This paper examines the required tax rate in a national retail sales tax (NRST). I show that recent proposals, such as one to replace virtually all federal revenues with a 23 percent taxinclusive NRST, are based on assumptions that real government spending would decline by $480 billion per year, and that there would be no tax avoidance, evasion or political erosion of the tax base in an NRST. Correct for these assumptions indicates that the required tax-inclusive rate would be over 50 percent and the required tax-exclusive rate would be over 100 percent.
The PDF won't load on my computer. Adobe Reader says the file is corrupted or is the wrong file type.
It's Adobe 5.0. Try and download the new Adobe or right-click 'save target as' on the hyperlink above.
Here's an HTML version that I can post: It's a lil' bit garbled, but it'll do.
The Required Tax Rate in a National Retail Sales Tax
William G. Gale
The Brookings Institution
1775 Massachusetts Avenue, NW
Washington, DC 20036
(202) 797-6148
wgale@brook.edu May, 1999
__________
*This paper was prepared for the National Tax Association Spring Symposium, Crystal City,
VA, May 10-11, 1999. The development of the ideas in this paper has benefited significantly
from discussions with Henry Aaron, David Bradford, J.D. Foster, Janet Holtzblatt, Evan Koenig,
Gib Metcalf, Pamela Moomau, Diane Rogers, John Sabelhaus, Joel Slemrod, and seminar
participants at NYU, Georgetown, and the Tax Economists’ Forum. The opinions expressed are
my own and should be not ascribed to the staff, officers, and trustees of the Brookings
Institution.
Page 2
2
ABSTRACT
This paper examines the required tax rate in a national retail sales tax (NRST). I show
that recent proposals, such as one to replace virtually all federal revenues with a 23 percent tax-
inclusive NRST, are based on assumptions that real government spending would decline by
$480 billion per year, and that there would be no tax avoidance, evasion or political erosion of
the tax base in an NRST. Correct for these assumptions indicates that the required tax-inclusive
rate would be over 50 percent and the required tax-exclusive rate would be over 100 percent.
Page 3
3
Over the last few years, proposals to replace much or all of the federal tax system with a
national retail sales tax (NRST) have attracted increased attention. The most prominent
proposals include those by Rep. Dan Schaefer (R-CO) and Billy Tauzin (R-LA) and by a group
called Americans for Fair Taxation. Schaefer and Tauzin (hereafter S-T) would replace existing
personal and corporate income taxes, the estate tax, and some excise taxes with what they call a
15 percent national retail sales tax, levied on a broad consumption base and on government
spending. Americans for Fair Taxation (AFT) would replaces taxes on personal and corporate
income, estates, and payroll with what they call a 23 percent national retail sales tax, on a base
similar to that proposed by S-T.
1
This paper examines the required tax rate in a national retail sales tax. A necessary
precursor to the analysis is noting the difference between tax-inclusive and tax-exclusive sales
tax rates. Suppose a good has a sticker price of $100, not including taxes, and there is a $30
sales tax placed on the good. The tax-exclusive sales tax rate would be 30 percent, or T/P, where
T is the tax payment and P is the price of the good, not including the tax. Sales taxes are usually
quoted in tax-exclusive terms, which correspond to the “mark-up at the cash register.” The tax-
inclusive tax rate would be about 23 percent, or T/(P+T). Income taxes are typically quoted in
tax-inclusive terms. Although there is no “true” method of reporting tax rates, it is crucial to
understand which approach is being used, since the tax-inclusive rate will always be lower than
the tax-exclusive rate, and the difference becomes quite large as the tax rate rises.
2
The tax rates
1
See H.R. 2001, “The National Retail Sales Act of 1997,” and Americans for Fair
Taxation (1997).
2
If the tax-exclusive rate is t, the tax-inclusive rate is t/(1+t). At t=.01, the difference is
negligible, but a 100 percent tax-exclusive rate corresponds to a 50 percent tax-inclusive rate.
Page 4
4
quoted by advocates of the S-T and AFT plans are in tax-inclusive terms. The corresponding
tax-exclusive rates would be about 17.6 percent and 30 percent, respectively.
This paper draws together several issues relating to a NRST and shows that the tax rates
quoted above are based on a series of problematic assumptions. First, the rates quoted above
require either large reductions (on the order of $300-$500 billion per year) in the real size of
government spending, or similar-sized increases in the deficit (or reductions in the surplus). This
hidden effect occurs because of inconsistent assumptions made in calculating the required
revenue. Second, the proposals make the incredibly optimistic assumptions that, under a
national retail sales tax, there would be no erosion of the statutory tax base for political,
economic, or administrative reasons, no tax avoidance behavior (legal behavior that reduces
taxes), and no tax evasion (illegal behavior that reduces taxes).
My estimates show that holding the real size of government constant in the AFT proposal
would require a 35 percent tax-inclusive sales tax rate, or a 53 percent “mark-up” at the cash
register, even if there is no base erosion, avoidance, or evasion. If, in addition, one makes what I
view as conservative, but realistic, assumptions about statutory base erosion, avoidance and
evasion--stipulating that each factor would be positive, but smaller in the NRST than it is in the
existing system-- the required tax-inclusive tax rate in the AFT proposal would be over 50
percent and the required tax-exclusive tax rate would be over 100 percent. The required tax-
exclusive rate in the S-T proposal would be about 60 percent as high.
Section I reviews major features of the sales tax proposals. Section II models and
discusses the treatment of government. Section III examines erosion of the statutory tax base,
avoidance and evasion. Section IV presents estimates of the required sales tax rate under a
variety of circumstances. Section V is a short conclusion.
Page 5
5
I. Overview of the sales tax proposal
Replacement Taxes As proposed by the AFT, the sales tax would replace the individual
income tax, corporate income tax, payroll tax, and estate tax. The S-T proposal would replace
the individual and corporate income taxes, estate tax, and certain excise taxes. The revenues
generated by these taxes in 1995 are shown in table 1.
Tax Base As shown in table 2, the tax bases in the two plans are quite similar. The tax
base would include all goods or services purchased for consumption purposes in the United
States. All personal consumption outlays would be taxed, other than college education, on the
grounds that it is an investment, and food produced and consumed on farms, for administrative
reasons. The proposed base includes the imputed value of financial intermediation services (e.g.,
household purchases of banking services via reduced interest rates on checking accounts).
The treatment of housing merits comment. Retail sales occur when businesses sell goods
or services to households. Thus, sales of newly constructed homes to families that intend to live
in the home would be subject to tax. But sales of an already existing owner-occupied home
would not be taxable, and sale of a newly constructed home to businesses or individuals that
intend to rent the home would not be taxed.
In a pure retail sales tax, business-to-business transactions would not be included either,
because the purchase is used as an input, not as household consumption. The AFT proposal,
however, would include business-to-business sales in the gross tax base and have businesses
apply to government to receive rebates of the taxes they paid on their purchases. Thus, business
purchases would not enter the net tax base, shown in table 2.
Domestic expenditures by U.S. non-residents would be taxed, but the base would exempt
expenditures abroad and half of foreign travel expenditures by U.S. residents. The S-T proposal
Page 6
6
would include state sales taxes in the base, while the AFT proposal would not.
Households consume, directly and indirectly, goods and services that are produced by the
government. Some of these occur via sales, such as national park admissions, but many occur in
the absence of any specific household-government transaction, such as police and fire protection.
Thus, the proposals adopt a "pre-payment" approach to taxing consumption of government-
provided goods and services, and would tax all federal, state, and local government purchases of
inputs, materials, services, wages, and all new government investment. Government would not
pay taxes on transfer payments. If and when the transfer payments were spent on consumption,
they would be subject to tax.
Demogrants The AFT proposal would provide an annual demogrant to each household,
equal to the sales tax rate times the Health and Human Services poverty guideline.
Proposed tax rates The proposals calculate the required tax-exclusive tax rate as simply
the ratio of revenues to be replaced divided by the sum of taxable purchases by consumers and
government less the consumption demogrant:
(1)
s = R/(C+G-X)
where s = the required tax-exclusive tax rate as calculated in the proposals,
R = current nominal revenue raised by taxes that are to be replaced by the sales tax,
C = current nominal expenditures on private consumption that would be subject to the
sales tax,
G = current nominal expenditures on government outlays that would be subject to the
sales tax, and
X = the amount of consumption to which the demogrant applies (=0 in the pre-sales tax
economy).
Page 7
7
Using this formula and the data in table 3, the implied tax-exclusive rates are 16.6
percent for S-T and 29.6 percent for AFT (Table 3). The implied tax-inclusive rates are given by
(2)
s
I
= s/(1+s),
and equal 14.2 percent for S-T and 22.8 percent for AFT. The proposal sponsors rounded these
figures up to 15 percent and 23 percent, respectively, implying tax-exclusive rates of 17.6
percent and 30 percent, as shown in Table 3.
The central theoretical point of the paper is that the required tax rate in equation (1) is
incorrectly specified.
3
The central empirical point is that making the appropriate adjustments
has a large effect on the estimated rates. These points are explored in the following sections.
II. Holding the Real Size of Government Fixed
4
Virtually everyone agrees it is appropriate to hold the real size of government constant
while analyzing fundamental tax reform plans. To examine the role of government, I add the
following definitions:
T = current nominal government transfers to households,
á = the proportion of such transfers that are untaxed (0<á<1), and
D = the current nominal cash-flow budget deficit (D<0 represents a surplus),
Note that G and T include the tax payments that government workers and transfer
recipients, respectively, make under the existing income tax, and that C, G and T represent
expenditures rather than physical quantities. Also, for future reference, in the pre-sales tax
3
Equation (2) correctly shows the relation between the tax-exclusive and tax-inclusive
rate, but of course gives the wrong answer for the tax-inclusive rate if the tax-exclusive rate is
misspecified.
4
This section is based on Gale, Koenig, Rogers and Sabelhaus (1998).
Page 8
8
economy, the government budget constraint is given by
(3)
R = G + T - D,
which indicates that government purchases of goods and services and transfer payments are
financed by a combination of tax revenues and borrowing.
To hold the real size of government fixed after tax reform requires three changes. First,
nominal government expenditures on goods and services must be adjusted for any change in
prices that the government has to pay as a purchaser of those goods and services. Since, in the
AFT and S-T proposals, all federal, state and local government purchases of goods and services
would be subject to federal sales taxation, nominal government spending has to be adjusted to
allow for any changes in the consumer price level (including the sales tax) if the sales tax were to
replace the existing system and the real size of government were held fixed.
Second, to retain their real, after-tax purchasing power, transfer payments need to be
adjusted for changes in prices. This is complicated by the fact that some transfers are currently
subject to income taxation--which would be removed in the switch to a sales tax--and some are
not. It is straightforward, however, to show that transfers that are currently untaxed should be
adjusted with changes in the producer price level, while transfers that are currently taxed at the
same tax-inclusive rate as would occur under the sales tax should be adjusted with changes in the
consumer price level.
5
If all transfers were currently taxed in the income tax at either a zero rate or the rate that
5
For example, a household that currently receives $100 in transfers and pays 20 percent
in income taxes has a net-of-tax transfer of $80. If, in a switch a sales tax with a tax-inclusive
rate of 20 percent, the producer price level stays constant, the household could make the same
net-of-tax purchase as before with a gross-of-tax nominal transfer of $100. That is, changing
nominal taxed transfers in line with the producer price level retains the after-tax value of the
transfers.
Page 9
9
would prevail in a sales tax, á could be thought of as the proportion of transfers that are currently
untaxed. In the more general case, where transfers are currently taxed at a variety of rates in the
income tax, the notion of the proportion of transfers that is untaxed is given by 1-(t
YI
/t
SI
), where
t
YI
is the average tax-inclusive income tax rate on all transfers currently received, including those
taxed at zero rates, and t
SI
is the tax-inclusive sales tax rate.
Third, to hold the real deficit constant, the nominal deficit should change in proportion to
changes in the consumer price level. To model these factors, I examine two cases, one where
nominal wages and the producer price level stay constant after tax reform, the other where
nominal wages and the producer price level fall by the full amount of the previously existing
taxes. In each case, the procedure is the same: (a) develop an equation for tax revenues by
multiplying the tax-exclusive sales tax rate by the statutory tax base, (b) develop a government
budget constraint that equates revenues with government spending on goods and services,
transfers, the demogrant, less the deficit (plus a surplus), all adjusted to hold the real size of
government constant; (c) combine the two equations to solve for the required tax rate.
(A) Nominal wages and producer prices are constant
When nominal wages and producer prices are constant after a switch to a sales tax, the
consumer price level will rise by the full amount of the sales tax; that is, consumer prices,
including the sales tax, will rise by t
1
percent, where t
1
is the required tax-exclusive sales tax
rate.
The statutory tax base is the sum of private consumption and government spending on
goods and services, so the tax revenue equation is
(4)
R
1
= t
1
(C+G),
where R
1
is the nominal revenue requirement under a sales tax.
Page 10
10
Since government spending is part of the statutory tax base, to maintain real government
spending on goods and services, nominal spending must rise by the same percentage as the
consumer price level; that is, it must rise from G to (1+t
1
)G. Nominal government spending on
transfers must rise from T to (1+át
1
)T
6
To maintain the real deficit, the nominal deficit can rise
from D to (1+t
1
)D. The budgetary cost of the demogrants is t
1
X. Thus, the government budget
constraint is given by
(5)
R
1
= (1+t
1
)G + (1+át
1
)T + t
1
X - (1+t
1
)D.
Using (3), (4), and (5) to solve for t
1
yields:
(6)
t
1
= R/(C-X-áT+D).
Equation (6) holds the real size of government constant and differs in several ways from
equation (1) which does not hold government constant. First, (6) does not have a G term in the
denominator. The simple reason why is that adding government spending to the tax base, but
maintaining the real size of government, does not affect the required tax rate. That is, even
though the statutory tax base includes government spending (G), the effective tax base omits G,
because to hold real government spending constant requires an increase in nominal spending that
exactly offsets the tax revenues collected on government spending. This point is both
fundamental and easy to derive: If government purchases were not subject to taxation under the
sales tax, the revenue equation would be R
1
= t
1
C instead of (4), and the spending equation
would be R
1
= G + (1+át
1
)T + t
1
X - (1+t
1
)D, instead of (5). Solving these equations for t
1
gives
the same result as in (6).
6
Nominal spending on currently untaxed transfers would have to rise to (1+t
1
)áT from áT
to retain their purchasing power. Nominal spending on currently taxed transfers would remain
constant at (1-á)T. The sum of the two is given in the text.
Page 11
11
The second difference is that (6) contains terms representing government transfers and
the deficit, which (1) omits. Consumption that is financed with currently untaxed transfers is not
part of the effective tax base because, in order to retain their real purchasing power, such
transfers must be changed with changes in the consumer price level. The effective tax base also
includes the pre-sales-tax deficit. Smaller deficits (or larger surpluses) require higher sales tax
rates.
Thus, all three adjustments--to G, T, and D--that are required to hold government
constant show up in the differences between (6) and (1). A simple, special case occurs when all
transfers are untaxed (á=1), and there is no deficit. In that case, the required tax rate is just t
1
=
R/(C-X-T). Thus, even when government purchases are included in the statutory tax base, the
effective base excludes government purchases, and includes only the portion of consumption that
is not financed by currently untaxed government transfers or demogrants, as long as the size of
government is maintained.
(B) Nominal wages and producer prices decline
The same tax rate formula results when nominal wages and producer prices decline by
the full amount of the removed income tax. In this case, the consumer price level (including the
sales tax) would remain constant after switching to a sales tax. The statutory tax base includes
government spending on goods and services as well as private consumption, but since wages and
producer prices have fallen, each component is deflated by the factor (1+t
2
), where t
2
is the
required tax-exclusive tax rate. Thus, the tax revenue equation is
(7)
R
2
= t
2
(C+G)/(1+t
2
).
Since the consumer price level remains constant, to maintain real government spending
on goods and services, nominal government spending, including the sales tax payments on
Page 12
12
government spending, is held constant. Nominal transfers fall from T to (1+át
2
)T/(1+t
2
).
7
Nominal deficits remain constant since the consumer price level is constant. The demogrant
would exempt X/(1+t
2
) worth of consumption. Thus, the government budget constraint is given
by
(8)
R
2
= G + t
2
X/(1+t
2
) + (1+át
2
)T/(1+t
2
) - D.
Equating (7) and (8) yields:
(9)
t
2
= R/(C-X-áT+D),
the same expression as in (6). This indicates that the required tax rate is not a function of what
happens to the price level after tax reform, as long as the real size of government is held constant
after tax reform.
8
(C) Why is equation (1) wrong?
At this point, it is easy to show why the sales tax advocates’ calculation of the required
tax rates is flawed. Equation (1) can be derived by solving for s using a revenue equation
(10)
R* = s(C+G)
and a government budget constraint
(11)
R* = G + T + sX - D.
The key point is that (10) and (11) are mutually inconsistent. The budget constraint (10)
treats government variables the same as in (4), which assumes that nominal wages and producer
prices are constant after tax reform. The government budget constraint (11) treats government
7
Taxed transfers should fall to (1-á)T/(1+t
2
). Untaxed transfers remain at áT. The sum of
these two is (1+át
2
)T/(1+t
2
).
8
It is easy to verify that--when the producer price level falls after tax reform--taxing or
not taxing government does not change the required rate, as long as the real size of government
is held constant.
Page 13
13
variables (G, T and D) the same way as the budget constraint (9) (and makes the additional
assumption that á=1) which assumes that nominal wages and producer prices fall after tax
reform. These assumptions are obviously inconsistent. As shown in equations (6) and (9),
making either assumption on a consistent basis leads to the same tax rate, and one that is higher
than in (1).
The implied decline in government spending using the tax rate implied by equation (1) is
given by
(12)
G+T - [G/(1+t) + áT/(1+t) + (1-á )T] = {t/(1+t)} [G + T]
where t is the tax-exclusive sales tax rate. The intuition for this equation is easiest to see in the
case where the producer price level stays constant after the switch to a sales tax. In the pre-sales
tax economy, government spending on goods and services and on transfers is G+T. Under the
AFT and S-T proposals, nominal government spending on goods and services, including sales
tax payments, and untaxed transfers is fixed. Therefore, real purchases and the real value of
currently untaxed transfers fall by the factor 1/(1+t). Nominal taxed transfers are also held fixed,
but they retain their original real value, since the producer price level is constant and the income
tax on such transfers has been removed.
9
III. Erosion of the Statutory Tax Base, Tax Avoidance, and Tax Evasion
10
Incorporating erosion of the statutory base, avoidance and evasion into the tax rate
formula is straightforward. Let
9
The last statement is approximate; it assumes that the income tax rate is the same as the
tax-inclusive sales tax rate would be.
10
More detailed analysis of the findings in this section may be found in Gale and
Hotlzblatt (1999) and Gale (1998).
Page 14
14
a = the rate of tax avoidance in a NRST,
e = the rate of tax evasion in a NRST, and
p = the rate of erosion of the statutory sales tax base for “political” reasons, broadly
defined.
Define the actual private consumption tax base in an NRST as
(13)
C* = (1-a)(1-e)(1-p)C
Then the required tax rate is given by (6) or (9), with C* substituted for C:
(14)
t = R/(C*-X-áT+D).
Comparing the tax rates in equations (1) and (14) will indicate the sensitivity of the
required tax rate to various assumptions and specifications. Before turning to that exercise, I
review issues surrounding base erosion, evasion and avoidance in order to help determine
reasonable values to employ for the parameters in (13). In examining these issues, it is worth
noting here--and will be shown below--that adjusting the tax rate just to hold the real size of
government constant would require a tax-inclusive tax rate of 35 percent in the AFT proposal
and a tax-exclusive rate of 53 percent. For the S-T proposals, the comparable rates are 24
percent and 31 percent respectively. Thus, considerations of the likely amount of base erosion,
avoidance, and evasion should made in light of these required rates, not the ones in the AFT or
S-T proposals.
(A) Erosion of the Statutory Tax Base
The sales tax proposals do not allow for any erosion of the statutory tax base; virtually all
consumption is assumed to be taxable. In practice, no income or consumption tax comes close to
meeting that standard. Some items are difficult to tax, some are exempted for reasons of
economic and social policy, and some are exempted due to powerful political constituencies.
Page 15
15
For example, the AFT proposal, net of demogrants, would tax 93 percent of personal
consumption expenditures (PCE), or 63 percent of GDP, in 1995. In contrast, European value-
added taxes (VATs) tax only about 41 percent of GDP (Tanzi, 1995). Feenberg, Mitrusi and
Poterba (1997) use a taxable private consumption base that constitutes 83 percent of personal
consumption expenditures (PCE). Metcalf (1997) develops a taxable sales tax base that is about
80 percent of PCE. The Congressional Budget Office (1997) defines a “broad” consumption tax
base that is about 80 percent of PCE. This base covers all transactions that could be easily taxed
on the product side. A narrower base, that restores some of the preferential treatment that exists
in the income tax is only 60 percent of PCE. Likewise, the sales taxes that already exist in 45
states tax only about half of private consumption of goods and services (Musgrave and Musgrave
1989). No state retail sales tax imposes taxes on imputed financial services, or on housing.
None tax health at the level envisioned in the NRST. Many states exempt food, services and
other items. In light of these considerations, and in light of 53 percent mark-ups even in the
absence of any statutory base erosion, I consider the assumption that political lobbies,
administrative factors, or the desire to introduce economic incentives will reduce the proposed
consumption tax base by 10 percent to be very conservative.
(B) Avoidance
Tax avoidance could arise in several ways. First, taxpayers could combine business
activity--which is generally exempt from retail sales taxation--with personal consumption. For
example, individuals may seek to register as firms, individuals may seek to purchase their own
consumption goods using a business certificate, or employers might buy goods for their workers
in lieu of wage compensation (GAO, 1998).
Second, up to $2,000 per year of goods imported from overseas would be exempt from
Page 16
16
taxation. Thus, buyers could avoid high sales taxes for the first $2,000 of purchases by
purchasing from firms that set up nominal off-shore affiliates. Sales by catalogue, mail or phone
order, or internet may prove especially easy to move off-shore.
Third, because the HHS poverty guideline rises less than proportionally with the number
of family members, the demogrant will, too. This will create incentives in many households for
citizens to try to claim the demogrant as individuals rather than families. For example, if the
(tax-inclusive) sales tax rate were 35 percent, a three-generational family, sharing common
living space, would find that total annual demogrants would increase by over $1,800 if it claimed
the existence of two households (a one-person household consisting of grandpa and a three-
person household consisting of his daughter and her children) rather than one household
(containing all three generations).
11
Likewise, unrelated adults living in the same household,
whether as a couple or simply as roommates, would have large incentives to try to claim their
demogrants as individuals rather than as a two-person household.
12
For all of these reasons, it is implausible to assume that the avoidance rate will be zero in
a NRST. I assume it will be 5 percent, which seems relatively low compared to the amount of
avoidance and sheltering that occurs in the income tax and other systems.
(C) Evasion
Sales tax proponents point to several factors that they believe imply that evasion will be
lower under the sales tax than under the income tax (Mastromarco 1998). These include the
11
Using 1998 data, with 4 persons in the households, the demogrant would be 35 percent
of $16,450, or $5,758. With a 1-person and a 3-person household, the demogrant would be 35
percent of $8,050 (for the individual) plus 35 percent of $13,650, for a total of $7,595.
12
For the same reasons, every married couple would face a sizable marriage penalty in the
sales tax (Gale 1998).
Page 17
17
presence of fewer taxpayers, which implies a higher proportion of taxpayers could be audited,
and a simpler tax system, which implies fewer unintentional errors. However, these claims are
probably overstated. Gale and Holtzblatt (1999), for example, note that if taxpayers have to file
monthly under an NRST, as has been proposed, the number of filings will rise, offsetting gains
from fewer taxpayers. In addition, the majority of filing requirements that would be eliminated
under the NRST would cover relatively simple returns that were already effectively completely
audited by the IRS because of information reporting requirements and withholding.
The presence of off-shore sales could pose evasion problems as well as the avoidance
problems mentioned above. For example, buyers could purchase more than $2,000 from off-
shore entities and not pay taxes on the purchases. Experience with state-level “use” taxes and
voluntary filings has been “dismal at best” (Murray 1997). The importation of services,
especially direct delivery of electronic and information services could cause additional evasion
problems.
Another potential problem is the appearance of many small, tax-evading retailers, who
are able to set up businesses with low overheads, and are effectively created by the prospect of
being able to undercut legitimate retailers by large margins. It will also prove very difficult to
collect high-rate sales taxes from a number of small-scale service industries--taxi cab drivers,
plumbers, handy men, painters, maids, etc. are classic examples. More generally, the two parties
to a sale will have incentives to report lower-than-accurate transaction prices to the government
coupled with side payments. The true transaction price and the side payments could easily be
shielded via complex financing arrangements.
One of the most important determinants of the level of evasion in the current system is
whether anyone other than the taxpayer withholds taxes or reports the tax to the government.
Page 18
18
The rate of evasion is currently around 17 percent in the income tax, but varies greatly by
withholding and reporting arrangements. For income where taxes are withheld and reported to
government by a third-party, the evasion rate is about 1 percent. This is predominantly
withholding of taxes on wages. At the other extreme, for income where taxes are not withheld
and there is no reporting, the evasion rate is 30 percent or more. In contrast, the pure retail sales
tax would be collected only from businesses that make retail sales, and there would be no
withholding or reporting by anyone other than the business itself. That is, the entity reporting
the tax payment would also be the entity legally responsible for the tax liability. Because the
pure retail sales tax would feature no third-party withholding or reporting, the possibility of high
rates of evasion needs to be taken quite seriously.
Sales tax advocates sometimes claim that the NRST would be more effective than the
current system at raising revenue from the underground economy. The classic example is that of
a drug dealer who currently does not pay income tax on the money he earns, but would be forced
to pay sales taxes under an NRST if he took the funds and bought, for example, a Mercedes. The
problem with this argument is laid out by Armey (1995). "If there is an income tax in place, he
[the drug dealer] won't report his income. If there is a sales tax in place, he won't collect taxes
from his customers" and send the taxes to government. In the end, to a first order approximation,
neither system taxes the drug trade. Some additional effects, though, may complicate the
analysis. For example, the effective tax rate on drug dealers and their customers may differ, and
the drugs may be purchased with income generated illegally.
Many other countries have attempted to implement some variant of a sales tax such as
wholesale level taxes or "ring" taxes (retail sales taxes with business exemptions certificates for
businesses "in the ring.") Almost all of these countries have switched to a value-added tax. In
Page 19
19
1967, 19 OECD countries had some form of wholesale, retail or turnover tax. By 1995, all 19 of
these countries had converted to value-added taxes. Developing countries as well have largely
abandoned retail sales taxes in favor of value-added taxes. Countries that have relied on RSTs
tend to charge around 4-6 percent of the tax exclusive value of goods, while standard VAT rates
tend to range between 14 percent and 25 percent. No country has maintained a sales tax rate
above 30 percent.
Ultimately, my reading of the evidence is that the interaction of (a) incentives to evade
taxes, caused by the lack of third-party withholding and the high marginal tax rates that would be
required even if there were no evasion, and (b) the political pressure to exempt certain goods,
given the high sales tax rate, will create somewhat of a vicious cycle: each factor raises the
required rate and increases incentives to raise the other factor, which in turn raises required rates
again, and so on. Thus, it is quite likely that evasion in the sales tax would be at least as high as
in the current system and it is quite possible that the sales tax would prove to be unworkable. It
is worth emphasizing that these conclusions are by no means extreme relative to the literature.
Concern and doubt about the adminstrability and enforceability of a high-rate national retail
sales tax have also been expressed by: Bartlett (1995), Casanegra (1987), McLure (1987),
Mikesell (1997), Murray (1997), OECD (1998), Slemrod (1996), Tait (1988), Tanzi (1995), and
the Wall Street Journal (1996).
13
Possible administrative improvements in the administration of a NRST, discussed by
13
Unfortunately, there is little evidence from the states to gauge how extensive evasion
would be under a NRST. Evasion in a federal sales tax would likely be significantly higher than
in existing state taxes, because state sales tax rates have substantially lower rates than an NRST
would, and states can piggyback on federal enforcement efforts, which are in turn aided by the
existence of an income tax with its various reporting requirements.
Page 20
20
Zodrow (1999), would likely reduce evasion, as would a VAT, compared to a retail sales tax.
But no one would argue seriously that the changes would eliminate tax evasion.
In light of these considerations, I assume below that the rate of evasion in an NRST
would be 15 percent. It should be emphasized that this is significantly lower than what I would
regard as a plausible evasion rate in a sales tax, but the goal here is to provide conservative
assumptions that, if anything, understate the required tax rate. Note also that this evasion rate is
lower than in the existing system, even though most revenues in the existing system are collected
via withholding of taxes on wages for the income and payroll taxes, a type of income for which
the evasion rate is very low.
III. Estimates of the required rate
A. Parameter Estimates
The data needed to generate tax rate estimates are provided in table 4 for the AFT
proposal.
14
All data refer to 1995.The proposed consumption base and government spending are
taken from table 2. The demogrant estimate is taken from Americans for Fair Taxation (1997). I
consider two specifications of government: one where real federal, state and local government is
held constant, and one where only the federal government is held constant. In the latter case,
purchases by state and local governments are taxable and thus are placed in consumption.
15
14
To simplify calculations, I use the same tax base for the AFT and the S-T proposals. In
practice, the proposed bases differ somewhat in their treatment of state and local taxes,
housing, and tax-exempt organizations, and their calculation of the demogrant. Because the
differences between the proposals largely offset each other in terms of their effects on the size of
the tax base, incorporating the differences would have only small effects on the results.
15
When holding only the size of federal government constant, the consumption figure is
the sum of private consumption ($4,381.7 billion) plus state and local purchases of $787.5
billion. The state and local purchase figure does not equal the sum of state and local
consumption and gross purchases shown in table 2, because the latter figure includes
Page 21
21
To estimate T, note that federal transfers to individuals totaled $709.4 billion in 1995 and
state and local transfers totaled $280.6 billion, for total transfers to persons of $990 billion. The
federal government made net interest payments of $224.8 billion, while state and local
governments made net interest payments of -$59.6 billion (that is, they received net interest).
Thus, total net interest payments were $165.2 billion, total transfers were $1,152.7 billion, and
total federal transfers were $934.2 billion (NIPA, Tables 3.1, 3.2 and 3.3).
Estimating á precisely is difficult and only a rough calculation is provided. The
following transfers can be subject to federal income taxation: net interest payments ($181.7
billion), 50 percent to 85 percent of social security benefits ($163.8 billion to $278.5 billion),
unemployment insurance ($21.7 billion), federal employee retirement ($67.3 billion), and state
and local employee retirement ($66.3 billion) (NIPA table 3.12). If only 50 percent of social
security benefits are subject to taxation, these transfers total $501 billion in 1995 or 43.5 percent
of total transfers, suggesting á=0.565. If 85 percent of social security benefits are taxed, taxed
benefits equal $616 billion in 1995, suggesting that á=0.466.
These figures should be raised, however, for two reasons. First, not all taxable transfers
are actually taxed (because of the zero tax rate bracket in the income tax). For example, only 15
percent of social security benefits and two-thirds of net interest paid on government debt are
taxed (Congressional Budget Office 1997, p. 16). Making these adjustments would reduce
taxable social security benefits by $114 billion and taxable net interest by $60 billion. Thus,
only $327 billion, or 28 percent, of transfers would be taxable, suggesting that á=.72. Likewise,
preliminary calculations using the Joint Committee on Taxation (JCT) individual tax model
depreciation of government capital, which would not be taxed under a sales tax.
Page 22
22
suggest that over three quarters of transfers are not taxed.
Second, some--presumably large--proportion of taxed transfers face the 15 percent rate
in the existing system, which is below the rate that would apply in a national retail sales tax.
Preliminary calculations from the JCT model indicate an average tax-inclusive tax rate of
roughly 21 percent on the small minority of transfers that are actually taxed.
16
This is below
what would be required in the AFT proposal, which should raise á further. Thus, setting á=0.75
may understate the true value. Higher values of á would generate higher sales tax replacement
rates than reported below.
To estimate D, note that the federal budget deficit measured on a current basis was
$174.4 billion in 1995. To this figure, we add federal gross investment ($66.1 billion) and
subtract depreciation of federal capital ($68.2 billion), for a federal cash flow deficit of $172.3
billion. The corresponding state and local figures are a current surplus of $103.1 billion, gross
investment of $147.4 billion, and depreciation of $54.2 billion, for a cash flow surplus of $9.9
billion. The overall government cash flow deficit was therefore $162.4 billion (NIPA tables 3.1,
3.2, 3.3, and 3.7).
To hold only federal government constant, we use estimates of G (or G*, as noted), T,
and D based on federal statistics only. State and local spending on goods and services is placed
in C--that is, it is treated just like private consumption. This raises C to $5169.2 billion. In this
case, state and local transfers are not protected against price changes.
B. Estimates of the Required Tax Rate
The required tax rate estimates are presented in table 5. The first panel shows the rates
16
I thank Pamela Moomau for providing the estimates using the JCT model.
Page 23
23
required to replace taxes in the AFT proposal: personal and corporate income taxes, the estate
tax, and payroll taxes. The first row shows the AFT proposal would require a tax-inclusive rate
of about 23 percent and a tax-exclusive rate of about 30 percent. But it also notes that this would
involve a reduction in real federal, state, and local spending of $483 billion (using equation (12))
and does not allow for any evasion, avoidance, or erosion of the base. The second row shows
that holding the real size of federal, state and local government constant would raise the tax-
inclusive rate to 35 percent, and the tax-exclusive rate to 54 percent. The third row shows that
the relatively conservative parameter choices used--a 5 percent rate of tax avoidance, a 15
percent rate of tax evasion, and a 10 percent rate of erosion of the statutory tax base--boosts the
required tax rates markedly, to over 100 percent on a tax-exclusive basis and over 50 percent on
a tax-inclusive basis.
The last two lines of the first panel show the effect of preserving only the size of the
federal government, letting real state and local spending fall by $219 billion (or about 22
percent). With conservative (but positive) rates of avoidance, evasion, and statutory base
erosion, the required tax-inclusive rate would be 39 percent, while the required tax-exclusive rate
would be 65 percent.
The second panel shows the required rates when replacing only the taxes specified in the
S-T proposal: individual and corporate income taxes, estate taxes, and selected excise taxes. The
required tax-exclusive rates are roughly 60 percent as high as in the first panel, because the
revenue target is lower (as shown in table 1).
These estimates could be extended in a number of ways. For example, if the sales tax
spurred economic growth, the rates would be lower in the long run. If consumption grew by 10
percent, which is on the optimistic end of forecasts for the economic growth due to realistic tax
Page 24
24
reforms, the tax-exclusive tax rate in case (c) in the top panel of table 5 would fall from 101
percent to 82 percent, while the tax-inclusive tax rate would fall from 50 percent to 45 percent.
However, in the short run, as consumption fell (saving increased), the required tax rate would
have to increase.
On the other hand, in 1995, state sales taxes and personal and corporate income taxes
raised about $410 billion in revenues. Moving to a NRST would likely require states to abandon
or significantly alter their income taxes and modify their sales tax. A plausible outcome would
be that states converted their income taxes to sales taxes (otherwise there would be little net
simplification from a national sales tax) and conformed their sales taxes to the federal sales tax
base. If this were done, the average combined federal and state sales tax-exclusive tax rates
would be about 30 percent (not percentage points) higher in the top panel of table 5 and more
than 50 percent higher in the bottom panel.
V. Conclusion
Proposals to replace the existing federal tax system have received increased attention in
recent years, but the claims made on behalf of some of these proposals are not credible. In
particular, claims that a national retail sales tax could replace the income tax system, the estate
tax and payroll taxes at a 23 percent tax-inclusive (30 percent tax-exclusive) tax rate are
fundamentally flawed. This paper shows that such claims hinge on hidden reductions in real
government spending of almost half a trillion dollars per year, and on the impossibly optimistic
assumptions that there will be no tax avoidance, no tax evasion, and no statutory erosion of the
proposed tax base. Corrections for these problems show that plausible national retail sales taxes
will require--at the federal level--tax-inclusive rates on the order of 50 percent and tax-exclusive
rates on the order of 100 percent. Accounting for economic growth would reduce this rate, but
Page 25
25
consideration of state and local taxes would raise the rate by even more than growth would
reduce it. Finally, given the experience of other countries with high-rate retail sales taxes, there
have to be serious questions about whether a sales tax with rates that high could actually be
enforced.
Page 26
26
References
Americans for Fair Taxation. “The National Retail Sales Tax, Tax Evasion and the Underground
Economy.” [
http://www.fairtax.org/impact/nationalretail.htm]. 1997.
Armey, Richard. “Caveat Emptor: The Case Against the National Sales Tax.” Policy Review.
(Summer 1995): 31-5.
Bartlett, Bruce. “Replacing Federal Taxes with a Sales Tax.” Tax Notes 21 Aug. 1995, 997-
1003.
Casanegra de Jantscher, Milka. “Problem in Administering a Consumption Tax,” in The
Consumption Tax, edited by Charles E. Walker and Mark A. Bloomfield. (Cambridge: Bullinger,
1987).
Feenberg, Daniel R., Andrew W. Mitrusi, and James M. Poterba. “Distributional Effects of
Adopting a National Retail Sales Tax.” Tax Policy and the Economy 11 (1997): 49-89.
Gale, William G. (1998), "An Evaluation of the National Retail Sales Tax," mimeo., The
Brookings Institution, February 17, 1998.
Gale, William G. and Janet Holtzblatt (1999), “The Role of Administrative Factors in Tax
Reform: Simplicity, Compliance, and Enforcement,” mimeo., Brookings and Treasury.
Gale, William G., Evan Koenig, Diane Lim Rogers, and John Sabelhaus. “Taxing Government
in a National Retail Sales Tax.” Tax Notes (October 5, 1998): 97-109.
Mastromarco, Dan. “The ‘Fair Tax’ and Tax Compliance: An Analytical Perspective.” Tax
Notes. (April 20, 1998): 379-387.
McLure, Charles E. Jr. The Value-Added Tax: Key to Deficit Reduction? (Washington, D.C.:
American Enterprise Institute, 1987).
Metcalf, Gilbert E. “The National Sales Tax: Who Bears the Burden?” Cato Policy Analysis no.
289, 1997.
Mikesell, John L. “The American Retail Sales Tax: Considerations on their Structure,
Operations, and Potential as a Foundation for a Federal Sales Tax.” National Tax Journal 50
(1997): 149-65.
Murray, Matthew N. “Would Tax Evasion and Tax Avoidance Undermine a National Retail
Sales Tax?” National Tax Journal 50 (1997): 167-82.
Musgrave, Richard A., and Peggy B. Musgrave. Public Finance in Theory and Practice. Second
Edition. New York: McGraw-Hill Book Company, 1976.
Page 27
27
The National Retail Sales Tax Act of 1997. H.R. 2001, 105 Cong. 1 sess. 1997.
Organization for Economic Cooperation and Development, Consumption Tax Trends. (Paris:
OECD, 1998).
Slemrod, Joel. “Which Is the Simplest Tax System of Them all?” In Economic Effects of
Fundamental Tax Reform, edited by Henry J. Aaron and William G. Gale, 355-391.
Washington, D.C.: Brookings, 1996.
Tait, Alan A. Value Added Tax: International Practice and Problems. Washington, D.C.:
International Monetary Fund, 1988.
Tanzi, Vito. Taxation in an Integrating World. Washington, D.C.: Brookings Institution, 1995.
United States Congressional Budget Office. Comparing Income and Consumption Tax Bases.
Washington, D.C.: Congressional Budget Office, 1997.
United States General Accounting Office. Tax Administration: Potential Impact of Alternative
Taxes on Taxpayers and Administrators. GGD-98-37. Washington, D.C.: General Accounting
Office, 1998.
“VAT in Drag.” Wall Street Journal, July 19, 1996.
Zodrow, George R., “The Sales Tax, The VAT, and Taxes in Between--or, is the Only Good
NRST a “VAT in Drag,” National Tax Journal, September, 1999.
Nope, I have 6.0. No go either way.
Thank you. That helps.
AFT used a tax base of 93% of all retail goods. This clown used 50-60%. What a schmuck.
He also assumed that wages would go down somehow.
Mr. LINDER. Mr. Speaker, I yield myself such time as I may consume.
Mr. Speaker, I want to say only this: those who choose to not put a sales tax on the Internet are picking winners and losers. The Government ought to be neutral. Our neighbors down the street ought to have the same treatment as the people that sell on the Internet in competition with them.
Bill Gale lies.
Will take me some time to pick his argument apart, but I'll get around to it sometime today.
I kinda figured you'd jump in here, LurkeyLiarLooneyLou.
You are quoting figures the LIEberal leaning Joint Committee on Taxation prepared for Rep. Charlie Wrangle for his use at the House Ways and Means Committee hearing in April, 2001.
All the Democrat Members of W&M quoted that figure, which was and remains, bogus.
The Joint Committee "scoring" of the FairTax was based almost entirely on Bill Gale's false assumptions as contained in his 1999 "study" of the FairTax. They even used some of the same language, without attribution, BTW.
And the possibility that any of those things would actually happens is . . . . ?