CONSUMPTION TAXES AND THE LEVEL AND COMPOSITION OF SAVING
CRS Report for Congress
Consumption Taxes and the Level and
Composition of Saving
Updated January 11, 2001
Steven Maguire
Analyst in Public Finance
Government and Finance Division
Congressional Research Service The Library of Congress
Consumption Taxes and the Level and Composition of
Saving
Summary
Many proposals for tax reform would shift the base of taxation from income to
consumption. Two arguments for a consumption tax are frequently made. One, a
consumption tax is argued to encourage saving, at least more so than the current
income tax. And two, a consumption tax is often considered a more efficient means
of tax collection. In this paper, the merits of these arguments are analyzed. To
accomplish this, the present income tax is compared to a generic consumption tax.
By definition, a consumption tax does not include non-consumed accretions to
wealth in the tax base. Hypothetically, each individual has a savings account where
unspent income accumulates. Over the course of the year, all net gains to the account
are deducted from wage earnings and net losses to the account added. The resulting
total is the individual’s consumption and represents the taxable base. The appropriate
tax rate is then applied to the derived annual consumption. All saving is in effect free
from taxation.
However, the added saving incentive (or lack of disincentive) implicit in a move
to a consumption tax may not be as large or effective as anticipated; the current tax
code already favors saving through a variety of income tax advantages for non-
consumption. Further, to the extent a consumption tax does ease the tax burden on
saving, evidence is ambiguous on the magnitude of response that would occur. Thus,
a shift from the current income tax to a consumption tax may not produce as large a
gain in saving as some theorists suggest. It would, however, likely result in a
reallocation of existing saving among different saving vehicles. The purpose of this
report is to explore the current tax treatment of non-consumed income and to analyze
the likely effect on saving, specifically retirement saving, of adopting a consumption
tax.
Contents
Saving Under Current Tax Laws Compared to a Consumption Tax..........2
Social Security..............................................4
Tax-Favored Employer-Sponsored Pensions.......................4
Defined Benefit Pension Plans..............................5
Defined Contribution Pension Plans..........................6
Tax-Favored Personal Saving...................................7
Individual Retirement Accounts (IRAs).......................7
Life Insurance..........................................9
Owner Occupied Housing.................................9
Taxed Personal Saving.......................................10
Equities .............................................. 10
Financial Intermediary Instruments..........................10
Direct Business Investment...............................11
Summary ................................................. 11
Saving Literature Review.........................................12
Tax Reform and Saving Literature..............................12
Specific Saving Vehicle Literature..............................14
IRAs ................................................ 14
401(k)s .............................................. 15
National Saving Literature....................................16
The Evidence in Sum........................................17
Conclusion ................................................... 17
References .................................................... 19
Consumption Taxes and the Level and
Composition of Saving
This report examines how the composition of saving might change with the
introduction of a consumption tax. Specifically, the report analyzes the possible
impact of a consumption based tax system on various forms and levels of saving1.
In 1999, annual personal saving in the United States was 2.2 percent of
disposable personal income, much lower than in other developed economies.2
Another measure of saving, gross saving as a percent of gross national product,
averaged 17.1% over the 1990's (1990 to 1999), a significant decline from 20.9% in
the 1960's (1960 to 1969).3 The cause of the relatively low saving rate is a frequent
topic of study for economists as well as policy makers. Many economists suggest that
the design of our tax code alters the saving decision, perhaps partially explaining the
relatively low saving balances. In response, policy makers have introduced legislation
in several recent congresses which taxes consumption instead of income hoping to
encourage more saving as well as improve economic efficiency.4 Logically, if
individuals can either consume or save and consumption is taxed and saving not, then
saving should increase and consumption decline under a consumption tax.
Several forms of consumption taxes, which are thought to encourage saving,
have been proposed in past years. Three prominent proposals in previous sessionsth
include the Nunn-Domenici USA Tax (104 Congress, S. 722), the Armey Flat Tax
(106th Congress, H.R. 1040) and the Tauzin retail sales tax (106th Congress, H.R.
2001). This trend may well continue through the 107th Congress. A review of tax
reform legislation in the 106th Congress can be found in CRS Issue Brief IB10013th
Major Tax Issues in the 106 Congress: A Summary. The stylized consumption tax
analyzed for this report can be modified to resemble any of the above proposals.
1For a textbook comparison of the consumption tax to the income tax, see: Musgrave, Richard
A., and Peggy B. Musgrave. Public Finance in Theory and Practice. Fifth Edition. New
York, McGraw-Hill, Inc., 1989. p. 224-226.
2Department of Commerce, Bureau of Economic Analysis. NIPA basis for aggregate savings
which reports the difference between disposable personal income and expenditures.
3For an overview of the various measures of saving, see CRS Report 98-580E, Saving in the
United States: Why Is It Important and How Has It Changed?, by Brian W. Cashell and Gail
Makinen.
4For a detailed taxonomy of the theoretical variations of the consumption tax, see CRS
Report 95-1141E, The Flat Tax and Other Proposals: Who Will Bear the Tax Burden. , by
Jane G. Gravelle, and CRS Issue Brief IB95060, Flat Tax Proposals and Fundamental Tax
Reform: An Overview, by James M. Bickley.
However, the added saving incentive (or lack of disincentive) implicit in a move
to a consumption tax may not be as large or effective as thought; the current tax code
already favors saving through a variety of income tax advantages for non-
consumption. Further, to the extent a consumption tax does ease the tax burden on
saving, evidence is ambiguous on the magnitude of response that would occur. A
consumption tax, however, likely would result in a reallocation of existing saving
among different saving vehicles. The purpose of this report is to explore the current
tax treatment of non-consumed income and to analyze the likely effect on saving,
specifically retirement saving, of adopting a consumption tax.
The first section describes the current tax treatment of various saving vehicles
and analyzes the effect on particular saving instruments when moving to a typical
consumption tax. The second section reviews the current literature addressing the
effectiveness of saving incentives. And the last section offers some concluding
comments.
Saving Under Current Tax Laws Compared to a
Consumption Tax
The present federal tax structure has two basic vehicles for revenue collection:
a payroll tax and an income tax. A first step in understanding the preferences for each
saving vehicle is to identify the location along the income and expenditure cycle where
the tax preference is applied. The payroll tax is applied at the earliest stage, when
wages are paid, and the employer and employee each pay an equal share. Here,
deferred compensation, such as employer contributions to a pension plan, are not
included in wages and thus are not subject to the payroll tax.5 This is the first tax
preference for long term saving. The remainder of the paycheck (including the
employee’s share of the payroll tax) is considered taxable income for the income tax.
Some wages paid after the payroll tax can, however, avoid the income tax.
Income, as typically defined in economics, includes all accretions to wealth over
the tax year and as such includes capital income as well as wage income. The second
tax preference appears at this stage. The "inside build-up" arising from investments
in certain tax favored retirement accounts (including IRAs) generally are not included
in the income tax base. Alternatively, most investments in liquid, non-retirement
vehicles do not enjoy such treatment and are subject to the income tax. The third tax
preference arises from the tax deferral of some deposits into retirement accounts. All
else equal, rational taxpayers prefer to pay taxes later rather than sooner. The
simplest example is an IRA where up to $2,000 (single filer) may be deductible and
thus not subject to the income tax. For an individuals in the higher marginal income
tax brackets, this advantage translates into a significant tax saving. When the funds
are withdrawn at retirement the income tax is applied, usually when the taxpayer
resides in a lower marginal tax bracket. A variation of this advantage is the back-
5Assuming that the retirement plan qualifies for such treatment. While 401(k) salary deferrals
are thought to be "elective employer contributions," these amounts are subject to the payroll
tax.
loaded "Roth IRA" where, instead of deposits being income tax free, withdrawals are
income tax free.
Summarizing, the current tax code has three primary tax preferences for saving:
payroll tax exemption, tax free "inside build-up", and tax deferral on funds deposited
until savings are received.6 The tax preferences generally apply to long term saving
rather than short term saving–for example, to pensions and not bank deposits– and7
accordingly the next section focuses primarily on long term saving instruments.
Moving to a consumption tax would have two countervailing effects. The
removal of the disincentive on short term saving (e.g., time deposits such as
certificates of deposit) will appear as an increase in future income (the rate of return
is now higher given interest earnings are not taxed) and individuals may respond to
the tax cut with increased consumption today and lower saving. In other words, less
needs to be saved today to meet future expenditure targets. Conversely, the
consumption tax will also raise the price of current consumption and lower the price
of future consumption (saving). Taxpayers may sacrifice current consumption for
future consumption which is now less expensive. This move away from current
consumption would counter the income generated push for greater consumption8
today. Which effect dominates dictates what happens to overall savings. Some
observers believe that retirement saving, or more generally long term saving, will
decline and short term saving increase as short term investments gain tax equivalency.
Another layer of analysis might explore the generational effects of the two
principal forms of a consumption tax.9 A wage tax and a retail consumption tax are
theoretically identical for the young. The same level of revenue can be raised through
collecting a fixed percentage of wage payments or a fixed percentage of consumption
expenditures, the only difference being the rate of taxation. The consumption base
is smaller thus the fixed percentage would have to be higher to raise the same level
of revenue.
Assuming there are two types of taxpayers, older and younger, where the older
generation is dissaving and the younger saving, the type of tax will have differential
effects on these cohorts. A retail consumption tax would increase the burden on the
older Americans who are dissaving and the younger generation would enjoy a
relatively lower tax burden. A wage tax would have the opposite effect where older
Americans, assumed to be living off of accumulated wealth, would recognize a tax
reduction to be paid for by the younger cohort. Thus, underlying the analysis is the
potential for differential distributional effects generated by the style of consumption
6See: Samwick, Andrew A. Tax Reform and Target Saving. NBER working paper 6640.
July 1998. He recognizes the same tax preferences. Tax-free refers to the fact that annual
gains to the fund are not taxed.
7In addition to providing different tax consequences, short and long term saving are motivated
by different objectives.
8Economists refer to these effects as the income and substitution effects respectively.
9Another aspect is the overall objective of the tax reform. We assume throughout this report
that the tax change is revenue neutral.
tax chosen. For ease of exposition, this paper assumes taxpayers do not change their
consumption patterns over the life-cycle.
Following is a more detailed review of savings vehicles categorized by their tax
treatment. The current tax preference (if any) of the vehicle is described then the
anticipated effect of a consumption tax on the vehicle is discussed. Theoretically,
assets in vehicles previously tax advantaged will shift to vehicles now on equal footing
in the tax code. The list is by no means comprehensive.
Social Security
Social Security is perhaps the most commonly known source of retirement
savings. Social Security is funded by the payroll tax where almost every working
American contributes 7.65% of their salary before income taxes to the Social Security
trust fund. The 7.65% is composed of 6.2% for Old Age and Survivors Disability
Insurance (OASDI) retirement up to adjusted gross income of $72,600 (1999 tax
year) and 1.45% for medicare (HI) which applies to all wages regardless of income.
Employers match the employee contributions. Calendar year 1998 payroll tax
contributions to the OASDI portion of the Social Security trust fund were
approximately $430 billion and the HI portion collected $124 billion.10
Social Security benefits are often not taxed. However, higher income individuals
gradually lose tax free status of their Social Security benefit disbursements. The
current tax code favors Social Security savings; however, the return is lower than
most other retirement vehicles even considering the privileged tax status. If payroll
tax funding for Social Security remained under a consumption tax, balances in the
Social Security trust fund would become partially dependent upon the employers
choice of employee compensation. Without the ability to predict employer trade-offs
between deferred compensation and wage compensation, the net impact on the Social
Security trust fund are difficult to project.
Tax-Favored Employer-Sponsored Pensions11
Employer-sponsored pensions do not cover as many participants as the nearly
universal Social Security system, though a majority of working adults have some type12
of retirement benefit other than Social Security. These plans usually enjoy all of the
tax preferences outlined earlier. However, to maintain their tax preferred status most
employer sponsored pensions are subject to nondiscrimination regulations to ensure
that highly compensated employees are not disproportionately favored in the pension
10U.S. Social Security Administration Board of Trustees, 1999 Annual Report.
11For an extensive review of the structure of various plans, see: Library of Congress.
Congressional Research Service. Retirement Plans with Individual Accounts: Federal Rules
and Limits. Report No. 98-171 EPW, by James R. Storey and Paul Graney.
12Social Security covered approximately 141 million workers in 1995. About 56% of full and
part-time workers participated in some type of retirement other than or in addition to Socialth
Security in 1993-1994. Source: EBRI Databook on Employee Benefits, 4 Edition.
Washington, DC: Employee Benefits Research Institute, 1997.
program. Some plan types, usually for small businesses, need not comply with
nondiscrimination regulations. Nevertheless, to maintain tax preferred status most
plans generally must satisfy the coverage and nondiscrimination tests.13
Defined Benefit Pension Plans. Defined benefit pensions provide a target
income for the employee's retirement. Often these plans are integrated with Social
Security to provide a given retirement package that meets a target income. Both
employer and employee contributions are deductible within limits. The limit on
deductible contributions are calculated based upon the future benefit distributions.
Generally, to remain tax favored, the annual indexed retirement benefit cannot exceed
100% of the participant's average compensation over the three highest consecutive
years of compensation. In addition to the benefit limitation, the computed average
annual compensation cannot exceed the indexed dollar amount of $130,000 for
limitations ending in 1999. If the caps are binding on an individual, the benefit
package is often augmented by another, non-tax favored retirement arrangement.
For almost all defined benefit pension plans, full vesting must occur within five
years of continuous employment or seven years under graded vesting. The vesting
requirement is designed to keep the employee with the company long enough to cover
much of the firm’s initial human capital expenditures. If a substantial portion of
retirement earnings are held contingent upon remaining with the firm for the vesting
period, then the firm can potentially recover some of the training and hiring costs
invested in the employee. However, the same objective could be achieved without
holding retirement contributions until employees are vested. Lump sum payments
after a benchmark number of years of employment are arguably a more efficient
method of remuneration for loyalty.
The present income tax favors retirement fund vesting over the lump sum design
through the deductibility of annual pension contributions and the exemption from
payroll taxes. For example, assume that Company X contributes 5% of an employee’s
salary to a defined benefit pension for five years at which time the employee becomes
fully vested.14 If the employee were to leave before becoming vested, the funds
residing in the pension fund remain in the pension trust. The present five year limit
on cliff vesting allows Company X to effectively shelter (avoiding taxation) earnings
in retirement plans for up to five years. Alternatively, a lump sum distribution under
the current income tax is less desirable from the firm’s perspective given the inability
to shelter periodic contributions, as is the case with pension contributions.
A consumption tax would treat loyalty incentives more equally, reducing the
advantage of deferred compensation, although, the existence of the payroll tax would
still discourage the lump sum apparatus and favor the deferred compensation
13The Minimum Participation Test, which does not apply (since 1997) to defined contribution
plans, requires that at least 50 employees participate in the defined benefit plan. If a small
employer, then the greater of 40% of employees or two employees must participate.
14Referred to as ‘cliff vesting,’ which implies that 100% vesting occurs at one point in time
as opposed to gradual increments.
retirement vehicle. However, the diminution of the tax advantage may induce a shift
from retirement pensions to lump sum distributions.15
Another complication affecting the defined benefit pension system when shifting
to a consumption tax is the effect on the insurance agency overseeing defined benefit
pension plans. In response to the expected decline of new participant enrollment, the
Pension Benefit Guarantee Corporation (PBGC) may need to increase premiums to
compensate for funding shortfalls. The increase in administrative costs of maintaining
the benefit plan would further discourage this style of compensation.
In summary, focusing on employer behavior, a shift from the present income tax
to a consumption tax could lead to lower retirement savings in general as employer
defined benefit pension contributions are no longer as tax advantaged. Without the
impetus of employer contributions and incentives, employees with more saving
flexibility may not choose the same level and time profile of saving that currently
characterizes defined benefit pensions.
Defined Contribution Pension Plans. For these plans, the most common of
which are referred to as 401(k) and 403(b) plans, employer contributions up to the
statutory limit of $10,000 are deductible business expenses and the total contribution,
including the employees share, is limited to the lower of 25% of pay or $30,000
annually. Employees defer taxes on accumulated interest and principal until funds are
withdrawn after age 59½. Early withdrawals are assessed a 10% excise tax unless the
participant has died, become disabled, taken an early retirement after age 55, or
incurred excessive medical expenses. Transfers to other tax deferred retirement
accounts may also be exempt from the excise tax.
The 401(k) and 403(b) plans' popularity stems from the comprehensive tax
advantage and the shift of potential investment risk (and return) from the employer
to the employee. Under the consumption tax, if pension contributions are included
in the tax base (as in the USA Tax), employers and employees will no longer enjoy
the tax advantages of contributing to 401(k) plans and as such employees will likely16
opt for direct wage compensation. The impact on saving will depend upon how
individuals choose to spend the additional income.
If the pension plan were inducing saving beyond the employee's target, the
income effect would dominate and saving would decline. On the other hand, if the
employee would save more if not for the strictures on 401(k) balances, then removing17
the liquidity constraints may induce greater saving, albeit possibly short term saving.
15Another factor is the administrative cost of retirement plans. Assuming lump sum plans are
not subject to the Pension Benefit Guarantee Corporation (PBGC) premium and non-
discrimination rules, they may be less expensive to administer. The PBGC is analogous to
the Federal Deposit Insurance Corporation (FDIC) which insures bank deposits.
16If pension contributions are still exempt from the payroll tax, they would still be favored
over wage payments. Consumption tax proposals are mixed in their treatment of the payroll
tax. Flat taxes and value added taxes generally leave the payroll tax intact.
17Though not explored in depth, another factor to consider is the role of anti-discrimination
(continued...)
In summary, employer sponsored defined contribution plans will more than likely
decline in popularity with a consumption tax. Wages and pensions (deferred
compensation) under the consumption tax are treated much more equally than under
the present income tax. If the equality is extended through the elimination of the
payroll tax, the dampening effect on defined contribution plans is even more extreme.
With both defined benefit and defined contribution plans, the typical consumption tax
eliminates all pecuniary incentives for employer sponsored pension plans. As such,
the likely effect on this style of 'forced' saving when moving from the current income
tax to a consumption tax would be to reduce long term saving. However, the
reallocation from employer sponsored plans to those initiated by the employee would
partially compensate for the impact on employer sponsored plans.
Tax-Favored Personal Saving
Tax-favored personal savings gain their advantage through their deferral of
income taxes and on their tax free inside build-up. The deferral can have an added
benefit for those who avoid taxes while still in the higher tax bracket and pay taxes
on deferred income when they enter a lower tax bracket. In contrast to the employer
contributions to employee pensions, contributions to personal accounts are subject to
the payroll tax (i.e., are not deductible) and amounts above statutory limits are
included in the income tax base. The following section focuses on the income tax
treatment rather than the overall tax liability incorporated into the return of the asset.
Individual Retirement Accounts (IRAs). Funds designated as an IRA account
are typically invested through banks, mutual funds, or securities firms.18 These plans
are not salary deferral plans and as such, employers have little direct influence on the
amount held in these accounts. Individuals with wage income can contribute up to
$2,000 per year in an account before income taxes prior to reaching the age 70 ½.
However, there is an income cap for full deductibility for those with an employer plan.
To maintain full deductibility from gross income, the maximum Adjusted Gross
Income (AGI) in 2001 for a single filer is $33,000 and for those filing jointly,
$53,000.19 Without the deductibility from income taxes, the sole advantage of the
IRA is the tax deferral of accumulated earnings until distribution after the age of 59
½. Most early withdrawals are assessed a 10% penalty.
A variation of the above account is the “Roth IRA.” A Roth IRA allows for
after tax contributions which grow without any future tax liability like a tax exempt
17(...continued)
laws presently in the tax code. Generally, retirement plans must follow guidelines that ensure
lower income workers are offered similar retirement benefits of the higher income workers.
To achieve equality between low and high income workers, employers will often induce the
low income employees to contribute to a retirement plan. Replacing the current income tax
with a consumption tax would remove the low-income worker forced saving.
18According to the Investment Company Institute (1998, Table 4), in 1997, 42% of IRA's
were in mutual funds, 38% in brokerages, 13% in banks, and 7% in life insurance companies.
19The deductibility phase out is complete when income for a single filer reaches $43,000 and
$63,000 for joint filers. The range of values increases until 2005 for single filers and through
bond. The annual contribution limit is the same as the standard IRA, $2,000, but
unlike the traditional IRA, contributions are allowed after age 70 ½. Also, the income
phase out of deductibility is set much higher for the Roth IRAs. For single filers, the
phase out begins at an AGI of $95,000 and is complete at $110,000. The phase out
range for joint filers is AGI of $150,000 through $160,000. Distributions are tax free
if the account is held for at least five years and the individual is over age 59 ½. If
those conditions are not met, early withdrawals of funds not previously taxed are
included in gross income and assessed a 10% penalty. However, withdrawals are
treated as coming first from taxed contributions. Younger investors in lower marginal
tax brackets would benefit the most from Roth IRA contributions.
Under a consumption tax, one could think of all saving as an IRA without a limit
except that you could not use existing assets to finance contributions. To analyze the
effect of a switch to a consumption tax on the level of assets in IRAs, it is informative
to subdivide savers into two types: those saving at the IRA limit and those not saving
at the IRA limit. For those at the limit, it is reasonable to assume that they have
additional assets beyond the IRA. More than likely, those additional assets are not
receiving the tax-favored treatment inherent in IRAs. However, in return for lack of
tax advantage, these assets are often more liquid and may have an after tax return
relatively close to the IRA. The change to a consumption tax will then bestow the tax
preference upon these assets which were previously not tax favored, which in turn
induces a shift from IRAs to these accounts. One might then expect limit savers to
abandon IRAs for the now more lucrative liquid assets.
The savers that have not reached the annual IRA limit probably do not have
significant discretionary saving beyond the IRA. Nevertheless, they are also likely to
switch to accounts that exhibit greater flexibility and rates of return, although, the
transition may not be as seamless as for those with other taxable saving accounts
already in place. Whether the entire IRA balance is shifted is dependent upon the
countervailing income and substitution effects alluded to earlier. These savers will be
induced to save more if IRA illiquidity were the predominant reason for saving less
than the limit. The increased liquidity of saving in general would appear to these
savers as an increase in the attractiveness of saving. However, if they are saving for
a target, the income effect would dominate and saving would decrease as the rate of
return has increased.
In summary, with the limit on deductibility for IRAs removed, those that are
presently contributing the maximum amount will likely save in different accounts.
Those that are not contributing at the limit may save less as the income effect
dominates for target savers. Engen, Gale, and Scholz (1994) calculate that only 5.7%
of all IRA contributors were 'falsely constrained' by the contribution limit.20 This
finding indicates that those who would react the removal of the limit are in the
significant minority of IRA savers.
20Falsely constrained meaning that they would have contributed more if not for the
contribution limit. For more information, see: Engen, Eric M., William G. Gale, and John
Carl Scholz. Do Saving Incentives Work? Brookings Papers on Economic Activity. Number
Life Insurance. Life insurance policies are free from taxation of the “inside
buildup.” Also, employers are allowed to deduct life insurance premiums if the
business is not a direct or indirect beneficiary. Though relatively insignificant in terms
of total assets invested, this form of saving does benefit from the current income tax
rules. A consumption tax would expand the deductibility of contributions though
other saving instruments with greater flexibility and would compete for assets
previously held in life insurance.
Owner Occupied Housing. The primary income tax advantage of owner
occupied housing is the absence of a tax on imputed rental income. That is, occupiers
of their own home do not pay tax on the stream of residential services the home, in
effect, provides. A comprehensive income tax would include such income. The best
way to exhibit this is through an example. First assume two individuals purchase
homes next to each other. If they trade places and rent to one another then they both21
would claim their rental income as regular income. The absence of a tax on the
imputed rental income creates an advantage for owner occupied housing over other
types of investment.
There are two other tax preferences for housing relative to other consumer
goods often cited (albeit for itemizers): income tax exemption for mortgage interest
payments and a deduction for local property taxes paid. Another advantage of
housing investment is the ability to exclude capital gains taxes on sales. Gains from
home sales (after May 6, 1997) are not included in taxable income (capped at
$250,000 for single filers and $500,000 for joint filers). The strongest condition for
receiving this tax break is that the taxpayer must have lived in the old house for at
least 2 of the 5 years before the sale to exclude the capital gain. Home ownership is
clearly a strongly favored asset under the current income tax.
With a consumption tax where taxes are paid when funds are removed from a
saving account, the effect on housing stock is uncertain. The financing and prices of
homes, however, will adjust. Home purchases can be financed with either debt or
equity. Drawing down financial assets to purchase a home with equity becomes
expensive because the consumption tax will apply to net changes in the savings
account balance. Buying a house with debt, on the other hand, becomes relatively
more lucrative. Paying off the mortgage appears as a net gain to the saving account
and can lower tax liability. Many may even refinance homes to take advantage of this
tax preference. As for the price of housing, Capozza, Green, and Hendershott
suggest that the price of housing will decline as the previous tax advantages, including22
the mortgage interest and property tax deductions, are eliminated. Indeed, with
lower prices, long term adjustments may include a decline in the stock of housing or
a slower growth than would have taken place under the present tax system.
21Indeed, one could make the argument that all consumer durables are tax favored. Although
true, homes represent a much larger expenditure and are most often considered assets.
22See: Capozza, Dennis R., Richard K. Green, and Patric H. Hendershott. Taxes, Mortgage
Borrowing, and Residential Land Prices. In Aaron, Henry J. and William Gale, eds.
Economic Effects of Fundamental Tax Reform. First Edition, 1996. The Brookings
Institution, p. 196. They estimate that prices will fall by approximately 10%.
Taxed Personal Saving
The current income tax does not have any provisions that favor the following
types of investments. However, the capital gains tax that applies to them in some
cases has been gradually reduced thus decreasing the tax liability over time.
Equities. Corporate equities have perhaps the least favorable income tax
treatment. Corporations are taxed on net income and thus indirectly so are the
shareholders. In addition, dividend payments and realized capital gains that occur
throughout the year are taxed again through the individual income tax. The double
taxation of shareholder’s investment in corporations discourages this type of saving.
However, the potential return of equity investments has been significantly greater over
the long term than other investment options.
Investment in equities by individuals, directly or through mutual funds, may
become more popular under a consumption tax. In theory, however, equity prices
under a flat tax style consumption tax would decline in value. Generally, new capital
investment is immediately expensed under the flat tax which raises the effective rate
of return of new capital relative to old capital. Thus, existing equity prices of firms
with old capital should drop to compensate for this lower relative return.
In addition, pension funds which currently hold a mix of equities, bonds and
cash, may shift away from equities into less volatile assets such as bonds to reflect the
generally risk averse nature of individuals. Presently, pension managers design
pension plans with long term growth objectives in mind which on most accounts
includes a significant share of assets held in equities. When individuals have a greater
role in asset allocation decisions, it is uncertain whether they will choose the same
profile of assets. Possibly, risk averse individuals will choose less volatile asset mixes
which could possibly imply a smaller share of equities.
The increase in demand for direct ownership of equities resulting from their new
found tax preference may be accompanied by a drop in demand for equities as short
term saving arrangements with less volatility become more prevalent and equity prices
in general decline. In short, it is difficult to predict which effect will prevail.
Financial Intermediary Instruments. Short term saving accounts such as
certificates of deposit (C.D.’s), traditional passbook savings accounts, and retail non-
retirement mutual fund accounts, generally do not possess any income tax
advantages.23 The liquidity offered by these savings vehicles explains much of their
popularity. If the relative tax disadvantage of short term savings accounts were
eliminated, the level of saving in these accounts would almost surely increase under
a consumption tax. Assets held in these accounts will clearly expand as their tax
disadvantage vanishes with the consumption tax. However, the additional funds in
these accounts may not represent new saving, rather a shift from previously tax
advantaged accounts, such as IRAs and other retirement accounts.
23However, many mutual fund families have tax advantaged funds that invest in income tax
free municipal debt.
Direct Business Investment. Capital investment by a small business is in a
sense saving. Entrepreneurs invest with the hopes that the business will earn a rate
of return that exceeds that of other potential investments. The tax code does little to
encourage such investment other than immediate expensing for capital expenditures
by small businesses and accelerated depreciation deductions. For tax year 2001,
$24,000 of business investment in qualified property (generally equipment) is
considered a business expense, which is essentially consumption tax treatment. This
limit rises to $25,000 in 2003 and is available only to qualified small businesses. Once
$200,000 worth of qualified property is put in place, the provision is gradually phased
out.
If all business investment were to qualify as saving under a proposed
consumption tax (the USA tax, for example), assets held in business investments
should theoretically increase. A consumption tax would appear to encourage new
investment as capital purchases are immediately deductible, without limits. Under our
present income tax, the deductibility is roughly stretched over the life of the
investment through depreciation. Thus, there could be an increase in new business
investment when moving to a consumption tax given the front loaded deductibility.24
Economic theory would also predict the consumption tax to encourage capital
investment over labor. Thus, saving in the form of business investment eligible for
expensing would likely increase under a consumption tax.
Summary
The current tax code clearly favors some types of saving over others. Most
economists would suggest that the variety of saving instruments available has
complicated the tax code and created distortions in taxpayer behavior. One could
argue that current tax provisions serve to encourage those who would not save for
retirement to do so. Others contend that inducing individuals to move into illiquid
retirement saving may not necessarily be desirable.
There are two general points to keep in mind. One, the relative tax treatment of
various saving vehicles would change under a consumption tax. Therefore, the
distribution of saving would also change. And two, since some saving is already tax
favored, its relative treatment may not be greatly improved by any of the
contemplated changes.
Following is a review of current literature addressing the individual’s saving
decision. Generally, there are macroeconomic issues such as the national saving rate
and microeconomic issues such as individual choice of saving instrument. Tax reform
which introduces some variant of a consumption tax has implications for both
branches of research.
24Somewhat paradoxically, existing capital owners may not fare as well. Thus, capital
intensive industries may not initially benefit from a move to a consumption tax.
Saving Literature Review
The purpose of this literature review is to explore the effect on saving when
moving from the current income tax to a stylized consumption tax. Certainly the
answer is dependent upon how individuals respond to a change in the relative price
of saving, especially retirement saving. Much of the existing literature on retirement
saving has focused primarily on the affect on saving of some external change in the
tax rules affecting a particular saving incentive. Others have explored a more general
question, that is, what happens to saving when the rate of return changes. Both
branches of the literature are relevant to the analysis of a consumption tax where all
saving vehicles are treated equally. Though not addressing precisely the same
question, previous literature can still be used successfully as a guidepost for explaining
what might happen when a consumption tax is introduced.
Tax Reform and Saving Literature
Recently, academic interest in retirement saving under a consumption tax has
intensified. Research by Christopher D. Carroll and Andrew A. Samwick introduced
an expanded life cycle model to explain the potential effects on saving when moving
to a consumption based tax.25 The basic life-cycle model is motivated by the
observation that individuals have a lifetime earnings profile that begins relatively low,
rises to a peak around mid-life, then declines after retirement. Consumption, on the
other hand, is relatively constant over time, perhaps growing gradually as one ages.
Thus, to maintain constant consumption throughout retirement, individuals should
save for retirement during their peak earnings years.
In Carroll and Samwick's expanded specification, saving can also be motivated
by the desire to hold a 'buffer-stock' of saving or precautionary balances.26 The
inclusion of precautionary balances, which are relatively insensitive to changes in
interest rates, reduces the overall sensitivity of saving to changes in the rate of return.
An increase in the rate of return of saving, be it through more favorable tax treatment
or changes in market conditions, thus does not have the same impact on the level of
saving as would be the case in the absence of buffer stock saving. The higher return
allows for the same future balances to be achieved with smaller annual contributions.
According to Samwick, switching to a consumption tax, which increases the rate of
return of all saving, would then not increase saving as significantly as some have
25Carroll, Christopher. Buffer-Stock Saving and the Life Cycle/Permanent Income
Hypothesis. Quarterly Journal of Economics. vol. 112, no. 448. February 1997; Engen,
Eric M., William G. Gale, and John Carl Scholz. Do Saving Incentives Work? Brookings
Papers on Economic Activity. Number 1, 1994; and Carroll, Christopher D. and Andrew A.
Samwick. How Important Is Precautionary Saving? Review of Economics and Statistics.
vol. 80, no. 3. August 1998. pp. 410-419, have all explored precautionary balances or
'buffer stock' saving.
26The amount held as a buffer is some target portion of permanent income.
suggested. Samwick concludes that "the theory and evidence presented in this paper
strongly suggest that this [tax reform] will result in a reduction in private saving."27
The inclusion of precautionary balances, or as some suggest, income uncertainty,
is not altogether new. Engen and Gale refer to previous analysis noting that:
the effects of consumption taxes within a stochastic life-cycle model [a
model that includes, for example, uncertainty about future income,
unemployment, disability or illness] yields increases in saving that are as
much as 80 percent smaller than those produced by a certainty life-cycle
model. 28
The analysis of Samwick agrees in principle with the proposition that the saving
sensitivity to the rate of return is not as strong as the life-cycle hypothesis would
suggest.29 To that end, Engen, Gravelle, and Smetters compare estimates of the
saving response to a change in the rate of return which accompanies consumption tax
reform.30 Two of the dynamic tax models examined are based on the life-cycle
hypothesis; one with precautionary saving, the other without. The inclusion of
precautionary balances or earnings uncertainty “...yields gains in savings and output31
that are only half the size of the gains when uncertainty is turned off.” The analysis
of Samwick and the finding of Engen, Gravelle, and Smetters serve as evidence that
the change in saving under a consumption tax is dependent upon how interchangeable
long term saving is with precautionary saving.
It is then not surprising that a consistent theme throughout the saving incentive
debate is whether or not the funds flowing into tax preferred accounts represent new
saving. On the one hand, if growth in the tax preferred accounts is new saving then
the incentive is effective. On the other hand, if the funds are simply transferred from
other saving accounts then the incentive simply benefits those who are already
saving.32 Further, if the tax expenditure were funded through public debt, the saving
incentive may actually reduce national saving by way of reduced federal tax revenues.
27See: Samwick, Andrew A. Tax Reform and Target Saving. NBER working paper 6640.
July 1998, p. 19.
28See: Engen, Eric M., William G. Gale, and John Carl Scholz. The Illusory Effects of Saving
Incentives on Saving. Journal of Economic Perspectives. vol. 10, no. 4. Fall 1996, p. 94.
29See: Samwick, Andrew A. Tax Reform and Target Saving. NBER working paper 6640.
July 1998.
30See: Engen, Eric, Jane Gravelle, and Kent Smetters. Dynamic Tax Models: Why They Do
the Things They Do. National Tax Journal. vol. L, no. 3. September 1997,657-682.
31Ibid., p. 678.
32If the profile of current savers is reviewed, one finds that the wealth of those that save is
significantly higher than those who do not save, a fact consistent with the argument that
saving is a 'normal good'. The tax break is then enjoyed by those that are relatively well off,
which would tend to flatten the progressivity of the income tax.
Specific Saving Vehicle Literature
The two saving vehicles most commonly studied to identify the saving response
to tax changes are the IRA and the 401(k). These saving devices are differentiated
in mechanics as well as participation profile. IRAs are personalized, retirement
accounts where the employer has little if any direct role in contributions. 401(k)s, on
the other hand, are contingent upon employer sponsorship and thus are influenced
directly by employer behavior. As for who participates, Hubbard and Skinner (1996)
note that: "Households who contribute to IRA accounts tend to be wealthier, older
and have higher incomes than those who do not."33
The literature reviewed below is structured to reflect the prominent role the tax
treatment of IRAs and 401(k)s have had in explaining what happens to savings when
the price of doing so changes. Both sides of the debate present rather vigorous
arguments to support their views.
IRAs. Venti and Wise assume that individuals choose between three avenues
for their income: tax preferred saving (IRAs), regular saving, and consumption. If tax
preferred saving is not a perfect substitute for regular saving, then the introduction of
a saving incentive (an IRA) will necessarily be funded partially from a reduction in
consumption. Using data from the Consumer Expenditure Survey, they find that "...if
the IRA limit were raised, about two-thirds of the increase in IRA saving would be
funded by a decrease in current consumption and about one third from a reduction in
taxes; only a very small proportion would come from other saving."34
There are two related criticisms of this study. One arises from the fact that as
the individual saving decision function is modeled, all non-IRA investment is assumed
to be relatively homogeneous.35 In reality, non-IRA saving can be any manner of
saving, some very similar in objective to an IRA and others very dissimilar. Clearly,
those components which are dissimilar will tend to drive the lack of substitutability
and ultimately the Venti and Wise results. A better specification as suggested by
Gravelle would group IRA's and other illiquid long term saving together as an
argument in the decision function which "...recognizes that IRAs will be perfect
substitutes for S2 [long term] saving but not necessarily for consumption or short term
saving."36 With this specification, whether IRA balances represent new saving is
uncertain.
The second criticism arises from the observation that many IRA contributors do
so at the limit. Individuals contributing at the limit, usually the financially well off,
33See: Hubbard, R. Glenn, and Jonathan Skinner. Assessing the Effectiveness of Saving
Incentives. Journal of Economic Perspectives. vol. 10, no. 4. Fall 1996, p. 75.
34See: Venti, Steven F. and David Wise. Have IRAs Increased U.S. Saving?: Evidence From
Consumer Expenditure Surveys. Quarterly Journal of Economics. vol. 5, issue 3. August
35See: Gravelle, Jane G. Do Individual Retirement Accounts Increase Savings? Journal of
Economic Perspectives. vol. 5, no. 2. Spring 1991.
36Ibid., p. 142.
simply transfer funds from other saving accounts to take advantage of the tax savings
inherent in an IRA. The fact that they contribute at the limit is tacit confirmation that
the saving incentive is not marginally effective in inducing additional saving.
With the consumption tax, where all saving is tax preferred, the comparative tax
benefit of IRA style saving is effectively removed. Consequently, the marginal
incentive to save becomes independent of the relative prices of different saving
vehicles. Rather, the choice is between consuming or saving directly, not between tax
preferred and non-tax preferred. Thus, the literature describing the effects of the
incentive effects of IRAs relative to other types of saving is not directly applicable
when analyzing the changes in saving upon a move to a consumption tax. What can
be gleaned from the IRA literature is how individuals respond to changes in the rate
of return of non-consumed income. When offered the IRA option, which increased
the rate of return of saving, there is no consensus among economists about the effect
on overall saving. However, there is a shift from other types of saving into retirement
saving. The second vehicle, the 401(k), requires a slightly different analysis.
saving is actually new saving. Utilizing the feature of 401(k) plans that eligibility is
contingent upon the employer, they compare the level of financial assets of non-
eligible to eligible workers. If the 401(k) incentive is effective, then the level of
financial assets of eligible workers should exceed that of non-eligible workers,
controlling for income and demographic characteristics. They show that for 1987, the
ratio of total financial assets of eligible employees to non-eligible was 1.62 and in38
1991, 2.22. Workers eligible for 401(k) programs appear to be lured into saving
more than their ineligible counterparts. This evidence is quite convincing but there
are potential flaws.
Engen, Gale, and Scholz provide a series of arguments that question the above
'experiment'.39 They do not disagree with the general assumption that 401(k)s will
increase saving for some, though they do dispute the size of the effect. Following are
their arguments. One, comparing a cohort which is eligible for a saving incentive plan
with an otherwise similar cohort which is not eligible for a saving incentive plan,
biases upward the effect of saving incentives. It is likely that the eligible cohort has
sought out and remained with an employer that offered the 401(k) saving plan. Thus,
the eligible cohort has a "... systematically stronger taste for saving than other40
households." Two, the Poterba, Venti, and Wise approach examines total assets as
opposed to net assets; it is conceivable that individuals have assumed more debt to
37Poterba, James M., Steven Venti, and David Wise. Personal Retirement Saving Programs
and Asset Accumulation: Reconciling the Evidence. NBER working paper 5599. May 1996;
and Poterba, James M., Steven Venti, and David Wise. How Retirement Saving Programs
Increase Saving. Journal of Economic Perspectives. vol. 10, no. 4. Fall 1996. pp. 91-112.
38Ibid, p. 99.
39See: Engen, Eric M., William G. Gale, and John Carl Scholz. The Illusory Effects of Saving
Incentives on Saving. Journal of Economic Perspectives. vol. 10, no. 4. Fall 1996, pp.113-
40Ibid., p.114.
contribute more to 401(k) programs. Three, substitutes for 401(k)s (e.g., social
security) and financial markets themselves have changed significantly over their period
of study thus limiting the reliability of the estimates. Four, the level of saving in
401(k)s is pretax which overstates their value. And finally, the Poterba, Venti, and
Wise approach ignores the total compensation package when comparing individuals
by income. Eligible workers are also receiving matching employer contributions
which should be considered income when comparing individuals based upon income.
Even though the benefit is deferred compensation, it is still compensation and should
be included as income.
National Saving Literature41
Another perspective on the debate is the effect on national saving incentives.
The estimated revenue loss on saving incentives is $56.2 billion in 1998.42 Some have
argued that this amount, if financed through larger budget deficits or smaller
surpluses, actually reduces national saving.43 To counter this argument, Martin
Feldstein suggests that the appropriate method of estimating the cost and benefit of
saving incentives (in his study, the incentive is the IRA which accounts for $3.3 billion
of the above tax expenditure estimate) is to include the growth in investment that
produces a taxable revenue stream at the corporate level.44 Through including this
revenue, Feldstein finds the cost of IRAs in terms of tax revenue lost is offset by
about one-third over the first five years and two-thirds over the first ten years.45
Engen, Gale, and Scholz again provide a counter argument to the above
findings.46 They argue that the estimates above are inflated for several reasons.
Perhaps the most contentious is the Feldstein assumption that at least half of IRA
saving is new saving; an assertion that is refuted by other studies sited earlier. With
such uncertainty concerning the degree to which IRA balances represent new saving,
predicting tax revenues well into the future is difficult.
41Again, for a more comprehensive review of the national saving literature see: Library of
Congress. Congressional Research Service. Saving in the United States: Why Is It
Important and How Has It Changed? by Brian Cashell and Gail Makinen. Report 98-580.
June 1, 1998.
42See: EBRI Databook on Employee Benefits, 4th Edition. Washington, DC: Employee
Benefits Research Institute, 1997., Table 5.3, p. 42-43. Government employers represent the
largest share of this sum.
43See: Engen, Eric M., William G. Gale, and John Carl Scholz. Do Saving Incentives Work?
Brookings Papers on Economic Activity. Number 1, 1994. pp. 85-151. They address the
effect on national saving of saving incentives.
44See: Feldstein, Martin. The Effects of Tax-Based Saving Incentives on Government
Revenue and National Saving. Quarterly Journal of Economics. Vol. 110, No. 2, May
45Ibid., p. 478.
46Engen, Eric M., William G. Gale, and John Carl Scholz. Do Saving Incentives Work?
Brookings Papers on Economic Activity. Number 1, 1994. pp. 85-151.
The Evidence in Sum
Taken in sum, the literature does not definitively support the proposition that the
use of tax incentives for saving increases private or national saving, at least in the
short run. The contentious debate is not surprising; economic theory suggests that
the tradeoff between the income and substitution effects associated with a tax cut for
saving yields ambiguous results. What we do know is a shift to a consumption tax
will effectively remove the preference for illiquid IRA contributions and encourage
a shift to short term, liquid saving. If individuals are as sensitive to the favorable tax
treatment of IRAs in a positive direction (as suggested by Venti and Wise), then
removal of the relative preference for IRAs will diminish their popularity. However,
as noted above, the applicability of the estimates provided by research into the
incentive effects of IRAs is questionable.
The literature concerning 401(k)s would also seem to suggest that upon removal
of the tax preference for retirement plan contributions, assets held in these accounts
would likely decline given the lack of employer encouragement through matching
contributions. As for saving incentive effects on national saving, it appears that a
consensus in the literature has not been reached.
Conclusion
Generally, an efficient tax regime minimizes the distortion of taxpayer behavior.
A consumption based tax which eliminates the peculiarities and preferences inherent
in the treatment of saving would eliminate many of the saving incentives presently in
place. In place of those incentives, the consumption tax would instead offer a broadly
applied incentive to save. A critical question explored here is whether the existing
incentives accomplish a similar objective. If the current array of incentives in the
present income tax are effective, then the consumption tax will not significantly
encourage more saving when they are removed.
Even though the potential increase in efficiency arising from less distortion in
taxpayer behavior might be reason enough for its proponents to endorse a
consumption based tax, it is not clear that saving under a consumption tax, relative
to the present income tax, would actually increase beyond existing levels. What is
fairly certain is that the allocation of saving assets will adjust. Short term saving will
increase and previously tax favored long term saving such as retirement will actually
decline.
References
Aaron, Henry J. and William Gale, eds. Economic Effects of Fundamental Tax
Reform. First Edition, 1996. The Brookings Institution.
Allen, Everett T., Joseph J. Melone, Jerry S. Rosenbloom, and Jack L.
Vanderhei. Pension Planning: Pension, Profit-Sharing, and Other Deferred
Compensation Plans. Eighth edition, 1997. Irwin/McGraw-Hill.
Carroll, Christopher. Buffer-Stock Saving and the Life Cycle/Permanent Income
Hypothesis. Quarterly Journal of Economics. vol. 112, no. 448. February 1997.
pp. 1-56.
Carroll, Christopher. Why Do the Rich Save So Much? NBER working paper
Carroll, Christopher D. and Andrew A. Samwick. How Important Is
Precautionary Saving? Review of Economics and Statistics. vol. 80, no. 3. August
Commerce Clearing House. 1999 U.S. Master Tax Guide. 82nd Edition, 1997.
CCH Incorporated.
Diamond, P.A. and J.A. Hausman. Individual Retirement and Savings Behavior.
Journal of Public Economics. vol. 23, 1984. pp. 81-114.
EBRI Databook on Employee Benefits, 4th Edition. Washington, DC: Employee
Benefits Research Institute, 1997.
Engen, Eric M., William G. Gale, and John Carl Scholz. Do Saving Incentives
Work? Brookings Papers on Economic Activity. Number 1, 1994. pp. 85-151.
Engen, Eric M., William G. Gale, and John Carl Scholz. The Illusory Effects of
Saving Incentives on Saving. Journal of Economic Perspectives. vol. 10, no. 4. Fall
Engen, Eric, Jane Gravelle, and Kent Smetters. Dynamic Tax Models: Why
They Do the Things They Do. National Tax Journal. vol. L, no. 3. September
Feldstein, Martin. The Effects of Tax-Based Saving Incentives on Government
Revenue and National Saving. NBER working paper 4021. March 1992.
Feldstein, Martin. The Effects of Tax-Based Saving Incentives on Government
Revenue and National Saving. Quarterly Journal of Economics. May 1995. pp.
Gravelle, Jane G. Do Individual Retirement Accounts Increase Savings?
Journal of Economic Perspectives. vol. 5, no. 2. Spring 1991. pp. 133-148.
Gravelle, Jane G. The Economic Effects of Taxing Capital Income. First
edition, 1994. Massachusetts Institute of Technology Press.
Hubbard, R. Glenn, and Jonathan Skinner. Assessing the Effectiveness of Saving
Incentives. Journal of Economic Perspectives. vol. 10, no. 4. Fall 1996. pp. 73-90.
Investment Company Institute. Fundamentals: Investment Company Institute
Research in Brief. vol. 7, no. 2. July 1998.
Kusko, Andrea, James Poterba, and David Wilcox. Employee Decisions with
Respect to 401(k) Plans: Evidence from Individual-Level Data. NBER working
paper 4635. February 1994.
McDonnell, Ken, Paul Fronstin, Kelly Olsen, Pamela Ostuw, Jack VanDerhei,
and Paul Yakoboski. EBRI Databook on Employee Benefits. Fourth Edition, 1997.
Employee Benefits Research Institute.
Musgrave, Richard A., and Peggy B. Musgrave. Public Finance in Theory and
Practice. Fourth Edition, 1984. McGraw-Hill, Inc.
Ozanne, Larry. What Have We Learned About IRAs and Saving? National Tax
Association Papers and Proceedings 1996.
Papke, Leslie, Mitchell Petersen, and James Poterba. Did 401(k) Plans Replace
Other Employer Provided Pensions? NBER working paper 4501. October 1993.
Poterba, James M., Steven Venti, and David Wise. Personal Retirement Saving
Programs and Asset Accumulation: Reconciling the Evidence. NBER working paper
Poterba, James M., Steven Venti, and David Wise. How Retirement Saving
Programs Increase Saving. Journal of Economic Perspectives. vol. 10, no. 4. Fall
Samwick, Andrew A. Tax Reform and Target Saving. NBER working paper
Stiglitz, Joseph E. Economics of the Public Sector. Second Edition, 1988.
W.W. Norton and Company.
Thaler, Richard, and H.M. Shefrin. An Economic Theory of Self-Control.
Journal of Political Economy. vol. 89, no. 2. April 1981. pp. 392-406.
Thaler, Richard. Psychology and Saving Policies. American Economic
Association Papers and Proceedings. May 1994. pp. 186-192.
Venti, Steven F. and David Wise. Have IRAs Increased U.S. Saving?: Evidence
From Consumer Expenditure Surveys. Quarterly Journal of Economics. vol. 5, issue