It's The Sunsets, Stupid!  CBO and JCT Should Show the Extended Cost of Expiring Provisions
Jeff Lemieux
May 3, 2004

The Budget Resolution for 2005 has been delayed by a dispute over "pay-as-you-go" rules.  But even the best pay-as-you-go rules can be circumvented politically by the creative use of "sunset" provisions, those abrupt terminations of tax cuts or spending programs intended to disguise the true long-term cost.  Sunset provisions are not always meant to be deceptive. In fact, it often makes sense to force Congress to re-evaluate spending programs or tax cuts at a later date.  However, the Congressional leadership is currently using sunset provisions for less honorable purposes.

There is a surprisingly easy solution.  The Joint Committee on Taxation (JCT) and the Congressional Budget Office (CBO) should show the full, 10-year cost of any spending or tax change that is scheduled in the law for expiration (and for which an extension is an obvious possibility).  Moreover, JCT should show the impact of a particularly deceptive sunset provision relating to the Alternative Minimum Tax (AMT) when it estimates the revenue effects of other tax bills. 

These extra calculations would not replace the official estimates or change "scoring" conventions.  Performing them wouldn't require a change in law or congressional rules.  JCT and CBO would simply show the 10-year cost of affected provisions (without the sunset assumption) in the footnotes or "memorandum" lines of their regular revenue and cost estimate tables.  Instead of waiting for an end-of-year compendium of the cost of extending sunset provisions, members of Congress and the public could see the potential long-term cost before the bills were enacted.

Outline:
How Sunset Provisions Distort the Budget
  Example #1, "Section 179" Expensing
  Example #2, The 2002 Tax Cuts
  Example #3, Medicare Physician Payment
Even Good Pay-As-You-Go Rules Can Be "Gamed" Using Sunsets
JCT and CBO Should Show 10-Year Extended Estimates

How Sunset Provisions Distort the Budget:  Congress has learned a new way to play games with the budget.  In the 1990s, Republicans and Democrats in Congress generally played the budget straight, insisting that all new spending programs or tax cuts be "offset" by equivalent budget cuts or tax increases.

Some of the offsets used in the 1990s were a little gimmicky, and some -- particularly some of the cuts in Medicare payments to health providers -- were destined for subsequent repeal or adjustment.

But in general, if legislators wanted to pass a 10-year, $100 billion tax cut, they really did "pay for" it with ten years' worth of offsetting spending cuts or tax increases. 

This "pay-as-you-go" discipline was very effective, helping to bring the federal deficit down from almost $300 billion in 1992 to virtually zero by 1997.  A subsequent surge in economic growth lifted the budget to a surplus of $236 billion in 2000 (see Figure 1).

Figure 1.



However, now Congress has learned a devastating new budget gimmick:  the sunset provision. 

Since 2001, legislators have used misleading sunset provisions to vastly understate the real long-term cost of several large tax cuts and some smaller spending provisions.  This is a major reason why the budget has plummeted into deep deficits.

To be fair, some laws are truly intended to expire on their sunset dates.  In theory, this is not a bad way to legislate.  Forcing Congress to re-evaluate spending programs or tax provisions in the future can be a good idea.

However, the sunset provisions in most recent legislation are intended simply to disguise the long-term cost of legislators' real intent:  permanent spending or tax cuts.  The use of sunsets grossly distorts the budget process and can make a mockery of budget rules.

Table 1 compares the official revenue estimates -- which are often cited in Congress and the media -- with the long-term cost of several recent tax provisions that contain sunset provisions.

Assuming the original intent was for these tax cuts to be permanent, the sunset provisions effectively hid as much revenue loss as was originally shown in the official estimates.

---

Table 1.

Notable Tax Laws with "Sunset" Provisions
(billions of dollars)
			Cost To Extend
		Original 10-Year	Permanently
		"Sunset" Score	(2004-2014)
2001 Income and Estate Tax Cuts		1,350	969
2002 and 2003 Business Tax Cuts		159	474
2003 Investors' Tax Cut		148	191
Memo:
Alternative Minimum Tax (AMT) Reform			300-900?
Sources: Centrists.Org, Joint Committee on Taxation,
Congressional Budget Office.
Note: The AMT often interacts with tax cuts, in effect negating their stated effects.  The
cost of many tax cuts would be much higher if the AMT thresholds were subsequently
raised or indexed, or if the AMT was repealed.

---

Moreover, the revenue lost from some of these tax cuts was already understated because of a particularly nefarious sunset provision in the Alternative Minimum Tax (AMT).  The AMT effectively raises the taxes or negates the tax cuts of many middle- and upper-income taxpayers.  In fact, the Congressional Budget Office (CBO), in a recent study, predicts that 90 percent of taxpayers with incomes between $100,000 and $500,000 will pay the AMT -- and therefore not receive as much of the scheduled tax cuts as advertised -- by 2010.

Estimates vary for the cost of ending the sunset on "fixes" to the AMT.  It depends on which other tax cuts are also assumed to extend beyond their sunset dates.  Estimates of repealing the AMT altogether, assuming other tax cuts scheduled in law are extended past their sunset dates, range as high as $1 trillion over the next 10 years.

Here are some particular examples of how sunset provisions can hide the real cost of Congressional initiatives.

Example #1, "Section 179" Expensing in the House Highway Bill.  Tucked into the tax measure that accompanies the House highway bill is a two-year extension of a tax cut for small businesses that was enacted in 2003 and is scheduled to sunset in December 2005.  

We have nothing against this particular tax cut, but it serves to illustrate the larger budget problems with sunset provisions very nicely. 

First, take a close look at the graph of the official budgetary cost of the two-year extension (see Figure 2).

Figure 2.



According to Congressional custom, a provision such as this is said to "cost" a mere $1.3 billion in lost revenue over the next ten years (the Congressional budgeting period).

However, Figure 2 shows that the near-term costs are much greater than that -- about $10 billion.  Those near-term costs are very likely to actually occur.

The official long-term revenue cost of extending this sunset provision is only $1.3 billion because after the sunset date -- presuming it wasn't subsequently extended by Congress -- revenue would increase, in a sort of "catch up" pattern.

But this is unlikely to actually happen, because Congress will keep extending the provision, never allowing it to expire on its sunset date.

Table 2 shows that the original revenue estimate when the Section 179 provision was enacted in 2003 was only $1.0 billion, because dubious new revenues after the original sunset (December 2005) were embedded in the total. 

However, the cost of eliminating the sunsets altogether is a cool $33.8 billion.  In fact, the cumulative cost of this provision between 2003 and 2014, assuming Congress keeps pushing back its sunset dates, would be about $55-60 billion.

---

Table 2.

Revenue Loss (-) from Section 179 Expensing
(Billions of Dollars)
		First 5 Years	Full 10 Years
Enactment in 2003 (sunset 2005)		-4.5	-1.0
Extension in 2004? (sunset 2007)		-10.9	-1.3
Permanent Extension (2006 - 2014)		-19.5	-33.8
Sources: Centrists.Org, Joint Committee on Taxation

---

Example #2, The 2002 Tax Provisions.  The main business tax cut enacted in 2002 was bonus depreciation, which has the effect of reducing tax liabilities for many businesses.  This tax cut was expanded (made more generous) in 2003.  

Congress will probably extend bonus depreciation beyond its scheduled expiration in December 2004, if for no other reason than to avoid a sudden reduction in corporate net profits (and a drop in stock market values).  In fact, the sunset date was set at December 2004 for political reasons.  Bonus depreciation gives businesses an incentive to make investments prior to the sunset date, which will presumably boost economic activity this fall -- during the political campaigns -- instead of in 2005, when the elections will have been decided.

But the main point is that the original estimate of the revenue cost (shown in Figure 3) vastly understated the long-term cost.  When it was enacted in 2002, the 10-year revenue loss was estimated at only 17.9 billion -- because of the sunset provision and "catch up" revenues" -- while the cost in the first three years alone was $108 billion.  

In 2003, Congress expanded this tax cut at a two-year cost of $44 billion.  However, the long-run, 10-year cost of the two-year expansion totaled only $9 billion due to the sunset provision.  In both the 2002 and 2003 estimates, the expectation of "catch up" revenues after the sunset -- which Congress will probably not allow the government to collect -- dramatically lowered the long-term estimate.

Figure 3.



Example #3, The Physician Payment Provision in the Medicare Drug Bill.  Spending provisions are also susceptible to sunset-like provisions, even if the law doesn't actually include explicit expiration dates.

For example, the physician payment system in Medicare is a mess.  In the Balanced Budget Act of 1997, Congress attempted to link Medicare's overall spending on physician services to GDP growth.  The idea was to create a budget for Medicare spending, and adjust fees paid to doctors so that the budget was met. 

However, the resulting fee setting system turned out to be much more volatile than expected, and the data used by Medicare administrators to compute the fees remains in dispute.  In the last several years, the system cut the fees to physicians more than was politically feasible.

As a result, Congress has been issuing short-term "fixes" to the physician fees -- in effect overriding the fee setting system on an ad-hoc basis.

In the Medicare Rx drug bill of 2003, Congress raised doctors' fees in the short term, but allowed the system to continue to generate huge predicted fee cuts in the future.  Thus the short term "fix" -- which costs money now -- creates the illusion of even larger long-term savings, which are incorporated into the official cost estimates (see Figure 4).

As a result, the official cost estimate of the drug bill showed that fixing physician payments was "budget neutral" -- that is, the short-term costs were matched by long-term savings.  However, everybody knows Congress will never really allow those savings to occur.

Figure 4. 



Congress won't allow an old fee system to cut doctors' payments by 5 percent a year for the next 7 or 8 years.  However, that's what's in the law, and that's what the official cost estimates and baseline show. 

Even Good Pay-As-You-Go Rules Can Be "Gamed" Using Sunsets:  Pay-as-you-go is an extremely important budgeting rule when it is used in good faith.  However, even good budget rules can be worked around by tricksters in Congress, especially in combination with a perverse use of budget rules.

The budget reconciliation process -- which allows tax and spending changes to pass Congress without stalling techniques or supermajority votes -- was designed to help legislators muster the courage to reduce budget deficits.  Over the last 3 years, however, it has been used to enact tax cuts that make the deficit much worse.

For example, suppose the Congressional budget resolution allows a tax cut of $100 billion over 10 years under the reconciliation process.

We would expect Congress to enact a tax cut that costs roughly $10 billion a year, right?  That would be the logical way to reduce taxes by $100 billion over 10 years.

Modern Congressional leaders are far too clever to do that, however.  To get the biggest possible tax cut, they are more likely to enact a tax cut that costs $100 billion a year, but sunset the tax after only 1 year.  Alternatively, they could enact a tax cut that costs $50 billion a year for two years.  Either way, the official revenue estimate would still say $100 billion over ten years.

In effect, this dares the subsequent Congresses to let the tax cut expire on schedule -- and be called "tax raisers" -- or pass their own extensions.  (Often, letting the tax cuts expire abruptly as the law dictates would cause enough economic disruption that no one, not even opponents of the tax cuts, will allow it to happen.) 

Congress can also use sunsets to sidestep pay-as-you-go rules.  Suppose Congress has a $10 billion spending cut over 10 years that it wants to use to offset the cost of a tax cut.

Congress will enact a tax cut of $1 billion a year for 10 years, right?  Wrong.  The current crop of Congressional leaders is more likely to cut taxes by $10 billion in the first year, and then -- via a sunset provision -- require the tax rate to snap back up to its previous level.  This would allow the current Congress to say it "paid for" its tax cut.  But the next Congress would have to figure out a way to extend the cut, or suffer the economic disruption.

JCT and CBO Should Show 10-Year Extended Estimates:  There is a simple solution to the sunset problem.  As sunlight is said to be the best disinfectant, the bright light of official cost and revenue estimates should be shined on sunset provisions. 

JCT and CBO should show the full 10-year cost of any tax cut or spending program that is directly or indirectly subject to a sunset provision in the "memorandum" lines or the footnotes of their official cost and revenue estimates. 

It would be easy to estimate the 10-year cost of provisions with direct expiration dates.  Usually JCT or CBO estimate the long-term cost of a provision anyway, and calculate the impact of a sunset provision as a second step.

It might be a little more difficult to compute the impact of indirect sunset provisions.  Sometimes a sunset provision in one part of the tax code or spending law can cause the cost of another provision to be understated.  The most notable example is the sunset provision on the indexation of the AMT, which causes the estimated revenue loss from income tax cuts to be dramatically understated.

However, JCT already shows the number of taxpayers subject to the AMT on many of its official estimates.  It shouldn't be hard to add footnote lines showing the extended revenue loss for provisions affected by substantial indirect sunsets.

Showing the impact of sunsets in the official cost and revenue estimates would allow journalists, members of Congress, and the public to have an authoritative source for the real cost of important Congressional proposals.

That way, journalists could describe both the short-term and potential long-run cost of proposals using estimates from the same authoritative source, before they are enacted. 

There is probably no way to ban the use of sunset provisions.  Creative and devious legislators -- intent on short-term political gains and blind to longer-term budgetary realities -- will always flock to gimmicks and distortions to achieve their goals.

Besides, we don't really want to ban sunset provisions.  Congress really should make some programs or tax changes temporary.

But shining the light of official cost estimates on the sorts of deceptions sunset provisions can create (even if it's just in the footnotes) would go a long way toward reducing policy misunderstandings and improving the public budget process. 

Links:
Centrists.Org A "Duck-The-Issues Budget -- Interpreting the Congressional Budget for 2005 (revised March 19, 2004)

Center on Budget and Policy Priorities Revised House Tax Cut Bill Swells Deficit and Continues Less Favorable Treatment of Lower-Income Couples by Joel Friedman (revised April 29, 2004)  This article, citing unpublished JCT estimates, argues that the "marriage penalty" tax cut recently passed in the House would cost almost double its official estimate if sunset provisions relating to the AMT were extended.

Congressional Budget Office The Alternative Minimum Tax (April 15, 2004)

Congressional Budget Office Budget and Economic Outlook, Table 4-10 "Effect of Extending Tax Provisions That Will Expire Before 2014" (updated February 11, 2004)

Joint Committee on Taxation Estimated Revenue Effects of Title IX of H.R. 3550 the House Highway Reauthorization Tax Act (April 1, 2004)

Joint Committee on Taxation Estimated Budget Effects of the Revenue Provisions Contained in the President's Fiscal Year 2005 Budget Proposal (March 3, 2004)

Joint Committee on Taxation Estimated Budget Effects of the Conference Agreement on H.R. 2 "The Jobs and Growth Tax Relief Reconciliation Act of 2003" (May 22, 2003)

Joint Committee on Taxation Estimated Revenue Effects of the "Job Creation and Worker Assistance Act of 2002" (March 6, 2002)

Joint Committee on Taxation Estimated Budget Effects of the Conference Agreement for H.R. 1836 (Bush Tax Cut of 2001) (May 26, 2001)