Revised May 21, 1997

The Tax Cuts in The Budget Agreement:
Phase-ins, Backloading, And Revenue Acceleration
Lock in Very Large Future Costs

by Iris J. Lav and Robert Greenstein

The fine print of the budget agreement indicates that the agreement envisions heavily backloaded tax cuts. The budget agreement documents indicate that the tax cuts will ultimately explode in cost and lose far more revenue over the long term than has been understood.

A table included in the budget agreement documents and entitled "Long Range Summary 1997-2007" includes numbers on the cost of the tax cut each year through 2007. These numbers are consistent with the year-by-year reconciliation instructions in the budget resolution for tax cuts in 1998 through 2002. As is well known, the net tax cut totals $85 billion over the first five years and $250 billion over the first 10 years. But if the tax legislation follows the year-by-year path set forth in the table, the tax cut will cost vastly more than this after the initial 10-year period ends.

The table in the budget agreement package shows that the net tax cut will begin to explode toward the end of the 10-year period, growing 32 percent in just the final two years of the period (i.e., between the eighth year and the tenth year).

Of the $250 billion cost of the tax cut over the 10-year period, 44 percent of the cost — $109 billion — is packed into the last three years.

The net tax cut would lose $41.6 billion in 2007, the tenth year. That is two and one-half times the tax cut's $17 billion average annual cost in the first five years.

Because critical features of the tax provisions would continue to grow in cost in years subsequent to 2007, the tax cut is likely to lose approximately $650 billion in the second 10-year period, from 2008 through 2017. This is the period when the baby boom generation begins to retire and deficits are forecast to shoot up to alarming levels that begin to damage the economy unless unprecedented austerity measures are instituted.


Why Does The Revenue Loss Number in the Agreement Explode in This Fashion?

The year-by-year revenue loss figures in the budget agreement table make clear that the agreement authorizes use of an array of backloading devices. In fact, the agreement accommodates tax cuts more backloaded than the tax cuts included in S. 2, the Senate Republican Leadership tax bill that Senator Trent Lott introduced in January 1997. S. 2 itself contained a heavy dose of backloading devices, but its costs did not rise as steeply in the final years of the 10-year period as the costs of the tax cuts in the budget agreement. These backloading devices would evidently be used to enable Congressional tax-writers to shoehorn as much of S. 2 as possible, along with education tax cuts, into the five-year and 10-year revenue figures the agreement contains.

An examination of the agreement's year-by-year revenue loss figures shows that these figures have peculiar characteristics. The revenue loss amounts jump sharply in some years, then decline in subsequent years, and then rise again. Specifically, after nearly doubling in cost between 1999 and 2000, the revenue loss levels off in 2001 and declines in 2002. Once 2002 passes, the revenue loss jumps again. It rises 34 percent in a single year between 2002 and 2003, keeps rising after 2003, and starts to explode in the years immediately preceding 2007. (See Figure 1 - below)

This pattern is not accidental. The year-by-year revenue loss figures reflect the cost of specific tax cuts, designed in a manner that minimizes their cost in the initial five-year and ten-year periods through delayed effective dates, slow phase-ins, and acceleration into the budget period of revenue that otherwise would be collected in future years.

Capital gains indexing is a case in point. The year-by-year revenue loss numbers reflect the inclusion of capital gains indexing, with the effective date for indexing delayed until January 1, 2001 and with several other backloading features as well. This produces an indexing provision that causes no revenue loss until 2004 or 2005 and actually increases revenue in 2000 and 2001. (The revenue increase in these years largely reflects the acceleration of capital gains tax revenue that otherwise would be collected in later years.) An identical backloaded indexing provision was part of the vetoed 1995 reconciliation bill.

As explained below, designing a capital gains indexing provision in this manner allows congressional tax cutters to shoehorn both capital gains indexing and a large capital gains rate cut or exclusion into the budget agreement totals, since capital gains indexing would not begin to lose money until the seventh or eighth year. Moreover, because a capital gains indexing provision of this nature would raise revenue in 2001 and especially in 2002, it has the perverse effect of financing additional tax cuts during the first five years.

Eventually, the capital gains indexing proposal would lose substantial amounts of money. In 1995, the Joint Committee on Taxation estimated the cost of the capital gains indexing provision included in the "Contract with America." The Joint Tax Committee found that the revenue loss from indexing would be three times as large in the second five years it was in effect as in the first five years.

Administration Rebuffed on Backloading Curbs

In the latter stages of the budget negotiations, the Administration proposed that the documents setting forth the tax cut component of the budget agreement state that the tax cuts would not include timing devices such as deferred effective dates and phase-ins that cause mushrooming costs in the out-years. The Administration also proposed language stating that the tax cuts would not include capital gains indexing. All of these proposals were rejected. None appear in the budget agreement.

The year-by-year revenue numbers in the budget agreement table also can accommodate an Individual Retirement Account provision that relies heavily on backloading gimmicks. In fact, IRA tax cuts can be designed in a manner that includes such a heavy dose of backloading that the tax cuts increase revenues in the first five years. This result is achieved by accelerating into the five-year budget window some billions of dollars in revenue that otherwise would be collected in subsequent decades. IRA tax cuts backloaded in this manner ultimately lose very large amounts of revenue each year.

Cost in Second Ten Years is Approximately $650 Billion

The backloading gimmicks responsible for the sharp rise in the cost of the tax package between the eighth and the tenth year also will continue to push up the cost of the tax cuts in years after 2007. The tax cut is likely to lose between $600 billion and $700 billion in the 10-year period from 2008 through 2017. Several methods of estimating the cost the tax cuts would have during the second 10-year period produce similar results.

One method of estimation is to assume that the cost of the tax cuts will grow by the same amount in each year from 2008 through 2017 as the average amount that the tables in the budget agreement show the cost will grow from 2004 through 2007. Using that assumption, the annual cost of the tax cut would rise from $41.6 billion in 2007 to $85 billion in 2017. Estimated in this manner, the total cost of the tax cut in the 10-year period from 2008 through 2017 would be $654 billion.1

To arrive at a more conservative estimate, one could assume that the backloading features would not cause costs to grow by the same amount throughout all of the second 10-year period. One could assume, for example, that costs through 2012 would continue to rise each year by the average annual amount that costs would rise from 2004 through 2007, but that after 2012, the cost of the tax cut would simply rise in tandem with the growth of the economy.2 Given the nature of capital gains indexing and of the backloaded IRA provisions, this approach may understate the revenue loss in the second 10 years. Under these assumptions, the cost of the tax cut would rise from $41.6 billion in 2007 to $79 billion in 2017, and the cost in the 10-year period from 2008 through 2017 would be $636 billion.

Using the average annual increment in cost between 2004 and 2007 as the basis for projection, as both of these methods do, does not fully capture the rate at which the tax cut is growing in cost at the end of the first 10 years. A third potential method of estimation would assume that the average rate at which the tax cut would grow between 2004 and 2007 would continue through 2010; for the remaining seven years through 2017, the growth in cost is assumed — under this third method — to fall back to the rate of GDP growth. This also appears to be a somewhat conservative assumption, since the effects of backloading are likely to result in growth in the cost of the tax cuts that exceeds the growth of the economy for more than three years beyond the initial 10. Under this set of assumptions, the cost of the tax cut would rise from $41.6 billion in 2007 to $83 billion in 2117, and the cost in the 10-year period from 2008 through 2017 would be $670 billion.

1 The increases in cost of the tax cut in the budget agreement year-by-year path are $2.8 billion from 2004 to 2005, $4.8 billion from 2005 to 2006, and $5.4 billion from 2006 to 2007. The average increase of $4.33 billion understates the extent to which the tax package is growing at the end of the budget period.
GDP growth is assumed to be 4.6 percent each year. This is the growth rate for the years 2006 and 2007 in the Congressional Budget Office March 1997 baseline.

Since the tax cuts would be heavily backloaded and the revenue losses are heavily concentrated in the final years of the budget agreement, one might assume these tax cuts could later be reversed or suspended if deficits began to grow again. This, however, is not the case. Large portions of these revenue losses would be locked in permanently.

This is because capital gains indexing and backloaded IRAs both utilize devices that induce taxpayers to pay additional taxes in the first five years of the budget agreement in exchange for very large tax breaks in later years. Once taxpayers have entered into such a bargain and paid additional taxes upfront, the government cannot renege on its promise of future tax breaks. Moreover, once taxes that otherwise would be collected in future years are accelerated into the initial five years covered by the budget agreement — as would be the case with both the capital gains indexing and IRA proposals — these taxes cannot again be collected in the future. The revenue "hole" created in future years is permanent.


Spending Initiatives and Tax Cuts

One noteworthy aspect of the budget agreement is that the Clinton domestic initiatives included in the package shrink over time in inflation-adjusted terms while the tax cuts grow larger.

In 1999, the domestic initiatives — at $6.1 billion — equal 54 percent of the $11.3 billion in net tax reductions.

Over the first five years as a whole, the domestic initiatives amount to 37 percent — or about three-eighths — of the net revenue reductions.

By 2007, the domestic initiatives are eroding in inflation-adjusted terms, having been held at $7 billion for four consecutive years. Meanwhile, the tax cuts have mushroomed to $41.6 billion a year. By this point, the domestic initiatives equal less than 17 percent — about one-sixth — of the net revenue reduction.

Over the second 10-year period, the initiatives would continue to shrink in relation to the net tax cut, eventually equaling about one-tenth of the size of the net tax cut.


Upper-Income Tax Cuts Grow in Relation to Middle-Income Tax Cuts

This shift over time in domestic initiatives as compared to tax cuts almost certainly would be paralleled by another shift — the erosion over time of middle-class tax cuts compared to upper-income tax cuts. Virtually all of the growth in the out-year costs of the tax cuts reflected in the budget agreement's 10-year table appears to be due to the mushrooming cost of three tax cuts primarily benefitting upper-income taxpayers — capital gains indexing, backloaded IRA tax cuts, and estate tax cuts. Although the middle-income tax provisions of the tax package initially constitute most of the tax relief provided, they dwindle over time as a proportion of the tax cuts overall. As the cost of the tax cut package burgeons in cost, the package increasingly becomes a tax relief package heavily tilted to upper-income households.

The capital gains and estate tax cuts and the expanded IRA provisions all are likely to be designed in ways that result in modest or no net revenue losses over the first five years of the agreement. As a result, the provisions that primarily benefit middle- income families — the child tax credit and the education initiatives — are likely to account for at least three-quarters of the cost of the package in the first five years.

In the second five years, however, the cost of the high-income tax cuts would begin to grow while the cost of the child tax credit is likely to begin declining (because the child tax credit is not expected to be indexed; it is not indexed in the Republican tax bills). It is probable that less than half of the tax cuts would be devoted to the middle-class tax benefits in the second five years. The capital gains, estate tax, and IRA tax breaks — which provide most of their benefits to upper-income taxpayers — would account for the majority of the tax relief during this period.

The high-income tax cuts would continue to grow as a proportion of the total tax package each year through 2007 and beyond. By the tenth year, it is likely that at least two-thirds of the total tax cut would be concentrated in the upper-income provisions.

During the first five years, the tax legislation to implement the budget agreement is likely to provide a greater share of its benefits to the middle class than the Senate Republican Leadership tax bill (S. 2) would have provided. By the end of the 10-year period, however, the distribution of tax reduction benefits is likely to be similar to that in S. 2. In the second five years of the Senate Republican Leadership tax bill, tax cuts aimed primarily at middle-income families would account for just 33 percent of the tax cuts. By the tenth year, less than 30 percent of the tax benefits in S. 2 are devoted to middle-income relief.(1)


The Design of Backloaded Tax Cuts

The budget agreement specifies the total amount of the tax cuts over the next 10 years and the types of tax cuts to be included in the package. The "side letter" from the Congressional leadership to President Clinton states: "It was agreed that the net tax cut shall be $85 billion through 2002 and not more than $250 billion through 2007. We believe these levels provide enough room for important reforms, including broad-based permanent capital gains tax reductions, significant death tax relief, a $500 per child tax credit, and expansion of IRAs." The letter goes on to say: " was agreed that the package must include tax relief of roughly $35 billion over five years for post-secondary education, including a deduction and a tax credit. We believe this package should be consistent with the objectives put forward in the HOPE scholarship and tuition tax proposals contained in the Administration's FY 1998 budget to assist middle-class parents."(2)

The agreement does not include details on the design of the tax cuts. The specificity of the year-by-year revenue loss numbers and the uneven pattern of those numbers, however, indicate they reflect a particular package and tax cut design. While the tax-writing committees can establish their own policies, the year-by-year revenue loss numbers included in the agreement suggest that one of the Congressional leadership's goals is to fit as much of its tax proposals (as reflected in S. 2) into the legislation as possible. Through aggressive backloading, that goal can be accomplished to a larger extent than many policymakers, journalists, and other observers have understood.

How could a heavily backloaded package fit the major components of S. 2 plus the President's education initiative into the tax cut levels specified in the budget agreement? As the following discussion explains, this is possible through the creative use of phase-ins, backloading, and revenue-acceleration techniques.


The Child Tax Credit

The child tax credit is likely to be phased in. For example, the credit could start at $300 in 1998, and rise to $400 in 1999 and $500 in 2000. A phase-in would likely be necessary to hold the cost of the child tax credit within the year-by-year revenue loss path reflected in the budget agreement table. The full $109 billion cost of the child tax credit in S. 2 over the first five years could not fit within the $85 billion net tax cut specified in the budget agreement.

It also is possible that the age limitations on eligibility and/or the income levels at which the child tax credit is phased out will be a compromise between the President's proposal and the Senate Leadership tax bill. The Senate Leadership proposal would allow families with incomes up to $150,000 to receive some child tax credit, while the Administration plan would limit receipt of the credit to families with incomes below $75,000. The Leadership credit would extend to children under age 18, while the Administration credit would be available to families for children under age 13. Eligibility limitations based on age and/or income could themselves be phased in, further holding down the near-term cost of the tax cut.

As under the Senate Leadership plan, the year-by-year revenue numbers in the budget agreement table imply that the $500 child credit would not be indexed for inflation and thus would likely decline in cost in the latter years of the 10-year period, as well as in years after that.(3) The tax credit in the Senate Leadership tax bill would cost $19 billion less in the second five years than in the first five years.

Capital Gains

The capital gains tax cut is likely to be similar to the one included in the November 1995 reconciliation bill vetoed by President Clinton, which included both a 50 percent capital gains exclusion and capital gains indexing. Under the 1995 bill and this year's Senate leadership tax bill, half of the profits from the sales of assets would be excluded from income. The legislation implementing the tax cuts in the new budget agreement could include a similar type of exclusion; alternatively, the maximum rate on capital gains income could be lowered in lieu of an exclusion. In addition, as noted earlier, the capital gains indexing provision of the vetoed 1995 legislation is expected to be part of the forthcoming tax bill. As noted above, the year-by-year revenue path in the budget agreement table was designed to accommodate capital gains indexing starting in 2001.

In the first few years following enactment of a capital gains exclusion or rate cut, only modest revenue losses would occur. The Joint Committee on Taxation assumes that if the effective rate of taxation on capital gains is reduced substantially, investors will respond by selling a large volume of assets to take advantage of the lower tax rates. The additional asset sales produce additional tax revenue in the short run, although most of this revenue simply represents an acceleration of revenue from subsequent years when the assets would otherwise have been sold. Within the five-year budget window, however, these accelerated revenue collections partially offset the revenue losses that result from reducing the effective capital gains tax rate. This keeps the size of the revenue loss from the capital gains tax cut much lower in the first five years than would otherwise be the case.

Beginning in 2001, a second, large capital gains tax cut would be added. From that point on, the purchase price of assets held for at least three years would be indexed for inflation when capital gains tax is figured at the time the asset are sold. This would essentially give investors a double tax break — in calculating the capital gains tax, part of an investor's profit would be shielded from taxation by indexing, while the remaining profit would be taxed at a much lower effective rate than under current law. As the box on page 10 explains, many very wealthy investors would actually pay about 10 percent of their profits in capital gains taxes, rather than 28 percent as at present; they would secure a capital gains tax cut of more than 60 percent.

The inclusion of capital gains indexing also substantially increases the backloading of the tax cuts. Indexing would apply to the sale of assets held for at least three years after the starting date in 2001. As a result, indexing would begin to apply to taxes paid on assets sold in 2004; the capital gains indexing provision would result in no revenue losses before that time. The losses then would mount with each passing year after 2004. (Assets purchased in 2001 and sold in 2004 would qualify for three years of indexing; assets purchased in 2001 and sold in 2011 would qualify for 10 years of indexing. The cost of indexing would consequently rise for a number of years, during which it would be shielding a steadily growing share of capital gains profits from taxation.)

Delaying the start of indexing until 2001 would result in a one-time increase in revenues in 2001 and 2002. Revenues would increase in these years because the indexing provisions would include an additional device that was part of the indexing provisions of both the vetoed reconciliation bill and S. 2. The additional device would accelerate the collection of a substantial amount of revenue that otherwise would be collected in future years into fiscal years 2001 and 2002. The device works as follows.

Capital Gains Indexing Results in Tax Rate Lower than Lowest Rate on Wages

The combination of a provision excluding half of the profits from the sale of assets from capital gains taxation and an inflation indexing provision (the capital gains provisions included in S. 2) would allow a substantial majority of capital gains income to escape any taxation. Consider a stock initially purchased for $100,000 and sold after five years for $134,000. Under current law, the investor would pay tax at a maximum rate of 28 percent on the $34,000 capital gain; the tax would equal $9,520. Under a 50 percent exclusion plus indexing, however, the price for which the investor purchased the stock would be adjusted for inflation and then only half the remaining gain would be subject to taxation.

Assuming inflation of three percent per year, the $100,000 purchase price first would be adjusted upward to approximately $116,000 to account for five years of inflation; the investor’s capital gain thus would be considered to be $18,000 (the $134,000 sale price minus the indexed purchase price of $116,000) rather than $34,000. Fifty percent of this $18,000 gain would be excluded, making only $9,000 of it taxable. If the taxpayer was in the top income tax bracket (the 39.6 percent bracket), the taxpayer would pay tax of $3,580. This would represent a 62 percent tax reduction from the $9,520 in capital gains tax the investor would pay under current law.

Looked at another way, under current law, the investor would pay tax equal to 28 percent of the $34,000 profit. Under the combination of a 50 percent capital gains exclusion and capital gains indexing, this investor would pay tax of $3,580 on this gain, an amount equal to only 10.5 percent of the $34,000 profit. This wealthy investor would pay a lower effective rate of tax on this profit than moderate-income families must pay on their wages and on interest they receive on modest savings accounts.

Investors who own assets on January 1, 2001 would be allowed to qualify these assets for indexing from 2001 forward. They would do so by paying capital gains taxes — at the new, lower effective tax rate on capital gains income — on the increase in the value of their assets from the time the assets were purchased until 2001. The bulk of these taxes would be paid in fiscal year 2002. Most of these taxes would not constitute additional revenue collections for the Treasury; to the contrary, this device accelerates the collection of revenues that otherwise would be collected in future years when the assets actually are sold.

Figure 2 (call 202-408-1080 for a faxed copy) shows the 10-year revenue loss caused by the capital gains provisions of the vetoed 1995 reconciliation bill. That bill included the same capital gains provisions, with the same backloading gimmicks, as those described here. As the Figure shows, those capital gains tax cuts would have produced a one-year revenue spike in 2002 but then began to lose substantial revenue, with the losses mounting as the years passed. The year-by-year revenue loss numbers in the budget agreement table are tailored to reflect a similar capital gains revenue loss pattern.

An indication of how large the future losses from indexing would be can be gleaned from the Joint Committee on Taxation's estimate of the cost of the capital gains indexing provision in the original "Contract with America." The Joint Committee estimated the cost of this indexing provision separately from the cost of other capital gains tax cut changes.

The indexing provision in the Contract would have allowed indexing for assets held at least one year, and the value of all assets would have been eligible for indexing after the legislation's effective date. In examining this provision, the Joint Tax Committee found the cost of indexing would triple between the first five years and the second five years. In addition, the Joint Tax Committee estimated that the revenue loss from indexing would reach $9.2 billion in the tenth year; the loss was continuing to grow at that point.(4)

In each year between the fifth year and the tenth year, the revenue loss would grow by more than $1 billion, according to these Joint Tax Committee estimates. Revenue losses resulting from indexing would be likely to continue growing at that pace for a number of years after the tenth year.

Estate Tax

The package is likely to include a phased-in increase from $600,000 to either $1 million or $1.2 million in the amount of an estate exempt from estate tax. In the Senate Leadership tax bill, the increase in the amount of an estate that is exempt from estate tax is phased in slowly over eight years in $50,000 annual increments. The exemption thus would reach $1 million for people who die and leave estates in 2004 and subsequent years.

The full revenue impact of this increase in the estate tax exemption would be delayed beyond 2004. Probate and settlement of an estate generally must precede tax filing, and most estate tax returns for people who die in 2004 will be filed for tax years 2005 through 2007 and will not affect revenues until fiscal years 2006 through 2008. As a result, the full impact of the increased estate tax exemption in S. 2 would not be seen until after the 10-year budget period ends in 2007.

The delays between the years for which increases in the estate tax exemption would be effective and the years in which taxes would be collected on the estates of individuals who died would reduce the cost of the cut in the estate tax in the five-year and ten-year budget windows. This is especially true in the five-year period. Although the estate tax exemption would reach $900,000 in 2002 under S. 2, estate taxes collected in fiscal year 2002 would be based on individuals who died in years for which the exemption ranged from $750,000 to $850,000.

These delays also mean that tax writers could decide to raise the estate tax exemption further to $1.2 million, as some Republican Congressional leaders have suggested, by continuing to increase the exemption by $50,000 each year. If tax writers did so, little revenue loss impact from raising the exemption to $1.2 million rather than $1 million would be felt prior to 2007. The vast majority of the additional revenue loss would occur only after the 10-year period ended.

Individual Retirement Accounts

An expansion of eligibility for Individual Retirement Accounts could be accommodated within the costs of the tax package through the creation of a new type of "backloaded" IRA. The most likely IRA expansion would allow taxpayers of all income levels to utilize a new form of "backloaded" IRA, regardless of whether or not they are covered by an employer-sponsored retirement plan.

Under current law, married filers with adjusted gross income exceeding $50,000 are not eligible to make deductible contributions to IRAs if they also have access to an employer-sponsored, tax-favored retirement savings plan. (Eligibility for IRA deductions was eliminated for these taxpayers in the 1986 Tax Reform Act as one of the base-broadening measures included in return for a substantial lowering of the top tax rate, then 50 percent.) The principal goal of IRA tax break proposals now before Congress is to restore the eligibility of these upper-middle and upper-income taxpayers for IRA tax preferences.

Restoring eligibility for a $2,000 IRA tax deduction would, however, be very expensive over the next 10 years. As a result, the IRA expansion likely to be included in tax legislation to implement the budget deal would enable taxpayers of all income levels to use a new type of "backloaded" IRA, under which most of the revenue loss that this tax break causes is postponed until after the end of the 10-year budget period. The vetoed reconciliation bill included a similar provision.

Under a conventional IRA, depositors receive a deduction of up to $2,000 in the year that they make the deposit into their IRA account. When they retire and withdraw funds from the account, the withdrawals are taxed as ordinary income. By contrast, under a backloaded IRA, deposits in the accounts would not be deductible. But interest earned on funds in the account would be permanently free of tax. When depositors withdrew funds from the account after they retired, the withdrawals would be entirely tax free.

Backloaded IRAs thus involve very little near-term revenue loss. But revenue losses mount over time as interest earned on a steadily increasing share of national savings is shielded from taxation.

In addition, virtually all recent IRA proposals that would establish backloaded IRAs have included "rollover" provisions that further reduce the near-term revenue loss from IRA tax breaks and further enlarge the long-term loss. A rollover provision permits taxpayers to take funds that are now in conventional IRAs and transfer them to backloaded IRAs on which all future interest earnings would be tax-free (and from which withdrawals upon retirement would be tax-free). At the time of the transfer, the funds moved from conventional to backloaded IRAs would be taxed as ordinary income. This rollover feature is another gimmick designed to reduce the near-term cost of a new tax break by "borrowing from the future"; it would accelerate into the current budget period some billions of dollars in tax payments that otherwise would be made in future years when IRA account-holders retire.

IRA proposals that combine the creation of backloaded IRAs with a rollover provision have severe backloading effects. In the first five years, such proposals cause little revenue loss since no deductions are taken when IRA deposits are made. Furthermore, the rollover provision would accelerate billions of dollars in tax revenue that would have been collected in subsequent decades into the five-year budget window. As a result, generous IRA tax breaks fashioned in this manner can have little or no cost in the first five years. In fact, the original version of the Contract with America included a backloaded IRA of this type that the Joint Tax Committee estimated would raise $2 billion over the first five years.

But the revenue losses then begin to mount. Backloaded IRA proposals can exceed $10 billion a year in cost by the tenth year, depending on their design. Moreover, the cost continues to mount for a number of years after the 10-year point, as interest on a steadily increasing share of national savings is sheltered from taxation (and as the effects are felt of having accelerated into the 1998-2002 period some billions of dollars in taxes on IRA deposits that otherwise would have been collected in later years). Past Joint Tax Committee estimates indicate that at the end of the initial 10-year period, the cost of backloaded IRA proposals is mounting at a rate of $1 billion a year.

Education Initiatives

The agreement specifies an intent to include in the tax legislation roughly $35 billion over five years in an education tax credit and deduction, consistent with President Clinton's education tax proposals. Nevertheless, it may be noted that the "side letter" from the Republican Congressional Leadership to President Clinton uses different language to describe various components of the tax bill. In the case of capital gains, the word "permanent" is used to describe the tax cut. By contrast, for the education initiatives, the language specifies "roughly $35 billion over five years." If the capital gains, estate tax, and IRA provisions are designed in such a backloaded manner that there is not room for $35 billion in education tax cuts in the second five years, it is unclear what the tax-writing committees will do. The tax legislation the committees write could include a decline in the amount of education credits or deductions available to students and their families in the second five years. Alternatively, Congress could seek to sunset these tax provisions after the initial five-year period.


High-Income Tax Cuts Become Locked-in for Future Years

The two most highly backloaded components of the forthcoming tax legislation are likely to be the capital gains tax cut, which would include indexing plus either an exclusion of a portion of capital gains income or a reduction in the maximum capital gains tax rate, and the backloaded IRA accounts. By 2007, these two provisions are likely to account for a majority of the cost of the tax package.

Both capital gains indexing and the backloaded IRA (along with the rollover provisions) contain features that would be difficult or impossible to repeal or modify once they have been enacted and taken effect. The revenue losses for future years that would be set in motion by the provisions could never be stanched. This is true because taxpayers would pay taxes up-front to "buy in" to these future tax benefits.

For capital gains indexing, the "buy in" is the gimmick that would allow assets currently held by investors to qualify for indexing. To qualify, taxpayers would pay taxes on the gain in the value of their assets between the date an asset was purchased and the date on which indexing would become effective, presumably January 1, 2001. (This is sometimes called "mark-to-market" because it is based on the market price of assets on the date the legislation specifies that indexing begins.) If taxpayers choose not to "buy in" to indexing, no taxes would be due until their assets actually are sold at some point in the future. Each taxpayer who chooses to use the buy-in would be making a financial judgement that the ability to index future gains on assets is worth more to him or her than continued deferral of taxes on the assets until they are sold. Once taxpayers have made that judgement, have accepted the terms of the government's offer, and have paid tax on the gain in the value of assets as of the date that indexing takes effect, the government cannot renege on its part of the bargain. Indexing must remain in place.

Similar circumstances exist with respect to the funds now held in conventional IRA accounts that would be rolled over into the new backloaded accounts. Taxpayers choosing to roll over their funds would be doing so because the ability to earn tax-free interest in future years is worth more to them, given their personal financial circumstances, than the continued deferral of taxes on deposits in current IRA accounts. Once taxpayers trade their continued deferral by paying taxes now in exchange for future tax-free interest, the government must honor its side of the bargain. The interest cannot be taxed.

Furthermore, both of these provisions accelerate into the first five years of the budget agreement many billions of dollars in revenues that otherwise would be collected in future years. Capital gains taxes that would have been collected in future years when assets are sold are collected in fiscal year 2002, as taxpayers qualify assets for indexing. Taxes on IRA deposits and earnings that would have been collected in future years when the holders of the accounts retire are collected over the next four or five years, as taxpayers roll over existing IRAs into the new backloaded accounts. Obviously, once these revenues are collected, they never can be collected again. In this way, the tax package borrows from the future to finance current tax cuts. Once the borrowing occurs, it cannot be undone.


Implications for Long-term Fiscal Health

The increasing costs over time of the backloaded tax reductions discussed in this report would aggravate the very serious fiscal problems the nation faces when the baby boom generation begins to retire. As reports from CBO, the GAO, and other fiscal watchdogs have repeatedly emphasized, the nation will face daunting fiscal challenges in those years. The costs of Medicare, Medicaid and Social Security will swell as the number of elderly individuals mounts, and deficits will rise to dangerous levels. In addition, the ratio of workers to retirees will fall to a low level.

The budget agreement sets in motion reductions in spending for discretionary programs, Medicare, and Medicaid. Because the budget cuts must pay, however, for tax cuts that mushroom in cost in the very years the baby boomers begin to retire, the progress the budget agreement makes toward addressing our long-term fiscal problems is considerably more modest than would otherwise be the case. Spending reductions used to finance burgeoning tax cuts are not available to help finance mounting retirement costs and control deficits.

Tax cuts such as those implied by the budget agreement may seem affordable in the short run. But they seem affordable only because they borrow revenues from future years and pass costs on to future generations. With the vanguard of the baby boom generation having already reached age 50, the nation can not afford to budget with this type of sleight of hand.

End Notes

1. Iris J. Lav, Senate Leadership Tax Proposals: Mushrooming Tax Cuts for High-Incomes Taxpayers Would Jeopardize Long-Term Budget Integrity, Center on Budget and Policy Priorities, March 1997.

2. Letter from Newt Gingrich and Trent Lott to The Honorable William J. Clinton, May 15, 1997.

3. Several factors account for the decline in cost. Over time fewer taxpayers near the bottom of eligibility income range would owe sufficient taxes to utilize the credit, because the personal exemption, standard deduction, and Earned Income Tax Credit are indexed to keep pace with inflation. Because the maximum income at which a family can qualify for the child credit is not indexed, there also would be a decline in the number of taxpayers with incomes near the phase-out range who would be eligible for the credit. Finally, there is an interaction between the child tax credit and the income tax provision known as the Alternative Minimum Tax. Because non-refundable credits are not allowed as an offset to minimum tax liability, some taxpayers that are required to pay AMT could lose some or all of their credit.

4. Joint Committee on Taxation, Estimated Revenue Effects of the Tax Provisions Contained in the "Contract with America," February 6, 1995. These losses were in addition to the cost of the 50 percent capital gains exclusion also included in the contract proposal.

Other budget agreement analyses:

The Budget Agreement: An Overview

Medicaid and Child Health Provisions of the Budget Agreement