The Size and Role of Government: Economic Issues
The Size and Role of Government:
Updated March 5, 2008
Specialist in Macroeconomics
Government and Finance Division
The Size and Role of Government: Economic Issues
The size and role of the government is one of the most fundamental and
enduring debates in American politics. Economics can be used to analyze the
relative merits of government intervention in the economy in specific areas, but it
cannot answer the question of whether there is “too much” or “too little” government
activity overall. That is not to say that one cannot find many examples of
government programs that economists would consider to be a highly inefficient, if
not counterproductive, way to achieve policy goals. Reducing inefficient government
spending would benefit the economy; however, reducing efficient government
spending would harm it, and reducing the size of government could involve either
one. Government intervention can increase economic efficiency when market
failures or externalities exist. Political choices may lead to second-best economic
outcomes, however, and some argue that, for that reason, market failures can be
preferable to government intervention. In the absence of market failures and
externalities, there is little economic justification for government intervention, which
lowers efficiency and probably economic growth. But government intervention is
often based on the desire to achieve social goals, such as income redistribution.
Economics cannot quantitatively value social goals, although it can often offer
suggestions for how to achieve those goals in the least costly way.
The government intervenes in the economy in four ways. First, it produces
goods and services, such as infrastructure, education, and national defense.
Measuring the effects of these goods and services is difficult because they are not
bought and sold in markets. Second, it transfers income, both vertically across
income levels and horizontally among groups with similar incomes and different
characteristics. Third, it taxes to pay for its outlays, which can lower economic
efficiency by distorting behavior. Not all taxes are equally distortionary, however,
so there are ways of reducing the costs of taxation without changing the size of
government. Furthermore, deficit spending does not allow the government to escape
the burden of taxation since deficits impose their own burden. Finally, government
regulation alters economic activity. The economic effects of regulation are the most
difficult to measure, in terms of both costs and benefits, yet they cannot be neglected
because they can be interchangeable with taxes or government spending.
There are many different ways to measure the size of the government, making
its economic effects difficult to evaluate. Budgeting conventions are partly
responsible: tax expenditures, offsetting receipts and collections, and government
corporations are all excluded from the budget. But some governmental functions,
like regulation, simply cannot be quantified robustly. Discussions about the overall
size of government mask significant changes in the composition of government
spending over time. Spending has shifted from the federal to the state and local
level. Federal production of goods and services has fallen, while federal transfers
have grown significantly. Today, nearly two-thirds of federal spending is devoted to
Social Security, Medicare, Medicaid, interest payments, and national defense. Thus,
there is limited scope to alter the size of government without fundamentally altering
these programs. The share of federal spending devoted to the elderly has burgeoned
over time, and this trend is forecast to continue.
How Does the Government Affect the Economy?.........................1
How Large Is the Government?.......................................2
Effect of the Government on Economic Efficiency.......................13
What Is a Market Failure?......................................15
Ex ternalities .............................................17
Failure to Optimize.......................................18
How Do Taxes Affect Economic Efficiency?.......................19
Balancing Economic Efficiency With Other Goals...................21
Effect of the Government on Economic Growth.........................22
Effect of Spending............................................23
Effect of Transfers............................................25
Effect of Taxes...............................................26
Effect of Regulation...........................................28
List of Figures
Figure 1. Federal Outlays and Receipts, 1950-2007.......................3
Figure 2. Federal Spending by Function, 1962-2007......................5
Figure 3. Government Purchases and Transfers, 1950-2007................6
Figure 4. Federal and State and Local Outlays, 1950-2007................11
Figure 5. Public Investment Spending, 1950-2007.......................24
List of Tables
Table 1. Estimated Revenue Loss from Income Tax Expenditures
Exceeding $10 Billion in FY2009.................................8
Table 2. Offsetting Receipts and Collections by Type, FY2006............10
The author would like to thank Justin Murray for helpful research assistance.
The Size and Role of Government:
The appropriate size and role of the government is one of the most fundamental
and enduring debates in American politics. What role does the state play in
economic activity? How is the economy affected by government intervention?
Many of the arguments surrounding the proper size of government are economic in
nature, and they will be discussed in this report.
How Does the Government Affect the Economy?
Government activity affects the economy in four ways:
!The government produces goods and services, including roads and
national defense. Less than half of federal spending is devoted to the
production of goods and services.
!The government transfers income through both the tax system and
outlays. Popular perception typically focuses on transfers across
income classes through the progressive income tax system and
means-tested benefits, referred to as vertical redistribution. But
vertical redistribution is dwarfed by horizontal redistribution,
transfers unrelated to income class. The largest beneficiaries of
transfers are the elderly, through programs such as Social Security.
!The government collects taxes, and that alters economic behavior.
For instance, taxes on labor change the incentives to work, while
taxes on specific goods (e.g., gasoline) change the incentive to
consume and produce those goods.
!The government regulates economic activity for a number of
reasons, including environmental protection, workplace safety, and
consumer protection. The economic impact of regulation is
probably the hardest and most contentious to measure of the four1
types of government economic activity.
1 The term regulation is used in this report in the popular sense to encompass laws,
mandates, and government regulations that affect commerce.
How Large Is the Government?
Before assessing how the government affects economic activity, it is necessary
to agree upon how to measure the size of the government. For a number of reasons,
this exercise is less straightforward than it may seem.
!The size of government can be expressed in a number of
different units of measurement. Should the size of government
be measured in dollars, on a per capita basis, by total employees,
or as a percentage of GDP?
Each measurement has its advantages, but some measurements have more
shortcomings than others. If measured in dollars, then those dollars should be
adjusted for inflation. The purpose of measurement is to gauge the resources at the
government’s disposal, and a dollar of tax revenue in 1946 would buy $10.63 of2
goods and services in 2007 because of inflation. Measuring the size of government
by the number of employees is imprecise because the government can substitute
capital for labor over time to accomplish the same tasks with fewer employees. For
example, the government’s purchase of computers has rendered many clerical jobs
obsolete. The federal government can also pay workers in the private labor force
through contracts and grants or allow state and local government workers to deliver
federal programs in place of federal public servants. One estimate puts the number
of private and state and local government workers working for the federal3
government at more than seven times the size of the federal workforce.
Comparisons over time that do not incorporate demographic change are arguably
misleading because government spending per capita is more meaningful than total
government spending: $458.4 billion in federal spending in 1946 amounted to $3,242
per person then, but would only finance $1,533 per person in 2006.
But over long periods of time, because of the power of compounding, any level
of government spending will appear to be insignificant unless it is expressed as a
fraction of gross domestic product (GDP) (a measurement that incorporates inflation
and population growth). In 1944, at the height of World War II, federal spending was
less than half of today’s federal budget in constant (inflation-adjusted) dollars. Yet
outlays in 1944 accounted for 43.7% of GDP, whereas the budget in 2007 accounted4
for only 20% of GDP. Nevertheless, some argue that stating the size of government
as a percentage of GDP understates increases in government spending in the short
term, particularly in years when growth is high. For example, those who claim that
government spending increased sharply in 2000 point to the fact that it increased by
2.5% in constant dollars. Those who claim the increase was modest point to the fact
that it fell by 0.2 percentage points of GDP. Since this report focuses on long-term
trends, all measurements are made as a percentage of GDP.
2 Federal Reserve Bank of Minneapolis, “What is a Dollar Worth?”,
[ h t t p : / / www.mi nneapol i s f e d.or g/ r e sear ch/ dat a/ us/ c al c/ ] .
3 Paul Light, The True Size of the Government (Washington: Brookings Institution Press,
4 All budget data are measured on a fiscal year basis unless otherwise noted.
!The size of the government can be measured by expenditures
(outlays) or revenues (receipts).5
At times when the budget deficit is large, the difference between the two
measures is significant, as seen in Figure 1. Measured by receipts, the size of
government in the post-war period peaked in 2000. Measured by outlays, the size of
government peaked in 1983. In 2000, the peak year as measured by receipts, outlays
were at their lowest level since 1966. Government has grown since 2000 when
measured by outlays, and shrunk measured by receipts.
Figure 1. Federal Outlays and Receipts, 1950-2007
3 9 56 59 62 9 65 9 68 71 74 9 77 9 80 83 86 9 89 92 95 98 0 01 04 07
19 50 19 5 1 19 19 1 1 19 19 1 1 19 19 1 19 19 19 2 20 20
Re c e i p ts Out la ys
Source: OMB, Budget of the U.S. Government, Historical Tables, Table 1.2
Note: Data measured on a fiscal year basis.
There are two main reasons why outlays might be considered a better measure
of the size of government than receipts. First, receipts are more volatile than outlays
and are only indirectly controlled by legislators. They are particularly sensitive to
economic conditions. Receipts did not peak in 2000 because of changes in the tax
code, but because of the interaction between the tax code and the extremely rapid
growth in (taxable) income.
Second, outlays and revenues can temporarily diverge because of budget
deficits. But eventually, the budget must be brought back into balance. Therefore,
cutting taxes without corresponding spending cuts does not permanently reduce the
size of government, and measuring the size of government by revenues gives the
misleading impression that government is smaller than it is. Furthermore, although
people often refer to the burden of high taxes, that burden cannot be avoided in the
long run through deficits because deficits impose a burden that is every bit as real as6
taxes. In other words, a given level of spending requires the resources of individual
5 The data reported in budget documents are net receipts and net outlays, and do not include
offsetting receipts and collections, which are described below. This report will follow the
standard convention of defining receipts and outlays as net measurements, unless otherwise
6 See also CRS Report RL30520, The National Debt: Who Bears Its Burden?, by Marc
Labonte and Gail Makinen; and CRS Report RL31775, Do Budget Deficits Raise Interest
Rates and Is This The Relevant Question?, by Marc Labonte.
taxpayers, whether deficit financed or tax financed; all that changes is the timing of
While it is sometimes argued that deficits hold down the growth in spending for
political reasons, deficits directly increase the outlays needed to maintain a fixed
level of government services in the future by increasing interest payments on the
national debt. In other words, if spending is constant over time, a $1 tax cut today
will lead to a tax increase of $1 plus compounded interest in the future. This raises
a further question: if interest payments are the direct result of deferring payments for
past spending to the present, should they be included in comparisons of the size of
government over time? Including net interest payments, as current practice does,
makes the government appear to be larger following periods of large deficits relative
to the period before the deficits. Excluding them gives a different budget picture: for
example, high interest payments in the 1990s and 2000s obscure the fact that other
outlays at the time were as small as they had been in the 1960s.
Since this report focuses on long-term trends, it measures the size of
government by spending. When measured by outlays, the size of government
followed an upward trend until 1983, and has followed a downward trend ever since.
Outlays rose from 14.7% of GDP in 1947 to 23.5% in 1983. They then fell to 18.4%
of GDP in 2000, and have increased since. Outlays were below 20% of GDP from
1996, and were below 20% of GDP from 1997 to 2002, but reached 20% of GDP
again in 2003 and 2005-2007.
!A look at total spending masks the large compositional changes
in spending over the post-war period.
In a nutshell, the government’s largest activity has gone from national defense
in the 1960s to transfers to the elderly today. Defense spending peaked at 9.5% of
GDP in 1968. It then fell to 4.7% of GDP in 1978, rose to 6.2% of GDP in 1986
before beginning a sharp decline, and stood at 4.0% of GDP in 2007. At the same
time, mandatory spending (excluding net interest) has risen from 4.9% of GDP in
1962 to 10.6% in 2007. Mandatory spending programs for which the elderly are
major beneficiaries make up nearly three-quarters of all mandatory spending.7 Non-
defense discretionary spending, which includes spending on transportation,
education, the environment, and numerous other government activities, grew from
and stood at 3.6% of GDP in 2007. Net interest on the publicly held national debt
grew significantly in the 1980s and 1990s because of the government’s budget
deficits. In the post-war period to the 1980s, it had always been below 2% of GDP;
in the 1980s and early 1990s, it exceeded 3% of GDP.
7 Mandatory spending programs for which the elderly are major beneficiaries include Social
Security, Medicare, Supplemental Security Income, veterans benefits, and federal (civilian
and military) pensions. This calculation does not include Medicaid, which finances long-
Figure 2. Federal Spending by Function, 1962-2007
0 974 4 8 992 06
1962196419661968197 19721 1976197819801982198 1986198 19901 19941996199820002002200420
Defense Discretionary SpendingNon-Defense Discretionary Spending
Mandatory SpendingNet Interest
Source: OMB, Budget of the U.S. Government, Historical Tables, Table 8.4.
Note: Data measured on a fiscal year basis. This data series not available before 1962.
It is useful to remember that federal spending is overwhelmingly devoted to a
handful of activities. Defense spending, Social Security, Medicare, Medicaid, and
interest payments on the national debt accounted for nearly two-thirds of all federal
outlays today. Thus, any proposal to reduce the government’s size would be unlikely
to make much of a dent in overall spending unless it reduced one or more of these
!Government transfers and government purchases of goods and
services have different and distinct effects on the economy.
Economists draw a distinction between government outlays spent on goods and
services (purchased from the private sector or produced directly by the government)
and outlays that transfer resources from one set of private individuals (taxpayers) to
another. Since a significant portion of government spending is devoted to
government transfers to individuals, much of the revenue collected through taxation
is ultimately spent by the private sector on private sector goods and services (after it
is transferred by the government). Government transfers do not employ U.S. capital
and labor (except to administer those transfers) in the same way as government
production of goods and services. Government transfers basically shift private sector
spending from one group of private individuals to another.8 By shifting income from
its market allocation, government transfers still have an effect on the economy,
however, because the transfers and taxes to finance them alter the incentives to work
The merits of government transfers cannot typically be evaluated on the basis
of economic efficiency alone, since they often implement social goals. By contrast,
government production of goods and services falls comfortably within the framework
of economic efficiency based on whether or not the spending addresses a market
failure, as explained in the next section.
Figure 3. Government Purchases and Transfers, 1950-2007
0 954 958 962 966 970 974 978 982 986 990 994 998 002 006
195 1 1 1 1 1 1 1 1 1 1 1 1 2 2
government purchasestransfers, interest, and subsidies
Source: OMB, Budget of the U.S. Government, Historical Tables, Table 14.2.
Note: Data measured on a fiscal year basis. Grants to state and local government not included.
Interest is defined as net interest paid on the national debt. Subsidies are measured as net outlays to
Transfers can result in vertical redistribution, transfers among income classes,
or horizontal redistribution, transfers unrelated to income class. Most transfers on
the spending side of the budget result in horizontal redistribution. For example, the
largest portion of government transfers are directed to the elderly, notably through
Social Security. (Social Security contains an element of vertical redistribution in that
the tax-benefit formula is more favorable as income declines, but its main function
is horizontal redistribution.) Another large category of transfers is devoted to interest
on the national debt, which represents a transfer from taxpayers to bondholders.
Transfers also include net subsidies to government corporations, which are discussed
below. The largest portion of government production is directed to national defense.9
8 Most transfers are from U.S. taxpayers to other Americans. Some transfers go to
foreigners, however. In 2005, the U.S. government paid $134 billion to foreigners in
interest payments on the national debt and $33 billion in other transfers (e.g., foreign aid.)
Transfers to foreigners made up 11% of total transfers. See BEA, National Income and
Product Accounts, Table 3.2.
9 Data on federal production of goods and services and transfers does not include grants to
The distinction between government transfers and government production of
goods and services becomes important when making historical comparisons. Total
federal outlays have remained in the range of 20% with a slow upward trend for most
of the post-war period, but this masks large changes in the composition of federal
outlays. Since the 1950s, outlays for federal production and transfers have moved
in the opposite direction. While government production was nearly three times as
large as government transfers in the early 1950s, production is only half as large as
!Measuring the size of government by receipts or outlays omits
Tax expenditures are defined in the Congressional Budget Act of 1974 (P.L. 93-
344) as “revenue losses attributable to provisions of the Federal tax laws which allow
a special exclusion, exemption, or deduction from gross income or which provide a
special credit, a preferential rate of tax, or a deferral of liability.” They take many
forms and cover many policy areas; a Senate Budget Committee compendium
identifies 160 corporate or income tax expenditures in current law.10 Revenue loss
attributable to these tax expenditures totaled $943 billion, or 7.2% of GDP, in
FY2006.11 This is almost as large as total discretionary spending (defense and non-
defense) in FY2006, yet tax expenditures do not show up directly on the outlay or
revenue side of the budget.12 Table 1 lists the largest tax expenditures. As a
comparison, the tax expenditure for the deduction of mortgage interest is more than
one and a half times as large as the entire FY2006 appropriation for the Department
of Housing and Urban Development (HUD).
state and local governments. A non-comparable data series on grants to state and local
governments from OMB suggests that grants for transfers are about twice as large as grants
for production in recent years.
10 U.S. Congress, Senate Committee of the Budget, Tax Expenditures: Compendium of
Background Material on Individual Provisions, committee print 109-72, 109th Cong., 2nd
sess. (Washington: GPO, 2006).
11 According to the compendium, this is the sum of the estimated revenue gains from
repealing each tax expenditure individually. It is not the estimated revenue gain from
repealing all tax expenditures simultaneously since that would lead to interactions among
the repeal of different expenditures.
12 Generally, the refundable portion of a tax expenditure is counted as an outlay, but very
few expenditures are refundable so only a small proportion of total tax expenditures is
counted as outlays. Refundable credits are tax credits which are paid when the taxpayer has
no tax liability.
Table 1. Estimated Revenue Loss from Income Tax
Expenditures Exceeding $10 Billion in FY2009
(projection, in billions of dollars)
FY2009 FY2009-FY2013( c umula t iv e)
Exclusion of employer medical insurance premiums$168.4$1052.0
and medical care
Deductibility of mortgage interest on owner-occupied100.8576.7
ho usi ng
Capital gains (except agriculture, timber, iron, ore, and55.9257.2
Net exclusion of pension contributions and earnings:51.0325.0
Deductibility of charitable contributions, other than47.0274.0
education and health
Net exclusion of pension contributions and earnings:45.7216.3
Accelerated depreciation of machinery and equipment44.1270.0
Step-up basis of capital gains at death36.7197.8
Capital gains exclusion on home sales 34.7191.8
Deductibility of non-business state and local taxes other33.2256.5
than on owner-occupied homes
Exclusion of interest on public purpose state and local25.9137.5
Exclusion of interest on life insurance savings23.5140.6
Social Security benefits for retired workers18.6106.0
Deductibility of state and local property tax on owner-16.6131.7
Deduction of U.S. production activities15.3119.3
Deferral of income from controlled foreign13.876.3
Accelerated depreciation on rental housing11.872.8
Individual Retirement Accounts11.767.4
Source: OMB, Budget of the U.S. Government, Analytical Perspectives, Table 19.3.
Note: Data measured on a fiscal year basis.
Economists would argue that, in many ways, tax expenditures are equivalent to
government spending, and which is preferable depends on which is a more effective
or efficient way of achieving any particular goal. For example, a $1,000 child tax
credit is equivalent to the government sending a check for $1,000 to parents who
have eligible children and meet the credit’s income requirements. Similarly, a tax
deduction for mortgage interest for a taxpayer in the 33% marginal income tax
bracket is equivalent to the government sending the taxpayer a check for 33 cents for
every dollar of mortgage interest paid. From a revenue perspective, the equivalence
comes from the fact that marginal rates must be raised to the same extent to finance
an expenditure whether it is a tax expenditure or an outlay. Because tax provisions
are permanent (unless they include an expiration date), however, revenue loss from
specific expenditures may rise over time automatically without congressional action,
unlike appropriated spending. If this equivalence argument is correct, measures of
the size of government that omit tax expenditures drastically underestimate its size.
!Measuring the size of government by receipts or outlays omits
offsetting receipts and collections.
The outlay and revenue totals discussed here and in the federal budget are net
measures, and do not include offsetting receipts and collections. These are the
income that the government receives, primarily from business-like activities (many
are user fees). Mostly, the offsetting receipts and collections go directly toward the
provision of those activities for which they were collected, in some cases without
appropriation. These include receipts collected directly by the government for health
care premiums through Medicare Part B, national park user fees, and proceeds from
the sale of government resources. They also include receipts collected by
government corporations (defined below) such as the Postal Service, the Export-
Import Bank, and the Federal Deposit Insurance Corporation. The receipts and
collections are not included in revenues, and the outlays that they fund are subtracted
from total outlays.
OMB justifies the exclusion of offsetting receipts and collections from the
budget on the grounds that “[t]he budget focuses on ... outlays and receipts that
measure governmental activity rather than a combination of governmental and market
activity.”13 What should be realized is that even if offsetting collections and receipts
are not included in the budget as revenues because they represent choices made in the
marketplace (i.e., they are not compulsory like taxes), removing them from revenues
also causes the activities that they finance to be removed from the outlay side of the
budget. Some budget analysts argue that this keeps many of the financed activities
outside the oversight and deliberation of the annual appropriation process.
Regardless of whether offsetting receipts and collections finance “public goods”
or (government-provided) “private goods,” when thinking about the size of the
government in relation to private economic activity, it may be sensible to include
them, and they are not an insignificant sum. In 2007, offsetting collections and
receipts totaled $320.7 billion, about 11.7% as large as outlays included in the
budget. Table 2 lists offsetting receipts and collections that exceeded $10 billion in
13 OMB, Analytical Perspectives, Budget of the U.S. Government, FY2004, Ch. 21.
removing them from the budget arguably underestimates the growth in the size of
Table 2. Offsetting Receipts and Collections by Type, FY2006
(in billions of dollars)
Military assistance program sales15.8
Sale of energy 12.8
Source: OMB, Budget of the U.S. Government, Analytical Perspectives, Table
Note: Data measured on a fiscal year basis.
!When considering government’s influence on the economy, it is
best to include government spending at the state and local level.
When state and local government spending is included, the decline in the size
of the government since 1983 is smaller because of the corresponding increase in the
size of state and local government. State and local government outlays grew from
7.1% of GDP in 1950 to 9.8% in 1983.15 But unlike the federal government, state
and local outlays have grown since 1983, reaching 11.6% of GDP in 2007. The
growth in state and local government is larger (and the growth in the federal
government is smaller) if one includes federal grants to state and local governments,
which have grown from 0.7% of GDP in 1950 to 2.7% of GDP in 2007. In contrast
to the federal government, at the state and local level government purchases
significantly exceeds government transfers.
14 For example, the General Accounting Office estimates that the user fees it studied
increased by 27% in real terms from 1991-1996. See GAO, Federal User Fees: Budgetary
Treatment, Status, and Emerging Management Issues, AIMD-98-11, December 1997.
15 These figures refer to own-source state outlays, and do not include state outlays financed
by grants from the federal government. Those grants are included in federal outlays.
Figure 4. Federal and State and Local Outlays, 1950-2007
1950 1955 1960 196 197 19 19 1 985 1 990 1995 2000 2005
Federal OutlaysState and Local Outlays
Source: OMB, Budget of the U.S. Government, Historical Tables, Table 15.3.
Note: Data measured on a fiscal year basis. State and local outlays include own-source outlays only.
State and local outlays from federal grants are included in federal outlays.
!Should government corporations be included in discussions
involving the size of government? Should government-
sponsored enterprises be included?
Government corporations are government agencies that provide market services
and raise revenue that partly or fully cover their expenses. The Postal Service,
AMTRAK, and the Federal Deposit Insurance Corporation are prominent examples.
For the most part, their revenues and expenses occur outside of the budget. Surpluses
are returned to the Treasury and the corporation may receive appropriated subsidies
or loans from general revenues. The activities of these agencies are recorded in the
federal budget only in-so-far as they receive subsidies or federal loans, or generate
surpluses. They do not have shareholders, and raise capital through the Treasury’s
Federal Financing Bank. Some raise capital independently of the government; for
example, the Tennessee Valley Authority issues its own bonds.
In economic terms (business decisions, budgets, products, and so on), these
corporations may more closely resemble private businesses, and have little input from
the executive branch or legislature in day-to-day decision making. Nevertheless, the
organizations are part of the government, and not privately owned. In that way, the
major difference between them and the rest of the government is that they provide
goods and services that are sold in the market, whereas typically the government
provides non-market goods (e.g., education, defense, and highways). Sometimes the
government corporations compete with private corporations (e.g., package delivery
by the Postal Service), and sometimes they have a monopoly (e.g., letter delivery by
the Postal Service).16 Government corporations account for about 1% of national
income, a fraction that has stayed relatively constant over time.17
In addition, there are “quasi-government” organizations with more distant ties
to the government than government corporations, but distinct from the private sector.
These include RAND, the Smithsonian Institution, and government sponsored
enterprises (GSEs), such as Fannie Mae and Freddie Mac.18 What they have in
common are legal characteristics that in some way link them to the government. For
example, the GSEs were federally chartered by the government and have special
borrowing privileges from the U.S. government. However, they are shareholder-
owned corporations with management structures independent of the government and
do not receive government appropriations (although their ties to government increase
the firms’ market value.)19
!The focus on an annual cash flow budget neglects the fact that
policy decisions made today can have long-run consequences
that are not reflected in today’s budget.
This can occur in a number of ways. Many recent tax cuts and spending
programs are being phased in over a number of years. For example, under P.L.107-
Spending on multi-year projects can lead to implicit future spending commitments
even if explicit commitments are not made. For example, once a military hardware
investment project begins, it may be highly impractical to stop funding the project
before it has been completed. Nevertheless, defense spending, like all discretionary
spending is assumed to grow at the rate of inflation in the budget baseline, and future
budget totals are not adjusted for multi-year spending projects.
Mandatory spending can increase with no change in law when the number of
eligible beneficiaries increases. The example of this phenomenon that dwarfs all
others under current policy is entitlement spending on the aged. Under current
policy, the retirement of the baby boomers and the increase in life expectancy are
projected to cause spending on Social Security, Medicare, and Medicaid to increase
from 7.4% of GDP in 2000 to 18.8% of GDP in 2080.20 While the spending increase
will not occur until the future, the promises to increase spending are being made
today. These promises are not reflected in the FY2008 budget, and unlike a private
pension plan, the payroll taxes being collected from baby boomers while they work
16 For more information on government corporations, see CRS Report RL30365, Federal
Government Corporations: An Overview, by Kevin Kosar.
17 BEA, National Income and Product Accounts, Table 1.13.
18 Some analysts argue that the primary motivation for the creation of certain GSEs was to
remove their costs from the federal budget. For example, see Roy Meyers, Strategic
Budgeting (Ann Arbor: University of Michigan Press, 1996), pp. 72-79.
19 For more information on the quasi-government, see CRS Report RL30533, The Quasi
Government: Hybrid Organizations with Both Government and Private Sector Legal
Characteristics, by Kevin Kosar.
20 OMB, Budget of the U.S. Government, Analytical Perspectives, FY2009, Table 13.2.
are not set aside to fund the future obligations they have secured under current law.
In all, one recent study estimated that the government’s total future unfunded
liabilities equaled $44.2 trillion in present value terms.21 Some budget analysts have
argued that the budget should be measured using a different accounting method (e.g.,
accrual accounting) so that liabilities incurred today are recorded in today’s budget.
Different accounting methods might tell a significantly different story about changes
in the size of government over time.22 The shortcoming of cash flow budgeting as
a measure of size of government is best expressed in the question: does the more than
doubling of elderly entitlement spending as a percentage of GDP projected under
current policy really imply any growth in the size of government when the benefits
each individual receives remain the same?
!Spending and tax revenue are not the only ways to think about
the size of the government.
The government can arguably have a bigger (smaller) role in the economy
without spending more (less). Government regulation (in the form of laws,
regulations, or mandates) undoubtedly has an economic impact every bit as real as
spending or taxation, but the cost of regulation is difficult to quantify and is not
measured overall by any official source.23 And yet to ignore it in discussions of the
size of government could be misleading. For example, consider a government
proposal to reduce the consumption of a product (gasoline, cigarettes, alcohol, and
so on). The government could reduce consumption by paying people not to use the
product or paying them to use an alternative, which would be counted as an increase
in government outlays. The government could tax the product to reduce its use by
making it more costly, which would be counted as an increase in government
revenues. Or it could pass regulations forbidding or restricting its use, which would
not show up on either side of the budget’s ledger. Yet all three proposals could be
crafted to have the exact same effect on the quantity of the good consumed.
Effect of the Government on Economic Efficiency
How can policymakers determine the appropriate role of the government in the
economy? Most economists judge the economic merit of any government program
based on its effect on economic efficiency. Does larger government raise economic
efficiency or lower efficiency? The only overarching answer that can be offered is:
it depends. Economic theory is clear that government intervention has the potential
21 Jagdeesh Gokhale and Kent Smetters, Fiscal and Generational Imbalances (Washington:
American Enterprise Institute Press, 2003).
22 See Mario Blejer and Adrienne Cheasty, “The Measurement of Fiscal Deficits: Analytical
and Methodological Issues,” Journal of Economic Literature, vol. XXIX, December 1991,
23 The most comprehensive official estimate of the costs and benefits of regulation is
probably found in Office of Management and Budget, Draft 2005 Report to Congress on the
Costs and Benefits of Federal Regulations, March 2005. It finds that the 10-year cost of
regulations reviewed by OMB was $34.8 billion to $39.4 billion.
to improve efficiency when market failures exist, but is likely to reduce efficiency
when markets are already “perfect,” which is defined below. In reality, government
intervenes both in cases of market failure and in cases where markets are already
operating relatively efficiently.
As a result, some government policies raise economic efficiency and some lower
efficiency. By no means are all spending decisions made by the government today
justified on efficiency grounds. If it were possible to isolate and eliminate all actions
which lowered efficiency, economic welfare could hypothetically be improved by
reducing the size of government. Likewise, one could identify areas where
government intervention could improve currently uncorrected market failures and a
larger government would theoretically improve economic efficiency.
Before discussing what constitutes a market failure, it is useful to define
economic efficiency, which differs from popular parlance. As opposed to its popular
usage, economic efficiency does not involve economic growth, wealth, or
productivity. In fact, one can think of examples where efficiency is at loggerheads
with these goals. Generally, an outcome is economically efficient if the marginal cost
of producing one more unit of a good is equivalent to the marginal benefit of
consuming one more unit of the good. When markets function perfectly, which is
defined as a market with many buyers and sellers, no barriers to entry, perfect
information, and the costs and benefits of the transaction are completely borne by the
buyer and seller, an economically efficient outcome will occur and government
intervention can only reduce efficiency. When there are market failures, government
intervention has the potential to improve efficiency by moving away from the
economically inefficient outcome produced by the market.
While economic efficiency is easy to define theoretically, discord arises when
applied to actual government policies. Typically, an efficiency-enhancing measure
cannot be produced without being accompanied by efficiency-reducing side effects.
For example, without our criminal justice system markets could not operate, but a
criminal justice system cannot be operated without taxes that are likely to take a
efficiency-reducing form. While there is a broad consensus that a tax-financed
criminal justice system is efficiency-improving on net at some level, there is likely
to be disagreement as to whether the benefit derived from a marginal increase in
resources devoted to criminal justice from current levels would exceed the costs of
a marginal increase in taxation to finance it.
Furthermore, the democratic process is conducive to compromises that include
a mixture of efficiency-enhancing and efficiency-reducing measures. Judging the
balance between the two is unlikely to produce wide consensus. Economic theory
can describe the economic benefits (and costs) of a broad policy approach, but cannot
predict how the compromise that emerges from the legislative process will differ
from the policy as originally conceived. Measuring efficiency gains and losses of any
proposal is more difficult when other policies are also distorting a market. (Personal
differences in opinion on these matters go a long way toward explaining the wide
ideological diversity within the economics profession.)
What Is a Market Failure?
To understand when government intervention in the economy can increase
economic efficiency, it is necessary to define a market failure. Before doing so, it is
useful to give examples of what is not a market failure: inequality, poverty, fraud,
discrimination, bankruptcy, layoffs, high prices, and so on are not market failures, as
defined by economic theory. While they are undesirable phenomena which may be
valid targets of public policy, they are problems that either are not economic in
nature, or do not meet the definition of economic inefficiency: they do not involve
a mismatch between marginal cost and marginal benefit. In economic theory the
following are major types of market failures:
Public Goods. There are some beneficial goods that will not be provided by
the market because they are “non-excludable” (people cannot be prevented from
using the good) and they are “non-rival” (one person’s use of the good does not
diminish another’s use). For those two reasons, a private producer has no incentive
to supply the good. Economists refer to goods meeting these two criteria as “public24
goods” which are often provided only by the government. The classic example of
a public good is national defense. While private armies might be capable of
defending the country, there is no incentive to form a private army because nobody
would voluntarily pay for its services (called the “free rider problem.”) That is
because once an army is in place, it has no means to defend its customers from attack
without also defending non-customers. Only government, with the power of
taxation, can raise the funds to finance an army. Similar public goods include basic
knowledge, which once discovered can be enjoyed by all, and the civil and criminal
justice system, much of which (e.g., property rights, dispute settlement, contract
enforcement) makes market transactions possible.25
Because public goods are not transacted through the marketplace, it is difficult
to determine whether government overspends or underspends on their provision. The
value of any given public good is different for different people, and the political
process must sort their preferences rather than the marketplace. For example, could
an acceptable level of national defense be attained at lower cost? Or is the nation not
secure enough at current levels of spending? During peacetime, these questions
cannot be definitively answered.
Common Resources. Another type of good that cannot be efficiently
provided by the market is a common resource, such as the environment, ocean
fishing, and certain water supplies. Unlike public goods, these resources are rival —
a fish or glass of water consumed by one person cannot be consumed by another. But
24 Private producers have found ways to provide some public goods. For instance, broadcast
television and radio is non-rival and non-excludable, but can nevertheless be privately
financed by advertising.
25 To clarify the economic meaning of public good, which may differ from popular usage,
it is useful to point out one government service that does not meet the economic definition
of a public good, education. Education is an excludable good: a school could conceivably
close its doors to keep individuals out if it wished. Education is also a rival good: a book
used by one student cannot be used by another.
the resources are not excludable because they cannot be assigned property rights. For
that reason, they may be overconsumed and can be depleted or even exhausted over
time in the absence of government intervention. As a result, government control or
regulation is necessary for an efficient and sustainable use of the resources. It is no
coincidence that some of the common resources in danger of “depletion” are those
such as ocean fishing and the environment that do not fall within the exclusive
jurisdiction of a single national government — with no single sovereign entity, over-
consumption is harder to prevent.
Although the government has the potential to achieve economic efficiency
through the regulation of common resources, its intervention may not move the
resource closer to efficient use in practice. For example, many governments have
subsidized the fishing industry, potentially exacerbating the depletion of ocean
fisheries. And because common resources also cannot be valued through the
marketplace, estimating the appropriate level of government intervention is difficult.
Monopoly Power. One reason perfect competition leads to economically
efficient outcomes is because any one producer does not have enough market power
to push prices above marginal cost. Monopoly producers can do so to earn
“economic rents” (excess profits) by reducing production to an inefficiently low
level. (Theoretically, there could also be consumers with monopoly power, in which
case the outcome would be economically inefficient because price would be driven
below marginal cost.)
Monopoly can occur for many reasons, including barriers to entry — legal or
natural — and economies of scale. In the narrowest definition of the word, a market
with a single producer, monopolies tend to exist only when marginal cost is
continually declining (producing one more good is always less expensive than
producing the previous good). This special case, known as a natural monopoly, tends
to occur in markets where consumption of a good is non-rival (one person’s
consumption does not come at the expense of another’s); utilities (electricity, water,
cable, telephone) are the most common examples.
In most markets, neither a natural monopoly nor perfect competition exists. The
best description for what does exist is termed monopolistic competition, where each
company makes a product that is distinct but highly substitutable with its rival (e.g.,
the Big Mac vs. the Whopper in the fast-food hamburger market), so that each
company has some market power. In monopolistic competition, production is still
inefficiently low, but closer to the efficient point than in a pure monopoly, and
economic rents do not exist. As the number of firms increases and difference
between products decreases, the monopolistic competition outcome approaches the
perfect competition case.
Economic theory suggests that governments can increase economic efficiency
by increasing a monopoly’s production to its efficient point through regulation or
direct ownership. Both approaches have been used historically for utilities. Over the
past three decades, questions have been raised whether government intervention can
truly raise economic efficiency even in the case of natural monopolies, given the
political intervention, complexity, lack of profit motive, and distorted incentives that
regulation produces for the monopoly.26 Some economists have argued that even
when significant market concentration exists, with the exception of the natural
monopolies, the potential for competition is powerful enough to deter producers from
maximizing monopoly rents even when robust competition does not currently exist.
It is noteworthy that government has rarely attempted to intervene to improve
efficiency in monopolistic competition, even though a theoretical case could be made
to do so. In cases where it has, such as the airline and trucking industries, economic
regulation has been widely deemed a failure and eliminated.
Monopolies are not always less efficient than perfect competition, and
sometimes they are fostered by the government for that reason. For example, as
required by the Constitution, the government grants patents and copyrights so that
inventors and authors can enjoy monopoly profits for their work. Without these
government-created monopolies, there would frequently not be sufficient incentive
to undertake those activities.
Externalities. In the market for many goods and services, all the costs and
benefits inherent in the consumption and production of a good are borne by the buyer
and seller. But some goods also create “externalities” in their consumption or
production. Positive externalities are benefits enjoyed by third parties, negative
externalities are costs borne by third parties. Again, it is difficult to determine how
much government intervention is required to correct an externality since the
externality cannot be valued in the marketplace.
Pollution is the classic case of a negative externality. Society as a whole bears
the cost of environmental degradation, and there is no incentive for the consumer or
producer to take these societal costs into account. As a result, from a societal
perspective the good is overproduced and overconsumed in the free market outcome.
If a good generates a negative externality, it does not mean that good should not be
consumed. It means that to maximize social welfare, the consumption of the good
should be reduced to the level that reflects its social costs.
Vaccines are an example of a positive externality. When someone is vaccinated
against a communicable disease, society as a whole benefits since that person can no
longer contract the disease and spread it to others. From a societal perspective,
vaccines would be underproduced and underconsumed in the absence of government
Some social goals are often popularly justified on the grounds that they generate
positive externalities, but the criteria to qualify as an externality are strict and
economists are divided if social goals qualify. For example, home ownership is often
viewed as generating positive externalities because home owners are viewed as
having higher incomes, having higher rates of civic participation, and committing
fewer crimes than renters. However, it is not clear that home ownership causes
incomes and civic participation to be higher and crime to be lower or if it just
26 For example, see Richard Posner, “The Social Costs of Monopoly and Regulation,”
Journal of Political Economy, vol. 83, no. 4, August 1975, p. 807.
happens to be correlated with other personal attributes that cause these outcomes.27
Similar arguments apply to education.
Asymmetric Information. Competitive markets only work efficiently when
both buyer and seller are well-informed. In some markets, the buyer may be more
informed than the seller, or vice versa. When this happens, the market outcome is
inefficient. For example, in insurance markets, buyers know more about their
riskiness than sellers. As a result, only buyers with higher risks will tend to purchase
more insurance because they are more certain that the benefit of the insurance will
exceed the cost. This pushes up the price of insurance and hampers insurers’ efforts
to pool risk. When government is able to provide information to the uninformed
party or make participation mandatory, it can move the market back to an efficient
outcome. For auto insurance, many state governments make (some) insurance
mandatory to avoid the problem of asymmetric information. For employment
insurance, the government provides the insurance directly. The insurance market is
also distorted by moral hazard, which occurs when the insured party acts more
recklessly as a result of the insurance. For example, some drivers may be more likely
to speed or run red lights when they become insured, pushing up the price of
Asymmetric information is also used as a rationale for financial regulation.
Referred to as the principal-agent problem, the manager of a company or bank may
not have the same incentives (e.g., the costs of risk-taking) as shareholders or
depositors. Information disclosure and accounting laws can increase the information
available to monitor the behavior of managers.
Failure to Optimize. Finally, the assumption in economic theory that people
make rational, optimal economic decisions that maximize their well-being may be
invalid in many cases. Without this assumption, an array of possible government
interventions has the potential to improve well-being. An assumption that
individuals do not optimize underlies diverse arguments such as Social Security is
necessary because people do not save enough, primary education should be
mandatory, the Federal Reserve should prevent stock market bubbles, and drug use
should be illegal.
By its nature, the failure to optimize in any given market is the hardest to prove
or disprove. Much economic analysis is based on the construction of theoretical
models based on optimizing individuals, which are then tested against empirical
evidence. In many cases, the models may be a good approximation of reality not
because everyone is rational, but because any individuals making different mistakes
nearly cancel each other out. A failure of the evidence to match the model is not
necessarily proof that individuals are not optimizing; the model could simply be mis-
specified. Thus far, robust, testable models in which individuals do not optimize
have not attained widespread use.
27 See Edward Leamer and Jesse Shapiro, The Benefits of the Home Mortgage Interest
Deduction, NBER, Working Paper no. 9284, October 2002.
How Do Taxes Affect Economic Efficiency?
Economic theory characterizes taxes as reducing economic efficiency by
“distorting” (changing) the behavior being taxed relative to all other behavior, thus
moving behavior away from the efficient market allocation where marginal benefit
equals marginal cost. (This is the case if the behavior being taxed does not result in
negative externalities; if it does, then the market allocation was not efficient to begin
with.) Thus, a sales tax on a specific good (e.g., passenger air travel) changes the
choice between the taxed good and non-taxed goods, as well as the relationship
between taxed goods and saving. Likewise, a wage tax shifts a worker’s allocation
of his time between labor and leisure away from an efficient balance (the balance that
maximizes his welfare).28 (In the real world, there is also the possibility that taxes
on labor inefficiently shift behavior from taxable compensation to non-taxable
compensation, such as non-taxable benefits.)29
Is economic efficiency reduced on balance by taxation? That question cannot
be answered in isolation of the other side of the government balance sheet. Taxes
finance government, and, as described above, certain government interventions can
increase economic efficiency.30 Whether any given policy increases or decreases
economic efficiency depends on whether the efficiency cost of the taxes needed to
finance it outweighs the policy’s benefits. Furthermore, efficiency losses are difficult
to estimate, particularly in the presence of other policies that change the market
outcome. This is why no broad statement can be made about whether the overall size
of government is inefficiently large or small. Economic theory suggests that the
efficiency loss from a tax increases geometrically as the tax increases.31 This
suggests that as government gets larger (and tax rates get higher), fewer new
government programs would generate efficiency gains large enough to offset the
efficiency losses caused by their financing.32
28 The tax on labor is not inefficient because it reduces labor supply. Theoretically, the tax
could result in less labor supply because the reward to work has diminished or more labor
supply because more work is needed to reach a given income level. It is the change in the
relationship between consumption and leisure — and not the direction of the change — that
makes the tax economically inefficient. These effects are called substitution and income
effects, respectively, and are discussed in detail below in the section on The Size of
Government and Economic Growth.
29 See Martin Feldstein, Tax Avoidance and the Deadweight Loss of Income Taxation,
NBER, Working Paper no. 5055, March 1995.
30 The link between taxes and spending can be temporarily broken through deficit financing.
Deficit-financed tax cuts do not necessarily raise efficiency, however. A $1 deficit-financed
tax cut in effect is a promise that taxes will be raised or spending will be cut by more than
$1 in the future, because of interest costs. Unless inefficient spending is cut in the future,
deficit financing results in a net loss in economic efficiency over time.
31 See, for example, N. Gregory Mankiw, Principles of Microeconomics (Fort Worth:
Dryden Press, 1997), p. 167.
32 When the government provides “private goods,” goods and services that are exclusive and
rival, it is more economically efficient for them to be purchased directly by the consumer
than financed through taxation. Many government-provided “private goods” are already
And yet, the efficiency losses of high taxes should not be overstated. That is
because not all taxes distort behavior. For example, a capitation tax (a fixed dollar
tax that everyone must pay) results in no efficiency loss no matter how high taxes are
set because there is no way to avoid the tax, short of death or emigration. While the
example of a capitation tax is extreme, in general economists assert that the
efficiency losses from taxation could be much lower than they are in our current
system if the tax base were broadened so that an equivalent amount of revenue could
be raised with lower marginal tax rates.33 It is exemptions, deductions, tax credits
and the like that narrow the tax base and require marginal tax rates to be as high as
they are; if the tax base were broadened, spending could be judged more on its own
merits than on the costs of financing it. Unlike the case for a smaller government,
where economists are sharply divided, the case for a broader tax base garners wide
consensus among economists. This suggests that considering the optimal size of
government is best done on the spending side. The efficiency case against taxes is
more a case for tax reform than tax reduction.
The case for broadening the tax base raises a further point: putting aside the
spending side of the budget, not all tax cuts increase economic efficiency. From an
economic perspective, unless they correct for an externality, tax expenditures
(credits, exemptions, exclusions, deductions, preferential rates, or deferrals) reduce
efficiency by creating distortions in economic behavior. In many ways, they are
economically equivalent to spending, and the economic distortions they create are no
different than the distortions created by an equivalent spending policy.34 A $1,000
child tax credit is no different than the government paying every eligible parent
$1,000 per child. Likewise, the mortgage interest deduction is no different than
paying every mortgage-holder in the 15% tax bracket $0.15 for every dollar in
mortgage interest they pay. It is only reductions in marginal rates that increase
economic efficiency. Just like spending, tax expenditures increase economic
efficiency only if they distort behavior that suffers from a market failure, and they are
not generally better or worse than an equivalent spending provision at doing so. If
tax expenditures are not aimed at correcting a market failure, they can divert
economic resources from efficient production to the inefficient pursuit of the
economic rents that the tax code creates, such as “tax shelters.” Other tax
expenditures cause marginal rates to be higher without any meaningful change in the
targeted behavior. For example, there is little evidence that more children are born
as a result of the child tax credit. If that is the case, the credit mainly serves as a
transfer to parents from taxpayers — in economic terms, a “windfall” since parents
need not change anything to receive the credit.
financed through user fees and are omitted from the budget (see the section above on
offsetting receipts and collections).
33 Most economists view the Tax Reform Act of 1986 (P.L.99-514) as a good example of a
tax reform that followed this principle.
34 Tax expenditures raise distributional issues compared to spending. Unless they are
refundable, income tax expenditures are only available to entities with taxable income,
which excludes many lower income households. Furthermore, when tax expenditures take
the form of tax deductions, they are regressive because the value of the deduction increases
as taxable income increases as the taxpayer’s marginal tax bracket increases.
The case for tax simplification is also based on the goals of minimizing
administrative cost, complexity, and evasion, all of which create economic costs in
addition to the efficiency costs discussed above.
Balancing Economic Efficiency With Other Goals
It is crucial to remember that any policy also has non-economic costs and
benefits. Economic efficiency is certainly not the only criterion for public policy.
Equity, fairness, and justice are just a few social goals that economists cannot judge
quantitatively. There are many social goals that policymakers may wish to pursue,
and they often consider whether social benefits outweigh the efficiency loss. For
example, a civil and criminal justice system is not maintained primarily for its effect
on economic efficiency, but for its effect on justice, equity, morality and fairness. A
progressive income tax system is less economically efficient than a capitation tax.
But economics says nothing about whether the capitation tax is more socially
desirable because it cannot weigh the efficiency loss against the equity motive for
having a progressive tax system. No economist can offer a precise estimate of
whether the economic costs and benefits of any proposal will outweigh its social
costs and benefits.35
Economists can, however, evaluate how to achieve a social goal in the least
economically costly way. For example, economists generally applauded the
government’s decision to deal with the social costs of worker displacement from
NAFTA by coupling free trade with worker retraining programs and extended
unemployment benefits for displaced workers. This was viewed by economists as
more economically efficient than preventing working displacement through trade
barriers, so that the gains from trade to consumers can still be enjoyed.36 And in
some cases public sector provision of a social goal may be more economically
efficient than private sector provision. For example, it may not be possible for the
private sector to efficiently provide unemployment insurance because of the
asymmetric information problems described above. Government provision of
unemployment insurance can potentially achieve efficiency while simultaneously
contributing to the social goal of preventing economic hardship. Likewise, some
economists have argued that income redistribution can only be carried out at an
efficient level by the state because it is a public good prone to under-provision in the
marketplace because of the free-rider problem.37
35 Many economists have devised sophisticated theories for achieving optimum income
redistribution based on various social principles. For example, economists have developed
different redistribution schemes based on an equal sacrifice principle that tax rates should
be set by income status such that taxes reduce the welfare of all taxpayers equally (so that
high income individuals pay higher taxes). But there is no way to use economic theory to
judge which theory of redistribution is most fair. For an introduction and bibliography, see
Richard Musgrave, “Fairness in Taxation,” in Joseph Cordes et al., The Encyclopedia of
Taxation and Tax Policy (Washington: Urban Institute Press, 1999), p. 117.
36 See CRS Report RL32059, Trade, Trade Barriers, and Trade Deficits: Implications for
U.S. Economic Welfare, by Craig Elwell.
37 See Harold Hochman and James Rodgers, “Pareto Optimal Redistribution,” American
Effect of the Government on Economic Growth
Arguments surrounding the size of government are often posed in terms of their
effects on economic growth. Like the section on efficiency, this section argues that
the effect of government spending on economic growth can only be judged on a case-
by-case basis. But more generally, the problem with using growth as a policy
evaluation criterion is that it tells nothing about overall welfare, which includes non-
economic benefits, such as quality of life. Unlike efficiency, economists view
growth, at best, as one effect of a policy to be considered and not a goal in and of
itself. Even in cases where the effect on growth is positive, society may be made
worse off. As a thought experiment, consider the effects of a mandatory 80-hour
work week: it would be expected to increase economic growth, but society would be
worse off. Nonetheless, since economic efficiency cannot be easily measured,
growth will often be the best alternative criterion available.
The question of the relationship between the size of government and economic
growth is of a long-term nature. Thus, a distinction should be made between short-
term fluctuations in growth due to the business cycle and the long-term, sustainable
growth rate of the economy. For that reason, arguments for or against larger
government cannot be based on the ability of an increase in the budget deficit to
increase aggregate spending in the economy in the short run. Notice that these short
run effects are consistent with certain definitions of both larger government (higher
spending with constant taxes) and smaller government (constant spending with lower38
Long-term growth (increases in output) is caused by increases in the labor
supply (hours worked or number of workers), the physical capital stock, or
productivity. When the size of the labor force grows with the population, there
would be no effect on per capita growth, however. Growth in the physical capital
stock is made possible by national savings (or borrowing from abroad). For the size
of the government to affect long-term growth, it must affect one of these three
sources of growth. All four types of government behavior (spending, transfers, taxes,
and regulation) have the potential to influence these three sources of growth.
Economic Review, vol. 59, no. 4, September 1969, p. 542.
38 An argument has been made by Fatas and Mihov that larger governments can more
effectively stabilize the business cycle in the short run because they have relatively larger
automatic stabilizers. Automatic stabilizers are changes in spending and revenue that
increase the deficit, thereby boosting aggregate demand, and occur without policy changes.
The authors offer international evidence supporting this theory. This argument can be
criticized, however, on the grounds that there is nothing preventing smaller governments
from using discretionary monetary and fiscal policy effectively, mitigating the reliance on
automatic stabilizers. See Antonio Fatas and Ilian Mihov, “Government Size and Automatic
Stabilizers: International and Intranational Evidence,” Journal of International Economics,
Effect of Spending
A portion of gross domestic product (GDP) is produced by the government.
Like a private firm, the government purchases inputs from the private sector (e.g., the
military purchases tanks and airplanes from private defense contractors) and labor
(e.g., soldiers) to produce a final good or service (e.g., national defense). But unlike
a firm, most government goods and services, like defense, are not bought and sold
in the private market, and so there is no way to value them. Nevertheless, these
goods and services are part of the nation’s GDP. To prepare the GDP accounts, the
Bureau of Economic Analysis values government production, unlike private
production, by measuring the inputs rather than the output. This makes it difficult
to measure the effect of government production on the economy empirically, even if
there are good empirical reasons for believing it to have an effect.
There is little reason to believe government spending has any direct effect on
increases in the labor force since the number of jobs available expands to
accommodate increases in the labor force over the long run, but it may affect
productivity gains over time. Some have argued that the lack of competition and the
profit incentive in the government sector leads to less innovation and lower
productivity gains over time.39 Empirical tests of the proposition may yield little
meaningful information, however, because of the way that government GDP is
recorded. In the private sector, productivity growth occurs when the same inputs
yield higher output than previously. Since government GDP is based on inputs and
output is not measurable (because it is not bought and sold in the marketplace), there
is no direct way to tell how much government productivity is growing over time.40
There is also the question of magnitude: are differences between public and
private sector productivity growth rates significant enough to have a meaningful
effect on economic growth? As a thought experiment, consider that government
output (including state and local) accounted for about 20% of GDP in 2007. If
private sector productivity growth is 2% and government productivity growth is only
half as high, total output growth will only be 0.2 percentage points lower a year than
if productivity were the same in both sectors. In that case, reducing government
production by half would only raise productivity growth by 0.1 percentage points.
(Likewise, if the government’s productivity growth rate were higher than the private
sector’s, the effect would be too small to make much of an impact on the overall
productivity rate.) If anything, this calculation may overstate the productivity
differential since some government output is produced by private sector contractors
who presumably have productivity growth rates similar to the rest of the private
39 See, for example, William Niskanen, Bureaucracy and Representative Government
(Chicago: Aldine-Atherton Publishers, 1971).
40 For a description of a BLS program that attempted to measure government productivity,
see Bureau of Labor Statistics, “The Federal Productivity Measurement Program,” Monthly
Labor Review, May 1997, p. 19. By its estimate, government productivity rose 1.1% a year
from 1967-1994, compared to 1.5% non-farm business productivity growth over that period.
(The article lists several reasons why the government and business productivity are non-
comparable.) This program has been discontinued.
Finally, some government goods and services, while possibly subject to lower
productivity growth in their production, alter private sector productivity growth. For
example, government enforcement of property rights may lead to more
entrepreneurial activity, and a new road may reduce the costs of shipping private
sector goods to market. In these cases, reducing government spending could lower
total productivity growth (by lowering private sector productivity growth) even if
government productivity growth is lower than private sector productivity growth.
Increasing other government goods and services could lower private sector
Thus far, this section has focused on government spending on consumption
goods and services. But government spending can also finance public capital goods,
which increases the national capital stock in the same way as private investment.
Government spending on capital investment (roads, structures, ports, and so on)
increases output by increasing the nation’s capital stock in the same way that private
capital investment does. Annual non-defense capital investment spending has stayed
relatively constant in recent decades at less than 0.5% of GDP at the federal level,
and 2% of GDP at the state level (which is partly financed by federal grants).
Defense capital investment has followed a downward trend in peacetime, from about
defense investment has increased above the trend (although not in recent years). It
is less clear that defense investment has the same positive effect on measured GDP
growth as non-defense investment since it is an investment in the intangible good
“securi t y.”
Figure 5. Public Investment Spending, 1950-2007
State and local
1950 1958 1966 1974 1982 1990 1998 2006
Source: BEA, National Income and Product Accounts, Table 3.9.5.
Note: Data measured on a calendar year basis.
Some would argue that government spending on research and development
(R&D) should be included in measures of investment spending, and some would
include spending on education and training because it increases the nation’s “human
capital stock.” Federal spending on R&D has trended down from about 2% of GDP
in the 1960s to about 1% of GDP today; throughout this period, it has been split
about half and half between defense and non-defense.41
Whether an additional dollar of spending on government output directly
increases or decreases growth depends on what it is replacing. To the extent that
government capital spending replaces private consumption, there will be a net
increase in economic growth through a higher capital stock. Conversely, to the
extent that government consumption spending replaces private investment, there will
be a net decrease in growth. Whether government or private investment yields a
higher rate of return for the economy will vary on a case-by-case basis, and both are
prone to diminishing returns to investment as investment spending is increased.42
Effect of Transfers
Government spending on transfers to individuals has no direct effect on the level
of aggregate private output, aside from administrative costs. Transfers only affect the
distribution of private output: the transfer recipient uses transfer funds to buy goods
and services from the private sector or to save rather than the taxpayer who finances
The effect on economic growth of transfers comes from the distortion in
incentives caused by the transfer, both on the taxpayer, as discussed in the next
section, and the recipient. Since the largest recipients of transfers are retirees, it is
useful to consider the effects on that group. Social Security payments and other
government pensions and age-based transfers may reduce the private saving rate by
replacing private saving for retirement (although the offset would not be one for one
since some retirees do not save adequately for their retirement). Since these transfers
are financed on a pay-as-you-go basis, there is arguably no public saving to offset the
reduction in private saving, reducing national saving. The programs may also lead
to earlier retirement than would otherwise occur, reducing output through a smaller
labor force. Transfers for the elderly also incorporate a number of insurance-like
functions, protecting the elderly against the risk of disability, outliving their assets,
spousal death, and so on. These insurance-like functions may reduce the need for
private precautionary saving.43
41 Government also influences private R&D spending through the R&D tax credit. For an
introduction to the evidence on R&D and economic growth, see Zvi Griliches, “Productivity
Puzzles and R&D: Another Nonexplanation, Journal of Economic Perspectives, vol. 2, no.4,
autumn 1988, p. 9.
42 Some research suggests that public investment can have a higher rate of return than
private investment or increase the rate of return on private investment. See Alicia Munnell,
ed., Is There A Shortfall in Public Capital Investment, Federal Reserve Bank of Boston,
Conference Series 34, June 1990; and David Aschauer, “Is Public Expenditure Productive?”
Journal of Monetary Economics, vol. 23, no. 2, p. 177.
43 For more information, see CRS Report RL31498, Social Security Reform: Economic
Issues, by Jane Gravelle and Marc Labonte.
Are transfers to the elderly ill-advised because they may reduce economic
growth? This example illustrates why efficiency is a better economic criterion for
judging programs than growth. The relevant question is not whether the insurance-
like qualities of these transfers reduce private saving, but whether the government
can provide insurance more efficiently than the private sector because of market
failures in the private insurance market (adverse selection, moral hazard, and
incomplete markets.) In other words, market failures in the insurance market can
lead to individuals saving too much, and government intervention has the potential
to make them better off by reducing the risky contingencies for which they were
previously saving. As a result of government intervention, saving (and economic
growth) would fall, but economic efficiency would increase.
Another major category of transfers are means-tested transfers. Means-tested
transfers can potentially reduce growth by creating an incentive for recipients to keep
income or wealth or hours worked below the point where benefits are phased out.
The incentive can be reduced by phasing transfers out more slowly as income
increases. The effect of means-tested transfers on economic growth are often
weighed against their effect on non-economic goals.
Economists are least fond of transfers that are economically inefficient and do
not serve any broad social goal. For example, subsidies to specific industries or
sectors of the economy may reduce economic efficiency by causing over-production
in those industries. The introduction of an industrial subsidy would be expected to
lead to a one-time reduction in GDP as the economy’s resources are reallocated to a
less efficient outcome. (It is less clear if industrial subsidies would reduce economic
growth on an ongoing basis; they may do so if they reduce competition and the
incentive to innovate.) At the same time, the benefits of the subsidy are highly
concentrated and generally do not accrue to broad groups on a non-discriminatory
basis. For example, workers in the industry being subsidized are made better off
even though their income level and employment situation may be superior to those
in other industries that are not being subsidized.
Effect of Taxes
By changing economic behavior, taxes have the potential to affect overall
growth. Taxes on saving could change saving rates, and thereby investment rates,
and taxes on labor could change the labor supply. Theoretically, income tax
reduction, which affects saving and labor, could increase or decrease economic
growth. That is because a tax cut affects behavior in two ways. First, it increases the
rewards to work relative to leisure, giving the worker an incentive to work more.
This is known as the substitution effect. Second, it increases the worker’s after-tax
income, such that he must work less to maintain his previous standard of living. This
gives him an incentive to work less, and is known as the income effect.
Theoretically, there is no way to know whether the substitution or income effect
dominates for any given tax; it is an empirical question.
Empirically, the evidence is divided on the size and even direction of the effects
on saving and labor.44 Some studies have concluded that labor supply increases in
response to tax cuts, others that it falls. In any case, most studies find the response
to be small. There is evidence that the response of certain demographic groups are
greater than others. Working-aged male adults are overwhelmingly already employed
full-time, so there is relatively little scope for them to enter the labor market or
greatly increase their hours in response to a tax cut. On the other hand, female
employment rates are lower, and there is some evidence that there is a larger labor
supply response to tax cuts among females in high-income households. The labor
supply of workers near the beginning or end of their career may also be more
sensitive to changes in tax rates.45 Casual observation is consistent with small overall
effects on labor. The average work week declined through the 1960s and 1970s, and
has stayed relatively constant since, despite the decline in marginal income tax rates
since the 1980s. The male labor force participation rate has followed a downward
trend since the 1960s, and the female participation rate rose dramatically for much
of the post-war period, but has been relatively constant since the 1990s.
The effect of a change in taxes on labor supply is likely to be a one-time effect:
once individuals have adjusted their labor supply in response to a tax cut, they will
not continue to increase or decrease it in the future. Thus, a change in taxes is
expected to have only a one-time effect on growth through the labor supply channel.
The empirical evidence on saving is particularly inconclusive. At a glance, one
can see that the household saving rate has been in steady decline over the past few
decades — and was close to zero in the past several years — despite a downward
trend in marginal income tax rates and taxes on capital, and a dramatic expansion in
tax-favored savings accounts since the 1980s. Since saving is not motivated solely
by tax rates, one cannot make a simple comparison between the two. To explain the
empirical behavior of saving requires matching the data to some theoretical notion
of why people save. Because the motives for saving are complex, the theoretical
models used by economists become very complex even under simple assumptions.
Not only the level of taxes, but also the structure of the tax code affects
economic growth. To the extent that “loopholes” in the tax code do divert resources
from the pursuit of efficient market activity, this too lowers economic growth,
although it is difficult to estimate the size of this effect, and whether there is a one-
time reduction in growth when the loophole is introduced or an on-going reduction
Evaluating the effects of tax cuts on growth cannot be done in isolation. To
finance a tax cut, spending must be reduced, other taxes must be raised, or the
government must borrow. If a tax cut is financed through lower spending or raising
other taxes, then the effect of those changes on long-run growth (which could be
negative or positive, depending on the policy change) would need to be weighed
44 For more extensive treatment of the issues discussed in this section, see CRS Report
RL31949, Issues in Dynamic Revenue Estimating, by Jane Gravelle.
45 See Orley Ashenfelter and Richard Layard, Handbook of Labor Economics (Amsterdam:
Elsevier Science Publishers, 1986), Ch. 1-2.
against the tax cut. If the tax cut is financed through borrowing, its effect on long-
run growth is likely to be negative since borrowing reduces national saving, unless
the positive effects on private saving and labor supply are large enough to offset the
reduction in national saving.
Effect of Regulation
Estimating the effect of regulation on growth is difficult, and can only be done
on a case-by-case basis. Some regulations reduce output by diverting resources to
non-economic compliance costs. Undoubtedly, other regulations enhance growth,
such as the positive effects of the rule of law on commerce. Some regulations may
reduce a firm’s output directly, such as environmental regulations, but be offset
(partially or wholly) by indirect positive economic effects, such as through an
improvement in public health.
It is important to distinguish between a regulation’s one-time effect on output
(and growth), and its permanent effect on growth. In the long-run most one-time
effects are swamped by the compounding of growth. For example, if environmental
regulations raise a firm’s costs, reducing the demand for their product, their output
will fall. But after the one-time fall, the regulation may have little effect on how
much their output grows in the future.
The government affects economic activity through four primary channels:
government production of goods and services, transfer payments, taxation, and
regulation. Measuring the size of government is not a straightforward exercise.
Relying on annual federal outlays excludes several types of government intervention
including state and local outlays, tax expenditures, government corporations, and
offsetting receipts and collections; as a result, the government appears to be smaller
than it is in reality. Annual measures of government also neglect future unfunded
liabilities implicit in current policy. Measuring the economic impact of regulation
is difficult, and is not done comprehensively or consistently.
Government intervention increases economic efficiency when it rectifies market
failures and reduces efficiency when it distorts perfectly competitive markets.
Political choices may lead to second-best outcomes, however, and some therefore
argue that accepting market failures can be preferable to government intervention in
some cases. To be efficiency-enhancing, government spending must have greater
benefits than the efficiency-reducing taxes that finance it.
Likewise, tax cuts will only increase overall efficiency if they are financed by
reductions in relatively less efficient spending or increases in relatively less efficient
taxes. In general, marginal tax reductions tend to be efficiency increasing, while
many tax expenditures are efficiency decreasing. Deficit-financed tax cuts do not
increase efficiency if they are financed by larger tax increases in the future.
Not all government spending is created equally. Economists universally agree
that some government spending, on a well-functioning legal system, for example,
increases economic efficiency and growth. Agreement is nearly as universal that
some government spending, on subsidies to industries, for example, reduces
economic efficiency or growth. In between, are policies that are a jumble of
efficiency-enhancing and efficiency-reducing provisions. For this reason, no general
conclusions can be drawn about the overall size of the government’s effect on the
economy. It is conceivable that a very large government could devote all of its
spending on efficiency-enhancing policies and a very small government could devote
all of its spending on efficiency-reducing policies, or vice versa.
Many policies that reduce economic efficiency or growth may meet social or
non-economic goals (equity, fairness, and so on) that make them worthwhile.
Currently, government transfers to individuals are nearly twice as large as
government spending on goods and services, and the primary goal of most transfers
is probably not economic efficiency.
Government intervention increases (decreases) long-term growth with policies
that foster (hinder) greater work effort, capital accumulation, and technological
innovation. Economic growth is not a clear normative goal such as economic
efficiency, however. Some growth-enhancing policy measures may reduce economic
welfare. In any case, economic growth has historically been stable over long periods
of time, suggesting that market forces tend to be stronger than the growth effects of
In sum, there is not an economic rationale for either “big” or “small”
government, per se. It is not so much the size of government as what government
does with its spending, transfer, tax, and regulatory policies that affects economic
efficiency and growth