The central contest over competing visions for federal taxes for
the next few years began on October 31. That is when Chairman
Rangel (D-NY) of the House Ways and Means Committee shepherded the
Temporary Tax Relief Act of 2007 (H.R. 3996) through his committee.
Citing analysis by the Joint Committee on Taxation (JCT), Chairman
Rangel described the bill as nearly revenue neutral, saying it
would raise $484 million over 10 years. Under the approach taken in
this bill, however, taxes would rise by nearly a trillion dollars
over 10 years. The difference isn't a matter of math or policy, but
The Congressional Budget Office (CBO) forecasts future receipts
and spending levels, and Congress uses these forecasts to inform
their deliberations over fiscal policy. The forecast, or baseline,
used for tax legislation is based on "current law" and assumes that
expiring tax provisions--like Alternative Minimum Tax (AMT) relief,
or the "patch"--will not be renewed. This stands in sharp contrast
to the CBO assumption that "current policy" with respect to federal
spending will continue even if a spending program is scheduled to
expire. For example, the CBO spending baseline assumes the
continuation of the State Children's Health Insurance Program
(SCHIP) even though that policy expired in 2007.
The result of the CBO receipts baseline assumption is that a tax
bill that appears revenue neutral when scored, like Chairman
Rangel's, can actually be a tax hike. To eliminate this structural
bias favoring higher taxes and to bring greater accuracy and
balance to tax policymaking, the CBO should revise its methodology
for expiring tax provisions to bring it in line with its practices
for spending programs. In the meantime, Congress should extend the
patch for 2007. And because extending the patch is not a tax cut
but rather the prevention of a tax increase, extension of the patch
should not be accompanied by other tax increases.
Where a journey ends often depends on where the journey begins.
This year's journey into threatened tax hikes begins with the CBO
receipts baseline. The CBO has, for many years, constructed a tax
receipts forecast for the federal government based on its forecast
for the economy and on current tax law. While the economic forecast
occasionally generated controversy, the use of current law led to
In part, the assumption of current law for tax policy was
non-controversial because current tax law usually extended current
policies, without interruption or change, into the future. For
example, next year's tax rates are usually the same as current
rates. Exceptions, such as the R&D tax credit and related "tax
extenders," have been relatively minor in terms of overall tax
policy. Thus there was no apparent reason to question the current
law assumption. Now, however, the current law assumption has become
a serious issue, and it is apparent that it has always been the
AMT and the Baseline
The AMT patch shines a bright spotlight on the CBO's current law
assumption in the receipts baseline. The AMT is a parallel income
tax; taxpayers pay the larger of the AMT and regular income tax
liability. The AMT patch is an increase in the AMT exemption
amount. It was first legislated in 2001 to ensure that taxpayers
would not be denied their regular income tax relief by falling
under the AMT. The AMT patch expired at the end of 2006
and would be extended for 2007 by H.R. 3996.
The AMT patch has, like the R&D tax credit, typically been
extended one year at a time. Yet the patch differs in two key
respects. First, the AMT patch represents an enormous amount of tax
revenue. The JTC scores the 2007 patch at almost $51 billion,
whereas a one-year extension of the R&D credit reduces revenues
by just under $9 billion. Second, the AMT patch prevents this tax
hike from falling on about 18 million individuals and families,
rather than a group consisting mostly of large, multinational
Though the issues the AMT patch raises in terms of the
formulation of the CBO baseline are not new, the magnitude of the
revenues and the number of taxpayers involved make correcting the
CBO baseline most urgent. Formulating the revenue baseline on the
basis of current law is neither reasonable nor justifiable for the
AMT patch, the R&D tax credit, or any other tax provision.
In 2006, millions of taxpayers were protected from the AMT by
the patch. In 2007, absent congressional action and the President's
signature, those millions will lose their protection and face an
average tax increase of over $2,000; some will face a tax increase
of over $5,000. From the perspective of these taxpayers, if the AMT
patch is not extended, Congress will have put into effect a tax
increase of $51 billion for one year and nearly a trillion dollars
over 10 years if this same approach were followed year after
Chairman Rangel has repeatedly acknowledged that failure to
extend the AMT patch for 2007 would result in an unacceptable tax
increase on millions of unsuspecting taxpayers. In this, he is
surely correct. His legislation would extend the patch for 2007,
thereby preventing a tax hike. But because the CBO baseline
reflects current law, the effect of extending the patch for 2007 is
not in the baseline, because the patch requires a change in the
law. Thus, extending the AMT patch is shown as a tax cut in the
For the overall bill to be shown as revenue neutral in the
official scoring, the Rangel bill increases taxes by $51 billion on
other taxpayers. Chairman Rangel is proposing to raise taxes on
other taxpayers to offset a tax cut that is not a tax cut. In
short, under this logic, allowing the AMT patch to expire is a tax
increase, but extending the AMT patch and raising $51 billion in
tax revenue elsewhere is not a tax increase. This is illogical.
Comparing Revenue and Spending
The flaw in the CBO baseline formulation is further highlighted
by contrasting the revenue and spending baselines. Most
congressional and administration budget deliberations on the
spending side are relative to current policy. The spending
baseline shows what spending would look like in
future years under current policy, adjusting for inflation, certain
legislated and regulatory changes that have not yet taken effect,
and changes in the population.
Consider, for example, SCHIP. The SCHIP program expired at the
end of September 2007. In formulating the current policy baseline,
the CBO assumed spending would continue in 2008 and beyond at the
level set in 2007. Thus, the debate in Congress is about the extent
of the change in the program beyond what is provided under current
If the CBO reflected the SCHIP program in the spending baseline
the way it treats the AMT patch in the revenue baseline, then SCHIP
would be assumed to expire for all time after 2007--that is, the
baseline would show no SCHIP funding in 2008 or any year
thereafter. This would not be a reasonable budget baseline for
SCHIP, and it is not a reasonable way to present the AMT patch.
Faulty Baseline Renders PAYGO
The problem with the CBO revenue baseline is brought into stark
relief by Congress's adoption of pay-as-you-go (PAYGO) rules. These
are general rules Congress imposes on its own deliberations to
constrain legislation from increasing the budget deficit. For
example, under the PAYGO rules now in effect, in most cases
legislation that changes taxes or non-appropriated spending may not
increase the deficit unless Congress votes to waive the rules.
PAYGO rules sum up the effects of changes in receipts and
spending relative to a budgetary starting point. That starting
point consists of the baseline forecasts for receipts and spending.
If those baseline forecasts include systematic biases or errors,
then PAYGO rules can fundamentally misinform legislators and the
public about the issues under consideration and critically bias the
outcome of the debate. This is exactly what is happening with
legislation to extend the AMT patch.
If the CBO revenue baseline were formulated correctly, then the
revenue score of extending the AMT patch would show it to be
revenue neutral, just as legislation to extend the SCHIP program at
current levels is budget neutral. Further, legislation such as
Chairman Rangel's H.R. 3996 that extends the AMT patch and includes
various tax increases as revenue offsets would correctly show a
significant tax increase. If there is no tax cut, then a revenue
offset is just a tax hike.
What Should Congress Do?
The CBO provides Congress with information needed to make
decisions. It is an agency of the congressional branch, and its
methods, products, and structure all reflect congressional
oversight and funding. Congress, therefore, has full latitude to
direct the CBO to alter its procedures and methods as needed and
The CBO itself may recognize the flaw in its methodology and
desire to correct the flaw, but it may feel bound by precedent to
continue with past practices until told or at least cleared to do
otherwise. In any event, the flaw in the baseline methodology can
be corrected by the CBO, and Congress can ensure that the flaw is
corrected by clearly instructing the CBO to do so. For example,
appropriately worded Sense of the Senate and Sense of the House
resolutions should accomplish the task.
These resolutions would state the desire of the Congress that
the CBO present unbiased, complete information and that, in so
doing, it is important that tax policies and spending policies be
treated alike in the budget baselines, using a consistent set of
rules based on current policy assumptions. In this way, legislation
extending a program without changes would have no budgetary effect
for scoring purposes, whereas programmatic changes that increase
(or decrease) spending would continue to be clearly scored as such.
Similarly, tax provisions that expire would be treated in the
receipts baseline as though they were extended indefinitely. Thus,
legislation that extends a tax provision would not be scored as
changing projected receipts.
Before the year is out, Congress should and likely will extend
the AMT patch for 2007, thereby preventing a $51 billion tax hike
that would otherwise hit millions of families. Congress should not
use this as an excuse to raise taxes by a like amount on some other
group of taxpayers.
Some Members' efforts to use the extension of an AMT patch as an
opportunity to raise taxes are greatly facilitated by a serious and
long-standing flaw in the CBO's methodology for developing its
revenue baseline. Unlike its spending baseline, which is formulated
based on current policy, the revenue baseline is calculated based
on current law. There is no justification for constructing these
The effect of the CBO baseline flaw is that efforts to prevent a
tax hike on millions of families are portrayed as a tax cut, rather
than preventing a tax increase. And because of Congress's PAYGO
rules, which rely on the CBO baseline and JTC scoring relative to
that baseline, these efforts to prevent a tax hike must themselves
be accompanied by tax hikes or else Congress must waive the
Congress should recognize this serious flaw in the CBO
methodology. It should waive the PAYGO rules for purposes of
passing an AMT patch extension. It should not use this as an
opportunity to raise taxes. And Congress should direct the CBO to
remedy the flaw in its methodology and treat both spending and
taxes on a current policy basis when developing next year's
J. D. Foster, Ph.D., is Norman
B. Ture Senior Fellow in the Economics of Fiscal Policy in the
Thomas A. Roe Institute for Economic Policy Studies at The Heritage
a more complete discussion of the AMT and the AMT patch, see J.D.
Foster, "Making Good Policy Out of a Bad AMT," Heritage Foundation
Backgrounder No. 2082, October 31, 2007, at www.heritage.org/Research/Taxes/bg2082.cfm,
and Joint Tax Committee, Present Law and Background Relating to
the Individual Alternative Minimum Tax, staff report,
JCX-10-07, March 7, 2007, at www.house.gov/jct/x-10-07.pdf.
Joint Tax Committee, "Estimated Revenue Effects
of the Chairman's Amendment in the Nature of a Substitute to H.R.
3996, 'The Temporary Tax Relief Act of 2007,'" JCX-105-07, at www.house.gov/jct/x-105-07.pdf.
current policy baseline is sometimes called the "current services"
baseline in budget parlance.