and competency (Lucas 1988).
Recent work by economist Robert Lucas (1993) outlines a model
that illustrates both of the above explanations. He begins by examining
differences in human capital across countries and finds that they
are not due just to disparities in formal education. Countries which
are similar with respect to labor and capital inputs and formal
education still can have significantly different growth rates. Lucas
argues that the essential differences in human capital across countries
are due to disparities in learning-by-doing on the job.
Lucas's theory suggests that economic growth is affected by the
amount of learning-by-doing on the job. It also indicates a route
by which trade barriers can lead to lower growth, and recent evidence
confirms that this route exists.
It follows from learning-by-doing that a relationship should be
observed between the productivity of workers and the length of their
experience working with a given technology. Lucas examines how the
number of worker hours to complete a given project varies as worker
experience grows. In particular, he refers to a Labor Department
study of worker hours needed to produce Liberty ships (cargo-carrying
ships built for the United States and its allies during World War
II). If that relationship (which is input per unit of output) is
inverted, a picture emerges of how worker productivity (that is,
output per unit of labor input) varies with worker experience. The
pattern seems intuitive and general. When workers begin a new technology,
their productivity is quite low. Everything is new to them. However,
after a slow start, their productivity climbs rapidly as they become
more familiar with their tasks and learn better ways of carrying
them out. Eventually, productivity levels off as workers master
the bulk of the necessary skills and procedures.
It also seems intuitive that productivity will initially fall
as a firm moves up the technology ladder. When a new technology
is adopted, productivity initially falls because workers must learn
new skills and procedures. However, with the new and improved technology,
productivity will eventually surpass what it was with the old one.
This pattern of learning-by-doing suggests that economic growth
depends on how fast old technologies are discarded and new ones
adopted. If technologies are discarded too rapidly, economies absorb
the initial costs of switching to new technologies but do not exploit
the gains from rapidly rising productivity on existing technologies.
This, in fact, seemed to be the experience of Singapore (Young 1992).
However, if technologies are kept for too long, economies are faced
with slow productivity growth because workers' learning-by-doing
has plateaued. This, in fact, seemed to be the experience of Great
Britain. Thus there is a rate of technological innovation that maximizes
Barriers to trade, then, can affect growth by slowing down the
rate of technological innovation. Protected industries do not have
to discard their outmoded technologies. Theoretically, these barriers
could promote growth if, without them, a country's rate of technological
innovation were too fast. In contrast, they could lower growth if,
without them, a country's rate were about right or too low.
Thus the relationship between growth and trade barriers becomes
an empirical issue. The new observations and new statistical studies
strongly suggest that freer trade promotes higher growth. The evidence
comes from casual observation, careful historical review, and formal
Since Friedman's Capitalism And Freedom was published in
1962, many new observations on the link between freer trade and
economic growth have amassed. During the 1950s, 1960s, and 1970s
economists persuaded developing countries to erect trade barriers
to protect their infant industries (Edwards 1993). The economies
of these countries generally stagnated. In the 1980s and 1990s countries
in East Asia and Latin America that opened their markets experienced
rapid growth. A study done at the Dallas Federal Reserve finds that
"outward-oriented policies are a much stronger conduit for economic
growth and advancement than protectionist import substitution policies"
(Gould, Ruffin, and Woodbridge 1993, p. 4). It cites an analysis
of 29 episodes of trade liberalization which finds that growth increased
in manufacturing and agriculture following the liberalizations (Papageorgiou,
Michaely, and Choksi 1991).
Other empirical studies, both formal and informal, support the
conclusion that freer trade promotes growth. One formal study finds
a positive relationship between openness and growth and concludes
when openness and the level of public infrastructure are taken into
account, physical investment becomes quantitatively more important
in the growth process, implying that a better quality of investment
is encouraged by a more liberal international trade regime and by
more government fixed investment. Particularly for the developing
countries, investment in human capital also becomes more quantitatively
important when a more open trading environment and a better public
infrastructure are in place. (Knight, Loayza, and Villanueva 1993,
Similarly, another recent study finds that economic growth was stimulated
in France by a freeing of trade within the European Community (Coe and
Moghadam 1993). Finally, an extensive survey of historical studies and
econometric analyses relating growth and trade protection in developing
countries was recently published (Edwards 1993). While critical of all
of the studies cited, the survey finds a positive relationship between
freer trade and growth in almost all of them.
The message from the new theory and evidence is clear: trade protection
inhibits growth. Thus, in erecting or maintaining trade barriers,
the fairness benefit of protecting workers should not be weighed
against just the one-time cost of lower efficiency due to reduced
comparative advantage. It also is necessary to include the cost
of lower economic growth.
Redistributive Tax and Transfer Policies
Redistributive taxes and transfers is a second area where government
interventions made in the name of fairness could be scaled back
to promote higher growth. It is generally observed that in the 1980s
the rich got richer and the poor got poorer (McKenzie 1992). The
government responded with changes to federal income tax and transfer
policies (beginning with the 1986 tax reform) to counter this increase
in income disparity (CBO 1994, pp. 54-57). In 1993, for example,
marginal tax rates were raised significantly on higher-income families
and the earned income tax credit was greatly expanded for lower-income
families. These changes were defended based on fairness arguments.
Standard economic analysis recognizes that these changes, especially
to higher marginal tax rates, had costs in terms of lower efficiency.
Higher marginal tax rates on the wealthy were seen as distorting
investment decisions by encouraging the use of tax shelters. A more
progressive tax system also was seen as reducing the supply of labor,
primarily that of households' secondary income earners. However,
these costs were seen by policymakers as low relative to the benefits
of a fairer income distribution, and so the policy changes were
The standard measure of redistribution costs understate the true
costs because they fail to include the cost of lower growth. The
costs are measured under the assumption that the income distribution
is static--the same individuals are either unemployed or working
at the same jobs year after year. Costs to redistribution, then,
are thought of in terms of how they affect individuals in given
income classes. (These classes are usually defined by quintiles,
which each contain one-fifth of the population.)
Recent research indicates that the static assumption is false
and provides a new way to think about the income distribution. Suppose
income quintiles are hotel rooms and individuals are the occupants
(Sawhill and Condon 1992). Standard analysis assumes the same individuals
are in the same rooms year after year. But, in fact, recent studies
indicate that a significant amount of room-changing is occurring
over time. It is reported that in both the 1970s and 1980s "some
three out of five adults changed income quintiles. A little less
than half the members of the bottom quintile moved up into a higher
quintile, and about half the members of the top quintile fell out
of that quintile" (Sawhill and Condon 1992, p. 3).
A U.S. Treasury Department study (1992) finds that a significant
degree of household mobility over time is explained by the life
cycle pattern of earnings. In their early years individuals typically
invest in acquiring skills and furthering their education and thus
earn low incomes. In their later years they earn progressively higher
incomes as they realize the returns to their investment in human
Thus, for many people, progressively higher marginal tax rates
represent reductions in the rate of return to their human capital.
In their early adult years they must sacrifice income to invest
in their human capital by taking time to study and train. They are
willing to invest because they foresee an adequate return in terms
of higher future income. Progressively higher marginal tax rates
reduce the return to their human capital investment by removing
progressively larger chunks of their future income.
There is good reason to believe that people's career decisions
are affected by changes in returns; that is, they go where the money
is. For example, one study reports that in the 1980s there were
large increases in the relative supplies of most of the types of
workers whose relative wages increased (Bound and Johnson 1992).
Another study finds a large enrollment response to a change in the
return to higher education (Blackburn, Bloom, and Freeman 1991).
These and other studies suggest people respond to rewards as theory
and common sense suggest. Thus lowering the returns to investment
in human capital will lower the amount of time and effort people
put into training and education.
The argument that redistribution lowers growth depends on the
existence of significant mobility across income classes, as is true
in the United States. In the United States, progressive taxes are
mainly just higher taxes on the same individuals as they move through
their careers, causing them to invest less in their human capital.
The argument need not hold for other countries which have little
mobility across classes. For those countries, taking money from
the few rich and giving it to the many poor could allow the poor
to invest more in their human capital, such as schooling, thus raising
total human capital investment and growth. But, again, in the United
States, where schooling is available to all and mobility is high,
this possibility seems remote.
That there is a conflict between fairness, or equity, and growth
can be seen clearly in the current U.S. environment. By almost all
accounts the income distribution has widened, and that in itself
has brought calls for a greater use of redistributive taxes and
transfers. However, this widening in the income distribution can
be traced primarily to changes in real, or technical, factors. Thus
greater redistribution, which mutes the message of markets, limits
individuals' response of either investing in the skills or going
to the jobs where the returns are highest.
Most studies conclude that the major cause of the widening U.S.
income distribution in the 1980s was a growing inequality in real
earnings. Tax and other policy changes caused little of the widening.
In fact, one study finds that most advanced industrialized countries
showed increases, often large, in wage inequality during the 1980s
(Davis 1992). Thus other industrial countries experienced similar
income distribution changes as in the United States even though
their policy courses were very different.
The changes in real earnings, meanwhile, are generally thought
to be due to changes in technology (Bound and Johnson 1992). In
the 1980s low- skilled jobs suffered a decline in real wages, while
high-skilled jobs experienced a rise in real wages. The changes
in real wages reflect a higher return to education, training, and
experience. The most likely explanation for the changes in returns
is the use of advanced technology that substitutes for the work
of low-skilled workers and at the same time makes high-skilled workers
more productive (Bound and Johnson 1992). One study finds that workers
who do use more advanced technology get paid higher wages (Dunne
and Schmitz 1992). Another study finds that the evolving use of
the computer can explain much of the measured technological advance
Since the changes in income distribution were largely technology-
induced, the government's effort to dampen the associated changes
in the reward structure certainly will slow growth. The market is
signaling higher returns to acquiring skills and education. When
these signals are clearly received, individuals are encouraged to
develop greater skills, such as computer training, and to acquire
more schooling, such as a college degree. Higher progressive taxes
mute those signals, however, and they decrease the incentives to
invest in these types of human capital. This reduction in investment
will slow growth.
The social security system is the final example I use to illustrate
my point that scaling back government interventions made in the
name of fairness would promote growth.
Social security serves a valuable function in society by assuring
that all older individuals have a modest amount of income when they
retire. But flaws in the way the system is designed and administered
make it both inefficient and growth-stifling. Correcting these flaws
is feasible, and the result would be gains in economic efficiency
The government's social security program is financed differently
than a private pension system. The program is financed pay-as-you-go:
current workers are taxed, and the proceeds are distributed to current
retirees. The federal government does not accumulate assets to meet
its future obligations as a private pension system does. If it did,
social security now would be recording a surplus as a trust fund,
while the federal budget with the social security trust fund removed
would be in rough balance on a present-value basis. However, current
budget projections indicate that the federal budget deficit excluding
social security will grow steadily from $242 billion in fiscal 1995
to $304 billion in fiscal 1999 (CBO 1994, p. 26). Presumably, this
deficit will continue to grow into the next century. The implication
is that instead of social security revenues net of expenditures
being used to acquire assets, they are being used to finance other
government expenditures. Thus future obligations under social security
will have to be financed by taxes on future generations of workers.
Because of the way social security is financed, its future appears
much less rosy than its past. In the past the program steadily increased
payments to retirees and paid retirees many times what they put
into the plan. This expansion in benefits was financed by rising
social security tax revenues fed by increases in the percentage
of the workforce paying into the plan and in both the social security
tax rate and tax base. However, in the future the return to retirees
looks much less favorable. Since the program now applies to virtually
all workers, no substantial increases in coverage are possible.
Demographics are also turning less favorable. In the 1980s there
were five workers supporting each retiree. By 2050 that 5-to-1 ratio
will fall to 2-to-1. These facts suggest that in the future retirees
will be getting less back from the plan than they put in (Goodman
and Musgrave 1992, chap. 13).
This unsettling future gives good reason to consider reforms of
the system. Two reforms are readily suggested: first, back future
obligations with funds of assets, and second, turn over the management
of those funds to the private sector. Because these reforms would
make the system more market-oriented, they would likely lead to
greater efficiency and higher growth.
Under this proposed privatization scheme, the government would
still have a role in the social security system. That role is dictated
on efficiency grounds. Due to adverse selection problems (which
are explained in the sidebar), unfettered private annuity markets
are unlikely to be able to provide adequate protection to workers.
However, the government can address these problems without managing
Thus the social security example is a little different from the
previous two. The argument is not that unfettered annuity and pension
markets are most efficient. In fact, some government intervention
seems necessary for efficiency. However, it is not necessary that
the government operate the system. The argument, then, is that changing
the government-operated system to a privately run, funded system
will increase both efficiency and growth.
The government, of course, could have opted for a privately run
social security system at the outset, but instead it chose to operate
the system itself. Since it could have addressed the adverse selection
problems without managing the system, it must have had other reasons
for doing so. It appears that the other reasons are based on fairness--fairness
dictated a government-run, pay-as-you-go system.
The social security system aims for fairness through income redistribution.
It defines benefits for retirees, and those benefits are only loosely
tied to contributions. This has allowed the system to redistribute
income from the newer generations to the older generations and among
individuals in the same generations.
The social security system, then, can be thought of as a mandatory,
unfunded pension system and a complicated tax/transfer system. Because
of the fairness aspects of the system, the government has chosen
to operate it. I argue that the United States would be better off
with private operation of a funded pension system subject to government
edicts addressing adverse selection and with an explicit tax/transfer
policy targeting income to the older poor.
The government-run system is less efficient than the one I am
advocating. Under the government-run system, workers view their
social security contributions as a tax. In fact, this is a major
reason why the CBO (1991, pp. 71-75) argues that social security
payments and contributions should be consolidated with ordinary
government expenditures and taxes. Workers view their contributions
as taxes because their future benefits are not closely tied to those
contributions. Their future benefits are promised, but the government
at times changes them. In fact, those benefits depend not only on
economic and demographic developments that affect future real income
and the size of the future workforce, but also on the willingness
of future workers to continue to participate in the plan.
In contrast, workers contributing to a private plan would own
their pension/annuity and thus be less prone to view it as a tax.
Benefits would depend on the amount they contribute and on the return
to their investment. Their payoff would be uncertain only because
the rate of return to their investment would be uncertain. there
would be no uncertainty with respect to the willingness of future
generations to participate.
Society might still want to redistribute income on fairness grounds
even with a privately run system. But then the government would have to
make explicit the taxes and transfers for carrying out the redistribution.
Making the redistribution explicit would likely make it less pervasive
and better targeted.
The privately run system is likely to lead to greater economic
efficiency than a government-run system for two reasons. First,
privately run funds would have to compete among themselves and would
likely be more efficient than a monopolistic, government-run fund.
Second, the redistribution scheme used with a privately run system
would likely be more limited, which would lead to smaller disincentives
to work and save.
Probably more important, though, a privately run system would
lead to higher growth. That is because the rate of return to investment
in a privately funded system would be greater than the rate of contributions
in a pay-as-you-go system. In a paper written well after Friedman's Capitalism and Freedom, economist Martin Feldstein (1987)
examines the economic consequences of shifting dollars from an unfunded
to a funded system. He finds that an extra dollar placed in a funded
system would yield to retirees the pretax rate of return on private
capital, which might be expected to lie in the 10-15 percent per
year range, as it has in the last 40 years. However, an extra dollar
contributed to the unfunded social security system would yield a
rate of return equal to the rate of growth of the economy, which
most analysts now expect to be around 3 percent per year. This suggests
that growth would be increased if resources were shifted from the
low-yielding social security system to a higher-yielding privately
Although shifting to a privately funded system would increase
growth, it would not make every generation better off. The shift
requires that society as a whole save more by paying less to current
retirees. Thus the current old generation would be made worse off.
Feldstein shows that under reasonable assumptions and using standard
ways of weighting the welfare of different generations, the gains
to the current young and future generations would greatly exceed
the losses to the current old generation. Nevertheless, current
retirees would be hurt by the shift. This suggests that if dollars
are shifted out of social security, the shift should be made gradually,
and the dollars should be taken from those most able to fend for
themselves--the people with high incomes and wealth.
This proposal to shift to a privately run social security plan
will strike many as unrealistic, if not ridiculous. However, it
should be noted that such a shift is being made in other countries.
Chile gave workers the option to put their money either in the state-run
system or in private funds. Most workers have been choosing the
private funds, so the state-run system is gradually being phased
out. Moreover, the Chilean plan has proved so successful that other
Latin American countries either have adopted similar programs (Mexico
and Peru) or are in the process of doing so (Argentina, Bolivia,
and Venezuela) (Survey 1993).
Public Attitudes Must Change
My argument that the costs of government interventions made in
the name of fairness are costlier than commonly thought does not
suggest that they should be eliminated. Rather, it suggests that
they should be scaled back. Economically, restraining or scaling
back government redistributions seems straightforward and easy to
do. Politically, it seems difficult. That is why public attitudes
toward government redistributions must change first.
Economically, all that is required is that programs to redistribute
resources be better targeted. The economic rationale for these programs
is to provide resources to the poor and needy. No economic rationale
exists for providing resources to individuals in the middle- and
upper-income classes. Yet that's where most of the money goes. A
study of how federal benefits are distributed across income classes
finds that the benefits scarcely redistribute income at all (Howe
1991). In fact, over 40 percent of the 1991 benefits went to households
earning over the average (median) cash income, which was three to
four times the poverty threshold for a family of two.
By making transfers to those who are not needy, the amount of
redistribution is much larger than it has to be for fairness. This
excess unnecessarily raises the costs to the economy in terms of
inefficiency and slow growth. Economically, interventions made in
the name of fairness can then be easily restrained by making all
benefits means-tested, that is, making sure the benefits go only
to people whose wealth and income are below some threshold levels.
Politically, this remedy may be hard to apply. The benefits of
government redistributions tend to be spread widely because that
is what the public wants. In order to get public support for such
programs, politicians find it necessary to distribute their benefits
far beyond just the truly needy. Many people have the attitude that
if they pay for a program, they deserve some of the proceeds. As
long as the public feels this way, there will be little support
for reining in government programs and interventions.
The public, however, has it dead wrong. If the government is to
intervene based on fairness, its role is to redistribute resources
from the middle- and upper-income classes to the poor. Period. People
in the middle and upper classes would be much better off if the
amount of taxes used for their own benefits was eliminated and they
used the increase in their after-tax incomes to arrange privately
for their benefits. In this way fairness could still be served,
but the government's role and the resulting inefficiencies could
Milton Friedman was right to stress the advantages of unfettered
markets. Yet his critics also were right to point out that some
individuals in society cannot attain an acceptable level of existence
without help. The task confronting society is to develop systems
that help the poor while interfering as little as possible with
the private markets' ability to foster efficiency and growth. That
means targeting redistributions carefully. It means the public must
become aware of the high cost of extending benefits to those who
do not truly need them. It also means that failure to change could
cause society to pay an increasingly high cost as time goes by.
How Adverse Selection Affects Annuity Markets
The adverse selection problem in annuity markets is easy
to see. Suppose an insurance company offers an annuity which
pays a fixed sum of money per month to individuals at age
60 for as long as they live and which charges a premium based
on the population's average life expectancy. Then individuals
who have sound, and in part private, information which suggests
that they will live longer than average would have an incentive
to buy the annuity. However, those who expect to live shorter
than average would have an incentive not to buy the annuity.
Thus those who buy the annuity would have life expectancies
longer than the population average, costing the insurance
company more than it had planned and forcing it to raise its
premium. But then those with poorer health who were willing
to buy before would no longer want to buy at the higher premium.
This adverse selection process could continue until it becomes
unprofitable for the insurance company to offer any annuity
at all. This could occur even though the public were better
off with active annuity markets. (See, for example, Eckstein,
Eichenbaum, and Peled 1985.)
The government can help attenuate the adverse selection
problem. It need only issue two simple edicts:
- Insurance companies cannot discriminate in their premiums
based on the perceived health of the applicants.
- All individuals must buy some minimal amount of annuities.
Edicts of this type are common in other types of insurance. For
instance, they are included in the administration's health care
plan to address the similar adverse selection problems in health
The views expressed in this annual report are
solely those of the author; they are not intended to represent a
formal position of the Federal Reserve System.